Evaluation of Unmanned Aircraft Flying Qualities Using A Stitched Learjet Model

Air Force Institute of Technology
AFIT Scholar
Theses and Dissertations Student Graduate Works
2-26-2019
Evaluation of Unmanned Aircraft Flying Qualities

Using a Stitched Learjet Model
Patrick M. Callaghan
Follow this and additional works at: https://scholar.afit.edu/etd

Part of the Aviation Commons, and the Navigation, Guidance, Control and Dynamics Commons
Recommended Citation
Callaghan, Patrick M., "Evaluation of Unmanned Aircraft Flying Qualities Using a Stitched Learjet Model" (2019). Theses and
Dissertations. 2213.
https://scholar.afit.edu/etd/2213
This Thesis is brought to you for free and open access by the Student Graduate Works at AFIT Scholar. It has been accepted for inclusion in Theses and
Dissertations by an authorized administrator of AFIT Scholar. For more information, please contact richard.mansfield@afit.edu.
Evaluation of Unmanned Aircraft Flying
Qualities Using a Stitched Learjet Model
THESIS
Patrick M. Callaghan, Captain, USAF

AFIT-ENY-MS-19-M-206
DEPARTMENT OF THE AIR FORCE

AIR UNIVERSITY
AIR FORCE INSTITUTE OF TECHNOLOGY

Wright-Patterson Air Force Base, Ohio
DISTRIBUTION STATEMENT A
APPROVED FOR PUBLIC RELEASE; DISTRIBUTION UNLIMITED.
The views expressed in this document are those of the author and do not reflect the
official policy or position of the United States Air Force, the United States Department
of Defense or the United States Government. This material is declared a work of the
U.S. Government and is not subject to copyright protection in the United States.
EVALUATION OF UNMANNED AIRCRAFT FLYING QUALITIES USING A

STITCHED LEARJET MODEL
THESIS
Presented to the Faculty

Department of Aeronautical Engineering
Graduate School of Engineering and Management
Air University
Air Education and Training Command
in Partial Fulfillment of the Requirements for the
Degree of Master of Science in Aeronautical Engineering
Patrick M. Callaghan, B.S.A.E.

Captain, USAF
February 26, 2019
DISTRIBUTION STATEMENT A

THESIS
Patrick M. Callaghan, B.S.A.E.

Captain, USAF
Committee Membership:
D. L. Kunz, PhD
Chair
R. G. Cobb, PhD
Member
M. F. Reeder, PhD
Member
Abstract
In recent years, military UAVs have taken over missions that were too dull, dirty,
or dangerous for manned aircraft. The increased demand has led to a build-fly-
fix-fly development mentality, plaguing the early lifecycle with staggering mishap
rates. Currently, MIL-STD-1797 lists flying qualities for UAVs as TBD, and the
standards for manned fixed wing are inadequate when applied to UAVs. In an effort
to expand the database of UAV flying qualities, an analysis was completed on a
Simulink model of an LJ-25D developed from Calspans Variable Stability System
aircraft at the United States Test Pilot School. Three maneuvers were simulated:
(1) a non-precision, non-aggressive climbing spiral, (2) a precision, non-aggressive
side step landing, and (3) a precision, non-aggressive aerial refueling task. These
maneuvers were chosen to evaluate the performance and workload of the aircraft as
four stability and control parameters were scaled. The data were utilized in identifying
trends between the scaled stability and control parameters and resulting workload and
performance metrics. Thumbprint plots were generated to identify Level 1, Level 2,
and Level 3 flying qualities and compared to MIL-STD-1797 plots. Results point to
utilizing a combination of classical aircraft literal factors, such as ζsp and CAP, with
newly developed mathematical techniques, such as L2 norm and TIC, to assess the
workload of the flight controller and performance during the maneuver.
iv
Table of Contents
Page
Abstract . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iv
List of Figures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . vii
List of Tables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi
List of Symbols . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii
List of Acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xv
I. Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.1 Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1
1.2 Research Problem Motivation and Description . . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Research Objectives . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Thesis Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
II. Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
2.1 Early Development of Flying Qualities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

2.2 Manned Fixed-Wing Flying Qualities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
2.2.1 Subjective Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2.2.2 Objective Criteria . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
2.3 Manned Rotary-Wing Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
2.4 Unmanned Aerial Vehicle Standards . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
2.5 Previous UAV Flying Qualities Research . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2.6 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
III. Research Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

3.1 Simulink Model of LJ-25D Aircraft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.1.1 Model Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34
3.1.2 Stability Parameter Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.1.3 Lower-Order Equivalent System Development . . . . . . . . . . . . . . . . 37
3.2 Flight Maneuvers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
3.2.1 Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.2.2 Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
3.3 Evaluation of Maneuver Workload and Performance . . . . . . . . . . . . . . . . . 47
3.3.1 Aerial Refueling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
3.3.2 Maneuver Workload Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
3.3.3 Maneuver Performance Metrics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
3.4 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
v
Page
IV. Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.1 Bare Airframe Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 52
4.1.1 Stability Parameter Scaling Analysis . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Climbing Spiral Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.2.1 Baseline Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
4.2.3 Longitudinal Controller Gain Scaling . . . . . . . . . . . . . . . . . . . . . . . . 71
4.3 Side Step Landing Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.1 Baseline Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
4.3.3 Lateral Controller Gain Scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
4.3.4 Cross-Coupling Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91
4.4 Aerial Refueling Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.4.1 Baseline Aerial Refueling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
4.4.3 Deficiency Identification in UAV Flying Qualities . . . . . . . . . . . . 106
4.4.4 Adjusting TIC of Altitude for Flight Test . . . . . . . . . . . . . . . . . . . 108
4.5 Chapter Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
V. Conclusions and Recommendations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.1 Conclusions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
5.2 Future Research Recommendation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 115
5.3 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 117
Appendix A. LOES Bode Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 118
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 126
vi
List of Figures
Figure Page
1 Flight Hours for Unmanned Systems between 1996 and

2011 [51] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
2 U.S. Military Aircraft and UAS Class A Mishap Rates,

1986-2006 [15] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
3 Learjet LJ25 Variable Stability System [8] . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
4 The Stinson Reliant played an important role in

developing early flying qualities metrics [1] . . . . . . . . . . . . . . . . . . . . . . . . . 11
5 The NT-33A aided in the development of the X-15,

A-10, F-15, F-16, F-18, F-117, and F-22 [6] . . . . . . . . . . . . . . . . . . . . . . . . . 12
6 The F-16 VISTA [36] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13
7 Cooper-Harper Pilot Rating Scale [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15
8 ωsp Tθ2 vs ζsp plot depicting how Cooper-Harper ratings

drove objective criteria [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
9 Short-period Dynamic Requirements [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19
10 Short-period Pitch Response to Pitch Controller [5] . . . . . . . . . . . . . . . . . 20
11 Time Domain Transient Response Parameters [5] . . . . . . . . . . . . . . . . . . . 20
12 Aircraft Classification in MIL-STD-1797B [5] . . . . . . . . . . . . . . . . . . . . . . . 23
13 Modified Neal-Smith Design Criteria for Pitch

Dynamics [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24
14 Williams’s UAV Class System [50] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
15 UAV Grouping by Reynolds Number and Aircraft

Weight [17] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29
16 UAV Grouping by Operational Altitude, Weight, and

Airspeed [51] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30
17 Cotting’s Modified Cooper-Harper Rating Scale for

UAVs [18] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
vii
Figure Page
18 Notional Relation between UAV Roles, Mission Task

Elements, and Task Categories [18] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
19 LJ-25D Model stitching block diagram [8] . . . . . . . . . . . . . . . . . . . . . . . . . . 35
20 Envelopes of Maximum Unnoticeable Added Dynamics

Example . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38
21 Block Diagram of Spiral Maneuver Longitudinal

Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
22 Block Diagram of a PID Controller [28] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
23 Block Diagram of Spiral Maneuver Lateral Controller . . . . . . . . . . . . . . . . 43
24 Block Diagram of Spiral Maneuver Sideslip Controller . . . . . . . . . . . . . . . 43
25 Block Diagram of Spiral Maneuver Throttle Controller . . . . . . . . . . . . . . . 44
26 Block Diagram of Side Step Maneuver Lateral Controller . . . . . . . . . . . . . 45
27 Block Diagram of Side Step Maneuver Heading

Controller . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
28 Longitudinal LOES for Baseline Bare Airframe, 200

KCAS 5,000 ft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 53
29 Lateral LOES for Baseline Bare Airframe, 200 KCAS

5,000 ft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
30 Baseline Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
31 Workload and Performance Metrics of Baseline

Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
32 Mq Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
33 Workload and Performance Metrics of Mq Scaled

Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
34 Flying Qualities Graphs of Mq Scaled Climbing Spiral . . . . . . . . . . . . . . . 63
35 Mδ e Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
36 Workload and Performance Metrics of Mδ e Scaled

Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66
viii
Figure Page
37 Flying Qualities Graphs of Mδ e Scaled Climbing Spiral . . . . . . . . . . . . . . 67
38 Lp Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
39 Workload and Performance Metrics of Lp Scaled

Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
40 Lδ a Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
41 Workload and Performance Metrics of Lδ a Scaled

Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
42 Kc Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
43 Workload and Performance Metrics of Kc Scaled

Climbing Spiral . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
44 Flying Qualities Graphs of Kc Scaled Climbing Spiral . . . . . . . . . . . . . . . 74
45 Baseline Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76
46 Mq Scaled Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
47 Workload and Performance Metrics of Mq Scaled Side

Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
48 Flying Qualities Graphs of Mq Scaled Side Step . . . . . . . . . . . . . . . . . . . . . 80
49 Mδ e Scaled Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 81
50 Workload and Performance Metrics of Mδ e Scaled Side

Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82
51 Flying Qualities Graphs of Mδ e Scaled Side Step . . . . . . . . . . . . . . . . . . . . 83
52 Lp Scaled Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
53 Workload and Performance Metrics of Lp Scaled Side

Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 86
54 Lδ a Scaled Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
55 Workload and Performance Metrics of Lδ a Scaled Side

Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
56 Kc Scaled Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
ix
Figure Page
57 Workload and Performance Metrics of Kc Scaled Side

Step . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 90
58 Flying Qualities Graphs of Kc Scaled Side Step . . . . . . . . . . . . . . . . . . . . . 91
59 Baseline Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 93
60 Baseline Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 94
61 Mq Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 95
62 Mq Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 96
63 Flying Qualities Graphs of Mq Scaled Aerial Refueling . . . . . . . . . . . . . . . 97
64 Mδ e Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98
65 Mδ e Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99
66 Flying Qualities Graphs of Mδ e Scaled Aerial Refueling . . . . . . . . . . . . . 100
67 Lp Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101
68 Lp Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 102
69 L2 Norm vs TIC of Roll Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103
70 Lδ a Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104
71 Lδ a Scaled Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
72 L2 Norm vs TIC of Roll Angle . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106
73 TIC vs ζsp Deficiency Overlay . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107
74 Baseline Aerial Refueling Tracking . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
75 Mq Scaling Bode Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 119
76 Melev Scaling Bode Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121
77 Lp Scaling Bode Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 123
78 Lail Scaling Bode Plots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125
x
List of Tables
Table Page
1 Equivalent Time Delay and Transient Peak Ratio

Requirements[5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
2 Rise Time Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3 Aircraft Classification in MIL-STD-1797B [5] . . . . . . . . . . . . . . . . . . . . . . . 22
4 Neal-Smith Criteria for Bandwidth [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
5 LJ-25D Control Surface Deflection Limits . . . . . . . . . . . . . . . . . . . . . . . . . 36
6 LJ-25D Control Surface Rate Limits . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
7 Stability and Control Parameters [52] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
8 Mission Task Elements [31] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
9 Gains for Climbing Spiral Autopilots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44
10 Gains for Side Step Autopilots . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
11 Lead Lag Compensator for Altitude Control Design

Values . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 49
12 Longitudinal Trim Results, 200 KCAS 5,000 ft . . . . . . . . . . . . . . . . . . . . . . 52
13 Lateral-Directional Trim Results, 200 KCAS 5,000 ft . . . . . . . . . . . . . . . . 53
14 Mq LOES Specifications, 200 KCAS 5,000 ft . . . . . . . . . . . . . . . . . . . . . . . . 55
15 Melev LOES Specifications, 200 KCAS 5,000 ft . . . . . . . . . . . . . . . . . . . . . . 55
16 Lp LOES Specifications, 200 KCAS 5,000 ft . . . . . . . . . . . . . . . . . . . . . . . . 56
17 Lail LOES Specifications, 200 KCAS 5,000 ft . . . . . . . . . . . . . . . . . . . . . . . 56
18 Baseline Climbing Spiral Flight Conditions, 200 KCAS

5,000 ft . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
19 Kc Scaling Factors for Climbing Turn . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
20 Kc Scaling Factors for Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . 89
xi
Table Page
21 TIC Calculations for Aerial Refueling Altitude

Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
22 Adjusted Input TIC Calculations for Aerial Refueling

Altitude Comparison . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110
xii
. List of Symbols
α angle of attack
β sideslip angle
δa aileron deflection
δe elevator deflection
δr rudder deflection
δt throttle position
θ pitch angle
ζd Dutch damping ratio
ζp phugoid damping ratio
ζsp short period damping ratio
τ pilot delay
φ roll angle
ψ yaw angle
ωBW bandwidth
ωd Dutch natural frequency
ωp phugoid natural frequency
ωsp short period natural frequency
h altitude
xiii
Lp Rolling moment due to roll
Lδ a Roll moment due to aileron deflection
Mq Pitching moment due to pitch
Mδ e Pitching moment due to elevator deflection
p roll rate
q pitch rate
r yaw rate
TR roll mode time constant
TS minimum doubling time for spiral mode
vt true velocity
x state vector
ẋ derivative state vector
xiv
. List of Acronyms
ADS Aeronautical Design Standard
CAP Control Anticipation Parameter
CG Center of Gravity
DoD Department of Defense
FAA Federal Aviation Administration
FC Flight Condition
FDM Flight Dynamics Model
FPM Foot per Minute
GUI Graphical User Interface
ISR Intelligence Surveillance Reconnaissance
HOS High-Order System
JSBSim Jon S. Berndt Simulation
kts Knots
KCAS Knots Calibrated Airspeed
KIAS Knots Indicated Airspeed
KTAS Knots True Airspeed
LHP Left Half Plane
LOES Low-Order Equivalent System
xv
MATLAB Matrix Laboratory
MIL-STD Military Standard
MTE Mission Task Element
NACA National Advisory Committee on Aeronautics
NASA National Aeronautics and Space Administration
NRC National Research Council
PIO Pilot Induced Oscillation
PID Proportional Integral Derivative
ROC Rate of Climb
RHP Right Half Plane
RPV Remotely Piloted Vehicle
SIO System Induced Oscillation
TIC Theil’s Inequality Coefficient
TPS Test Pilot School
UA Unmanned Aircraft
UAS Unmanned Aerial System
UAV Unmanned Aerial Vehicle
USAF United States Air Force
V/STOL Vertical and/or Short Takeoff and Landing
xvi
VISTA Variable In-flight Test Simulator
VSS Variable Stability System
xvii
I. Introduction
As man entered into the world of flight, early aviation pioneers recognized the
importance of the interface between man and machine. Over decades, engineers and
aviators built and tested thousands of aircraft, steadily converging on the properties
of aircraft that improved mans control of the aerial machine. These properties, called
flying qualities, are defined by the United States Air Force (USAF) Test Pilot School
(TPS) handbook as “those stability and control characteristics which influence the
ease of safely flying an aircraft during steady and maneuvering flight in the execution
of the total mission” [4]. As technology advances, unmanned aerial vehicles (UAVs)
have become a commonplace in both the civilian and military world. Despite the
progress in both fixed wing and rotorcraft flying qualities over the past 100 years, UAV
flying qualities have remained largely unexplored. The technical specifications for
neither fixed wing nor rotorcraft make direct translations to adequate flying qualities
of an unmanned vehicle, and the current “build-fly-fix-fly” remains an inefficient and
mishap prone methodology. The focus of this research is to continue the advancement
of unmanned flying qualities by identifying key characteristics of UAVs that result in
satisfactory workload and performance for various mission task elements (MTE).
1.1 Background
After the 2011 raid on Osama Bin Laden’s compound, a note was found in his
handwriting stating “Brothers said they were frankly exhausted from the enemies air
1
bombardments” [49]. The bombardments were due in part to the relentless use of
UAVs in the fight against the Taliban. Their ability to stay aloft for 20+ hours at a
time, out of reach of ground fire, provides the military with a unique asset in the war
on terror. Maintaining a constant “eye in the sky,” they also have been instrumental
in reducing the number of fratricide events: “U.S. troops are three times safer from
friendly fire attacks when deployed in war zones covered by drones compared with
traditional warfare” [49]. As a result, the number of UAV missions has increased
significantly and is projected to increase even further. Figure 1 shows the exponential
increase of UAV missions conducted by the US government.
Figure 1. Flight Hours for Unmanned Systems between 1996 and 2011 [51]
Not only have drones been effective in saving soldiers on the battlefield, but they
have also been used in the civilian world to save lives in several capacities. In 2017,
DJI, a Chinese manufacturer of drones, reported that drones have saved at least 59
people since 2013. Drones have been used to find lost hikers in a snowy field, drop
ropes and life vests to drowning swimmers, and find lost kayakers in remote areas
[42]. Their small size and high maneuverability allows them to fly at low altitudes
and in tight spaces where manned aircraft cannot operate. UAVs do not require the
on board life support systems needed for manned flight. In turn, they weigh less,
2
allowing for a larger payload, and are smaller, making them cheaper. Thus, their
lower costs make them less of a liability when an asset is lost.
In recent years, military UAVs have taken over missions that were too ‘dull, dirty,
or dangerous’ for manned aircraft [24]. Although they have been effective at complet-
ing these missions, their mishap rates have been nearly 10-20 times as high as their
manned counterparts have incurred [15]. Figure 2 shows the staggering rate of UAV
mishaps over the lifecycle of each aircraft.
Figure 2. U.S. Military Aircraft and UAS Class A Mishap Rates, 1986-2006 [15]
The data shows that as flight hours accumulate, mishap rates decrease exponen-
tially. But from the mishap data early in the development of the UAVs, it is clear
a “build-fly-fix-fly” mentality drove the development of the Global Hawk, Hunter,
Shadow, and Pioneer [24]. Instead of using a “hunt and peck” method of design,
a database of UAV flying handling qualities design knowledge based on historical
data could prevent some of the mishaps early in the lifecycle of a UAV. Lowering the
mishap rate will decrease both the cost and risk of developing UAVs.
3
1.2 Research Problem Motivation and Description
As UAVs become commonplace in both military and civilian applications, an over-

