Mader Charles A 201206 PHD Thesis

Stability-Constrained Aerodynamic Shape Optimization with
Applications to Flying Wings
by
Charles Alexander Mader
A thesis submitted in conformity with the requirements

for the degree of Doctor of Philosophy
Graduate Department of Institute for Aerospace Studies
University of Toronto
c 2012 by Charles Alexander Mader

Copyright
Abstract
Stability-Constrained Aerodynamic Shape Optimization with Applications to Flying Wings
Doctor of Philosophy
Graduate Department of Institute for Aerospace Studies
University of Toronto
2012
A set of techniques is developed that allows the incorporation of flight dynamics metrics
as an additional discipline in a high-fidelity aerodynamic optimization. Specifically, techniques
for including static stability constraints and handling qualities constraints in a high-fidelity
aerodynamic optimization are demonstrated. These constraints are developed from stability
derivative information calculated using high-fidelity computational fluid dynamics (CFD). Two
techniques are explored for computing the stability derivatives from CFD. One technique uses
an automatic differentiation adjoint technique (ADjoint) to efficiently and accurately compute a
full set of static and dynamic stability derivatives from a single steady solution. The other technique uses a linear regression method to compute the stability derivatives from a quasi-unsteady
time-spectral CFD solution, allowing for the computation of static, dynamic and transient stability derivatives. Based on the characteristics of the two methods, the time-spectral technique
is selected for further development, incorporated into an optimization framework, and used to
conduct stability-constrained aerodynamic optimization. This stability-constrained optimization framework is then used to conduct an optimization study of a flying wing configuration.
This study shows that stability constraints have a significant impact on the optimal design of
flying wings and that, while static stability constraints can often be satisfied by modifying the
airfoil profiles of the wing, dynamic stability constraints can require a significant change in the
planform of the aircraft in order for the constraints to be satisfied.
ii
Dedication
To my parents:
For nurturing my curiosity and giving me the tools to succeed
iii
Acknowledgements
The end product of a doctoral program is a thesis with a single name on it. However, the
creation of that thesis would not be possible without the support of many, many people. I
would like to take this opportunity to thank the various people who have supported me in this
endeavor.
First and foremost I would like to thank my supervisor, Professor Joaquim Martins. When
this project first started, it was, in many ways, out in left field. Professor Martins allowed me
the freedom to pursue the research where it led. While perhaps costly in the short term, I think
that in the end this proved very fruitful. Further, Professor Martins enthusiasm for research in
general and optimization research in particular provided a constant impetus pushing this work
along. Finally, Professor Martins advice and guidance has been invaluable over the course of
my degree. Thank you.
I would also like to thank Professor Zingg and Professor Damaren, my other committee
members, for their time and effort. Their insightful questions pushed me to better my understanding of various subjects and to fill in holes in certain areas of my research. This has led
to significantly better final product. I would also like to thank Professor Kyle Anderson, my
external examiner, for taking the time to review my thesis and provide constructive feedback
on my work.
I would like to express my gratitude to Dr. Ruben Perez, whose insights into aircraft design
and in particular, stability and control have been instrumental in my work.
Finally, I would like to thank all of my colleagues in the MDO lab. We have succeeded
in creating an excellent collaborative atmosphere that has been a pleasure to work in. The
impromptu brainstorming sessions and theoretical discussions are invigorating and the positive
energy in the lab helps keep things going on those occasional days when things dont go exactly
as planned. Thanks Everyone!

University of Toronto Institute for Aerospace Studies
February 2012
iv
Contents
1 Introduction
1.1
The Case for Unconventional Configurations . . . . . . . . . . . . . . . . . . . . .
1.2
The Need for Stability Constraints in High-Fidelity Optimization . . . . . . . . .
1.3
Overview of the Stability Constraint Approach . . . . . . . . . . . . . . . . . . .
1.4
Thesis Layout . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
1.5
Original Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
2 Linear Flight Dynamics
2.1
Decoupled Longitudinal Equations . . . . . . . . . . . . . . . . . . . . . . . . . .
2.2
Short-Period Approximation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
2.3
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11
3 CFD-Based Stability Derivatives
12
3.1
Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12
3.2
Sensitivity Analysis Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
3.3
ADjoint Stability Derivative Method . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.4
3.3.1
Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16
3.3.2
ADjoint Approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23
3.3.3
Verification of the ADjoint for Stability Derivatives . . . . . . . . . . . . . 25
Time-Spectral Stability Derivative Method
. . . . . . . . . . . . . . . . . . . . . 26
3.4.1
Time-spectral CFD . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27
3.4.2
Linearized Aerodynamic Forces . . . . . . . . . . . . . . . . . . . . . . . . 31
3.4.3
Verification and Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 36
4 Time-Spectral ADjoint Method
46
4.1
Background . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
4.2
Theory . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48
4.3
Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
v
4.4
4.5
4.3.1
Single-Cell Routine . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.3.2
R/ and R/x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 51
4.3.3
I/ and I/x . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.3.4
Solution of the Adjoint System . . . . . . . . . . . . . . . . . . . . . . . . 56
Verification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4.1
Test case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4.2
Accuracy . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
4.4.3
Performance
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5 Auxilliary Analyses
64
5.1
Center of Gravity Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64
5.2
Moment of Inertia Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
5.3
Root Bending Moment Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . 68
5.4
Stability Constraints . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 70
5.5
5.4.1
Static Longitudinal Stability Definition . . . . . . . . . . . . . . . . . . . 70
5.4.2
Static Margin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71
5.4.3
Control Anticipation Parameter . . . . . . . . . . . . . . . . . . . . . . . . 72
Geometry Handling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 73
6 Design Optimization Results
76
6.1
Multidisciplinary Optimization Overview
6.2
Optimization Study Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.3
6.4
6.5
. . . . . . . . . . . . . . . . . . . . . . 77
6.2.1
Design Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80
6.2.2
Additional Constraints for Shape Variables . . . . . . . . . . . . . . . . . 84
Optimization Problem Statements and Qualitative Results . . . . . . . . . . . . . 85

