Handling Qualiities

Evolution of Flying Qualities Analysis: Problems for a
New Generation of Aircraft

by
M. Christopher Cotting
Dissertation submitted to the Faculty of the
Virginia Polytechnic Institute and State University
in partial fulllment of the requirements for the degree of
Doctor of Philosophy
in
Aerospace Engineering
Craig A. Woolsey, Chair
Wayne C. Durham, Co-Chair
Kenneth A. Bordignon
William H. Mason
Leigh S. McCue-Weil
March 29, 2010
Blacksburg, Virginia
Keywords: UAV, Flying Qualities, Flight Mechanics, Education
Copyright 2010, M. Christopher Cotting
Evolution of Flying Qualities Analysis: Problems for a New Generation of Aircraft
M. Christopher Cotting
Abstract
A number of challenges in the development and application of ying qualities criteria for modern
aircraft are addressed in this dissertation. The history of ying qualities is traced from its origins
to modern day techniques as applied to piloted aircraft. Included in this historical review is the
case that was made for the development of ying qualities criteria in the 1940s and 1950s when
piloted aircraft became prevalent in the United States military. It is then argued that UAVs today are
in the same context historically as piloted aircraft when ying qualities criteria were rst developed.
To aid in development of a ying qualities criterion for UAVs, a relevant classication system for
UAVs. Two longitudinal ying qualities criteria are developed for application to autonomous UAVs.
These criteria center on mission performance of the integrated aircraft and sensor system. The rst
criterion is based on a sensor platforms ability to reject aircraft disturbances in pitch attitude. The
second criterion makes use of energy methods to create a metric to quantify the transmission of
turbulence to the sensor platform. These criteria are evaluated with airframe models of different
classes of air vehicles using the CASTLE 6 DOF simulation.
Another topic in ying qualities is the evaluation of nonlinear control systems in piloted aircraft.
A L
1
adaptive controller was implemented and tested in a motion based, piloted ight simulator.
This is the rst time that the L
1
controller has been evaluated for piloted handling qualities. Results
showed that the adaptive controller was able to recover good ying qualities from a degraded air-
craft.
The nal topic addresses a less direct, but extremely important challenge for ying qualities re-
search and education: a capstone course in ight mechanics teaching ight test techniques and
featuring a motion based ight simulator was implemented and evaluated. The course used a mix-
ture of problem based learning and role based learning to create an environment where students
could explore key ight mechanics concepts. Evaluation of the courses effectiveness to promote
the understanding of key ight mechanics concepts is presented.
iii
To my beautiful wife Lisa,
thank you for all your sacrice
in making this possible.
I cant wait for our next adventure together!
iv
Acknowledgments
No work of signicance can be accomplished by one person alone, and there have been many that
have accompanied me on this journey of growth academically, personally, and professionally. First
I would like to thank my God for fullling his promise to provide in all things; even when the bank
account was low, the time was scarce, or my sanity was far from view. I would also like to express
my love and gratitude for my wife, Lisa, who has selessly sacriced in supporting me while I pur-
sued this degree. This degree is as much yours as it is mine. I also wish to thank my parents, Mac
and Jeanne Cotting, who have always encouraged me to dream, and then supported me in chasing
those dreams over many horizons.
I would like to express my gratitude to Dr. Wayne Bull Durham for giving me the opportunity
to come to Virginia Tech. Teaching and developing the ight test course with you was the most fun
I had while I was on campus, and I am grateful to have a chance to collaborate with you impacting
the lives of students. I also appreciate your willingness to advise and assist me during your re-
tirement. I am also very grateful to Dr. Craig Woolsey for taking me on as his student, and being
willing to let me run in a direction that pushed both our limits. Your patience, candor, support, and
enthusiasm has been a great help as I have worked towards this degree. Many thanks are in order to
Dr. Bill Mason for giving me a place of T/A refuge in his design class. It has been a privilege to be
able to tap into your wealth of knowledge of aircraft. I would like to thank Dr. Leigh McCue-Weil
for encouraging me to pursue my education interests academically, and offering your support in
fullling me degree. Thank you also to Dr. Ken Bordignon, who started this process by introducing
v
Acknowledgments
me to his old advisor, and then being supportive with long phone calls and exible with time zone
challenges.
I am very appreciative to John Hodgkinson for his willingness to review my work, his candid
remarks, and giving me a voice in the ying qualities community. I would also like to thank
NAVAIR PMA-262 and PMA-263 for their continued support of this and many other specica-
tion and standard development and update efforts within NAVAIR. In particular, Dr. Steve Cook,
with the NAVAIR Airworthiness Team, and Jessica Holmberg in the Flight Dynamics Branch of
Naval Air Systems Command, Patuxent River. Special thanks is due to Jessica Holmberg from
NavAir for sponsoring my research on UAV ying qualities, and providing valuable direction, and
to her branch head Steve Hynes for offering his input to the research. I would also like to thank
the AOE department for its support in my piloted ying qualities research, and in providing my
funding while conducting the ight test course. Thanks are also in order to Dr. Naira Hovakimyan
for allowing me to use her L
1
controller in my research, and to Dr. Chengyu Cao for his help in my
implementation of the controller. I would also like to say a special thank you to Wanda Foushee and
the ladies in the AOE ofce that have helped me navigate the required trail of paperwork in order to
complete my degree.
Thanks are also due to Lt. Col. Bob Kraus for sharing an ofce, helping with the ight simulator,
and providing his ight test pilots experience to an engineers questions. I would also like to thank
all the students that have passed through the Nonlinear Systems Lab during my time at Virginia
Tech. Especially thanks to Mark Monda for his help with the aircraft camera systems, Laszlo Techy
for his help with the airdata boom, Justin Murtha for designing and building the SPAARO, Artur
Wolek for his help in reducing ight test data, and Craig Sossi and Tyler Aarons for their help in
keeping the SPAARO ying. Thanks also to Nina Mahmoudian and Amanda Dippold for their help
and advice during my visits to the NSL.
vi
Acknowledgments
I would like to offer a special thank you to Dr. David G. I. Kingston, and his wife Beverly for
providing valuable mentorship to both my wife and me during our stay in Blacksburg. Finally, I
would like to thank all the Navigators at Virginia Tech for allowing me a chance to share in your
journey. It has been an honor to share this time with you.
I dont believe in mathematics.
- Albert Einstein
vii
Table of Contents
Acknowledgments v
List of Figures xiii
List of Tables xviii
List of Symbols xx
1 Introduction 1
2 Literature Review 5
2.1 History of Piloted Flying Qualities . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Unmanned Air Vehicle Flying Qualities . . . . . . . . . . . . . . . . . . . . . . . 10
2.2.1 The Case for Unmanned Air Vehicle Flying Qualities . . . . . . . . . . . . 10
2.2.2 Flying Qualities for UAVs . . . . . . . . . . . . . . . . . . . . . . . . . . 12
2.2.3 Current Work in UAV Flying Qualities . . . . . . . . . . . . . . . . . . . . 16
2.3 Nonlinear Control Flying Qualities . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.4 Teaching Flight Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
3 Aircraft Equations of Motion 22
viii
Table of Contents
4 Review of Linear Systems 37
4.1 Transfer Functions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
4.2 Denition of Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
5 Classication 44
5.1 Creating a Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 45
5.1.1 Manned Aircraft Requirements Framework . . . . . . . . . . . . . . . . . 45
5.1.2 Aerodynamic Effects on Aircraft Dynamics . . . . . . . . . . . . . . . . . 47
5.2 Study of Available Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2.1 Sources of Data Considered . . . . . . . . . . . . . . . . . . . . . . . . . 49
5.2.2 Results of Available Aircraft Study . . . . . . . . . . . . . . . . . . . . . 49
5.3 Payload/Sensor Study . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6 Review of Manned Flying Qualities 61
6.1 AFFDL-TR-T25, RPV Flying Qualities Design Criteria . . . . . . . . . . . . . . . 62
6.2 MIL-F-8785C . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6.3 MIL-HDBK-1797 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.3.1 LOES . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.3.2 Neal-Smith . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3.3 Bandwidth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
6.3.4 Time Domain Methods . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.4 SAE94900 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.5 NASA TN D-5153 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.6 ADS-33E-PRF . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
6.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
7 UAV Flying Qualities 87
7.1 UAV Criteria Guidance . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
ix
Table of Contents
7.1.1 Classication of Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
7.1.2 Categorization of Flight Phases and Aircraft Maneuvers . . . . . . . . . . 88
7.1.3 Evaluation of UAV Task Performance . . . . . . . . . . . . . . . . . . . . 94
7.1.4 Applicable Manned Flying Qualities Criteria . . . . . . . . . . . . . . . . 94
7.2 Forming a New Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.2.1 Frameworks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
7.2.2 Classical Bandwidth Compared to Flying Qualities Bandwidth . . . . . . 99
7.2.3 Accommodating MIMO Analysis . . . . . . . . . . . . . . . . . . . . . . 102
7.3 Proposed Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
7.3.1 A Simplied Pitch Criterion . . . . . . . . . . . . . . . . . . . . . . . . . 103
7.3.1.1 Modeling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
7.3.1.2 Algebraic Reduction . . . . . . . . . . . . . . . . . . . . . . . . 104
7.3.1.3 SISO Analysis Case . . . . . . . . . . . . . . . . . . . . . . . . 106
7.3.1.4 MIMO Analysis Case . . . . . . . . . . . . . . . . . . . . . . . 110
7.3.2 A Criterion with Winds . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
7.3.2.1 Wind Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7.3.2.2 Energy Methods . . . . . . . . . . . . . . . . . . . . . . . . . . 115
7.3.2.3 Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
7.4 Simulation Trials . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.4.1 Simulation Testing . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
7.4.2 Aircraft Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
7.4.3 Sensor Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120
7.4.4 Testing the Simplied Pitch Criterion . . . . . . . . . . . . . . . . . . . . 120
7.4.5 Testing the Wind Criterion . . . . . . . . . . . . . . . . . . . . . . . . . . 123
7.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131
8 Nonlinear Flying Qualities 134
8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
x
Table of Contents
8.2 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135
8.2.1 L
1
Adaptive Control Architectures and Their Verication and Validation . . 135
8.2.2 Aircraft Model and SAS . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
8.2.3 Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 138
8.2.3.1 Simulator . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.2.3.2 IADS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 140
8.3 L
1
Controller Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142
8.4 Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
8.4.1 Predicted HQR of Plant Models . . . . . . . . . . . . . . . . . . . . . . . 146
8.4.2 Simulation Traces (Linear and Nonlinear) . . . . . . . . . . . . . . . . . . 147
8.5 Piloted Simulation Trial . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
8.5.1 Piloted Task Denition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 147
8.5.2 Simulation Trial Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
8.5.2.1 Task 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
8.5.2.2 Task 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 152
8.6 Open Issues in Prediction of Nonlinear Control Flying Qualities . . . . . . . . . . 155
9 Flight Mechanics and Flying Qualities in Engineering Education 157
9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 158
9.2 Combining Pedagogues . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161
9.3 Technology Utilization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
9.3.1 Student Immersion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
9.3.2 Facility . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166
9.4 Classroom Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 167
9.4.1 Course Goals and Objectives . . . . . . . . . . . . . . . . . . . . . . . . . 167
9.4.2 Course Organization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170
9.4.3 Flight Testing Positions . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
9.4.3.1 Test Pilot . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
xi
Table of Contents
9.4.3.2 Flight Test Engineer (FTE) . . . . . . . . . . . . . . . . . . . . 174
9.4.3.3 Test Conductor . . . . . . . . . . . . . . . . . . . . . . . . . . . 174
9.4.3.4 Discipline Engineer . . . . . . . . . . . . . . . . . . . . . . . . 174
9.4.4 Teaching and Student Participation . . . . . . . . . . . . . . . . . . . . . . 174
9.4.5 Using the Proper Vocabulary . . . . . . . . . . . . . . . . . . . . . . . . . 176
9.4.6 Using Reference Books Instead of Textbooks . . . . . . . . . . . . . . . . 176
9.4.7 Team Building . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177
9.4.8 Evaluation of Students; More Than Just Grades . . . . . . . . . . . . . . . 178
9.5 Evaluation of Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179
9.6 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189
10 Conclusions and Recommendations 190
Bibliography 194
A Appendix: Self Assessment Surveys for AOE4984 Flight Test Techniques 211
A.1 Student survey . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 211
A.2 Post graduation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 214
A.3 Employer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219
xii
List of Figures
2.1 Timeline for ying/handling qualities development. . . . . . . . . . . . . . . . . . 8
3.1 Body axis xed coordinate system. . . . . . . . . . . . . . . . . . . . . . . . . . . 24
3.2 Aircraft symmetry in the X-Z plane. . . . . . . . . . . . . . . . . . . . . . . . . . 26
3.3 Gravity vector in relation to aircraft body coordinates. . . . . . . . . . . . . . . . . 29
3.4 Angle of attack and angle of sideslip in relation to aircraft body coordinates. . . . . 31
3.5 Typical sign conventions for aircraft control surface deections. . . . . . . . . . . 33
4.1 Basic block diagram example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
4.2 Plot of the complex plane. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
4.3 Example Bode diagram where U(s) =

2
n
s
2
+2ns+
2
n
. . . . . . . . . . . . . . . . 40
4.4 Example feedback block diagram reduction to single transfer function. . . . . . . . 41
4.5 Example system diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
4.6 Example system diagram. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
5.1 Operational aircraft comparison in user guide for MIL-F-8785B. . . . . . . . . . . 47
5.2 Aircraft operational comparison by Reynolds and Mach numbers. . . . . . . . . . 51
5.3 Aircraft operational comparison by Reynolds and Mach numbers with manned air-
craft classes overlaid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
xiii
List of Figures
5.4 Aircraft operational comparison by Reynolds number and weight. . . . . . . . . . 52
5.5 Aircraft operational comparison by Reynolds number and weight with manned air-
craft classes overlaid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
5.6 Aircraft operational comparison by Reynolds number and weight, with UA 2007
Roadmap Levels overlaid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.7 Aircraft operational comparison by Reynolds number and weight, with proposed
UAV classes overlaid. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
5.8 Aircraft operational comparison by Reynolds number and weight, with proposed
UAV classes and trend lines overlaid. . . . . . . . . . . . . . . . . . . . . . . . . . 54
5.9 JUAS COE classication boundaries as described in the 2009 Unmanned Systems
Integrated Roadmap. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
6.1 Natural frequency specication for Category B ight phases, MIL-F-8785C. . . . . 66
6.2 Example transfer functions to illustrate LOES. . . . . . . . . . . . . . . . . . . . . 69
6.3 Example transfer function time histories to illustrate LOES. . . . . . . . . . . . . . 69
6.4 LOES boundaries for matching high order to low order transfer functions. . . . . . 70
6.5 Pilot model used with Neal-Smith criterion. . . . . . . . . . . . . . . . . . . . . . 73
6.6 Neal-Smith criterion for use in piloted aircraft with sample data. . . . . . . . . . . 73
6.7 Bandwidth longitudinal control block diagram. . . . . . . . . . . . . . . . . . . . 76
6.8 Root locus longitudinal control diagram with bandwidth frequency. . . . . . . . . . 76
6.9 Denition of bandwidth frequency,
BW
. . . . . . . . . . . . . . . . . . . . . . . 77
6.10 Bandwidth requirements from MIL Handbook. . . . . . . . . . . . . . . . . . . . 77
6.11 Revised bandwidth requirements to aid in PIO prediction. . . . . . . . . . . . . . . 78
6.12 Denition of dropback by Gibson. . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.13 Relation of ight path angle and pitch attitude as shown by Gibson. . . . . . . . . 81
6.14 Denition of acceleration weighting function w(f). . . . . . . . . . . . . . . . . . 82
6.15 Cooper Harper rating scale. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84
xiv
List of Figures
7.1 Proposed categories. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
7.2 Flying qualities cycle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
7.3 Mission task elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
7.4 A modied form of the Cooper Harper scale for use in sensor/UAV evaluation. . . 95
7.5 Denition of bandwidth frequency,
BW
compared to the classical bandwidth de-
nition. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100
7.6 Model used to approximate sensor and aircraft system. . . . . . . . . . . . . . . . 105
7.7 Transfer function denitions used to dene aircraft and sensor system . . . . . . . 105
7.8 Reduced block diagram of sensor and aircraft system. . . . . . . . . . . . . . . . . 105
7.9 Angle descriptions for sensor and aircraft system. . . . . . . . . . . . . . . . . . . 106
7.10 Example of analysis of Equation (7.3). . . . . . . . . . . . . . . . . . . . . . . . . 108
7.11 Time history of SPAARO short period model and sensor using TARGET SISO model.108
7.12 Bode plot of individual systems of the SPAARO short period and sensor model. . . 109
7.13 Time history of Cessna 172 short period model and sensor using TARGET SISO
model. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
7.14 Frequency analysis of SISO system for aircraft and sensor dynamics. . . . . . . . . 111
model with 1/6 system gain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111
model with 1/3 system gain. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
7.19 Interaction between sensor and aircraft . . . . . . . . . . . . . . . . . . . . . . . . 114
7.20 The inuence of wind turbulence on aircraft motion and sensor errors. . . . . . . . 115
7.21 An example of boundaries placed on for a pitch doublet MTE. . . . . . . . . . . 118
7.22 Bode comparison of sensor models with SPAARO aircraft dynamics. . . . . . . . . 121
xv
List of Figures
7.23 Linear simulation comparison of sensor models with SPAARO aircraft dynamics,
input at 1 rad/sec. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.24 Linear simulation comparison of sensor models with SPAARO aircraft dynamics,
input at rad/sec. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122
7.25 CASTLE 6-DOF simulation of SPAARO with sensor model gain K = 1. . . . . . 124
7.26 CASTLE 6-DOF simulation of Cessna 172 with sensor model gain K = 1. . . . . 124
7.27 CASTLE 6-DOF simulation of F-16 with sensor model gain K = 1. . . . . . . . . 125
7.28 CASTLE 6-DOF simulation of all aircraft sensor pointing error. . . . . . . . . . . 125
7.29 Bode comparison of all aircraft with sensor model gain K = 1. . . . . . . . . . . . 126
7.30 Linear simulation comparison of all aircraft with sensor model gain K = 1, input
excitation of rad/sec. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
7.31 calculation using CASTLE 6-DOF simulation of SPAARO with no turbulence. . 128
7.32 calculation using CASTLE 6-DOF simulation of SPAARO with moderate turbu-
lence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 128
7.33 calculation using CASTLE 6-DOF simulation of Cessna 172 with no turbulence. 129
7.34 calculation using CASTLE 6-DOF simulation of Cessna 172 with moderate tur-
bulence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 129
7.35 calculation using CASTLE 6-DOF simulation of F-16 with no turbulence. . . . . 130
7.36 calculation using CASTLE 6-DOF simulation of F-16 with moderate turbulence. 130
8.1 L
1
adaptive control architecture. . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
8.2 Augmented F-16 Plant. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
8.3 Diagram of Virginia Tech Flight Simulation System. . . . . . . . . . . . . . . . . 141
8.4 Sample IADS
R
data screens used for the study. . . . . . . . . . . . . . . . . . . . 143
8.5 Block diagram of L
1
controller. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144
8.6 MIL-F-8785C Short Period Criteria. . . . . . . . . . . . . . . . . . . . . . . . . . 148
8.7 Doublet command used to test L
1
controller response. . . . . . . . . . . . . . . . . 149
8.8 Doublet short period response with L
1
controller. . . . . . . . . . . . . . . . . . . 150
xvi
List of Figures
8.9 Pilots view for pitch capture task. . . . . . . . . . . . . . . . . . . . . . . . . . . 151
8.10 Piloted Cooper Harper Ratings Results. . . . . . . . . . . . . . . . . . . . . . . . 153
9.1 First page of Laboratory 1 ight test cards. . . . . . . . . . . . . . . . . . . . . . . 171
9.2 Illustration of ight test tasks/stations. . . . . . . . . . . . . . . . . . . . . . . . . 173
xvii
List of Tables
3.1 Variable denitions for transfer functions found in Equation (3.69) - Equation (3.73). 36
5.1 Manned aircraft classications as dened in MIL-F-8785C. . . . . . . . . . . . . 46
5.2 Manned aircraft Flight Phase Categories as dened in MIL-F-8785C. . . . . . . . 46
5.3 Denition of classication separation lines. . . . . . . . . . . . . . . . . . . . . . 55
5.4 Aircraft (manned and unmanned) used in comparison . . . . . . . . . . . . . . . . 57
5.5 Sample payloads found in research for this report . . . . . . . . . . . . . . . . . . 59
5.6 Payloads from Table 5.5 reduced to general classes. . . . . . . . . . . . . . . . . . 60
6.1 Gain and phase margin requirements as dened in AS94900. . . . . . . . . . . . . 81
6.2 Probability of exceedance of ride discomfort indices (vertical and lateral) in AS94900. 82
7.1 Flight Phases as dened in MIL-F-8785C. . . . . . . . . . . . . . . . . . . . . . . 90
7.2 Example maneuvers for proposed categories. . . . . . . . . . . . . . . . . . . . . 91
7.3 Aircraft operational points used for testing in simulation. . . . . . . . . . . . . . . 119
8.1 Reference of vehicle states. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
8.2 Test Flight Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 139
8.3 Reference of L
1
parameters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 146
xviii
List of Tables
8.4 Cooper Harper Ratings from Flying Qualities Test of L
1
controller augmenting
Short Period. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153
xix
List of Symbols
angle of attack
c mean aerodynamic chord
q dynamic pressure
aircraft sideslip
energy based ying qualities metric
pitch acceleration
change in horizontal control surface deection
e
change in aircraft elevator position
n
ss
change in steady state n
z
e
elevator position
dH
dt
change in angular momentum with respect to time
dp
dt
change in linear momentum with respect to time
h aircraft sink rate

xx
List of Symbols
ight path angle
bw gain
Denition of gain frequency for ying qualities bandwidth
bw phase
Denition of phase frequency for ying qualities bandwidth
bw
Classical denition of bandwidth frequency
sp
short period natural frequency
aircraft roll attitude
u
(f) Von Karmen gust power spectral density of intensity specied in MIL-STD-1797
aircraft yaw attitude
air density
pilot delay
e
equivalent time delay
p1
pilot lead compensation
p2
pilot lag compensation
p
phase delay approximation for equivalent time delay
e
pitch attitude
Target
pitch angle from aircraft to the target
a/cc
aircraft pitch attitude command
ac
aircraft pitch attitude from inertial to body axis
cs
sensor pitch attitude command
c
pitch attitude command
xxi
List of Symbols
err
pitch angle pointing error used in metric
e
pitch attitude error
sensor
sensor pitch attitude from inertial to body axis
F =< F
X
, F
Y
, F
Z
> inertial space coordinate vector
V
i
=< |, 1, J > inertial frame velocity vector
sp
short period damping ratio
a
z
aircraft vertical acceleration
C
L sens
closed loop sensor transfer function
C
L SP
closed loop short period transfer function
C
L
lift curve slope,
C
L
C
L
coefcient of lift
C
m
longitudinal control power,

Cm
C
m
damping derivative,
Cm

c
2V
C
m
C
L
static margin
C
m
pitching moment coefcient
D
i
ride discomfort index
E aircraft energy
E
h
energy height
f frequency Hz
f
m
mode frequency in Hz(FCS engaged)
xxii
List of Symbols
f
t
truncation frequency
g acceleration due to gravity
g acceleration due to gravity
G
sens
open loop sensor transfer function
G
SP
open loop short period transfer function
h aircraft altitude
I
yy
pitch moment of inertia
K variable used for a gain
K
p
pilot model gain
K
p
pilot pitch gain
L aircraft rolling moment
L
dimensional form of C
L
l
t
tail moment arm
M aircraft pitching moment
m vehicle mass
m vehicle mass
M
e
change in pitching moment with respect to elevator deection.
M
w
change in pitching moment with respect to vertical acceleration.
M
w
change in pitching moment with respect to vertical speed.
N aircraft yawing moment
xxiii
List of Symbols
n
aircraft load factor per angle of attack

n
z
vertical acceleration or aircraft loading
P
s
specic excess power
range
err
virtual distance error used in metric
S aircraft wingspan
s s = + i , Laplace integration variable
T
2
pitch numerator time constant for short period approximation
T
cs
(f) transmissibility, at crew station,
g
ft/sec
u aircraft body axis forward velocity
V magnitude of aircraft velocity
v aircraft body axis size velocity
V
0max
maximum operational airspeed (MIL-STD-1797)
V
0min
minimum operational airspeed (MIL-STD-1797)
V
l
limit airspeed (MIL-A-8860)
V
t
aircraft true velocity
W aircraft weight
w aircraft body axis vertical velocity
w(f) acceleration weighting function per g
Z
e
change in height with respect to elevator deection
Z
w
change in height with respect to vertical speed.
xxiv
List of Symbols
6 DOF Six Degree of Freedom
V
ex
vehicle exhaust velocity
APC aircraft pilot coupling
CAP control anticipation parameter
FCS ight control system
GM gain margin, the minimum change in loop gain, at nominal phase, which results in an insta-
bility.
JUAS COE Joint Unmanned Aircraft Systems Center of Excellence
LOES low order equivalent system
Mode A characteristic aeroelastic response of the aircraft as described by an aeroelastic character-
istic root of the coupled aircraft/FCS dynamic equation of motion.
MTE mission task element
MUAD maximum unnoticeable added dynamics
NACA National Advisory Committee on Aeronautics
NAS national air space
PBL Problem Based Learning
PIO pilot induced oscillation
PM phase margin, the minimum change in phase, at nominal loop gain, which results in an
instability.
RPV remotely piloted aircraft
xxv
List of Symbols
STEMS standard test evaluation maneuver set
UAV uninhabited (unmanned) aerial vehicle
xxvi
Chapter 1
Introduction
A
TMOSPHERIC ight mechanics is an applied discipline that draws together many of the the-
oretical and fundamental sciences often associated with pure research.
1
Unlike classical me-
chanics, uid dynamics, or basic control system theory which all have foundations where funda-
mental research can be a rich topic, atmospheric ight mechanics seeks to nd work done in funda-
mental research and apply that work to air vehicles. The research in atmospheric ight mechanics
therefore involves nding work done in other more fundamental disciplines, and then applying it
to the special case of air vehicles. The research in atmospheric ight mechanics has often been
driven by problems found in the application of fundamental research to the air vehicles. Examples
of this are numerous, but can be traced to the formation of the discipline of aircraft stability and
control that was formed by Lancaster
2
and Bryan.
3, 4
They have been credited with the foundation
of aircraft stability and control which made possible the application of control techniques to aircraft,
and ultimately the discipline of aircraft ying qualities. Aircraft ying qualities can be dened as
the measuring of the aircrafts dynamic response to a command. Their specication is intended
to assure ying qualities that provide adequate mission performance and ight safety regardless of
design implementation or ight control system mechanization.
5
For this reason the study of ying qualities has often lagged behind fundamental research in other
1
1 Introduction
elds as well as aircraft design innovation. Flying qualities research exists to ensure the application
of new technology to an air vehicle will either enhance or not impede the operation of the aircraft
to accomplish a desired mission. The new innovations that increase an aircrafts operational ability
must still fall within either a pilots or an operators ability to control an aircraft. Therefore, the
role of ying qualities is to quantify an aircrafts performance so that its suitability to a task can be
judged. Believing that no further study of ying qualities was warranted in the application of current
technology to aircraft, the USAF declared ying qualities a sunset science in the 1980s, and ceased
signicant research into ying qualities of aircraft. The occurrence of the YF-22 PIO soon after this
declaration proved that there was still plenty to be learned about human / aircraft interface when new
technology is applied to aircraft. Continued work in PIO prediction and performance has continued,
but no emphasis on revision of ying qualities standards to keep pace with modern advances in air-
craft has been made. This work denes two current areas where ying qualities standards must be
changed to keep pace with the application of modern technology to aircraft. Specically the areas
are in dealing with nonlinear control theory, and a revolutionary change in the aircraft paradigm,
unmanned aerial vehicles (UAVs).
This dissertation rst provides a historical overview of ying qualities development. It then
illuminates the current need for a ying qualities criterion for UAVs. A review of currently used
manned ying qualities standards is given, with the motivation of nding modications to them
to be used for UAVs. A ying qualities standard for use in specifying longitudinal performance
in airplanes like UAVs is then presented. Flying qualities as applied to manned vehicles using
nonlinear control is also addressed. Many open problems in this area still exist, and this dissertation
gives an example of one controller that is applied and tested in a ight simulation. Finally current
issues in ight mechanics education are addressed. Modern teaching techniques are explored in
augmenting a curriculum with relevant ight mechanics problems to encourage students to study
ight mechanics and the eld of ying qualities.
The contributions of this work to the state of the art in ight mechanics and ying qualities touch
2
1 Introduction
on several areas in this eld. They are as follows:
A classication standard for UAVs is proposed. This standard would be used in a similar way
as the standard rst used in MIL-F-8785B.
6
This classication standard allows for weight and
the aircrafts operational ight envelope to be used in classifying UAVs for ying qualities
testing, whereas current military classications only rely on gross takeoff weight to classify
aircraft.
Two longitudinal ying qualities criteria are proposed for autonomous UAV operation. While
previous criteria have been proposed in literature for RPVs, only the modication of previous
piloted criteria boundaries have been proposed. These two proposed criteria in this work use
the heritage of piloted ying qualities criteria as a starting point, but then derive new criteria
specically for autonomous UAVs. The new criteria focus on the interaction of the aircraft
and payload sensor package to judge overall mission performance. Specically one criterion
focuses on the short period interaction with sensor performance, and the other is a metric to
account for both short period and phugoid interaction with sensor performance in the pres-
ence of turbulence.
A nonlinear adaptive controller was implemented in the Virginia Tech manned ight simu-
lator, and then evaluated. This was the rst time that the L
1
controller was own by a pilot
and evaluated for ying qualities. The study was done to investigate the ability of current
piloted evaluation techniques for use in evaluation of a nonlinear adaptive controller. The
study showed the the recovery of good longitudinal ying qualities by the L
1
controller when
a degraded aircraft model was being controlled.
A new approach to teaching ight mechanics was successfully implemented by Dr. Wayne
Durham and the author. By using a full motion ight simulator, ight testing techniques were
3
1 Introduction
taught to aerospace engineering seniors as a capstone course in ight mechanics. The course
was able to foster deep learning by reinforcing previously covered concepts, and introducing
new, relevant concepts to the students while employing nonconventional methods of student
engagement. The course received overwhelmingly successful reviews by the students both
after the course and after they had entered the workplace. Their employers further endorsed
the course as a means for accelerating the students integration into the workforce.
4
Chapter 2
Literature Review
I
N 1908 the United States Army issued a sole source procurement specication to Orville and
Wilbur Wright for the rst purchase of an airplane for military use. That specication stated
that It should be sufciently simple in its construction and operation to permit an intelligent man
[sic] to become procient in its use in a reasonable length of time.
7
While that statement still holds
true for modern day aircraft, the criteria dening the words sufciently simple, procient, and
reasonable length of time must be precisely dened. These denitions have evolved over time to
reect the complexities that have arisen from advancing aircraft designs. A new round of evolution-
ary changes has occurred in aircraft design. This evolutionary change has removed the pilot from
the aircraft creating a new paradigm known as unmanned air vehicles (UAVs), or unmanned aerial
systems (UAS). This literature review traces the development of ying qualities for piloted aircraft,
and then makes an argument for the need of ying qualities criteria for unmanned aircraft.
2.1 History of Piloted Flying Qualities
Aircraft ying qualities design specications are intended to assure ying qualities that provide
adequate mission performance and ight safety regardless of design implementation or ight control
5
system mechanization.
5
These specications are intended to address problems that were originally
identied before the rst airplane took ight.
8
Chanute compiled a list of ten problems that he
identied as needing to be overcome before an aircraft could be successfully operated. Three of
his listed ten problems were directly related to what we now classify as ying qualities: (1) the
maintenance of equilibrium (or trimmed ight), (2) the guidance in any desired direction (controlled
ight), (3) the alighting safely anywhere (landing). The Wright brothers heeded this advice, and
spent considerable time ying gliders to understand the nuances of aircraft controllability before
ying their rst powered aircraft.
9
While addressing the Western Society of Engineers in 1901,
Wilbur Wright stated:
Men already know how to construct wings or aeroplanes which when driven through
the air at sufcient speed, will not only sustain the weight of the wings themselves,
but also that of the engine, and the engineer as well. Men also know how to build
engines and screws of sufcient lightness and power to drive these planes at sustaining
speed. . . Inability to balance and steer still confronts students of the ying problem
. . . When this one feature has been worked out, the age of ying machines will have
arrived, for all other difculties are of minor importance.
10
The Wrights ability to confront and solve this problem led them to be the rst successful aircraft
manufacturer. The importance of controllability was not lost on their rst customers as well. When
the U. S. Army contracted the Wright brothers to manufacture the rst aircraft for military use,
requirement 10 in their specication addressed the need for operational simplicity so that an intel-
ligent man could be quickly trained.
7
Widely regarded as the rst rst ying qualities requirement
for aircraft, it is still relevant to todays aircraft.
11
The rst formal work in ying qualities analysis was performed by the National Advisory Com-
mittee on Aeronautics (NACA) . The initial work was done by Soul e to determine the instrumen-
tation and techniques required to measure aircraft ying qualities.
12
Gilruth
13
continued this work,
and wrote what was is acknowledged as the rst technical ying qualities specication for the
6
NACA in 1943. This specication was mirrored by similar documents by the U. S. Army and
Navy.
14, 15
These early documents reected a need to be specic in aircraft requirements, as the
number of aircraft being procured by the military increased due to the war effort. A desire to have
uniformly ying aircraft was predicated by the desire to have a uniform pilot training program. With
the large quantity and variety of aircraft being procured for the war effort, the armed forces recog-
nized that pilot training efforts could be streamlined if all aircraft were operated in a similar manner,
and all aircraft responded to pilot control within given guidelines. Shortly after the conclusion of
World War II, the rst multi-service ying qualities document was issued.
16
MIL-F-8785 was the
beginning of modern ying qualities documents in that for the rst time desired aircraft dynamic
responses were specied, and all military services agreed upon a unifying standard. The aircraft
dynamic responses were quantied with natural frequencies and damping ratio targets for the short
period, phugoid, dutch roll, and spiral modes. A detailed timeline of the evolving ying qualities
specications can be found in Figure 2.1.
In 1948 ying qualities research transitioned from its infancy into the modern era of analytical
work. During this time frame technology expanded in many elds directly affecting aircraft in-
cluding propulsion, materials, control systems, and analytical techniques. The speed and altitude
capabilities of aircraft saw a signicant increase and aircraft congurations began to move in very
different directions. During this time Cornell Aeronautical Laboratory began experimenting with
variable stability aircraft. Work was also done in the area of control anticipation parameter (CAP),
which species the aircrafts pitch acceleration to change in steady state load factor. These advances
coupled with the onset of the Korean War led the U. S. Air Force and Navy to collaborate on the
MIL-F-8785 in 1954
11, 17
.
As advances to aircraft technology allowed for varied uses, distinguishing characteristics of broad
categories of aircraft became apparent. These broad categories of aircraft dictated different require-
ments for each category, and led to revisions of MIL-F-8785
18
with a nal version of MIL-F-8785C
5
7
1
9
1
0
1
9
2
0
1
9
3
0
1
9
4
0
1
9
5
0
1
9
6
0
1
9
7
0
1
9
8
0
1
9
9
0
2
0
0
0
2
0
1
0
M
I
L
-
H
N
B
K
-
1
7
9
7
A
2
0
0
0
A
S
9
4
9
0
0
2
0
0
7
N
A
C
A
R
e
p
o
r
t
7
5
5
(
G
i
l
r
u
t
h
)
1
9
4
3
K
o
r
e
a
n