arching governing document is required to ensure the airworthiness of each UAV. A
piece of that qualification includes assessing the flying qualities of the aircraft. Cooper
and Harper stated that flying qualities refer to “those qualities or characteristics of
an aircraft that govern the ease and precision with which a pilot is able to perform
the tasks required in support of an aircraft role” [16]. Several standards have been
developed for military fixed-wing, military rotary-wing, and civilian aircraft, but none
accurately classify the varying size and missions UAVs assume.
The Air Force’s fixed-wing document governing flying and handling qualities is
MIL-STD-1797. Its purpose is “to assure flying qualities for adequate mission per-
formance and flight safety regardless of the design implementation or flight control
system augmentation” [5]. Currently, no requirement has yet to be developed to fit
UAVs into MIL-STD-1797, and the section for UAVs reads TBD. The current MIL-
STD-1797 which governs the flying qualities of Air Force aircraft has four classes of
aircraft based on maximum gross weight and maximum g-load [5]. Each of these cat-
egories corresponds to an aircraft type which also coincide with the mission set the
aircraft accomplishing [5]. For example, an F-16 has high g-loading and a medium
maximum takeoff weight, classifying it as a Class IV aircraft, typical of fighters. A
C-17 with high gross and low g-loading would be a Class III aircraft, typical of heavy
transport aircraft. But because fixed-wing UAVs can vary widely in weight and oper-
ating altitude yet accomplish the same intelligence, reconnaissance, and surveillance
(ISR) mission, they do not easily fit into the existing manned fixed-wing categories
in MIL-STD-1797. One UAV may operate in Class A airspace with a payload of over
3000 lbs, while another may operate below 1000 feet and weigh a mere 30 lbs. The
two UAVs may have vastly different weights and operating altitudes, but have the
4
same maneuverability and accomplish the same mission.
ADS-33E addresses the flying qualities of rotary aircraft based on mission task
performance. The goal of the document is to “assure that no limitations on flight
safety or on the capability to perform intended missions will result from deficiencies in
flying qualities” [19]. ADS-33E requires the aircraft fly various mission task elements
representative of an operational flight envelope, while specifying varying performance
standards that must be met. Certain performance standards are further broken down
by aircraft primary mission to provide tighter tolerances for more maneuverable air-
craft such as scout/attack and relaxed tolerances for less maneuverable aircraft such
as cargo/utility and slung load. Because ADS-33E is focused solely on rotorcraft, its
direct applicability to fixed-wing UAV flying qualities is minimal. But, the philoso-
phy of utilizing mission task elements and maneuverability to evaluate flying qualities
could prove useful in evaluating UAVs in both precision and non-precision maneuvers.
The Federal Aviation Administration (FAA) must also deem an aircraft airworthy,
but their standards are less stringent than MIL-STD-1797. Their primary concern is
safety and meeting certain standards in performance, structural design, and several
other categories [22]. They ensure the safety of the passengers on board with little
concern for the commercial success of an aircraft. On the other hand, the success of a
military aircraft can be life or death on the battlefield, and therefore advanced flying
qualities parameters are paramount in determining the effectiveness of an aircraft
before it is built.
Despite the airworthiness requirements set by MIL-STD-1797, ADS-33E, and the
FAA, UAV flying qualities still lack an overarching document that specifies the design
principles necessary for satisfactory mission performance. The development of both
fixed-wing and rotary-wing aircraft is aided through the use of a database of design
knowledge compiled over decades of testing. Without a similar UAV design database
5
to build future UAVs, the Department of Defense (DoD) risks the success of a UAV
program even before its inception.
By exploring new parameters based on workload and performance during mission
oriented maneuvers, a knowledge base about what parameters correlate with certain
flying handling qualities can be determined. These workload and performance values
from known aircraft with satisfactory handling qualities, such as the LJ-25D, provides
a basis for engineers to design future UAVs.
1.3 Research Objectives
The primary goal of this research is to further the database of available UAV
handling qualities knowledge and provide new metrics to accurately correlate subjec-
tive flying qualities with UAV engineering design principles. To do this, a stitched
model of the CALSPAN Variable Stability System (VSS) LJ-25D, shown in Figure 3,
will be used to simulate various flight maneuvers and post process results from the
simulation.
Figure 3. Learjet LJ25 Variable Stability System [8]
The stitched model refers to a “full flight-envelope, real-time simulation model

composed of individual linear models and trim data for discrete flight conditions” [8].
It was developed at the Air Force Test Pilot School using flight data collected on the
LJ-25D VSS. The model was built in Matrix Laboratory’s (MATLAB) Simulink, a
6
proprietary, graphical block diagramming software effective for simulating complex
dynamic systems [34]. The results of this research can be compared to data from
future flight tests on the LJ-25D. This research builds upon the work of Lt Col.
Kara Greene’s Toward a Flying Qualities Standard for Unmanned Aircraft [24], Lt
Joshua Kim’s Evaluation of Unmanned Aircraft Flying Qualities Using JSBSIM [30],
and Capt Ali Hamidani’s Evaluating the Autonomous Flying Qualities of a Simulated
Variable Stability Aircraft [26], with the following objectives:
1. Evaluate the flying qualities of an unmanned LJ-25D through various precision

and non-precision maneuvers.
2. Identify trends between stability and control parameter scaling, workload and
performance metrics, and classical control literal factors.
1.4 Thesis Overview
Chapter I described the background and motivation behind developing flying han-
dling qualities metrics for UAVS. Chapter II, the literature review, includes a concise
summary of the development of manned flying qualities and previous efforts on ad-
vancing unmanned aircraft (UA) flying qualities. Next, Chapter III, research method-
ology, is presented, which includes a detailed description of the LJ-25D stitched model
and the performance metrics that will be calculated for the maneuvers. Chapter IV
then presents the results of the research, including the evaluation of each maneuver
using previously researched performance and workload metrics. Finally, Chapter V
provides the conclusions of the research and future recommendations for research.
7
II. Literature Review
This chapter first provides an introduction to the history and development of

modern day flying handling qualities metrics. Next, the flying qualities standards for
manned fixed wing and manned rotary wing are explained. Finally, previous research
accomplished in the area of UAV flying qualities is presented.
2.1 Early Development of Flying Qualities
Even before the Wright Brothers’ first flight, an Englishman by the name of G.H.
Bryan had developed the theoretical groundwork of stability [12]. Stability refers to
the response of an aircraft in a trimmed state to a perturbation [13]. Bryan termed the
initial response to a perturbation as the static response and the oscillatory response
as the dynamic response. He correctly identified that aircraft of the time needed to be
both statically and dynamically stable in the longitudinal axis and lateral-directional
axis. He even identified the two primary modes of longitudinal oscillation: the phugoid
mode and the short-period mode. His work continued to the lateral-directional modes,
distinguishing the spiral mode, roll mode, and Dutch roll mode [40]. Although the
calculations developed from Bryan’s work were often too computationally intensive to
be utilized by aircraft designers, Bryan had laid the theoretical foundation for future
investigations in flying qualities.
Since the first Wright Flyer, a cornerstone of flight has been the interface between
man and machine. On one of their first flights, Wilbur noted, “the machine seemed
to steer all right laterally, but after attaining high speed began to undulate somewhat
and suddenly turned downward” [35]. Although it is only a crude description of the
phenomenon taking place, it is one of the first descriptions of “flying and handling
qualities.” The Wright Brothers continuously modified their designs, seeking one
8
that, in the pilot’s opinion, met acceptable flying and handling characteristics. But
only so much progress could be made with little understanding of aerodynamics and
a lack of correlation between pilot opinion and engineering design.
As the Wright brothers continued to refine their design, the US Army Signal Corps
took interest in new “heavier-than-air” flying machines. In 1907, they released a
request for proposal for a heavier-than-air machine, along with a list of specifications.
Although most specifications were performance related, buried within the request is
the first US military flying qualities requirement. It read:
Before acceptance, a trial endurance flight will be required of at least one

hour during which time the flying machine must remain continuously in
the air without landing. It shall return to the starting point and land
without any damage that would prevent it immediately starting upon
another flight. During this trial flight of one hour it must be steered in
all directions without difficulty and at all times under perfect control and
equilibrium.
– Advertisement and Specification for a Heavier-than-Air Flying

Machine, Signal Corps Specification No. 486, December 1907 [7]
Early aviators knew that if the machine was difficult to control in flight, it would
be increasingly demanding or even impossible to accomplish the mission at hand. But
simply stating “steered in all directions without difficulty” leaves the specification up
for interpretation based on the pilot’s opinion. A standardized system was needed for
both the subjective flying qualities determined by the pilot’s opinion and the objective
qualities based on engineering theory.
As WWI broke out, trench warfare dominated the battlefield, and the airplane
added another dimension. Although airplanes were still underdeveloped, their merits
were undeniable. They were primarily used in the form of reconnaissance for artillery
strikes and aerial photography [48]. Soon after, both the fighter airplane and the
bomber airplane were developed. Clearly, the unnamed British general who had
9
stated “the airplane is useless for the purposes of war” [48] had been proven wrong
by the rapid advancement in technology.
Across the Atlantic Ocean, the United States established National Advisory Com-
mittee for Aeronautics (NACA) in 1915, with goal of advancement in aeronautics [46].
NACA made its home at Langley Field, Virginia, and by 1922, flight test research
had begun on several aircraft from both U.S. and foreign services. The dozen man
team was initially tasked to “supervise and direct the scientific study of the problem
of flight, with a view to their practical solution, and to determine the problems which
should be experimentally attacked” [10]. The Navy Bureau of Aeronautics contacted
NACA to utilize these aircraft in studying the design features that correlated to sta-
bility, controllability, and maneuverability. The research first focused on obtaining
accurate data to understand stall characteristics, takeoff speeds, and landing speeds.
Down the road, NACA began developing instrumentation needed to measure data
in flight, which enabled the study of pressure distributions along wings. With this
new instrumentation, a series of flight tests which previously lasted months could be
shortened to a single day [10]. The pressure data allowed aircraft design parameters
to be correlated to stability and control characteristics.
In the 1930s, commercial aviation became a profitable venture due to the increased
size and performance of aircraft. In the commercial world, a new emphasis was put
on safety and operational efficiency. Edward Warner, both the chief scientist for
NACA and a consultant for the Douglas Aircraft Company with a vested interest in
the success of the new DC-4, convinced NACA to investigate flying qualities in an
attempt to find those qualities desired by pilots [10]. Warner tasked Hartley Soulè to
conduct a study on writing a set of requirements which resulted in satisfactory flying
qualities [40].
As a result, a report titled Preliminary Investigation of the Flying Qualities of
10
Figure 4. The Stinson Reliant played an important role in developing early flying
qualities metrics [1]
Airplanes suggested further refinement of flying qualities requirements and recom-

mended further testing on assorted aircraft [44]. Flying a Stinson Reliant aircraft
similar to Figure 4, the test team made measurable control inputs which were then
correlated to aircraft design specifications. Tests continued on a total of 12 aircraft,
creating a small design database focused on flying qualities for aircraft of various
sizes [10]. The data was used to forge crude numeric formulas to be used as rules
of thumb for future designs. The database benefited aircraft designers through the
1940s, especially in the development of combat aircraft [10].
In the 1950s, aircraft had become bigger, faster, and more maneuverable than
ever before. Designers also became more experimental, creating variable stability
aircraft to vary aircraft parameters via an in flight control system. The first of these
aircraft was the NT-33A, pictured in Figure 5. The NT-33A was programmed to
create the feel of an aircraft with different flying qualities characteristics than its own
11
[6]. As a result, further research could be done in determining what aircraft design
characteristics resulted in satisfactory flying qualities without having to physically
design and build the geometry’s of an aircraft with those characteristics.
Figure 5. The NT-33A aided in the development of the X-15, A-10, F-15, F-16, F-18,
F-117, and F-22 [6]
The NT-33A provided the Air Force with nearly 40 years of service and was
instrumental in developing the flying qualities standards used today [6]. Far before
the end of its service life, it was clear that a replacement was needed to continue
research and development via variable stability aircraft [14]. The solution was to
modify an F-16D to create the NF-16D Variable Stability In-flight Test Simulator
(VISTA), shown in Figure 6. The first major project that employed the NF-16D was
to demonstrate the effectiveness of thrust vectoring on the F-16’s maneuverability in
flight [33]. At the conclusion of the program, the aircraft was refitted to the original
VISTA configuration and assigned to the United States Test Pilot School where it is
operated today [33].
Over time, aircraft designers refined their skills and created higher performance
12
Figure 6. The F-16 VISTA [36]
planes with improved flying qualities. New rotating wing aircraft called helicopters
drove the separation of flying qualities standards for fixed wing and rotary-wing air-
craft. These standards will be discussed separately in the subsections 2.2 and 2.3.
2.2 Manned Fixed-Wing Flying Qualities
By 1969, a standalone document titled Flying Qualities of Piloted Airplanes rep-

resented the Army, Air Force, and Navy’s requirements for fixed wing aircraft. Also
termed MIL-F-8785(ASG), the goal of the military specification was to “assure fly-
ing qualities that provide adequate mission performance and flight safety regardless
of design implementation or flight control system mechanization” [3]. Within the
specification, aircraft were further broken down into classes based on weight and ma-
neuverability and were assessed in three different flight phases, termed categories,
based on aircraft mission. This allowed designers to access aircraft design parameters
for aircraft of varying sizes and mission sets [3]. MIL-F-8785B was further revised to
MIL-F-8785C in 1980 and although MIL-F-8785C is now obsolete, it set the stage for
MIL-STD-1797B, today’s gold standard in flying qualities specifications.
MIL-STD-1797 was first published 31 March 1987 for use in developing USAF
13
aircraft [5]. MIL-STD-1797 and its subsequent versions added to MIL-F-8785C by
incorporating more design criteria from research conducted in the 1970s and 1980s
[37]. MIL-STD-1797 also added in subjective criteria based on Cooper-Harper ratings,
which was not included in MIL-F-8785C. The current version, MIL-STD-1797B, was
published in 2006 and continues to be used for assessing fixed wing flying qualities
standards. The subsequent subsections discuss the various subjective and objective
criteria of MIL-STD-1797B used to assess fixed wing aircraft.
2.2.1 Subjective Criteria.
In the 1960s, pilot rating scales became a prominent technique to assess the han-
dling qualities of aircraft. These scales provided pilots and engineers a method to
assess vehicle performance and pilot workload in the completion of specific mission
tasks [16]. But without a single standardized scale to be used across airframes and
test programs, the job of comparing pilot ratings from one aircraft to another became
increasingly difficult. In 1969, George Cooper and Robert Harper published a paper
titled The Use of Pilot Rating in the Evaluation of Aircraft Handling Qualities which
sought to standardize the pilot rating scale not only across the US armed services,
but also the world [16].
2.2.1.1 Cooper-Harper Rating Scale.
The resulting Cooper-Harper Pilot Rating Scale, shown in Figure 7, utilizes a

decision tree to guide the pilot to a numbered rating based on the demands of the
pilot in the selected task or required operation. The scale ranges ten values with the
worst rating being a 10 and the best being a 1. The decision tree is a series of three
questions asking:
1. Is the vehicle controllable?
14
2. Is adequate performance attainable with a tolerable pilot workload?
3. Is it satisfactory without improvement?
Figure 7. Cooper-Harper Pilot Rating Scale [5]
The Cooper-Harper rating scale utilizes these questions to guide the pilot to an
objective handling qualities rating between 1 and 10 to the aircraft. A Cooper-Harper
rating between 1 and 3 indicates Level 1 flying qualities, termed satisfactory. A rating
between 4 and 6 indicates Level 2 flying qualities, termed acceptable. Finally, a rating
between 7 and 9 indicates Level 3 flying qualities, termed controllable [5].
The rating scale can be used in a wide variety of operational tasks, such as takeoff,
landing, and in-flight refueling. The scale seeks to identify when the pilot workload
is excessive and his mission performance begins to degrade. As Cooper and Harper
state in their paper, “In a specific task, he is capable of attaining essentially the same
performance for a wide range of vehicle characteristics, at the expense of significant
15
reductions in his capacity to assume other duties and to plan subsequent operations”
[16]. In this particular instance, Cooper and Harper observed that the test pilots
could maintain the same performance in poorer handling qualities vehicles, but at
the expense of accomplishing basic mission tasks.
The evaluation thus seeks to assess the pilot-vehicle operation in the accomplish-
ment of the mission. If pilot workload is excessive in accomplishing the mission task,
then he will rate it poorly on the scale. Conversely, if the mission task was less dif-
ficult to accomplish and the pilot maintained his capacity to complete operational
tasks, he will rate it highly on the scale.
The Cooper-Harper rating scale has become the standard in the assessment of
fixed wing aircraft. Because the scale is dichotomous, it lends itself to repeatability
by guiding the pilot in evaluations via a series questions regarding performance and
workload [27]. It is also included in MIL-STD-1797, which states, “we use it as the
principal way to relate flying qualities requirements to operational needs” [5]. Down
the road, the rating scale would be modified for rotary aircraft and even a UAV
derivative of the scale was created, which will be presented later in Section 2.5 of this
chapter.
2.2.2 Objective Criteria.
As flying qualities research advanced via programs such as the NT-33A and
NF-16D, large databases of pilot ratings and subsequent aircraft parameters were
amassed. Engineers worked to create thumbprint plots, such as the one in Figure 8,
to correlate pilot ratings on the Cooper-Harper scale to objective aircraft parameters.
Aircraft design parameters such as the short-period natural frequency times the pitch
rate transfer function zero associated with the short-period mode, ωsp Tθ2 , and the
equivalent short-period damping ratio, ζsp , are good indicators of fixed wing aircraft
16
mission performance in the longitudinal axis [45].
The short-period mode is dominates a pilot’s opinion of longitudinal flying qual-
ities. The magnitude of short-period damping is ζsp , which is representative of how
much the short-period mode oscillation is damped. A low ζsp will cause the aircraft
to be too sensitive and difficult to control in higher precision and aggression tasks.
A high ζsp will be too sluggish and again difficult to control in certain tasks. During
early research to develop plots such as Figure 8, a ‘sweet spot’ was determined to
be around a ζsp of 0.7 [45], which correlated to good pilot opinion of the aircraft.
ωsp Tθ2 was chosen to be used in several MIL-STD-1797 flying qualities graphs due to
its correlation to nα , the aircraft load factor response to angle of attack [45]. Once
again, an optimum value was determined for ωsp Tθ2 from flight test, around 2.25,
which correlated to good pilot opinion.
With a large sample sizes of data, engineers were able to draw lines around the
areas where pilots reported satisfactory, adequate, and controllable flying qualities.
The areas were then deemed Level 1, Level 2, and Level 3 flying qualities, respec-
tively, producing a “thumbprint” outline for varying design values. Without building
aircraft, manned fixed wing designers can predict aircraft flying qualities based on
objective aircraft designer parameters.
2.2.2.1 Literal Factors.
From the correlation of large sample sizes of data to Cooper-Harper ratings came
several literal factors which predict aircraft flying qualities. From these correlations,
engineers were able to produce thumb print charts bounding Level 1, Level 2, and
Level 3 flying qualities ratings. One of the most important literal factors is the
Control Anticipation Parameter (CAP). CAP is the ratio of the instantaneous pitch
acceleration, q˙0 , to the change in steady-state load factor, ∆nss [27].
17
Figure 8. ωsp Tθ2 vs ζsp plot depicting how Cooper-Harper ratings drove objective criteria
[5]
CAP is a literal factor that describes the pilot’s perception threshold of angular
pitching acceleration which accompanies any minor changes to the aircraft’s flight
path [9]. The control anticipation parameter correlates well with pilot opinion because
aircraft that are deemed difficult to control often require either too much precise path
adjustment and are considered “sensitive,” or not enough precise path adjustment
and are considered “sluggish.” Figure 9 from MIL-STD-1797B depicts how CAP and
short-period damping are used to set Level 1, Level 2, and Level 3 boundaries on
18
short-period dynamic requirements.
Figure 9. Short-period Dynamic Requirements [5]
It should be noted that this definition of CAP was developed from data on classical
aircraft without highly augmented control systems [11]. Further research has since
been conducted by Bischoff in his report, The Control Anticipation Parameter for
Augmented Aircraft, to create a modified CAP that more accurately represents the
higher-order dynamics of augmented aircraft [11].
Another literal factor that accurately predicts longitudinal flying qualities is ωsp Tθ2 .
In addition to being related to nα , ωsp Tθ2 is also representative of the time response
between aircraft pitch attitude and flight path [5]. As a result, ωsp Tθ2 is proportional
to CAP and can be used in creating thumbprint plots describing the short-period
pitch response. Figure 9 and Figure 10 show similar shaped outlines as ζsp is varied
with changing CAP and ωsp Tθ2 , respectively.
In addition to frequency domain analysis of literal factors, a time domain analysis
of literal factors also correlates to pilot ratings. By working directly from the pitch
rate response of the aircraft, aircraft designers can create satisfactory systems without
19
Figure 10. Short-period Pitch Response to Pitch Controller [5]
direct designation of dominant system poles and zeros [5]. Figure 11 shows the pitch
rate response of an aircraft to a pilot step input.
Figure 11. Time Domain Transient Response Parameters [5]
Several aircraft performance specifications arose from these transient literal factors
to aid in identifying Level 1, 2, or 3 aircraft. Equivalent time delay (t1 ) requirements,
the analog to τθ in the frequency domain, are specified in Table 1. Transient peak ratio
∆q2
( ∆q1
) requirements, correspondent to short-period damping factor in the frequency
20
domain, are listed in Table 1 as well. Finally, effective rise time requirements, which
are analogous to CAP in the frequency domain, are spelled out in Table 2.
Table 1. Equivalent Time Delay and Transient Peak Ratio Requirements[5]