6.3.1
Reference Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 85
6.3.2
Static Stability Constrained Problems . . . . . . . . . . . . . . . . . . . . 95
6.3.3
Dynamic Stability Constrained Problems . . . . . . . . . . . . . . . . . . 104
Results Tables
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.4.1
Mach = 0.5 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
6.4.2
Mach = 0.7 Cases
6.4.3
Mach = 0.85 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112
Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 114
7 Conclusions
116
vi
8 Future Work
118
References
120
Appendix
129
A Linear Flight Dynamics
130
B Results Data Tables
134
C Relevant Optimization Parameters
139
D FFD Coordinate Tables
145
vii
List of Tables
3.1
Comparison of lift and moment coefficients for the NACA 0012 at = 0 for
various values of q (results from Limache [55] in parentheses) . . . . . . . . . . . 22
3.2
Sensitivity verification: NACA 0012 test case, Mach = 0.5, 131,072 cells: 1012
relative convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25
3.3
ADjoint stability derivatives for a NACA 0012 airfoil at = 0.0 degrees . . . . . 25
3.4
Effect of point location on time-spectral derivative location . . . . . . . . . . . . 36
3.5
SACCON test case conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41
4.1
Sensitivity verification: ONERA M6, pitching motion, 917,000 cells: 1012 relative convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.2
Sensitivity verification: ONERA M6, plunging motion, 917,000 cells: 1012 relative convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60
4.3
Time-spectral ADjoint computational cost breakdown for ONERA M6 (normalized with respect to a total flow solution cost of 160.3 seconds) . . . . . . . . . . 61
5.1
Stability bounds for damping ratio and CAP parameter . . . . . . . . . . . . . . 73
6.1
Baseline wing: geometry specifications . . . . . . . . . . . . . . . . . . . . . . . . 77
6.2
Primary design variables and their bounds . . . . . . . . . . . . . . . . . . . . . . 81
6.3
Compatibility design variables and their bounds . . . . . . . . . . . . . . . . . . . 81
6.4
NACA 0012 wing: planform only optimization results: 1107k cells, M = 0.5 . . . 111
6.5
NACA 0012 wing: shape optimization results: 1107k cells, M = 0.5 . . . . . . . . 112
6.6
NACA 0012 wing: planform only optimization results: 1107k cells, M = 0.7 . . . 113
6.7
NACA 0012 wing: shape optimization results: 1107k cells, M = 0.7 . . . . . . . . 114
6.8
NACA 0012 wing: shape optimization results: 1107k cells, M = 0.85 . . . . . . . 114
B.1 NACA 0012 wing: planform only optimization results summary: 1107k cells,
M = 0.5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134
B.2 NACA 0012 wing: shape optimization results summary: 1107k cells, M = 0.5 . . 135
viii
B.3 NACA 0012 wing: planform only optimization results summary: 1107k cells,
M = 0.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 136
B.4 NACA 0012 wing: shape optimization results summary: 1107k cells, M = 0.7 . . 137
B.5 NACA 0012 wing: shape optimization results summary: 1107k cells, M = 0.85 . 138
C.1 Relevant parameters: M=0.5 cases, planform only . . . . . . . . . . . . . . . . . 140
C.2 Relevant parameters: M=0.5 cases, with shape variables . . . . . . . . . . . . . . 141
C.3 Relevant parameters: M=0.7 cases, planform only . . . . . . . . . . . . . . . . . 142
C.4 Relevant parameters: M=0.7 cases, with shape variables . . . . . . . . . . . . . . 143
C.5 Relevant parameters: M=0.85 cases, with shape variables . . . . . . . . . . . . . 144
D.1 FFD coordinates for baseline case, M = 0.5 . . . . . . . . . . . . . . . . . . . . . 146
D.2 FFD coordinates for twist-only case, M = 0.5 . . . . . . . . . . . . . . . . . . . . 147
D.3 FFD coordinates for bending moment constrained case, M = 0.5 . . . . . . . . . 148
D.4 FFD coordinates for Cm constrained case, M = 0.5 . . . . . . . . . . . . . . . . 149
D.5 FFD coordinates for Kn constrained case, M = 0.5 . . . . . . . . . . . . . . . . . 150
D.6 FFD coordinates for CAP constrained case, M = 0.5 . . . . . . . . . . . . . . . . 151
D.7 FFD coordinates for baseline case with shape variables, M = 0.5, part 1 . . . . . 152
D.8 FFD coordinates for baseline case with shape variables, M = 0.5, part 2 . . . . . 153
D.9 FFD coordinates for Cm constrained case with shape variables, M = 0.5, part 1 154
D.11 FFD coordinates for Kn constrained case with shape variables, M = 0.5, part 1 . 156
D.13 FFD coordinates for CAP constrained case with shape variables, M = 0.5, part 1 158
D.15 FFD coordinates for bending moment constrained case, M = 0.7 . . . . . . . . . 160
D.16 FFD coordinates for Cm constrained case, M = 0.7 . . . . . . . . . . . . . . . . 161
D.17 FFD coordinates for Kn constrained case, M = 0.7 . . . . . . . . . . . . . . . . . 162
D.18 FFD coordinates for CAP constrained case, M = 0.7 . . . . . . . . . . . . . . . . 163
D.19 FFD coordinates for bending moment constrained case with shape variables,
M = 0.7, part 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 164
M = 0.7, part 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165
ix
M = 0.85, part 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172
M = 0.85, part 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 173
D.29 FFD coordinates for Cm constrained case with shape variables, M = 0.85, part 1174
D.30 FFD coordinates for Cm constrained case with shape variables, M = 0.85, part 2175
D.31 FFD coordinates for Kn constrained case with shape variables, M = 0.85, part 1 176
D.32 FFD coordinates for Kn constrained case with shape variables, M = 0.85, part 2 177
D.33 FFD coordinates for CAP constrained case with shape variables, M = 0.85, part 1178
D.34 FFD coordinates for CAP constrained case with shape variables, M = 0.85, part 2179
List of Figures
3.1
Cp contour and streamline comparison, rotating NACA 0012: M=0.2,=0.00 . . 21
3.2
Flowfield streamline comparison, rotating NACA 0012: M=0.2, =0.00 . . . . . 22
3.3
Pure angle of attack motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.4
Pure pitching motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33
3.5
Time-spectral solution: motion: Mach = 0.1 . . . . . . . . . . . . . . . . . . 33
3.6
Time-spectral solution: CL vs : Mach = 0.1 . . . . . . . . . . . . . . . . . . . . 34
3.7
Time spectral solution: CL hysteresis vs :

Mach = 0.5 . . . . . . . . . . . . . . 35
3.8
NACA 0012 meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.9
NACA 0012 mesh convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.10 Plunging NACA 0012: time-spectral stability derivative verification . . . . . . . . 39