W
a
r
1
9
5
0

1
9
5
3
M
I
L
-
F
-
8
7
8
5
(
A
S
G
)
R
e
l
e
a
s
e
d
1
9
5
4
M
I
L
-
S
T
D
-
1
7
9
7
A
(
t
r
i
-
s
e
r
v
i
c
e
)
1
9
9
0
M
I
L
-
F
-
9
4
9
0
D
1
9
7
5
M
I
L
-
S
T
D
-
1
7
9
7
(
U
S
A
F
)
1
9
8
7
M
I
L
-
F
-
8
7
8
5
B
(
A
S
G
)
, w
i
t
h
U
s
e
r
's
G
u
i
d
e
1
9
6
8
W
r
i
g
h
t
B
r
o
t
h
e
r
s
F
i
r
s
t
F
l
i
g
h
t
1
9
0
3
e
A
D
S
-
3
3
, U
s
e
s
M
T
E
s
a
n
d
S
y
s
t
e
m
s
E
n
g
i
n
e
e
r
i
n
g
A
p
p
r
o
a
c
h
1
9
8
5
A
r
m
y
S
i
g
n
a
l
C
o
r
p
s
S
p
e
c
4
8
8
, F
i
r
s
t
F
Q
S
p
e
c
i
c
a
t
i
o
n
1
9
0
8
A
F
F
D
L
-
T
R
-
7
6
-
T
2
5
1
9
7
6
F
A
R
P
a
r
t
2
5
1
9
6
5
s
W
e
s
t
b
r
o
o
k
P
a
p
e
r
a
t
A
I
A
A
1
9
6
5
F
i
r
s
t
A
u
t
o
p
i
l
o
t
F
l
i
g
h
t
(
L
a
w
r
e
n
c
e
S
p
e
r
r
y
)
1
9
1
3
0
U
T
M
I
L
-
F
-
8
7
8
5
C
(
L
O
E
S
i
n
c
l
u
d
e
d
)
1
9
8
0
J
S
S
G
-
2
0
0
1
2
0
0
1
M
I
L
-
S
T
D
-
1
7
9
7
B
2
0
0
6
Figure 2.1: Timeline for ying/handling qualities development.
(Wright Flyer), http:en.wikipedia.orgwikiFile:First ight2.jpg, 2010, public domain.
(Army Signal Corps Specication No 486), http:www.ascho.wpafb.af.milbirthplacePG14-1.HTM, 2010, fair use.
(Sperry Autopilot Aircraft), http:www.historynet.comlawrence-sperry-autopilot-inventor-and-aviation-innovator.htm, 2010, public
domain.
(NACA), http:www.dfrc.nasa.govGallerygraphicsLogosHTMLNACA logo.html, 2010, fair use.
(FAA), http:www.faa.gov, 2010, fair use.
(U. S. Army Aviation), http:www.army.milinfoorganizationrucker, 2010, fair use.
(U. S. A. F.), http:upload.wikimedia.orgwikipediacommons002Usaf.png, 2010, public domain.
(U. S. Department of Defense), http:www.af.milartmediagallery.asp?galleryID=4979, 2010, fair use.
8
in 1980. This nal version of MIL-F-8785 was considered controversial because it did not contain
guidance for the emerging class of digitally controlled aircraft.
11
The MIL-F-8785C specication
was notable, however, for rst containing the low order equivalent system (LOES) approach to
evaluating complex aircraft control systems. The LOES approach gives guidelines for pilot input
to aircraft response in the form of transfer functions that are simplied models of the total aircraft
response. By matching an aircrafts response to a known, good LOES response an aircraft is pre-
dicted to have favorable ying qualities. Good LOES response aids in predicting favorable margins
for dynamic stability.
19
Aircraft dynamic stability prevents unintentional excursions into dangerous
ight regimes. The stable aircraft, which is predicted by meeting ying qualities specications is
also resistant to external forces such as wind that may require added pilot compensation. These
characteristics not only improve ight safety, but allow the pilot to perform maneuvering tasks with
smoothness, precision, and minimum effort.
20
By following piloted ying qualities criteria an air-
craft designer can aid the pilot in maximizing aircraft mission effectiveness while minimizing safety
concerns to the pilot during ight.
While advances in technical specications progressed, the nal test of ying qualities remained
the subjective judgement of the aircraft operator. Unlike performance requirements for aircraft, y-
ing qualities specications are a design guideline used as a means to achieve an end, and are not
an end unto themselves. As stated by Vincenti:Thus, for the designer, the quantities set down in
performance specications are themselves objective ends; the quantities prescribed in specications
of ying qualities are objective means to an associated subjective end.
21
In the 1980s government
procurement underwent a change in philosophy from specifying hard and fast requirements to
providing guidance to designers, and allowing aircraft designers to make nal decisions on criteria
that were applicable to their specic aircraft.
11
In an effort to reect new trends in military acqui-
sition, a new standard MIL-STD-1797
22
was released. This new standard incorporated standards
in MIL-F-8785C, but also included ight test methodologies, and a pilot opinion rating scale that
had been widely used but not ofcially adopted. Known as the Cooper-Harper Rating Scale,
23, 24
it
9
2.2 Unmanned Air Vehicle Flying Qualities
allowed for a less stringent pass/fail criterion in judging aircraft. MIL-STD-1797 also reected
a change from specic requirement to design guidance allowing requirements to be customized for
each aircraft. MIL-STD-1797 included guidance from research conducted in the 1970s and 1980s
regarding digital control for aircraft, although at the time no one criterion was found that could be
applicable to all aircraft. LOES specications that were rst presented in MIL-8785C were consid-
ered controversial and suggested revisions were made to begin to account for time delay effects of
digital control systems.
25
Specically the Neal-Smith database
26
had brought to light these issues
with ight testing of various aircraft. This database was later augmented by further research by
Smith into time delay and its effect on ying qualities.
27, 28
MIL-STD-1797 brought revisions to
the LOES specications as well as other suggested criteria to deal with the previously mentioned
time delay issues. As well as including references to the Neal-Smith database and work by Smith
and Geddes, it also suggested criteria based on pitch bandwidth,
29
and a mixed frequency and time
based criterion.
30
MIL-STD-1797 was reissued in 2000 as MIL-HNBK-1797 further underscoring
the change in criteria from a requirements document to a design guidance document. The specica-
tions were again amended, primarily to reect further acquisition strategy changes in 2002.
31
These
changes reected a desire by procurement agencies to keep up with the ever changing and more
complex aircraft that were being designed for manned operation. Their intent, however, was still
the same: to assure ying qualities that provide adequate mission performance and ight safety
regardless of design implementation or ight control system mechanization.
22
2.2.1 The Case for Unmanned Air Vehicle Flying Qualities
The use of unmanned aircraft has risen sharply since the end of the Cold War.
32
The future use of
both military and civilian unmanned aircraft is projected to increase sharply. Technological limits
that previously required aircraft to be controlled by pilots no longer exist, and barriers to unmanned
10
aircraft are ones of logistics and ensuring mission capability.
33, 34
The Department of Defense (DoD)
expects to have over 400 UAVs operating in the eld, investing over $10 billion in their use. The
DoD expects to be operating full scale aircraft as UAVs by 2012. Advantages of UAVs to manned
systems include the dull, dirty, and dangerous jobs where exposure of human life is either consid-
ered too costly or unsafe. The DoD has identied the goal of Decreasing the annual mishap rate of
larger model UAVs to less than 20 per 100,000 ight hours by FY09 and less than 15 per 100,000
ight hours by FY15.
33
Procurement costs are considered a major inuence in choosing UAVs over traditional manned
aircraft. The aviation industry recognizes an informal rule that the production cost of an aircraft is
directly proportional to its empty weight. That gure is currently $1500 per pound (FY94 dollars).
Ten to fteen percent of a manned aircrafts empty weight is allotted to pilot systems (cockpit, ejec-
tion seat, etc.). Unmanned aircraft do not need these systems and pose a savings based on reduced
weight alone. Although these systems may still be present in the form of ground control stations,
they are usable for more than one UAV and are not considered lost in the event of an aircraft in-
cident. Additional savings can be found in that most manned aircraft systems spend a majority
of their combat life in training scenarios keeping pilots procient. Modern strike aircraft spend
95% of their 8000 hour in-ight life in conducting training scenarios, and only 400 hours in-ight
supporting combat operations. Current UAV designs for strike aircraft are focused on a 4000 hour
in-ight life where little to no in-ight training time is required. UAV ground control stations can
be used with modern ight simulators to provide high delity training without ying the UAV itself.
Seventy-ve percent of non-combat aircraft losses are attributed to human error. While pilots aug-
ment their training with simulation, no substitute is currently available for a pilot ying an aircraft
in training scenarios. With UAVs, however, since the pilot is not present in the aircraft, simula-
tion becomes a much more viable alternative for training, and reduces the exposure of airframes to
loss during training scenarios.
33
Pervasive use of UAVs is compelling, and the need to ensure that
they will perform their designed missions is paramount with large investments being made. Current
11
ying qualities specications focus only on ensuring good ying qualities in relation to a human
pilot. An example of this is the CAP criterion, whose boundaries are drawn based on the human
vestibular systems ability to sense pitch acceleration. This is just one of many examples where the
applicability of current ying qualities specications is suspect, since a pilot is no longer ying the
aircraft. Proper ying qualities requirements tailored for UAVs are needed to ensure that UAVs will
be able to adequately perform their missions.
2.2.2 Flying Qualities for UAVs
Early attempts at ying qualities specications provided guidance for basic aircraft of the day.
13
In
1948 ying qualities research transitioned from its infancy into the modern era of analytical work.
During this time frame technology expanded in many elds directly affecting aircraft including
propulsion, materials, control systems, and analytical techniques. The speed and altitude capabili-
ties of aircraft saw a signicant increase and aircraft congurations began to move in very different
directions. These advances coupled with the onset of the Korean War led the U. S. Air Force and
Navy to collaborate on the MIL-F-8785 in 1954
11, 17
.
In 1965 Westbrook
17
made a strong case for revisions to ying qualities requirements for aircraft
that eventually led to MIL-F-8785C and MIL-STD-1797. Referring to the modern piloted aircraft
of the day he stated:
It becomes very clear that every effort possible in dening the basic criteria, perform-
ing complete analyses, and checking the characteristics of the vehicle in the design
stage must be made.
Very few new aircraft are being procured... When a new vehicle is procured it is ex-
pected to be a signicant advance over previous aircraft.
12
The contractor is under extreme pressure to meet his schedule and to meet the deni-
tive guarantee of the contract.
It is an obvious trend that aircraft have become much more complex. There has been a
proliferation of congurations and a spreading of the regimes of ight that the aircraft
traverse. There has been a great advance in ight control technology... These trends
have several effects on handling qualities criteria. One of the most important is that
criteria based only on the dynamics of the vehicle become less and less meaningful
from a total system point of view. Another is the widening spread of characteristics,
missions, conditions and regimes of ight, automatic and emergency modes of ight,
etc. that must be considered and provided for a given vehicle by the criteria.
Safety, reliability, and maintainability, have always been of importance to the Air
Force. With the immense cost and importance of a single vehicle and the possible loss
of life and property of a crash, what would have been considered acceptable attrition
in the past can no longer be tolerated. With the great increase in complexity, mainte-
nance costs have zoomed and reliability has dropped. The trend toward a stronger and
stronger need for a close tie between these factors and handling qualities criteria will
continue.
Modern UAVs are experiencing growth and evolution similar to manned aircraft during the
1950s. These statements made by Westbrook regarding piloted aircraft are now directly applicable
to UAVs. With large investments in UAV technology by the DoD, and expectations of signicant
reduction in mishap rate, managed advances in UAV technology are expected by the DoD. Gov-
ernment agencies must have a way to judge the performance of new vehicles against a common
standard to ensure their effectiveness. With new operational scenarios emerging for UAV use in
Iraq and Afghanistan, the pressures on contractors to deliver new capability is increasing. Flying
13
qualities specications were created for manned aircraft to aid in their design and assessment. The
time has now come to revisit those specications to create similar standards for UAVs.
One central theme in all the previously mentioned ying qualities documents is that the aircrafts
performance and response is tailored to optimize the pilot-aircraft interaction. No guidance is given
in these documents for aircraft that are un-piloted. An effort to address unmanned aircraft ying
qualities was begun in the 1970s in response to the growing number of remotely piloted vehicles
(RPVs).
35
While RPVs differ from UAVs in that they are not autonomous, they are similar in that
the pilot is removed from the aircraft. This study was a tailoring to remotely piloted vehicles of the
eventually accepted MIL-STD-1797 for manned aircraft. While it did not address fully autonomous
aircraft, it was a rst step toward addressing aircraft performance when a pilot is no longer present
in the aircraft. Further work to highlight problems associated with ying qualities and remotely
piloted vehicles was presented by Breneman,
36
although his work served more to highlight cur-
rent problems than offer solutions. Small unmanned aircraft present challenges not present in full
scale aircraft ight test.
37, 38
For example, current ight testing techniques rely on pilot technique
to overcome handling deciencies in early ight control implementation. Also, small aircraft are
not subject to the same aerodynamic effects as large aircraft, since effects that are largely ignored
for full scale aircraft become signicant for small aircraft. Fundamentally these smaller aircraft
respond to disturbances differently than their larger counterparts due to the low Reynolds Num-
ber effects that amplify turbulent ow effects and that would be considered insignicant in high
Reynolds Number airow.
39
Williams later highlights the need to specify ying qualities criteria.
40
Current ight testing techniques do not attempt to quantify low Reynolds Number effects. A later
attempt to dene ying qualities focused on using a dynamic scaling approach to current manned
ying qualities criteria, specically those found in MIL-F-8785C.
41
The criteria neglects, however
time delay effects and also only focuses on one type of UAV.
Of all current attempts to dene ying qualities for UAVs several issues remain open. One pri-
14
mary issue is the question of what are ying qualities of an unmanned system?
40
Second, current at-
tempts have focused purely on mirroring manned specications to unmanned aircraft, assuming that
a UAV must y like a manned aircraft. Returning to the original purpose of ying qualities speci-
cations, they are intended to assure ying qualities that provide adequate mission performance and
ight safety regardless of design implementation or ight control system mechanization.
5
While
the word pilot is not directly used, the safety of the pilot is implicit in the denition of ight safety.
The safety of the pilot is no longer a required constraint, and the computer controlling the aircraft is
not necessarily limited by the same inuences that a pilot may be limited. Since the fully augmented
UAV no longer has to directly interface with the pilot to achieve its mission objective, the desired
mission performance should no longer be linked to criteria that are designed to maximize the pilots
ability to perform a task. UAV ying qualities should remain focused on achieving the mission per-
formance goals, but without the constraint of human intervention.
36
For UAVs the accommodation
should be changed from the manned perspective of enabling a pilot to perform a task to enabling a
sensor mounted on the aircraft to perform its desired mission. This study aims to leverage current
manned criteria in creating new UAV criteria by changing the focus of current manned criteria away
from pilot accommodation to sensor/payload accommodation.
Criteria referenced in MIL-STD-1797 may still be applicable for UAVs, with a focus changed
from the pilot to the sensor. The LOES approach as well as the pitch bandwidth and Gibson cri-
terion all rely on a simplied dynamic system to model aircraft response to a disturbance.
27, 29, 42
Keeping the new requirements in the familiar context of existing ying qualities requirements will
also allow easier adoption of the new requirements by those in the aviation community.
Further evidence of the need to dene ying qualities for unmanned systems can be found out-
side of military applications for UAVs. The National Research Council conducted an assessment
of NASAs aeronautics technology programs in 2004. In this review the council found that NASA
should increase efforts in research in the area of aircraft ight controls and handling (ying) quali-
15
ties. Specically the report states:
NASAs past work in ight controls and handling qualities provided the reference
standard for todays system designs. However, as we move toward unmanned systems,
the existing standards, which are for manned systems, may be too restrictive. Further
evolution of the base work done by NASA to include unmanned systems is essential to
creating a competitive advantage for US products as this market becomes more price-
driven.
43
NASAs aeronautics research interest has traditionally complemented the US militarys interests.
While NASA has done research in areas that have direct military application (such as its long history
in X-plane research,
44
as well as such programs as the High Alpha Research Vehicle (HARV)
45
), it
has also done research for civilian benet as well.
34
2.2.3 Current Work in UAV Flying Qualities
Flying qualities and airworthiness criteria in use today measure an aircrafts ying qualities
based on the prediction of an average pilots ability to perform a given task with an aircraft. Since
there is no pilot on board to y the UAV, a central theme in ying qualities determination for airwor-
thiness is no longer present. The pilots role has been replaced by a ight control system performing
a prescribed maneuver. While the mission of the UAV may be similar to that of a manned aircraft,
the critical link to a successful mission is no longer the pilots ability to y the aircraft. The critical
link is now the payloads ability to perform its task while integrated with the UAV. UAVs therefore
require a new set of criteria to describe their performance in relation to their payload and designed
mission.
To date work done to tailor ying qualities for UAVs has been focused on remotely piloted ve-
hicles (RPVs). In 1976 Rockwell International compiled applicable sections of MIL-8785B, MIL-
F-83300, MIL-F-9490D, and MIL-C-18244A into a new design guidance document for use with
RPVs.
35
Specic tailoring of requirements to UAVs was not done in this study; the work was done
16
2.3 Nonlinear Control Flying Qualities
specically for RPVs without mention of autonomously operating UAVs. The Naval Air Warfare
Center published results from applying and then testing modied manned ying qualities criteria
to small scale RPVs.
36
Work has also been done to show that short period natural frequency re-
quirements should be different for small UAVs compared to traditional aircraft.
41, 46
Using dynamic
scaling, both reports suggested different natural frequency boundaries for small UAVs versus full
scale piloted aircraft.
Dynamic scaling
47, 48
denes a relationship between a full scale vehicle and a small scale model.
Scaling can be performed on a wide variety of parameters
49
affecting dynamic response including
natural frequency (
n
), time constant () and moments of inertia (I) based on a scale factor N 1.
n
model
=
N
n
fullscale
model
=
N
fullscale
I
model
=
1
N
5
I
fullscale
This method has shown success when comparing a sub-scale model to a full-scale vehicle using
remotely piloted control.
48
These RPV frequency techniques on dynamic scaling require a full scale
vehicle to determine an appropriate scaling factor. Representative large scale vehicles for different
UAV classes must be identied if these techniques are to be applicable to all RPVs.
Aircraft cannot be modeled throughout their entire ight envelope as linear differential equations
due to nonlinear affects. Linear control theory has been used, however, as the defacto-standard of
aircraft control systems. Linear models of aircraft are created to at discrete ight conditions through-
out an aircrafts ight envelope. These ight conditions are then patched together to create a piece-
17
wise continuous linear aircraft model used for analysis as well as control design. Linear methods
give insight to control synthesis that can be difcult to obtain with nonlinear control systems.
10, 50
As sophistication of aircraft increased, and the demands on basic linear control theory advanced,
the application of modern control theory began to become more preeminent in aircraft,
51, 52
but it
still can not account for uncertainties such as control surface failure or reconguration. To account
for aircraft control system failures, a piecewise control system approach has been applied to full
scale aircraft,
53
but can be time intensive, and can not guarantee full ight envelope coverage. With
the relatively recent application of nonlinear control theory to aircraft, new possibilities for control
robustness can be explored. Full control surface failure or outer mold line morphing / change can
be accounted for within the control system, so long as the physical system can generate the required
moments to overcome disturbances.
54
The exciting increase of capability has brought with it a
new set of problems associated with human interface with vehicles with nonlinear, adaptive control.
Early testing of adaptive control has occurred to date with piloted vehicles.
5558
While these adap-
tive methods can account for large uncertainties, their validation for use in piloted aircraft is still
untested. Signicant issues relating to the human interface to these control systems still remain.
59
Adaptive control algorithms have given hope to a new level of ight vehicle robustness previously
unachievable with traditional control methods. Recent Air Force programs RESTORE (X-36) and
JDAM have demonstrated the potential of adaptive controllers for compensation of modeling uncer-
tainties and component failures. However, the major lesson learned during those programs was that
the conventional model reference adaptive control scheme is very sensitive to time-delay, especially
if the controller attempts to adapt quickly.
60
This has consequently led to a new paradigm for design
of adaptive controllers, specically The Theory of Fast and Robust Adaptation.
6168
The Theory of
Fast and Robust Adaptation provides an ability for a priori prediction of the uniform performance
bounds for transient and steady-state for a systems input and output signals, and also analytical
quantication of the gain and time-delay margin, similar to linear systems gain and phase margin
concepts. Various architectures of this theory, known as L
1
adaptive controllers, have been ight
18
2.4 Teaching Flight Mechanics
tested and validated against the theoretical claims.
6977
While it may be proven that an aircraft can
be stabilized when in a degraded or unknown state, the ying qualities of the ight vehicle must be
favorable if a piloted system is to be controlled, especially during critical phases such as landing or
formation ight, e.g. air to air refueling.
Teaching students in an Aerospace Engineering curriculum presents challenges that are not usually
present in traditional engineering programs. Because of the size, cost, and liability associated with
operating full scale aircraft, most aerospace engineering students only encounter pictures of aircraft
in a classroom. While scientic knowledge and theoretical concepts can be clearly communicated
in a traditional classroom, practical experience with aircraft is generally infeasible. This is espe-
cially problematic in the teaching of atmospheric ight mechanics and ying qualities, since these
disciplines study the integration of the entire aircraft and how it responds to pilot or operator com-
mands. Modern use of remote control aircraft in the aerospace curriculum aids students by giving
them some practical experience, but it only gives students an experience by proxy, and not the im-
mersive experience that ight in an aircraft brings. This creates a barrier to learning for students
eager to fully understand aerospace engineering concepts with no way to practice the concepts they
have learned outside of traditional homework problems and tests. The result is that an aerospace
engineering curriculum attracts students with great mathematical prowess and often excludes stu-
dents with more practically based engineering skills. Education in engineering requires a balance
of scientic knowledge and practical experience to cultivate well rounded engineers.
The National Academy of Engineering (NAE) challenges educators in Educating the Engineer of
2020 to pursue student oriented education. They encourage :
1. better alignment of engineering curricula and the nature of academic experiences
with the challenges and opportunities graduates will face in the workplace and
19
2. better alignment of faculty skill sets with those needed to deliver the desired cur-
riculum in light of the different learning styles of students.
78
The NAE also highlights a growing separation between practicing engineers and engineering fac-
ulty. The NAE states
The great majority of engineering faculty, for example have no industry experience.
Industry representatives point to this disconnect as the reason that engineering students
are not adequately prepared, in their view, to enter todays workforce.
78
The National Research Council proposes in How People Learn that technology can be used to
support learning by bringing exciting curricula based on real-world problems into the classroom
and building local and global communities that include teachers, administrators, students, parents,
practicing scientists, . . .
79
By using a Learner-Centered approach,
80
and Problem Based Learning
(PBL)
81, 82
as a baseline pedagogy, the authors have been able to use available technology to create
a curriculum that directly answers the challenge offered by the NAE.
Aerospace engineering education needs to attract creative students that are more than just math
whizzes. Students are attracted to engineering because of other reasons besides liking math. Em-
ployers need students that understand airplanes conceptually as well as theoretically. Students that
just understand basic equations are not effective in advancing a complex system such as an aircraft
where the integration of disciplines is key to a successful design. Engineering education has swung
from being overly practical to in some cases overly theoretical where students are not engaged in
the hands on practice of engineering. The balance needs to be realigned. This can be done by us-
ing problem based learning and connecting students and teachers with industry experts in order to
assure classroom assignments and example problems are relevant and real.
Traditional design classes taught at the senior level serve as capstone courses to most curricu-
lums. These design classes offer a chance for students to see the different courses they have taken
20
integrated together, but they rarely allow students a chance at experiential knowledge. Often these
classes turn into another analysis course that is nothing more than a revisiting of the courses they
have taken before. The medical communities have found signicant reductions in student errors
after participating in simulations.
83, 84
Students using this method get a chance to practice their
skills before being exposed to real patients where serious harm can be done by making mistakes
that could be avoided with simple experience with their newly learned skills. Just as the engineering
educational community has been inuenced through medical education with Problem Based Learn-
ing
81, 82
learning through simulated immersion is another tool that engineering educators can take
and apply from the medical educational eld. Before entering the workforce students must have rst
hand exposure to more than just analytical skills. Students must apply their knowledge to learn by
doing. This type of simulated immersion extends beyond the goals of a typical design class where
the goal is to submit an analytical paper. Students need to have a means to work on something that
gives them an output for their input. The output should not be easily gamed or manipulated by the
students to produce an inaccurate response to the inputs. These immersive, simulated learning envi-
ronments lend motivation and credence to the theoretical material they learn in analytical classroom
settings.
85
21
Chapter 3
Aircraft Equations of Motion
T
HE analytical work done for this research has at its foundation the assumption that the aircraft
analyzed are rigid bodies. In the most general case a rigid body moving freely through space
can be dened by six coordinates, three in translation, and three in rotation.
86
Using this assumption
six degrees of freedom (6 DOF) nonlinear differential equations have traditionally been used to
model aircraft rigid body motion for analysis. Derivations of the exact equations used can take
many different forms of varying complexity depending on the assumptions used in their derivation.
They have been derived in different forms in various texts, and can be readily found in Perkins and
Hague,
87
McGruer et al.,
88
Stevens and Lewis,
89
and Etkin
1
as well as many others. For a historical
overview of the development of the aircraft equations of motion McGruer gives detail to not only the
equation development, but the sources where the original concepts were derived. These equations
have been developed in many forms for various uses, and they will be derived here for use in near
Earth ight. There will be four assumptions made in deriving the nonlinear equations of motion for
aircraft ight.
Assumption 3.1: The airframe is a rigid body.
Assumption 3.2: The Earth is at, and xed in space.
Assumption 3.3: The mass and its distribution in the vehicle are constant.
22
3 Aircraft Equations of Motion
Assumption 3.4: The X-Z plane in the vehicle is symmetric.
Using assumption 3.1 the degrees of freedomof the body can be limited to six degrees of freedom,
whose position in space can be described in translation and rotation about its center of mass. From
assumption 3.2 an inertial coordinate frame relative to the Earth can be established for relatively
short ight times and distances. Once an inertial reference frame is established, Newtons second
law can be applied. Linear momentum is dened as p = mV
i
, and the time rate of change of linear
momentum is equal to the sum of all externally applied forces.
dp
dt
=
F (3.1)
The rate of chance of angular momentum is equal to the sum of all externally applied torques.
dH
dt
=
M (3.2)
The time rate of change of momentum can be described as
dp
dt
= m
dV
i
dt
V
ex
dm
dt
= F (3.3)
or rearranged as
m
dV
i
dt
= F +V
ex
dm
dt
= F +T. (3.4)
T is dened as V
ex
dm
dt
and is the thrust component due to expulsion of vehicle mass. Equa-
tion (3.1) can be used for vehicles that do not use vehicle mass for momentum exchange, and
Equation (3.4) can be used for vehicles where vehicle mass is used for momentum exchange, such
as rockets. For conventional aircraft, the thrust term T can be contained within the external forces
F. Inertial space can be represented as a right hand Cartesian coordinate frame < A, }, : > and
the velocity vector V
i
=< |, 1, J >, and a total force F =< F
X
, F
Y
, F
Z
>. Equation (3.4) can
23
be rewritten as
m
d|
dt
= F
X
(3.5)
m
d1
dt
= F
Y
(3.6)
m
dJ
dt
= F
Z
(3.7)
The rotary equations can be further dened by the denition of angular momentum
H = [I] (3.8)
where [I] is the inertia matrix and is the angular velocity. The time derivative can then be dened
as
dH
dt
= [I]
d
dt
+
d [I]
dt
= M. (3.9)
In order to simplify Equation (3.9) a body xed coordinate frame can be established as can be
seen in Figure 3.1.
X
Y
Z
L,p
M,q
N,r
u
v
w
CG
Figure 3.1: Body axis xed coordinate system.
24
Using a body xed coordinate frame where the vehicles mass is assumed constant allows the time
rate of change of inertia to be dened as
d[I]
dt
= 0, simplifying the analysis of the vehicle equations
of motion. In order to determine the rectilinear acceleration of the vehicle in body coordinates, both
linear and rotational velocities must be considered. Using the theorem of Coriolis:
a =
dV
b
dt
=

V
b
+
b
V
b
. (3.10)
the linear and rotational velocities in the body frame can be used to nd the acceleration of the
vehicle with respect to an inertially xed frame. Taking
V
b
= u
i +v
j +w
k (3.11)
= p
i +q
j +r
k (3.12)
(3.13)
the rectilinear acceleration terms can be dened as:
a
X
= u +qw rv (3.14)
a
Y
= v +ru pw (3.15)
a
Z
= w +pv qu (3.16)
In a similar manner to Equation (3.4), the time rate of change in angular momentum can be
dened as
dH
dt
=

H +H (3.17)
where
H = [I] . (3.18)
25
Expanding H yields
H = [pI
xx
qI
xy
rI
xz
]
i + [pI
xy
+qI
yy
rI
yz
]
j + [pI
xz
qI
yz
+rI
zz
]
k. (3.19)
Using assumption 3.4 that there is a plane of symmetry in the X-Z plane as seen in Figure 3.2
Plane of Symmetry
y
x
y
z
Figure 3.2: Aircraft symmetry in the X-Z plane.
further simplication can be achieved. This assumption allows the right and left hand sides of the
aircraft to be considered mirror images of each other. Taking advantage of this assumption gives
I
yz
= I
xy
= 0. The full set of expanded equations of motion for an aircraft referenced to a body-
26
xed axis can then be written as:
F
X
= m( u +qw rv) (3.20)
F
Y
= m( v +ru pw) (3.21)
F
Z
= m( w +pv qu) (3.22)
L = pI
xx
rI
xz
+qr(I
zz
I
yy
) pqI
xz
(3.23)
M = qI
yy
+pr(I
xx
I
zz
) r
2
I
xz
+p
2
I
xz
(3.24)
N = rI
zz
pI
xz
+pq(I
yy
I
xx
) +qrI
xz
(3.25)
Relating the rotation of the body xed frame to the inertially xed earth frame requires the use of
rotation matrices to create a direction cosines matrix. Standard convention for aircraft uses a 3-2-1
rotation system dened as:
1. A right-handed rotation about the inertial :-axis (positive )
2. A right-handed rotation about the new y-axis (positive )
3. A right-handed rotation about the new x-axis (positive ), which is also the body roll axis.
This creates a transformation between body axis and earth xed inertial dened as:
_
_
x
y
z
_
_
body
=
_
_
1 0 0
0 cos sin
0 sin cos
_
_
_
_
cos 0 sin
0 1 0
sin 0 cos
_
_
_
_
cos sin 0
sin cos 0
0 0 1
_
_
_
_
x
y
z
_
_
earth
(3.26)
This rotation denes the relationship between the orientation of the earth xed coordinate frame
and the body xed coordinate frame with the three Euler angles (, , ). Using this relationship a
27
further relationship between body angle rates and the Euler angles rates can be dened as
_
_
p
q
r
_
_
=
_
_
1 0 sin
0 cos cos sin
0 sin cos cos
_
_
_
_
(3.27)
Remembering assumptions 3.1 - 3.3, and assuming ight close to the Earth the force of gravity
can be considered constant as applied to the aircraft. Gravity is considered to act normal to the local
horizontal plane in the Earth xed coordinate frame. The gravity vector can therefore be projected
into the aircraft body axis coordinate frame using the predened Euler angles (, , ) as seen in
Figure 3.3. The Earth relative gravity vector has a non-zero component in the z direction. Using
Equation (3.26) to transform earth xed coordinates to body xed coordinates,
(F
x
)
gravity
= mg sin (3.28)
(F
y
)
gravity
= mg cos sin (3.29)
(F
z
)
gravity
= mg cos cos . (3.30)
The force of gravity can therefore be added to the equations of motion to yield:
F
X
= m( u +qw rv +g sin ) = ma
x
(3.31)
F
Y
= m( v +ru pw g cos sin ) = ma
y
(3.32)
F
Z
= m( w +pv qu g cos cos ) = ma
z
(3.33)
L = pI
xx
rI
xz
+qr(I
z
I
y
) pqI
xz
(3.34)
M = qI
yy
+pr(I
xx
I
zz
) r
2
I
xz
+p
2
I
xz
(3.35)
N = rI
zz
pI
xz
+pq(I
yy
I
xx
) +qrI
xz
(3.36)
28
x
b
z
b
local horizontal
mg
local horizontal
y
b
z
b
mg
Figure 3.3: Gravity vector in relation to aircraft body coordinates.

These equations can be rearranged and solved in terms of their state derivatives.
89
Using the form
found in Stevens and Lewis,
89
they form four sets of equations commonly referred to as :
Force Equations
u = rv qw g sin +
F
x
m
(3.37)
v = ru pw +g sin cos +
F
y
m
(3.38)
w = qu pv +g cos cos +
F
z
m
(3.39)
Kinematic Equations
= p + tan (q sin +r cos ) (3.40)
= q cos r sin (3.41)
=
q sin +r cos
cos
(3.42)
Moment Equations
p = (c
1
r +c
2
)q +c
3
L +c
4
N (3.43)
q = c
5
pr c
6
(p
2
r
2
) +c
7
M (3.44)
r = (c
8
p c
2
r)Q+c
4
L +c
9
N (3.45)
29
Navigation Equations
p
N
= ucos cos +v(cos sin + sin sin cos )
+w(sin sin + cos sin cos ) (3.46)
p
E
= ucos sin +v(cos cos + sin sin sin )
+w(sin cos + cos sin sin ) (3.47)
h = usin v sin cos wcos cos (3.48)

where c
1
c
9
are dened
89, 90
as
c
1
=
(I
yy
I
zz
)I
zz
I
2
xz
(3.49)
c
2
=
(I
xx
I
yy
+I
zz
)I
xz
(3.50)
c
3
=
I
zz
(3.51)
c
4
=
I
xz
(3.52)
c
5
=
I
zz
I
xx
I
yy
(3.53)
c
6
=
I
xz
I
yy
(3.54)
c
7
=
1
I
yy
(3.55)
c
8
=
I
xx
(I
xx
I
yy
) +I
2
xz
(3.56)
c
9
=
I
xx
(3.57)
= I
xx
I
zz
I
2
xz
(3.58)
30
Often the states u, v, w are replaced with total velocity V
t
, and angle of attack and angle of
sideslip as dened in Figure 3.4. When u, v, w are replaced, they can be calculated using
_
_
u
v
w
_
_
=
_
_
V
t
cos cos
V
t
sin
V
t
sin cos
_
_
(3.59)
and the new states are dened as
V
t
=
u u +v v +w w
V
t
(3.60)
=
u w w u
u
2
+w
2
(3.61)
=
vV
t
v

V
t
V
2
t
cos
(3.62)
The solution to Equation (3.37) - Equation (3.48) or their wind derived equations in Equa-
tion (3.60) - Equation (3.62) is done numerically using a numerical integrator.
x axis (body)
z axis (body)
y axis (body)
x axis (wind)
relative
wind
x axis (stability)
Figure 3.4: Angle of attack and angle of sideslip in relation to aircraft body coordinates.
31
The primary forces acting on an aircraft beside gravity are lift and drag, and when nondimen-
sionalized their variation can be considered linear in small regions used for vehicle study. The lift
and drag force, which do not always act at the center of gravity, act to impart forces and moments
in all axes of the aircraft. Control surface movements (such as elevators, ailerons, and rudders)
also impart forces and moments as well as the engine(s), and any stabilizing surfaces, such as the
horizontal or vertical tail, wings and fuselage. Their effects are quantied in an aircrafts stability
derivatives. The standard convention used in control surface deection is shown in Figure 3.5.
Using the linear variation of the forces and moments over small regions of ight, another approach
in analyzing the equations of motion is often used instead of solving them directly. This approach
requires the equations to be simplied into a linear form. Traditionally the linearized equations
have been used in order to gain insight into the different aircraft modes of motion. The linearization
process for aircraft equations of motion has been well documented in many sources,
1, 88, 89
and a
more generalized process for linearizing nonlinear equation of motion can be found in many texts
as well.
86, 91
By rst picking a trim point where all the vehicle states are constant (not necessarily
zero), a set of linearized equations can be formed by either analytical or numerical techniques. For
the work presented in this document analytical techniques were used to create the linear equations
of motion. A Taylors series expansion
92
about a steady state (trim) condition for each state can
be accomplished where only the rst order terms of the Taylors series expansion are kept. This
expansion forms the linear equations of motion for an aircraft.
If an assumption is made that steady horizontal ight is present (
0
= 0) then the linearized,
longitudinal aircraft equations of motion referenced to the stability axis in frequency domain can be
32
A negative elevator deection (
e
<0) will
yield a tail down force and a positive pitching moment
A negative aileron deection (
a
<0) will yield
more lift on the left wing than on the right,
causing the aircraft to roll right wing down
(positive rolling moment).
A negative rudder deection (
r
<0) will
yield a side force, and a positive yawing moment
N>0
L>0
M>0
Figure 3.5: Typical sign conventions for aircraft control surface deections.
33
simplied to:
88
(s x
u
)u x
w
w +g = x
x
u
u
g
x
w
w
g
(3.63)
z
u
u + (s z
w
)w u
0
s = z
z
u
u
g
z
w
w
g
(3.64)
m
u
u (m
w
s +m
w
)w +s(s m
q
) = m
m
u
u
g
[(m
w
m
q
/u
0
)s +m
w
]w
g
(3.65)
s = q (3.66)
a
z
= w u
0
s = h (3.67)
Transfer functions can be generated from the linear longitudinal equations by solving the equa-
tions for the desired variable, with all the input variables except for the desired input set to zero. As
an example Kramers rule can be used to determine transfer functions.
w(s)
(s)
=
s X
u
X
g
Z
u
Z
U
0
s
M
u
M
s(s M
q
)
S X
u
X
w
g
Z
u
s Z
w
U
0
s
M
u
(M
w
s +M
w
) s(s M
q
)
(3.68)
34
These transfer functions as derived in McRuer et al.
88
can then be dened as:
(s)
(s)
=
A
s
2
+B
s +C
long
=
A
(s + 1/T
1
)(s + 1/T
2
)
long
(3.69)
w(s)
(s)
=
A
w
s
3
+B
w
s
2
+C
w
s +D
w
long
=
A
w
(s + 1/T
w
1
)(s + 1/T
w
2
)(s + 1/T
w
3
)
long
(3.70)
u(s)
(s)
=
A
u
s
3
+B
u
s
2
+C
u
s +D
u
long
=
A
u
(s + 1/T
u
1
)(s + 1/T
u
2
)(s + 1/T
u
3
)
long
(3.71)
h(s)
(s)
=
A
h
s
3
+B
h
s
2
+C
h
s +D
h
long
=
A
h
(s + 1/T
h
1
)(s + 1/T
h
2
)(s + 1/T
h
3
)
long
(3.72)
a
z
(s)
(s)
= s
2
h(s)
(s)
(3.73)
long
= As
4
+Bs
3
+Cs
2
+Ds +E = (s
2
+ 2
p
p
s +
2
p
)(s
2
+ 2
sp
sp
s +
2
sp
) (3.74)
Equation (3.74) is the characteristic polynomial for the aircraft longitudinal motion. The equation
is fourth order, and represents two second order modes of motion. These modes are dened as the
short period and phugoid modes. The natural frequency and damping ratio for the short period can
be shown
89
to be :
nsp
=
_
M
q
Z
V
T
M
_1
2
(3.75)
sp
=
(M
q
+M

+Z
/V
T
)
2
nsp
(3.76)
Only the longitudinal equations of motion are used for the work in this dissertation. The corre-
sponding lateral-directional linearized equations can be found in McRuer et al.
88
35
A
B
C
D
+
Z
w
X
(
Z
u
M
w
+
M
u
)
+
Z
(
M
w
X
w
w
)
(
X
u
+
Z
w
)
X
(
Z
u
M
w
Z
w
M
u
)
+
Z
(
M
u
X
w
M
w
X
u
)
+
M
Z
w
X
u
X
w
Z
u
w
Z
Z
u
(
X
u
+
M
q
)
+
M
U
0
X
(
U
0
M
u
Z
u
M
q
)
+
Z
X
u
M
q
U
0
M
X
u
g
(
Z
M
u
Z
u
)
u
X
(
Z
w
+
M
q
+
M
+
Z
X
w
X
(
Z
w
M
q
(
X
w
M
q
+
g
M
w
)
+
M
(
X
g
)
g
(
M
Z
w
M
w
)
h
Z
u
+
Z
(
M
q
+
M
+
X
u
)
M
Z
u
(
M
q
+
M
[
X
u
(
M
q
+
M
)
]
(
Z
M
u
Z
u
)
+
Z
[
M
u
(
X
g
)
X
u
]
M
[
Z
X
u
Z
u
(
X
g
]
M
q
Z
w
X
u
Z
w
M
q
X
w
Z
u
+
X
u
(
M
q
+
M
+
Z
w
)
X
u
(
Z
w
M
q
)
+
Z
u
(
X
w
M
q
+
g
M
w
)
M
u
(
X
g
)
E
=
g
(
Z
u
M
w
M
u
Z
w
)
T
a
b
l
e
3
.
1
:
V
a
r
i
a
b
l
e
d
e
n
i
t
i
o
n
s
f
o
r
t
r
a
n
s
f
e
r
f
u
n
c
t
i
o
n
s
f
o
u
n
d
i
n
E
q
u
a
t
i
o
n
(
3
.
6
9
)
-
E
q
u
a
t
i
o
n
(
3
.
7
3
)
.
36
Chapter 4
Review of Linear Systems
L
INEAR systems are often used to model more complex dynamical systems. While their appli-
cation is limited typically to a small range of the full systems domain, linear systems can be
used to approximate a dynamical system and greatly reduce the complexity of the analysis of the
system. This analysis employs the use of a homogeneous linear system of differential equations,
where the principle of superposition applies . The most general form of these systems can be taken
to be:
x
j
=
n
i=1
A
ji
(t)x
j
j = 1, 2, . . . , n
4.1 Transfer Functions
A transfer function is a method used to map a systems inputs to its outputs. Transfer functions
can be represented both in mathematical and graphical forms. An example of a graphical form of a
transfer function can be seen in Figure 4.1.
Mathematically, assume a system can be described as a linear differential equation
37
Transfer
Function
Input Output
Figure 4.1: Basic block diagram example
_
d
m
dt
m
+b
1
d
m1
dt
m1
+ +b
m1
d
dt
+b
m
_
y(t)
=
_
d
n
dt
n
+a
1
d
n1
dt
n1
+. . . a
n1
d
dt
+a
n
_
x(t)
(4.1)
where m > n. To analyze a function of this kind, the Laplace transform can be applied in order to
analyze the system in a frequency domain. The Laplace transform can be dened as
L[f(t)] = F(s) =
_

0
f(t)e
st
dt (4.2)
where s is a complex variable dened by s = +j. Applying the Laplace transform as dened in
Equation (4.2) to the system dened in Equation (4.1) the differential equation can be transformed
to the algebraic equation
_
s
m
+b
1
s
m1
+ +b
m1
s +b
m
_
Y (s)
=
_
s
n
+a
1
s
n1
+ +a
n1
s +a
n
_
X(s).
(4.3)
Rearranging Equation (4.3) a transfer function can be found where
U(s) =
_
s
n
+a
1
s
n1
+ +a
n1
s +a
n
_
(s
m
b
1
s
m1
+ +b
m1
s +b
m
)
(4.4)
and Y (s) is described as
Y (s) = U(s)X(s) (4.5)
38
and the time domain solution can be found by taking
y(t) = L
1
[Y (s)]. (4.6)
For a simple, second order system setting x(t) = (t) so that X(s) = 1, and setting the initial
conditions to zero, and dening an undamped natural frequency
n
=
_
K/J and a damping ratio
of = B/(2
KJ) the transfer function can be dened as