∆q2
Level Equivalent Time Delay t1 Transient Peak Ratio ∆q1
∆q2
1 t1 ≤ .12 sec ∆q1
≤ .30
∆q2
2 t1 ≤ .17 sec ∆q1
≤ .60
∆q2
3 t1 ≤ .21 sec ∆q1
≤ .85
Table 2. Rise Time Values[5]
Non-terminal Flight Phases Terminal Flight Phases

Level Min ∆t Max ∆t Min ∆t Max ∆t
1 9/Vt 500/Vt 9/Vt 200/Vt
2 3.2/Vt 1600/Vt 3.2/Vt 645/Vt
2.2.2.2 Classification Category.
Several mission sets exist within manned fixed-wing aircraft which require vastly
different aircraft to accomplish them. In turn, pilots have different desired flying
qualities based on the mission set of the aircraft. For example, a cargo aircraft
would be ill suited for carrying sensitive payloads if it were highly maneuverable,
while a fighter would be a poor dogfighting aircraft if it had sluggish maneuverabil-
ity. MIL-STD-1797B incorporated a classification category to aid in designing an
aircraft tailored to a specific mission set. The document breaks down aircraft into
four primary Classes: Class I is small light aircraft, Class II is medium weight with
low to medium maneuverability, Class III is heavy, low maneuverability, and Class IV
is high maneuverability [5]. Table 3 shows the full aircraft classification guide from
MIL-STD-1797B.
Aircraft were separated into classes by using two primary design factors: maxi-
mum g-load (n) and gross weight [5]. G-load separates high and low maneuverability
21
Table 3. Aircraft Classification in MIL-STD-1797B [5]
aircraft, while maximum gross weight primarily isolates the heavier transport aircraft
from all others. Plotting maximum g-load vs. gross weight yields Figure 12. Once
again based on historical data of past aircraft, lines are drawn to aid in delineating
one class from another, although exceptions may exist [5]. MIL-STD-1797B incor-
porates these class categories by relaxing or tightening certain aircraft performance
parameters based on aircraft class.
2.2.2.3 Modified Neal-Smith Criteria.
The Modified Neal-Smith Criteria, often referred to as the Neal-Smith criteria, is

a design method that utilizes a specified pilot model in a pilot-aircraft pitch attitude
control system with unity feedback [5]. With this model, varying adjustments can
be made to pilot delay, pilot gain, and develop lead or lag compensation [5]. The
open-loop response of the system is plotted on a Nichols Chart, where the closed-loop
dynamics can be ascertained. The closed-loop droop bounds the minimum gain while
the closed-loop resonance sets the maximum allowable pilot gain to meet certain
22
Figure 12. Aircraft Classification in MIL-STD-1797B [5]
flying qualities levels. Figure 13 pictures a sample Nichols chart for determining
pilot-in-the-loop pitch dynamics.
The overarching usefulness of the Neal-Smith Criteria is in determining the band-
width of the closed-loop systems with the pilot in the loop. From where the open-loop
plot intersects the closed-loop phase of -90 degrees is the bandwidth. Table 4 depicts
the minimum bandwidth required by the Neal-Smith Criteria for various categories
of flight.
Table 4. Neal-Smith Criteria for Bandwidth [5]
Flight Phase Bandwidth

Category A 3.5 rad/s
Category B 1.5 rad/s
Landing 2.5 rad/s
Other Category C 1.5 rad/s
23
Figure 13. Modified Neal-Smith Design Criteria for Pitch Dynamics [5]
2.3 Manned Rotary-Wing Standards
Fixed-wing flying qualities were much further developed by the 1950s as opposed
to rotary-wing flying qualities. As a result, MIL-F-83300, titled Flying Qualities of
Piloted V/STOL Aircraft, was not released until December of 1970 [2]. One of the
primary goals of MIL-F-83300 was for the development of V/STOL (Vertical and/or
Short Takeoff and Landing) flying qualities criteria via first developing techniques
to analyze and evaluate V/STOL flying qualities and second generate flying quali-
24
ties requirements and design criteria [29]. The program utilized the XV-4B variable
stability test vehicle to develop baseline flying qualities data [29], much in the way
the NT-33 was utilized in fixed-wing flying qualities. Part of the research program
included in-depth reviews of draft specifications by 10+ industry entities and several
government entities across the Air Force, Army and Navy [29]. Despite the work
towards improving design requirements, helicopter mission performance continued to
lack [23].
In the mid 1980s, Aeronautical Design Standard (ADS) 33 replaced the previous
flying qualities documents governing vertical lift aircraft. ADS-33 sought to focus
on the mission of the aircraft, and therefore separated aircraft into four primary
missions categories regardless of weight: attack, scout, utility, and cargo. During
evaluation, aircraft complete certain MTEs representative of the operational mission
[19]. Examples of MTEs are takeoff, lateral reposition, and side step; each MTE has
criteria to meet desired performance and adequate performance [19]. The advantage
to this standard is the performance criterion for each MTE can be based on both
aircraft precision and quickness without reference aircraft gross weight [23]. In other
words, ADS-33 assesses MTE and operational environment to determine the class to
evaluate the rotorcraft [23]. ADS-33 went through several revisions until becoming
ADS-33E in 2000 and remains the standard for rotary-wing aircraft flying qualities
today.
2.4 Unmanned Aerial Vehicle Standards
Air Force Maj. Gen. David Baker stated in 1956 that “We can readily see that
except for certain types of missions, the manned combat aircraft will become tech-
nically obsolete in the future” [21]. In the 1960s, the mission of unmanned aerial
vehicles had expanded from primarily being expendable target drones to carrying
25
surveillance equipment and completing aerial reconnaissance missions [21]. The in-
telligence community quickly identified unmanned drones as an asset to be utilized
in gathering critical information over hostile territory without risking American lives.
Still, no design criterion existed to assess the flying qualities and its effects on mission
performance.
In 1976, AFFDL-TR-76-125, Remotely Piloted Vehicle Flying Qualities Design
Criteria, became the first attempt at creating flying qualities standards for unmanned
vehicles [25]. The report was the second part of a four part program for developing a
remotely piloted vehicle (RPV) flying qualities specification [41]. The document takes
a top down approach in developing requirements: “performance-oriented criteria are
developed in hierarchical fashion from mission requirements, through total system
requirements, to subsystem requirements. The criteria encompass automatic and
manual controls, command and data link, controls and displays as well as vehicle
stability and control” [41]. The final goal of phase II was to develop flying qualities
requirements that could later be validated by simulation and analysis [30]. As the
final phases were never completed, the document stands as a technical report, not a
mandatory standard.
The primary issue in applying AFFDL-TR-76-125 to modern aircraft lies in the
document’s definition of RPV: “unmanned air vehicle which has the capability of
being controlled by a remote operator during some flight phase of an operational mis-
sion” [41]. AFFDL-TR-76-125 was developed at a time when completion of the RPV
mission relied on human interaction with the system. Today, UAVs are expected to
achieve satisfactory mission performance without the intervention of human opera-
tors, thus reducing the skill requirements of operators and increase system reliability
[50]. Since AFFDL-TR-76-125 was published in 1976, it has yet to be updated in
any form to reflect the shift of focus in requirement development from pilot-aircraft
26
interface to full aircraft autonomy.
One of the challenges still being addressed by UAV researchers is the classification
of UAVs [30]. AFFDL-TR-76-125 utilized the four class criteria from MIL-F-8785C
based on gross weight and maneuverability. That criterion works well with manned
fixed-wing aircraft and is used in MIL-STD-1797B, but fails in accurately describing
the mission and performance for UAVs. Flying qualities design challenges vary greatly
depending on the size of the UAV; a small disturbance to a 5000 lb UAV is a large
disturbance to a 20 lb UAV. The longitudinal, lateral, and directional sensitivities
associated with the controls also vary depending on the size of a UAV [50], and
therefore the relationships between small and large UAVs cannot be assumed to be
linear. Williams also alludes to the lack of pilot comments, writing, “the challenges
relate to the absence of control stick force feedback, an absence of vibration and buffet
response” [50]. The challenges faced in developing UAV flying qualities have led to
the lack of a leading standard in the UAV community [23].
In addition to the military extensively researching flying qualities, National Air
and Space Administration’s (NASA) had continued the flying qualities work that
NACA had pioneered in the 1920s. A 2004 study conducted by the National Re-
search Council (NRC) assessed the overall scientific and technical quality of NASA’s
Pioneering Revolutionary Technology Program [38]. In evaluation of NASA’s UAV
research, the NRC highlighted that the existing flying quality standards utilized for
manned aircraft may be too restrictive in development of UAVs, and “further evo-
lution of the base work done by NASA to include unmanned systems is essential
to creating a competitive advantage for U.S. products as this market becomes more
price-driven” [38]. The NRC identified the vested interest for not only the military
but the civilian sector in the development of a flying qualities standard for UAVs.
27
2.5 Previous UAV Flying Qualities Research
One of the overarching issues with developing an unmanned flying qualities stan-
dard is determining how to classify the aircraft. AFFDL-TR-76-125 used the maneu-
verability and gross weight metrics from manned aircraft to classify RPVs, as those
frequently aligned with the general mission of the manned aircraft. This scale fails
when applied to UAVs because recent advances in technology allow an ultra-light,
highly maneuverable quad-copter to accomplish the same reconnaissance mission as
a heavy, sluggish Predator drone.
Recently, several researchers have devised new UAV specific classes which seek to
resolve the dilemma. For example, Williams classified UAVs via control system and
reuseability. Figure 14 pictures how Williams utilizes autonomy to separate RPVs
from UAVs and reusablility to exclude single mission UAVs such as missiles and target
drones from multi-mission UAVs such as Predator, Global Hawk, and Pioneer [50].
Figure 14. Williams’s UAV Class System [50]
Cotting proposed an alternate class system for UAVs shown in Figure 15, based
on Reynolds Number and aircraft weight [17]. With this scale, lighter, slower UAVs
28
which operate at lower Reynolds Numbers are classed together, just as heavier, faster
UAVs at higher Reynolds Numbers are grouped. The Unmanned Systems Integrated
Roadmap: FY 2011-2036 offered a UAV class system based off operational altitude,
weight, and airspeed. Figure 16 shows the five classes of UAVs: penetrating, persis-
tent, tactical, small tactical, and micro/mini tactical. Despite these innovative class
alternatives to maneuverability and gross weight, Greene points out that basing UAV
classes on physical and performance metrics of the UAV may be flawed all together:
“the correlation between class and mission tends to break down for unmanned aircraft,
because unmanned aircraft of very different sizes often perform the same missions”
[23]. But, without sufficient UAV flying qualities research or a database of flight test
data, an alternative mission-based classification approach cannot be validated either
[23].
Figure 15. UAV Grouping by Reynolds Number and Aircraft Weight [17]
In addition to UAV classification changes, several adapted Cooper-Harper rating
29
Figure 16. UAV Grouping by Operational Altitude, Weight, and Airspeed [51]
scales have been contrived as an alternative to the original. Cotting proposed a modi-
fied version of the Cooper-Harper Piloted Rating Scale for unmanned systems, shown
in Figure 17. He defined UAV flying qualities as “those qualities or characteristics
of an aircraft and sensor system that govern the ease and precision with which an
operator is able to perform the tasks required in support of its mission role” [18].
Cotting’s scale focuses on the operator’s ability to perform the tasks necessary to
complete the mission. A rating scale of 1 to 10, where 10 is worst and 1 is best, is
utilized by the evaluator to give an ordinal rating with several logic gates similar to
the original Cooper-Harper Rating Scale [18]. Although Cotting successfully adapts
the scale for UAVs, there still remains disagreement amongst the UAV community
on what “good” UAV performance entails [23]
30
Cotting’s focus was similar to ADS-33E in that a specific MTE would be evaluated
using the modified Cooper-Harper Scale. Each task reflects the mission, thereby
evaluating the aircraft and sensor integrated systems necessary for mission success
[18]. He breaks down the MTEs into task categories, sorted by aggressiveness and
precision. Figure 18 illustrates how aircraft roles relate to MTEs and various task
categories representative of the aircraft mission. Cotting recognized that validation
of the scale precludes its widespread adoption by the flying qualities community, as
well as a standard set of MTEs for evaluation [18].
Figure 17. Cotting’s Modified Cooper-Harper Rating Scale for UAVs [18]
Several research endeavors involving UAV flying and handling qualities have been
completed at the Air Force Institute of Technology. This thesis builds on the foun-
dation that Greene first laid for developing an unmanned flying qualities standard.
In Greene’s dissertation Toward a Flying Qualities Standard for Unmanned Aircraft,
she stated the end goal of her research was to kick-start a process utilized in the air-
31
Figure 18. Notional Relation between UAV Roles, Mission Task Elements, and Task
Categories [18]
craft design and model validation that increases the overall success rate of UAVs [23].
Using the j2 Universal Tool-Kit software, she evaluated three UAV models through
four maneuvers: the non-precision non-aggressive climbing spiral, the non-precision
aggressive air-to-ground tracking, the precision non-aggressive offset landing, and the
precision aggressive pitch and roll discrete tracking [23]. From these maneuvers, flying
qualities criteria such as pitch quickness, bandwidth, and time delay were calculated
and recorded. From her analysis, she found that weight is a poor classifier for UAVs
[23].
As expected, Greene recommended incorporating verified models into the analy-
sis, which Kim did in his follow-on research. Kim investigated a Cessna 172 (C172)
climbing spiral maneuver using Jon S. Berndt Simulation (JSBsim) [30]. The C172,
a validated model based on publicly available data, was analyzed using workload and
32
performance metrics with varying levels of aggressiveness during the spiral maneu-
ver. He also assessed the effects of varying both longitudinal and lateral-directional
stability and control derivatives to mimic the research techniques of variable stabil-
ity aircraft. His findings stated that how much the autopilot works in completing a
maneuver is not as much of a concern as the actuator position limits and rate limits
degrading aircraft performance [30].
Subsequent research by Hamidani investigated workload and performance metrics
during two precision aggressive tasks: pitch attitude tracking and altitude track-
ing. Using an F-16 model in JSBsim, he also varied longitudinal stability and control
derivatives and assessed the tracking ability of the aircraft [26]. Hamidani also recom-
mended further investigation into the relationship between control surface deflection
and aircraft performance [26].
2.6 Summary
A history of the development of modern day flying qualities was presented, starting
from the first pioneers discovering the concepts of static stability to modern day
criteria used to assess aircraft flying qualities. Large databases of flight test data
were utilized in the development of documents such as MIL-STD-1797B and ADS-
33E. These standards drive contemporary flying qualities criteria, yet these criteria
are inadequate for UAVs. Several alternate classifications have been proposed to aid
in developing new flying qualities criteria, but without actual flight test data, they
remain unproven hypotheses. Previous research leveraged simulations of UAVs as a
cost-effective alternative to flight test. Chapter III details the methodology of using
such simulations to assess workload and performance of an LJ-25D.
33
III. Research Methodology
The overarching goal of this research is to contribute to a database of UAV flying

handling qualities. A stitched model of an LJ-25D is utilized in various maneuvers
and analyzed using workload and performance metrics. Ideally, correlations can be
drawn from the data indicating design parameters that equate to satisfactory flying
qualities.
3.1 Simulink Model of LJ-25D Aircraft
As previously stated, the LJ-25D model is a stitched, quasi-linear, parameter-

varying model developed at the USAF Test Pilot School [8]. For the LJ-25D aircraft
under consideration, a full flight envelope simulation model was created utilizing
actual flight data from five different flight and loading conditions [8]. The model was
then implemented in Simulink and validated at various flight conditions. Reference 5
details the exact procedures to create such a stitched model.
3.1.1 Model Development.
After taking the flight test data, the model was stitched primarily using the x-body
axis velocity state U as a scheduling parameter [8]. Density ratio scaling permitted
state and trim data to be interpolated between available point model data [8]. The
stitched model can also scale forces and moments based on changes in aircraft weight,
moment of inertia, and center of gravity (CG) location [8]. In summary, “model stitch-
ing is accomplished by implementing lookup tables of the aircraft state trim values,
control input trim values, and stability and control derivatives based on point models
and trim data” [8]. Figure 19 shows a detailed block diagram of how the schedul-
ing parameter U interfaces with the lookup tables and subsequent pilot commands,
34
aircraft states, and nonlinear equations of motion to output the next aircraft state.
Figure 19. LJ-25D Model stitching block diagram [8]
An important aspect of the model is control surface actuator saturation. Satura-

tion occurs if either the position of the actuator reaches a limit or the rate of actuator
movement reaches its limit. Saturation can cause undesirable flying qualities [5] and
will be one of the primary focuses of performance and workload metrics in analysis
of aircraft maneuvers. Table 5 shows the LJ-25D control surface deflection limits and
Table 6 shows the LJ-25D control surface rate limits.
35
Table 5. LJ-25D Control Surface Deflection Limits
Elevator Deflection min -15◦

max 15◦
Aileron Deflection min -40◦
max 40◦
Rudder Deflection min -30◦
max 30◦
Table 6. LJ-25D Control Surface Rate Limits
Elevator Rate min -100◦ /s

max 100◦ /s
Aileron Rate min -150◦ /s
max 150◦ /s
Rudder Rate min -100◦ /s
max 100◦ /s
3.1.2 Stability Parameter Scaling.
Within the LJ-25D model are two lookup table data structures: an A matrix for
the stability parameters and a B matrix for the control parameters. The stitched
model is composed of 8 point models, with 4 of them at flaps-up and 4 at flaps 20
degrees. Each of the flap settings are at approximately 185 kts, 220 kts, 250 kts, and
300 kts. Reference 5 plots the flaps up derivatives as they change with airspeed.
As part of this research, four stability and control parameters will be scaled as
a method of altering the bare airframe dynamics of the LJ-25D. The two stability
parameters are Mq , the pitching moment due to pitch rate, and Lp , the rolling moment
due to roll rate. The two control parameters are Mδ e , the pitching moment due to
elevator deflection, and Lδ a , the rolling moment due to aileron deflection. Table 7
depicts the four stability and control parameters.
Previous research by Kim [30] and Hamidani [26] scaled the stability and control
derivatives Cmq , Clp , Cmδe , and Clδa instead of the stability and control parameters.
Due to limitations in the design of the LJ-25D Simulink model, only the stability
and control parameters could be scaled. Because there is a proportional relationship
36
Table 7. Stability and Control Parameters [52]
Stability Parameter Definition Units