3.11 SACCON meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.12 SACCON mesh convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
3.13 SACCON UCAV: CL derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.14 SACCON UCAV: Cm derivatives . . . . . . . . . . . . . . . . . . . . . . . . . . . 42
3.15 SACCON UCAV: CL and Cm comparison . . . . . . . . . . . . . . . . . . . . . . 43
3.16 SACCON UCAV: derivatives: frequency comparison . . . . . . . . . . . . . . . 44
3.17 SACCON UCAV: derivatives: frequency comparison . . . . . . . . . . . . . . . 44
3.18 SACCON UCAV: q and derivatives: frequency comparison . . . . . . . . . . . . . 45
4.1
Single cell stencil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50
4.2
Call-graph: current residual computation: steady and time-spectral cases . . . . 51
4.3
Call-graph: improved residual computation: time-spectral case . . . . . . . . . . 53
4.4
Call-graph: force and moment computation . . . . . . . . . . . . . . . . . . . . . 54
4.5
Steady case sparsity pattern: 24 block H-H mesh . . . . . . . . . . . . . . . . . . 56
4.6
Time-spectral sparsity pattern: 24 block H-H mesh: 3 time instances . . . . . . . 57
4.7
ONERA M6 test case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
4.8
Time-spectral ADjoint scaling . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 62
5.1
Diagram of CG calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65
xi
5.2
Discrete surface mesh with lumped masses at the cell centroids . . . . . . . . . . 66
5.3
Cell area calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
5.4
Bending moment calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69
5.5
Geometry manipulation setup . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 74
5.6
Deformed geometry . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 75
6.1
Initial geometry of wing test case . . . . . . . . . . . . . . . . . . . . . . . . . . . 77
6.2
Generic MDF architecture [51] . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.3
Generic IDF architecture [51] . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
6.4
Thickness constraints: constraint locations and initial values represented by the

vertical line segments marked by the black spheres . . . . . . . . . . . . . . . . . 84
6.5
FFD control point constraints: initial configuration (upper left), perturbation in

section twist (upper right), unconstrained perturbation of leading edge control
points (lower left), constrained perturbation of leading edge control points as
well as a perturbation of mid-chord control points (lower left) . . . . . . . . . . . 86
6.6
XDSM for the baseline problem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 87
6.7
Baseline problem: M=0.5, e = 0.964 . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.8
Baseline problem: M=0.7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
6.9
Baseline problem: M=0.85 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89
6.10 XDSM for the twist optimization problem . . . . . . . . . . . . . . . . . . . . . . 90

6.11 Twist optimization: M=0.5, e = 0.977 . . . . . . . . . . . . . . . . . . . . . . . . 90
6.12 XDSM for the root bending moment constrained problem . . . . . . . . . . . . . 92
6.13 Bending moment constrained optimization: planform variables only: M=0.5,
e = 0.964 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
6.14 Bending moment constrained optimization: planform variables only: M=0.7 . . . 93
6.15 Bending moment constrained optimization: planform and shape variables: M=0.7 93
6.16 Bending moment constrained optimization: planform and shape variables: M=0.85 94
6.17 XDSM for the Cm constrained problem . . . . . . . . . . . . . . . . . . . . . . . 96
6.18 Cm constrained optimization: planform variables only: M=0.5, e = 0.963 . . . . 97
6.19 Cm constrained optimization: planform and shape variables: M=0.5, e = 0.968 . 97
6.20 Cm constrained optimization: planform variables only: M=0.7 . . . . . . . . . . 98
6.21 Cm constrained optimization: planform and shape variables: M=0.7 . . . . . . . 99
6.22 Cm constrained optimization: planform and shape variables: M=0.85 . . . . . . 100
6.23 XDSM for the Kn constrained problem . . . . . . . . . . . . . . . . . . . . . . . . 102
6.24 Kn constrained optimization: planform variables only: M=0.5, e = 0.944 . . . . . 102
6.25 Kn constrained optimization: planform and shape variables: M=0.5, e = 0.968 . 103
xii
6.26 Kn constrained optimization: planform variables only: M=0.7 . . . . . . . . . . . 103

6.27 Kn constrained optimization: planform and shape variables: M=0.7 . . . . . . . 104
6.28 Kn constrained optimization: planform and shape variables: M=0.85 . . . . . . . 105
6.29 XDSM for the CAP constrained problem . . . . . . . . . . . . . . . . . . . . . . . 106
6.30 CAP constrained optimization: planform variables only: M=0.5, e = 0.942 . . . . 107
6.31 CAP constrained optimization: planform and shape variables: M=0.5, e = 0.970 108
6.32 CAP constrained optimization: planform variables only: M=0.7 . . . . . . . . . . 108
6.33 CAP constrained optimization: planform and shape variables: M=0.7 . . . . . . 109
6.34 CAP constrained optimization: planform and shape variables: M=0.85 . . . . . . 110
xiii
Nomenclature
Introduction:
Cp
coefficient of pressure
Chapter 2:
g
gravitational acceleration
Ii
moment of inertia components
L, Li
roll moment in the body frame, stability derivative with respect to parameter i
mass
M , Mi
pitch moment in the body frame, stability derivative with respect to parameter i
N , Ni
yaw moment in the body frame, stability derivative with respect to parameter i
p, p
roll rate in the body frame, time derivative of p
q, q
pitch rate in the body frame, time derivative of q
r, r
yaw rate in the body frame, time derivative of r
u, u
velocity in x direction in the body frame, time derivative of u
v, v
velocity in y direction in the body frame, time derivative of v
w, w
velocity in z direction in the body frame, time derivative of w
X, Xi
force in x direction in the body frame, stability derivative with respect to parameter i
Y , Yi
force in y direction in the body frame, stability derivative with respect to parameter i
Z, Zi
force in z direction in the body frame, stability derivative with respect to parameter i
angle of attack
short-period frequency
vehicle Euler angle
vehicle Euler angle
vehicle Euler angle
short-period damping ratio
Chapter 3:
b
span
chord
Dt
spectral derivative operator
et
total energy
generic function
xiv
fi
flux term in direction i
Fi
force in the ith coordinate direction
h
i
step length
identity matrix (Section 3.3.1), function of interest (Section 3.3.2)
thermal conductivity of fluid (Section 3.3.1)
frequency index (Section 3.4.1)
reduced frequency (Section 3.4.3)
time instance index
Mi
moment about the ith coordinate axis
time instance index
number of spectral time intervals
pressure (Section 3.3.1), roll rate (Section 3.3.2)
pitch rate
yaw rate
source terms
intermediate variable in automatic differentiation (Section 3.2), time
temperature (Section 3.3.1), time period of periodic flow problem (Section 3.4.1)
fluid velocity in fixed frame
velocity of fluid with respect to moving grid
aircraft speed, cell volume (Section 3.4.1)
velocity of the moving grid
grid coordinates (Section 3.3.1), design variables (Sections 3.3.2, 3.2)
flow residual
normalized roll rate (pb/2V )
normalized pitch rate (qc/2V )
normalized yaw rate (rb/2V )
angle of attack
side slip angle
ith control surface deflection (Section 3.3.2)
elements of the identity matrix (Section 3.3.1)
adjoint vector
specific heat ratio
rotational rate of the moving grid
density
viscous stress tensor
xv
generalized time variable
flow states for entire mesh
cell
flow states for a single cell
CD
aircraft drag coefficient
CL
aircraft lift coefficient
Cl
airfoil lift coefficient (Section 3.3.1), aircraft roll moment coefficient (Section 3.4.2)
Cm
airfoil pitch moment coefficient (Section 3.3.1)
Cm
aircraft pitch moment coefficient (Section 3.4.2)
Cn
aircraft yaw moment coefficient
Cp
CY
aircraft side force coefficient
Ci0
initial value of coefficient Ci
Cij
stability derivatives (derivative of Ci with respect to j)
Chapter 4:
Dt
spectral derivative operator
h
i
step size
function of interest
number of spectral time intervals
Ncells
number of cells in the mesh
NI
number of functions of interest
Nx
number of design variables
number of states per cell
N T
total number of states
cell volume
design variables
steady flow residual
RT S
time-spectral flow residual
angle of attack
flow states
CD
average drag coefficient
Cmi
stability derivatives (derivative of Cm with respect to i)
CL
average lift coefficient
Cm
average pitch moment coefficient