Y (s) =
K
Js
2
+Bs +K
(4.7)
and a time domain solution can be derived as
y(t) =

n
_
1
2
e
nt
sin
n
_
1
2
t, < 1. (4.8)
Various techniques have been developed to analyze transfer functions. One technique in particu-
lar that has gained wide use in ying qualities analysis is the Bode diagram. The traditional Bode
diagram, also know as the j-Bode diagram is constructed by varying s along the imaginary axis,
holding constant, and varying the frequency as seen in the plot of the complex plane in Fig-
ure 4.2. By varying the frequency , the complex solution to Y (s) can be plotted on a diagram.
Since the complex solution is comprised of both magnitude and phase components, a three dimen-
sional plot can be constructed, or two two dimensional plots can be made. The Bode diagram is the
plotting of the magnitude and phase portions of the complex solution to Y (s) by varying the fre-
quency . Traditionally these plots are done on a semi-log axis. The Bode plot using Equation (4.4)
can be seen in Figure 4.3.
Using transfer functions and block diagrams allows for a simple, graphical form where systems
can be created and represented graphically as well as mathematically. Blocks can be combined and
simplied to support mathematical analysis. An example of a feedback loop that is reduced can be
seen in Figure 4.4.
39
j
s= j
s=
cos
-1

Figure 4.2: Plot of the complex plane.
60
50
40
30
20
10
0
10
20
M
a
g
n
i
t
u
d
e

(
d
B
)
10
1
10
0
10
1
10
2
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bode Diagram,
n
=4, = 0.65
Frequency (rad/sec)
Figure 4.3: Example Bode diagram where U(s) =

2
n
s
2
+2ns+
2
n
40
4.2 Denition of Bandwidth
G
_
+
C
R E
=
R C
P
Figure 4.4: Example feedback block diagram reduction to single transfer function.
The 6 - DOF equations of motion of an aircraft as found in Equation (3.37) through Equa-
tion (3.48) lend themselves to linear systems analysis, and to further be reduced by splitting between
longitudinal and lateral-directional motion. The longitudinal motion can be described by three de-
grees of freedom with the linear systems found in Equation (3.63) through Equation (3.67). The
longitudinal motion linear equations can further be decomposed into two modes of motion that oc-
cur at separate frequencies, specically the phugoid and short period modes. Since these modes can
be considered separate, often analysis of each mode is done in a separate sense. The response of the
two rigid body modes of an aircraft in longitudinal motion are the primary focus of aircraft ying
qualities, since they dictate the response of the aircraft to pilot input.
Two different denitions for bandwidth are used in analysis of aircraft. One denition is the classical
denition of bandwidth as specied by electrical engineering applications. This denition of the
bandwidth of a system is classically dened as the region of frequency where the gain of the closed
loop system does not drop below -3 dB.
91
The system found in Figure 4.5 is given as an example
to illustrate the denition of bandwidth. A Bode plot of the closed loop system can be found in
Figure 4.6, showing that the system gain response drops to -3 dB at 2 rad/sec, dening the system
bandwidth as 2 rad/sec. The bandwidth therefore denes the range of frequency where the input
signal can be tracked by the system, and the bandwidth frequency denes the frequency where the
gain falls off from its low frequency value.
The ying qualities community has a second denition for system bandwidth as it is applied
41
_
+
Figure 4.5: Example system diagram.
20
15
10
5
0
5
10
M
a
g
n
i
t
u
d
e

(
d
B
)
10
1
10
0
10
1
270
225
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
-3 dB
Figure 4.6: Example system diagram.
42
to ying qualities analysis. This denition of bandwidth was rst offered by Hoh et al.
93
The
ying qualities denition of bandwidth is dened for handling quality criterion purposes [as] the
frequency at which the phase margin is 45 deg or the gain margin is 6 dB, whichever frequency is
lower.
93
Using the aircraft short period with a pilot closing the total system loop as seen in the
example system in Figure 6.7, a Bode plot can be constructed as in Figure 7.5. In this example,
the
bw gain
would be the ying qualities bandwidth frequency, and the system would be seen as
being gain limited. Using the gain and phase margin as parameters, the pilot can then have margin
to double his gain or adjust time delay without causing an instability in aircraft control. It can be
thought of as a measure of the limits of the closed loop systems robustness. For a pilot, it is the
highest frequency at which he can excite the aircraft while still feeling condent that the aircraft
will have the inherent robustness to reject outside disturbances. The ying qualities bandwidth is
shown as
bw gain
in Figure 7.5 and the traditional bandwidth value is found in as
bw
in Figure 7.5.
These two denitions for bandwidth were created by two communities for different purposes.
They will be compared and contrasted in their application to UAVs in Section 7.2.2.
43
Chapter 5
Classication
F
OLLOWING guidance from Chalk
18
in suggesting changes to MIL-F-8785(ASG), the rst step
in dening a new ying qualities criterion is to determine classications for the aircraft that
will be evaluated by the new ying qualities criterion. While classications exist for piloted aircraft,
no ight mechanics based classication is currently available for UAVs. The DoD has issued four
roadmaps for UAVs, in 2002,
33
another in 2005,
94
another in 2007,
95
and the most recent in 2009.
96
While each roadmap offers classes for UAVs, each one presents markedly different classes with
each revision.
The 2002 roadmap simply lists UAVs as either Operational, Developmental, or Other. The 2005
roadmap expands this view and also moves away from the UAV acronym and uses the acronym
UAS (Unmanned Aerial System) to encompass the entire air and ground system required to operate
the air vehicle. It denes categories as : Major UAS , Concept Exploration UAS, Special Operations
UAS, and Small UAS. In the Small UAS category, vehicle types are further broken down into Mini
UA, Micro Air Vehicles (MAV), and Organic Air Vehicles (OAV). This view of UAVs (or UASs) is
a systems engineering perspective of dividing UAVs. While this distinction creates a division based
on mission and size of UAVs, it does not create a clear distinction between UAVs for delineating
the dynamic response of aircraft. The 2007 roadmap changes focus from just UAVs (or UASs) to
44
5.1 Creating a Framework
unmanned systems (USs) of all kinds. It creates another set of categories from Level 0 to Level 5
based primarily on weight. This class description begins to follow more closely the classication
used in MIL-F-8785(ASG). The 2009 roadmap classies the air vehicle portion of USs by weight
and FAA airspace integration boundaries.
By classifying UAVs they can be grouped together with other aircraft of similar characteristics.
Due to the relatively small sample size of relevant data for UAVs, both RPVs and autonomous
vehicles are analyzed together to nd similarity in basic airframe dynamic response, focusing on
airframe classes more than control classes. This study investigates airplane like, xed-wing vehi-
cles, excluding hovering vehicles, or launch (rocket powered) vehicles.
Using the manned aircraft requirements framework as a guide, a new framework was constructed
for unmanned aircraft. The work done for manned aircraft has been widely accepted, and provides
a good foundation for unmanned aircraft. Understanding the justication for the manned aircraft
requirements framework gives insight that is required to create a new framework for unmanned air-
craft. While unmanned aircraft show large differences from manned aircraft, similarities to manned
aircraft should be exploited in creating a new set of requirements for unmanned aircraft.
5.1.1 Manned Aircraft Requirements Framework
In MIL-F-8785C
5
piloted ying qualities are broken down by aircraft class, ight phase, and
levels of performance. The manned aircraft classications have traditionally been categorized into
four different classes found in Table 5.1.
The classes are primarily delineated by gross weight and limiting load factor. This is because
they are considered to be gages of the maneuvering capability of an airplane and they reect the
mission for which it will be used.
18
The Flight Phase Categories are dened in Table 5.2.
These categories allow aircraft operation to be divided into different operational categories based
45
Class I small, light-weight, medium maneuverability airplanes
Class II medium-weight, low to medium maneuverability airplanes
Class III large, heavy-weight, low maneuverability airplanes
Class IV high-maneuverability airplanes
Table 5.1: Manned aircraft classications as dened in MIL-F-8785C.
Category A: Nonterminal ight phases that require rapid maneuvering,
precise tracking, or precise ight path control.
Category B: Nonterminal ight phases that are accomplished with gradual
maneuvers (climb, cruise, descent)
Category C: Terminal ight phases such as takeoff or landing
Table 5.2: Manned aircraft Flight Phase Categories as dened in MIL-F-8785C.
on tasks to be performed in the aircraft. The Levels then allow rating for ight safety and mission
capability (1 - 3). Boundaries for a given ying qualities specication are then listed by class and
category. These boundaries then specify a ying qualities level of one, two, or three for a given
vehicle class and maneuver category.
As explained in the user guide for MIL-F-8785B(ASG)
6
aircraft classications were created us-
ing Figure 5.1. The chart uses a log axis for aircraft maximum gross weight, and a linear axis for
load factor. It is important to note that there is overlap between the different aircraft classes. While
todays aircraft designers primarily use intuition to state whether an aircraft is Class I, II, III, or IV
it is important to note that there is a quantitative foundation for that intuition.
Signicant interest has been placed on creating UAV classes based on mission.
33, 94, 95
MIL-
F-8785ASG
16
originally did this with Class I: Primary trainer or observation, Class II: Bomber or
cargo, and Class III: Fighter or interceptor. Aircraft grouped by these missions were later reclassied
according to weight and load factor. The 2007 DOD US Roadmap lists 21 separate current and
future missions for UAVs. All of these missions can be combined into the broad classes of strike
46
Figure 5.1: Operational aircraft comparison in user guide for MIL-F-8785B.
6
and support, ISR and communication relay, and cargo / refueling tanker. Unlike manned aircraft
systems, size and weight of the UAVs that support these missions are not specic to the mission.
For example the Global Hawk and Hunter could both be considered ISR UAVs.
5.1.2 Aerodynamic Effects on Aircraft Dynamics
Aircraft motion is directly related to aerodynamic forces and moments. While this may seem
a simplistic statement, it is worth reviewing the prevailing contributors to the forces and moments
that affect aircraft motion. Anderson
97
describes lift, drag, and pitching moment as
L = f
L
(V
, S,
, a
) (5.1)
D = f
D
(V
, S,
, a
) (5.2)
M = f
M
(V
, S,
, a
) (5.3)
47
for a given angle of attack (). In order to perform dimensional analysis on drag (D), assume force
depends polonomially on the parameters:
D = ZV
b
S
e
a
f
(5.4)
where Z, b, d, e, f, g are dimensionless constants. Dimensional analysis can then be used to rewrite
drag as
D = ZV
2
S
_
a
_
f
_

S
0.5
_
g
. (5.5)
Recognizing that
M =
V
Re =
_

S
0.5
_
1
(5.6)
drag can then be rewritten as
D = 2Z qS
_
1
M
_
f
_
1
Re
_
g
. (5.7)
Using standard denitions of
q =
1
2
V
2
C
D
=
D
1
2
V
2
S
(5.8)
the drag relationship can be rewritten as
C
D
= Z
_
1
M
_
f
_
1
Re
_
g
. (5.9)
The same analysis can be performed on lift (L) and pitching moment (M) to yield
C
L
= f
1
(M
, Re, ) C
D
= f
2
(M
, Re, ) C
M
= f
3
(M
, Re, ). (5.10)
A similar result can be shown for lateral and directional forces and moments. While this is well
known, it is worthwhile to emphasize these relationships and that Mach number, Reynolds number,
and angle of attack (or sideslip) have direct functional relationships to the forces and moments that
48
5.2 Study of Available Aircraft
effect aircraft motion.
5.2.1 Sources of Data Considered
Data for forty-two different aircraft were recorded from various sources.
98102
These aircraft ranged
from small UAVs to all four classes of piloted aircraft. Calculations were made to create plots of an
operational Reynolds number vs. Mach number for each aircraft as well as Reynolds number vs
maximum takeoff/launch weight for each aircraft. The data are shown in Figures 5.2 - 5.7. For the
piloted aircraft, the calculations were done for a cruise conguration. The UAV calculations were
done in a loiter conguration. Both cruise and loiter are considered Category B maneuvers in the
piloted MIL-F-8785C specication. Cruise and loiter congurations were chosen since most of the
aircrafts operational ight time would be spent in one or the other conguration. The data sets were
plotted in this manner to determine possible groupings of aircraft that would conrm 8785C classes
and possibly lend guidance towards classes of UAVs. Mach and Reynolds numbers were chosen
because of their relationship to lift, drag, and pitching moment coefcients that in turn have a large
impact on short period and phugoid mode natural frequency and damping ratio. Weight was also
considered because of its effect on range and mission performance. Another motivating factor in
choosing weight was that the aviation industry recognizes an informal rule: the production cost of
an aircraft is directly proportional to its empty weight. That relationship is generally considered to
be $1500 per pound (FY94 dollars).
33
Adjusted for ination, the FY10 rate is $2200 per pound.
103
5.2.2 Results of Available Aircraft Study
A plot of operational Reynolds Number vs Mach Number was made for all forty-two aircraft, and
can be seen in Figure 5.2. From this plot of data, a separation can be found between UAVs and
manned aircraft. Distinguishing characteristics between different manned aircraft classes can also
be determined. For example, transport category aircraft (Class III) are scattered, but well above other
49
aircraft. Fighter aircraft (Class IV) are grouped closely together at Mach 0.85 and at a Reynolds
number of approximately 40 million. Medium weight aircraft (Class II) can be found lower and
to the left of ghter aircraft. Small aircraft (Class I) can be found near the lower left of the chart.
This chart also lends another perspective and some further insight into MIL-F-8785C Classes II, III,
and IV beyond that found in Figure 5.1. In the far lower left corner, well below Mach 0.3 a large
concentration of UAVs group together. Figure 5.3 shows a possible breakout of different manned
aircraft classes and UAVs using the data from Figure 5.2. One observation from Figure 5.2 is that
the different piloted aircraft begin to form groupings that reect their Classes as found in MIL-F-
8785C. Another observation is that UAVs collect in their own grouping, distinctly outside of the
other class groupings. This would seem to suggest that UAVs would merit their own ying qualities
specications just as each class of aircraft does. The UAVs do not however seem to form distinct
groups in the Figure 5.3.
Using the weight data gathered, another plot was created for all the aircraft in this study. This
time Reynolds number vs. Gross Takeoff Weight was used to compare the aircraft as seen in
Figure 5.4. Again, Class I, II, II, and IV aircraft group together, and an overlay of the different
groupings can be seen in Figure 5.5. In this plot there is considerable overlap between Class IV and
Class II manned aircraft. This is intuitive, since they cruise (operate) at similar conditions and are
primarily distinguished by maneuverability. As in Figure 5.1 the primary discriminator for these
aircraft is design load factor. With the exception of Global Hawk, the UAVs again are found at
the lower left of the chart. Figure 5.5 highlights a separation between UAVs and manned aircraft.
Another interesting observation is that all the aircraft tend to form a linear trend when viewed on a
log-log scale, with the UAVs dominating the bottom end of the scale. Figure 5.6 adds boundaries
denoting Domestic Use UAS levels as specied in Figure A.4 in the Unmanned Systems Roadmap
2007-2032.
95
These levels only make use of system weight as the discriminator between aircraft
levels. Reexamining these boundaries, a new trend can be found to discriminate UAV classes. A
new trend can be identied in Figure 5.7, where aircraft of similar operation, mission, and size are
50
0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1
0
0.2
0.4
0.6
0.8
1
1.2
1.4
1.6
1.8
2
x 10
8
F16
F15
C130
C141
C5
B747
DC8
C21
F104
F4
A7
T37
Shadow Shadow600 Sentry
Altus
Gnat
GBLHWK
Pred
Reaper
Scane Pionr HuntA HuntB HuntC
C172
J3
King Air
Citation X Horizon
Dakota Exdrone Finder
U2S
R
e
y
n
o
l
d
s

N
u
m
b
e
r
Mach Number
Comparison of Aircraft
Figure 5.2: Aircraft operational comparison by Reynolds and Mach numbers.
98102
Figure 5.3: Aircraft operational comparison by Reynolds and Mach numbers with manned
aircraft classes overlaid.
5, 98102
51
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
5
10
6
10
7
10
8
10
9
F16 F15
C130
C141
C5
B747
DC8
C21
F104
F4
A7
T37
Shadow
Shadow600
Sentry Altus
Gnat
GBLHWK
Pred
Reaper
Scane
Pionr
HuntA
HuntB
HuntC
C172
J3
King Air
Citation X Horizon
Dakota
Exdrone
Finder
U2S
Aerosonde
AeroStar
Buster
Unicorn
Silverfox
Vertex
SPAARO
C152
R
e
y
n
o
l
d
s

N
u
m
b
e
r
Gross Takeoff Weight (lbs)
Figure 5.4: Aircraft operational comparison by Reynolds number and weight.
98102
Figure 5.5: Aircraft operational comparison by Reynolds number and weight with manned
aircraft classes overlaid.
95, 98102
52
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
5
10
6
10
7
10
8
10
9
L
e
v
e
l

0
L
e
v
e
l

1
L
e
v
e
l

2
L
e
v
e
l
s

3

a
n
d

4
L
e
v
e
l

5
F16 F15
C130
C141
C5
B747
DC8
C21
F104
F4
A7
T37
Shadow
Shadow600
Sentry Altus
Gnat
GBLHWK
Pred
Reaper
Scane
Pionr
HuntA
HuntB
HuntC
C172
J3
King Air
Citation X Horizon
Dakota
Exdrone
Finder
U2S
Aerosonde
AeroStar
Buster
Unicorn
Silverfox
Vertex
SPAARO
C152
R
e
y
n
o
l
d
s

N
u
m
b
e
r
Figure 5.6: Aircraft operational comparison by Reynolds number and weight, with UA 2007
Roadmap Levels overlaid.
95, 98102
Figure 5.7: Aircraft operational comparison by Reynolds number and weight, with proposed
UAV classes overlaid.
98102
53
10
0
10
1
10
2
10
3
10
4
10
5
10
6
10
5
10
6
10
7
10
8
10
9
L
e
v
e
l

0
L
e
v
e
l

1
L
e
v
e
l

2
L
e
v
e
l
s

3

a
n
d

4
L
e
v
e
l

5
F16 F15
C130
C141
C5
B747
DC8
C21
F104
F4
A7
T37
Shadow
Shadow600
Sentry
Altus
Gnat
GBLHWK
Pred
Reaper
Scane
Pionr
HuntA
HuntB
HuntC
C172
J3
King Air
Citation X Horizon
Dakota
Exdrone
Finder
U2S
Aerosonde
AeroStar
Buster
Unicorn
Silverfox
Vertex
SPAARO
C152
R
e
y
n
o
l
d
s

N
u
m
b
e
r
C
l
a
s
s

I
I
Class 0,I
C
l
a
s
s

I
V C
l
a
s
s

I
I
I
C
l
a
s
s

V
Scane Scane
Pio
Dako
Exdrone EE
Finder n
de Aerosonde o A de A
r
icorn co co
Ae Ae AAAA
Silverfoxx Si SSi S
Scane Scane
SPA
r
ni
Pio
ako
SPA
Shado d Sh Sh ww
Shadow had
adow adoww
Sentry SS
Sh dddd
Gna GG
ionr n
HuntA HHHH
HuntB H
AA
Hunt Hunt H
ota
AAA
Sentry Sentry yy SSSS
ARO
oww
dow
w
y
na
ion
B
untt
ota
try try yy
ARO
w600
Altus AAAA
Pred
Alt Alt
Reaper
C
C
J3
arr t
CCCC152 152
CC
w60
per
C
C
a tt
522
1
L
e
v
L
ee
L
e
v
e
l
s

3

a
e
FFF16 FFF 5 66
CC21
F104 0 110
FFFFF
FFFFFFFFFFF
FFFF
4
16 16 AAA7777777FFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFFF
TTT37
GBLHWK H GBLHWK
172
King Air KKi Ki Ki KKKK
F 10 1
n X CCCCi
F16 F 55 666 77777777FFFFFFFFFFFFFFFFFFFFFFFFFFFFF
o oo Citation itation o Ci Ci CCCCCit Ci CCCC
FFFFFF16 16 16 16 FFFFFFFFFFF 1115 15 15 1555 66666 77777777FFFFFFFFFFFFFFFFFFFFFFF
U2SS
e
v
a
5
17
n X n
Figure 5.8: Aircraft operational comparison by Reynolds number and weight, with proposed
UAV classes and trend lines overlaid.
98102
54
b (y-intercept) a (slope)
7E6 -0.75
7E7 -0.75
7E8 -0.75
7E9 -0.75
3E11 -0.75
Table 5.3: Denition of classication separation lines.
grouped together. These classes can be further rened in Figure 5.8. Here the aircraft are separated
by lines of the form
y = bx
a
(5.11)
where b and a are dened in Table 5.3. Using the coefcients as dened in Table 5.3, Equa-
tion (5.11) takes the form of Equation (5.10) showing an inversely proportional relationship be-
tween Reynolds number and weight, similar the relationship between Reynolds number and coef-
cient of lift. The Reynolds number is nondimensionalized with a unit of length, usually the wing
chord. The aircraft weight in trimmed ight is the aircrafts lift, which is nondimensionalized with a
unit of area, usually the wing planform area. These two parameters combined can therefore take into
account the wing geometry of the vehicle. By taking into account Reynolds number and weight;
derivative aircraft can now be grouped together. This also allows such similar aircraft as the Sentry,
Shadow, and Hunter series to all be grouped together. The inclusion of Reynolds number provides
for the consideration of operational altitude and operational speed when classifying aircraft, which
is currently lacking in a weight-only distinction as seen in Figure 5.6. Another advantage is that
Predator derivative aircraft can be compared directly to small aircraft such as the Piper J-3 Cub
and Cessna 172. The Global Hawk aircraft can also be placed into a similar classication with the
U-2 and other mid-sized aircraft. Further, the possibility now exists for of a much lighter UCAV
to be placed into a classication with current manned strike aircraft. The Class II vehicles shown
in Figure 5.8 include smaller and slower UAVs such as the Scan Eagle, Dakota, Pioneer, Exdrone,
and Finder. Class I vehicles would include the Vertex, Unicorn, and Buster. The Class 0 vehicles,
55
or very light weight UAVs (such as micro air vehicles) were not included, due to a lack of available
data. Intuitively, these would merit a separate class.
Figure 5.9 shows the same data used in Figure 5.8 within the boundaries dened in the JUAS
COE classication. Currently no UAVs fall into class 4 of the gure. Table 5.4 then compares the
proposed classication method against the JUAS COE method. The 8785C Class column gives the
piloted aircraft classication, while the JUAS COE (Joint Unmanned Aircraft Systems Center of
Excellence) column gives the classication as dened in the 2009 Unmanned Systems Integrated
Roadmap.
96
The proposed changes to UAV classications can be found in the Proposed Class col-
umn of this table. This new classication would allow specications to be tied to aircraft weight
and ultimately cost, as well as an operational Reynolds number to aid in determining aerodynamic
effects.
0
100
200
300
400
500
600
1 10 100 1000 10000 100000 1000000
A
i
r
s
p
e
e
d

(
K
t
s
)

Weight (lbs)
JUAS COE Categories
Piloted
UAV
0 1 2 3
4
5
J3
Reaper
C152
C172
Predator
Global Hawk
U2
F16
F15
C5
B747
Hunter
ScanE
SPAARO
SHADOW
King Air
C21
Figure 5.9: JUAS COE classication boundaries as described in the 2009 Unmanned Systems
Integrated Roadmap.
96
56
5.3 Payload/Sensor Study
Piloted Aircraft 8785C Class Unmanned Aircraft JUAS COE Levels Proposed Class
F-16 IV Shadow 2 III
F-15 IV Shadow600 2 III
C-130 III Sentry 2 III
C-141 III Altus 3 IV
C-5 III Gnat 3 III
B-747 III Global Hawk 5 V
DC-8 III Predator 3 IV
C-21 II Reaper 3 IV
F-104 IV Scan Eagle 2 II
F-4 IV Pioneer 2 II
A-7 IV Hunter A 3 III
T-37 II Hunter B 3 III
Cessna 172 I Hunter C 3 III
Piper J-3 Cub I Dakota 2 II
King Air II Exdrone 2 II
Citation X II Finder 2 II
Horizon II Aerosonde 2 II
Cessna 152 I AeroStar 2 III
U2-S II Buster 1 I
Unicorn 1 I
Silverfox 2 II
Vertex 1 I
SPAARO 2 II
Table 5.4: Aircraft (manned and unmanned) used in comparison
5.3 Payload/Sensor Study
Part of the original intent of this study was to not only classify UAVs but to also classify the pay-
load/sensor packages they carry. By classifying payload/sensor packages, ultimately a compatible
set of classications of UAVs and payloads could be dened. Unfortunately very little informa-
tion describing payload/sensor packages for military applications is publicly available. Thirty-two
payloads were found in Janes Unmanned Aerial Vehicles and Targets
98
and collected in Table 5.5.
From the available data the payloads were then collected into similar types in Table 5.6. Since
sensor dynamic information was not available for this study, an alternate method was employed to
57
5.4 Conclusion
account for sensor dynamics in the creation of the ying qualities criterion. For the purpose of this
study, simple rst order system will represent the CloudCap
R
TASE gimbaled camera.
5.4 Conclusion
UAVs are at a similar point of development as conventional aircraft during the 1950s where ying
qualities requirements were developed to aid in procurement of future systems. While early work
has been done to dene RPV ying qualities by dynamic scaling, further work is needed to dene
full scale aircraft that are analogous to different UAVs. Work is also required to determine ying
qualities criteria for autonomous UAVs, which may not have full scale analogues. The rst step to-
ward focusing UAV ying qualities development is to dene classes of UAVs, an example of which
is presented in this paper.
From inspection of Reynolds Number and Mach number it is apparent that piloted aircraft do
collect into groupings that fall along the lines of traditional classes used for ying qualities analysis.
UAVs have been shown to fall outside of these traditional classes. Using Reynolds number and
weight it has been shown that UAVs begin to form groups or classes that can be divided similar
to piloted vehicles. This work has been done to dene a method for UAV classication for ying
qualities analysis. One vehicle from each UAV class will be chosen as representative of its class to
aid in development of a UAV ying qualities criterion.
Results that would aid in the further renement of UAV class denition boundaries would be
design limit normal acceleration, as well as bare airframe short period and phugoid modal responses.
While this data was not available for this study, collecting this information on future UAVs could
aid in future renement of UAV classication.
58
5.4 Conclusion
Sensor Manufacturer Type
D-500 BAE Infra-red linescanner
Series 52 BAI laser range nder PTZ camera
Series 66 BAI daylight TV and IR
Apache FLIR Systems Camera IR
BRITE Star FLIR Systems IR, TV, laser designator & rangender
MK II FLIR Systems Sensor Turret
Photon FLIR Systems IR Camera core
StarSAFIRE III FLIR Systems IR, TV, laser illuminator, rangender
Ultra 7000/MicroSTAR FLIR Systems IR , TV
Ultra 8500 FLIR Systems IR, TV, laser designator & rangender
AN/APY-8 Lynx General Atomics SAR
Lynx II General Atomics SAR
microSAR IMI IR Camera
miniSAR IMI SAR
CAM-NOIR Irvine SAR
RM-3000 L3 Solid state recorder
RM-8000 L3 Solid state recorder
AD-G/EXJAM Lockheed Martin Comm Jammer
IR-G600 Mitsubishi Thermal Imager
IR-M700 Mitsubishi Thermal Imager
AN/ZPQ-1 TESAR Northrop Grumman SAR
LR-100 Northrop Grumman ESM/ silent sensor
Raven Eye I Northrop Grumman electro-optic
Raven Eye II Northrop Grumman electro-optic
TUAVR Northrop Grumman SAR
AN/AAS-44V Raytheon Thermal Imager
AN/AAS-52 MTS Raytheon Multisensor targeting
UltraLite Raytheon silent payload
SU320MVis-1.7RT SUI visible light and SWIR camera
SU320MX-1.7RT SUI NIR camera
SU320US-1.7RT SUI SWIR camera
SU640SDV-1.7RT SUI SWIR camera
Table 5.5: Sample payloads found in research for this report
59
5.4 Conclusion
Payload Type Total Found
IR Camera 8
TV Camera 9
Turret 1
SAR 6
Recorder 2
Jammer 1
Imager 3
Passive 2
SWIR 3
Rangender 4
Designator 3
Table 5.6: Payloads from Table 5.5 reduced to general classes.
60
Chapter 6
Review of Manned Flying Qualities
T
O develop ying qualities standards for unmanned aircraft, the rst step taken was to catego-
rize UAVs according to parameters that inuence their dynamic response. This categorization
method is documented in Chapter 5. The next step in creating a new ying qualities standard is
to review prominently used piloted ying qualities standards to determine if any are applicable to
unmanned aircraft. Modifying existing standards will allow for a faster development of a criterion
as well as general familiarity and acceptance within the aerospace community. This report provides
a review of current manned ying qualities criteria that are in use today with a focus on their appli-
cability to unmanned aircraft. Documents reviewed include the following as well as their original,
source works where applicable.
1. AFFDL-TR-T25, RPV Flying Qualities Design Criteria
35
2. MIL-F-8785C, Military Specication, Flying Qualities of Piloted Aircraft.
5
3. MIL-HDBK-1797, Flying Qualities of Piloted Aircraft
104
4. SAE94900 Aerospace - Flight Control Systems - General Specication for Design, Installa-
tion and Test of Piloted Military Aircraft
105
5. NASA TN D-5153, The Use of Pilot Rating in the Evaluation of Aircraft Handling Qualities
23
61
6.1 AFFDL-TR-T25, RPV Flying Qualities Design Criteria
6. ADS-33E-PRF, Aeronautical Design Standard Performance Specication Handling Qualities
Requirements for Military Rotorcraft
106
The review of these documents focused on longitudinal ying qualities criteria, especially in the
frequency domain. The ndings of that review are presented in this chapter.
6.1 AFFDL-TR-T25, RPV Flying Qualities Design Criteria
This document was the product of phase II of a four phase AFFDL/FGC program plan to develop a
RPV Flying Qualities Specication. Its primary purpose was to develop a preliminary framework
for RPV ying qualities criteria which could be expanded, rened, and validated by simulation
and analysis...
35
The document contains a compilation of MIL-8785B, MIL-F-83300, MIL-F-
9490D, and MIL-C-18244A into a new design guidance document for use by RPVs.
35
Specic
tailoring of requirements to autonomous UAVs was not done in this study, since the work was done
specically for RPVs without mention of autonomously operating UAVs. The Naval Air Warfare
Center has also published results from applying and then testing modied manned ying qualities
criteria to small scale RPVs
36
using similar techniques to those used in AFFDL-TR-T25. Work
has also been done to show that short period natural frequency requirements should be different for
small UAVs compared to traditional aircraft.
41, 46
Using dynamic scaling, both reports suggested
modifying the CAP criterion found in Section 6.2 using different natural frequency boundaries for
small UAVs versus full scale piloted aircraft. While they identied the need for added time delay,
and cited specications for remote human operation, the CAP criterion is only applicable to bare
airframe dynamics or stability augmented control. It does not address autonomous aircraft control
in enough detail to be applicable to autonomous UAVs. The criteria in AFFDL-TR-T25 is therefore
not reviewed in this section, but is reviewed later in the discussion of MIL-HDBK-1797 and MIL-
F-8785C.
62
6.2 MIL-F-8785C
6.2 MIL-F-8785C
The primary focus of longitudinal ying qualities requirements in MIL-F-8785C is unaugmented,
piloted aircraft. Specically the short period and phugoid modes are specied using criteria based
on a human operators ability to act as the aircrafts augmentation and control system. The phugoid
mode specications listed in MIL-F-8785C are not changed from MIL-F-8785B.
107
Requirements
for phugoid stability in MIL-F-8785B are based primarily on pilot comments, specically during
landing tasks.
6
The ight path stability requirements in MIL-8785C also trace heritage to MIL-
F-8785B, and specically to landing tasks. All of these requirements were empirically derived
from pilot comment. The phugoid requirements are directly linked to gust rejection ability of the
pilot/aircraft system. While these values may be used as a guide for determining gust rejection char-
acteristics for aircraft with a control system base on human response times and perceptions, there is
no quantitative correlation between these values and unmanned aircraft. Since the primary guide to
determine these values was pilot input on unaugmented aircraft, their applicability to autonomous
UAVs is limited to those that are designed to match piloted aircraft dynamics for landing purposes
or gust rejection, irrespective of the UAVs control system.
Original ying qualities criteria for short period dynamics were based on a thumbprint plot
of short period undamped natural frequency vs. short period damping ratio.
19
This criterion did
not account for the zero in the numerator of the short period transfer function, specically 1/T
2
.
Bihrle
108
introduced the control anticipation parameter (CAP) to account for the short period trans-
fer function numerator. It is considered one of the earliest and most used piloted ying qualities
criteria, especially for unaugmented aircraft. While CAP was rst introduced in 1965, it continues
to be used in modern standards, including MIL-HDBK-1797. Modern updates to pitch stability
requirements can still trace heritage to CAP, including a dynamic margin constraint proposed by
Phillips and Niewoehner.
109
The CAP is ratio of the aircrafts pitch acceleration to change in steady state load factor. This ratio
63
6.2 MIL-F-8785C
correlates the sensitivity of the human vestibular organ to pitch acceleration to a sensed g-loading
of an aircraft. The human vestibular system is very sensitive to pitch acceleration and is the rst
feedback to the pilot that the aircraft is changing motion. The desired aircraft pitch change results
in an angle of attack (), and ultimately a change in g-loading on the aircraft as the ight path
changes. This g-loading (n
z
) of the aircraft is sensed by the pilot and is used to command aircraft
ight path. Closing this loop by the human requires that the dynamic relationship between pitch
attitude acceleration and aircraft loading be within human sensing capabilities. To derive CAP, rst
observe that the aircraft pitch acceleration can be represented as
=
qS cC
m
I
yy
(6.1)
and take the change in aircraft loading as
n
ss
=
_
1 +
Cm
C
L
lt/ c
_
C
m
C
m
C
L
C
L
+
cg
2V
2
C
m
(6.2)
By dividing Equation (6.1) by Equation (6.2) CAP can be dened as
CAP =
n
ss
=
W cC
m
C
L
+ 0.25S c
2
gC
m
I
y
_
1 +
Cm
C
L
lt/ c
_ (6.3)
The natural frequency of the short period mode can be derived
108
to be
n
=
W
W cC
m
C
L
+ 0.25S c
2
gC
m
I
yy
(6.4)
where
L
= qSC
L
(6.5)
64
6.2 MIL-F-8785C
Combining Equation (6.3) and Equation (6.4) yields
CAP =
W
L
2
n
_
_
1
1 +
Cm
C
L
lt/ c
_
_
(6.6)
The fractional term is approximately 1, leaving
CAP =
W
L
2
n
(6.7)
The aircraft load factor per unit angle of attack can be taken to be
n
=
L
W
. (6.8)
CAP can therefore be reduced to
CAP

2
n
n
(6.9)
This expression gave rise to the short period frequency requirements found in MIL-F-8785C,
5
specically charts such as the one found in Figure 6.1.
Chalk et al.
6
offer another derivation of CAP. Using Equation (3.69) and Equation (3.73) as a
starting point, they assume constant-speed in order to focus on the short period mode. They then
apply the initial value theorem to the

/
e
= s
2
/
e
transfer function and the nal value theorem
to n/
e
= n/(g
e
) transfer function. After algebraic reduction, the following expression can be
obtained
CAP =
n
ss
=

2
n
V
g
1
T
2
. (6.10)
(1/T
2
) is the numerator lead factor in the constant speed /
e
transfer function.
1
T
2
=
Z
e
M
w
M
e
Z
w
M
e
+Z
e
M
w
(6.11)
65
6.2 MIL-F-8785C
100
10
1.0
0.1
1.0 10 100
NOTE: THE BOUNDARIES FOR VALUES OF
OUTSIDE THE RANGE SHOWN ARE DEFINED
BY STRAIGHT-LINE EXTENSIONS
10.0
3.6
0.085
0.038
n/
nSP
~
RAD/SEC
n/ ~ g's/RAD
L
E
V
E
L
1
L
E
V
E
L
2
L
E
V
E
L
S
2
&
3
L
E
V
E
L
1
nSP
2
n/
Figure 6.1: Natural frequency specication for Category B ight phases, MIL-F-8785C.
5
66
6.2 MIL-F-8785C
It is valid in this case to assume |M
e
| |Z
e
M
w
| and that |M
e
| |Z
e
M
q
|, reducing
1
T
Z
e
M
w
M
e
Z
w
M
e
g
V
n
(6.12)
Substituting Equation (6.12) into Equation (6.10) then yields Equation (6.9).
Bihrle recommends that CAP be maintained above 25
degrees
second
2
g
, and that the lower critical value
is 15.
108
These recommendations lead to the boundaries as seen in Figure 6.1.
The relation between ight path and attitude is reported by Mitchell
25
as
=
1
T
2
s + 1
(6.13)
The values recommended by Bihrle reect the human pilots perception of motion of the aircraft,
and the pilots ability to react and then control the aircrafts short period mode. While the mathe-
matical concepts of CAP would be applicable to an unpiloted aircraft, the numerical relationships
would not be applicable, unless the vehicles sensors and control system are designed to mimic hu-
man function. One major drawback to using a purely CAP criterion is that almost all autonomous
UAVs possess a stability augmentation system, which would modify the bare airframe dynamics.
The higher order systems that arise from vehicle augmentation can not be specied using CAP. The
need to apply a Low Order Equivalent Systems (LOES) approach to account for higher order modes
is discussed by Moorhouse,
107
although no explicit requirements criterion is called out in MIL-F-
8785C. LOES as well as other higher order criteria are mentioned in MIL-F-8785C, although they
appear in a more mature form in MIL-STD-1797 (later revised to MIL-HDBK-1797), and are cov-
ered in that section of this report. It is important to note that ignoring the effects of high order
systems can have disastrous results. One example of ignoring the effects of the higher order modes
can be seen in Harris and Black
110
where they discuss the YF-22 PIO that resulted in the loss of an
aircraft.
67
6.3 MIL-HDBK-1797
6.3 MIL-HDBK-1797
As aircraft with stability augmentation systems became more prevalent, the need to specify their
ying qualities with criteria that expanded beyond CAP increased. The primary consideration be-
yond that specied with CAP was the time delay associated with the higher order modes introduced
with stability augmentation systems. This added time delay adversely affected aircraft handling,
requiring the new criteria to be introduced. The reference that documents these new ying qualities
criteria is MIL-HDBK-1797.
6.3.1 LOES
Early work by DiFranco,
111
as well as Neal and Smith
112
and Hodgkinson
113
determined that classi-
cal systems for aircraft specications were not adequate for digital control systems where high order
modes are introduced. High frequency phase lags associated with higher order systems can not be
accounted, and required a methodology to map higher order systems into lower order systems. This
gave rise to the low order equivalent systems (LOES) approach.
114
Hodgkinson
114
cites an example
using transfer functions to explain the basic principles of LOES. Figure 6.2 and Figure 6.3 are Bode
plots of :
transfer function 1 K
s + 0.7
(s
2
+ 4s + 4)(s
2
+ 16.8 s + 144)
(6.14)
s + 0.7
(s
2
+ 4s + 4)
e
0.12s
(6.15)
s + 0.7
(s
2
+ 4s + 4)
(6.16)
The traces in Figure 6.2 clearly show the difference in phase when part of the denominator from
Equation (6.14) is omitted. By approximating this loss with a time delay as in Equation (6.15) the
time delay incurred from the high order transfer function can be recovered. Boundaries have been
68
6.3 MIL-HDBK-1797
30
25
20
15
10
5
0
5
10
M
a
g
n
i
t
u
d
e

(
d
B
)
10
1
10
0
10
1
10
2
270
225
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
transfer functon 1
transfer functon 2
transfer functon 3
Figure 6.2: Example transfer functions to illustrate LOES.
114
0 1 2 3 4 5 6 7 8 9 10
0.2
0
0.2
0.4
0.6
0.8
1
1.2
1.4
p
i
t
c
h

r
a
t
e
time (s)
transfer functon 1
transfer functon 2
transfer functon 3
Figure 6.3: Example transfer function time histories to illustrate LOES.
114
69
6.3 MIL-HDBK-1797
10
1
10
0
10
1
10
2
20
10
0
10
20
M
a
g
n
i
t
u
d
e

(
d
B
)
Magnitude Limit
Frequency (rad/sec)
Upper Boundary
Lower Boundary
10
1
10
0
10
1
10
2
180
90
0
90
180
P
h
a
s
e