2
q¯1 Sc̄
Mq 2Iyy U1 Cmq s−1
q¯1 Sc̄
Mδe Iyy Cmδe s−2
q¯1 Sb2
Lp 2Ixx U1 Clp s−1
q¯1 Sb
Lδa Ixx Clδa s−2
between the stability and control derivatives scaled by Kim and Hamidani and the
respective stability and control parameters, the end result will always be a propor-
tionally scaled parameter.
For each maneuver, a single stability and control parameter will be varied. The
scaling factor was generally multiplied between 0.25 and 2 of the original bare air-
frame parameter. Scaling factors varied between stability and control parameters.
The reasoning was twofold: first, it was to ensure an accurate model of the new bare
airframe was developed, as data from an inaccurate model would be unusable. Sec-
ond, it allows full completion of each maneuver, as some large changes of stability
derivatives alter bare airframe dynamics to a point of instability.
3.1.3 Lower-Order Equivalent System Development.
In order to quantify the literal factors utilized in the objective criteria, MIL-
STD-1797 sets out a methodology to simplify the higher-order systems (HOS) into a
Lower-Order Equivalent System (LOES). In order to accurately model the HOS, the
LOES and HOS must be analyzed by evaluating the difference in gain and phase [5].
The overall goal is to minimize the cost function, described by Equation 1,
nω
20 X
J= [(GHOS − GLOES )2 + .02(φHOS − φLOES )2 ] (1)
n ω
1
37
where J is the overall cost, n is the number of frequencies, G is the gain in decibels,
and φ is the phase in degrees. The primary goal is to minimize the cost enough that
the difference in magnitude and phase between the HOS and LOES fits within bounds
described by MIL-STD-1979. Figure 20 depicts an acceptable LOES, as the differences
in both magnitude and phase remain between the black boundaries. Outside of those
limits indicates a LOES that may be inaccurate at that frequency, and should be used
with caution.
Gain Envelope
30
Limit
20
HOS−LOS
Gain [dB]
10
−10
−20
−2 −1 0 1 2
10 10 10 10 10
Frequency ω [rad/s]
Phase Envelope
200
150 Limit
Phase [degree]
100
HOS−LOS
50
−50
−100
−150
−2 −1 0 1 2
10 10 10 10 10
Frequency ω [rad/s]
Figure 20. Envelopes of Maximum Unnoticeable Added Dynamics Example
With the simplification of the HOS, a LOES allows the designer to extract im-
portant literal factors that drive the aircraft dynamics. For longitudinal aircraft
dynamics, the LOES takes the form of Equation 2,
θ Kθ (s + 1/Tθ1 )(s + 1/Tθ2 )e−τe s

= 2 (2)
δe (s + 2ζp ωp s + ωp2 )(s2 + 2ζsp ωsp s + ωsp
2 )
where ζp and ζsp represent the phugoid and short-period modal damping respectively.
38
Similarly, for lateral aircraft dynamics, the LOES takes the form of Equation 3,
φ Kφ (s2 + 2ζφ ωφ s + ωφ2 )e−τeφ s

= (3)
δa (s + 1/TS )(s + 1/TR )(s2 + 2ζd ωd s + ωd2 )
where ζφ and ζd represent the roll and Dutch roll modal damping respectively. These
literal factors described in both Equation 2 and Equation 3 are then used in various
graphics to depict where satisfactory, acceptable, and controllable flight levels exist.
In assessing the effects of altering stability and control parameters, a LOES was
developed for each scaled parameter. For changes to Mq and Mδ e , a longitudinal
LOES as described in Equation 2 was developed. For scaling of the parameters Lp ,
Lδ a , a lateral LOES as described in Equation 3 was developed. To ensure an accurate
model, MATLAB’s fmincon function matched a lower-order equivalent system to the
bare airframe by minimizing Equation 1. This was done utilizing a Bode plot analysis,
as described above.
3.2 Flight Maneuvers
As stated in Chapter II, varying levels of precision and aggressiveness are required
for an aircraft to complete a task maneuver. The report WL-TR-97-3100 lists ma-
neuvers based on precision and aggressiveness and are shown in Table 8 [31]. Three
flight maneuvers were chosen in the analysis of aircraft workload and performance: a
climbing spiral task, an aerial refueling task, and a side step landing maneuver. The
climbing spiral maneuver is representative of a non-precision, non-aggressive maneu-
ver while the aerial refueling task and side step landing task are representative of
precision, non-aggressive tracking tasks.
39
Table 8. Mission Task Elements [31]
3.2.1 Climbing Spiral.
The climbing spiral maneuver is a non-precision, non-aggressive maneuver rep-

resentative of a typical ISR task that a UAV would accomplish. The task includes
a simultaneous change in both altitude and heading for a specified period of time.
The flight controller should have enough control authority to complete the maneuver
without saturation issues and achieve satisfactory performance.
40
3.2.1.1 Control System Design.
In designing the aircraft control systems, the desire was to fully automate the
model thereby simulating a fully autonomous UAV. An autonomous UAV “receives
goals from humans and translates them into tasks to be performed without human
interaction” [51]. By automating control within the LJ-25D model, desired maneuvers
can be repeated and subsequent performance evaluated.
Prior to evaluation of the aircraft in completion of the maneuver, a longitudinal
controller was developed using classical control concepts. The longitudinal controller,
shown in Figure 21, incorporates three feedback loops and a Proportional, Integral,
Derivative (PID) controller. First, a pitch-attitude tracking autopilot was developed
closing the loop on pitch rate, q, and then pitch angle, θ. With satisfactory tracking of
pitch, altitude rate of climb, denoted by ‘z dot’ in the block diagram, can be fed back
and subtracted from a target rate of climb (ROC). The difference between the target
ROC and aircraft ROC is the error e, which is then sent to the PID controller. The
controller, modeled by Figure 22, was tuned to provide satisfactory rise time, settling
time, and overshoot. By varying Kp , Ki , and Kd , the proportional, integral, and
derivative gains, respectively, satisfactory aircraft ROC performance was achieved.
Figure 21. Block Diagram of Spiral Maneuver Longitudinal Controller
41
Figure 22. Block Diagram of a PID Controller [28]
The PID controller then outputs an error command, ec , which is added to θe , the
difference between the aircraft pitch angle and trimmed pitch angle. The result of
that is subtracted from q, which is now the controller output. The command output
enters an elevator actuator block modeled as a second-order system with position and
rate limits. Finally, that is sent to the bare airframe system which then outputs the
various aircraft parameters.
Next a lateral controller was developed to track a roll command. Figure 23 depicts
a block diagram of the proportional feedback controller used during the spiral com-
mand. A similar design to the pitch attitude tracker was utilized. First by feeding
back roll rate, p, in an inner loop then by closing an outer loop with roll angle φ
and subtracting that from the desired roll angle. The error is then again fed into a
second-order elevator actuator which outputs into the bare airframe dynamics. The
feedback gains for φ and p were tuned to provide satisfactory tracking performance
aircraft roll angles.
To ensure a coordinated turn, a directional controller was implemented to track
a zero degree sideslip angle, β. Figure 24 shows the proportional feedback controller
utilized to coordinate turns during the climbing spiral maneuver.
Finally, a throttle controller was designed to ensure a steady velocity during the
maneuver. Figure 25 depicts the throttle controller. It first feeds back aircraft ve-
42
Figure 23. Block Diagram of Spiral Maneuver Lateral Controller
Figure 24. Block Diagram of Spiral Maneuver Sideslip Controller
locity in KCAS, which is subtracted from the trim velocity and fed into a phase-lead
compensator. The output of that is added to the throttle trim position and then fed
into an engine model that limits both throttle position and throttle rate. The result is
sent to an engine scaling model utilizing table lookup data which outputs the thrust
to the bare airframe. Table 9 summarizes the gains used in the controllers developed
for the climbing spiral maneuver.
3.2.1.2 Target Inputs.
Two separate inputs were made for the climbing spiral maneuver. First, a ROC
input is sent to the longitudinal controller to trigger a climb at a certain point in
time. At that same time, a target roll angle is sent to the lateral controller, initiating
the turning portion of the climbing spiral. The trim speed is fed into the throttle
controller, prompting it to add throttle to maintain trim speed during the maneuver.
43
Figure 25. Block Diagram of Spiral Maneuver Throttle Controller
Table 9. Gains for Climbing Spiral Autopilots
Autopilot System Gain Parameter Control Parameter Value

Longitudinal Kp Proportional 1.5
Ki Integral 1
Kd Derivative 0.3
Kq Pitch Rate 0.952
Kθ Pitch Angle 1
Lateral Kp Roll Rate 0.28
Kφ Roll Angle 1.02
Directional Kβ Sideslip 4.0
Throttle KT Throttle 1
The sideslip controller continually minimizes the sideslip angle to zero throughout
the maneuver, ensuring a coordinated turn.
3.2.2 Side Step Maneuver.
Under report WL-TR-97-3100, the side step landing maneuver can be considered
‘Precision Offset Landing’ and thus is a precision, non-aggressive maneuver [31]. The
maneuver investigates lateral-directional control as well as longitudinal control, as it
requires the aircraft to roll while maintaining heading and flaring to maintain altitude.
44
In accomplishment of the side step maneuver, several controllers needed to be

employed to accurately control the pitch, roll, and yaw of the aircraft during the
maneuver. In order to track the correct ROC, the same controller used in the climbing
turn was again used in the side step. The climbing turn throttle controller also
provides adequate performance and was employed again for the side step maneuver.
A more complicated controller was developed for tracking ycg , or side step distance,
via aileron control. First, P , φ, ẏ, and ycg were closed via feedback loops, with the
target side step distance at the input. A lead compensator added control to the high
frequency areas of the response. Figure 26 depicts the block diagram of the controller.
Figure 26. Block Diagram of Side Step Maneuver Lateral Controller
A slightly different rudder controller was used to track heading instead of sideslip.
By tracking a heading of zero, this ensures that the aircraft’s nose remained as parallel
as possible in relation to the runway, as opposed to the relative wind in the lateral
axis. The heading controller fed back R and ψ. The addition of a lead compensator
ensured satisfactory control at higher frequencies. Figure 27 depicts the block diagram
of the controller.
Several gains, poles, and zeros were iterated in the design to obtain a satisfactory
45
Figure 27. Block Diagram of Side Step Maneuver Heading Controller
response from the system. Table 10 summarizes the gains used in the controllers
developed for the side step maneuver. Note the two lead compensators use the same
poles and zeros, but different gains.
Table 10. Gains for Side Step Autopilots
Autopilot System Parameter Control Parameter Value

Longitudinal Kp Proportional Gain 1.5
Ki Integral Gain 1
Kd Derivative Gain 0.3
Kq Pitch Rate Gain 0.952
Kθ Pitch Angle Gain 1
Lateral Kp Roll Rate Gain 0.28
Kφ Roll Angle Gain 1.02
Kẏ Horizontal Rate Gain 0.275
Ky Horizontal Distance Gain 1
K Lead Compensator Gain 0.778
z1 Lead Compensator Zero -1.5
p1 Lead Compensator Pole -37.5
Directional KR Yaw Rate Gain 1
Kψ Yaw Angle Gain 1.25
K Lead Compensator Gain 5
z1 Lead Compensator Zero -1.5
p1 Lead Compensator Pole -37.5
Throttle KT Throttle Gain 1
To setup the approach, the aircraft starts in a 750 foot per minute (FPM) descent
at 7000 ft MSL. In order to replicate a glide slope of approximately 3 degrees, the
46
aircraft trim speed was set to 140 knots. The gear was also set down, with flaps at 10
degrees, again to simulate an instrument approach setup. When the plane passes 6750
ft, approximately 10 seconds into the maneuver, the aircraft begins to track the side
step to an adjacent runway, which is at a 500 foot horizontal increment from where
the aircraft began its descent. The 6750 ft mark is considered the maneuver start
point, where all workload and performance data are first analyzed. In the vertical
plane, the aircraft tracks a height above touchdown of 50 ft above the imaginary
runway at 6250 feet; therefore the controller aims to flare at 6300 feet where a human
pilot would take control. Ground effect is not modeled in this simulation. The
parameters for the maneuver termination point are a 2% steady-state error in both
the z and y axis. When the aircraft is within 2% of the side step distance of 500 ft and
within 2% vertical height of 6300, and remains within those bounds for the remaining
simulation time, the termination point is marked. All workload and performance data
are analyzed from the maneuver start point to the maneuver termination point.
3.3 Evaluation of Maneuver Workload and Performance
Previous research in UAV flying qualities by Hamidani [26] and Kim [30] focused
in assessing the UAV’s workload and performance during a climbing turn maneuver,
a pitch tracking maneuver, and an altitude tracking maneuver. The software utilized
in both Hamidani and Kim’s research, JSBSim, only output actuator position, and
not actuator rate. Therefore, rate limiting had to be inferred from position rate
graphs. Previous UAV research [30] recommended investigating the effects of rate
limiting on UAV performance and workload. In manned fixed-wing flying qualities,
pilot induced oscillation (PIO) has been known to be caused by “nonlinear events
such as saturation of control rate or position limits at too low a command” [37].
Rate limiting has been known to be a cause of PIO in manned aircraft for some
47
time [32]. Although PIOs are not an issue in UAVs, system induced oscillations
(SIO) may become problematic during precision aggressive maneuvers [30]. The LJ-
25D model allows for the investigation into actuator rate limiting and its effects on
aircraft workload and performance during a task.
3.3.1 Aerial Refueling.
Under report WL-TR-97-3100 lists, the aerial refueling maneuver can be consid-
ered ‘tanker boom tracking’ and thus is a precision non-aggressive maneuver [31].
The maneuver elicits precision control by simultaneously tracking altitude, roll angle,
and airspeed.
In designing a longitudinal control system, a similar approach to the ROC con-

troller was taken. Again, the pitch-attitude tracking autopilot previously developed
was utilized for the two inner loops of the controller. The final outer loop was then
modified to feedback aircraft altitude. Instead of a PID compensator, a lead-lag
compensator, shown in Equation 4, was chosen to provide a robust design to track
altitude.
s + z1 s + z2
Lead Lag Compensator = K (4)
s + p1 s + p2
Gain K is varied based on task: a large step 1000 foot altitude change requires a
+
lower gain than an aerial refueling tracking task with a margin of 10 feet. Table 11
depicts the values for K, T1 and T2 for large and small amplitude altitude tracking
tasks.
For roll, sideslip, and throttle control, the previously developed controllers uti-
lized in the climbing spiral provided satisfactory performance in completing an aerial
48
Table 11. Lead Lag Compensator for Altitude Control Design Values
Task K z1 p1 z2 p2
Large Amplitude Tracking -1 -1.0282 -9.3469 -0.0714 -0.0060
Precise Amplitude Tracking -4.87 -1.0282 -9.3469 -0.0714 -0.0060
refueling task. Therefore, these controllers were not modified for the aerial refueling
task.
In order to simulate an aerial refueling task, the tracking input for both target
altitude and roll angle needed to mimic the randomness of turbulent airflow at alti-
tude. Previous research [26] utilized a multi-sine input for a pitch attitude tracking
task and an altitude tracking task. These were accomplished separately as two tasks;
this research will employ a multi-sine input to both altitude tracking and roll track-
ing to adequately model a UAV controller adjusting to maintain refueling tracking
tolerances. In addition, the throttle controller will be set to maintain trim speed by
making adjustments to thrust. Equation 5 depicts the form of the multi-sine input
used for altitude and roll command inputs.
hCM D − or − φCM D = A1 sin(ω1 t + φ1 ) + A2 sin(ω2 t + φ2 ) + A3 sin(ω3 t + φ3 ) (5)
3.3.2 Maneuver Workload Metrics.
The workload is the relationship between resource supply and task demand [43].
Previous research [26, 30] employed the L2 error norm of the control surface deflection
away from the trimmed condition as primary metric of workload. A value of zero
indicates the control system did no work during the maneuver, and a value of 1
indicates the control system work was maximized during the maneuver. The L2 error
49
norm is shown in Equation 6:
v
u n
u1 X
L2 = t (x1 (i) − x2 (i))2 (6)
n i=1
In addition to assessing the L2 error norm for elevator, aileron, and rudder deflection
positions during a maneuver, the L2 error norm for elevator, aileron, and rudder
deflection rates will also be assessed.
3.3.3 Maneuver Performance Metrics.
Originally developed for economic forecasting [47], Theil’s Inequality Coefficient

(TIC) has been adapted to assess time histories of simulation to determine their
accuracy. It is primarily used in assessing nonlinear systems such as the LJ-25D model
utilized in this thesis, and “culminating in the validation of an aircraft simulation
model using flight test data” [20]. TIC is shown in Equation 7.
q P
1 n
n i=1 (xi − x̃i )2
T IC = q P q P (7)
1 n 2 1 n
n i=1 xi + n i=1 x̃2i
where x is the simulation time history, x̃ is the desired track, and n is the number
of data samples. A TIC of less than 0.25 is considered an accurate model. Ideally,
correlations can be made between workload metrics from the L2 error norm of actuator
positions, actuator rates, and TIC performance metrics.
3.4 Summary
An explanation of the LJ-25D Simulink model and its development at the USAF
Test Pilot School was presented. A discussion of the climbing spiral maneuver, aerial
refueling maneuver, and side step maneuver was provided, along with the method-
ology of developing the controllers for each maneuver. The L2 error norm and TIC
50
equations showed how workload and performance are calculated for each mission task.
An analysis of the LJ-25D model during each flight maneuver using the evaluation
techniques described will be presented in Chapter IV.
51
IV. Results
4.1 Bare Airframe Analysis
Before altering the stability and control parameters, a bare airframe analysis of
the LJ-25D was conducted to establish a baseline of aircraft parameters. First, the
aircraft was trimmed, and using previously described techniques in Section 3.1.3, a
LOES was developed for both the longitudinal and lateral-directional stability axes.
The aircraft was trimmed to a flight path angle of 0 degrees at 200 KCAS and 5000
ft, with zero flaps and gear up. Table 12 shows the longitudinal literal factors in
baseline trim condition for the LJ-25D.
Table 12. Longitudinal Trim Results, 200 KCAS 5,000 ft
Kc Tθ1 Tθ2 τθ ωp ωsp ζp ζsp ωsp Tθ2 CAP J

-1.27 2.98 2.97 0.01 0.23 0.83 -0.30 0.76 2.48 0.33 12.47
First, the cost function value using Equation 1 of 12 indicates an accurate model.
For the purposes of this analysis, an acceptable model has a cost value under 25 and
remains within the MIL-STD-1979 HOS − LOS bounds. When examining the poles
and zeros, any negative number indicates the pole or zero exists in the right half plane
(RHP), indicating it is unstable. A positive number indicates the pole or zero lies
in the left half plane (LHP), and thus stable. Table 12 indicates the bare airframe
has a satisfactory short-period damping (ζsp ) value of 0.76. The phugoid mode,
denoted by ζp , has a value of -0.30. The second order transfer function associated
with the phugoid mode can be written as Equation 8. The poles of the polynomial
in Equation 8 are 0.0695 ± 0.219i, indicating that the pair of poles associated with
the phugoid mode lie in the RHP and thus are unstable.
1 1
= (8)
(s2 + 2ζp ωp s + ωp2 ) (s2 − 0.139s + 0.0529)
52
The controllers developed for each maneuver mitigate the unstable phugoid mode.
Figure 28 shows the Bode plot analysis of the longitudinal LOES developed for the
bare airframe. Figure 28a shows the HOS and LOES are nearly identical Bode plots
in both magnitude and phase and Figure 28b confirms that with the HOS − LOS
envelope plot.
(a) Overlay Bode Plot (b) Envelope Plot
Figure 28. Longitudinal LOES for Baseline Bare Airframe, 200 KCAS 5,000 ft
An analysis of the bare airframe lateral-directional dynamics was also completed.