xvi
Chapter 5:
A
area
cref
reference chord length
CG%
location of the center of gravity expressed as a percentage of the distance

between the two wing spars, referenced from the leading edge spar.
gravitational acceleration
hCG
location of the center of gravity normalized by the MAC
hNP
location of the neutral point normalized by the MAC
second moment of area
Izz
pitch moment of inertia
Kn
static margin
generic length scale
mass
Mach number
bending moment
normal acceleration of the aircraft
change in normal acceleration of the aircraft with
dynamic pressure
pitch acceleration of the aircraft
distance
cell area
thickness
velocity
x, y, zCG center of gravity location

x, y, zmass point mass location
xbendingi
root location of bending axis
xrefi
moment coefficient reference point
beam height for stress computation
angle of attack
domain of integration
density
bending stress
short-period natural frequency
Cb
total coefficient of bending

xvii
Cbendingx
coefficient of bending moment about x axis
Cbendingz
coefficient of bending moment about z axis
CL
coefficient of lift
Cfx
force coefficients in the x direction
Cfy
force coefficients in the y direction
Cfz
force coefficients in the z direction
CLi
stability derivatives (derivative of CL with respect to i
CmCG
moment coefficient about the CG location
CmNP
moment coefficient about the NP location
Cmx
moment coefficients about the x axis
Cmy
moment coefficients about the y axis
Cmz
moment coefficients about the z axis
Cmi
stability derivatives (derivative of Cm with respect to i)
Chapter 6:
A
area
span
croot , ctroot
root chord length, target root chord length
CG%
location of the center of gravity expressed as a percentage of the distance

between the two wing spars, referenced from the leading edge spar.
span efficiency factor
Kn
static margin
Mach number
MACt
target mean aerodynamic chord
PIzz
multiplication factor on computed Izz
xCG , xtCG
location of the CG, target location of the CG
coupling variables
angle of attack
sweep
section twist at station i
FFD
change in FFD volume control points
sp
short-period damping ratio
Cb
total coefficient of bending
t
CD , CD
drag coefficient, target drag coefficient
Cl
section lift coefficient

xviii
CL
lift coefficient
Cm , Cmz
pitch moment coefficient
CmCG
moment coefficient about the CG location
CmNP
moment coefficient about the NP location
Cp
CLi ,CLt i
t
Cmi ,Cm
i
stability derivatives (derivative of CL with respect to i), target stability derivatives

stability derivatives (derivative of Cm with respect to i), target stability derivatives
Abbreviations:
AD
Automatic differentiation
BLISS
Bi-level integrated system synthesis
BWB
Blended-wing-body
CAP
Control anticipation parameter
CFD
Computational fluid dynamics
CG
Center of gravity
CO
Collaborative optimization
CSSO
Concurrent subspace optimization
DATCOM
United States air force data compendium
DES
Discrete eddy simulation
ESDU
Engineering sciences data unit
FFT
Fast Fourier transform
FFD
Free form deformation
ICAO
International civil aviation organization
IDF
Individual discipline feasible
MAC
Mean aerodynamic chord
MDF
Multidisciplinary feasible
MDO
Multidisciplinary design optimization
NACA
National advisory committee for aeronautics
NATO RTO North Atlantic treaty organization research and technology organization
NLFD
Nonlinear frequency domain
NP
Neutral point
ONERA
Office National dEtudes

et de Recherches Aerospatiales
RANS
Reynolds averaged Navier-Stokes
SACCON
Stability And Control CONfiguration
SAND
Simultaneous analysis and design