(
d
e
g
)
Phase Limit
Frequency (rad/sec)
Upper Boundary
Lower Boundary
Figure 6.4: LOES boundaries for matching high order to low order transfer functions.
115
70
6.3 MIL-HDBK-1797
set for LOES matching by Wood and Hodgkinson.
115
An example of these boundaries can be seen
in Figure 6.4. These boundaries were primarily set using the NT-33 in-ight simulation aircraft
with work done by Neal and Smith.
112
The boundaries are set to allow the maximum unnoticeable
added dynamics (MUAD) as perceived by a human pilot. If a response goes beyond the MUAD
boundary, the aircraft response will no longer be close enough to a typical airplane so that an
average pilot would be able to comfortably control the vehicle. These boundaries shrink around the
region where the cross over frequency for a given mode should occur. This weights the matching
of the low and high order systems at the frequency region that is the most important, while relaxing
the matching criteria at low and high frequencies. The boundaries specied in Figure 6.4 can be
described mathematically as:
Upper Magnitude Boundary
3.16s
2
+ 31.61s + 22.79
(s
2
+ 27.14s + 1.84)
(6.17)
Lower Magnitude Boundary
9.55E 2s
2
+ 9.92s + 2.15
(s
2
+ 11.60s + 4.95)
(6.18)
Upper Magnitude Phase
68.89s
2
+ 1100.12s + 275.22
(s
2
+ 39.94s + 9.99)
e
0.0059s
(6.19)
Lower Magnitude Phase
475.32s
2
+ 184100s + 29456.1
(s
2
+ 11.66s + 3.89E 2)
e
0.0072s
(6.20)
The boundaries dened focus on the crossover frequency range where piloted modes typically are
found. These boundaries, drawn to minimize MUAD, may not be applicable to an unpiloted vehicle
whose modes can be dened as desired. For unmanned aircraft the boundaries would have to be
redened to allow focus to remain on the crossover frequency region, no matter where it may lay in
the frequency spectrum. Further Mitchell and Hoh
25
do cite concerns where lead-lag augmentation
occurs near a vehicles short period frequency. Hodgkinson
114
however does point out that all
ying qualities criteria must use some level of higher order to lower order matching, and that LOES
provides the least level of mismatch. Another concern in using LOES relates to the boundaries as
seen in Figure 6.4. These boundaries are drawn under the assumption that a conventional response
71
6.3 MIL-HDBK-1797
is expected from the aircraft. In the case that a rate command attitude hold, or an attitude command
attitude hold response is required, the boundaries for LOES would have to be redened. The use
of nonconventional response types to best meet different mission task elements (MTEs) may be
required, and in the case of UAVs the possibility exists for other permutations of response types
beyond those typically used with a human pilot.
6.3.2 Neal-Smith
The Neal-Smith criterion
26, 112
is widely known for its large database of piloted data that was used in
its creation. The criterion uses frequency based analysis to specify requirements for a pitch tracking
task. This criterion was one of the rst to use a simplied pilot model to create a closed-loop
analysis method for piloted aircraft. The model used by this criterion can be found in Figure 6.5.
The criterion assumes that the pilot closes a visual-tactile loop around the controlled aircraft. The
assumption is made that the pilot will adjust his gain and phase compensation to permit the pitch
tracking task. The pilots gain and phase compensation then determines the pilots perception of
the aircrafts ying qualities. The criterion uses pitch rate command response to measure a closed
loop resonance, which can vary depending on the pilots ability to add lead or lag compensation. By
xing the a minimum bandwidth frequency of 3.5 rad/sec, and knowing the open loop frequency
response, the pilots compensation required to complete the task can be measured. This criterion
is geared specically to the human pilots ability to adapt to a given system, and does not show a
direct correlation to a system that is unmanned. Further, later work with this criterion led to the
Bandwidth criterion
114
where the bandwidth frequency is not specied, but allowed to vary with
system performance. An example of the Neal-Smith criterion plot can be see in Figure 6.6.
6.3.3 Bandwidth
The bandwidth criterion is an evolution of the Neal-Smith criterion.
116
In the bandwidth criterion,
the minimum closed loop pitch attitude tracking bandwidth is not specied, unlike Neal-Smith
where it is specied at 3.5 rad/sec. Allowing bandwidth to vary gives broader application of the
72
6.3 MIL-HDBK-1797
_
+
Pilot
Flight Control System
and Airframe
Figure 6.5: Pilot model used with Neal-Smith criterion.
26, 104, 112
Figure 6.6: Neal-Smith criterion for use in piloted aircraft with sample data.
26, 104, 112
73
6.3 MIL-HDBK-1797
criterion to aircraft of different sizes, so that a nimble ghter may have a higher bandwidth, or
lumbering cargo aircraft may have a smaller bandwidth. The criterion is frequency based, and
denes the bandwidth as the maximum frequency at which the phase margin is at least 45
and
the gain margin is at least 6 dB. Using these parameters as a metric allows the pilot to double his
gain or adjust time delay without causing an instability in aircraft control. While the criterion was
developed specically for the pitch attitude control loop, it can be applied to any system control
loop. A summary of the criterion is given by Hoh, et al. :
93
It follows that airplanes capable of operating at large values of bandwidth will have
superior performance when regulating against disturbances. When ying an aircraft
with low bandwidth, the pilot nds that attempts to rapidly minimize tracking errors
result in unwanted oscillations. He is, therefore, forced to back off and accept some-
what less performance (larger and more sustained tracking errors). It is not difcult to
imagine a clear cut preference on the part of pilots for aircraft with increased bandwidth
capabilities.
Hoh et al. further claim that Most, if not all, familiar handling quality metrics are, in fact, a measure
of bandwidth.
93
This criterion, along with most others rst rely on the low order equivalent system
reduction of a higher order system to a short period approximation
88, 93
represented as
e
=
M
e
_
s +
1
T
2
_
s
_
s
2
+ 2
sp
sp
s +
2
sp
_ (6.21)
The pilot is then modeled as an idealized system
e
= K
p
e
es
(6.22)
capturing a simple time delay for pilot response, and a gain from pilot input. The full closed loop
system including pilot compensation can then be modeled as Figure 6.7. By increasing pilot gain,
74
6.3 MIL-HDBK-1797
the system poles can be moved until reaching the bandwidth frequency (
BW
) as seen in Figure 6.8.
Using a Bode plot, the value of the bandwidth frequency can be found. The methodology for nding
the bandwidth frequency is as follows, and references Figure 7.5. The phase bandwidth frequency
(
bw
phase
) is the frequency at which the phase is 45
above the 180
crossing on the phase plot.

The gain bandwidth frequency (
bw
gain
) is the frequency at which the gain is 6 dB above the value
found at the 180
crossing frequency (
180
). The bandwidth frequency
bw
is then the lesser of
either
bw
gain
or
bw
phase
. If
bw
=
bw
gain
the system is said to be gain margin limited, and if
bw
=
bw
phase
the system is said to be phase margin limited.
The shape of the phase curve above
bw
is also used in this criterion. It is used to measure time
delay, since rapid roll offs in time delay can be represented as pure time delay. The change in phase
due to time delay is linear in frequency, and can be approximated as
e

p
=
1
+ 180
57.3
1
(6.23)
where
1
is a frequency above the neutral stability frequency and
p
is the approximation to
e
. The
criterion was originally expressed as Figure 6.10. In later application, this criterion has been used to
predict PIO (Pilot Induced Oscillation, also referred to as APC, Aircraft Pilot Coupling). In further
study by Mitchell and Klyde
117
the pitch attitude bandwidth limits in MIL-STD-1797 have been
found to be too stringent and have been signicantly adjusted. The new boundaries are presented
in Figure 6.11
Shortcomings to this approach have been highlighted by Hodgkinson et al.
29
where a shelf like
frequency response can cause small pilot gain changes to cause large changes in bandwidth fre-
quency. Hodgkinson et al. suggest using a Nichols chart to represent the short period transfer func-
tion. This method, while applicable is not as widely used as the original Bode plot form. This crite-
rion has been primarily applied to the short period mode of aircraft, but has also seen application to
ight path angle control as well as roll axis control. While the PIO boundaries to the bandwidth cri-
terion do not apply to UAV applications, the concept of nding the systems maximum bandwidth
75
6.3 MIL-HDBK-1797
_
+
Pilot
Short Period
Airframe Dynamics
Figure 6.7: Bandwidth longitudinal control block diagram.
93
X
X
Figure 6.8: Root locus longitudinal control diagram with bandwidth frequency.
93
76
6.3 MIL-HDBK-1797
60
40
20
0
20
40
M
a
g
n
i
t
u
d
e

(
d
B
)
10
2
10
1
10
0
10
1
10
2
270
225
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bandwidth Definition
Frequency (rad/sec)
6dB
bw gain
45 deg
bw phase
180
Figure 6.9: Denition of bandwidth frequency,
BW
.
93
Figure 6.10: Bandwidth requirements from MIL Handbook,
22
77
6.3 MIL-HDBK-1797
0 1 2 3 4 5 6 7
0
0.05
0.1
0.15
0.2
0.25
0.3
0.35
0.4
P
i
t
c
h

A
t
t
i
t
u
d
e

P
h
a
s
e

D
e
l
a
y
Pitch Attitude Bandwidth
Susceptible to PIO
PIO if ight path bandwidth
bw
< 0.58 rad/sec
No PIO
(pitch bobble if pitch rate overshoot > 9dB)
PIO if
pitch rate
overshoot
>12dB
Figure 6.11: Revised bandwidth requirements to aid in PIO prediction.
117, 118
78
6.3 MIL-HDBK-1797
frequency to gage system performance does hold some merit for UAV applications, specically
when the system is judged in relation to the bandwidth of a sensor and fullling mission require-
ments where both systems must work in concert. By specifying a minimum total aircraft/sensor
system bandwidth this criterion could be used to ensure proper performance for a desired mission.
The applicability of this criterion to other modes beside the short period, and to any response type
makes this criterion particularly attractive.
6.3.4 Time Domain Methods
The time domain criterion discussed in MIL-STD-1797 by Gibson
119
was originally dened in
AGARD-CP-333 and ICAS-86-5.3.4. This criterion uses dropback as a means to quantify accept-
able pitch response in the time domain. The time domain criterion does not necessarily break
categories into level one, two, or three, but does provide guidance for short period time response.
The criterion gives guidance to classical response of a short period, but does not give correlation to
higher order systems where control systems may be involved to control the aircraft. MIL-STD-1797
states:
A conclusion to be drawn is that for low-order aircraft with elementary pitch damper
augmentation, a low maneuver margin should be aimed for, with its inherently high
natural damping, if precision pitch handling is required.
22
Interestingly Hoh et al. have found that This criterion has been proven to appear in the bandwidth
criterion as well. The upper limit on bandwidth as stated in MIL-STD-1797A has been found to be
characterized by the pitch attitude dropback.
120
The dropback criterion has traditionally been used
to judge pitch attitude and its relationship to ight path angle during the landing task. Dropback is
dened as seen in Figure 6.12, and the relationship between pitch attitude and ight path angle can
be seen in Figure 6.13. When A = B in Figure 6.12 then an ideal integral response in pitch attitude
is achieved. When A > B overshoot may occur, and the pitch response is deemed sluggish. When
A < B the vehicle is overly responsive in pitch and dropback at the end of the maneuver will occur.
79
6.4 SAE94900
According to Gibson A = B is ideal, but dropback is preferred to overshoot. The ratio of q
m
to
steady state pitch rate during the maneuver must be 1.0 q
m
/q
ss
3.0.
Figure 6.12: Denition of dropback by Gibson.
22
Gibson also investigated criteria to predict PIO, and offered the phase rate criterion as a possible
solution. It is similar to the Neal-Smith criterion in that it attempts to measure required phase
compensation by a pilot during a maneuver. While others have shown applicability of this criterion
to remotely piloted vehicles (RPVs) ,
116
its applicability to UAVs is limited, since its boundaries
and metrics relate specically to a humans ability to compensate for a decient control system.
6.4 SAE94900
The primary scope of SAE94900 (formerly MIL-F-9490D) is to establish general performance,
design, test, development and quality assurance requirements for the Flight Control Systems of
military piloted aircraft.
105
This document does not necessarily give criteria that may be useful
in specifying the dynamic response of aircraft, but it does lend guidance to the specication of the
control systems that determine the dynamic response of modern, augmented aircraft. Signicant
80
6.4 SAE94900
Figure 6.13: Relation of ight path angle and pitch attitude as shown by Gibson.
22
Mode
Frequency
(Hz)
Airspeed
Below V
0min
V
0min
to V0max
At
Limit Airspeed (V
l
)
At
1.15 (V
l
)
fm < 0.06
GM = 6 dB,
no phase requirement
GM = 4.5 dB
PM = 30
GM = 4.5 dB
PM = 20
GM = 0
PM = 0
0.06 fm
First Elastic Mode
GM = 6 dB,
GM = 6.0 dB
PM = 45
GM = 4.5 dB
PM = 30
GM = 0
PM = 0
fm > First
Elastic Mode
GM = 6 dB,
GM = 8.0 dB
PM = 60
GM = 6.0 dB
PM = 45
GM = 0
PM = 0
Table 6.1: Gain and phase margin requirements as dened in AS94900.

105
portions of this document specify the safety and reliability of components of a ight control system,
which are out of scope for this study. The document does, however list robustness requirements
for ight control systems stating gain and phase margin requirements for piloted aircraft during
aerodynamic closed loop operation. The gain and phase margin requirements are listed in Table 6.1.
AS94900 also describes intensity exceedance probability, for gusts and turbulence, referencing
MIL-STD-1797 for specic turbulence requirements. These turbulence values place basic robust-
ness requirements on a ight vehicle. Another criterion that is listed in AS94900 that ties directly to
turbulence relates to ride smoothing. Specically Section 3.2.4.2.8.1 describes the ride discomfort
index as
D
i
=
__
ft
0.1
[w(f)[
2
[T
CS
(f)[
2
u
(f)df
_
1/2
(6.24)
81
6.4 SAE94900
Ride Discomfort Index Probability of Exceedance
0.7 0.20 vertical
0.28 0.01 lateral
Table 6.2: Probability of exceedance of ride discomfort indices (vertical and lateral) in
AS94900.
105
where a lower index D
i
denotes a more comfortable ride. The acceleration weighting function
w(f) is shown as Figure 6.14. T
cs
is the transmissibility of the disturbance at the crew station, and
u
(f) is the Von Karmen gust power spectral density of intensity specied in MIL-STD-1797.
10
1
10
0
10
1
10
2
10
1
10
0
10
1
Frequency (Hz)
A
c
c
e
l
e
r
a
t
i
o
n

W
e
i
g
h
t
i
n
g
,

W
(
f
)

(
1
/
g
)
Vertical Vibration
Lateral Vibration
Figure 6.14: Denition of acceleration weighting function w(f).
105
To ensure adequate ride smoothing AS94900 requires a probability of exceedance as seen in Ta-
ble 6.2. This criterion was originally created to reduce crew fatigue, especially during long missions
in large aircraft, such as bombers, where crew would be stationed at various points in the aircrafts
fuselage.
121123
While UAVs will not have human crew aboard the vehicle, the sensors that the
UAV carries will require ride conditioning to ensure proper operation. The ride quality index can be
82
6.5 NASA TN D-5153
modied to give a measure of frequency vibration imparted to a sensor onboard an aircraft based on
a predicted gust load. While this may not be directly seen as ying qualities, modication to this
criterion may give design guidance to gust rejection requirements for a vehicle that will become a
part of the vehicles ying qualities.
6.5 NASA TN D-5153
This document is the primary reference for the Cooper-Harper pilot rating scale.
23
Later revisited
in a Wright Brothers Lectureship in Aeronautics by Harper and Cooper,
24
it creates a rating scale
that can be used by a pilot to judge an aircrafts ability to perform a given task. As important as
providing an accurate rating, the task itself is key in judging the pilot/aircraft system. Along with
the numerical rating, pilot comments are required to give justication to the rating. Often pilot
comments are just as, if not more informative than the rating number given. Beyond giving a stan-
dard scale to judge aircraft handling qualities, it serves as an aid to probe for deciencies in aircraft
handling qualities. An example of the Cooper Harper rating scale can be seen in Figure 6.15. While
this document was written specically to give human aircraft operators a way to quantify data that
was purely qualitative, it can still remain applicable to UAVs so long as the tasks are appropriately
dened. The Cooper Harper scale could be used to judge a vehicles mission effectiveness in rela-
tion to an operators ability to perform a given task with a sensor package, or an operators ability
to launch or recover a vehicle.
6.6 ADS-33E-PRF
The scope of ADS-33-PRF states that This specication contains the requirements for the ying
and ground handling qualities of rotorcraft.
106
While the targeted aircraft of this document is ro-
torcraft, there are lessons to be learned from this document from the process used in dening ying
qualities requirements. These lessons can be used in creating a UAV ying qualities criteria. Un-
like MIL-STD-1797 that contains multiple criteria for short period ying qualities with no guidance
83
6.6 ADS-33E-PRF
Figure 6.15: Cooper Harper rating scale.
23
(NASA, TN D-5153, fair use.)
84
6.7 Conclusion
as to when one may be more appropriate in a given situation, ADS-33E-PRF species only one
criterion. The philosophy used in ADS-33E-PRF owes its heritage to technical report written by
Mitchell et al.
124
suggesting revisions to MIL-STD-1797a. The document species mission task
elements that are to be used in aircraft design and evaluation. The mission task elements then rate
different tasks in terms of task aggressiveness and precision. The task elements are broken down
to allow for different standards of precision and aggressiveness following recommendations from
WL-TR-94-3162.
124
Specically, Table 1 in ADS-33-E-PRF lists specic mission task elements
that must be satised for compliance. The performance standards for each task can then be tailored
based on different levels of desired precision and aggressiveness. Unlike other documents reviewed
for this study, ADS-33-PRF uses a system engineering approach to dening requirements for the
aircrafts performance. This is accomplished by rst dening mission tasks that are to be preformed
by the aircraft, and then specifying performance requirements to verify that requirements are met
for those mission tasks, and then nally giving ight test maneuvers that can be used to validate the
vehicles ability to perform the desired mission tasks.
6.7 Conclusion
Piloted ying qualities criteria were rst created to categorize and then standardize aircraft. Later
ying qualities criteria were created to respond to the growing changes in aircraft ight control
technology. As more complex aircraft emerged, different techniques were employed to analyze and
predict vehicle-pilot interaction and performance. As newer technologies such as nonlinear control
and unmanned aircraft continue the evolution of aircraft ying qualities criteria must also evolve
to address these new technologies. By basing new criteria on techniques historically accepted, a
sense of comfort and intuition that has been fostered in the aviation community can be leveraged
to introduce newer standards. The concepts underpinning criteria such as bandwidth, the Cooper-
Harper rating scale, mission task elements, and metrics similar to the ride discomfort index can be
transitioned forward in the creation of new criteria. The systems approach taken in ADS-33E-PRF is
85
6.7 Conclusion
especially attractive in creating a mission based criterion for UAV use. Taking the knowledge from
this chapter, the next two chapters look to apply some of the concepts outlined in this chapter to
nonlinear control systems and to UAVs. Chapter 7 proposes new ying qualities criteria for UAVs,
and Chapter 8 details a study to use the Cooper-Harper rating scale to rate a pitch tracking task with
an adaptive controller.
86
Chapter 7
UAV Flying Qualities
T
HE application of ying qualities analysis to UAVs presents a unique problem. Since ying
qualities analysis traditionally focuses on pilot opinion, a major component of ying qualities
analysis must be rethought. There is however, a large body of work of piloted criteria that does pro-
vide guidance for the application of ying qualities analysis to UAVs. Any application of a criterion
to UAVs must follow some of the basic tenants used in piloted analysis. It must be simple enough
to use, effective enough to make worth using, and familiar enough to users so that there is a feeling
of intuitive comfort in using it. Taking this guidance a methodology is described in this chapter that
can be used to specify UAV ying qualities requirements.
7.1 UAV Criteria Guidance
Based on documents reviewed in chapter 5 and chapter 6, guidance is offered in the effort to cre-
ating a ying qualities criterion for UAVs. Realizing the extensive work that has been done in the
aviation community to create ying qualities specications for piloted aircraft, this section provides
an outline to be used to create a new UAV ying qualities criterion.
87
7.1.1 Classication of Aircraft
Traditional ying qualities methods break their specications into different aircraft classications
and ight phase categories. The manned aircraft classication as specied in MIL-F-8785C can be
found in Table 5.1 of in Section 5.1.1.
The classes are primarily delineated by gross weight and limiting load factor. This is because
they are considered to be gages of the maneuvering capability of an airplane and they reect the
mission for which it will be used.
18
In an effort to create aircraft classes that mimic the utility
of manned aircraft categories a proposed classication for UAVs is proposed in Figure 5.8 of Sec-
tion 5.2.2. Chapter 5 highlights the need for UAV classication, as well as the methodologies chosen
for the classes created.
7.1.2 Categorization of Flight Phases and Aircraft Maneuvers
Just as manned aircraft have been traditionally categorized by standards used in MIL-F-8785C,
the ight phase categories as dened by MIL-F-8785C are still used as todays standards. These
categories were introduced in Table 5.2 of Section 5.1.1, and are not fully expanded as dened in
MIL-F-8785C in Table 7.1. These categories are based on manned aircraft that were developed
in the 1950s. Mitchell et al. proposed changes to these categories as well as how ying qualities
standards are included in the design process by incorporating mission-oriented tasks.
124
Mitchell et
al. give the following reasoning for breaking the traditional three ight phase categories into four
categories:
Category A is too broad, ranging from air-to-air combat to reconnaissance.
Category B is too lenient and should apply only to ight in Visual Meteorological
Conditions (VMC).
Category C is not sufciently stringent for the precision landing typically preformed in
ight research, and Category A is too stringent for that MTE (Mission Task Element).
88
Instead Mitchell et al.
124
suggest dividing ight phase categories into four categories, based on
precision of the task and aggressiveness required to accomplish the task. An example of how these
new categories relate to the previous categories dened in MIL-F-8785C can be found in Figure 7.1.
Mission tasks that a given aircraft would be required to perform would then be divided into one of
the four new categories. These mission tasks were called mission task elements (MTEs), and they
would be chosen from a catalogue of predetermined tasks. A proposed breakdown for these MTEs
is given as in Table 7.2. Specic ight test maneuvers for standard test evaluation maneuver sets
(STEMS) are also given
125
as a method for validating the ying qualities requirements. The STEMS
would be chosen to verify that the MTE requirements had been implemented correctly. Following
this structured systems engineering approach, ADS-33-PRF implemented this methodology.
120
In order to establish a UAVying qualities criterion, UAVs should followthe same four categories
presented by Mitchell et al.,
124
and implemented in ADS-33-PRF. Each of these new categories
would have MTEs prescribed to them, an example of which would be Table 7.2. Using Figure 7.3
as an example, aircraft with different roles would be assigned MTEs that would only be relevant
to that role. As roles in the UAV force change and evolve, a basic pool of MTEs would remain
relatively static, allowing different combinations of MTEs for any new or changing aircraft role.
These MTEs would then be assigned a MTE category of A, B, C, or D, determined by the MTEs
required combination of precision and aggression, following those previously shown in Figure 7.1.
The MTE categories would then contain requirements for the UAV ying qualities criterion. Using
this method ensures that only four sets of requirements would be created for the ying qualities
criterion, and would be relevant across UAV classications. The MTEs would also be linked to
ight test elements as shown in Figure 7.2. This would allow vehicle requirements to ow down
to vehicle design and analysis as well as verication of the vehicle requirements based on MTEs.
Unlike criteria based on ight phase categories in Table 7.1, the criteria could specify compliance
to a desired set of maneuvers linked to a MTE category as opposed to a general maneuver category.
89
Category A
Nonterminal ight phases that require rapid maneuvering,
precise tracking, or precise ight path control.
Air to air combat (CO)
Ground Attack (GA)
Weapon delivery/launch (WD)
Aerial recovery (AR)
Reconnaissance (RC)
In-ight refueling (receiver) (RR)
Terrain following (TF)
Antisubmarine search (AS)
Close formation ying (FF)
Category B
Nonterminal ight phases that are accomplished with gradual
maneuvers and without precision tracking, although accurate
ight path control may be required.
Climb (CL)
Cruise (CR)
Loiter (LO)
In-ight refueling (tanker) (RT)
Descent (D)
Emergency descent (ED)
Emergency deceleration (DE)
Aerial delivery (AD)
Category C
Terminal ight phases are normally accomplished
using gradual maneuvers and usually require accurate ight-path control.
Takeoff (TO)
Catapult takeoff (CT)
Approach (PA)
Wave-off/go-around (WO)
Landing (L)
Table 7.1: Flight Phases as dened in MIL-F-8785C.
90
8785C
Category A & C
Precision Tasks
8785C
Category B
Non-Precision Tasks
New Category A
Precise and
Aggressive
New Category C
Precise and
Non-Aggressive
New Category B
Non-Precise and
Non-Aggressive
New Category D
Non-Precise and
Aggressive
Figure 7.1: Proposed categories.
124
Non-Precision Tasks Precision Tasks
Non-Aggressive Aggressive Non-Aggressive Aggressive
Category B Category D Category C Category A
Reconnaissance Gross Acquisition Aerial Recovery Tracking a Moving Target
In-ight Refueling (tanker) Max g turn In-ight Refueling (receiver) Ground Attack
Takeoff Anti-Submarine Precision Landing Terrain Following
Non-precision Landing Split-S, Loop, etc. Formation Flight Precision Aerobatics
Table 7.2: Example maneuvers for proposed categories.
124
91
Mission Task Elements (MTEs)
Task Categories
Group A
Agressive /
Precision
Group B
Non-Agressive /
Non-Precision
Group C
Non-Agressive /
Precision
Group D
Agressive /
Non-Precision
Battle
Damage
Assesment
Target
Illumination
Weapons
Delivery
Urban Over the Hill
High Altitude
Target ID Recon
Flight Test Elements (FTEs)
Battle
Damage
Assesment
Target
Illumination
Weapons
Delivery
Urban Over the Hill
High Altitude
Target ID Recon
Figure 7.2: Flying qualities cycle
92
Mission Task Elements (MTEs)
Aircraft Roles
Strike ISR Comm SAR
Target
Illumination
Weapons
Delivery
Battle
Damage
Assesment
High Altitude
Over the Hill
Urban
Task Categories
Group A
Agressive /
Precision
Group B
Non-Agressive /
Non-Precision
Group C
Non-Agressive /
Precision
Group D
Agressive /
Non-Precision
Air /
Air
Air
Refuel
Figure 7.3: Mission task elements
93
7.1.3 Evaluation of UAV Task Performance
The evaluation of the sensor/aircraft system performance of a task must be done on a scale that can
be applicable across tasks. While predicted performance analysis can be used to measure compli-
ance with design requirements and specications listed in the ying qualities criteria, a scale used
to measure ight test performance to aid in validation testing is also required. The Cooper Harper
scale as seen in Figure 6.15 has traditionally been used in piloted aircraft to perform this function.
The Cooper Harper scale can be modied as seen in Figure 7.4 to become a task generic scale for
UAVs. Just as in piloted evaluation, the task description and comments by the UAV operator will be
key in using the scale to properly evaluate the sensor/aircraft system. Using this new rating scale the
UAV and the sensor system can be evaluated as an integrated system against mission performance.
This rating scale also includes the ability to check an autonomous UAV against disturbances so that
the system can be evaluated in a relevant operating environment. The ratings of 7, 8, and 9 are de-
pendent only on the UAV without the sensor, and could be used to aid in establishing airworthiness
independent of the integrated sensor system. Just as in piloted evaluations adequate and desired
criteria would be established, and would be key to the evaluation of the task with the rating scale.
Disturbance levels would also have to be included with the adequate and desired criteria for a given
task in order evaluate mission effectiveness. The Results of a System to a Given Task column
replace the Demands on a Pilot in Selected Task or Required Operation in the Figure 6.15. By
starting at the bottom left of the chart, and then systematically working through the decision tree a
UAV operator can provide feedback from a task evaluation in structured manner.
7.1.4 Applicable Manned Flying Qualities Criteria
The ying qualities criteria reviewed in this report make an import assumption in their analysis.
They all assume that a pilot is operating the aircraft. While that is a valid assumption for their
intended purpose, it is not for the application of ying qualities criteria to autonomous UAVs. The
question then remains if any of the criteria are still applicable if this assumption is violated. The
answer is a qualied yes.
94
Is the UAV
Stable?
Improvement
Mandantory
Is adequate
mission
performance
possible?
Is mission
performance
satisfactory without
improvement?
UAV Operator
Evaluation
Deciencies
warrant
improvement.
Deciencies
warrant
improvement.
Major deciencies
Control will be lost during some portion of the required
operation
10
Major deciencies Control maintained with specied disturbance levels 9
Major deciencies
UAV tracks commands without specied disturbance
levels
8
Major deciencies
UAV tracks commands with specied disturbance
levels
7
Very objectionable but
tolerable deciencies
Adequate performance of integrated system without
specied disturbance levels
6
Moderately objectionable
deciencies
Adequate performance of integrated system with
5
Minor but annoying
deciencies
Desired performance of integrated system without
4
Fair, mildly unpleasant
deciencies
Desired performance of integrated system with
3
Good, negligible
deciencies
Desired performance of integrated system with some
margin and specied disturbance levels
2
Excellent, highly
desireable
Desired performance of integrated system with
signicant margin and specied disturbance levels.
Expansion of performance envelope possible.
1
Adequacy for Selected
Task or Required Operation
Integrated System
Characteristics
Results of System to a Given Task
Operator
Rating
Yes
No
Yes
No
No
Yes
Note: disturbance levels, desired, and adequate criteria must
be specied for task evaluation
Figure 7.4: A modied form of the Cooper Harper scale for use in sensor/UAV evaluation.
95
The CAP criterion from MIL-F-8785C also assumes that a control system with signicant time
lags is not present. While some UAVs may rely on bare airframe dynamics instead of a control
system to augment stability, it is not realistic to assume that this will be the normal case with a fully
autonomous UAV. For fully autonomous UAVs, one of the criteria that take into account higher or-
der systems is a more obvious choice. The structure of MIL-STD-1797, and later MIL-HDBK-1797
present several options for ying qualities criteria with narratives explaining each method without
giving clear guidance as to when to apply a given criterion. This has led to confusion in their appli-
cation, and contracting issues between government oversight and contractor analysts. Alternatively
the methodology chosen by ADS-33-PRF, where one criterion that can be broadly applied, limits
confusion and ensures a lack of ambiguity in application.
All the criteria reviewed in MIL-HDBK-1979 model the human pilot as a transfer function that
augments the ight of the vehicle. These criteria then place bounds on the aircrafts response based
on the human operators limitations. The question then remains; if all of these criteria were chosen
to dene the human aircraft interface, are any of them suitable for use in an autonomous UAV?
Several of these criteria also attempt to predict a PIO. Can a PIO occur without a pilot?
In the case of large scale UAVs that would require integration into the NAS (National Air Space)
following manned requirements for unmanned aircraft does make some intuitive sense, since the
UAV would be required to interface with standards currently set for piloted aircraft. The same argu-
ment can be made for UAVs that are required to take off and land from ships in a similar manner to
piloted aircraft, or make use of standard runways. The target of developing a UAV ying qualities
criterion should then be for aircraft that are not intended to interface directly with NAS operational
requirements. Aircraft that are capable of operating in the NAS could still fall under UAV ying
qualities criteria if they are required to operate in a manner that would exceed manned ying quali-
ties boundaries.
96
As chosen in ADS-33-PRF, the bandwidth criterion is recommended to serve as the basis for the
UAV ying qualities criterion. Choosing one criterion would aid in reducing the current confusion
of knowing when to apply a given criterion as is currently the case in MIL-HDBK-1797. The band-
width criterion does incorporate several of the ying qualities criteria found in MIL-HDBK-1797,
and has the added benet of predicting PIO,
120
or in the case of an UAV a limit cycle oscillation
driven by a control system. Another advantage to the bandwidth criterion is its independence from
vehicle response type, whether it be conventional, rate command attitude hold, or attitude command
attitude hold. In the case of an UAV the response types may become even more varied, and this
criterion would allow for those variations. The criteria also lends itself to application outside of the
short period mode, giving it further reason to be used in application to UAVs where modal frequen-
cies may not necessarily be easily separated.
The pilot model of the bandwidth criterion is no longer applicable to UAVs. The boundaries set
by that criterion are based on a human operators ability to control the aircraft. The pilot model
would need to be replaced by a sensor model, where the ultimate goal is no longer to y the aircraft
to a given task, but to keep a sensor pointed at a given position. The sensors dynamics now become
a major part of the overall systems ability to perform a MTE, and the sensor/airframe integration
becomes the key piece to ensure a MTE is successfully accomplished. In the event that the sensor
dynamics can be modeled as a simple gain and time delay a model similar to the one used in Fig-
ure 6.7 could be potentially used. The criterion would then require that the total system bandwidth
be greater than a bandwidth specied to complete a given task. This would allow the total system
bandwidth to be large enough to achieve the goal of keeping the sensor pointed at the desired loca-
tion.
The ride quality index found in SAE94900 also holds promise for use in integrating the UAV
and sensor package into a complete system. The ride quality index as currently constituted is used
97
7.2 Forming a New Criterion
to measure the fatigue of a human subject in an aircraft when exposed to long term vibration. By
calculating the transferred inputs from the turbulence to the place of the aircraft where the sensor
is located, vibration input to the sensor can be estimated. These vibration inputs can the be used
to determine if sensor stabilization is required for a given airframe, or if current sensor vibration
requirements are adequate. Incorporating this criterion into UAV development would aid in the
sensor/UAV system integration to ensure proper operation during encountered turbulence.
A new criterion must be simple enough to use, effective enough to make worth using, and familiar
enough to users so that there is a feeling of intuitive comfort in using it. When piloted ying
qualities criteria were reviewed for possible use in UAVs it became apparent that no one criteria
was going to be able to be used without modication. Linear systems theory was chosen as an
analysis means in creating a UAV criterion in order to align with currently accepted piloted ying
qualities criteria. The bandwidth criterion was chosen as a starting point while also paying attention
to suggested modications to current standards in order to provide an updated and relevant method
for evaluating UAVs.
7.2.1 Frameworks
The approach taken to application of the criteria is similar to that used in ADS-33-PRF.
106
While
this standard was written for rotorcraft, the systems engineering approach it takes to requirements
specication is more attractive than MIL-STD-1797B.
126
The basic precept used in ADS-33-PRF
is the denition of a mission task element (MTE), and then the application of analysis to the MTE.
By breaking down a vehicles overall mission into specic elements, a basic list of tasks can be
dened independent of the total mission of the vehicle. This allows vehicles with varying missions,
but overlapping mission task elements to be directly compared. The MTE approach also allows a
total mission to be built from a set of standard MTEs. Each MTE would then be given a rating on
aggressiveness and precision, ultimately allowing a chart such as the one in Figure 7.3 to be used.
98
This approach as seen in Figure 7.2 also allows ight test to replace modeling analysis tests where
modeling data for a UAV may not be obtainable. Further, ight test can be used to verify analytical
results where a combination of modeling and ight test is practical.
The use of linear systems theory has traditionally been very attractive in creating a ying qualities
criterion. It is the primary form of analysis used in modern ying qualities analysis, and provides a
familiar platform for continued analysis of vehicles. Linear systems theory also allows abstraction
of more complex systems and forms a basic, simple framework for complex systems to be modeled.
While the simplication process of taking more complex systems and representing them in a linear
fashion may introduce modeling inaccuracies, careful application of this technique using LOES has
been used for piloted aircraft. Linear systems presents analytical tools that can be readily used that
are not available if a nonlinear approach is taken. The goal of creating a new criterion is to make a
system that is as simple as possible, but still representative of most implementations of sensors and
aircraft, much in the same way that the bandwidth criterion did this for piloted aircraft with feed-
back control systems. By using linear systems, a block diagram approach can be taken to visually
represent the more complex systems that are being analyzed.
7.2.2 Classical Bandwidth Compared to Flying Qualities Bandwidth
The ying qualities criterion for bandwidth uses a denition for bandwidth that is nonstandard out-
side of the world of ying qualities. The methodology for nding the ying qualities bandwidth
frequency is as follows, and references Figure 7.5. The phase bandwidth frequency (
bw
phase
) is
the frequency at which the phase is 45
above the 180
crossing on the phase plot. The gain band-

width frequency (
bw
gain
) is the frequency at which the gain is 6 dB above the value found at the
180
crossing frequency (
180
). The bandwidth frequency
bw
is then the lesser of either
bw
gain
or
bw
phase
. If
bw
=
bw
gain
the system is said to be gain margin limited, and if
bw
=
bw
phase
the system is said to be phase margin limited. The classical denition for bandwidth is the rst
99
frequency where the gain drops below 70.79 percent (-3 dB) of its DC value.
91
Both denitions for
bandwidth are shown in Figure 7.5, and it is apparent that the values for bandwidths can be quite
different.
60
40
20
0
20
40
M
a
g
n
i
t
u
d
e