Table 13 shows the lateral-directional literal factors in baseline trim condition for the
LJ-25D.
Table 13. Lateral-Directional Trim Results, 200 KCAS 5,000 ft
Kφ ζφ ωφ τ Tr Ts ζd ωd BW J
-3.83 6.70 3.26 0.045 -0.016 1.18 -0.25 0.67 2.04 23.89
From the baseline analysis, the Dutch roll mode, denoted by ζd , is -0.25. Utilizing the
same analysis technique used in Equation 8, it is determined the poles associated with
the Dutch roll mode lie in the RHP, indicating instability. The zero Tr representing
the roll mode time constant also lies in the RHP. That is not as concerning for an
aircraft developer but does indicate that the plane eventually would demonstrate
53
instability for that mode. Figure 29 shows the Bode plot analysis of the lateral-
directional LOES developed for the bare airframe. Similar to the longitudinal Bode
plots, the lateral-directional Bode plots indicate the LOES is an accurate model.
(a) Overlay Bode Plot (b) Envelope Plot
Figure 29. Lateral LOES for Baseline Bare Airframe, 200 KCAS 5,000 ft
4.1.1 Stability Parameter Scaling Analysis.
In order to do an accurate analysis of literal factors, a LOES was developed for

every stability parameter at every scaling factor. Cost function was used to determine
if the LOES accurately modeled the HOS. The LOES is based on the LJ-25D at
5,000 ft and 200 KCAS, which were the parameters utilized in the climbing turn
and aerial refueling maneuver. Separate LOESs were developed for the side step
maneuver, which required a different altitude, airspeed, flap setting and gear setting.
Those LOES used for the side step maneuver slightly differ from the LOES used in
the climbing spiral maneuver and aerial refueling maneuver. The lone outlier is the
1.5 x Melev , which had a significantly different LOES for the side step maneuver as
compared to the LOES for climbing turn and aerial refueling. Bode plots of the LOES
overlaid on the HOS as well as the MIL-STD-1797 plots of HOS − LOES within gain
and phase envelopes can be found in Appendix A.
54
Table 14. Mq LOES Specifications, 200 KCAS 5,000 ft
Scale Kc Tθ1 Tθ2 τθ ωp ωsp ζp ζsp ωsp Tθ2 CAP J

0.25 x -1.27 3.12 3.12 0.01 0.26 0.70 -0.51 0.72 2.18 0.24 10.17
1x -1.27 2.98 2.98 0.01 0.23 0.83 -0.30 0.76 2.48 0.33 12.47
2x -1.27 2.74 2.74 0.01 0.20 1.03 -0.09 0.84 2.82 0.47 15.51
4x -1.30 2.60 2.60 0.01 0.17 1.28 0.25 1.08 3.32 0.68 6.70
6x -1.34 1.35 1.35 0.01 0.47 0.69 1.63 2.27 0.93 0.10 20.08
Table 14 presents the LOES developed from scaling Mq . Mq was the sole stabil-
ity parameter capable of being scaled greater than twice its original value without
sacrificing the integrity of the LOES. Note that as Mq increases, phugoid damping
changes from unstable to stable and short-period becomes more stable, as indicated
by the increasing ζsp .
Next, Melev was scaled and a LOES developed at each point. Table 15 depicts the
LOES developed from scaling Melev .
Table 15. Melev LOES Specifications, 200 KCAS 5,000 ft
Change Kc Tθ1 Tθ2 τθ ωp ωsp ζp ζsp ωsp Tθ2 CAP J

0.25 x -0.35 6.36 6.36 0.01 0.28 0.66 -0.48 0.84 4.19 0.44 2.36
0.75 x -0.96 3.32 3.32 0.01 0.25 0.82 -0.38 0.76 2.73 0.36 11.44
1x -1.27 2.98 2.98 0.01 0.23 0.83 -0.30 0.76 2.48 0.33 12.47
1.25 x -1.57 2.65 2.65 0.01 0.20 0.84 -0.16 0.78 2.22 0.30 13.11
†
1.5 x -1.89 2.42 2.42 0.01 0.17 0.78 0.18 0.85 1.88 0.23 6.47
∗
1.5 x -2.69 4.91 0.01 0.01 0.16 10.90 0.16 6.79 0.11 0.27 10.51
† indicates model used for climbing turn and aerial refueling maneuver
* indicates model used for side step maneuver
Melev did not experience as much of a change in short-period damping, but did
yield a steady increase in phugoid damping, again changing from unstable to stable.
An important point is the different models used at 1.5 x Melev . The side step maneuver
at 1.5 x Melev has vastly different literal factors from the models used during the
climbing turn and aerial refueling. The difference is attributed to the changes in
altitude, velocity, gear extension, and flap deployment. This is the only instance
55
where LOES significantly differed across the three maneuvers. Also, past research
1
was criticized for large changes in Melev ( 10 x to 10 x) as unrealistically large changes
to the sizes of the control surfaces. This research attempts to mitigate that by keeping
the changes smaller and thus within the realm of possibility for an aircraft designer.
Next, lateral-directional LOES’s were developed at each scaled value of Lp . Ta-
ble 16 shows the LOES developed from scaling Lp .
Table 16. Lp LOES Specifications, 200 KCAS 5,000 ft
Change Kφ ζφ ωφ τ Tr Ts ζd ωd BW J
0.25 x -18.83 6.61 3.24 0.03 -0.003 1.24 -0.30 0.69 2.04 23.08
0.75 x -17.82 6.65 3.24 0.03 -0.003 1.21 -0.27 0.68 2.04 23.60
1x -3.83 6.70 3.26 0.05 -0.016 1.18 -0.26 0.67 2.04 23.89
1.25 x -4.51 6.60 3.20 0.04 -0.014 1.17 -0.24 0.66 2.03 24.15
1.5 x -20.00 6.74 3.26 0.03 -0.003 1.15 -0.22 0.66 2.03 24.40
To note on Lp LOES is the Dutch roll mode damping, as it remains unstable through-
out the changes but does move towards the LHP as Lp increases. The roll mode
time constant, Tr also remains unstable throughout the scaling. Spiral mode time
constant, Ts steadily decreases, indicating an increasing spiral mode time constant
and improved lateral handling qualities. That, combined with the steady increase
in Dutch roll damping, indicates workload should decrease and performance increase
during each maneuver as Lp is scaled up.
Finally, Table 17 presents the LOES developed from scaling Lail .
Table 17. Lail LOES Specifications, 200 KCAS 5,000 ft
Change Kφ ζφ ωφ τ Tr Ts ζd ωd BW J
0.5 x 1.37 26.48 0.004 0.01 -15.62 1.37 -0.28 0.65 1.48 1.25
0.75 x -3.91 6.80 3.62 0.04 -0.01 1.17 -0.25 0.67 1.78 25.59
1x -3.83 6.70 3.26 0.05 -0.02 1.18 -0.26 0.67 2.04 23.89
1.25 x -5.03 7.01 3.23 0.04 -0.01 1.19 -0.26 0.67 2.27 22.78
1.5 x 4.08 0.03 0.10 0.01 -300.00 -1.66 -10.00 0.03 2.40 620.87
The 1.5 x Lail LOES was not used due to its high cost value, over 600, indicating
the LOES does not match the HOS. The Bode and envelope plots in Figure 78,
56
Appendix A verify the inaccuracy graphically as well. The 1.5 x Lail points are
included in the analysis for trend identification purposes but should be used cautiously
when drawing conclusions.
4.2 Climbing Spiral Analysis
The first maneuver analyzed was the non-precision, non-aggressive climbing spiral
maneuver. First, a baseline analysis was completed varying bank angle and climb
rate. Bank angle and climb rate were then fixed for the remaining runs which varied
the stability and control parameters.
4.2.1 Baseline Climbing Spiral.
Before altering the stability and control parameters of the bare airframe, a short
performance analysis was completed on the aircraft with controllers in the loop. Three
separate runs were completed, each starting at a trim condition of 200 KCAS and
5,000 ft, for a duration of 180 seconds. The exact flight conditions are shown in
Table 18 and the resulting aircraft paths in Figure 30.
Table 18. Baseline Climbing Spiral Flight Conditions, 200 KCAS 5,000 ft
Flight Condition (FC) Rate of Climb (FPM) Bank Angle (deg)

1 1000 10
2 1500 20
3 2000 30
Further analysis was completed to determine L2 norms of actuator rate and po-
sition as well as TIC values for rate of climb, roll angle, and velocity. Figure 31a
and Figure 31b depicts these values for elevator, aileron, and rudder actuator rates
and actuator positions, respectively. Aileron and rudder actuator rates and actuator
positions increase slightly with maneuver aggressiveness, while elevator rates and po-
sitions remains constant. Throttle sees a steady increase in position as aggressiveness
57
Figure 30. Baseline Climbing Spiral Maneuver
increases.
58
(a) L2 Norm of Actuator Rate (b) L2 Norm of Actuator Position
(c) TIC Values
Figure 31. Workload and Performance Metrics of Baseline Climbing Spiral
These actuator position and rate values indicate that performing the most aggres-
sive climbing spiral maneuver, FC 3, takes slightly more work from the aileron and
rudder, but primarily more throttle to achieve. Figure 31c shows the TIC values for
roll angle, rate of climb, and velocity for the three flight conditions remain nearly
constant across flight conditions.
After assessing the data, FC 3 was chosen as the nominals to be utilized moving
forward with stability and control parameter scaling runs of the climbing spiral. As
evidenced by the nearly identical L2 norms and TIC values across all flight conditions,
FC 3 provides an ample level of aggressiveness in the turning spiral and still maintains
59
adequate aircraft performance throughout the maneuver.
In order to assess how the stability and control parameters affect aircraft workload
and performance, the four stability and control parameters Mq , Mδ e , Lp , and Lδ a were
scaled according to their values in Section 4.1.1. Subsequent runs were completed
using Flight Condition 3: 2000 FPM climb and 30 degree bank angle.
4.2.2.1 Mq Scaling.
The first parameter Mq represents the pitching moment due to pitch rate, and
was scaled by 0.25 x Mq , 1 x Mq , 2 x Mq , 4 x Mq , and 6 x Mq . Figure 32 shows the
resulting five aircraft tracks plotted on a single graph. Visual inspection shows each
of the tracks is nearly identical, a result of the non-precision, non-aggressive nature of
the maneuver. However, important data can be gathered from the actuator positions
and rates as well as the literal factors associated with the bare airframe dynamics.
Figure 33a is a view of how the L2 norm of elevator, aileron, and rudder actuator
rates change with Mq scaling. For this particular plot, the elevator rate L2 norm
steadily decreases as Mq is scaled: from 11% of its maximum elevator rate to about
6%. Figure 33b shows elevator, aileron, rudder, and throttle actuator positions stay
nearly constant as Mq is scaled. Finally, Figure 33c depicts the TIC values of ROC,
bank angle, and velocity as Mq is scaled. TIC of ROC appears to steadily increase as
Mq is scaled, but the y-axis scale indicates it is a small amount. TIC of bank angle
and TIC of velocity stay nearly constant.
Figure 34 presents a four pane of aircraft analysis graphs. Figure 34a and Fig-
ure 34b are two analysis plots from MIL-STD-1797 (Figures 44 and 45 in MIL-STD-
1797, respectively). The bare airframe was plotted utilizing the aircraft literal factors
60
Figure 32. Mq Scaled Climbing Spiral Maneuver
derived from the LOES and aircraft level determined via boundaries set in MIL-
STD-1797. According to the MIL-STD-1797 graphs, points 0.25 x Mq through 4 x
Mq remain within Level 1 flying qualities, but 6 x Mq strays outside that bound.
This is driven by the high ζsp value of 1.63, indicating the aircraft controls are very
sluggish causing difficulty for the pilot to control the aircraft.
In addition to assessing the bare airframe dynamics based off of MIL-STD-1797
criteria, additional criteria were created to assess performance. In the case of Fig-
ure 34c, TIC is plotted vs ζsp , with ζsp boundaries for Level 1, 2, and 3 flying qualities
from MIL-STD-1797. In this case, a high ζsp will cause the aircraft to be overdamped
and not have the maneuverability to complete the task. A low ζsp will cause the
aircraft to be borderline unstable, again unsuitable for the task. TIC then accounts
61
(c) TIC Values
Figure 33. Workload and Performance Metrics of Mq Scaled Climbing Spiral
for how well the aircraft performs the task, with cutoff points for Level 1 at a TIC of
0.15, Level 2 at 0.25, and Level 3 for any TIC values higher than 0.25. These numbers
were chosen based on previous research by Kim [30] and are subject to modification
as more data is gathered. The advantage of this plot is that it can identify what the
aircraft designer needs to change when aircraft performance is unsatisfactory. If the
TIC is acceptable but the ζsp is outside of the limits, then the bare airframe dynamics
of the aircraft should be adjusted. If the bare airframe has good flying characteristics
but the TIC is too high, then the controller needs to be modified to allow the aircraft
to fly the mission task. Altering controller gains demonstrates the case of a high TIC
62
(a) ωsp Tθ2 vs ζsp (b) CAP vs ζsp
(c) TIC vs ζsp

(d) L2 Norm vs TIC ROC
Figure 34. Flying Qualities Graphs of Mq Scaled Climbing Spiral
occurring with acceptable levels of ζsp .

In this particular case of scaling Mq , all values are within Level 1 limits with the
exception of 6 x Mq , which has a ζsp of 1.63. Although the aircraft model exhibits
acceptable performance of the maneuver, as indicated by a TIC value of around 0.05,
the overdamped bare airframe will cause sluggish control. A more aggressive or more
precise maneuver may not be possible with that bare airframe design.
Figure 34d was created in order to identify further trends in actuator positions
and rates with parameter scaling. The colors depict which scaling factor was applied
63
for that data point, and the symbols represent either the aileron actuator or elevator
actuator’s L2 norm of rate or position. The plot provides a singular view of how
scaling a specific stability and control parameter affects the workload of the actu-
ators. It also gives perspective on how much more workload is being done and the
corresponding performance at that workload. In this case of scaling Mq , the workload
for the elevator actuator bottoms out at 2 x Mq to 4 x Mq . The low values for L2
norm of aileron actuator position and rate show that changes to Mq have no effect on
aileron workload.
Figure 34d highlights a weakness of TIC. On the x-axis is TIC of ROC, which
indicates Level 1 flying qualities for all scaled values of Mq . Yet the MIL-STD-1797
plots, which utilize the literal factor ζsp , indicate Level 3 flying qualities for 6 x Mq .
Both are correct, as a ζsp of 1.63 is too high and Figure 32 shows aircraft performance
is satisfactory during the climbing maneuver. But because the climbing turn is a
benign maneuver, a high ζsp will not effect performance, and hence provide a low
TIC value. The MIL-STD-1797 graphs are based on bare airframe literal factors and
applicable in all flight phases, indicating that for maneuvers requiring more aggression
or precision the aircraft will display Level 3 flying qualities at 6 x Mq . TIC, on
the other hand, does not take bare airframe dynamics into account, and therefore
satisfactory performance during a non-precision, non-aggressive maneuver, should
not be extrapolated to precision or aggressive maneuvers. A designer can avoid this
pitfall by utilizing the hybrid Figure 34c, which incorporates the both critical bare
airframe literal factor ζsp and aircraft performance for that particular maneuver.
4.2.2.2 Mδ e Scaling.
The next parameter scaled was Mδ e , the pitching moment due to elevator deflec-
tion, and was scaled by 0.25 x Mδ e , 0.75 x Mδ e , 1 x Mδ e , 1.25 x Mδ e , and 1.5 x Mδ e .
64
Figure 35 depicts the resulting climbing turn while scaling Mδ e .
Figure 35. Mδ e Scaled Climbing Spiral Maneuver
Figure 36a and Figure 36b show an interesting interaction between elevator actuator
rates and positions as Mδ e is scaled. The L2 norm of elevator rate increases while the
L2 norm of elevator position bottoms out at 0.75 x Mδ e and then steadily increases
upward. The scaling of Mδ e essentially increases the effectiveness of the elevator,
which explains the large drop off in L2 norm position between 0.25 x Mδ e and 0.75
x Mδ e . TIC of ROC steadily decreases with increasing Mδ e , but the changes are
small. TIC of roll angle and TIC of velocity stay constant. In assessing these results,
a designer could decrease elevator size by 25% (from 1 x Mδ e to 0.75 x Mδ e ) to limit the
workload of the elevator while still maintaining adequate climbing turn performance.
The subfigures of Figure 37 all indicate Level 1 flying qualities for all values of
65
(c) TIC Values
Figure 36. Workload and Performance Metrics of Mδ e Scaled Climbing Spiral
Mδ e . Figures 37a, 37b, and 37c all show a trends towards Level 2 flying qualities,
but only on the graph of ωsp Tθ2 vs ζsp does 1.5 x Mδ e come close to that boundary.
Figure 37d shows a high aileron actuator L2 norm at 0.25 x Mδ e , but the TIC of ROC
is relatively low regardless of scaling factor. The L2 norms of elevator actuator rate
also remain low across the board, indicating once the proper deflection is achieved
for the bank and climb, the controller only needs to make small changes to maintain
good performance. Another point to note is the consistent low values for L2 norm of
aileron actuator position and rate, indicating the changes to Mδ e have no effect on
aileron workload and thus no cross-coupling effects were present.
66
(c) TIC vs ζsp (d) L2 Norm vs TIC ROC
Figure 37. Flying Qualities Graphs of Mδ e Scaled Climbing Spiral
4.2.2.3 Lp Scaling.
Next, the lateral stability parameter Lp representing the roll moment due to roll
rate was scaled by 0.25 x Lp , 0.75 x Lp , 1 x Lp , 1.25 x Lp , and 1.50 x Lp . Fig-
ure 38 shows the resulting five aircraft courses plotted on a single graph. Due to the
non-aggressive, non-precision nature of the maneuver, visual inspection of the five
maneuvers appears to indicate similar results.
Despite the scaling of Lp , Figures 39a and 39b indicate the L2 norm of all actuator
rates and positions stay relatively constant across the board. In addition, TIC values
for ROC, velocity, and roll angle all remain nearly constant as well. This indicates
67
Figure 38. Lp Scaled Climbing Spiral Maneuver
that changes in Lp on the scales demonstrated appear to have little to no effect on

the workload and performance of the LJ-25D in a climbing spiral maneuver.
4.2.2.4 Lδ a Scaling.
Finally, the lateral control parameter Lδ a was scaled by 0.5 x Lδ a , 0.75 x Lδ a ,

1 x Lδ a , 1.25 x Lδ a , and 1.5 x Lδ a . Figure 40 depicts the climbing maneuver with
the scaled values of Lδ a . Visual inspection of the graph indicates that there is a
variation between each run, as the turns have a slightly wider radius with increasing
Lδ a . Whether or not that the wider turns are closer to the ideal bank angle of 30
degrees will have to be determined by the data.
Figure 41c shows that as Lδ a is scaled up, the TIC of roll angle decreases. Although
68
(c) TIC Values (d) L2 Norm vs TIC of Roll Angle
Figure 39. Workload and Performance Metrics of Lp Scaled Climbing Spiral
the difference in TIC of roll angle between 0.5 x Lδ a and 1.5 x Lδ a is only 0.01,
this supports the notion that a larger control surface provides more control and can
thus slightly increase aircraft turning performance in the climbing spiral maneuver.
Figure 41d also affirms previous conclusions that larger control surfaces both decrease
workload and increase turning performance. No improvements were seen in velocity
tracking performance or ROC performance.
69
Figure 40. Lδ a Scaled Climbing Spiral Maneuver
70