xix
UCAV
Unmanned combat aerial vehicle
XDSM
Extended design structure matrix
xx
Chapter 1
Introduction
Demand for aircraft transportation is increasing at a tremendous rate around the world. According to Boeing and Airbus projections, global airline passenger traffic will increase at an
average annual rate of close to 5% for the next 10 to 20 years [2, 12]. In particular, Boeing
expects there to be over 35,000 aircraft in passenger service by 2028, of which 29,000 will have
been newly constructed between 2008 and 2028 [12]. At the same time, the public is demanding
increased accountability regarding the environmental impact of all carbon consuming industries,
air transportation included. To put aviations place in this area into perspective, at 2006 levels,
aviation contributed approximately 2% of global CO2 emissions [45]. While this contribution
is relatively minor at 2006 levels, ICAO projects that with the aforementioned growth in air
traffic demand, aviation CO2 emissions will grow from the 2006 level of 632 Mega tonnes(Mt)
to between 1,422 and 1,738 Mt in 2036 [45]. This is more than a factor of two increase from
2006 levels.
In an attempt to mitigate the effects of this growth, the commercial aviation industry is
investigating a variety of possibilities to reduce its environmental footprint. Some of these
possibilities include the use of alternative fuels, improvements to operational practices and
technical innovation to improve the efficiency of the world fleet. While each one of these areas
promises to provide significant efficiency gains, the magnitude of the expected increase in traffic
is such that no one field on its own will be able to offset all of the increase. Therefore, all possible
avenues of improvement must be explored.
This thesis relates primarily to the improvement in efficiency of the aircraft itself. Two main
areas will be considered, the use of unconventional designs to improve the efficiency of aircraft
and the use of design optimization to facilitate the design of these unconventional concepts. In
particular, this thesis will focus on the design and optimization of flying wing configurations
and show that high-fidelity aerodynamic shape optimization can be used to produce stable and
flyable, flying wing designs.
1
Chapter 1. Introduction
1.1
The Case for Unconventional Configurations
The aviation industrys track record for technological improvement is impressive. According to
Lee [52], average fuel consumption per revenue passenger mile improved by approximately 70%
between 1959 and 1995. This improvement has come from a variety of new technologies such
as high-bypass ratio turbofans, supercritical airfoils and improved structural materials as well
as improvements in design and analysis methods. While these improvements are impressive, the
consensus in the design community seems to be that most of the benefit available from these
improved technologies and design methods has been realized. Given the expected increase in
demand for air travel, the incremental improvement still available from these approaches will
not be sufficient to achieve the performance targets set for the industry. In response to this,
a variety of alternative technologies are being explored, with the goal of finding solutions that
will provide significant improvements in efficiency.
One of the major ideas being advanced as a solution to this issue is the use of unconventional airframe designs. For each of these alternative concepts such as the blended-wingbody (BWB) [53], the joined wing [113, 28] or the strut-braced wing [36, 29] there is some
physical characteristic that has been shown, at least on paper, to provide significant efficiency
improvements over conventional aircraft, for a given technology level. For example, Liebeck [53]
demonstrates a blended-wing-body design that, through aerodynamic and structural improvements, has approximately 30% lower fuel burn per passenger than an equivalently sized A380
aircraft. Therefore, while these configurations may be relatively untried, they certainly deserve
to be developed further in order to identify what is needed to take advantage of their potential.
1.2
The Need for Stability Constraints in High-Fidelity Optimization
While the blended-wing-body promises improved aerodynamic and structural performance [53,
13, 90, 85], the improved efficiency does not come for free. Because of the unconventional
layout of the passenger cabin, the configuration raises questions about passenger comfort, market acceptance and emergency egress that would have to be addressed in any commercialized
design [53, 13]. Also, because the configuration is essentially a single, large, lifting surface, it
introduces additional stability and control considerations into the design problem [103, 53, 109].
Further, there is a strong coupling between the aerodynamic performance of the configuration
and its trim and stability characteristics. For example, Liebeck notes that there is a conflict
between the high aft camber characteristics common to the class of transonic airfoils needed
for good high speed performance and the need to maintain a low pitch moment airfoil in order
to be able to trim a BWB at cruise [53]. More specifically, transonic airfoils are designed to
limit the peak Cp values generated by the airfoil to a value below the critical Cp that will cause
shock waves to form on the wing. This tends to force the generation of lift further back on the
airfoil. Low moment airfoils are trying to balance the lift generation fore and aft of the quarter
chord of the foil. Given the ratio of area fore and aft of this point, this fundamentally requires
relatively large Cp values forward of the quarter chord. This effect is clearly demonstrated in
the Mach = 0.85 cases in Chapter 6. While multidisciplinary coupling such as this is inherent in all aircraft design, issues such as these tend to become significantly more restrictive in
unconventional configurations.
To further complicate matters, by their very nature, unconventional designs are well outside
the typical range of experience for most designers. As a result, there is not a large body
of experiential knowledge to be drawn upon in the design of such aircraft. In response to
this, the design of these configurations tends to gravitate towards the use of higher fidelity
analysis methods and more powerful design techniques such as design optimization, as these
methods allow for accurate and efficient exploration of the design space. This is also true
of this thesis. The core analysis involved in this work is a high-fidelity computational fluid
dynamics (CFD) solver, capable of solving Euler and RANS flows for steady, unsteady and timespectral problems. This solver is used to model the aerodynamic performance of the various
configurations considered. However, as shown by Martins et al. [64], simply having a highly
accurate single discipline optimization is not sufficient to achieve a good design. Optimization
of a single discipline on its own often leads to a suboptimal design of the full multidisciplinary
system. Indeed, as will be shown in Chapter 6, in the context of flying wing design, pure
aerodynamic shape optimization can lead to unstable designs if stability characteristics are not
considered. Thus, the goal of this work is to develop techniques that can be used to calculate
stability and control information from the high-fidelity CFD solver that can then be used in
conjunction with high-fidelity aerodynamic shape optimization techniques.
1.3
Overview of the Stability Constraint Approach
To compute the stability and control metrics required for the aircraft, and to keep the cost of
analysis reasonable for numerical optimization, linear flight dynamics will be used to model the
stability of the aircraft. This linear flight dynamics model will be used to formulate a variety
of flight dynamics constraints, which will ensure that the optimization produces a stable design