(
d
B
)
10
2
10
1
10
0
10
1
10
2
270
225
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bandwidth Definition
Frequency (rad/sec)
6dB
bw gain
45 deg
bw phase
180
Classical
bw
-3
FQ Bandwidth
Classical Bandwidth
Figure 7.5: Denition of bandwidth frequency,
BW
93
compared to the classical bandwidth
denition.
The question then arises as to which bandwidth denition to use for dening UAV ying qualities
denition. Sensor manufacturers will use a classical denition of bandwidth when specifying their
capabilities, whereas aircraft have traditionally been specied using the ying qualities bandwidth
denition. Both denitions have their advantages and disadvantages for use in specifying a criterion
for an integrated aircraft sensor system. Using ying qualities bandwidth UAVs could be compared
not only against other UAVs, but also with piloted aircraft. The ying qualities bandwidth is also
a more familiar quantity among engineers used to analyzing vehicle ying qualities. The ying
qualities bandwidth also has a level of robustness built into its denition, since it requires 6 dB of
100
gain margin and 45 degrees of phase margin. The use of the classical denition for bandwidth also
has its own merits. It yields the most optimistic measure of performance, which would be popular
among vendors of systems. It also is the denition that sensor manufacturers will most likely use to
describe their sensors, making a comparison between aircraft and sensor bandwidths much easier
and intuitive. The ying qualities bandwidth was also dened with the preferences of the human
pilot in mind for safely controlling a vehicle. With a human no longer part of the system in an
autonomous UAV, this driving factor is less important.
Another factor in using ying qualities bandwidth for piloted systems is that most piloted systems
have rapid phase roll offs which can include abrupt loss in stability margin. The rapid phase roll off
is often quantied as phase delay, which is similar to equivalent system time delay.
127
Choosing a
ying qualities bandwidth ensures that stability margin is maintained, and that abrupt phase roll off
is avoided. In UAV application, the shape of the Bode phase plot may or may not be the same as
a piloted aircraft. Only in the case that the phase plot is similar between UAV and piloted aircraft
would ying qualities bandwidth would provide insurance from encountering the phase cliff gener-
ally found in piloted systems.
The lack of phase and gain margin accounting in the classical bandwidth denition would create
an uneven playing eld for all UAVs to be measured against piloted aircraft. The issue of com-
parison between different UAVs is a bit more subtle. Robustness to different wind requirements
are specied when a new vehicle is contracted. These requirements are often not similar between
different platforms. The use of traditional bandwidth to compare UAVs would only allow a com-
parison of maximum performance of the platform without a measure of its robustness. Ultimately
the question must be answered: is this criterion going to be used to compare UAVs against piloted
aircraft, or just UAVs to their sensor payloads to judge integrated system mission performance?
101
7.3 Proposed Criteria
7.2.3 Accommodating MIMO Analysis
The use of linear systems analysis for piloted aircraft has always assumed a SISO (single input, sin-
gle output) system. This key assumption allows basic transfer function analysis to occur, and gives
easy to interpret results when using Bode, Nyquist, and root locus analysis. MIMO (multiple input
multiple output) analysis complex has fewer established tools and analysis methods. An integrated
UAV and sensor package is a MIMO system. The commands to the aircraft and the sensor create
multiple inputs, even if there is a single output for the system. A single output case might be a
non-zoom camera whose gimbal can only move in one direction. A camera with a zoom, or a multi
axis gimbaled camera would then be a multi-output system. A technique to simplify this MIMO
system to a SISO system can simplify analysis and cast the criterion in a form that is familiar to the
aviation community. Where it is not possible to do SISO analysis, a relevant yet simple metric for
MIMO analysis can provide an option for a criterion.
This section proposes two methods in forming a criterion for UAV longitudinal ying qualities anal-
ysis. In both criteria, mission performance is used to determine the ying qualities of the aircraft.
By using the MTE concept the ight safety segment of ying qualities must be covered by the air-
worthiness certication external to these criteria. The use of MTEs in piloted aircraft has received
some criticism in prior application because they do not account for the operation of the aircraft be-
tween mission tasks. While this is a shortcoming to the MTE method, a holistic approach would be
much more complex. UAVs of different sizes and roles often overlap mission tasks, but they rarely
overlap operationally outside of their direct use to accomplish a mission objective. An example
of this is the MQ-9 Reaper and the ScanEagle which are both used for surveillance, but have very
different ight envelopes and takeoff/recovery systems.
The goal of accomplishing a mission objective was kept rmly in view during the derivation of
102
these criteria. The integrated aircraft and sensor system must perform together to achieve a desired
goal to achieve mission success. With this vision the aircraft now becomes a platform for sensor op-
eration. The entire system must be measured rst for the sensors capabilities, and then second for
the platform stability the sensor requires to perform its mission. In this light, the aircraft becomes a
stabilizing platform for the sensor, and the driving question for the aircrafts design is: how stable
must the aircraft be to allow the integrated system to perform its mission?
Two different candidate criteria are proposed below. The rst is a simplied system that only
considers the short period of the aircraft, and the sensor dynamics. The goal of this simplied pitch
criterion is to reduce a MIMO system into a SISO system where traditional analysis tools can be
applied. The second candidate criterion embraces the MIMO concept and accounts for turbulence
and the coupling between phugoid and short period modes than occur during ight. It departs
from traditional criteria by taking an energy methods approach to dene a metric to predict mission
success.
7.3.1 A Simplied Pitch Criterion
This criterion was patterned after the approach taken with the bandwidth criterion where a pilot
was used to close a control loop around the short period mode of the aircraft. In this case a control
system is used instead of a pilot to stabilize and control the short period. This system includes
an added piece not normally seen in a ying qualities criterion, namely the sensor system model.
Using a block diagram modeling approach the goal is to reduce a MIMO system to a SISO system
for analysis. The reduced block diagram can then be used to compare relative bandwidths between
the aircrafts short period response and the sensors pitch command and response.
7.3.1.1 Modeling
The model created to approximate the sensor and aircraft system can be found in Figure 7.6, where
a/cc
is the aircraft pitch attitude command,
cs
is the pitch command to the sensor, and the output
103
is the systems total target output (pitch attitude towards the desired target,
Target
). A diagram
of sign conventions used to dene the different angles in this system can be found in Figure 7.9.
The transfer function G
SP
is the aircraft short period and the transfer function G
sens
is the sensor
transfer function, both dened in Figure 7.7(a). The G
SP
and G
sens
denitions are made here
as nominal denitions, although they could be changed to represent a more complex sensor model.
Using a well known identity, closed loop systems can be represented as a transfer function as dened
in Figure 7.7(b). Using the closed loop denition in Figure 7.7(b) to reduce Figure 7.6a block
diagram can be formed as seen in Figure 7.8, where C
L SP
is the closed loop short period and
C
L sens
is the closed loop sensor model.
7.3.1.2 Algebraic Reduction
The block diagram in Figure 7.8 can be represented algebraically as:
Target
=
_
cs
a/cc
C
L SP
_
C
L sens
+
a/cc
C
L SP
(7.1)
which can be rearranged as:
Target
=
cs
C
L sens
ac/c
C
L SP
(C
L sens
1) (7.2)
Using this form of the equation several observations can be made. First, as C
L sens
1,
Target

cs
. As the closed loop sensor system approaches an ideal system the aircraft dynamics will be
overcome, and the sensor will always track the desired target, no matter the aircraft dynamics.
Second, as C
L sens
0, or the sensor becomes stationary, the
Target

a/cc
C
L SP
, or the
dynamic response of the aircraft. If
a/cc
= 0, then
Target
will be only dened by the sensor
dynamics since the aircraft dynamics will be commanded to be zero. Finally, two special cases that
can be used for analysis can be considered. The rst case is when
cs
= 0, the system reduces to
the SISO system:
Target
=
a/cc
C
L SP
(C
L sens
1) (7.3)
104
_
+
G
SP
G
sens
cs
_
+
-
+
+
+
Target
Figure 7.6: Model used to approximate sensor and aircraft system.
G
SP =
K
SP
e
s
sens
1
sA
sens
+ 1
K
sens
G
sens =
(a) Bandwidth longitudinal control block diagram.
93
G
_
+
CL
=
=
G
G+ 1
(b) Closed loop system transfer function.
Figure 7.7: Transfer function denitions used to dene aircraft and sensor system
C
L_SP
C
L_sens
cs
-
+
+
+
Target
Figure 7.8: Reduced block diagram of sensor and aircraft system.
105
ac
target
sensor
+
-
-
Inertial Horizontal
B
o
d
y
A
x
is
S
e
n
s
o
r

P
o
in
t
in
g

V
e
c
t
o
r
Figure 7.9: Angle descriptions for sensor and aircraft system.
which still involves both the aircraft and sensor dynamic responses. This allows the MIMO system
to be analyzed using SISO techniques for a region where
cs
is zero, or near zero. It is important to
remember that this analysis is being done in a linear fashion, and the target command is a command
away from a steady state condition, and not necessarily from a true zero angle position. The second
case that can be used for analysis is where both
cs
and
a/cc
are nonzero. While this is not a SISO
case there are several useful relationships from SISO analysis that can still be used.
7.3.1.3 SISO Analysis Case
With the sensor commanded value set to zero in Equation (7.2) to form the SISO case in Equa-
tion (7.3),
Target
should approach zero as well. To overcome the aircraft dynamics C
L SP
the
sensor dynamics C
L sens
must serve to minimize the output. Specically, as C
L sens
1 the out-
put approaches zero. The relationship between C
L SP
and C
L sens
can be used to pick the proper
sensor for a given aircraft by inspecting the Bode plot generated from the system in Equation (7.3).
106
By nding the peak gain value, and using the relationship for a decibel
dB = 20 log
10
_
output
input
_
(7.4)
the sensors ability to reject the aircrafts motion can be analyzed. This becomes a task of specify-
ing the ratio of output to input, where the output is the sensors pointing error and the input is the
disturbance imparted to the aircraft.
As an example of using the SISO case to design a sensor and aircraft system, models for the
Virginia Tech SPAARO UAV
128, 129
and a CloudCap
R
TASE gimbaled camera are used to popu-
late Equation (7.3). A Bode plot can be created for this system as seen in Figure 7.10. Using
Equation (7.4) the region of the magnitude peak where the magnitude exceeds the desired dynamic
response can be dened. This peak region is dened as the Peak of Disappointment. The region
to the right of the peak is the frequency region where the aircraft can not track the desired aircraft
command. The boundary line can be chosen based on the aircraft short period response. The area
to the left of the peak of disappointment is where the sensor can null the aircraft dynamics. The
right boundary of this region is dened by the sensor dynamics. By choosing the right sensor and
aircraft system the Peak of Disappointment can be shaped such that it does not impact the total
systems mission performance. As an example, a time history of the total system can be found in
Figure 7.11, and it shows a two degree oscillation of the sensor pointing on the target. Based on
mission performance requirements, a decision could be made if this is within acceptable operational
limits. As an additional note, Bode plots of the aircraft short period and the sensor can be found in
Figure 7.12. The aircraft in this case has a similar phase relationship to the sensor, and the primary
inuence of the integrated system performance is the gain relationship.
Using data for a Cessna 172
130
and the same CloudCap
R
TASE gimbaled camera a time history
of the output of Equation (7.3) can be generated as found in Figure 7.13. The Bode representation
of this system can be found in Figure 7.14. The input to the time history was chosen to be at the
107
30
25
20
15
10
5
0
5
10
M
a
g
n
i
t
u
d
e

(
d
B
)
10
1
10
0
10
1
10
2
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
30
25
20
15
10
5
0
5
10
M
a
g
n
i
t
u
d
e

(
d
B
)
10
1
10
0
10
1
10
2222
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
Aircraft
cant track
command
Aircraft tracks
command &
sensor nulls
aircraft dynamics
Peak
of Dis-
appointment
Choice of
sensor
Choice of
aircraft
Figure 7.10: Example of analysis of Equation (7.3).
0 2 4 6 8 10 12 14 16 18 20
10
8
6
4
2
0
2
4
6
8
10
time (s)
P
i
t
c
h

A
t
t
i
t
u
d
e
Input
TARGET Output
Figure 7.11: Time history of SPAARO short period model and sensor using TARGET SISO
model.
108
100
80
60
40
20
0
20
M
a
g
n
i
t
u
d
e

(
d
B
)
10
1
10
0
10
1
10
2
10
3
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
CLSP
CLSENS
Figure 7.12: Bode plot of individual systems of the SPAARO short period and sensor model.
gain peak frequency found in Figure 7.14(a), the limiting case of this system. A Bode plot of the
closed loop transfer functions of both the sensor and the aircraft can be found in Figure 7.14(b).
Comparing the two systems in Figure 7.14 can be used as an aid in choosing the proper sensor and
aircraft combination to achieve the desired performance in Figure 7.14(a) and Figure 7.13. In the
case of piloted aircraft, and especially larger aircraft, the short period Bode plot will reveal that in
most cases the vehicle is phase limited, just as in the Cessna 172 case. This is due to the vehicles
high moment of inertia in comparison to smaller UAVs. Phase as well as the gain must be accounted
when inspecting the individual systems of the sensor and the aircraft short period.
One way the sensor bandwidth can be changed is by changing the system gain on the sensor.
Reducing the sensor bandwidth will reduce overall system performance. Figure 7.15 shows a sensor
gain reduction of 1/6 on the sensor model with the Cessna 172 aircraft model. The corresponding
Bode plots can be found in Figure 7.16. A sensor gain reduction of 1/3 using the same Cessna 172
109
0 5 10 15 20 25 30 35 40
10
8
6
4
2
0
2
4
6
8
10
time (s)
P
i
t
c
h

A
t
t
i
t
u
d
e
Input
TARGET output
Figure 7.13: Time history of Cessna 172 short period model and sensor using TARGET SISO
model.
model can be found in Figure 7.17 with corresponding Bode plots in Figure 7.18. By changing the
sensors bandwidth alone, the ability of the sensor to reject the aircrafts dynamics is dramatically
effected. From these plots it can be inferred that the bandwidth of the sensor should extend beyond
the peak gain frequency of the aircrafts short period in order to assure the sensor will be able to
stabilize the vehicle dynamics.
7.3.1.4 MIMO Analysis Case
The MIMO case must also be analyzed, specically for the case where the sensor is commanded
to move to a new target location while still rejecting aircraft motion. The output of the MIMO
case in Equation (7.2) is the commanded Target pitch attitude, while the inputs are the commanded
sensor pitch angle and the commanded aircraft pitch attitude. The MIMO case can be consid-
ered to be the superposition of two SISO cases, rst the term used in the previous SISO analysis,
a/cc
C
L SP
(C
L sens
1), and second
cs
C
L sens
. The rst case serves as a penalty on the sensor
110
50
40
30
20
10
0
M
a
g
n
it
u
d
e

(
d
B
)
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
(a) Bode plot of SISO analysis using Cessna 172 short
period and sensor model.
60
50
40
30
20
10
0
10
M
a
g
n
it
u
d
e

(
d
B
)
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
CLSP
CLCAM
(b) Bode plot of individual systems of the Cessna 172
short period and sensor model.
Figure 7.14: Frequency analysis of SISO system for aircraft and sensor dynamics.
0 5 10 15 20 25 30 35 40
15
10
5
0
5
10
15
P
i
t
c
h

A
t
t
i
t
u
d
e
time (s)
Input
TARGET Output
model with 1/6 system gain.
111
60
50
40
30
20
10
0
10
20
M
a
g
n
it
u
d
e

(
d
B
)
10
3
10
2
10
1
10
0
10
1
10
2
135
90
45
0
45
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
period and sensor model with 1/6 system gain.
60
50
40
30
20
10
0
10
20
M
a
g
n
it
u
d
e

(
d
B
)
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
CLSP
CLSENSE
short period and sensor model with 1/6 system gain.
0 5 10 15 20 25 30 35 40
10
8
6
4
2
0
2
4
6
8
10
P
i
t
c
h

A
t
t
i
t
u
d
e
time (s)
Input
TARGET Output
model with 1/3 system gain.
112
60
50
40
30
20
10
0
10
20
M
a
g
n
it
u
d
e

(
d
B
)
10
3
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
period and sensor model with 1/3 system gain.
60
50
40
30
20
10
0
10
20
M
a
g
n
it
u
d
e

(
d
B
)
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
CLSP
CLSENSE
short period and sensor model with 1/3 system gain.
dynamics in achieving the desired outcome. This penalty can be analyzed as in the SISO case to nd
the reduction in ability to track the desired sensor command due to the aircrafts dynamic response.
Just as before, the sensors bandwidth must be larger than the aircrafts peak gain frequency in order
to counter the dynamics of the aircraft. The second case
cs
C
L sens
determines the sensors ability
to track a given target without aircraft dynamics. The superposition of these two will then give the
integrated vehicles performance. Figure 7.19(a) is an example of the aircraft moving and the sensor
being commanded to remain stationary. Figure 7.19(b) is an example of the aircraft moving while
the sensor is being commanded to move as well.
7.3.2 A Criterion with Winds
While the simplied system shown above may prove to be useful in a large subset of UAV maneu-
vers, it is not all inclusive. The destabilizing effect that wind has on a vehicle must include both
phugoid and short period modes. In order to account for both modes and the wind inputs, as well as
vehicle and sensor commands, a SISO system is no longer practical for consideration. An approach
must now be used to characterize the integrated aircraft and sensor system. It is tempting to include
all inputs and outputs, and to in effect overly complicate the system. The key to nding a criterion
113
0 1 2 3 4 5 6 7 8 9 10
10
8
6
4
2
0
2
4
6
8
10
time (s)
T
h
e
t
a

(
d
e
g
r
e
e
s
)
(a) Frequency interaction between aircraft and sensor
with zero command to sensor. The solid lines are re-
sponses, dashed lines are commands, and the large am-
plitude response is the aircraft, and the small amplitude
response is the sensor.
0 1 2 3 4 5 6 7 8 9 10
10
8
6
4
2
0
2
4
6
8
10
time (s)
T
h
e
t
a

(
d
e
g
r
e
e
s
)
(b) Frequency interaction between aircraft and sensor
with frequency command to sensor. The solid lines are
responses, dashed lines are commands, and the large
amplitude response is the aircraft, and the small ampli-
tude response is the sensor.
Figure 7.19: Interaction between sensor and aircraft
of this sort is to create a minimal set of inputs and outputs that can be used to create a meaningful
metric for analysis. Special attention must be paid to determine the correct input and output planes
for analysis as well as their content.
7.3.2.1 Wind Effects
Wind turbulence models treat an aircraft as a point mass, and provide linear velocity perturbations
to the vehicles trajectory. Whether choosing the Dryden
131
or Von Karman
132
turbulence models,
their basic effect on an aircraft is the same.
133
The linear displacement of the aircraft will cause a
pointing angle error for the sensor, and changes in angle of attack of the aircraft will cause changes
in the rotation (short period mode) of the aircraft as well. A graphical illustration of this system can
be found in Figure 7.20.
While turbulence effects the linear motion of the aircraft disturbing it as a point mass, the cou-
pling may occur between short period and longitudinal modes, causes a pitch response. Taking the
knowledge that turbulence imparts a point mass acceleration, turbulence can be treated mathemati-
114
Turbulence Type
Turbulence Intensity
Altitude
Inputs
u,v,w
p,q,r
Outputs
Steady
altitude
velocity
pitch attitude
Inputs
Energy
altitude
velocity
Orientation
pitch attitude
Outputs
Pitch attitude
Constant radius
Inputs
Target "in view"
Outputs
Figure 7.20: The inuence of wind turbulence on aircraft motion and sensor errors.
cally as quickly adding and taking away power from the vehicle. Using this knowledge a metric can
be constructed from the vehicle energy equations.
7.3.2.2 Energy Methods
The wind effects on the aircraft are taken as changes in aircraft energy state. As changes in energy
they can be dened as aircraft power, and will impact the aircrafts trajectory just as an aircraft
engines thrust change would impact a trajectory. The equations of focus for this metric then become
not the typical state equations, but the aircraft energy equations. The kinetic and potential energy
for an aircraft can be described as
E = mgh +
1
2
mV
2
(7.5)
115
where V is the total inertial velocity of the aircraft. Energy height E
h
or specic energy is energy
per unit weight, and can be described as :
E
h
= h +
1
2g
V
2
(7.6)
Specic power can then be represented as :
P
s
=

h +
1
g
V

V (7.7)
The wind turbulence models directly effect the altitude rate

h and inertial velocity V of the vehicle.
Only by adding rotational energy can the short period be added into the power equation. The total
specic energy with rotation can then be described as:
E
h
= h +
1
2g
V
2
+
Iyy
2mg
2
(7.8)
and the total specic power can now be described as:
P
s
=

h +
1
g
V

V +
Iyy
mg
(7.9)
7.3.2.3 Metrics
Past work
134
has shown that nding detailed data on UAVs can be very difcult. Often, mass
properties data such as pitch moment of inertia are not publicly available, or even known by the
manufacturer. This creates a large challenge in using P
s
in a criterion form. The solution is to use
when possible what is available, and modify P
s
to create a metric that can be used with available
data for UAVs. While rigorously using an energy or power equation for a criterion is not practical
due to a lack of information available for UAVs, the question then arises, can a metric be dened to
include rotational, potential, and kinetic energy or power without all the information present? The
rst step to answering this question is to dene the relevant, minimum set of information required
116
to quantify performance in winds. Using the MTE methodology a given task would have a level
of wind turbulence. Specically if the Dryden or Von Karman models were used the MTE would
include an altitude above the ground and a probability of occurrence. Given a wind turbulence level,
a change in allowable power must then be specied for a given objective. The MIMO outputs used
in this metric would then be a vector whose individual components consist of the effects of kinetic,
potential, and rotational energy. A factor can be placed in front of each vector component in order
to weight the relative importance of each component to the full vector. These factors would replace
the mass properties data that is lacking in UAV databases. The vectors magnitude would then be
a scalar quantity that can be used to determine the magnitude of the vector space. The MIMO
input for this metric would also be a vector whose components are dened by allowable error for
the sensor to track a given target, both in pitch attitude and range to target. The comparison of the
MIMO input to outputs can then be done through relative magnitudes of the MIMO input and output
vectors. This would give rise to the following equation:
=
|K
i1
err
, K
i2
range
err
|
_
_
_
_
K
o1
h, K
o2
V

V , K
o3

__
_
_
=
_
(K
i1

err
)
2
+ (K
i2
range
err
)
2
_
_
K
o1

h
_
2
+
_
K
o2
V

V
_
2
+ (K
o3
)
2
(7.10)
where the values of K are chosen to change the emphasis of the various components of the metric. It
should be noted that the values of K should be inuenced such that is a non-dimensional number,
so that comparison between aircraft can still be made. A bound is then a set of such as :
min

max
(7.11)
The values for
min
and
max
are determined by nding the maximum allowable magnitudes
for the input and output vectors for a given task and placing bounds on their ratio. is the relative
magnitude of sensor error to aircraft dynamic motion, where an ideal value of 0, meaning the
sensor could track its target no matter the aircraft motion. If the sensor error remains constant while
the aircraft motion increases, the value of reduces, noting the more difcult task of maintaining
117
7.4 Simulation Trials
20 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19
20
0
5
10
15
Time (s)
max
min
turbulence
no turbulence
Figure 7.21: An example of boundaries placed on for a pitch doublet MTE.
a constant error in larger aircraft excursions. If the sensor error increases while the aircraft motion
becomes constant, the value of would increase, showing poorer performance. The value of
can be monitored during a specic task dened in a MTE, and the boundaries on would then be
required to be met during the task performance. An example of the application of boundaries can
be seen in Figure 7.21, where the red region is the allowable value above zero for the MTE of a
pitch doublet. The green line is the MTE performed without turbulence, and resides in the center of
the allowable region, and the blue line is the MTE performed with turbulence.
7.4.1 Simulation Testing
In order to test the new criteria two simulation environments were chosen. A linear, longitudinal
aircraft simulation was created in Matlab
R
to primarily test and evaluate transfer functions. A full
118
nonlinear six degree of freedom simulation was also used based on NavAirs CASTLE simulation
environment.
7.4.2 Aircraft Models
Three aircraft were chosen as the basis for airframe models in CASTLE. The choice of aircraft
represents a broad range of UAVs taken from different classications based on Classes II, IV, and
V in Figure 5.8. The aircraft models used are based on the F-16 model found in Stevens and
Lewis
89, 135
as well as a Cessna 172 model from Roskam,
130
and the Virginia Tech SPAARO.
128, 129
In order to match the representative Reynolds numbers in Figure 5.8, each aircraft was trimmed
at a different operational point. These points can be found in Table 7.3. Longitudinal only models
were constructed for linear analysis in Matlab
R
, and full six degree of freedom airframe models
were created for each airframe in CASTLE.
Aircraft Altitude (ft, MSL) True Airspeed (Kts)
SPAARO 1000 ft 53 Kts
Cessna 172 10,000 ft 120 Kts
F-16 30,000 ft 500 Kts
Table 7.3: Aircraft operational points used for testing in simulation.
For the CASLTE simulation, each aircraft starts off trimmed, and then is commanded to perform a
pitch doublet at 2 seconds. The pitch doublet has a period of two seconds, and has a peak amplitude
of 10 degrees. The aircraft begins the simulation at inertial coordinates (A, }) = (0, 0), with units
of ft. During the simulation, the target is kept at a constant position of (A, }) = (1000, 1000) at
MSL. The sensor on the aircraft is commanded to track the target on the ground while the aircraft
ies over the target.
119
7.4.3 Sensor Model
The sensor model used for all four airframes was taken from experimental data gathered from a
CloudCap
R
TASE gimbaled camera. The TASE is modeled to only move in pitch and was charac-
terized on a hardware bench with the following transfer function:
cmd
=
G
sens
G
sens
+ 1
(7.12)
where G
sens
is dened as:
G
sens
=
1
s
1
K

3
0.1293s + 1
(7.13)
where K can be varied to reduce the system bandwidth as required to adjust sensor dynamics for
testing.
7.4.4 Testing the Simplied Pitch Criterion
The simplied pitch criterion can be tested by assuming that a MTE requires a commanded pitch
attitude doublet at rad/sec, or a period of 2 seconds while holding the sensor at a constant target.
The aircraft mentioned in Section 7.4.2 are used to evaluate this task along with the sensor model
in Section 7.4.3, with the gain K = 1, 2, 3, and 6. In Figure 7.22 the SPAARO airframe is used
along with the different sensor gains in order to create a Bode plot of Equation (7.3). This system
is excited at a frequency of 1 rad/sec and shown in Figure 7.23. This gure shows the variation
of responses when there is separation in the different gain plots. Figure 7.24 is the linear response
of the same system when excited at rad/sec, showing small changes in the gain response of the
different systems. This is a conrmation of the Bode plot gain in Figure 7.22. If the only task of
interest is at rad/sec, the best answer might be to nd a different sensor for the SPAARO, but if
this amount of sensor oscillation is acceptable, any gain set for the sensor would be acceptable. If a
wider range of frequency operation is desired, the choice then becomes the K = 1 sensor, at the 1
rad/sec excitation case.
120
60
50
40
30
20
10
0
10
M
a
g
n
i
t
u
d
e

(
d
B
)
Bode Diagram
Frequency (rad/sec)
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
K=1
K=2
K=3
K=6
Figure 7.22: Bode comparison of sensor models with SPAARO aircraft dynamics.
121
0 1 2 3 4 5 6 7 8 9 10
10
8
6
4
2
0
2
4
6
8
10
time (s)

(
d
e
g
)
SPAARO Airframe with Various Sensors
K=1
K=2
K=3
K=6
cmd
Figure 7.23: Linear simulation comparison of sensor models with SPAARO aircraft dynamics,
input at 1 rad/sec.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
10
8
6
4
2
0
2
4
6
8
10
time (s)

(
d
e
g
)
SPAARO Airframe with Various Sensors
K=1
K=2
K=3
K=6
cmd
Figure 7.24: Linear simulation comparison of sensor models with SPAARO aircraft dynamics,
input at rad/sec.
122
Continuing this example, the next step is to explore the use of other aircraft with the K = 1
sensor. Figure 7.29 is a Bode plot of Equation (7.3) with all the aircraft in this example using the
K = 1 sensor. The three aircraft show similar results, with the SPAARO model showing a smaller
magnitude at rad/sec, the Cessna 172 next larger, and the F-16 with the most gain. Figure 7.30
shows the linear simulation of the MTE with excitation at rad/sec, conrming that the SPAARO
has the least sensor movement.
Using the CASTLE simulation a pitch doublet was commanded to each vehicle. Each aircraft was
initially congured as described in Section 7.4.2. The simulation was run twice with each aircraft,
once with no pitch doublet to establish a baseline, and again with the pitch doublet. The SPAARO
results can be found in Figure 7.25, while the Cessna 172 results can be found in Figure 7.26, and
the F-16 results can be found in Figure 7.27. Taking the difference of the two traces in these plots
yields the sensor tracking error during the maneuver, and the error for each aircraft is plotted in
Figure 7.28. The SPAARO is shown to have the smallest tracking error, as also predicted in the
linear analysis case.
7.4.5 Testing the Wind Criterion
The simplied pitch criterion does predict sensor pitch attitude errors based on aircraft pitch attitude
displacement, but it can not account for the added sensor pitch motion associated with changes in
aircraft energy. It is primarily a concern when airspeed or altitude changes, especially if these result
from abrupt changes caused by turbulence that are too quick for the sensor to compensate. This
can become a signicant issue, especially if a camera sensor is being used, and the sensor is limited
to a certain desired level of vibration. The same simulation tests as before were repeated in CAS-
TLE with the Dryden turbulence model turned on, and with various levels of turbulence selected.
Figure 7.31 shows the calculation for the SPAARO case with no turbulence, and the Figure 7.32
is the case for moderate turbulence. The spikes at two and four seconds represent the tracking
errors previously shown in the simplied pitch criterion test. The area of interest, however, is the
123
0 1 2 3 4 5 6 7 8 9 10
50
45
40
35
30
25
20
time (s)

(
d
e
g
)
SPAARO
no doublet test
doublet test
Figure 7.25: CASTLE 6-DOF simulation of SPAARO with sensor model gain K = 1.
0 1 2 3 4 5 6 7 8 9 10
110
105
100
95
90
85
80
time (s)

(
d
e
g
)
C172
no doublet test
doublet test
Figure 7.26: CASTLE 6-DOF simulation of Cessna 172 with sensor model gain K = 1.
124
0 1 2 3 4 5 6 7 8 9 10
110
105
100
95
90
85
80
time (s)

(
d
e
g
)
F16
no doublet test
doublet test
Figure 7.27: CASTLE 6-DOF simulation of F-16 with sensor model gain K = 1.
0 1 2 3 4 5 6 7 8 9 10
10
8
6
4
2
0
2
4
6
8
10
time (s)

(
d
e
g
)
6 DOF aircraft comparison
F16
C172
SPAARO
Figure 7.28: CASTLE 6-DOF simulation of all aircraft sensor pointing error.
125
10
2
10
1
10
0
10
1
10
2
180
135
90
45
0
45
90
P
h
a
s
e

(
d
e
g
)
Bode Diagram
Frequency (rad/sec)
60
50
40
30
20
10
0
10
M
a
g
n
i
t
u
d
e

(
d
B
)
SPAARO
C172
F16
Figure 7.29: Bode comparison of all aircraft with sensor model gain K = 1.
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
10
8
6
4
2
0
2
4
6
8
10
time (s)

(
d
e
g
)
Various Aircraft with Sensor K=1
SPAARO
C172
F16
cmd
Figure 7.30: Linear simulation comparison of all aircraft with sensor model gain K = 1, input
excitation of rad/sec.
126
straight and level ight after the doublet, where the turbulent case shows frequency content in the
calculation. Figure 7.33 and Figure 7.34 reect the same no turbulence and moderate turbulence
cases as those performed for the Cessna 172. This time the vehicle is moving faster and at a higher
altitude, and the increased turbulence from being at altitude manifests in larger spikes and higher
frequency of the spikes. This trend continues in the F-16 cases and can be seen in Figure 7.35 and
Figure 7.36.
The increase in turbulence results in higher frequency content in as well as higher trending
spikes. These values represent added vibration to the payload platform, which is the aircraft. The
sensor would possibly have to overcome these vibrations to operate effectively. The boundaries on
the levels, and possibly the frequency of oscillation would need to be established for a given mis-
sion or sensor tolerance. These vibrations are not quantied in the simplied pitch criterion because
it does not take into account energy changes in the aircraft. Using this criterion with the simple
pitch criterion, a full picture of the operational effectiveness of the sensor and aircraft platform can
be achieved.
127
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
35
40
time (s)
no doublet
doublet with no turbulence
Figure 7.31: calculation using CASTLE 6-DOF simulation of SPAARO with no turbulence.
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
35
40
time (s)
no doublet
doublet with moderate turbulence
Figure 7.32: calculation using CASTLE 6-DOF simulation of SPAARO with moderate tur-
bulence.
128
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
35
40
time (s)
no doublet
Figure 7.33: calculation using CASTLE6-DOF simulation of Cessna 172 with no turbulence.
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
35
40
time (s)
no doublet
Figure 7.34: calculation using CASTLE 6-DOF simulation of Cessna 172 with moderate
turbulence.
129
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
35
40
time (s)
no doublet
Figure 7.35: calculation using CASTLE 6-DOF simulation of F-16 with no turbulence.
0 2 4 6 8 10 12 14 16 18 20
0
5
10
15
20
25
30
35
40
time (s)
no doublet
Figure 7.36: calculation using CASTLE 6-DOF simulation of F-16 with moderate turbu-
lence.
130
7.5 Conclusion
7.5 Conclusion
The framework proposed in this chapter is inspired by piloted ying qualities criteria that were
developed to predict the capabilities of human/aircraft interaction to perform a given task. In the
case of a UAV where the pilot no longer present, a new criterion must be formed. While the new
criterion can be inspired by previous piloted criteria, the new criterion must place as its rst ob-
jective the ability to predict mission performance. The simplied pitch criterion and criterion with
winds proposed in this chapter pay homage to piloted criteria, but are not limited by them. The
overall framework proposed is a departure from those listed in MIL-STD-1797 and MIL-F-8785C,
and proposes following the systems approach outlined in ADS-33-PRF, providing a more targeted
and systematic approach to the use of the new criteria.
Open questions relating to this research still remain. The rst open issue is the need to dene a
set of MTEs that can be used to establish boundaries for the proposed criterion. These mission tasks
should be dened by operational considerations with the goal of creating a database of maneuvers
that are representative of UAV mission segments. From these MTEs a guideline for allowable error
in sensor pitch tracking could be established for each task. The MTEs would also dene operational
environments and give guidance to wind and turbulence conditions for a given task. This informa-
tion could then be used to dene parameter boundaries for the criterion with winds.
Flight and simulation testing of a larger set of UAVs would also be helpful in further evaluation
of the proposed criterion. Testing should be conducted after a sample set of MTEs is dened to
ensure relevance to future UAV mission operation. With a larger set of aircraft tested, a database
such as the Neal-Smith database can be developed for UAVs aiding in further criterion development
and evaluation.
The proposed simplied pitch criterion takes a complex system and abstracts it to a level where
simple analysis can be performed to estimate mission performance of an aircraft and sensor system.
131
7.5 Conclusion
By using a linear systems approach the system can be analyzed using standard techniques famil-
iar to those currently in the ying qualities community. Equation (7.3) gives the designer of the
integrated system the ability to estimate the interaction between a sensor and the aircraft, and then
choose the right combination of the two systems that will be operating at a know excitation fre-
quency. As long as both the sensor dynamics and the aircraft dynamics can be modeled as a transfer
function the system can be analyzed in this manner.
The proposed criterion with winds is a departure from standard ying qualities criteria in that it
uses a metric based on energy methods to quantify the aircrafts motion in both the short period
and phugoid modes. This allows for disturbances such as wind or turbulence to be factored into the
criterion. With a sensor pointing at a target at a distance, the vehicles center of gravity location as
well as its pitch attitude are signicant in determining the pointing vector for the sensor. With dis-
turbances such as turbulence, sharp disturbances are imparted to the aircraft creating a vibrational
environment that may have too high a frequency for the sensor to compensate. This criterion exposes
that frequency content and allows for the prediction of high frequency disturbances to be considered.
Evaluation of UAVs and sensor systems during ight test will require a rating scale that can func-
tion in a similar manner to the Cooper Harper rating scale. While the questions lling the decision
tree may need to be changed, the basic decision tree should be left in tact. The Cooper Harper rating
scale was developed during the ight testing of aircraft for ying qualities evaluation. A new rating
scale for UAVs should be developed in a similar manner. During ight testing unknown issues often
occur, and the creation of this new scale could be done during early MTE development ight testing.
Ultimately the adoption of new ying qualities standards for UAVs will require both the United
States government as well as industry approval. There will most likely be several different criteria
proposed before one is adopted for wide use. By using the MTE approach, the criteria proposed here
presents a targeted approach to UAV ying qualities. While this does not present a holistic approach
132
7.5 Conclusion
to UAV longitudinal ying qualities, and it does provide a starting point for future development. It
is hoped that this work will spurn on further conversation in the ying qualities community and
provide a starting point for others in the future development of UAV ying qualities criteria.
133
Chapter 8
Nonlinear Flying Qualities
A
DAPTIVE control has been considered as an attractive control architecture for future UAVs
and piloted aircraft.
59
The use of nonlinear controllers that can adapt offers an alternative to
classical controls to aircraft designers. Nonlinear controllers can be adapted to multiple airframes
with minimal changes, allowing a generic control architecture to be developed and then reused for
different vehicles. An example of this concept is the Honeywell MACH controller that was success-
fully ight tested on the X-35 and X-38 vehicles.
59, 136, 137
Multiple applications from this concept
have stemmed from research originally sponsored by the Air Force Research Laboratory.
138
The
ying qualities evaluation of nonlinear controllers has offered large challenges to the ying qualities
community, for both piloted and unpiloted aircraft.
59
This chapter describes a study to determine
if an adaptive controller can recover level 1 handling qualities on a piloted, simulated aircraft in a
degraded state. Using a piloted, motion-based ight simulator an evaluation of ying qualities was
conducted to collect pilots opinions and ratings using the Cooper Harper
23
rating scale. A sim-
plied F-16 model
89
was used as the plant model to evaluate the controller. Four different aircraft
models of varying handling qualities levels were used as a test data-set to determine if all four could
be augmented via the adaptive controller to level 1 handling qualities. Two experienced test pilots
were used to judge the aircraft ying qualities using the Cooper Harper Rating Scale, and then an
attempt to correlate those results to a handling qualities rating was made. While simulator evalua-
134
8.1 Introduction
tions are limited in comparison to full ight test, it is a safe and cost effective way to conduct initial
research in a realistic environment. At the end of this chapter, open issues associated with nonlinear
control and prediction of their handling qualities are discussed.
8.1 Introduction
A longitudinal study to evaluate the ying qualities of an adaptive controller was performed. The
adaptive controller was used to compensate for a poorly performing aircrafts handling characteris-
tics. Using a piloted, motion-based ight simulator an evaluation of ying qualities was conducted
to collect pilots opinions and ratings using the Cooper Harper
23
rating scale. A simplied F-16
model was used as the plant model to evaluate the controller. While simulator evaluations are lim-
ited in comparison to full ight test, it is a safe and cost effective way to conduct initial research in
a realistic environment.
8.2 Background
8.2.1 L
1
Adaptive Control Architectures and Their Verication and Validation
Adaptive controllers have been successfully ight tested for recovering nominal performance in the
presence of modeling and environmental uncertainties.
60, 139
While the stability of these schemes
follows from Lyapunovs direct method, the robustness analysis (stability margins) are being de-
termined from Monte-Carlo simulations. It has been observed that increasing the adaptation gain
for better performance leads to high frequency oscillations in the control signal and reduces the
systems tolerance to the time-delay in the control and the sensor channels.The Theory of Fast and
Robust Adaptation with its L
1
adaptive control architectures has addressed these limitations for a
wide class of systems,.
63, 64, 6668, 140146
It allows for fast adaptation with guaranteed, bounded away
from zero time-delay margin. The speed of adaptation is limited by the hardware, specically the
135
8.2 Background
Adaptive Law
x
x
r
u
, ,
r k x u
g
T
+ +
) / 1 ( ) ( s s D =
u
) ( ) ( t k t u =
Control Law
) ( + + + = x u b x A x
m
&
)
( + + + = x u b x A x
m
&
predictor
system
x
~
Adaptive Law
x
x
r
u
, ,
r k x u
g
T
+ +
) / 1 ( ) ( s s D =
u
) ( ) ( t k t u =
Control Law
) ( + + + = x u b x A x
m
&
)
( + + + = x u b x A x
m
&
predictor
system
x
~
Figure 8.1: L
1
adaptive control architecture.
computer processor.
Originally introduced by Cao and Hovakimyan
141, 142
for the class of systems with known high-
frequency gain and constant unknown parameters, the underlying design paradigm of L
1
adaptive
control was later extended to systems with unknown high-frequency gain, time-varying unknown
parameters and disturbances without restricting the rate of variation.
64, 140
Cao et al.
64, 140
predict
the uniform performance bounds a priori for the systems input and output signals simultaneously,
and also quanties the stability margins, including the time-delay margin.
The underlying design paradigm of this new theory is the use of a low-pass lter in the feedback
path that appropriately attenuates high-frequencies in the control signal typical for fast adaptation
rates. Selection of the lters bandwidth follows from a small-gain type argument, which then leads
to guaranteed, bounded away from zero time-delay margin in the presence of fast adaptation. While
136
8.2 Background
the adaptation law is still derived from conventional gradient minimization, the error signal x(t)
used in that process is the one between the system state and its predictor, Fig. 8.1. As in con-
ventional Model Reference Adaptive Control (MRAC), the adaptive law retains its structure and
the upper bound on the norm of x(t) remains proportional to the rate of variation of uncertainties
and inversely proportional to the adaptation gain. However, with the low-pass lter in the feed-
back path the L
1
adaptive controller readily avails the opportunity to increase the adaptation rate
without leading the system into high-frequency oscillations, while the small-gain theorem invoked
for the stability proof leads to uniform performance bounds inversely proportional to the adaptive
gain, and guaranteed, bounded away from zero time-delay margin in the presence of fast adapta-
tion.
64, 66, 140
The L
1
adaptive control architecture and its variants have been veried and validated for various
systems: ight test of augmentation of an off-the-shelf autopilot for path following,
69, 147
autopi-
lot design and ight test for Micro Air Vehicles (MAV),
73
Tailless Unstable Military Aircraft,
76, 77
Aerial Refueling Autopilot design,
71, 72
exible aircraft ( Sensorcraft),
75
control of wing rock in the
presence of faults
70
and air-breathing hypersonic vehicles.
74
It is important to note that the L
1
adaptive control system is only intended to be active during a
mismatch with desired aircraft performance. This mismatch may occur rst as a transient and then
as an error from the desired state that may or may not be constant. The adaptive control systems
ability to drive this mismatch to zero will be key in determining the aircraft ying qualities while
the adaptive controller is active. It is important, however, to not only look at an initial transient, but
also at the long term response of the controller while active. For example if the adaptive controller
is requiring 90% of pitch effectors to keep the aircraft trimmed and stable, a high rate pitch task may
not be possible, and aircraft handling qualities would be affected purely by lack of control authority
or possibly actuator rate limiting, and not just the dynamic controller response. While total system
interaction by structural mode coupling and aeroservoelastic effects would be a major concern in
137
8.2 Background
aircraft implementation, it was not investigated in this study due to simulation complexity.
8.2.2 Aircraft Model and SAS
The aircraft model used for this study was a simplied F-16 model
148
that supports six degree of
freedom motion. The aircraft model is a clean F-16 (no external loads), and does not contain math
modelling for high lift devices (aps or slats), landing gear, or ground effects. The model does con-
tain an after-burning turbofan jet engine with appropriate throttle lags modeled. The control inputs
are
PLA
,
e
,
a
, and
r
.
PLA
command is the power lever angle command (or throttle position),
and its inputs range from 0 (ight idle) to 0.5 (full military thrust) to 1.0 (max afterburner), and is
referenced as idle, MIL or MAX thrust, respectively. The three aerodynamic effectors are elevators
(
e
), ailerons (
a
), and rudders (
a
) all with a range of +/- 25
o
. The elevators are primarily used for
pitch control, the ailerons for roll control, and the rudders for yaw control, with PLA being used as
engine control. The aircraft model does not include a sensor model, but relies on use of the actual
aircraft states. A simplied atmospheric model was used as well. The aircraft model places the
center of gravity location at 35% of the mean aerodynamic chord, resulting in an open loop unstable
short period. The aircraft motion is described by the states found in Table 8.1.
A simple control system was implemented to modify the short period characteristics of the F-16
model. A block diagram of the system is shown in Figure 8.2. By feeding back the pitch rate and
angle of attack the damping ratio and natural frequency of the short period was modied. This
modication occurs by manipulating gains K1 and K2. The result is an augmented plant that can
be easily changed to match Handling Qualities Ratings that are well documented.
5, 22
Four specic
cases that can be found in Table 8.2 were created at a trim condition of steady, level ight, 10,000
ft MSL, 360 KEAS (709 ft/sec or 0.65 Mach), and an angle of attack of 1 degree.
8.2.3 Facility
138
8.2 Background
State Index Name Description Units
x(1) V body axis total velocity ft/sec
x(2) angle of attack rads
x(3) sideslip rads
x(4) roll angle rads
x(5) pitch angle rads
x(6) yaw angle rads
x(7) p roll rate rads/sec
x(8) q pitch rate rads/sec
x(9) r yaw rate rads/sec
x(10) N horizontal travel, North ft
x(11) E horizontal travel, East ft
x(12) h altitude (MSL) ft
x(13) pow power, state used to nd thrust 0 - 100 %
Table 8.1: Reference of vehicle states.
-
+
F-16 Model
Stick Cmds
(delevc)
K1
K2
q
+
+
Figure 8.2: Augmented F-16 Plant.
Case Natural Frequency
n
Damping ratio K
1
K
2
1 4.63 rad/sec 0.704 0.29 1.25
2 2.47 rad/sec 0.448 0.0 0.444
3 4.0 rad/sec 0.2 0.05 1.25
4 unstable unstable 0.0 0.0
Table 8.2: Test Flight Cases
139
8.2 Background
8.2.3.1 Simulator
The simulator used for this study was delivered to the Aerospace and Ocean Engineering Depart-
ment of Virginia Tech on March 5th of 1996 from NAS Oceana in Virginia Beach. Originally an
A-6E Intruder Operational Flight Trainer (OFT), the simulator was declared in excess when the
Navy retired its A-6Es and replaced them with F/A-18s. The transfer was made possible with
the help and support of research sponsors at Naval Air Systems Command Headquarters and at the
Manned Flight Simulator branch of the Naval Air Warfare Center, Patuxent River, Maryland. A
diagram of the simulation system can be found in Figure 8.3. The left (pilots) seat of the trainer
cockpit represents the cockpit of a A-6E Intruder. The right seat has been modied to accommo-
date either an instructor or a ight test engineer with a computer driven CRT which can be custom
congured with instrumentation as desired. The simulation computer has been converted to a SGI
Origin 2000 computer. This allows the simulation of many different aircraft models, from a F-16 to
a Cessna 152 to a Boeing 737 to a F/A-18. The A-6E Intruder aircraft ight controls, instruments,
and systems, as well as its visual, aural, environmental, and motion sensations are combined with
the desired aircraft software model to create a realistic ight experience. The three window visual
display shows the surrounding terrain throughout take-off, maneuvers, and landing approach as a
function of the aircraft attitude, altitude, and speed. Motion cues are provided by a three-degree of
freedom cantilevered motion system. The simulator was originally procured by the Navy to provide
pilot and air crew training for carrier based takeoffs and landings. While heavily upgraded, it retains
its original capability.
149
8.2.3.2 IADS
Symvionics Inc. of Arcadia, CA has donated its ight test instrumentation software package called
IADS
R
to Virginia Tech for use in their ight simulation laboratory. IADS
R
is a real-time data
viewing tool that allows its user to view parameters from the aircraft simulation while the test is
occurring. IADS
R
is an industry standard tool that is used in ight test by NASA, the US Air
Force, and US Navy. IADS
R
also archives the data it displays for analysis purposes after the test.
140
8.2 Background
Figure 8.3: Diagram of Virginia Tech Flight Simulation System.
141
8.3 L
1
Controller Implementation
IADS
R
allows the user to customize the data displayed on the computer screen allowing the user to
create data screens that are tailored to each tests requirements. IADS
R
has served as an important
evaluation tool allowing realtime data visualization during and post ight test. Examples of IADS
R
displays used in this study can be found in Figure 8.4.
8.3 L
1
A high level diagram of the L
1
architecture implemented for this study can be found in Figure 8.5.
For design and analysis purposes the controller was rst implemented with a linear system aircraft
model in Matlab Simulink
R
. The controller was then ported to the nonlinear 6-DOF CASTLE
simulation. For both the linear and nonlinear aircraft model cases the same L
1
controller was used.
The augmentation scheme referenced in Table 8.2 was also used for both linear and nonlinear plants.
The combination of augmentation schemes in Table 8.2 and the open loop aircraft model were used
to create the Stabilized Plant block in Figure 8.5.
The primary interest of this study was the short period mode of the longitudinal motion. The
phugoid was left out of the L
1
design because it had no impact on the short period aircraft response.
Lateral-directional control was implemented in the L
1
controller for future use, but since no uncer-
tainty in the lateral or directional motion is incorporated in this study, the L
1
controller does not add
any lateral-directional augmentation to the aircraft.
For this study the L
1
controller was rst designed with a linear system approximation of the
aircraft equations of motion. The linear approximation was then replaced with the full nonlinear
6-DOF equations of motion. The general equations for aircraft dynamics can be presented in state
space form as:
x(t) = A
m
x(t) +b
_
u(t) +
T
(t)x(t) +(t)
_
(8.1)
142
8.3 L
1
(a)
(b)
Figure 8.4: Sample IADS
R
data screens used for the study.
143
8.3 L
1
+
+
Companion
Model
Stabilized
Plant
Low Pass Filter
C(s)
_
+
u Command
u total xhat
X
u adaptive
Adaptive
Law
error
Control
Law
hat
Figure 8.5: Block diagram of L
1
controller.
where A
m
corresponds to case 1 in Table 8.2, and the uncertainties are , (t), and (t). The
predictor is dened as
x(t) = A
m
x(t) +b
_
u(t) +