4.2.3 Longitudinal Controller Gain Scaling.
In addition to scaling stability and control parameters and observing their effects
on performance and workload, controller gains were scaled to observe and compare
the effects as well. For the climbing spiral maneuver, Kc , the proportional gain of
the PID controller for rate of climb, was scaled according to Table 19. The high end
value of 2.25 x Kc was chosen as it nears the limit of an uncontrollable aircraft.
Table 19. Kc Scaling Factors for Climbing Turn
Run Number Kc Scaling Factor

1 0.25 x
2 0.5 x
3 1x
4 2x
5 2.25 x
Figure 42. Kc Scaled Climbing Spiral Maneuver
Observing the visual tracks of the climbing spiral with Kc variations clearly in-
71
dicates varying performance levels, particularly at values of 2 x Kc and 2.25 x Kc .
Figure 43a indicates that with changing the gains, the elevator actuator is more active
at higher gains. Figure 43b shows elevator position also follows suit, and sees increases
in workload at higher gain values. An interesting point is the change in throttle con-
trol, as the data indicates about a 20% throttle increase from the baseline 1 x Kc to
the 2.25 x Kc run. In previous runs, throttle control stayed constant throughout the
changes to all stability and control derivatives. Finally, Figure 43c indicates varying
trends for the three TIC calculations. First, TIC ROC bottoms out at 1 x Kc , with
higher values at both low and high gain changes. TIC velocity steadily increases with
the scaling of Kc , indicating there were unforeseen consequences on the performance
of the throttle controller. TIC roll angle remains constant throughout the maneuver.
72
(c) TIC Values
Figure 43. Workload and Performance Metrics of Kc Scaled Climbing Spiral
Figures 44a and 44b show every point stacked on 2.25 x Kc because the bare air-
frame dynamics remain unchanged in this sequence. Figure 44c presents an interesting
trend in TIC vs ζsp not previously observed. As Kc scales up, ζsp remains constant,
but TIC of ROC increases. Therefore on a plot of TIC vs ζsp , large changes in TIC
of ROC without an accompanying change in ζsp indicates to an aircraft designer that
a change in the elevator controller is needed to increase aircraft performance. This
idea will be further fortified with scaling of Kc during the side step maneuver.
73

(c) TIC vs ζsp
Figure 44. Flying Qualities Graphs of Kc Scaled Climbing Spiral
Lastly, Figure 44d shows another new trend in data, with the L2 norm of elevator
position and rates showing a trade off between performance and workload. The
highest performance, as indicated by the lowest TIC ROC value, is at 1 x Kc , but
it is not the lowest workload. There is a slightly lower workload on 0.5 x Kc and
0.25 x Kc , but at the cost of a higher TIC. 2 x Kc and 2.25 x Kc are the worst
cases, as workload increases and performance decreases. Therefore using Figure 44d,
a designer could modify controller gains to weight aircraft performance or aircraft
workload, as well as make a trade-off between the two.
74
4.3 Side Step Landing Analysis
The second maneuver assessed was the side step landing task. This precision,
non-aggressive task requires both lateral and longitudinal tracking. Lateral tracking
requires the use of aileron and rudder to center the aircraft on the runway and main-
tain proper heading, as well as elevator and throttle control to achieve a controlled
descent and flare at the proper time.
4.3.1 Baseline Side Step Maneuver.
In order to properly gauge the results of scaling stability and control parameters,
a baseline run was conducted to achieve datum values for workload and performance.
The target parameters were 140 KIAS, 750 FPM descent rate, gear down and flaps at
10 degrees, as described in Section 3.2.2.2. Figures 45a and 45b show a perspective
and top-down view, respectively, of the baseline side step maneuver.
75
(a) Perspective View
(b) Top-down View
Figure 45. Baseline Side Step Maneuver
As part of the analysis of the side step maneuver, the stability and control param-
eters Mq , Mδ e , Lp , and Lδ a were scaled according to their values in Section 4.1.1. The
76
simulations were utilized to calculate L2 norm and TIC values, and then assessed via
MIL-STD-1797 and newly developed criteria.
4.3.2.1 Mq Scaling.
The first scaled parameter was Mq , representing the pitching moment due to pitch
rate, and was scaled by 0.25 x Mq , 1 x Mq , 2 x Mq , 4 x Mq , and 6 x Mq . Figure 46
shows the resulting five aircraft tracks. The black circle represents the maneuver start
point for all runs, where all L2 norm and TIC calculations begin, and each colored X
in the figure represents where each individual run ended, and thus all L2 norm and
TIC calculations stop. The finality criterion was determined according to parameters
explained in Section 3.2.2.2.
77
(b) Side View
Figure 46. Mq Scaled Side Step Maneuver
Figure 47 shows the workload and performance of the LJ-25D during the side step
maneuver. First, L2 norms for aileron and rudder remain unaffected with changes
to Mq . Elevator workload varies non-linearly, as evidenced by the elevator curves in
both Figure 47a and Figure 47b. There is a local minimum at 2 x Mq for L2 norm
of elevator rate, but despite this valley, L2 norm of elevator position always grows
as Mq increases. This matches the trend with TIC of ROC, which steadily increases
with increasing Mq . TIC of velocity stays at nearly zero for all runs, indicating good
velocity tracking at all values of Mq . For the side step maneuver no TIC of roll angle
is calculated because lateral distance was utilized as the command input. To avoid an
overly complex controller, only one tracking input was made to the aileron controller.
The results in Figure 47 present the aircraft designer with the ability to alter Mq
78
between 0.25 x to 2 x the current values with minimal effect on ROC performance.
(c) TIC Values
Figure 47. Workload and Performance Metrics of Mq Scaled Side Step
MIL-STD-1797 charts present a similar conclusion that longitudinal flying perfor-

mance steadily decreases with increasing Mq . Figures 48a and 48b have 4 x Mq and
6 x Mq well beyond Level 2 flying qualities limits. Figures 48c and 48d concur with
conclusions of the MIL-STD-1797 criteria. The values for 0.25 x Mq , 1 x Mq , and 2
x Mq all remain well within the Level 1 flying qualities boundaries, while 4 x Mq and
6 x Mq stray outside to Level 3. Figure 48c indicates that 4 x Mq is unsatisfactory
based on the ζsp limits and 6 x Mq is unsatisfactory due to both TIC of ROC and
ζsp . Figure 48d shows that based on L2 norm values and TIC values, the aircraft is
79
on the verge of Level 3 flying qualities at high values of Mq . The results of scaling
Mq during the side step maneuver shows that mathematical tools such TIC and L2
norm can provide an alternative classification of flying qualities levels to the classical
literal factors in MIL-STD-1797.
Figure 48. Flying Qualities Graphs of Mq Scaled Side Step
The next parameter assessed during the side step maneuver was Mδ e , scaled by
0.25 x Mδ e , 0.75 x Mδ e , 1 x Mδ e , 1.25 x Mδ e , and 1.5 x Mδ e . Figure 49 depicts the
side step maneuver while scaling Mδ e .
80
(b) Side View
Figure 49. Mδ e Scaled Side Step Maneuver
81
0.75 x Mδ e to 1.5 x Mδ e all have similar path performance throughout the maneuver,
but 0.25 x Mδ e has noticeably more overshoot and steady-state error as compared to
the other runs. This is confirmed in the data shown in Figure 50c, which indicates a
large drop off in TIC from 0.25 x Mδ e to 0.75 x Mδ e . A similar trend is depicted in
Figures 50a and 50b, which exhibit a minimum workload at 0.75 x Mδ e . Workload
then increases from 0.75 x Mδ e to 1.5 x Mδ e , with a slight increase in performance as
well. The trends would give an aircraft designer the option to increase performance
slightly at the expense of a higher elevator actuator workload.
(c) TIC Values
Figure 50. Workload and Performance Metrics of Mδ e Scaled Side Step
The subfigures of Figure 51 show an interesting dynamic when comparing the
82
MIL-STD-1797 results to the developed graphs. 0.25 x Mδ e , which is Level 1 in
Figures 51a and 51b, is Level 3 in Figure 51d due to the high TIC value for ROC.
The opposite is true for 1.5 x Mδ e , in which Figures 51a and 51b identify the point
as Level 3, with a ζsp of 6.7, but Figure 51d indicates the configuration is Level 1.
Figure 51c, which combines aspects of the MIL-STD-1797 literal factors and the TIC
performance metric, correctly identifies both 0.25 x Mδ e and 1.5 x Mδ e as Level 3.
This further supports the notion of classifying UAV handling qualities using aspects
of the literal factors criteria in MIL-STD-1797 and new criterion such as TIC.
(c) TIC vs ζsp

Figure 51. Flying Qualities Graphs of Mδ e Scaled Side Step
83
4.3.2.3 Lp Scaling.
The lateral stability parameter Lp representing the roll moment due to roll rate
was scaled by 0.25 x Lp , 0.75 x Lp , 1 x Lp , 1.25 x Lp , and 1.50 x Lp . Figure 52 shows
the resulting aircraft tracks. When observing the top-down view shown in Figure 52b,
its clear the 0.25 x Lp has degraded performance when approaching the runway, and
continues on a limit cycle of approximately ± 10 feet when centered on the runway.
84
(b) Top-down View
Figure 52. Lp Scaled Side Step Maneuver
The poor performance of 0.25 x Lp evident in Figure 52 manifests itself in the L2

norm of aileron actuator position and rate. The high L2 norm values for aileron in
Figures 53a and 53b for 0.25 x Lp drop off at 0.75 x Lp , where the L2 norm and TIC
of ROC values remain relatively constant for the remainder of the scaling. Figure 53c
indicates that the poor performance and high aileron workload put 0.25 x Lp into
Level 2 flying qualities, informing the aircraft designer changes should be made. An-
other interesting point is that thus far cross-coupling effects between longitudinal and
lateral have been negligible. Here, the lateral parameter Lp is effecting a longitudinal
performance metric, TIC of ROC.
85
Figure 53. Workload and Performance Metrics of Lp Scaled Side Step Maneuver
Lastly, the control parameter Lδ a was scaled by 0.5 x Lδ a , 0.75 x Lδ a , 1 x Lδ a , 1.25

x Lδ a , and 1.5 x Lδ a . Figure 54 depicts the side step maneuver from a perspective
view and a top-down view. First glance indicates there is satisfactory performance
for all scaled values, with the exception of slightly increasing lateral overshoot as Lδ a
is increased.
86
(b) Top-down View
Figure 54. Lδ a Scaled Side Step Maneuver
Figures 55a and 55b show that the workload continually increases a small amount
for rudder and aileron as Lδ a is scaled up. Figure 55c indicates that the TIC values for
both ROC and velocity stay relatively constant despite the changes to Lδ a . Overall,
the scaling to Lδ a negligibly changed aircraft performance and workload during the
87
side step landing maneuver.
Figure 55. Workload and Performance Metrics of Lδ a Scaled Side Step Maneuver
4.3.3 Lateral Controller Gain Scaling.
In addition to scaling the stability and control parameters, the gains on the lateral
controller were also scaled. Kc , the gain on the lead compensator tracking the side
step distance, was scaled by the values shown in Table 20. The values were chosen
to represent a system that started overdamped and steadily transitioned to being
underdamped. The same aircraft workload and performance parameters of L2 norm
and TIC were then calculated for each maneuver and an analysis was completed.
Figure 56 shows the resulting tracks of the Kc scaled side step maneuvers.
88
Table 20. Kc Scaling Factors for Side Step Maneuver
Run Number Kc Scaling Factor

1 0.5 x
2 1x
3 2x
4 3.5 x
5 4.5 x
89
(b) Top-down View
Figure 56. Kc Scaled Side Step Maneuver

Observing the top-down view in Figure 56b indicates that 0.5 x Kc is clearly
overdamped, while 2 x Kc and 3.5 x Kc are underdamped and 4.5 x Kc is on the
verge of instability. The graphs in Figures 57a and 57b show that workload is steadily
increasing for all control surfaces as Kc increases. TIC value of rate of climb also
steadily climbs, which is the second instance during the side step maneuver that a
change to a lateral parameter affected longitudinal performance.
(c) TIC Values
Figure 57. Workload and Performance Metrics of Kc Scaled Side Step
The flying qualities graphs in Figure 58 indicate a similar trend to the Kc scaling
results in the climbing turn. Since the bare airframe is unchanged, the MIL-STD-1797
graphs in Figures 58a and 58b are each five points stacked on each other in the Level
90
1 category. Figure 58c shows that as Kc increases, TIC also increases to the point
where the aircraft is considered Level 2 at 3.5 x Kc and then Level 3 at 4.5 x Kc .
A large TIC value accompanied by a ζsp value within Level 1 boundaries indicates
to an aircraft designer that the bare airframe design is adequate but changes to the
controller need to be made to improve the aircraft performance.
Figure 58. Flying Qualities Graphs of Kc Scaled Side Step
4.3.4 Cross-Coupling Effects.
As previously stated, cross-coupling occurred in both the Lp and Kc scaling during

the side step maneuver. The cross-coupling manifesting itself in the side step and not
the climbing turn indicates that the side step’s precision nature reveals otherwise
91
unknown flying qualities issues, information important to an aircraft designer. The
cross-coupling discovery reinforces the importance of a robust test plan when assessing
UAV flying qualities, including all maneuvers related to the operational mission of
the aircraft.
4.4 Aerial Refueling Analysis
The final maneuver assessed was the precision, non-aggressive aerial refueling task.
This task requires precision control to simultaneously track a specified altitude and
a specified bank angle while maintaining a constant velocity.
4.4.1 Baseline Aerial Refueling.
Before scaling any stability and control derivatives, a baseline analysis of the aerial
refueling maneuver was completed to achieve a datum for workload and performance
of the unscaled bare airframe with the controller in the loop. The same aircraft
nominals were used for the baseline maneuver and the subsequent scaled parameter
maneuvers: 5000 ft, 200 KTAS, with gear up and flaps up. Both the desired aircraft
roll angle and altitude were set to the same sum-of-sines function based on research
by Hamidani and discussed in Section 3.3.1.2. The sum-of-sines tracking portion
lasted 30 seconds, began 10 seconds into the simulation, and terminated with 10
seconds remaining. The dashed lines seen in Figure 59a show the tracking input to
the controllers and the red lines show the resulting aircraft track.
From inspecting the altitude tracking portion, there is steady-state error and over-
shoot evident from the first 10 seconds and last 10 seconds of the maneuver, respec-
tively. The first 10 seconds indicate a steady-state error of about 10 feet, and the last
10 seconds show about a 10 foot overshoot of the desired altitude. To mitigate this,
the performance and workload metrics were calculated only in the 30 second band of
92
(a) Altitude Tracking (b) Roll Angle Tracking
Figure 59. Baseline Aerial Refueling Tracking
tracking the sum-of-sines input. Since the same controller was used throughout the
scaling of all parameters, the changes in performance and workload can be attributed
to changes in the bare airframe dynamics, and not controller performance. In assess-
ing bank angle tracking, Figure 59b indicates excellent performance throughout the
maneuver, only showing a slight delay in control and minor overshoot after each step
input.
An assessment was also done of the workload and performance of the aircraft
during the baseline maneuver using L2 norm and TIC. Figures 60a and 60b show
a very active elevator actuator as compared to both the rudder and aileron. This
matches the data shown in Figures 59a and 59b, which shows the aircraft struggled to
track altitude as compared to roll angle. But the interesting point comes in Figure 60c,
which shows TIC for roll angle having the highest value by a large margin, with TIC
for altitude and velocity only a fraction of that amount.
The unusually low TIC values of altitude arise due to the large trim value of
altitude. The TIC calculation sees a 10 foot altitude tracking error as minuscule
when compared to the trim altitude of 5,000 feet. A small one or two degree bank
angle error is magnified due to the target angles being less than 25 degrees. Therefore,
93
the relativity in size between the trim value and target value directly correlates to
the magnitude of the TIC value. Although the magnitudes of TIC change, the trends
observed during stability and control parameter scaling remain preserved. In the case
of altitude for the aerial refueling maneuver, the subsequent performance plots and
cutoffs for Level 1, 2, and 3 flying qualities have to be adjusted accordingly.
(c) TIC Values
To assess how the stability and control parameters affected aircraft workload and
performance during the aerial refueling task, the four stability and control parameters
Mq , Mδ e , Lp , and Lδ a were scaled according to their values in Section 4.1.1.
94
4.4.2.1 Mq Scaling.
First, Mq was scaled by 0.25 x Mq , 1 x Mq , 2 x Mq , 4 x Mq , and 6 x Mq . Figure 61

shows the resulting five aircraft tracks for altitude and roll angle.
(b) Roll Angle Tracking

(a) Altitude Tracking
Figure 61. Mq Scaled Aerial Refueling Tracking
When assessing the initial results, its clear that scaling in Mq has only an effect
on altitude tracking, as the roll angles plotted in Figure 61b are nearly identical.
Looking at Figure 61a shows clear variations in performance as Mq is scaled up. 0.25
x Mq through 2 x Mq show similar tracking performance, while 4 x Mq and 6 x Mq
have discernibly more overshoot and oscillations, with 6 x Mq reaching limit cycle
oscillation throughout the maneuver. The graphs shown in Figure 62 confirm that
as Mq is scaled up, L2 norm of elevator position steadily increases, as does TIC of
altitude. This indicates that the aircraft is both doing more work and performing
worse. TIC of velocity continually decreases until 6 x Mq , at which point it spikes,
indicating that changes in Mq also have an effect on throttle control. Figure 62d
shows that TIC of roll angle is changing, but a trivial amount, as indicated by the
y-axis values.
95
(c) Altitude and Velocity TIC Values (d) Roll Angle TIC Values
Figure 62. Mq Scaled Aerial Refueling Tracking
The MIL-STD-1797 plots and newly developed criteria plots of Figure 63 agree
that unsatisfactory flying qualities exist at 6 x Mq . Its location on Figure 63c shows
that it falls outside both the ζsp Level 3 limit and the TIC Level 3 limit, indicating
to the aircraft designer the possibility that both the controller and the bare airframe
need to be modified. In addition, 0.25 x Mq , 1 x Mq , 2 x Mq , and 4 x Mq are closely
grouped within Level 1, but as stated earlier, the Level 2 and Level 3 boundaries are
flexible, as the values set for those boundaries are rough estimates based on aircraft
responses. With more research and subsequent data, the boundaries can be adjusted
to better match the aircraft’s response.
96
(c) TIC vs ζsp (d) L2 Norm vs TIC Altitude
Figure 63. Flying Qualities Graphs of Mq Scaled Aerial Refueling
The second scaled maneuver assessed during the aerial refueling task was Mδ e ,
the pitching moment due to elevator deflection. Mδ e was scaled by 0.25 x Mδ e , 0.75
x Mδ e , 1 x Mδ e , 1.25 x Mδ e , and 1.5 x Mδ e . Figure 64 depicts the resulting aerial
refueling plots.
97
Figure 64. Mδ e Scaled Aerial Refueling Tracking
The changes to Mδ e show similar results as the climbing spiral and side step ma-
neuver. A smaller value of Mδ e causes the aircraft to be sluggish in pitch, indicating
less control. A higher Mδ e increases the aircraft’s responsiveness in pitch and thus im-
proves both rise time and overshoot performance. Figure 65b shows a sharp decrease
in workload required to accomplish the task from 0.25 x Mδ e to 0.75 x Mδ e , but then
a slight increase from 0.75 x Mδ e to 1.5 x Mδ e . These trends coincide with a TIC
of altitude that steadily decreases with increasing Mδ e . From the graphs, an aircraft
designer could choose to minimize elevator workload or maximize performance based
on changes to Mδ e .
98
Figure 65. Mδ e Scaled Aerial Refueling Tracking
The MIL-STD-1797 graphs in Figures 66a and 66b primarily indicate that the
aircraft has Level 1 flying qualities throughout the scaling of Mδ e . In contrast, the
0.25 x Mδ e exceeds the Level 1 flying qualities limit, based on the TIC of altitude
limit set at 0.8 x 10−3 . As previously stated, these TIC limits are floating until more
data is available. But referencing Figure 64a indicates that the tracking performance
of 0.25 x Mδ e has larger delays and overshoots in comparison to the other five runs.
In this case, the developed criteria of flying qualities more accurately categorizes the
aircraft response than the classical methods used in MIL-STD-1797.
99
(c) TIC vs ζsp