with the desired handling characteristics. While not as accurate as a fully nonlinear flight
dynamic model, this approach will provide sufficient information to examine the effects of
stability considerations on the optimal design of flying wing configurations.
Linear flight dynamics requires information about three characteristics of the aircraft: how
its aerodynamic characteristics change with the dynamic states (stability derivatives), how much
mass it has (mass) and how its mass is distributed (moments of inertia). A primary focus of this
work is examining methods for the computation of stability derivatives and their sensitivities
for use in optimization. Two approaches, both based on three-dimensional CFD, are introduced
and discussed. The first approach is based on the automatic differentiation adjoint (ADjoint)
sensitivity analysis method developed by Mader et al. [60]. This method is very efficient at calculating the full suite of static (, , V ) and dynamic (p, q, r) derivatives, but can not calculate
the transient (,
q,
etc.) derivatives. It is also not suited to large scale optimization, as the
cost of computing the gradients of the stability derivatives with respect to the design variables
scales with the number of design variables. Therefore, a second approach, based on unsteady,
forced oscillation techniques was developed. This method uses a time-spectral CFD method to
compute a forced oscillation solution, which is then used to generate the stability derivatives
algebraically. This whole system can then be differentiated using the ADjoint technique to
provide the sensitivities required for gradient-based optimization. Furthermore, because the
stability derivatives are based on an unsteady solution, estimation of the dot derivatives is
possible. The inclusion of the dot derivatives allows for the computation of handling qualities
estimates, which are not possible with only the static and dynamic derivatives.
The second two characteristics mass and moment of inertia receive a rather cursory
treatment in this work. The calculation of these parameters are based on relatively simple,
preliminary and conceptual level methods as described in Chapter 5. The author acknowledges
that the results of this work would benefit from more rigorous treatment of these calculations.
However, as the focus of this work was the aerodynamic side of the problem, that portion of
the analysis was left for future research.
1.4
Thesis Layout
The work presented in this thesis is laid out in five parts. Chapter 2 introduces the basics of
the linear flight dynamics models used to generate the dynamic stability constraints for the
optimization problems. Chapter 3 focuses on the computation of the stability derivatives, with
sections on both the ADjoint approach and the time-spectral approach. Each approach is introduced, discussed and verified against benchmark results. Chapter 4 introduces and discusses the
sensitivity analysis methods used to compute the sensitivities required for gradient-based optimization. In particular, it introduces and discusses the time-spectral ADjoint method, which
is a key element in enabling efficient stability-constrained optimization. Chapter 5 outlines the
simple methods used to calculate the center of gravity (CG) location and moment of inertia of
the aircraft, as well as the specific methods used to implement the various stability constraints
in the optimization. Finally, Chapter 6 presents the optimization results obtained during the
study.
1.5
Original Contributions
Previous work has shown that the inclusion of flight dynamics and control in MDO can have a
significant effect on the optimal design of an aircraft [88, 40]. While inclusion of stability and
control information has been accomplished at the low-fidelity level [110, 88, 40], the effects of
this discipline have not been explored at the high-fidelity level. The work in this thesis seeks
to start addressing this deficiency.
Since a linear flight dynamic model has been selected to describe the flight dynamics of the
aircraft, the question of computing the aerodynamic characteristics of the aircraft for the stability discipline is reduced to one of computing the stability derivatives for the aircraft. Therefore,
the first step in addressing this deficiency is developing a method for the computation of aircraft
stability derivatives that is compatible with CFD based aerodynamic optimization. Several researchers for example, Park et al. [83, 82], Murman [74] and Limache and Cliff [54] have
explored the use of CFD in the computation of stability derivatives. However, all of these
methods focus strictly on the computation of the stability derivatives and make no consideration of the implications associated with using these methods in optimization. The work in this
thesis contributes to this area in two ways. First, the ADjoint approach to the computation
of stability derivatives is introduced and demonstrated. This method exhibits extremely good
numerically accuracy and is an efficient method for computing full sets of static and dynamic
stability derivatives from a single CFD solution. Further, because it is based on an adjoint
method, the method is well suited to the computation of derivatives in cases involving large
numbers of control effectors. However, because this method is itself based on advanced sensitivity analysis techniques, it is not well suited to use in optimization. For this task, the second
major contribution of this work the time-spectral stability derivative method is much more
useful. Like the methods of Murman [74] and DaRonche et al. [94], the time-spectral stability
derivative approach is based on a forced oscillation technique. However, it has some unique
qualities. First, two unique motions are used that allow the derivatives and q derivatives
to be computed independently. This allows the derivatives to be computed independently
of the q derivatives, which is useful in the assembly of the flight dynamic constraints. Also,
because this method uses relatively simple algebraic operations on top of a periodic unsteady
solution, it is relatively straight forward to apply an adjoint technique to the problem to obtain
the sensitivities necessary for gradient-based optimization. This leads to the third contribution
in this thesis, the time-spectral ADjoint. Previous work has demonstrated the use of adjoint
methods for spectral flow solution techniques. Thomas et al. [104] have demonstrated an adjoint technique for the two dimensional harmonic balance technique, Nadarajah and Jameson
have implemented an adjoint for the Nonlinear Frequency Domain method [76] and Choi et
al. [20] have demonstrated an adjoint method for the time-spectral equations. However, the
time-spectral adjoint method of Choi et al. contained approximations for the derivatives of the
spectral radius and artificial dissipation computations, thereby limiting its numerical accuracy.
The time-spectral ADjoint approach introduced in this work avoids those approximations by
using automatic differentiation to compute the components of the adjoint system. Further,
because the time-spectral equations are formulated entirely in the real domain, the complex
step method can be used to verify the sensitivities. As is shown in Section 4.4, this allows
for verification of the derivatives to extremely small tolerances. These two techniques the
time-spectral stability derivative method and the time-spectral ADjoint can be combined
to enable the final major contribution of this thesis, the application of stability constraints in
high-fidelity aerodynamic shape optimization. As is shown in Chapter 6, the inclusion of stability constraints has a significant impact on the optimal shape for an aircraft across a wide range
of Mach numbers. To the best of the authors knowledge, this is the first time this capability
has been demonstrated in the context of high-fidelity aerodynamic shape optimization.
Chapter 2
Linear Flight Dynamics