T
(t)x(t) + (t)
_
(8.2)
where the adaptive estimates (t),

(t), and (t) are updated according to the following adaptation
laws:
(t) =
Proj
_
x(t) x
T
(t)Pb,

(t)
_
(8.3)
(t) =
Proj
_
x
T
(t)Pb, (t)
_
(8.4)
(t) =
Proj
_
x
T
(t)Pbu(t), (t)
_
(8.5)
=
(8.6)
with being the adaptation rate, while
P = P
T
> 0 A
T
m
P +PA
m
= Q Q > 0
144
8.3 L
1
The projection operator
150
is dened as:
c
R
n
[f() c, 0 c 1
f : R
n
R
f() =

T

2
max
2
max

0
Proj(, y)
_
_
y if f() < 0
y if f() 0 and f
T
y 0
y
f
f
_
f
T
f
, y
_
f() if f() 0 and f
T
y > 0
_
_
A control law was then dened to provide command inputs to the augmented aircraft control system:
u(s) = kD(s)r
u
(s) (8.7)
r
u
(s) is the Laplace transform of r
u
(t) = (t)u(t) +

T
x(t) + (t), and D(s) is any transfer
function such that it leads to a strictly proper stable lter transfer function
C(s) =
kD(s)
1 +kD(s)
(8.8)
The choice of D(s) and the gain k is subject to the following requirements:
|G(s)|
L
1
L < 1, G(s) = (sI A
m
)
1
b(1 C(s)) (8.9)
L = max
(t)
n
i=1
[
i
(t)[

0
= [
lo
,
hi
], (t) , [(t)[
0
, t 0
Letting D(s) = 1/s Equation (8.8) becomes
145
8.4 Results
Controller Parameter Value
k(1) 100
k(2) 100
k(3) 100
17000
D(s) 1/s
lim
100
lim
100
lim
[0.5, 2]
Table 8.3: Reference of L
1
parameters.
C(s) =
k
s +k
(8.10)
For digital implementation, the 1/s is approximated by the Tustin approximation where
s =
2(z 1)
T(z + 1)
For this particular implementation of the L
1
controller the values used in Table 8.3 were used to
congure the controller.
8.4 Results
8.4.1 Predicted HQR of Plant Models
Predicted longitudinal handling qualities were based on eigenvalue analysis of the short period
found in MIL-F-8785C.
5, 22
Cases 1, 2, and 3 in Table 8.2 were investigated using these criteria
to ensure a broad range of airframe characteristics during ying qualities testing. The load factor
(n/) and short period frequency (
n
SP
) comparison can be seen in Figure 8.6(a). Using this
criterion alone Cases 1 and 3 appear in the Level 1 category while Case 2 appears in the Level 2/3
category. Investigating Control Anticipation Parameter (CAP) vs. damping ratio
SP
, the second
part of the criterion yields further separation between the cases, as seen in Figure 8.6(b). From these
146
8.5 Piloted Simulation Trial
gures it can be seen that the cases are set to be correlated so that Case 1 has a predicted HQR of
level 1, Case 2 has a predicted HQR of level 2, and Case 3 has a predicted HQR of level 3, with
Case 4 being not classiable because the short period is unstable.
8.4.2 Simulation Traces (Linear and Nonlinear)
Simulation test cases were run on each plant test case before L
1
adaptive control was applied and
then afterwards to contrast performance with and without L
1
controller augmentation. For brevity
only the responses with L
1
are shown here. A double command (Figure 8.7) was given to both
linear and nonlinear models to assure that the L
1
controller and underlying models were behaving
similarly. Figure 8.8 shows responses for all four cases in Table 8.2. For reference, Case 1 is the
desired case for all responses. While Cases 2 - 4 do not identically match Case 1, the response in
all cases showed similarities, even for the open-loop unstable case. Further the CASTLE (nonlinear
model) and the Simulink (linear model) are not identical but very similar as well, showing that the
system is responding as desired for both linear and nonlinear models.
For this study a piloted simulation trial was conducted with two experienced test pilots, referred to
as Pilots A and B. One pilot is a retired U. S. Navy Test Pilot with over 3300 ight hours, and a
graduate of the Empire Test Pilot School. The other pilot is currently a U. S. Air Force Test Pilot
with 3000 ight hours and a graduate of the U. S. Air Force Test Pilot School.
8.5.1 Piloted Task Denition
The simulation trial was broken into two parts, the rst (referred to as task 1) being a fully disclosed
trial where a stick rap impulse was used to excite the different aircraft plant cases, both with and
without L
1
augmentation to familiarize the pilot with the different responses from each case. The
instruction to the pilot can be found below.
147
10
0
10
1
10
2
10
1
10
0
10
1
10
2
n
S
P

r
a
d
/
s
e
c
n/ g/rad
Category A Short Period Requirements, Natural Frequency
Level 1
Level 2/3
Level 2
Level 3
Level 3
1
2
3
(a) Natural Frequency Criterion(n).
10
1
10
0
10
2
10
1
10
0
10
1
SP
C
A
P

=

2n
S
P

/

(
n
/

g
*
r
a
d
Category A Short Period Requirements, Damping Ratio
Level 1
Level 2
Level 3
1
2
3
(b) Damping Ratio Criterion (SP ).
Figure 8.6: MIL-F-8785C Short Period Criteria.
148
0 1 2 3 4 5 6 7 8 9 10
5
4
3
2
1
0
1
2
3
4
5
Doublet Command from CASTLE
Time (s)
e

(
d
e
g
r
e
e
s
)
Figure 8.7: Doublet command used to test L
1
controller response.
Objective: Observe and qualitatively comment on the short period natural frequency
and damping for each airframe. This is not a blind test.
Task: Obtain level ight, 360 KEAS at 10000 ft. Trim aircraft, and verify with hands
off stick. Quickly pull aft on the control stick and release to excite the short period
mode of the aircraft with an impulse force. Observe and comment on the rise time and
overshoot of the response to estimate ying qualities of the aircraft.
After the pilot was introduced to all eight cases (four with L
1
augmentation, and four without
L
1
augmentation), a new set of tasks were performed (referred to as task 2). A target aircraft that is
another F-16 using the Stevens and Lewis model
148
with a maneuver autopilot
151
was own ahead of
the piloted aircraft at a constant altitude of 10,000 ft and 360 KEAS (709 ft/sec). The piloted aircraft
was positioned 1000 ft behind and 200 ft below the target aircraft. The pilot was to aggressively
capture the target aircraft within a bulls-eye found on the cockpit windshield. An example of the
149
0 1 2 3 4 5 6 7 8 9 10
1
0
1
2
3
4
5
Time (s)

(
d
e
g
r
e
e
s
)

0 1 2 3 4 5 6 7 8 9 10
1
0
1
2
3
4
5
time (s)

(
d
e
g
r
e
e
s
)
Doublet Command from Simulink

case 1
case 2
case 3
case 4
case 1
case 2
case 3
case 4
(a) Angle of attack () response from linear and nonlinear models with L1 controller.
0 1 2 3 4 5 6 7 8 9 10
20
10
0
10
20
Time (s)
q

(
d
e
g
r
e
e
s
/
s
e
c
)

0 1 2 3 4 5 6 7 8 9 10
20
10
0
10
20
Doublet Command from Simulink
time (s)
q

(
d
e
g
r
e
e
s
/
s
)

case 1
case 2
case 3
case 4
case 1
case 2
case 3
case 4
(b) Pitch rate (q) response from linear and nonlinear models with L1 controller.
Figure 8.8: Doublet short period response with L
1
controller.
150
Figure 8.9: Pilots view for pitch capture task.
pilots view can be found in Figure 8.9. The pilot was to rate the difculty of the task using the
Cooper Harper Rating Scale.
23
The pilot was not told which aircraft model he was ying during the
testing. The instruction to the pilot can be seen below.
Objective: Evaluate longitudinal ying qualities in a dynamic tracking task using
Cooper-Harper rating scale. Note this is a blind test.
Task: Place tail light of lead aircraft just outside of the top ring in the bulls-eye, while
maintaining 9800 ft altitude and near level ight, and approximate co-speed with the
lead aircraft. When ready pull back on the control stick abruptly to move the target
aircraft into the lower ring of the bulls-eye within 1 second. Then move the aircraft into
the center of the bulls-eye within 1 second. The entire task should take approximately
2 seconds.
Adequate: Outer Ring
Desired: Inner Ring
151
8.5.2 Simulation Trial Results
8.5.2.1 Task 1
When asked to characterize the four aircraft models without L
1
augmentation both pilots gave re-
sponses that correlated with predicted model behavior. Pilots A and B both described model 1 as
deadbeat. Both pilots noticed the lower damping and slower frequency in model 2 as well. Pi-
lot A responded that model 3 had a lot of overshoot and poor damping and that the aircraft
felt like it was going to diverge but did not. Pilot B concurred that model 3 had poor damping
as well. Pilot A referred to model 4 as having an aperiodic divergence. When the L
1
controller
was engaged Pilot B could not tell any difference in all four model responses, although he did state
that the controls felt heavier. He further noted that all four models felt deadbeat, and right on.
Both pilots noted that the aircraft had to be re-trimmed with L
1
engaged. With L
1
engaged Pilot A
noticed small differences between models 3 and 4 but felt that models 1 and 2 behaved the same
when L
1
was engaged.
8.5.2.2 Task 2
The Cooper Harper ratings given by each pilot are presented in two forms in Table 8.4 and in Fig-
ure 8.10. Pilot ratings as presented in Figure 8.10 show that a signicant improvement in aircraft
handling does occur with L
1
on versus L
1
off. Further, agreement between aircraft model 1
with L
1
off is found with all L
1
on aircraft model cases. Both pilots judged all L
1
controller
cases as well as case 1 L
1
off (baseline reference case) as being either a 3 or 4 on the Cooper
Harper scale. Pilot B correctly identied cases where the L
1
controller was on versus off with
the exception of the L
1
off aircraft model 1 case. As Pilot A was given more exposure to the
different airframes with L
1
controllers he noted that he was learning the airplane and that the task
became easier. With airframe case 2 and L
1
on, Pilot A remarked nothing to it, and with air-
frame case 4 and L
1
on he remarked very easy, although some pilot compensation was required.
Pilot A felt that the task as described for the case 2 and case 4 airframes with no L
1
compensation
was not achievable.
152
Aircraft Model Pilot A Pilot B
L
1
Off L
1
On L
1
Off L
1
On
1 3 4 3 4
2 10 2 8 3
3 6 3 9 4
4 10 3 9 3
Table 8.4: Cooper Harper Ratings from Flying Qualities Test of L
1
controller augmenting
Short Period.
0
2
4
6
8
10
0
2
4
6
8
10
Case 1
Case 2
Case 3
Case 4
Pilot A L1 Off
Pilot B L1 Off
Pilot A L1 On
Pilot B L1 On
CH
Level 1 FQ
Figure 8.10: Piloted Cooper Harper Ratings Results.
Fromthe eight different aircraft models presented, overall both pilots noted a considerable amount
of re-trimming required for each aircraft model presented. Also, a lack of control harmony between
the pitch and roll axes caused both pilots to pay more attention to the lateral aircraft response than
expected. In general, a considerable separation between ratings was found with the L
1
controller
153
on vs off for all but the case 1 airframe.
According to MIL-F-8785C, Level 1 ying qualities are clearly adequate for the mission Flight
Phase, Level 2 ying qualities are adequate to accomplish the mission Flight Phase, but some
increase in pilot workload or degradation in mission effectiveness or both, and Level 3 ying qual-
ities are such that the airplane can be controlled safely, but pilot workload is excessive or mission
effectiveness is inadequate, or both. Taking this as guidance, and to simplify data correlation, this
study takes achievement of desired criteria (Cooper Harper rating of 1 - 4) as a level 1 ying quality,
achievement of adequate criteria (Cooper Harper ratings of 5 and 6) as level 2 ying quality, and
Cooper Harper rating of 7 -9 as Level 3 ying quality. Using this as a correlation between Cooper
Harper rating and ying quality rating, all L
1
cases were level 1, although several were on the lower
edge of level 1. Most L
1
off cases, however were level 3.
While aircraft simulation and ying real aircraft are similar, deciencies in a simulation will make
drawing direct conclusions from ying qualities data imprecise. This principle may explain why
aircraft model case 1 was a marginal level 1 ying quality aircraft, although preliminary analysis
shows it lying directly in the heart of the MIL-F-8785 level 1 envelope. Further complicating the
analysis of the pilot rating is that only one pilot rated one airframe as a Level 2 ying quality aircraft
although by prediction using MIL-F-8785C both pilots should have rated aircraft model case 2 as
level 2 and case 3 as level 3. Noting that the majority of L
1
off cases were rated as level 3 by the
pilots also raises the possibility that the task assigned may have been too difcult and over-exposed
the weaknesses of the predicted level 2 and level 3 aircraft.
154
8.6 Open Issues in Prediction of Nonlinear Control Flying Qualities
8.6 Open Issues in Prediction of Nonlinear Control Flying
Qualities
While the simulator used in this study may introduce deciencies in the testing for proper ying
qualities, the pilot comments that the adaptive controller felt heavy or at times sluggish indicate
that there are characteristics of the nonlinear controller that are not being captured in the ying
qualities criteria prediction. Current ying qualities criteria use linear systems analysis techniques
to predict piloted ying qualities. These linear systems analysis techniques rst rely on the system
being analyzed to be stable. From Lyapunovs indirect method a linear analysis of a stable system
can also be inferred to be exponentially convergent.
152
While this fact is taken for granted in linear
systems analysis, often only stability and not a rate of convergence can be guaranteed in nonlinear
systems. Nonlinear controllers such as the adaptive controller in this study therefore can produce
characteristics that are not captured by linear analysis techniques. The equivalent time delay sensed
by the pilots and described as heaviness in the control is an example of this problem. Specically
in adaptive controls the adaption rate, or rate of convergence of the companion model can introduce
either lag or unwanted dynamics in the vehicle response. The question then rises as to how fast must
the convergence be in order for the vehicle to have desired ying qualities? If the convergence can
be bounded, what are the bounds so that adequate performance can be guaranteed for pilot control?
Another issue associated with nonlinear controls relates to the zero dynamics,
153
especially in
controllers that use forms of model following that are not adaptive. Zero dynamics can be consid-
ered to be the error between Equation (8.1) and Equation (8.2). The zero dynamics are therefore
the dynamics that are left unaccounted by the nonlinear controller and experienced by the aircraft
while the control law generates zero error. Adaptive controllers modify (t),