(d) L2 Norm vs TIC Altitude
Figure 66. Flying Qualities Graphs of Mδ e Scaled Aerial Refueling
4.4.2.3 Lp Scaling.
Third, the lateral stability parameter Lp representing the roll moment due to roll
rate was scaled by 0.25 x Lp , 0.75 x Lp , 1 x Lp , 1.25 x Lp , and 1.50 x Lp . Figures 67a
and 67b show the resulting five aircraft courses.
100
(c) Initial Roll Angle Capture
Figure 67. Lp Scaled Aerial Refueling Tracking
First glance indicates the decoupled nature of the longitudinal and lateral stability
and control parameters during the aerial refueling maneuver, as altering Lp has nearly
no effect on the altitude tracking performance of the aircraft. Figure 67c is an zoomed-
in view at the first step input of the roll angle tracking. Roll tracking shows larger
overshoot as Lp decreases, along with a slower rise time, indicating a trade-off that a
higher value of Lp decreases overshoot and increases rise time.
101
Figure 68. Lp Scaled Aerial Refueling Tracking
Figure 68d shows that as Lp increases, roll tracking performance decreases. This
appears counterintuitive, as 0.25 x Lp clearly has the largest overshoot, indicating
poor performance in the transient response. But upon closer inspection of Figure 67b,
0.25 x Lp has the best steady-state tracking performance, which over the 30 second
interval equates to the lowest TIC of roll angle. When assessing L2 norm vs TIC of
roll angle, Figure 69 shows little change in TIC over the course of scaling Lp , with all
five runs tightly packed well within the Level 1 flying qualities boundaries, indicated
by the bold black lines on the top and right of the plot. Since changing Lp has
little effect on the workload and performance during the aerial refueling maneuver,
102
an aircraft designer could then use Lp as a free variable to meet other requirements.
Figure 69. L2 Norm vs TIC of Roll Angle
Lastly, the lateral control parameter Lδ a was scaled by 0.5 x Lδ a , 0.75 x Lδ a , 1 x

Lδ a , 1.25 x Lδ a , and 1.5 x Lδ a . Figure 70 depicts the track of the aircraft in altitude
and roll angle. Figures 70b and 70c indicate a lower overshoot and faster rise time
with increasing Lδ a . This matches with previous results during the climbing spiral
and side step maneuver, as changing the size of the control surface and maintaining
the same actuator performance will provide more control authority.
103
(c) Roll Angle Tracking
Figure 70. Lδ a Scaled Aerial Refueling Tracking
Figure 71c confirms that increasing Lδ a both decreases workload, as evidenced by

the decreasing L2 norm of the aileron position. Figure 71d shows that performance
increases as well, with TIC of roll angle starting above 0.13 and decreasing to below
0.09 over the course of scaling Lδ a . TIC of velocity and altitude remains largely
unchanged.
104
Figure 71. Lδ a Scaled Aerial Refueling Tracking
Figure 72 shows that all tested values of Lδ a are within the Level 1 boundaries.
A trend suggests that increasing Lδ a draws the aircraft further away from the Level
1 to Level 2 boundary, shown as the bold outlines on the top and right side of the
figure. The presented results would provide an aircraft designer with information on
how small an aileron could be made without compromising significant performance
during a precision, non-aggressive task.
105
Figure 72. L2 Norm vs TIC of Roll Angle
4.4.3 Deficiency Identification in UAV Flying Qualities.
Trends identified in Figure 48c indicated poor performance at 6 x Mq was due to

a large ζsp value, and thus a bare airframe issue. Trends from Figure 58c indicated
that gain increase correlated to poor performance, and thus a controller design issue.
Combining the results from these two graphs, an aircraft designer who just simulated
a maneuver could plot results on a graph of TIC vs ζsp and determine what is causing
the deficiencies of the aircraft. Figure 73 identifies to the aircraft designer if the
aircraft has a controller issue, a bare airframe issue, or both.
106
Figure 73. TIC vs ζsp Deficiency Overlay
Figure 73 has the same bold black lines plotted on it to show Level 1, Level 2,
and Level 3 flying qualities, with the addition of red, blue, and purple dotted lines
overlaid. The red dotted lines are areas where ζsp is either too large or too small, and
thus data points in those areas would indicate a bare airframe issue. The blue dotted
lines indicate areas of poor performance caused by the controller, and thus to improve
the performance a controller redesign is required. Finally, the purple lines indicate
areas where both the controller and bare airframe are causing issues, and thus both
may require redesign. A graph such as Figure 73 provides critical information to
an aircraft designer pertaining to the deficiencies of the aircraft and how to improve
them.
107
4.4.4 Adjusting TIC of Altitude for Flight Test.
In Section 4.4.1, it was explained that TIC of altitude had orders of magnitude
smaller values than TIC of bank angle. This was due to the baseline altitude being
a large number, and the deviation from that proportionally much smaller, therefore
indicating a nearly insubstantial error when using the TIC calculation. For the aerial
refueling analysis completed in this thesis, altitude was held to a constant 5,000 feet
for stability and control parameter scaling. As previously stated, holding altitude
constant preserves the trends present between the scaling factors.
The issue arises when changing the datum altitude for completing the same aerial
refueling task at different altitudes. To demonstrate, the same baseline profile was
completed at altitudes of 5,000 feet, 10,000 feet, and 15,000 feet, and then plotted as
a net altitude change on the same plot. Only the altitude at which the maneuver was
completed was changed between runs. The multi-sine input was held constant, along
with the aircraft velocity. Figure 74 shows the resulting aircraft altitude tracking
plots.
The plot indicates nearly identical performance at each altitude, with the excep-
tions that as altitude increases, there is a small increase in overshoot and a slight
decrease in steady-state error. Yet, TIC calculations indicate something different.
Table 21 shows that as altitude increases, there is also a significant increase in perfor-
mance. According to the traditional TIC calculation, the aircraft saw a 50% increase
in performance at 10,000 feet over the 5,000 foot run and a 67% increase in perfor-
mance at 15,000 feet over the 5,000 foot run. Clearly this is a mathematical effect and
not an indicator of performance, and therefore an adjustment to TIC was developed
for the altitude calculation.
108
Table 21. TIC Calculations for Aerial Refueling Altitude Comparison
Altitude (ft) TIC Value Percent of Performance Increase
5,000 0.000681 -
10,000 0.000338 50.3
15,000 0.000223 67.1
Adjusted TIC uses a datum altitude to scale down all results to that altitude. In
the case of this analysis, the 10,000 foot and 15,000 foot altitudes were scaled down
to 5,000 feet by subtracted 5,000 feet and 10,000 feet, respectively, from every data
point in the matrix. Now, TIC of altitude can be evaluated on the same scale and
subsequent performance issues can be identified without the confounded mathematical
effects of TIC. The simple change maintains the same magnitude of TIC across several
altitudes while preserving the correct trends of TIC changing across altitudes.
109
The aerial refueling task was then completed at the same three altitudes, this time
calculating adjusted TIC instead of the original TIC calculation. Table 22 shows the
results of using the normalized TIC calculation. These results indicate a much more
realistic increase in performance: a 0.77% increase at 10,000 feet over the 5,000 foot
run, and 1.54% increase at 15,000 feet over the 5,000 foot run.
Table 22. Adjusted Input TIC Calculations for Aerial Refueling Altitude Comparison
Altitude (ft) Adjusted Input TIC Value Percent of Performance Increase

5,000 0.000681 -
10,000 0.000676 0.77
15,000 0.000671 1.54
The Air Force Test Pilot School plans to do similar aircraft performance tests
to compare simulation results from this thesis to actual flight test data. The runs
conducted in the aerial refueling simulations hold altitude and velocity constant,
and therefore the large scaling issues seen with TIC between altitude would not be
present. But in actual flight test, a certain altitude may not be available to complete
the maneuver, and if a higher or lower altitude is chosen, TIC will incorrectly indicate
large increases or decreases in performance. Adjusted input TIC may be employed in
flight test in order to provide the flexibility of completing the aerial refueling task at
a large range of altitudes.
4.5 Chapter Summary
Results of the bare airframe analysis were presented to establish the baseline
aircraft parameters. Bode plots of the HOS and LOES were compared and from those
literal factors were derived. This process was repeated for both the longitudinal axis
and lateral-directional axis. Next, stability and control parameters were scaled to
modify the existing bare airframe. The resulting literal factors of each stability and
control scaled bare airframe were then presented. Then, the analysis of the climbing
110
spiral was completed for each parameter. Workload and performance comparisons
were made utilizing L2 norm and TIC as well as several flying qualities graphs. After,
the same previous steps were completed for the side step maneuver and the aerial
refueling maneuver. Results point to utilizing a combination of classical aircraft literal
factors, such as ζsp and CAP, with newly developed mathematical techniques, such
as L2 norm and TIC, to assess the workload of the flight controller and performance
during the maneuver.
111
V. Conclusions and Recommendations
5.1 Conclusions
The primary objective of this research was to provide data on flying qualities
assessments in an effort to expand the research of UAV flying qualities. The research
completed indicated that the Simulink model of the LJ-25D is an excellent platform
for simulating various precision and non-precision maneuvers as well as assessing
workload and performance metrics. In summary, the objectives of this research were
as follows:
1. Evaluate the flying qualities of an unmanned LJ-25D through various precision
and non-precision maneuvers.
2. Identify trends between stability and control parameter scaling, workload and
performance metrics, and classical control literal factors.
Each of the three maneuvers completed provided valuable data pertaining to the
trends of flying qualities. First, the non-precision, non-aggressive climbing turn iden-
tified the importance of retaining some of the original literal factors from MIL-STD-
1797 in the assessment of UAVs. During the scaling of Mq , TIC indicated that all
scaled values would provide Level 1 flying qualities, while MIL-STD-1797 predicted
that 6 x Mq would have Level 3 flying qualities due to the high ζsp . All scaled values
of Mq performed adequately during the climbing turn, but did not during another
maneuver: the aerial refueling task. The result points to the importance of ζsp across
all maneuvers, and that its inclusion in a future UAV flying qualities standard is vital.
The scaling Mδ e during the aerial refueling task showed positive results for using
TIC to predict flying qualities levels. 0.25 x Mδ e had large overshoots and poor steady-
state tracking performance during the maneuver, indicative of Level 2 or Level 3 flying
qualities. MIL-STD-1797 predicted 0.25 x Mδ e as Level 1 flyinq qualities, while TIC
112
indicated Level 2. This particular case shows the merit of using TIC in the assessment
of flying qualities over some classical control literal factors. The side step maneuver
also presented surprising results when assessing the effects of scaling Mδ e . The results
indicate that neither MIL-STD-1797 nor L2 norm and TIC correctly classify all the
scaled values. Combining aspects of the two into a graph of TIC vs. ζsp creates a
resultant plot that correctly categorizes each instance of Mδ e . This result further
solidifies the notion that new metrics and classical literal factors should both be used
in the development of a UAV flying qualities standard.
The identification of trends between the scaling of stability and control param-
eters and resultant workload and performance metrics provides an aircraft designer
with trade-offs during the design phase. Several important trends were discovered in
the assessment of the LJ-25D which if presented to the aircraft design team in the
early phases of the program would provide a valuable trade space between aircraft
performance and workload. One example is the scaling of Mδ e during the aerial refu-
eling task. Between 0.75 x Mδ e and 1.5 x Mδ e , the workload and performance metrics
during the task are nearly identical. But at 0.25 x Mδ e , the performance encounters
a large decrease to Level 2. That information indicates to a design team to keep the
limits of Mδ e between 0.75 x and 1.5 x its current value or risk significantly decreasing
performance during the aerial refueling task. This was also true for Mδ e during the
side step maneuver, which showed that scaling Mδ e below 0.75 x caused a significant
decrease in performance. The sharp drop off in performance indicates to the designer
the lower performance limit of Mδ e , also correlated to the size of the elevator. Sim-
ilar trends were present in the scaling of Lδ a during the aerial refueling task. As
Lδ a increased, the aileron effectively became larger while holding actuator dynamics
constant, which both increased performance and decreased workload. Additionally,
scaling some parameters indicated no change in workload and performance during
113
the maneuver. For example, scaling Lp during aerial refueling made only minuscule
changes to how the aircraft performed during task. That information relays to the
aircraft designer that Lp is a free variable, and can be modified within the scaled
boundaries without hindering aircraft performance.
The scaling of Kc during the climbing spiral maneuver and the side step maneuver
point out the dual nature of UAV flying handling qualities. Small changes to the
controller gain significantly changed the performance of the aircraft, but important
literal factors such as ζsp remained unchanged with these gain changes and indicated
the aircraft was well within limits of Level 1 flying qualities predicted by MIL-STD-
1797. The results indicate that an aircraft with satisfactory bare airframe dynamics
will perform unsatisfactorily when paired with poorly designed controller. The highly
augmented flight control systems of modern UAVs play a key role in UAV flying
qualities and should be accounted for in a UAV flying qualities standard.
The inclusion of the precision, non-aggressive side step maneuver also reinforced
the notion of a diverse flying qualities assessment. For example, cross-coupling was
present only during Lp scaling and Kc scaling of the side step maneuver, as scaling
Lp and Kc both significantly degraded the longitudinal rate of climb tracking perfor-
mance. This cross-coupling discovery highlights the need for a robust test plan that
includes all operational maneuvers in order to assess a UAVs flying qualities.
One point that was present in previous research by Kim [30] but not seen in the
assessment of the LJ-25D was position or rate saturation of actuators. The precision,
non-aggressive tasks showed varying trends in workload. Little can be stated on
the effects of position and rate saturation on performance since the demonstrated
maneuvers were non-aggressive in nature and never fully exercised the actuators to
their limits.
114
5.2 Future Research Recommendation
The Simulink LJ-25D model proved useful in providing a new understanding on

how scaling stability and control parameters affects aircraft workload and perfor-
mance. With that, there are still further discoveries to be made in pursuit of UAV
flying qualities using the LJ-25D model.
In this research, the primary focus was altering the open-loop bare airframe dy-
namics while maintaining the same flight controller. But the flight controller is an-
other piece of the puzzle which can be further investigated. Including the controller
in the open-loop analysis will provide more information on how a controller can over-
come poor bare airframe dynamics by assessing workload and performance in various
maneuvers. Studying how changes in controller design affect in workload and perfor-
mance could uncover new trends leading to better controller designs for the task at
hand.
Further research could be conducted in developing a better performance measure
based on TIC. As previously stated, TIC works well for comparing performance at
the same aircraft nominals. Yet when changing those nominals between runs, such
as altitude, TIC becomes scaled which skews the performance results. Developing
a TIC that mitigates the variances of aircraft nominals and provide a more robust
performance metric that could be used across any altitude, airspeed, or specified
nominal. Additionally, researchers could investigate the use of the capability machine
index as a way of measuring performance of the aircraft. Capability machine index
measures how well a machine performs in relation to given tolerance limits [39]. This
would allow the engineer to specify a tolerance and measure how well the aircraft
maintains that tolerance throughout the maneuver.
The last recommendation is to perform more maneuvers with the LJ-25D model.
Specifically, a precision, aggressive maneuver would provide more workload and per-
115
formance data to verify the trends identified in the precision, non-aggressive and
non-precision, non-aggressive maneuvers. It would also furnish more data points to
be utilized in the development of thumb print plots in setting Level 1, Level 2, and
Level 3 flying qualities limits.
A planned follow-on of this research is to perform the same maneuvers in the
CALSPAN LJ-25D at the USAF Test Pilot School, Edwards AFB, California. In an
effort to collect more data on the role pilots play in rating flying qualities, the author
recommend three overarching sets of test points. In the first, a pilot in the seat will
fly each maneuver with varying stability and control parameters and provide Cooper-
Harper ratings. This will provide a baseline level of Cooper-Harper ratings at each
scaled stability and control parameter. Next, a pilot in the remote facility will fly
each scaled stability and control parameter and attempt to rate each maneuver. Last,
a controller either adapted from this research, developed by CALSPAN, or developed
by the USAF test pilot school will fly each maneuver at each scaled stability and
control parameter.
Each pilot will fly a maneuver at a chosen scaled stability and control parameter. It
is important to randomize both the stability and control derivative and the magnitude
of the scaling. As such, all 20 runs for each maneuver in which a pilot is flying should
be randomly ordered, thus removing any indications to the pilots of trends in an
increase or decrease in performance over the course of the test. During the test,
pilots will provide Cooper-Harper ratings to obtain information on their perceived
workload during each maneuver. The L2 norm of actuator deflections and rates will
also be recorded. TIC will be calculated to determine the pilot performance during
each maneuver. Adjusted input TIC, as described in Section 4.4.4, may also be
utilized for calculating TIC during the aerial refueling task. The desired value input
to the pilot will be x̃ and the actual aircraft track will be x. With this large set of
116
data, comparisons can be made between the pilot-in-the-seat, the remote pilot, the
controller in the aircraft, and the simulation results from this thesis. The Level 1,
Level 2, and Level 3 limits on the developed flying qualities graphs of TIC vs ζsp
and L2 norm vs TIC can be compared and adjusted based on pilots’ Cooper-Harper
ratings of the aircraft during that maneuver. These resulting graphs from flight test
can be compared to the simulation graphs and aid in developing new thumb print
plots which more accurately predict UAV flying qualities.
5.3 Summary
In conclusion, the LJ-25D Simulink model has provided invaluable information on

identifying trends between scaled aircraft stability and control parameters and air-
craft workload and performance. The three maneuvers conducted each offer another
data point in the growing database of UAV flying qualities. Neither the MIL-STD-
1797 standards or developed L2 norm and TIC metrics alone accurately predict UAV
flying qualities at every data point. Instead, a combination of the two utilized in
a thumb print plot more accurately predict UAV performance over precision, non-
aggressive and non-precision, non-aggressive maneuvers. By using the LJ-25D model,
comparisons from simulations can be made to flight test and bridge the gap between
theoretical flying qualities performance metrics and measured flight test performance
metrics.
117
Appendix A. LOES Bode Plots
This appendix contains data from the LOES development of the LJ-25D. The Bode
plot overlays of LOES and HOS are a graphical depiction verifying LOES fidelity. The
boundaries described by MIL-STD-1797, displayed in black, mathematically confirm
a LOES as acceptable at all frequencies.
(a) Mq x 0.25 Bode Plot (b) Mq x 0.25 Envelope Plot
(c) Mq x 1 Bode Plot (d) Mq x 1 Envelope Plot
118
(e) Mq x 2 Bode Plot (f ) Mq x 2 Envelope Plot
(g) Mq x 4 Bode Plot (h) Mq x 4 Envelope Plot
(i) Mq x 6 Bode Plot (j) Mq x 6 Envelope Plot
Figure 75. Mq Scaling Bode Plots
119
(a) Melev x 0.25 Bode Plot (b) Melev x 0.25 Envelope Plot
(c) Melev x 0.75 Bode Plot (d) Melev x 0.75 Envelope Plot
(e) Melev x 1 Bode Plot (f ) Melev x 1 Envelope Plot
120
(g) Melev x 1.25 Bode Plot (h) Melev x 1.25 Envelope Plot
(i) Melev x 1.5 Bode Plot (j) Melev x 1.5 Envelope Plot
Figure 76. Melev Scaling Bode Plots
121
(a) Lp x 0.25 Bode Plot (b) Lp x 0.25 Envelope Plot
(c) Lp x 0.75 Bode Plot (d) Lp x 0.75 Envelope Plot
122
(e) Lp x 1 Bode Plot (f ) Lp x 1 Envelope Plot
(g) Lp x 1.25 Bode Plot (h) Lp x 1.25 Envelope Plot
(i) Lp x 1.5 Bode Plot (j) Lp x 1.5 Envelope Plot
Figure 77. Lp Scaling Bode Plots
123
(a) Lail x 0.5 Bode Plot (b) Lail x 0.5 Envelope Plot
(c) Lail x 0.75 Bode Plot (d) Lail x 0.75 Envelope Plot
124
(e) Lail x 1 Bode Plot (f ) Lail x 1 Envelope Plot
(g) Lail x 1.25 Bode Plot (h) Lail x 1.25 Envelope Plot
(i) Lail x 1.5 Bode Plot (j) Lail x 1.5 Envelope Plot
Figure 78. Lail Scaling Bode Plots
125
Bibliography
1. “Stinson Reliant”. Wikipedia, 1935. URL https://en.wikipedia.org/wiki/