The fundamental idea pursued in this thesis is that, in order to properly optimize an aircraft,
one needs to account for both the stability characteristics and handling characteristics of the
aircraft during the optimization process. As shown by the results in Chapter 6, without this
information, the optimization produces an aerodynamically optimal design that may or may
not be stable and flyable. Such a design would then require follow-up modifications to achieve
the desired level of stability and handling, yielding a suboptimal design. This is of particular
importance with unconventional designs, such as a BWB or a flying wing, because of the strong
coupling between aerodynamic performance, trim and stability in these designs. However, while
this is a simple enough goal to describe, it actually turns out to be challenging to implement
in practice.
To provide a basis for the following discussion, consider the equations of motion for an
aircraft, as written by Etkin [27]:

X
sin
(u + wq rv)

Y + mg sin cos = m (v + ru pw)

Z
cos cos
(w + pv qu)
Ix

p
0 r q
Ix
Ixy Izx
p
Iy
Iyz q
Iyz q + r
0 p Ixy
Izx Iyz
Iz
r
Iz
r
q p
0
(2.1)
Ixy Izx

M = Ixy
Iy

Izx Iyz
N
1 sin tan cos tan

= 0
cos
sin

q
0 sin sec cos sec
r
(2.2)
(2.3)
These equations assume that the aircraft is rigid, that the effect of spinning rotors is negligible,
that the mass of the aircraft is constant and that the aircraft is flying through a still air mass.
7
Chapter 2. Linear Flight Dynamics
Note that the I variables represent the various aircraft moments of inertia, m represents the
aircraft mass and , , and are the vehicle Euler angles.
As can be seen from these equations, the relations governing the motion of an aircraft are
quite complex. Both sets of equations (2.1) and (2.2) are fully nonlinear. In addition, the forces
X, Y , Z and moments L, M , N are dependent on the current as well as the past
values of the motion variables in the equations, namely the velocities u, v and w and the
rotational rates p, q, and r. Therefore, to accurately evaluate this system, one would need
to run a time accurate evolution of the motion equations, evaluating the forces on the aircraft
at each instance in time. Given that computationally intensive CFD methods are being used
to compute the forces on the aircraft, this would be prohibitively expensive in the context of
an optimization problem.
A common methodology for reducing the complexity of this system of equations is to formulate a linearized form of the equations. This is accomplished in two stages. First, using small
disturbance theory and assuming small angle theory to be valid, equations (2.1) through (2.3)
can be rewritten as:
sin 0 cos 0
u + w0 q v0 r
X0 + X
= m v + u0 r w0 p)
Y0 + Y + mg
cos
w + v0 p u0 q)
cos 0 0
Z0 + Z

M0 + M = Ixy
Iy
Iyz q
N0 + N
Izx Iyz
Iz
r
1 0 tan 0
p

= 0 1

0

q .
0 0 sec 0
r
L0 + L
Ix
(2.4)
Ixy Izx
(2.5)
(2.6)
where values with the subscript 0 represent initial values. One can then express the linearized
forces as a sum of their derivatives with respect to the motion variables:
XControl Control
Xj j
Xi i
X
Y Yi i Yj j YControl Control

Z Zi i Zj j ZControl Control
+
+

L Li i Lj j LControl Control
M M i M j M
i j Control Control
NControl Control
Nj j
Ni i
N
(2.7)
where i = u, v, w, p, q, r, xE , yE , zE , , , and j = di/dt. The subscribed quantities are derivatives of each force or moment quantity with respect to the corresponding subscript.
After substituting in the initial conditions, this allows the flight dynamic system to be
rewritten as a state-space system of the form:
x = Ax + bu.
(2.8)
This is a set of coupled ordinary differential equations (ODEs) in state-space form, which can
be solved in a standard fashion. The full derivation of this system is shown in Appendix A.
2.1
Decoupled Longitudinal Equations
If one considers the aircraft to be symmetric about the mid-plane, both the cross moment of
inertia elements Ixy and Iyz and the change in lateral forces with respect the longitudinal
degrees of freedom are zero. Further, one can also assume that the change in longitudinal
forces and moments with respect to lateral motions is small enough to neglect. This allows the
longitudinal degrees of freedom to be separated from the lateral degrees of freedom.
Continuing with the derivation above, by setting Ixy and Iyz to zero and removing the cross
dependence of the forces and moments, one can write the longitudinal set of equations as:
(Xu m)
Xw
Xq
Zu
Zw m
Zq
Mu
Mw
Iy + Mq
w
q +
Xu
Xw
(Xq mw0 )
mg cos 0
Zu
Zw
muo + Zq
mg sin 0
Mu
Mw
Mq

w ZControl
q + M
Control = 0.
XControl
(2.9)
Control
These equations can be solved for the time derivative terms to yield:
w
q =
Xu
m
Zu
(Zw m)
(Mw Zu )
u
M
(Iy (Zw m))
Iy
Xw
m
Zw
(Zw m)
(Mw Zw )
w
M
(Iy (Zw m))
Iy
(Xq mw0 )
m
(Zq +mu0 )
(Zw m)
(Mw (Zq +mu0 ))
M
Iq
(Iy (Zw m))
y
(gm sin(0 ))
(Z m)
w
(M gm sin(0 ))
(I w(Z m))
y
w
w
+
q
XControl
g cos(0 )

ZControl
M
Control.
(2.10)
Control
Again, this is an equation in standard state-space form, which can be solved using standard
differential equation solution techniques.
10
2.2
Short-Period Approximation
While in general, one could solve the full flight dynamics model to determine stability and
control characteristics for an aircraft, for this work, a simplified short-period approximation is
used. This approximation assumes that the variation in forward velocity (u) is negligible and
that the short-period characteristics of the aircraft can be modeled by changes in pitch rate (q)
and angle of attack () alone. In this case, vertical velocity (w) is used in place of the angle of
attack where:
w
.
(2.11)
u
To facilitate this approximation, the second and third equations are extracted from equatan =
tions (2.10). Further, it is assumed, as per McRuer et al. [71] that:

Xw = Xq = Zw = Zq = 0,
(2.12)
as these derivatives are generally small.

With these assumptions, the relevant equations become:
"
"
#
# "
#
Zw
h
i
u
w
w
Z
0
Control
= M Z m M
+
.
Control
M
w
w
Iyw MIyw u0 Iyq
q
q
MControl
Iy m
(2.13)
Ultimately, as will be seen in Chapter 5, the handling qualities analysis requires the frequency and damping ratio of the system. These values can be determined from the characteristic
equation of the 2 2 system in equation (2.13). The characteristic equation of this system can
be calculated as det(sI A) or:

w

s Z
m

Mw Zw Mw
Iy m
Iy

u0

Mq
Mw u0
s
Iy
Iy
(2.14)
which evaluates to:

s2 +
Mw u0 Mq
Zw
+
+
Iy
Iy
m

s+
Zw Mq
Mw u 0
.
mIy
Iy
(2.15)
Using the mass and inertia normalizations of McRuer et al. [71]:

i = Mi
M
Iy
(2.16)
and
Zi
Zi =
(2.17)
m
where i = w,
w, or q one ends up with the second order characteristic equation shown by
McRuer et al. [71]:

w u0 + M
q + Zw + Zw M
q Mw uo .
s2 + M
(2.18)
11

Comparing this to the typical second order characteristic equation:
s2 + 2s + 2
(2.19)
yields:
=
q Mw uo
Zw M
(2.20)
where is the short-period natural frequency, and

w u0 + M
q + Zw
2 = M
or
=
w u0 + M
q + Zw
M
.
2
(2.21)
(2.22)
where is the short-period damping ratio.
2.3
Summary
By making the simplifications demonstrated above, the complex, nonlinear flight dynamics
problem is reduced to a set of coupled, ordinary differential equations which are much simpler to
solve. Further, rather than having to compute time dependent forces, the forces are represented
by a set of aerodynamic derivatives, typically called stability derivatives [27], which can be
computed about a single operating point. Using these stability derivatives as a basis for the
model, simple calculations can then be performed to find the frequency and damping ratio of
the system, which can in turn be used to quantify the handling qualities of the aircraft. Now
the key part is to find an efficient way of computing stability derivatives for use in optimization,
which is the subject of Chapter 3 and one of the goals of this thesis.
Chapter 3
CFD-Based Stability Derivatives