(t), and (t) based
on adaptive laws to minimize the zero dynamics. Control methodologies such as dynamic inversion
do not modify these parameters, and as a result of modeling error zero dynamics will exist. The
error and the dynamic shape of this error can then impact the aircrafts response to disturbances.
The aircrafts response will then directly impact the vehicles piloted ying qualities. Enns et al.
154
155
8.6 Open Issues in Prediction of Nonlinear Control Flying Qualities
addresses the zero dynamics problem by careful selection of control variables in the dynamic inver-
sion controller, however even in this case the zero dynamics may still exist. By dening a bound
on the zero dynamics as well as their required characteristics a relaxing of the requirements for a
vehicles on board model may be possible.
156
Chapter 9
Flight Mechanics and Flying Qualities in
Engineering Education
E
NGINEERING education requires a careful balance of theoretical understanding and practical
application in order to foster an understanding of the complex principles covered in a typical
curriculum. It can often be difcult to introduce practical applications into an engineering classroom
that are complex enough to make an impact, yet simple enough to facilitate. This chapter documents
the combination of concepts from several student centered pedagogues along with a unique asset
to create a learning environment that could facilitate the reinforcing of theoretical concepts with
practical applications to deepen the students understanding of basic engineering principles. This
particular combination was unique to Virginia Tech because of its possession of a unique asset, but
the same application of pedagogues could be made elsewhere with other assets other disciplines of
engineering.
157
9.1 Introduction
9.1 Introduction
Teaching students topics in ight mechanics in an Aerospace Engineering curriculum presents chal-
lenges that are not usually present in traditional engineering programs. Because of the size, cost,
and liability associated with operating full scale aircraft most aerospace engineering students only
encounter pictures of an aircraft in a classroom. While scientic knowledge and theoretical con-
cepts can be clearly communicated in a traditional classroom, practical experience with aircraft is
not possible. Modern use of remote control aircraft in the aerospace curriculum aids students by
giving them some practical experience, but it only gives students an experience by proxy, and not
the immersive experience that ight in an aircraft brings. This creates a barrier to learning for stu-
dents eager to fully understand aerospace engineering concepts in ight mechanics with no way to
practice the concepts they have learned outside of traditional homework problems and tests. The
result is that an aerospace engineering curriculum attracts students with great mathematical prowess
and often excludes students with more practically based engineering skills. Education in engineer-
ing requires a balance of scientic knowledge and practical experience to cultivate well rounded
engineers.
Due to safety issues, the cost of ownership of aircraft, liability concerns, and the cost of operating
an aircraft, application of aerospace concepts learned in a classroom is often delayed until students
arrive in industry. While a limited number of academic programs, particular those with aviation
technology degrees, service academies, and Universities with facilities such as Mississippi States
Raspet Flight Research Laboratory can sustain actual ying coursework,
155
for the vast majority
of aeronautical programs, the liability concern alone precludes putting students in an airplane. For
example, in the late 90s, Penn State University was forced by insurance constraints to sell the Piper
Arrow they used for aerospace education purposes.
156
As another example the University of Michi-
gan offers a Flight Test Engineering and Design course that has been radically changed in recent
years due to cost of operation of their Cessna R182.
157
While their students can still plan ight tests
and analyze the limited data-set recorded from a production general aviation aircraft, the students
158
9.1 Introduction
are not allowed access to the aircraft during ight. Access to modern military craft is exceedingly
limited outside the service academies. In industry employees are expected to perform their tasks
without time to revisit concepts learned in school. A means to allow students to apply lessons
learned in a conventional classroom is needed in the aerospace curriculum so that when students
arrive in the workplace they already have the requisite experience in applying their academic skills
to real-world problems.
The National Academy of Engineering (NAE) challenges educators in Educating the Engineer of
2020
78
to pursue student oriented education. They encourage :
1. better alignment of engineering curricula and the nature of academic experiences
with the challenges and opportunities graduates will face in the workplace and
2. better alignment of faculty skill sets with those needed to deliver the desired cur-
riculum in light of the different learning styles of students.
The NAE also highlights a growing separation between practicing engineers and engineering fac-
ulty. The NAE states
The great majority of engineering faculty, for example have no industry experience.
Industry representatives point to this disconnect as the reason that engineering students
are not adequately prepared, in their view, to enter todays workforce.
78
The National Research Council proposes in How People Learn that technology can be used to
support learning by bringing exciting curricula based on real-world problems into the classroom
and building local and global communities that include teachers, administrators, students, parents,
practicing scientists, . . .
79
By using a Learner-Centered approach,
80
and Problem Based Learning
(PBL)
81, 82
as a baseline pedagogy, available technology can be used to create a curriculum that
directly answers the challenge offered by the NAE.
159
9.1 Introduction
In the Spring semesters of 2006 and 2007, Virginia Techs Department of Aerospace and Ocean
Engineering offered an experimental course titled Flight Test Techniques, AOE 4984.
158, 159
The
course was designed to expose senior level undergraduates to industry and government accepted
methods used in aircraft ight testing. Offering this course allowed real world problems to be intro-
duced into the departments curriculum in a controlled environment. Building on the concept that
hands-on application of concepts learned in an academic setting builds conceptual understanding,
and is therefore a key to deep understanding,
160
this course serves as a capstone where concepts
previously taught in several classes are integrated to give students an overarching view of aircraft
ight mechanics, putting the theory students are taught in previous courses into practice. In order
to facilitate a learning environment and mitigate safety issues associated with using real aircraft,
two modern high performance aircraft models in the Virginia Tech Flight Simulation Laboratory
are used instead of test aircraft. This aids completion of targeted learning objectives, while alleviat-
ing operational costs, weather concerns, and liability and safety concerns. Using the simulator also
gives students the ability and opportunity to serve in all of the various roles required to ight test an
aircraft, from test pilot, to test conductor, to discipline engineer.
The ight test techniques course aims to do more than instruct students in ight testing concepts.
The ight test techniques class is designed to provide hands on experience with aircraft to foster
application of the theoretical concepts taught in the aerospace curriculum. The courses curriculum
is designed to create an environment where problem based learning facilitates social and role based
learning under the guidance of teachers that are also industry experts. By guiding students as they
convert their theoretical (procedural) understanding of aerospace engineering into a conceptual un-
derstanding of aerospace engineering they are no longer just dealing with abstract formulas, but are
gaining uency in the language of aerospace engineering and learning to speak airplane.
While the Flight Simulation Laboratory at Virginia Tech is a facility that would be difcult to
reproduce at most universities, the pedagogical concepts used in this course can be used to aid all
160
9.2 Combining Pedagogues
engineering curricular development. By using industry experts to frame real-world problems for
students to solve we have been able to refocus the students learning experience through immersion.
This immersive experience is a a carefully constructed environment that simulates how a practicing
engineer would be required to perform the basic skills that they have been taught in their undergrad-
uate curriculum.
According to Weimer,
80
in order to be considered Learner-Centered, teaching must include ve
elements. First the balance of power in the classroom must be shifted from the instructor to the
students. By empowering students to become responsible for their own learning, the instructor be-
comes more of a facilitator and less of an authoritarian. Second, content is used as an aid to construct
knowledge, instead of presenting a series of facts and gures. Concepts are taught to students, and
they are encouraged to seek specic details to accomplish their assigned tasks from sources outside
of the classroom. Third, the teacher must become more interactive with students, engaging the class
in other forms besides just a lecture. Forth, the responsibility for learning must be shifted from the
teacher to the students. Lifelong learning habits can be fostered by having students take ownership
of the learning process. Finally, targeted assessment must be made to encourage students to not just
learn material for a one time test, but to foster deep learning of concepts for long term retention.
Problem Based Learning ts nicely into a Learner-Centered scheme, and was used to create the
foundational pedagogy for the course.
Using problem based learning as a baseline pedagogy, cooperative learning, and role base learn-
ing can be used to enhance the learning experience. By using a multidimensional approach to
designing a curriculum the impact to student learning can be maximized over a short period of
time. This study uses an integrated approach to course design. Utilizing the interconnections and
strengths between these three teaching approaches a realistic, yet safe environment for posing open-
161
ended problems can be formed.
Problem based learning can trace its formal roots to a desire to reform teaching in the medical
eld. Using a problem based learning approach, the McMaster University Faculty of Health Sci-
ence graduated its rst class in 1972. The College of Human Medicine at Michigan State University
soon followed by implementing an alternative track using problem based learning. The success of
these schools created a wave of medical schools adopting problem based learning tracts in their
curriculum.
82
Spaulding states the motivation for problem based learning as Current dissatisfac-
tion with medical education imposes on a new medical school a responsibility of experimenting
with novel approaches.
161
Problem-based learning was documented in the United States by the
Report of the Panel on the General Professional Education of the Physician and College Preparation
for Medicine, known as the GPEP report
162
sponsored by the Association of American Medical
Colleges. This report gave credence to problem based learning to medical school deans and faculty
. Curriculum changes were implemented to reduce lecture hours, increase independent learning and
problem solving.
Engineering educators have long recognized the need for design in the curriculum. The Edu-
cational Development Program of UCLA considered design the essence of engineering and have
dened it as an iterative decision making process that we now recognize as the common discipline
of all professions.
163
Problem based learning lends itself well to incorporating design into a cur-
riculum. Giving students open ended problems allows them to weight multiple options for a solution
and is the essence of design. Traditional educational techniques use convergent thinking to focus
on a solution with a specic set of answers that are veriable and contain a truth value. In a design
based instruction the questioner is not necessarily concerned with the truthfulness or veriability of
potential answers..
164
Whereas convergent thinking operates on the knowledge domain, divergent
thinking operates on the conceptual domain, which may not be directly veriable.
164
162
Barrows
82
identies six core features of PBL:
1. Learning is student-centered.
2. Learning occurs in small student groups.
3. Teachers act as facilitators or guides.
4. Problems are the organizing focus and stimulus for learning.
5. Problems are vehicle for the development of clinical problem-solving skills.
6. New information is acquired through self-directed learning.
Student excitement over solving real-world problems can be easily lost if the instructor is not con-
sidered an expert in the eld.
79
It is not enough to be an academic expert in the eld; real-world
problems can only be accurately posed by those who have real-world experience. The real-world
experience allows the instructor to pose the problem correctly and ensure that proper emphasis is
placed.
165
The real-world experience is what separates the text book homework problem from the
real-world problem. Without experience in real-world implementation, the posed problems of the
instructor are no better than those found a text book.
An added benet to problem based learning is that it lends itself well to group learning. Smith
et. al.
81
draw a distinction between structured cooperative learning and placing students into study
groups. Formally structured cooperative learning groups must contain:
1. Positive interdependence
2. Face-to-face promotive interaction
3. Individual accountability/personal responsibility
4. Teamwork skills
5. Group processing
163
Students in a cooperative learning environment are encouraged to discuss the concepts presented to
them by their instructor. By having an open discourse on the subjects they are learning, deep learn-
ing is encouraged. The students no longer just accept information presented to them, they explore
the topic and adopt principles only after they have been vetted within their peer group.
166
Coopera-
tive learning groups are very similar to highly productive teams outside of a learning environment.
Katzenbach and Smith describe high-performance teams as:
A team is a small number of people with complementary skills who are committed to
a common purpose, performance goals, and approach for which they hold themselves
mutually accountable.
167
The instructors role is no longer that of a teacher, but more a facilitator. Guidance for methods
of facilitation can be found outside of traditional academia in programs that are designed to build
high performance teams. Rohnke and Bulter state that the leader/facilitator doesnt provide all the
answers to the group; primarily the participants learn from each other.
168
The authors go on to con-
trast leading versus facilitating by stating leading the group helping them to learn or facilitating
helping them to learn from each other.
168
Role-based learning also lends itself well into integration with problem based learning. By in-
troducing students to an environment that is modeled on what will be found when they enter the
professional workforce the problems given to the students take on an added sense of realism. By
giving students specic roles within a team environment to solve broadly dened problems the stu-
dents are challenged to act like professional engineering problem solvers. Using role-based learning
allows the student to fully participate in the practices of a given discipline.
169, 170
The risk associated
with a student or group of students not achieving a desired goal is minimized while their perceived
risk of not accomplishing a given task is still kept at a high level. The students are therefore al-
lowed to work as an engineering team while still in an academic environment. By being immersed
in a professional environment while still being in an academic setting students learn to speak the
language of their profession in a safe environment.
164
9.3 Technology Utilization
Taking advantage of appropriate technology, instructors can introduce real-world problems into the
curriculum. When used appropriately multimedia can become a tool to aid in teaching engineer-
ing principles. According to Mayer, multimedia is a sense-making activity in which the learner
seeks to build a coherent mental representation from the presented material.
171
It then becomes the
instructors job to aid the student in the sense-making. By using a motion based ight simulator
maintained by Virginia Techs Department of Aerospace and Ocean Engineering a complex multi-
media presentation of material is presented to the students. The ight simulator has already been
used to augment the traditional curriculum in the department on both the undergraduate and gradu-
ate level.
172
An expansion of the departments curriculum to include the Flight Test Techniques
course created a chance to create a non-traditional learning environment for teaching aerospace en-
gineering concepts. Using a student teaming approach to create an articial ight test environment
similar to that found at a government ight testing facility, the instructors were able to move beyond
basic course instruction and classic textbook problems. Key to creating this environment was the use
of a realistic ight simulator that is much more capable of approximating ight than a commercially
available desktop ight simulation.
9.3.1 Student Immersion
Often examples of problem based and role based learning are engineering design classes where
students are asked to use concepts from other courses to solve a specic problem. One failing in
trying to transform a classroom environment into a design environment is that the students still per-
ceive that they are in the classroom. The reason is because the proper environment is not set by
the instructor. The reasons can range from an inability to make assignments beyond analysis, to
not having the proper resources to go beyond analysis, to instructor ignorance. Many senior student
design competitions result in only a paper study without students have the opportunity to do hands
on work. This leads to an inability to reach all four learning styles as well as the students thinking
of the class as just another setting to generate a lab report. By using assignments where students
165
are required to do more than just analysis, but to actually perform a task set in a properly estimated
real world environment student perceived risk to the students goes beyond missing a question on
the next test and elevates the urgency to learn a skill taught in the course.
9.3.2 Facility
The facility used to teach this course is the same piloted ight simulator laboratory that was used
to conduct the nonlinear ying qualities research found in Chapter 8. A detailed description of the
simulator and instrumentation can be found in Section 8.2.3. One issue that must be addressed in
the use of a simulator is its ability to approximate reality well enough to frame problems posed to
students. Approximations and simulations can not encompass an entire experience as well as the
real thing. An aircraft simulation is nothing more than an approximation of a real aircraft. This
approximation provides both positive and negative impacts that must be weighed to ensure that the
use of a simulator is relevant. As computers have increased in complexity and capability aircraft
simulators have been able to model the ight of a real aircraft with increasing accuracy. The use
of simulators in approximating an aircraft ight should be targeted to areas of their strength, while
avoiding areas where simulators do not give accurate representation of aircraft ight. Simulators do
not provide the full acceleration cues that are experienced in aircraft ight. Simulators provide only
a limited visual perspective as compared to ying in the real world. On the other hand, aircraft
require extensive maintenance and would not always be available for student use during the course
of a semester. A signicant detractor in allowing student access to real aircraft is the purchase of
liability insurance. Only licensed pilots are legally allowed to pilot aircraft, whereas in a simulator
students can be quickly trained to fulll the position of test pilot. Like an aircraft, a simulator
requires large initial investment to acquire the facility, but unlike an aircraft the maintenance and
operational costs are much less. Another advantage to using the simulator is that the environment
for the simulated aircraft y can be tightly controlled. Flights are not cancelled by bad weather,
and uncertainty in measurements can be tightly controlled so that uncertainties can be introduced or
166
9.4 Classroom Implementation
removed depending on the concepts to be emphasized. While simulators may not be able to fully
reproduce an aircraft ight experience, they do pose an ideal setting for basic instruction of aircraft
ight.
The Flight Test Techniques class is designed to present real-world ight testing problems to teams of
students performing in roles that can be found on a ight test team working on real aircraft. Instead
of using a formal text book, the class uses reference documents that are used in real aircraft ight
test. The role of the instructors change as the course progresses. At the beginning of the course the
instructors give very specic problems to the ight test teams of students. As the course progresses
and the teams mature the instructors transition to a role of facilitators presenting the teams with
more open-ended problems culminating with the test teams designing and performing their own
ight tests.
9.4.1 Course Goals and Objectives
The course goals can be divided into learning goals specic to aerospace engineering and goals
specic to growth in general engineering practice. The high level goals of the course related to
aerospace engineering are threefold: namely to reinforce concepts taught in aircraft performance
and stability and control classes, expose students to ight testing by reproducing the ight test
environment in a classroom setting, and teach students ight test techniques based on currently
used manuals in government evaluation of aircraft to prepare them for careers on ight test teams.
By the end of the course, each student should be able to:
158
1. Apply concepts taught in aircraft performance and stability and control classes.
2. Explain why different aircraft feature different handling characteristics.
3. Describe, based upon hands-on experience, the connection between aircraft ight theory and
aircraft ight.
167
4. List key factors used to dene aircraft performance and stability and control.
5. Apply theory from classrooms into a big picture aircraft perspective, or from a systems
perspective.
6. Describe standard ight test processes and procedures for safety.
7. Outline a procedure for gathering results from stated requirements.
8. Identify challenges unique to the roll of test pilot based upon personal experience in the
simulator cockpit.
9. Identify challenges unique to the position of test conductor, running a ight test as an engineer.
10. Interpret gathered ight test data, and present that data in a standard, technical form.
11. Explain ight test techniques based on currently used manuals in government evaluation of
aircraft.
12. Classify both qualitative and quantitative aircraft characteristics in terms of MIL-F-8785C
13. Calculate level performance data based on student-collected ight data as described in the
US Navy Performance Manual.
173
14. Calculate excess power performance data based on student-collected ight data as described
in the US Navy Performance Manual.
173
15. Calculate sawtooth climb performance data based on student-collected ight data as de-
scribed in the US Navy Performance Manual.
173
16. Calculate climb schedules based on student-collected ight data as described in the US
Navy Performance Manual.
173
17. Evaluate longitudinal ying qualities based on student-collected ight data as described in
the US Navy Stability and Control Manual.
174
168
18. Evaluate lateral-directional ying qualities based on student-collected ight data as de-
scribed in the US Navy Stability and Control Manual.
174
19. Design a ight test synthesizing results from prior ight tests.
20. Identify potential safety hazards for a ight test.
21. Classify the risk inherent to a ight test.
22. Evaluate options/opportunities to mitigate risks.
23. Justify choice of acceptability of risk.
The goals of the course related to growth in general engineering practice are threefold as well:
foster a deep conceptual understanding of aerospace engineering principles, foster student compe-
tence and condence in their knowledge of aerospace engineering so that their knowledge is readily
maintained after taking the test, and give students a hands-on apprenticeship in engineering prior
to leaving school.
These objectives were developed to address the six levels of Blooms taxonomy in the cognitive
domain; knowledge, comprehension, application, analysis, synthesis, and evaluation.
175, 176
An
emphasis is placed upon the higher level objectives of analysis, synthesis, and evaluation as they are
oft neglected in undergraduate education yet are an important component in developing students and
future engineers.
177, 178
Due to the laboratory nature of the course, performance objectives consider
both procedural (e.g. describe standard ight test processes and procedures for safety) and product
outcomes (e.g. design a ight test synthesizing results from prior ight tests).
179
Through tasks
such as developing procedural knowledge, laboratory skills (including, but in no way limited to, the
act of piloting the simulator), and writing laboratory reports, students are able to develop their skill
set in the cognitive, psychomotor, and affective domains.
179
169
9.4.2 Course Organization
The Flight Test Techniques class met twice weekly for 75 minute sessions. This time was split into a
two week rotation, where the rst week was spent in classroom instruction and the second week was
spent in simulation testing. Two optional laboratory times of one hour and fteen minutes each were
established for each test team to work with the simulator as they desired. The twenty student class
was divided into four test teams of ve students each. The students grades were based solely on
ight test reports that were due one week after a test was performed. With the exception of the rst
report, used to gauge each individual students report writing capabilities, each report was submitted
as a group effort thus requiring the students to work together as a ight test team. Students were
given a basic reporting format to follow, and were required to provide percentage of work by each
team member on the report.
Each section of the course begins with a specic, complicated, real-world problem to solve. The
in-class portions serve to supplement the students knowledge by addressing procedures and overar-
ching principles required for the ight test.
180
At the start of each laboratory, students are presented
with a nal task which must be accomplished, such as quantifying aircraft climb performance. They
must then use knowledge from prior classes to determine the data required for the ight test.
Lectures are presented to address student questions and to assist with the general formulation
of the presented problem. Specically, the classroom portion of this course was used to revisit
students prior coursework on aircraft performance, equations of motion, basic stability and trim
analysis, and energy management. Additionally, ight testing procedures were introduced. After
discussing the theory behind the test, a pre-ight brieng for each test was done in class. This
pre-ight brieng included the test cards to be used (sample page of test cards given in Figure 9.1,
reporting objectives, safety requirements, and any special procedures required for the test. The
exception to this pattern is during ight tests 4 and 5, where students must develop their own test
cards and present a ight safety board (FSB) brieng.
After classroom instruction, students proceed to the simulator to conduct the ight test, and re-
duce the data for their team-generated topical report. The students time spent in the Flight Simula-
170
Figure 9.1: First page of Laboratory 1 ight test cards.
171
tion Laboratory was split into two dened periods. The rst period was listed as a practice session
for each ight test group. Each group was given a total of 3 hours a week simulator time that could
be used for anything they wished, from basic aircraft familiarization, to night carrier landings, to
rehearsing the next ight test. This time was not required for class participation, but was made
available to the students so that they could get the most exposure possible to the ight simulator.
The second period was during the ofcial course time, and was when their actual ight test was
performed. During this time each team reported to the ight simulation lab for their assigned test.
This time was highly structured, and student participation was required. While roll was not taken
during this time, each test was designed to require a full test team. In order for the data to be taken
for a given test report a full team was required, and students learned quickly to work together to
ensure all were present and engaged during testing.
9.4.3 Flight Testing Positions
Specic responsibilities were assigned to different stations during the ight test. The ight test tasks
were broken down by position as described in Figure 9.2. While not required to do so, students were
encouraged to rotate positions so that they could experience all aspects of ight testing during the
semester.
9.4.3.1 Test Pilot
This position requires great skill in piloting the aircraft. The test pilot must be capable of doing
much more than just y the aircraft. He must be capable of ying the precise maneuvers required
by the ight test cards to generate the necessary data for the test. At the beginning of the semester,
the instructor would demonstrate the rst half of the test, and then help the student pilot through the
maneuvers during the second half of the test. As the semester progressed and student prociency
improved, the students assumed the full role of test pilot enabling them to y entire test missions.
It should be noted that several students in the course had experience as private pilots, and quickly
learned this position. Interestingly enough however, other students in the class had considerable
172
Operations:
Ground
Data1 Data2
Con Test
Pilot FTE
Workstation Plan for Flight
Test Lab Course
Aircraft: Hokie 1
Pilot: Test Pilot
FTE: Flight Test Engineer
Operations: Ground
Test: Test Conductor
Con: Console Operations
(simulation operator)
DataRoom:
Data1: IADS display
Data2: IADS display
Aircraft: Hokie1
Data Room
Server Room
Simulator Bay
Figure 9.2: Illustration of ight test tasks/stations.
173
video game ight simulator experience, and also quickly learned this position.
9.4.3.2 Flight Test Engineer (FTE)
For this course the position of Flight Test Engineer was dened as the person in the right seat of the
cockpit. While sitting in this position, the FTE would have access to data displays that could not be
shown to the pilot. The FTE would work with the pilot to ensure the test criteria were being met by
reading off instrumentation and augmenting the pilots information during the test. The FTE also
served as a second set of eyes and ears in the cockpit during the test, and would comment on pilot
workload and any anomalies during the test.
9.4.3.3 Test Conductor
The test conductor was positioned at the operation console for the simulation. The test conductor
was responsible for the overall test, and would direct the test from the test cards. He would also
mark the time the events occurred during the test, so that during data playback in post-test analysis
the proper timeline for data analysis could be identied.
9.4.3.4 Discipline Engineer
While for this course specic discipline engineers are not identied, students monitoring the test
data were referred to as discipline engineers. They took data to ll out test cards, monitored pa-
rameters during the test, and commented on the test margins to ensure the aircraft remained within
desired parameters.
9.4.4 Teaching and Student Participation
Flight testing relies on multiple individuals with various talents working together as a team to be
successful. Understanding each role and how it relates to the team is crucial to effective team
performance. The Flight Test Techniques course reinforces this concept by grouping students into
ight test teams of ve students. Each test team must prepare, execute, and report on the test as a
174
team. Further, each student is given an opportunity to ll various roles on the team. By rotating team
positions the student can nd areas of strength and areas of needed improvement in their skill set.
Allowing students to experience each team position also gives a sense of appreciation for each team
positions responsibilities. When students complete this course, they have a basic understanding of
the nature of each position in a ight test team and a good idea of where they would best t into a
ight test team when they enter the workforce.
A unique aspect of the Flight Test Techniques course is that it accommodates a variety of learning
styles. According to Fleming and Mills
181
students address the learning of information in one of
four ways; Visual, Aural, Read/Write, and Kinesthetic. The Flight Test Techniques course presents
its topics to the students in all four ways, maximizing the chance that a student has to learn the
material. Each topic is presented in a classroom lecture and pre-ight brieng where the theory and
technical background for the test is presented, representing the aural learning style. Students are
given text and reference information regarding the topic, aiding in their preparation for the ight
test, and addressing the read/write learning style. During the test data is displayed in a visual format
that the students must watch, and judge as to its quality,allowing the exercising of the visual learning
style. The test also allows the students to see rst hand how their actions to the aircraft impact the
results of the test, giving them an experiential learning environment, accommodating a kinesthetic
learning style. Finally, a report is written at the end of the test allowing the students to demonstrate
their understanding of the material presented, further reinforcing the read/write learning style.
To appeal to and retain students from diverse backgrounds, this course aims to serve all psycho-
logical and learning types. Typical engineering curricula teach toward intuitive students with the
majority of faculty falling into the Myers-Briggs category of intuitors.
182
However, most engineer-
ing students are sensors.
176
Thus Wankat and Oreovicz
176
suggest designing courses to appeal to
sensors and serial learners with global summaries presented at the end of each class thus serving
both learning styles. By designing the course to serve both learning types, requiring both recall of
prior information and open hypothesis, stressing both how to evaluate data and the possibilities that
arise from the data, and requiring team-work in the completion of ight tests and reports, diversity
175
in the attraction of students with varied learning approaches and backgrounds is fostered.
9.4.5 Using the Proper Vocabulary
The vocabulary of aerospace engineering includes contributions from academia, technicians, and
pilots, as well as working engineers to create a mixture of terms that can often be confusing or
intimidating to an engineer directly out of school. For example, academia typically only uses SI
units, whereas the US aircraft industry as well as international ight regulations only use English
units. While students become familiar with velocity in meters per second, they will be asked to
relate to others in industry that are familiar with nautical miles per hour (knots). Aircraft also re-
port velocities in equivalent airspeed, calibrated airspeed, true airspeed, ground reference speed,
measured mach number, true mach number to name a few examples. In typical undergraduate cur-
riculums only one velocity is used to reference an aircraft, and the distinction between the different
values is not always made clear. It is also important that students begin to recognize synonyms in
the vocabulary such as angle of attack, and its typically referenced variable name . Pilots will
often communicate is short bursts of very technical jargon in order to convey a qualitative analysis
of their situation, and it is the job of the aerospace engineer to translate that jargon into a form that
can be used for quantitative analysis. By immersing the students into a world where these terms are
used on a regular basis, and them expecting them to communicate using those terms, the students
quickly become conversant in a vocabulary that they would not have been exposed to during their
undergraduate education.
9.4.6 Using Reference Books Instead of Textbooks
This class uses industry reference documents and asks students to reuse textbooks from their previ-
ous classes. Taking this approach exposes students to documents used in industry. This approach
also teaches students how to search their personal libraries for information. Government references
used in the course include U.S. Naval Test Pilot School Flight Test Manual, Fixed Wing Stability and
Control: Theory and Flight Test Techniques,
174
U.S. Naval Test Pilot School Flight Test Manual,
176
Fixed Wing Performance,
173
MIL-SPEC-8785C,
183
and NASA TN D-5153.
23
Students are asked
to review texts and course notes from their courses in aircraft performance, and aircraft dynamics
and control. The course structure is centered on the fullling eight laboratory assignments or ight
tests with each assignment building on material presented from previous tests. Instead of following
a topical survey similarly to how most engineering texts are organized, the students follow a sim-
ulated aircraft through a series of ight tests as if the aircraft was a new aircraft going through an
evaluation / test program. Instead of using classroom time to teach concepts learned from previous
classes, the classroom lectures are spent doing a quick review of concepts to be used for each ight
test followed by testing techniques that are used to test for different aircraft characteristics. Students
are required to review on their own basic aircraft concepts found in their text books.
9.4.7 Team Building
Course progression is designed to allow teams to build up to a point of solving problems presented
by the instructors with minimal intervention. Since students require basic instruction into ight
testing at the beginning of the semester the rst assigned lab is labeled as a familiarization with
the aircraft and basic testing techniques. Test teams are formed, but kept under tight supervision
with an instructor taking the lead as a test conductor. Students must rely on each other to perform
and analyze data for the rst test, but must write individual reports. The next lab assignment the
team of students is given formal instruction as to the testing techniques but is given full autonomy
in conducting the test. This lab is also the rst group lab report where the entire group must rely
on each member to contribute to the nal report. Each member collects part of the data-set during
the ight test, and the team as a whole must assemble the pieces of data taken in order to form
a full report. A third lab is then conducted under the same rules, where instructors give specic
instruction on how to conduct the lab, but students are given full autonomy to conduct the test and
analyze the data. During this time the students begin to understand the strengths and weaknesses
of each team member, and begin to get comfortable conducting the ight tests. The fourth lab in
the series builds directly from the third lab data and its concepts. Students are given a tutorial in
177
how to analyze the data for aircraft climb performance, and then given the task of predicting the
aircrafts climb performance and then testing their predictions. The students are required to create
their own ight test plan and then brief the plan before it is executed to a Flight Safety Board (FSB)
which consists of the instructors. Included in the plans are hazard analysis and mitigation as well
as a detailed ight test plan. The ight test teams must create this brieng, and are quizzed on each
portion to ensure all students were equal participants in creating the plans. Upon FSB approval the
team conducts the ight test. They must then must analyze their data and create a group ight test
report. The fth lab is designed to take data from the second and fourth labs to create a minimum
fuel ight prole for an aircraft to follow to achieve a landing on dry land after an aborted air craft
carrier landing. The students must design the ight prole based on the data from their previous
ight tests. They are then required to present their plan at another FSB brieng. After FSB approval
the teams conduct their ight test and analyze their data to create another group ight test report.
The nal two labs transition to a subject known as ying qualities analysis. The students are once
again given instruction as to proper test techniques, but this in these labs multiple students from
the test team must y the mission and results are based off of the pilots opinion of the aircrafts
ying qualities. Students must assemble possible conicting results and come to a consensus on a
collection of qualitative and quantitative data. By the end of the semester the students have not only
applied lessons taught in the classroom, but they have used that information to create solutions to
open ended problems presented to them by their instructors.
9.4.8 Evaluation of Students; More Than Just Grades
The formal grading for the semester is based on the grades given for each lab report. While the
grading may seem very simplistic, the work required to generate the lab report requires complex
interactions between the students. The students must work together to compile the data that each
student records into a full data-set. The simulator allows the instructors to create a lab so that there is
a predetermined right answer, but the right answer is not required for students to perform well on
a lab report. Instead of stressing a right answer the grading of the lab reports stresses the students
178
9.5 Evaluation of Objectives
ability to justify the results provided. A wrong answer that is well justied with shortcomings to
the methods used and a possible solution for correcting the data receives just as much credit as a
right answer that describes how the data was correctly taken. A right answer that does not detail the
methods used in the ight test and explain the methods can be given as poor a grade as an incorrect
answer that is poorly justied. Students are informed at the beginning of the semester that the data
they take will build upon itself over the course of the semester, and that incorrect data will skew
results not just for one lab, but for the entire semester. This adds extra motivation to ensuring that
the team members perform their duties and that all work is understood fully by each team member.
In the event that erroneous data is generated and the team can describe the error, and how to correct
the procedure the team can be allowed to rerun the test and generate better quality data, but only
after the team has come to a conclusion as to how to correct their work.
The evaluation of the course goals and objectives as stated in Section 9.4.1 was done with grading
of student reports, and the use of anonymous surveys. The surveys were given to the students at the
end of the course, to the students after they had entered the workplace, and to the students employ-
ers. The results of these surveys are collected in Appendix A.
The rst high level goal of reinforcing concepts taught in the aircraft performance and stability
and control classes was targeted rst by assuring the topics found in rst eighteen specic course
goals were adequately addressed in the lesson planning. This included readdressing issues during
class time when a student report showed deciencies in a particular area. The surveys then showed
that the concepts were reinforced. Questions 6 and 7 in Section A.1 listed below
This class has improved my overall understanding of aerospace engineering.
Strongly Agree
1
5
Strongly Disagree
Average Response: 1.415
179
I better understand the theories and concepts taught in my other classes based on the
topics taught in this course.
Strongly Agree
1
5
Strongly Disagree
show that the students felt strongly that their knowledge of concepts taught in other classes was
enhanced by this course. Further comments 2, 4, 5, and 7 in Section A.1 shown below
Actually applying knowledge learned in other classes
It provided a new conduit through which to demonstrate the material taught
through the AE curriculum
Helps you see how everything from previous courses comes together.
It reinforced topics learned in other courses and gave those concepts real-world
meaning.
reinforce the perception that students were able to enhance their understanding of topics from pre-
vious courses. The post graduation survey further enforced the accomplishment of the goal. In
Section A.2 questions 3 and 8 below
My understanding of basic aerospace engineering principals was enhanced by taking
AOE4984 Flight Test Techniques.
Strongly Agree
1
5
Strongly Disagree
The use of a ight simulator enhanced my understanding of aerospace engineering con-
cepts.
Strongly Agree
1
5
Strongly Disagree
180
show that aerospace principles were reinforced with the course, as well as comments 2, 9, 10, 13,
and 15 shown below.
The ight simulator provided a great hands-on experience to reinforce concepts
learned in lecture. An hour ying the simulator experiencing the different modes
of stability was comparable in value to weeks of lecture. Being able to calculate
a max range climb/cruise/descent prole in an effort to make it to an airstrip, and
then successfully y the prole with a limited amount of fuel was much better
than any scenario that could be presented in a Performance class.
I think it was the most important class that I took at Virginia Tech. It exposed
me to the type of job I wanted as well as enabling me to get a better grasp on the
concepts that I had spent the last 3 years of study learning about.
The ight test class made for an excellent capstone to my engineering degree and
allowed me to put everything I had learned into practice and apply to practical
applications. It was the rst class that made me honestly feel like I was in the
right major, which is a shame considering it was taken in my last semester.
Flight test techniques was without a doubt my favorite class I took in the aerospace
curriculum. It helped reinforce the principles learned in other classes and gave me
the connection to the real world that I needed. Too much of the aerospace cur-
riculum is abstract, ight test was much more realistic.
I think ight simulator assignments should be incorporated into intro to aerospace
engineering and in aircraft ight dynamic courses... I can honestly say I didnt un-
derstand several aerospace principals until I was exposed to an actual aircraft. I
feel that it is highly unlikely that the department can expose students to real air-
craft, but I think a simulator could help ll the gap.
The second high level goal was to expose students to ight testing by reproducing the ight test
181
environment in the classroom. This was done by rst making sure the instructors could speak in an
authoritative fashion about ight testing. The labs were arranged so that students would be using
industry standard software for analysis (IADS
R
) as well as ight test cards that are representative
of what would be used in industry. The labs were all initiated with pre-ight briengs formatted
after typical ight test briengs, and they were completed with ight test reports that were similar
to those that would be found in industry. The self assessment surveys attest to the realism of the
class. Specically comments 1, 2, and 9 in Section A.1 below
This class actually covered material and procedures required in the Real World,
not just the theory behind it.
Actually applying knowledge learned in other classes
I liked that the course required us to come up with our own ight test cards. It
really made us think about what we were doing.
show the students understanding the relevance of the class. In Section A.2 the result from survey
question 1 states that 6 of the respondents were currently employed in ight testing. Comments 3
and 16 in Section A.2 and shown below
Looking back on the class, everything in the real life ight test eld is exactly
like the class.
Ive been to a few performance ight tests and they are exactly how I remember
them from the ight test course. It wasnt until I went to the ight tests that I truly
appreciated how close the ight test class was to the real world.
further illustrate from students that have gone to ight testing that the proper ight test environment
was created in the class.
The third high level goal was to teach students ight test techniques based on manuals used in
government evaluation of aircraft. The two primary texts of the course were U.S. Naval Test Pilot
School Flight Test Manual, Fixed Wing Performance
173
and U.S. Naval Test Pilot School Flight Test
182
Manual, Fixed Wing Stability and Control: Theory and Flight Test Techniques.
174
These two manu-
als were used in conjunction with the students texts from previous courses to show how their course
texts could be used as a reference when applying techniques specied in a professional manual. Ev-
idence of a student understanding the importance of this can be found in Section A.1 comment 4
also seen below.
It provided a new conduit through which to demonstrate the material taught through
the AE curriculum
Questions 4 and 5 in Section A.2 and questions 3, 4, and 5 in Section A.3 all collected below
My PROCEDURAL understanding of aerospace engineering principals was enhanced
by taking AOE4984 Flight Test Techniques.
Strongly Agree
1
5
Strongly Disagree
My CONCEPTUAL understanding of aerospace engineering principals was enhanced
by taking AOE4984 Flight Test Techniques.
Strongly Agree
1
5
Strongly Disagree
The AOE4984 graduate has a higher PROCEDURAL understanding of aerospace en-
gineering principals than a typical employee of his tenure.
Strongly Agree
1
5
Strongly Disagree
183
The AOE4984 graduate has a higher CONCEPTUAL understanding of aerospace en-
gineering principals than a typical employee of his tenure.
Strongly Agree
1
5
Strongly Disagree
New hires working for me are required to have a stronger than average PROCEDURAL
understanding of aerospace engineering principals.
Strongly Agree
1
5
Strongly Disagree
show that both procedural as well as conceptual understanding of engineering material was accom-
plished during the class. The concept of aiding design thinking with divergent thought was evident
in the comments 7,8, and 9 made by the students in Section A.1 as seen below.
It reinforced topics learned in other courses and gave those concepts real-world
meaning.
I feel that this course should be required at the junior or senior level because
it brings hands-on, real meaning to all of the theory and number crunching we
do in other classes. This was one of the few classes where I actually got to do
engineering .
I liked that the course required us to come up with our own ight test cards. It
really made us think about what we were doing.
184
Addressing the general engineering practice goals, they were primarily targeted with the specic
course goals 19 - 23. Deep conceptual understanding of aerospace concepts was evaluated by the
response found in Section A.1 comments 5 and 6 listed below
Extensive teamwork was required to complete the work required for the mission
on the simulator, as well as complete the report. The same type of teamwork is
required of me in my current job.
I came to work with a good understanding of the team mentality that is critical
to perform my job effectively, and also gave me real life examples of problems
that could occur.
as well as Section A.3 comment 1
I cant stress the importance of relevant ight test experience. I go out of my way to
nd students that have had the VT or [another] ight test class because of the knowl-
edge and experience gained by the students. From the basics of how to develop a test to
the specics of what actually occurs during a ight, the ight test course exposes stu-
dents to the skill set necessary to excel at [ORG-X]. [ORG-X] has spoken with several
universities in the past and encouraged their engineering departments to incorporate a
class similar to the VT ight test class. It would be a shame to lose that competitive
edge that VT grads have with [ORG-X].
Students knowledge was improved beyond just the nal exam and the course as evidenced by the
response in Section A.1 comments 10 and 11 seen below,
This was one of the most interesting and practical courses Ive taken. It showed
us why we need to know what we are learning.
My favorite course of my 5 years here. I learned more in this class than any
other. The rst time I was excited to be an engineer.
as well as Section A.3 comment 2
185
The AOE4984 curriculum clearly prepared my employee for the avionic systems work
that he was hired for. He came in with surprising breadthwise knowledge and was hun-
gry to dig into the more complex and ambiguous problems our engineers are challenged
with in the industry.
The hands on apprenticeship was accomplished as demonstrated in Section A.2 questions 8 and
9 as well as comments 5 - 10 of that section.
The use of a ight simulator enhanced my understanding of aerospace engineering con-
cepts.
Strongly Agree
1
5
Strongly Disagree
The teaming arrangements of AOE4984 Flight Test Techniques prepared me for my
current work environment.
Strongly Agree
1
5
Strongly Disagree
that could occur.
I work with teams of various people almost everyday at my job. Without work-
ing together not much would get done. Time working on teams in college helps
prepare for this.
186
I currently work in a team environment. Being able to communicate and work
with others with a variety of personality traits is difcult. The team arrangement
in this class prepared me for my current work environment.
I think it was the most important class that I took at Virginia Tech. It exposed
me to the type of job I wanted as well as enabling me to get a better grasp on the
concepts that I had spent the last 3 years of study learning about.
The ight test class made for an excellent capstone to my engineering degree and
allowed me to put everything I had learned into practice and apply to practical
applications. It was the rst class that made me honestly feel like I was in the
right major, which is a shame considering it was taken in my last semester.