Stinson_Reliant.
2. Flying Qualities of Piloted V/STOL Aircraft. Technical Report MIL-F-83300,

Department of Defense, 1970.
3. Flying Qualities of Piloted Airplanes. Technical Report MIL-F-8785C, Depart-

ment of Defense, 1980.
4. USAF Test Pilot School Flying Qualities Textbook. Edwards AFB, California,
1986.
5. Flying Qualities of Piloted Aircraft. Interface Standard MIL-STD-1797A, NO-

TICE 1, Department of Defense, 1995.
6. “Lockheed NT-33A”. National Museum of the Air Force Fact Sheet,

2016. URL https://www.nationalmuseum.af.mil/Visit/Museum-Exhibits/
Fact-Sheets/Display/Article/195773/lockheed-nt-33a/.
7. Allen, James. Advertisement and Specification for a Heavier-than-Air Flying

Machine. Washington D.C., 1907. URL https://www.libraries.wright.edu/
special/wrightbrothers/packet/signal.
8. Berger, Tom, Mark Tischler, Steven G. Hagerott, M Christopher Cotting,

William R. Gray, James Gresham, Justin George, Kyle Krogh, Alessandro
D’Argenio, and Justin Howland. “Development and Validation of a Flight-
Identified Full-Envelope Business Jet Simulation Model Using a Stitching Ar-
chitecture”. AIAA Modeling and Simulation Technologies Conference, 2017.
126
9. Bihrle, William. A Handling Qualitites Theory for Precise Flight Path Control.
Air Force Flight Dynamics Laboratory, Wright Patterson AFB, Ohio, 1966.
10. Bilstein, Roger E. Flight Patterns: Trends of Aeronautical Development In the

United States, 1918-1929. University of Georgia Press, Athens, GA, 2008.
11. Bischoff, D. E. The Control Anticipation Parameter for Augmented Aircraft.

Technical Report NADC-81186-60, Naval Air Development Center, Warminster,
PA, 1981.
12. Bryan, G. H. Stability in Aviation. MacMillan, London, 1911.
13. Caughey, David A. “Introduction to Aircraft Stability and Control”. Sibley

School of Mechanical & Aerospace Engineering, Cornell University, 2011. URL
https://courses.cit.cornell.edu/mae5070/Caughey_2011_04.pdf.
14. Chalk, C. R. Study for a Fighter Total In-Flight Simulator. Calspan Flight
Research Memorandum 566, 1982.
15. Clapper, J. R., J. J. Young, J. E. Grimes, S.C. Payton, S. Stackley, and D. Popps.
“FY 2009-2034 Unmanned Systems Integrated Roadmap”. Department of De-
fense: Office of the Secretary of Defense Unmanned Systems Roadmap, 92–93,
2009.
16. Cooper, G. E. and R. P. Harper. “The Use of Pilot Ratings in the Evaluation
of Aircraft Handling Qualities”. Advisory Group for Aerospace Research and
Development, 1–56, 1969.
17. Cotting, M Christopher. Evolution of Flying Qualities Analysis: Problems for

a New Generation of Aircraft. Dissertation, Virginia Polytechnic Institute and
State University, 2010.
127
18. Cotting, M Christopher. “UAV Performance Rating Scale Based on the Cooper-
Harper Piloted Rating Scale”. 49th AIAA Aerospace Sciences Meeting including
the New Horizons Forum and Aerospace Exposition, 2011.
19. Daniel, Larry, Barry Baskett, Samuek Crews, and William Smith. Aeronautical
Design Standard Performance Specification Handling Qualities Requirements for
Military Rotorcraft. Technical Report ADS-33E, United States Army Aviation
and Missile Command, Redstone Aresenal, Alabama, 2000.
20. Dorobantu, Andrei, Peter Seiler, and Gary Balas. “Validating Uncertain Aircraft
Simulation Models using Flight Test Data”. AIAA Atmospheric Fligth Mechanics
Conference, 2013.
21. Ehrhard, Thomas P. “Air Force UAVs: The Secret History”. The Mitchell
Institute for Airpower Studies, 89, 2010.
22. Federal Aviation Administration. Airworthiness Standards: Transport Category

Airplanes, volume 14 CFR § 2. FAA, 2018.
23. Greene, Kara M. Toward a Flying Qualities Standard for Unmanned Aircraft.
Dissertation, Air Force Institude of Technology, 2017.
24. Greene, Kara M and Donald L Kunz. “Quickness Criteria for Large Unmanned
Aircraft in Non-Precision , Aggressive and Non-Aggressive Maneuvers”. AIAA
Atmospheric Flight Mechanics Conference, 2015.
25. Greene, Kara M., Donald L. Kunz, and M Christopher Cotting. “Toward a
Flying Qualities Standard for Unmanned Aircraft”. AIAA Atmospheric Flight
Mechanics Conference, 2014.
26. Hamidani, Ali M. Evaluating the Autonomous Flying Qualities of a Simulated

Variable Stability Aircraft. Thesis, Air Force Institute of Technology, 2017.
128
27. Hodgkinson, John. Aircraft Handling Qualities. Blackwell Science, Oxford, 1999.
28. Johansen, I. Autopilot Design for Unmanned Aerial Vehicles. Thesis, Norwegian
University of Science and Technology, Norway, 2012.
29. Key, David L. Generation of a Military Specification for Flying Qualities of

Piloted V/STOL Aircraft. Technical Report AFFDL-TR-71-23, Cornell Aero-
nautical Laboratory, 1971.
30. Kim, Joshua P. Evaluation of Unmanned Aircraft Flying Qualities Using JSBSim.
Thesis, Air Force Institute of Technology, 2016.
31. Klyde, D. and D. Mitchell. Handling Quality Demonstration Maneuvers for Fixed-
Wing Aircraft. Technical Report WL-TR-97-3100, Systems Technology, Inc., Oct
1997.
32. Leggett, David B. “Lecture Notes in Flying Qualities”, July 2018.
33. Leone, Dario. “The Fighter Jet that can Fly as a Cargo Plane: the F-16
VISTA”. The Aviationist, 2013. URL https://theaviationist.com/2013/
09/10/vista-f-16/.
34. Mathworks. “Simulink Overview”, 2018. URL https://www.mathworks.com/

products/simulink.html.
35. McFarland, M. W. The Papers of Wilbur and Orville Wright. McGraw-Hill, New
York, 1st edition, 1953.
36. Miller, T. “F-16 Vista”. Aviation Photo Digest, 2015. URL http://
aviationphotodigest.com/history-is-made-at-edwards-afb.
37. Mitchell, David G., David B. Doman, David L. Key, David H. Klyde, David B.
Leggett, David J. Moorhouse, David H. Mason, David L. Raney, and David K.
129
Schmidt. “Evolution, Revolution, and Challenges of Handling Qualities”. Journal
of Guidance, Control, and Dynamics, 27:12–28, 2004.
38. National Research Council. Review of NASA’s Aerospace Technology Enter-

prise: An Assessment of NASA’s Aeronautics Technology Programs. National
Academies Press, 2004.
39. Nielson, Michael. “Statistical Process Control”. URL http://www.

statisticalprocesscontrol.info/glossary.html.
40. Philips, W. Hewitt. Journey in Aeronautical Research: A Career at NASA Lan-

gley Research Center. NASA History Division of Policy and Plans and Office
of Space Science, Washington, DC, 1998. URL https://history.nasa.gov/
monograph12/ch4.htm.
41. Prosser, C.F. and C. D. Wiler. RPV Flying Qualities Design Criteria. Technical
Report AFFDL-TR-76-125, Rockwell International Corporation, 1976.
42. Reisinger, Don. “Here’s How Many Lives Drones Have Saved Since 2013”. For-
tune, 2017. URL http://fortune.com/2017/03/14/drones-save-lives/.
43. Sarno, K.J. and C.D. Wickens. “The role of multiple resources in predicting
timesharing efficiency”. The International Journal of Aviation Psychology, 107–
130, 1995.
44. Soulé, H. A. Preliminary Investigation of the Flying Qualities of Airplanes. Tech-

nical Report NACA TR-700, National Advisory Committee for Aeronautics, Lan-
gley, VA, 1940.
45. Stevens, Brian L and Frank L Lewis. Aircraft Control and Simulation. John
Wiley & Sons, Inc, Hoboken, New Jersey, 2nd edition, 2003.
130
46. Suckow, Elizabeth. “NACA Overview”. NASA History Program Office, 2009.
URL https://history.nasa.gov/naca/overview.html.
47. Theil, H. Economic Forecasts and Policy. North-Holland Publishing Company,

1970.
48. Trueman, C. N. “Aircraft and World War One”. The History Learning
Site, 2015. URL https://www.historylearningsite.co.uk/world-war-one/
aerial-warfare-and-world-war-one/aircraft-and-world-war-one/.
49. Weiner, Robert and Tom Sherman. “Drones spare troops, have powerful impact”.
San Diego Tribune, 2014. URL http://www.sandiegouniontribune.com/
opinion/commentary/sdut-drones-troops-impact-2014oct09-story.html.
50. Williams, Warren and Michael Harris. “The Challenges of Flight-Testing Un-
manned Air Vehicles”. Systems Engineering, Test & Evaluation Conference, 2002.
51. Winnefeld, J.a. and F. Kendall. “Unmanned Systems integrated roadmap: FY

2011-2036”. Department of Defense, 1–74, 2010.
52. Yechout, Thomas R., Steven L. Morris, David E. Bossert, and Wayne F. Hallgren.
Introduction to Aircraft Flight Mechanics. American Institute of Aeronautics and
Astronautics, Reston, VA, 2003.
131
Form Approved
REPORT DOCUMENTATION PAGE OMB No. 0704–0188
The public reporting burden for this collection of information is estimated to average 1 hour per response, including the time for reviewing instructions, searching existing data sources, gathering and
maintaining the data needed, and completing and reviewing the collection of information. Send comments regarding this burden estimate or any other aspect of this collection of information, including
suggestions for reducing this burden to Department of Defense, Washington Headquarters Services, Directorate for Information Operations and Reports (0704–0188), 1215 Jefferson Davis Highway,
Suite 1204, Arlington, VA 22202–4302. Respondents should be aware that notwithstanding any other provision of law, no person shall be subject to any penalty for failing to comply with a collection
of information if it does not display a currently valid OMB control number. PLEASE DO NOT RETURN YOUR FORM TO THE ABOVE ADDRESS.
1. REPORT DATE (DD–MM–YYYY) 2. REPORT TYPE 3. DATES COVERED (From — To)

22–03–2019 Master’s Thesis Sept 2017 — Mar 2019
4. TITLE AND SUBTITLE 5a. CONTRACT NUMBER
5b. GRANT NUMBER

Evaluation of Unmanned Aircraft Flying Qualities
Using A Stitched Learjet Model
5c. PROGRAM ELEMENT NUMBER
6. AUTHOR(S) 5d. PROJECT NUMBER
5e. TASK NUMBER

Callaghan, Patrick M., Capt, USAF
5f. WORK UNIT NUMBER
7. PERFORMING ORGANIZATION NAME(S) AND ADDRESS(ES) 8. PERFORMING ORGANIZATION REPORT

NUMBER
Graduate School of Engineering and Management (AFIT/EN)
2950 Hobson Way AFIT-ENY-MS-19-M-206
WPAFB OH 45433-7765
9. SPONSORING / MONITORING AGENCY NAME(S) AND ADDRESS(ES) 10. SPONSOR/MONITOR’S ACRONYM(S)
Department of Aeronautics and Astronautics AFRL/RQ

2950 Hobson Way
WPAFB OH 45433-7765 11. SPONSOR/MONITOR’S REPORT
DSN 271-0690, COMM 937-255-3636 NUMBER(S)
Email: patrick.callaghan@afit.edu
12. DISTRIBUTION / AVAILABILITY STATEMENT
DISTRIBUTION STATEMENT A:
13. SUPPLEMENTARY NOTES
This material is declared a work of the U.S. Government and is not subject to copyright protection in the United States
14. ABSTRACT MIL-STD-1797 lists flying qualities for UAVs as TBD, and the standards for manned fixed wing are
inadequate when applied to UAVs. In an effort to expand the database of UAV flying qualities, an analysis was
completed on a Simulink model of an LJ-25D developed from Calspans Variable Stability System aircraft at the United
States Test Pilot School. Three maneuvers were simulated: (1) a non-precision, non-aggressive climbing spiral, (2) a
precision, non-aggressive side step landing, and (3) a precision, non-aggressive aerial refueling task. These maneuvers
were chosen to evaluate the performance and workload of the aircraft as four stability and control parameters were
scaled. The data were utilized in identifying trends between the scaled stability and control parameters and resulting
workload and performance metrics. Thumbprint plots were generated to identify Level 1, Level 2, and Level 3 flying
qualities and compared to MIL-STD-1797 plots. Results point to utilizing a combination of classical aircraft literal
factors, such as ζsp and CAP, with newly developed mathematical techniques, such as L2 norm and TIC, to assess the
workload of the flight controller and performance during the maneuver.
15. SUBJECT TERMS
Flying Qualities, UAVs, Flight Dynamics Model, Maneuver Performance, Flight Control System, Workload, Theil’s
Inequality Coefficient
16. SECURITY CLASSIFICATION OF: 17. LIMITATION OF 18. NUMBER 19a. NAME OF RESPONSIBLE PERSON
ABSTRACT OF Dr. D. Kunz, AFIT/ENY
a. REPORT b. ABSTRACT c. THIS PAGE PAGES
19b. TELEPHONE NUMBER (include area code)
U U U UU 149 (937) 255-3636, x4547; donald.kunz@afit.edu
Standard Form 298 (Rev. 8–98)
Prescribed by ANSI Std. Z39.18

Evaluation of Unmanned Aircraft Flying Qualities Using A Stitched Learjet Model

Uploaded by

Copyright:

Available Formats

Evaluation of Unmanned Aircraft Flying Qualities Using A Stitched Learjet Model

Uploaded by

Document Information

Original Title

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

Evaluation of Unmanned Aircraft Flying Qualities Using A Stitched Learjet Model

Uploaded by

Copyright:

Available Formats

Air Force Institute of Technology

Evaluation of Unmanned Aircraft Flying Qualities

Follow this and additional works at: https://scholar.afit.edu/etd

Patrick M. Callaghan, Captain, USAF

DEPARTMENT OF THE AIR FORCE

AIR FORCE INSTITUTE OF TECHNOLOGY

EVALUATION OF UNMANNED AIRCRAFT FLYING QUALITIES USING A

Presented to the Faculty

Patrick M. Callaghan, B.S.A.E.

February 26, 2019

EVALUATION OF UNMANNED AIRCRAFT FLYING QUALITIES USING A

Patrick M. Callaghan, B.S.A.E.

2.1 Early Development of Flying Qualities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

III. Research Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 34

1 Flight Hours for Unmanned Systems between 1996 and

2 U.S. Military Aircraft and UAS Class A Mishap Rates,

3 Learjet LJ25 Variable Stability System [8] . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

4 The Stinson Reliant played an important role in

5 The NT-33A aided in the development of the X-15,

6 The F-16 VISTA [36] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13

7 Cooper-Harper Pilot Rating Scale [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

8 ωsp Tθ2 vs ζsp plot depicting how Cooper-Harper ratings

9 Short-period Dynamic Requirements [5] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

10 Short-period Pitch Response to Pitch Controller [5] . . . . . . . . . . . . . . . . . 20

11 Time Domain Transient Response Parameters [5] . . . . . . . . . . . . . . . . . . . 20

12 Aircraft Classification in MIL-STD-1797B [5] . . . . . . . . . . . . . . . . . . . . . . . 23

13 Modified Neal-Smith Design Criteria for Pitch

14 Williams’s UAV Class System [50] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

15 UAV Grouping by Reynolds Number and Aircraft

16 UAV Grouping by Operational Altitude, Weight, and

17 Cotting’s Modified Cooper-Harper Rating Scale for

18 Notional Relation between UAV Roles, Mission Task

19 LJ-25D Model stitching block diagram [8] . . . . . . . . . . . . . . . . . . . . . . . . . . 35

20 Envelopes of Maximum Unnoticeable Added Dynamics

21 Block Diagram of Spiral Maneuver Longitudinal

22 Block Diagram of a PID Controller [28] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

23 Block Diagram of Spiral Maneuver Lateral Controller . . . . . . . . . . . . . . . . 43

24 Block Diagram of Spiral Maneuver Sideslip Controller . . . . . . . . . . . . . . . 43

25 Block Diagram of Spiral Maneuver Throttle Controller . . . . . . . . . . . . . . . 44

26 Block Diagram of Side Step Maneuver Lateral Controller . . . . . . . . . . . . . 45

27 Block Diagram of Side Step Maneuver Heading

28 Longitudinal LOES for Baseline Bare Airframe, 200

29 Lateral LOES for Baseline Bare Airframe, 200 KCAS

30 Baseline Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58

31 Workload and Performance Metrics of Baseline

32 Mq Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61

33 Workload and Performance Metrics of Mq Scaled

34 Flying Qualities Graphs of Mq Scaled Climbing Spiral . . . . . . . . . . . . . . . 63

35 Mδ e Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

36 Workload and Performance Metrics of Mδ e Scaled

37 Flying Qualities Graphs of Mδ e Scaled Climbing Spiral . . . . . . . . . . . . . . 67

38 Lp Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68

39 Workload and Performance Metrics of Lp Scaled

40 Lδ a Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70

41 Workload and Performance Metrics of Lδ a Scaled

42 Kc Scaled Climbing Spiral Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71

43 Workload and Performance Metrics of Kc Scaled

44 Flying Qualities Graphs of Kc Scaled Climbing Spiral . . . . . . . . . . . . . . . 74

45 Baseline Side Step Maneuver . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76