Section A.3 comment 3 also reinforced the concept of an apprenticeship environment during the
course.
This gives the new hire a leg up in bridging the gap between theory and practice. This
has taken some load off of me and our senior leads.
One interesting conclusion from the surveys conducted is that high impact can be made in deep
understanding of a subject without causing extra work for students. In Section A.1 questions 1 and
2 seen below
This class required time than my other classes.
Considerably More
1
5
Considerably Less
I learned than I expected from this class.
Considerably More
1
5
Considerably Less
the students rated the ight test course as requiring no more work than any of their other courses,
but at the same time stated that they received more benet from the class than most of their other
187
traditionally taught courses. The blending of problem based learning with cooperative learning and
role based learning creates a very powerful teaching strategy as shown by the results of surveys
conducted. The realism of the course was strengthened by the two instructors having rst hand
knowledge of the ight testing environment. This lends credence to the statement from the Engi-
neer of 2020
78
calling for educators to have industry experience.
An added benet to using cooperative learning is that students foster teamwork skills that can be
used outside of the narrow focus of ight testing as offered in the course. Although Section A.2
question 1 states that most students that responded were not a part of ight testing eld the teaming
skills taught in the course were highly valued as found in Section A.2 question 9, and comments 5-8.
The teaming arrangements of AOE4984 Flight Test Techniques prepared me for my
current work environment.
Strongly Agree
1
5
Strongly Disagree
that could occur.
I work with teams of various people almost everyday at my job. Without work-
ing together not much would get done. Time working on teams in college helps
prepare for this.
I currently work in a team environment. Being able to communicate and work
188
9.6 Conclusions
with others with a variety of personality traits is difcult. The team arrangement
in this class prepared me for my current work environment.
The survey in Section A.2 also suggested that both conceptual as well as procedural understand-
ing in aerospace engineering principles were enhanced. By having a practical application of previ-
ously studied skills one would expect that conceptual principles would receive greater benet, but
according to the participants both were almost equally enhanced. It is also interesting to note the
balance with which employers value conceptual as well as procedural understanding as found in
Section A.3.
9.6 Conclusions
The teaching experiment with the ight test techniques course was successful in motivating students
and having them learn from themselves while being immersed in a realistic environment. Courses
such as this one allow students to apply the scientic knowledge that have learned from previous
classes and begin to gain condence in their understanding of the principles. This gain in condence
and conceptual understanding is what allows students to be able to speak airplane in a competent
manner. While many students desire application of the engineering skills taught in standard courses,
the context of the application is just as important as application itself. By combining pedagogues in a
way that closely mimicked a real life scenario the ight test techniques course was able to foster the
learning and remembering of concepts that were useful in both the students and their employers
opinion. With courses designed in this manner signicant gains in student understanding can be
achieved and measured.
189
Chapter 10
Conclusions and Recommendations
T
HIS work has highlighted several areas of research in ight mechanics and ying qualities in
an effort to raise awareness of open issues in ight mechanics, and to promote future work in
this eld. The growing use of UAVs in particular has necessitated a fresh look at existing ying
qualities criteria that were originally developed for piloted applications. In addition, the desire to
use nonlinear control systems in aircraft for failure reconguration in moment and commands has
resulted in signicant research in control design, and a need now arises to evaluate those controllers.
Finally hands on instruction in ight mechanics education presents unique challenges that can make
access to piloted aircraft difcult and often prohibitive. Work has been done to overcome some of
these challenges by using UAVs in classroom instruction
128, 129, 184
to provide experiential access to
aircraft in ight mechanics education. Although UAVs overcome some of these challenges, they
still do not yield the rst hand experience of ight as experienced with piloted aircraft.
A new method to classify UAVs has been presented that allows for a more appropriate grouping
of vehicles for ying qualities evaluation. Current methods used by the military focus on gross
takeoff weight, and do not allow for the operational envelope of the vehicle when the classication
is made. Borrowing from some of the concepts used to classify piloted aircraft, a new classica-
tion method has been proposed using both Reynolds number and gross takeoff weight. Instead of
190
10 Conclusions and Recommendations
picking aircraft arbitrarily by size, this method allows ying qualities criteria to be evaluated using
aircraft from representative groups with parameters that directly relate to the forces and moments
that affect ight. This classication could be expanded, however. Future work to include maximum
load factor for maneuvering UAVs would enhance the proposed classication system for UAVs. A
similar classication method is also needed for sensors that are used on UAVs. For camera related
sensors the gimbal dynamics and focusing dynamics, as well as eld of view would possibly allow
cameras to be classied in a similar manner to aircraft. While this data has been difcult to obtain,
the ultimate goal of integrating payloads and aircraft to form a UAV ying qualities standard must
have both aircraft and sensors classied.
The UAV ying qualities criteria presented in this work are breaking new ground in ying qual-
ities analysis. They should be considered the rst steps to creating an adopted standard between
government and industry partners. The simplied pitch criterion offers the benet of using simple
analysis techniques to analyze an aircraft and sensor system that can be characterized by a transfer
function. It treats the mixed aircraft and sensor dynamics as a complete system whose ultimate goal
is to follow a desired pointing vector towards a target. In this criterion the integrated aircraft and
sensor dynamics are measured to ensure a stable platform for the sensor to perform its mission effec-
tively. The criterion with winds, or criterion goes further than the simplied pitch criterion and
measures the interaction of the aircrafts phugoid and short period in performing the mission task.
Of particular interest is the turbulence dynamics that can not be rejected by sensor stabilization.
Coupled together these two criteria give the UAV designer the ability to measure the total UAV and
sensor systems ability to achieve a desired task. Future work is needed in creating a catalogue of
MTEs. Once these MTEs are created appropriate boundaries can be drawn for achieving different
levels of ying qualities. Once these boundaries have been dened, ight test maneuvers can be
designed for the ight test and evaluation of UAV ying qualities.
Nonlinear ight control systems have seen considerable research and advancement toward imple-
191
mentation on both piloted and unpiloted aircraft. Their evaluation for ying qualities testing must
also be considered. A L
1
adaptive controller has been evaluated for a pitch tracking task in a piloted
simulation. This marks the rst time that this type of controller has had a piloted evaluation. The
successful recovery of good ying qualities by the adaptive controller for a degraded system shows
promise for future implementation of this controller for piloted ight testing. However, just as ying
qualities standards must evolve to support UAV applications, ying qualities criteria must also be
developed to test nonlinear control systems.
The new challenges in ying qualities analysis require that future generations of engineers have
hands-on training in aircraft ight mechanics in order to further develop and apply future ying
qualities criteria to the next generation of aircraft. Engineering educators have been challenged to
rise to new levels of effectiveness in teaching tomorrows engineers. The subject of atmospheric
ight mechanics is a difcult subject to apply hands on teaching since the cost and risk of operating
piloted aircraft can be prohibitive. The use of a piloted ight simulator to augment ight mechanics
curriculum was demonstrated with great success. By combining problem based learning and role
based learning in a learner centered environment students were challenged to learn while doing,
fostering deep learning of ight mechanics concepts. The ight simulator served as a key catalyst in
creating a relevant environment where students could explore and learn key ight mechanics prin-
ciples. It is unfortunate that very few schools have a resource of this caliber available to aid in ight
mechanics instruction.
With the advent of new aircraft control systems, and new paradigms of aircraft such as UAVs
ight mechanics research has renewed relevance. UAVs today are gaining prominence just as pi-
loted systems gained prominence during the 1950s when the rst piloted ying qualities criteria
were developed. Continued work is needed to classify UAV payloads, catalogue MTEs in order to
develop a systematic approach to creating ying qualities criteria, and in the development of lateral-
directional ying qualities criteria. The application of ying qualities criteria to nonlinear control
192
systems for both piloted and unpiloted systems must also continue to be explored.
193
Bibliography
[1] B. Etkin, Dynamics of Flight, Stability and Control. New York, NY: John Wiley and Sons,
second ed., 1982.
[2] F. W. Lancaster, Aerodonetics. London, England: A. Constable & Company Ltd., 1908.
[3] G. H. Bryan and W. E. Williams, The Longitudinal Stability of Aerial Gliders, Proceedings
of the Royal Society of London, Vol. 73, The Royal Society, Janurary 7 1904, pp. 100116.
[4] G. H. Bryan, Stability in Aviation. London, England: Macmillian Co., 1911.
[5] anon, U.S. Military Specication MIL-F-8785C. Department of Defense, 5 November 1980.
[6] C. R. Chalk, T. P. Neal, T. M. Harris, F. E. Pritchard, and R. J. Woodcock, Background
Information and User Guide for Mil-F-8785B (ASG), Military Specication-Flying Quali-
ties of Piloted Airplanes, Tech. Rep. AD860865, Air Force Flight Dynamics Laboratory,
Wright Patterson AFB, OH, June 1971.
[7] anon, Advertisement and Specication for a Heavier Than-Air Flying Machine, Army Sig-
nal Corps Specication No. 486, December 23 1907.
[8] O. Chanute, Progress in Flying Machines. Originally Published in American Engineer and
Railroad Journal, New York, 1894, Mineloa, New York: Dover, 1997.
[9] O. Wright, How We Invented the Airplane. First Published in Harpers Magazine, June 1953,
original manuscript 1920, Dover, 1988.
[10] D. T. McRuer and D. Graham, Flight Control Century: Triumps of the Systems Approach,
194
Bibliography
Journal of Guidance, Control and Dynamics, Vol. 27, March-April 2004, pp. 161173.
[11] D. G. Mitchell, D. B. Doman, D. L. Key, D. H. Klyde, D. B. Leggett, D. J. Moorhouse,
D. H. Mason, D. L. Raney, and D. K. Schmidt, Evolution, Revolution, and Challenges
of Handling Qualities, Journal of Guidance, Navigation, Control, and Dynamics, Vol. 27,
January-Feburary 2004, pp. 1228.
[12] H. A. Soule, Preliminary Investigation of the Flying Qualities of Airplanes, NACA Report
700, 1943.
[13] R. R. Gilruth, Requirements for Satisfactory Flying Qualities of Airplanes, NACA Report
755, 1943.
[14] anon, Stability and Control Requirements for Airplanes, U. S. Navy Bureau of Aeronautics,
Washington D. C., BuAer Specication No. 119B, August 1943.
[15] anon, Stability and Control Requirements for Airplanes, U. S. Army-Air Forces Wright
Patterson AFB, OH Specication No. C-1815, August 1943.
[16] anon, Miltary Specication, Flying Qualities of Piloted Airplanes, MIL-F-8785(ASG),
September 1954.
[17] C. B. Westbrook, The status and future of ying qualities requirements, 2nd Annual Meet-
ing of the American Institute of Aeronautics and Astronautics, No. Paper 65-313, San Fran-
sisco, CA, July 26-29 1965.
[18] C. R. Chalk and R. K. Wilson, Airplane Flying Qualities Specication Revision, Journal of
Aircraft, Vol. 6, May - June 1969, pp. 232239.
[19] J. Hodgkinson, Aircraft Handling Qualities. AIAA, 1999.
[20] anon, Flying Qualities Textbook, Vol. 2. Edwards Air Force Base, CA: USAF TPS, 1986.
[21] W. G. Vincenti, What Engineers Know and How They Know It, Analytical Studies from Aero-
nautical History. Baltimore, MD: Johns Hopkins University Press, 1st ed., 1990.
[22] anon, MIL-STD-1797. Department of Defense, March 1987.
[23] G. Cooper and R. Harper, The Use of Pilot Rating in the Evaluation of Aircraft Handling
Qualities, Tech. Rep. TN D-5153, NASA, April 1969.
195
Bibliography
[24] R. Harper and G. Cooper, Handling Qualities and Pilot Evaluation, Journal of Guidance,
Control and Dynamics, Vol. 9, September-October 1986, pp. 515529.
[25] D. G. Mitchell and R. H. Hoh, Low-Order Approaches to High-Order Systems: Problems
and Promises, Journal of Guidance, Control and Dynamics, Vol. 5, September-October
1982, pp. 482489.
[26] T. P. Neal and R. E. Smith, A Flying Qualities Criterion for Design of Fighter Flight-Control
Systems, AIAA Journal of Aircraft, Vol. 8, October 1971, pp. 803809.
[27] R. Smith and N. Geddes, Handling Quality Requirements for Advanced Aircraft Design:
Longitudinal Mode, Technical Report AFFDL-TR-78-154, AFRL, Flight Dynamics Labo-
ratory, August 1979.
[28] R. E. Smith, Effects of Control System Dynamics on Fighter Approach and Landing Longi-
tudinal Flying Qualities, Tech. Rep. AFFDLTR78-122, AFRL, March 1978.
[29] J. Hodgkinson, J. R. Wood, and R. H. Hoh, An Alternate Method of Specifying Bandwidth
for Flying Qualities, AIAA Guidance and Control Conference, No. AIAA-82-1609, San
Diego, CA, AIAA, August 1982, pp. 571578.
[30] J. Gibson, The Development of Alternate Criteria for FBW Handling Qualities, tech. rep.,
AGARD-508, October 1990.
[31] anon, Department of Defense, Joint Service Specication Guide, Air Vehicle, Rept. JSSG-
2001A, October 2002.
[32] L. R. N. Newcome, Unmanned Aviation, A Brief History of Unmanned Aerial Vehicles.
AIAA, 1st ed., 2004.
[33] anon, Unmanned Aerial Vehicles Roadmap, 2002-2027, Ofce of the Secretary of Defense,
December 2002.
[34] T. H. Cox, C. J. Nagy, M. A. Skoog, I. A. Somers, and R. Warner, A Report of the Civil
UAV Capability Assessment, NASA Aeronautics Research Mission Directorate, December
2004.
[35] anon, RPV Flying Qualites Design Critera, Technical Report AFFDL-TR-76-T25, Rock-
196
Bibliography
well International Corperation, Columbus, OH, December 1976.
[36] K. Breneman and M. Lower, Flying Qualities of Remotely Piloted Vehicle, No. AIAA-
1992-4083, Patuxent River, MD, U. S. Navy, Naval Air Warfare Center, 1992.
[37] W. Williams, A Simulator Based Approach to DT&E of Small UAVs, Masters thesis, Uni-
versity of South Australia, Mawson Lakes, South Australia 5095, December 2002.
[38] W. Williams and M. Harris, The Challenges of Flight-Testing Unmannded Air Vehicles,
Systems Engineering Test and Evaluation Conference, Sydney, Australia, October 2002.
[39] G. R. Spedding and P. B. S. Lissaman, Techical Aspects of Microscale Flight Systems,
Journal of Avian Biology, Vol. 29, December 1998, pp. 458468.
[40] W. Williams, UAV Handling Qualities..... You Must be Joking, Aerospace Sciences Cor-
peration Pty. Ltd., 2003.
[41] T. M. Foster and W. J. Bowman, Dynamic Stability and Handling-Qualities of
Small Unmanned-Aerial-Vehicles, 43rd AIAA Aerospace Sciences Meeting and Exhibit,
No. AIAA-2005-1023, 2005.
[42] J. Hodgkinson, History of Low-Order Equivalent Systems for Aircraft Flying Qualities,
Journal of Guidance, Navigation, Control, and Dynamics, Vol. 28, July 2005, pp. 557583.
[43] National Research Council, Review of NASAs Aerospace Technology Enterprise, An Asses-
ment of NASAs Aeronautics Technology Programs. Washington, D. C.: National Academies
Press, 2004.
[44] S. Pace, X-Planes, Pushing the Envelope of Flight. Zenith Press, 1st ed., October 2003.
[45] A. H. Bowers, J. W. Pahle, R. J. Wilson, B. C. Flick, and R. L. Rood, An Overview of the
NASA F-18 High Alpha Research Vehicle, Technical Memorandum 4772, NASA DFRC,
Edwards, CA, October 1996.
[46] M. E. Peters and D. A. II, The determination of longitudinal ying qualities requirements
for light weight unmanned aircraft, AIAA Guidance, Navigation and Control Conference,
No. AIAA-1997-3701, New Orleans, LA, August 11-13 1997.
[47] B. Mettler and M. B. T. a. Takeo Kanade, System Identication of Small-Size Unmanned
197
Bibliography
Helicopter Dynamics, American Helicopter Society 55th Forum, Montreal, Quebec, Canada,
American Helicopter Society, May 25-27 1999.
[48] S. M. Burk, Jr. and C. F. Wilson, Jr., Radio-Controlled Model Design and Testing Tech-
niques for Stall/Spin Evaluation of General-Aviation Aircraft, SAE National Business Air-
craft Meeting, Wichita, KS, NASA-TM-80510, April 8-11 1975.
[49] C. H. Wolowicz, J. S. Bowman, Jr., and W. P. Gilbert, Similitude Requirements and Scal-
ing Relationships as Applied to Model Testing, Technical Paper TP-1435, NASA LaRC,
Hampton, VA, August 1979.
[50] D. T. McRuer and D. Graham, Eight Years of Flight Control: Triumps and Pitfalls of the
Systems Approach, Journal of Guidance and Control, Vol. 4, July - August 1981, pp. 353
362.
[51] W. L. Brogan, Modern Control Theory. Upper Saddle River, NJ: Prentice-Hall, second ed.,
1982.
[52] D. Gangsaas, K. R. Bruce, J. D. Blight, and U.-L. Ly, Application of Modern Synthesis to
Aircraft Control: Three Case Studies, IEEE Transactions on Automatic Control, Vol. AC-31,
November 1986, pp. 9951014.
[53] M. C. Cotting and J. J. Burken, Recongurable Control Design for the Full X-33 Flight En-
velopw, Technical Memorandum 2001-210396, NASA DFRC, Edwards, CA, August 2001.
[54] J.-J. E. Slotine and W. Li, Applied Nonlinear Control. Upper Saddle River, NJ: Prentice-Hall,
1st ed., 1991.
[55] J. T. Bosworth and P. S. Williams-Hayes, Flight Test Results from the NF-15B Intelligent
Flight Control System (IFCS) Project with Adaption to a Simulated Stabilator Failure, Tech-
nical Memorandum 2007-214629, NASA DFRC, Edwards, CA, December 2007.
[56] J. T. Bosworth, Flight Results of the NF-15B Intelligent Flight Control System (IFCS) Air-
craft with Adaption to a Longitudinally Destabilized Plant, AIAA Guidance, Navigation and
Control Conference and Exhibit, No. AIAA-2008-6985, Honolulu, Hawaii, August 18-21
2008.
198
Bibliography
[57] F. Soares, J. J. Burken, and T. Marwala, Neural Infomation Processing, Vol. 4234/2006 of
Lecture Notes in Computer Science, ch. Neural Network Applications in Advanced Aircraft
Flight Control System, a Hybrid System, a Flight Test Demonstration, pp. 684691. Springer,
October 2006.
[58] J. E. Tomayko, The Story of Self-Repairing Flight Control Systems. No. 1 in Dryden Histor-
ical Study, Edwards, CA: NASA Dryden Flight Research Center, 2003.
[59] G. G. Balas and J. Hodgkinson, Control Design Methods for Good Flying Qualities, AIAA
Atmospheric Flight Mechanics Conference and Exhibit, No. AIAA 2009-6319, Chicago, Il,
AIAA, August 10-13 2009.
[60] K. A. Wise, E. Lavretsky, and N. Hovakimyan, Adaptive Control in Flight: Theory, Applica-
tion, and Open Problems, In Proceedings of American Control Conference, 2006, pp. 5996
5971.
[61] C. Cao and N. Hovakimyan, Design and Analysis of a Novel L
1
Adaptive Controller, Part
I: Control Signal and Asynptotic Stability, Proceedings of the 2006 American Control Con-
ference, June 14-16 2006.
1
Adaptive Controller, Part II:
Guaranteed Transient Performance, Proceedings of the 2006 American Control Conference,
June 14-16 2006.
[63] C. Cao and N. Hovakimyan, A Novel L
1
Neural Network Adaptive Control Architecture
with Guaranteed Transient Performance, IEEE Transactions on Neural Networks, Vol. 18(4),
July 2007, pp. 11601171.
[64] C. Cao and N. Hovakimyan, Guaranteed Transient Performance with L
1
Adaptive Con-
troller for Systems with Unknown Time-Varying Parameters: Part I, Proc. of American
Control Conference, New York, NY, July 2007, pp. 39253930.
[65] C. Cao and N. Hovakimyan, Stability Margins of L
1
Adaptive Controller, Part II, American
Controls Conference, 2007.
[66] C. Cao, V. Patel, K. Reddy, N. Hovakimyan, and E. Lavretsky, Are the Phase and Time-delay
199
Bibliography
Margins Always Adversely Affected by High-gain?, Proc. of AIAA Guidance, Navigation
and Control Conf., 2006. AIAA-2006-6347.
[67] C. Cao and N. Hovakimyan, L
1
Adaptive Output Feedback Controller for Systems with
Time-varying Unknown Parameters and Bounded Disturbances, American Control Confer-
ence, 2007, pp. 486491.
1
Adaptive Output Feedback Controller for Systems of Un-
known Dimension, American Control Conference, 2007, pp. 11911196.
[69] C. Cao, N. Hovakimyan, I. Kaminer, V. Patel, and V. Dobrokhodov, Stabilization of Cas-
caded Systems via L
1
Adaptive Controller with Application to a UAV Path Following Prob-
lem and Flight Test Results, American Control Conference, 2007, pp. 17871792.
[70] C. Cao, N. Hovakimyan, and E. Lavretsky, Application of L
1
Adaptive Controller to Wing
Rock, In Proc. of AIAA Guidance, Navigation and Control Conference, 2006. AIAA 2006-
6426.
[71] J. Wang, C. Cao, V. Patel, N. Hovakimyan, and E. Lavretsky, L
1
adaptive neural network
controller for autonomous aerial refueling with guaranteed transient performance, In Proc.
of AIAA Guidance, Navigation and Control Conference, 2006. AIAA 2006-6206.
[72] J. Wang, V. Patel, C. Cao, N. Hovakimyan, and E. Lavretsky, L
1
adaptive neural network
controller for autonomous aerial refueling in the presence of unknown actuator failures, In
Proc. of AIAA Guidance, Navigation and Control Conference, 2007. AIAA 2007-6313.
[73] R. Beard, N. Knoebel, C. Cao, N. Hovakimyan, and J. Matthews, An L
1
Adaptive Pitch
Controller for Miniature Air Vehicles, Proc. of AIAA Guidance, Navigation & Control Conf.,
Keystone, CO, 2006. AIAA-2006-6777.
[74] Y. Lei, C. Cao, E. Cliff, N. Hovakimyan, A. Kurdila, M. Bolender, and D. Doman, Design of
an L
1
Adaptive Controller for an Air-Breathing Hypersonic Vehicle Model with Unmodelled
Dynamics, Proc. of AIAA Guidance, Navigation & Control Conf., Hilton Head Island, SC,
2007. AIAA 2007-6527.
[75] I. Gregory, V. Patel, C. Cao, and N. Hovakimyan, L
1
Adaptive Control Laws for Flexible
200
Bibliography
Wind Tunnel Model of High-Aspect Ratio Flying Wing, Proc. of AIAA Guidance, Naviga-
tion & Control Conf., Hilton Head Island, SC, 2007. AIAA 2007-6525.
[76] V. V. Patel, C. Cao, N. Hovakimyan, K. Wise, and E. Lavretsky, L
1
Adaptive Controller for
Tailless Unstable Aircraft, In Proc. of American Control conference, 2007, pp. 52725277.
[77] V. Patel, C. Cao, N. Hovakimyan, K. Wise, and E. Lavretsky, L
1
Adaptive Controller for
Tailless Unstable Aircraft in the Presence of Unknown Actuator Failures, Proc. of AIAA
Guidance, Navigation and Control Conf., Hilton Head Island, SC, 2007. AIAA 2007-6314.
[78] National Academy of Engineering, Educating the Engineer of 2020. National Academies
Press, 1st ed., 2005.
[79] National Research Council, How People Learn. Washington, D. C.: National Academy Press,
expanded ed., 2000.
[80] M. Weimer, Learner-Centered Teaching. San Franscisco, CA: Jossey Bass, A Wiley Com-
pany, 1st ed., 2002.
[81] K. A. Smith, S. D. Sheppard, D. W. Johnston, and R. T. Johnson, Pedagogies of Engage-
ment: Classroom-Based Practicies, Journal of Engineering Education, Vol. 94, January
2005, pp. 87101.
[82] H. S. Barrows, Problem-Based Learning in Medicine and Beyond: A Brief Overview, New
Directions for Teaching and Learning, Vol. 1996, No. 68, 1996, pp. 312.
[83] M. P. Fried, R. Satava, S. Weghorst, A. G. Gallagher, C. Sasaki, D. Ross, M. Sinanan, J. I.
Uribe, M. Zeltsan, H. Arora, and H. Cuellar, Identifying and Reducing Errors with Surgical
Simulation, Quality and Safety in Healthcare, Vol. 13, No. Supplement 1, 2004, pp. 1926.
[84] A. Liu, F. Tendick, K. Cleary, and C. Kaufmann, A Survey of Surgical Simulation: Appli-
cations, Technology, and Education, Prescence, Vol. 12, December 2003, pp. 599614.
[85] M. Wesch, Anti-Teaching: Confronting the Crisis of Signicance, Education Canada,
Vol. 48, Spring 2008, pp. 47.
[86] L. Meirovitch, Methods of Analytical Dynamics. Mineloa, New York: Dover, 1998.
[87] C. D. Perkins and R. E. Hage, Airplane Performance Stability and Control. Princeton, New
201
Bibliography
Jersey: John Wiley and Sons, 1949.
[88] D. McRuer, I. Ashkenas, and D. Graham, Aircraft Dynamics and Automatic Control. Prince-
ton, New Jersey: Princeton University Press, 1973.
[89] B. Stevens and F. Lewis, Aircraft Control and Simulation. New York, NY: John Wiley and
Sons, 1st ed., 1992.
[90] R. E. McFarland, A Standard Kinematic Model for Flight Simulations at NASA-Ames,
Contractor Report CR-2497, NASA ARC, Moffett Field, Ca, January 1975.
[91] K. Ogata, Modern Control Engineering. Upper Saddle River, NJ: Prentice-Hall, third ed.,
1997.
[92] D. Kincaid and W. Cheney, Numerical Analysis. Pacic Grove, CA: Brooks/Cole Publishing,
1st ed., 1991.
[93] R. H. Hoh, D. G. Mitchell, and J. Hodgkinson, Bandwidth - A Criterion for Highly Aug-
mented Airplanes, AGARD Conference Proceedings No.333, Criteria for Handling Quali-
ties of Military Aircraft, Ft. Worth, Texas, NATO, April 1982.
[94] anon, Unmanned Aerial Vehicles Roadmap, 2005-2030, Ofce of the Secretary of Defense,
August 4 2005.
[95] anon, Unmanned Systems Roadmap, 2007-2032, Ofce of the Secretary of Defense, De-
cember 10 2007.
[96] anon, Unmanned Systems Integrated Roadmap, 2009-2034, Ofce of the Secretary of De-
fense, 2009.
[97] J. D. Anderson, Jr., Introduction to Flight. New York, NY: McGraw Hill, 5th ed., 2005.
[98] K. Munson and M. Streetly, Janes Unmanned Aerial Vehicles and Targets. Janes Informa-
tion Group, 24 th ed., May 2005.
[99] J. Pike, http://www.globalsecurity.org/. May 2008.
[100] A. L. Velocci,Jr, Aerospace Sourcebook. New York, NY: Aviation Week and Space Technol-
ogy, McGraw Hill, January 2008.
[101] A. L. Velocci,Jr, Aerospace Sourcebook. New York, NY: Aviation Week and Space Technol-
202
Bibliography
ogy, McGraw Hill, January 2007.
[102] T. R. Yechout, S. L. Morris, D. E. Bossert, and W. F. Hallgren, Introduction to Aircraft Flight
Mechanics: Performance, Static Stability, Dynamic Stability, and Classic Feedback Control.
Reston, Va.: AIAA Education Series, 1st ed., 2003.
[103] U. S. Department of Labor, CPI Ination Calculator,
http://www.bls.gov/data/ination calculator.htm, November 2010.
[104] anon, MIL-HDBK-1797. Department of Defense, December 19 1997.
[105] anon, Aerospace - Flight Control Systems - General Specication for Design, Installation
and Test of Piloted Military Aircraft, Aerospace Standard AS94900, SAE Aerospace, July
2007.
[106] anon, Aeronautical Design Standard Performance Specication Handling Qualities Require-
ments for Military Rotorcraft, Design Standard ADS-33E-PRF, US Army Aviation and Mis-
sile Command, Aviation Engineering Directorate, Redstone Arsenal, Alabama, March 2000.
[107] D. J. Moorhouse and R. J. Woodcock, Background Information and User Guide for Mil-F-
8785C), Military Specication-Flying Qualities of Piloted Airplanes, Tech. Rep. AFWAL-
TR-81-3109, Air Force Flight Dynamics Laboratory, Wright Patterson AFB, OH, July 1982.
[108] W. Bihrle, A Handling Qualities Theory for Precise Flight Path Control, Tech. Rep.
AFFDL-TR-65-195, AFRL, Wright Patterson AFB, OH, June 1965.
[109] W. F. Philips and R. J. Niewoehner, Characteristic Length and Dynamic Time Scale As-
sociated with Aircraft Pitching Motion, Journal of Aircraft, Vol. 46, March-April 2009,
pp. 572582.
[110] J. J. Harris and G. T. Black, YF-22 Control Law Development and Flying Qualities, At-
mospheric Flight Mechanics Conference, No. AIAA-96-3379, San Diego, CA, AIAA, July
29-31 1996.
[111] D. A. DiFranco, Higher-Order Control System Dynamics and Longitudinal Handling Qual-
ities, Journal of Aircraft, Vol. 7, September - October 1970, pp. 457464.
[112] T. P. Neal and R. E. Smith, An In-Flight Investigation to Develop Control System Design
203
Bibliography
Criteria for Fighter Airplanes, Tech. Rep. AFFDLTR70-074, AFRL, December 1970.
[113] J. Hodgkinson, The Application of Equivalent Systems to MIL-F-8785B, Contrails, De-
cember 1978, pp. 203218.
[114] J. Hodgkinson, Equivalent Systems Criteria for Handling Qualities of Military Aircraft,
AGARD Conference Proceedings No.333, Criteria for Handling Qualities of Military Air-
craft, Ft. Worth, Texas, NATO, April 1982.
[115] J. R. Wood and J. Hodgkinson, Denition of Acceptable Levels of Mismatch for Equiva-
lent Systems of Augmented CTOL Aircraft, Tech. Rep. MDC-A6792, McDonnell Aircraft
Company, St. Louis, MO, September 17 1980.
[116] P. Fitzgerald, Modern Flight Control System Design for the Jindivik UAV, Masters thesis,
Craneld University, Craneld, Bedfordshire, UK, September 2000.
[117] D. G. Mitchell and D. H. Klyde, A Critical Examination of PIO Prediction Criteria, AIAA
Atmospheric Flight Mechanics Conference and Exhibit, No. AIAA-98-4335, Boston, MA,
[118] D. G. Mitchell and R. H. Hoh, Development of a Unied Method to Predict PIO, AIAA
Atmospheric Flight Mechanics Conference, No. AIAA-96-3435, San Diego, CA, AIAA, July
29-31 1996.
[119] J. C. Gibson, Piloted Handling Qualities Design Criteria for High Order Flight Control Sys-
tems, AGARD Conference Proceedings No.333, Criteria for Handling Qualities of Military
Aircraft, Ft. Worth, Texas, NATO, April 1982.
[120] R. H. Hoh and D. G. Mitchell, The Role of Handling Qualities Specications in Flight Con-
trol SystemDesign, Active Control Technology: Applications and Lessons Learned (L. M. B.
d. Costa Campos and H. Wunnenberg, eds.), Vol. 1, Turin, Italy, NATO, Advisory Group for
Aerospace Research and Development (AGARD), May 9-13 1994, pp. 11 112.
[121] S. Jacobson and J. Moynes, Ride Qualities Criteria for the B-2 Bomber, AIAA / AHS /
ASEE Aircraft Design, Systems, and Operations Meeting, No. AIAA-90-3256, Dayton, OH,
September 17-19 1990.
204
Bibliography
[122] NASALaRC, Proceedings of Conference on Ride Qualities of air, surface, and water vehicles
with emphasis on human factors engineering for comfort in transportation, No. NASA-TM-
X-2620, Hampton, VA, October 1 1972.
[123] J. W. Rustenburg, Ride Quality Design Criteria for Aircraft with Active Mode Control Sys-
tems, Technical Report ASD-TR-72-64, Aeronautical Systems Division, Wright-Pattersom
Air Force Base, Wright Patterson AFB, OH, October 1972.
[124] D. G. Mitchell, R. H. Hoh, B. L. Aponso, and D. H. Klyde, Proposed Incorporation of
Mission-Oriented Flying Qualities into MIL-STD-1797A, Technical Report WL-TR-94-
3162, Wright Laboratory, Wright Patterson AFB, OH, October 1994.
[125] D. J. Wilson, D. R. Riley, and K. D. Citurs, Aircraft Maneuvers for the Evaluation of Flying
Qualities and Agility, Tech. Rep. WL-TR-93-3081, Wright Laboratory, Wright Patterson
AFB, OH, 1993.
[126] anon, Flying Qualities of Piloted Aircraft, Department of Defense Interface Standard MIL-
STD-1797B, Department of Defense, Wright Patterson AFB, OH, February 15 2006.
[127] J. Hodgkinson, J. R. Wood, and R. H. Hoh, An Alternate Method of Specifying Bandwidth
for Flying Qualities, AIAA Guidance, Navigation and Control Conference, No. AIAA-82-
1609, San Diego, CA, AIAA, August 1982, pp. 571578.
[128] M. C. Cotting, A. Wolek, J. F. Murtha, and C. A. Woolsey, Developmental Flight Testing of
the SPAARO UAV, 48th Annual AIAA Aerospace Sciences Meeting and Exhibit, No. AIAA-
2010-295, Orlando, FL, January 4-7 2010.
[129] J. F. Murtha, M. C. Cotting, A. Wolek, T. Aarons, and C. A. Woolsey, The Educational
Impact of Creating a NewUAVfor CurriculumEnhancement, 27th AIAA Atmospheric Flight
Mechanics Conference and Exhibit, No. AIAA-2009-5851, Chicago, Il, August 10-13 2009.
[130] J. Roskam, Airplane Design, Vol. VI. Ottowa, Kansas: Roskam Aviation and Engineering
Corperation, 1986.
[131] H. L. Dryden, A Review of the Statistical Theory of Turbulence, Quarterly of Applied
Mathematics, Vol. 1, April 1943, pp. 742.
205
Bibliography
[132] T. V. Karman, Progress in the Statistical Theory of Turbulence, Proceedings of the National
Academy of Sciences, Vol. 34, August 1948, pp. 530539.
[133] J. C. Houbolt, R. Stiener, and K. G. Pratt, Dynamic Response of Airplanes to Atmospheric
Turbulence Including Flight Data on Input and Response, Technical Report TR-R-19, NASA
LaRC, Hampton, VA, June 1964.
[134] M. C. Cotting, An Initial Study to Categorize Unmanned Aerial Vehicles for Flying Qualities
Evaluation, 47th Annual AIAA Aerospace Sciences Meeting and Exhibit, No. AIAA-2009-
307, Orlando, FL, January 5 - 8 2009.
[135] L. T. Nguyen, M. E. Ogburn, W. P. Gilbert, K. S. Kibler, P. W. Brown, and P. L. Deal, Simu-
lator Study of Stall/Post Stall Characteristics of a Fighter Airplane with Relaxed Longitudinal
Static Stability, Technical Paper 1538, NASA LaRC, Hampton, VA, December 1979.
[136] G. P. Walker and D. A. Allen, X-35B STOVL Flight Control Law Design and Flying Qual-
ities, Biennial International Powered Lift Conference and Exhibit, No. AIAA-2002-6018,
Williamsburg, VA, AIAA, November 5-7 2002.
[137] R. Wacker, D. F. Enns, D. J. Bugajski, S. Munday, and S. Merkle, X-38 Application of Dy-
namic Inversion Flight Control, Advances in the Astronautical Sciences, Vol. 107, No. AAS
01-016, 2001, pp. 233250.
[138] Anon, Application of Multivariable Control Theory to Aircraft Control Laws, Technical
Report WL-TR-3099, Wright Laboratory, Dayton, OH, May 1996.
[139] K. Wise and E. L. e. al, Adaptive Flight Control of a Sensor Guided Munition., Proc. of
AIAA GNC Conf., 2005.
[140] C. Cao and N. Hovakimyan, Stability Margins of L
1
Adaptive Controller: Part II, Proc. of
American Control Conference, New York, NY, July 2007, pp. 39313936.
1
Adaptive Control Architec-
ture, Part I: Control Signal and Asymptotic Stability, Proc. of American Control Conference,
Minneapolis, MN, June 2006, pp. 33973402.
1
Adaptive Control Archi-
206
Bibliography
tecture, Part II: Guaranteed Transient Performance, Proc. of American Control Conference,
Minneapolis, MN, June 2006, pp. 34033408.
1
Adaptive Controller for Systems with Unknown Nonlinear-
ities: Part I, Submitted to IEEE Conference on Decision and Control, 2007.
1
Adaptive Controller for Nonlinear Systems in the Presence
of Unmodelled Dynamics: Part II, Submitted to IEEE Conference on Decision and Control,
2007.
1
Adaptive Controller for Systems in the Presence of Unmod-
eled Actuator Dynamics, Submitted to IEEE Conference on Decision and Control, 2007.
1
Adaptive Controller for Multi-Input Multi-Output Systems
in the Presence of Unmatched Disturbances, Submitted to IEEE Conference on Decision and
Control, 2007.
[147] I. Kaminer, O. Yakimenko, V. Dobrokhodov, A. Pascoal, N. Hovakimyan, V. V. Patel, C. Cao,
and A. Young, Coordinated path following for time-critical missions of multiple UAVs via
L
1
adaptive output feedback controllers, AIAA Guidance, Navigation and Control Confer-
ence, 2007.
[148] B. L. Stevens and F. L. Lewis, Aircraft Control and Simulation. Wiley, 2 ed., 2004.
[149] K. C. Scalera and W. C. Durham, Modication of a Surplus Navy 2F122A A-6E OFT
for Flight Dynamics Research and Instruction, AIAA Modeling and Simlation Technologies
Conference and Exhibit, No. AIAA-1998-4180, Boston, MA, AIAA, August 10-12 1998.
[150] J.-B. Pomet and L. Praly, Adaptive Nonlinear Regulation: Estimation from the Lyapunov
Equation, IEEE Transactions on Automatic Control, Vol. 37, June 1992, pp. 729 740.
[151] M. C. Cotting and T. H. Cox, A Generic Guidance and Control Structure for Six-Degree-
of-Freedom Conceptual Aircraft Design, AIAA Aerospace Sciences Meeting and Exhibit,
No. AIAA20050032, Reno, NV, January 2005.
[152] H. K. Khalil, Nonlinear Systems. Upper Saddle River, NJ: Prentice-Hall, third ed., 2002.
[153] A. Isidori, The matching of a prescribed linear input-output behavior in a nonlinear system,
207
Bibliography
Automatic Control, IEEE Transactions on, Vol. 30, Mar 1985, pp. 258265.
[154] D. F. Enns, D. J. Bugajski, R. Hendrick, and G. Stein, Dynamic Inversion: An Evolving
Methodology for Flight Control Design, Internation Journal of Control, Vol. 59, January
1994, pp. 7179.
[155] M. Rais-Rohani, K. Koenig, and T. Hannigan, Keeping students engaged: an overview of
three introductory courses in aerospace engineering, 2003 ASEE Annual Conference and
Exposition, Nashville, Tennessee, 2003.
[156] H. C. Smith, Teaching ight test engineering with a PC-based simulator, 2001 ASEE An-
nual Conference and Exposition, Albuquerque, New Mexico, 2001.
[157] T. Liu and M. Schulte, Flight Testing Education at Western Michigan University, 45th AIAA
Aerospace Sciences Meeting and Exhibit, Reno, Nevada, 2007.
[158] M. C. Cotting, L. S. McCue, and W. C. Durham, The instructional design and redesign of an
undergraduate-level, simulator-based course on Flight Test Techniques, ASEE Conference
and Exposition, No. AC2007-93, Honolulu, Hawaii, June 2007.
[159] M. C. Cotting, L. S. McCue, and W. C. Durham, Simulator-based ight test engineering as
a capstone to the dynamics and control curriculum, 45th AIAA Aerospace Sciences Meeting
and Exhibit, No. AIAA-2007-0699, Reno, NV, January 2007.
[160] R. M. Felder and R. Brent, Understanding Student Differences, Journal of Engineering
Education, Vol. 94, January 2005, pp. 5772.
[161] W. B. Spaulding, The Undergraduate Medical Curriculum (1969 Model) McMaster Univer-
sity, Canadian Medical Association Journal, Vol. 100, April 1969, pp. 659664.
[162] S. Muller, Physicians for the Twenty-First Century: Report of the Project Panel on the Gen-
eral Professional Education of the Physician and College Preparation for Medicine. No. 59,
Journal of Medical Education, 1984.
[163] A. B. Rosenstein, A Study of Profession and Professional Education, Tech. Rep. EDP 7-68,
School of Engineering and Applied Science, University of California, Los Angeles, 1968.
[164] C. L. Dym, A. M. Agogino, O. Eris, D. D. Frey, and L. J. Leifer, Engineering Design Think-
208
Bibliography
ing, Teaching, and Learning, Journal of Engineering Education, Vol. 94, January 2995,
pp. 103120.
[165] K. C. Cohen, Internet links for science education: student-scientist partnerships. Plenum
Press, 1997.
[166] K. A. Bruffee, Collaborative Learning, Higher Education, Interdependence, and the Author-
ity of Knowledge. Johns Hopkins University Press, 1998.
[167] J. R. Katzenbach and D. K. Smith, The Wisdom of Teams: Creating the High-Performance
Organization. Harvard Business School Press, 1st ed., 1993.
[168] K. Rohnke and S. Butler, Quicksilver. Project Adventure, Inc., Dubuque, Iowa: Kendall/Hunt
Publishing Co., 1st ed., 1995.
[169] B. M. Slator and H. C. Chaput, Learning by Learning Roles: A Virtual Role-Playing Envi-
ronment for Tutoring, Intelligent Tutoring Systems, 1996, pp. 668676.
[170] B. M. Slator, J. Clark, P. McClean, B. Saini-Eidukat, and A. R. White, Research on role-
based learning technologies, Proceedings. IEEE International Conference on Advanced
Learning Technologies, IEEE, Madison, WI 2001, pp. 3740.
[171] R. E. Mayer, Multimedia Learning. Cambridge University Press, rst ed., 2001.
[172] R. Beck and W. C. Durham, Flight simulation in ight dynamics education, AIAA At-
mospheric Flight Mechanics Conference and Exhibit, No. AIAA-1999-4361, Portland, OR,
[173] G. L. Gallagher, L. B. Higgins, L. A. Khinoo, and P. W. Pierce, U.S. Naval Test Pilot School
Flight Test Manual, Fixed Wing Performance, Tech. Rep. USNTPS-FTM-NO. 108 (Prelim-
inary), Veda Incorporated Contract N00421-90-C-0022, 30 September 1992.
[174] Aircraft Division, U.S. Naval Test Pilot School Flight Test Manual, Fixed Wing Stability
and Control: Theory and Flight Test Techniques, Tech. Rep. USNTPS-FTM-No. 103, Naval
Air Warfare Center, Patuxent River, Maryland, January 1997.
[175] B. Bloom and D. Krathwohl, Taxonomy of educational objectives, Handbook 1: Cognitive
domain. New York: Addison-Wesley, 1984.
209
Bibliography
[176] P. C. Wankat and F. S. Oreovicz, Teaching Engineering. 1992. Reprinted at https:
//engineering.purdue.edu/ChE/News_and_Events/Publications/
teaching_engineering/index.html.
[177] R. Felder and R. Brent, Objectively speaking, Chemical Engineering Education, Vol. 31,
No. 3, 1997, pp. 178179.
[178] J. E. Stice, A First Step Toward Improved Teaching, Engineering Education, Vol. 66,
February 1976, pp. 394398.
[179] N. Gronlund, How to write and use instructional objectives. Upper Saddle River, NJ:
Prentice-Hall, 6 ed., 2000.
[180] M. J. Prince and R. M. Felder, Inductive Teaching and Learning Methods: Denitions,
Comparisons, and Research Bases, Journal of Engineering Education, Vol. 95, April 2006,
pp. 123138.
[181] N. D. Fleming and C. Mills, Not another inventory, rather a catalyst for reection, To
Improve the Academy, Vol. 11, 1992.
[182] S. Montgomery, Chemical Engineering 580/Engineering 580: Teaching Engineering, Lec-
ture Notes, Winter 2004.
[183] anon, U.S. Military Specication MIL-F-8785C. Department of Defense, 5 November 1980.
[184] M. C. Cotting, J. F. Murtha, L. Techy, and C. A. Woolsey, Examples of Augmentation of
an Atmospheric Flight Mechanics Curriculum using UAVs, 27th AIAA Atmospheric Flight
Mechanics Conference and Exhibit, No. AIAA 2009-5852, Orlando, FL, August 10-13 2009.
210
Appendix A
Self Assessment Surveys for AOE4984
Flight Test Techniques
A.1 Student survey
The surveys were anonymous and contained both short answer and questions with a rating scale
from 1 to 5 to rate pre-written statements. A total of 39 students completed the surveys over the
course of two course offerings. Selected results from these surveys are listed below:
1. This class required time than my other classes.
Considerably More
1
5
Considerably Less
2. I learned than I expected from this class.
Considerably More
1
5
Considerably Less
3. The pacing of the course was:
211
A.1 Student survey
Too Fast
1
5
Too Slow
4. Basing my grade solely on the ight test reports reects my understanding of the material
presented in this class.
Strongly Agree
1
5
Strongly Disagree
5. I would recommend this class to my peers.
Strongly Agree
1
5
Strongly Disagree
6. This class has improved my overall understanding of aerospace engineering.
Strongly Agree
1
5
Strongly Disagree
7. I better understand the theories and concepts taught in my other classes based on the topics
taught in this course.
Strongly Agree
1
5
Strongly Disagree
8. I discussed this class with people outside of my test group.
Often
1
5
Never
212
A.1 Student survey
9. I worked than my other classes on this course.
Harder
1
5
Less
10. I have a understanding of the basic concepts involved in ight testing aircraft
after taking this course.
Excellent
1
5
Poor
11. My Overall Rating of this course is:
Excellent
1
5
Poor
A sampling of written Comments from the students survey are also included below:
1. This class actually covered material and procedures required in the Real World, not just the
theory behind it.
2. Actually applying knowledge learned in other classes
3. I got into engineering to be hands on and this was one of the best ways to do it.
4. It provided a new conduit through which to demonstrate the material taught through the AE
curriculum
5. Helps you see how everything from previous courses comes together.
6. It gives a feel for mathematical values.
213
A.2 Post graduation
7. It reinforced topics learned in other courses and gave those concepts real-world meaning.
8. I feel that this course should be required at the junior or senior level because it brings hands-
on, real meaning to all of the theory and number crunching we do in other classes. This was
one of the few classes where I actually got to do engineering .
9. I liked that the course required us to come up with our own ight test cards. It really made
us think about what we were doing.
10. This was one of the most interesting and practical courses Ive taken. It showed us why we
need to know what we are learning.
11. My favorite course of my 5 years here. I learned more in this class than any other. The rst
time I was excited to be an engineer.
A.2 Post graduation
An attempt was made to contact students that have taken the ight test techniques course after at
least one year past course completion. Seventeen students that were contacted completed another
anonymous survey. Again the survey was a mixture of short answer and questions with a rating
scale from 1 to 5 to rate pre-written statements.
1. I am currently working in the ight test eld
6 True, 11 False
2. AOE4984 Flight Test Techniques was discussed in the interview for my current position
11 True, 6 False
3. My understanding of basic aerospace engineering principals was enhanced by taking AOE4984
Flight Test Techniques.
214
A.2 Post graduation
Strongly Agree
1
5
Strongly Disagree
For the next set of questions please use the following denitions in responding:
PROCEDURAL understanding: a persons ability to use algorithms, rules, or formal lan-
guage to represent a system.
CONCEPTUAL understanding: a persons ability to represent major concepts within a sys-
tem.
4. My PROCEDURAL understanding of aerospace engineering principals was enhanced by tak-
ing AOE4984 Flight Test Techniques.
Strongly Agree
1
5
Strongly Disagree
5. My CONCEPTUAL understanding of aerospace engineering principals was enhanced by tak-
ing AOE4984 Flight Test Techniques.
Strongly Agree
1
5
Strongly Disagree
6. My current position requires me to have a strong PROCEDURAL understanding of aerospace
engineering principals.
Strongly Agree
1
5
Strongly Disagree
7. My current position requires me to have a strong CONCEPTUAL understanding of aerospace
engineering principals.
215
A.2 Post graduation
Strongly Agree
1
5
Strongly Disagree
8. The use of a ight simulator enhanced my understanding of aerospace engineering concepts.
Strongly Agree
1
5
Strongly Disagree
9. The teaming arrangements of AOE4984 Flight Test Techniques prepared me for my current
work environment.
Strongly Agree
1
5
Strongly Disagree
10. Exposure to a ight simulator earlier in the aerospace engineering curriculum would have
increased my understanding of aerospace engineering concepts.
Strongly Agree
1
5
Strongly Disagree
11. This course played a role in impressing a potential employer during the interview
process.
Signicant
1
5
Insignicant
1. Flight Test was one of the few labs that I understood not only what we were testing for
216
A.2 Post graduation
and what we were supposed to get, but much more importantly I understood why I got what
I did and why it agreed or disagreed with the expected answer. Using a ight simulator
allowed the strongest correlation between text book ideal cases and real world relevance and
considerations. This should be a mandatory course for all AOE majors during their sophomore
year. You can talk concepts and procedures all day, but without some rst hand knowledge
or experience it is very difcult to turn the text-book denitions in to real life effects and
applications. Within the rst week of any AOE course you can easily pick out the majority of
people with and without ight experience.
2. The ight simulator provided a great hands-on experience to reinforce concepts learned in
lecture. An hour ying the simulator experiencing the different modes of stability was com-
parable in value to weeks of lecture. Being able to calculate a max range climb/cruise/descent
prole in an effort to make it to an airstrip, and then successfully y the prole with a limited
amount of fuel was much better than any scenario that could be presented in a Performance
class.
3. Looking back on the class, everything in the real life ight test eld is exactly like the class.
4. Using a simulator to learn about ight testing is ideal. It allowed us to go through everything
we would have for a real ight test without the need to worry as much about safety and cost.
5. Extensive teamwork was required to complete the work required for the mission on the
simulator, as well as complete the report. The same type of teamwork is required of me in my
current job.
6. I came to work with a good understanding of the team mentality that is critical to perform
my job effectively, and also gave me real life examples of problems that could occur.
7. I work with teams of various people almost everyday at my job. Without working together
not much would get done. Time working on teams in college helps prepare for this.
217
A.2 Post graduation
8. I currently work in a team environment. Being able to communicate and work with others
with a variety of personality traits is difcult. The team arrangement in this class prepared
me for my current work environment.
9. I think it was the most important class that I took at Virginia Tech. It exposed me to the type
of job I wanted as well as enabling me to get a better grasp on the concepts that I had spent
the last 3 years of study learning about.
10. The ight test class made for an excellent capstone to my engineering degree and allowed
me to put everything I had learned into practice and apply to practical applications. It was
the rst class that made me honestly feel like I was in the right major, which is a shame
considering it was taken in my last semester.
11. The simulator was the best hands on ight experience that we received with our degree. Far
too often we would learn a topic in the classroom with no way to apply it. I personally think
it is essential for engineers to be in touch with what is actually going on with the operational
side of the aircraft we are designing.
12. While I would have thought the International Design Team would have been the main topic
of my interviews, the project was barely mentioned after they asked about Flight Test Tech-
niques. My rst three job offers were for ight testing positions, and Im certain I wouldnt
have been offered those opportunities without this course.
13. Flight test techniques was without a doubt my favorite class I took in the aerospace curricu-
lum. It helped reinforce the principles learned in other classes and gave me the connection to
the real world that I needed. Too much of the aerospace curriculum is abstract, ight test was
much more realistic.
14. The ight test class alone made a greater impact on my desire to be a test pilot more than all
the other classes put together.
218
A.3 Employer
15. I think ight simulator assignments should be incorporated into intro to aerospace engineer-
ing and in aircraft ight dynamic courses... I can honestly say I didnt understand several
aerospace principals until I was exposed to an actual aircraft. I feel that it is highly unlikely
that the department can expose students to real aircraft, but I think a simulator could help ll
the gap.
16. Ive been to a few performance ight tests and they are exactly how I remember them from
the ight test course. It wasnt until I went to the ight tests that I truly appreciated how close
the ight test class was to the real world.
17. I think the ight test techniques course would be advantageous to any aerospace engineer
whether they are pursuing ight testing or not. For me it has solidied the foundation of my
skills for what I want to continue doing and has reminded me why I wanted to become an
engineer in the rst place.
A.3 Employer
An attempt was made to contact employers of former students that have taken the ight test tech-
niques course after at least one year past course completion. Three employers completed the anony-
mous survey. Again the survey was a mixture of short answer and questions with a rating scale from
1 to 5 to rate pre-written statements.
1. AOE4984 Flight Test Techniques was discussed in the interview for the AOE4984 graduate
3 True, 0 False
2. I was impressed that the AOE4984 graduate was exposed to ight testing at an undergraduate
level
Strongly Agree
1
5
Strongly Disagree
219
A.3 Employer
For the next set of questions please use the following denitions in responding:
PROCEDURAL understanding: a persons ability to use algorithms, rules, or formal lan-
guage to represent a system.
CONCEPTUAL understanding: a persons ability to represent major concepts within a sys-
tem.
3. The AOE4984 graduate has a higher PROCEDURAL understanding of aerospace engineer-
ing principals than a typical employee of his tenure.
Strongly Agree
1
5
Strongly Disagree
4. The AOE4984 graduate has a higher CONCEPTUAL understanding of aerospace engineer-
ing principals than a typical employee of his tenure.
Strongly Agree
1
5
Strongly Disagree
5. New hires working for me are required to have a stronger than average PROCEDURAL un-
derstanding of aerospace engineering principals.
Strongly Agree
1
5
Strongly Disagree
6. New hires working for me are required to have a stronger than average CONCEPTUAL un-
derstanding of aerospace engineering principals.
Strongly Agree
1
5
Strongly Disagree
220
A.3 Employer
7. I would prefer that all my new hire engineers have been exposed to ight testing at the under-
graduate level.
Strongly Agree
1
5
Strongly Disagree
8. Exposure to ight testing prior to arrival at my facility has the effectiveness of
AOE4984 graduate as compared to the average new hire in his position.
Strongly Increased
1
5
Strongly Decreased
9. I would be interested in having all my new hires participate in a ight testing experience sim-
ilar to AOE4984 .
Strongly Agree
1
5
Strongly Disagree
1. I cant stress the importance of relevant ight test experience. I go out of my way to nd
students that have had the VT or [another] ight test class because of the knowledge and
experience gained by the students. From the basics of how to develop a test to the specics
of what actually occurs during a ight, the ight test course exposes students to the skill set
necessary to excel at [ORG-X]. [ORG-X] has spoken with several universities in the past and
encouraged their engineering departments to incorporate a class similar to the VT ight test
class. It would be a shame to lose that competitive edge that VT grads have with [ORG-X].
221
A.3 Employer
2. The AOE4984 curriculum clearly prepared my employee for the avionic systems work that
he was hired for. He came in with surprising breadthwise knowledge and was hungry to dig
into the more complex and ambiguous problems our engineers are challenged with in the
industry.
3. This gives the new hire a leg up in bridging the gap between theory and practice. This has
taken some load off of me and our senior leads.
222

Handling Qualiities

Uploaded by

Copyright:

Available Formats

Handling Qualiities

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Handling Qualiities

Uploaded by

Copyright:

Available Formats

Evolution of Flying Qualities Analysis: Problems for a

New Generation of Aircraft

h aircraft sink rate

longitudinal control power,

aircraft load factor per angle of attack

Figure 3.3: Gravity vector in relation to aircraft body coordinates.

= p + tan (q sin +r cos ) (3.40)

= q cos r sin (3.41)

h = usin v sin cos wcos cos (3.48)

KJ) the transfer function can be dened as

above the 180

crossing on the phase plot.

Table 6.1: Gain and phase margin requirements as dened in AS94900.

above the 180

crossing on the phase plot. The gain band-

You might also like