Having reduced the aerodynamic problem from one of computing a full set of time accurate
forces for a given maneuver to that of computing the stability derivatives about a given flight
condition, one can now examine the methods available for computing these derivatives. Historically, these derivatives have been estimated using empirical methods (e.g., DATCOM [106],
ESDU [26]), wind tunnel testing, or lower fidelity CFD methods such as panel codes. While
each of these methods is useful, they each have their own disadvantages with respect to various
factors such as cost, range of applicability, and computational time. Given that this work is
focused on conducting CFD-based aerodynamic shape optimization, it is necessary to be able
to compute accurate stability derivatives over the same range of flight conditions for which the
flow is being analysed. This leads to the idea of computing the stability derivatives with the
same high-fidelity methods that are being used to compute the flow.
3.1
Background
Over the last 15-20 years, several research groups have examined the idea of computing stability derivatives using high-fidelity CFD. Some early work in this area was conducted by
Charlton [19], who conducted simple and sweeps to get the force and moment information
required for falling-leaf predictions for tailless aircraft. The study concluded that, in most instances, the required stability data could be computed accurately, but that highly nonlinear
regions of the flow required larger test cases and longer run-times than desired. It is these
highly nonlinear cases which are often critical in aircraft design. While not explored in detail in
this work, the methods proposed here extend to those flow regimes, provided the CFD methods
used provide accurate solutions in those cases.
Godfrey and Cliff [33] explored the use of analytic sensitivity methods in particular, the
direct method for the computation of static stability derivatives. They show the computation
12
Chapter 3. CFD-Based Stability Derivatives
13
of the and derivatives of the YB-48 flying wing using a three dimensional inviscid flow solver.
Their results match existing semi-empirical methods to between 5% and 30%, depending on the
derivative. However, no comment is made regarding the computational efficiency of the method.
Limache and Cliff [54] followed up this work by examining the use of the same method for the
computation of dynamic stability derivatives. Limache and Cliff show the computation of the
dynamic pitching derivatives of an airfoil using a two-dimensional Euler CFD code. This study
shows the significant promise of analytic sensitivity methods for the computation of stability
derivatives, showing that multiple stability derivatives including the rotational derivatives
can be computed from a single steady solution. The study also highlights the importance of
using, at the minimum, Euler CFD to capture the shock waves seen in transonic flow.
Another approach that has been pursued for the computation of stability derivatives is
the use of automatic differentiation (AD). Park et al. [83, 82] applied ADIFOR [8] an
automatic differentiation (AD) tool to a three-dimensional viscous flow solver to compute
the static and dynamic derivatives of various configurations. The results from this work showed
promise, providing accurate results across a variety of flight conditions. However, the additional
computational cost of running the differentiated code significantly impacted the computational
efficiency of the approach, leading to an eight-fold increase in the computational time required
to compute the derivatives when compared to a single flow solution.
The previous two approaches to dynamic derivative computation those of Limache and
Park rely on a non-inertial reference frame CFD formulation to handle the rotational rates
needed for the dynamic derivatives. Babcock and Arena [5] handle the dynamic derivatives
by modifying the boundary conditions in a finite element based Euler CFD solver to separate
the velocity and position boundary conditions. With this approach they are able to perturb
the static states (, ) and the dynamic states (p,q,r) independently to determine the stability
derivatives using finite differences. The results from this approach compare relatively well with
theoretical, empirical and experimental results, confirming the validity of the method. However,
no computational cost details are included in the study.
A separate approach for computing the dynamic derivatives one that has been used in the
experimental community for many years is the forced oscillation approach. In this approach,
the test aircraft is driven thorugh a prescribed oscillatory motion. The resulting oscillatory
aerodynamic data can then be used to determine various characteristics of the aircraft. A
number of papers from the recent NATO RTO Task Group AVT-161 have explored the use
of this technique in conjuction with a variety of CFD solution methods [72]. In particular,
it has been demonstrated with RANS [96, 95], DES [22] and harmonic balance [94] solution
techniques. These techniques were shown to correlate well with experimental data.
In an independent study, Murman [74] also presents a method for computing stability deriva-
14
tives based on the traditional forced oscillation approach. In this work, Murman uses a frequency domain method to produce periodic data for the forced oscillation of the configuration
of interest. This data can then be analyzed with the same techniques used to produce stability
derivatives from forced oscillation wind tunnel data, which allows the method to take advantage
of the large body of knowledge that has been developed in that field. The study demonstrates
good accuracy for a variety of configurations and flight conditions, however because of the time
periodic nature of the solution, the computational cost is higher than would be the case for an
equivalent steady-state solution of the same configuration.
In the following chapter, two approaches for computing stability derivatives using CFD
are introduced. The first, based on adjoint sensitivity methods, is an efficient method for
calculating static and dynamic stability derivatives about a single operating point. The second
approach, based on time-spectral CFD, includes unsteady information, thereby allowing for
the computation of the dot derivatives. It is also more conducive to optimization than the
ADjoint approach. The details of these two methods are explained in the following sections.
3.2
Sensitivity Analysis Review
Before proceeding to desribe the methods developed for computing stability derivatives from
CFD, it is worth taking a brief aside to talk about sensitivity analysis for CFD in general.
Sensitivity analysis is typically performed using one of four main methods: finite-difference,
complex-step, analytic methods or automatic differentiation. The characteristics of each method
are as follows:
Finite-Difference: Finite-difference methods [15] are probably the most common methods
for calculating derivatives. The method comes in many forms, but the basic form of the method
can be expressed as:
df (x)
f (x + h) f (x)
=
.
dx
h
(3.1)
this method is popular because it requires no specialized knowledge of the code being differentiated. However, the method is subject to subtractive cancelation errors, which limit its overall
accuracy [102]. Further, the basic forward method shown above requires N + 1 function evaluations to compute a full gradient, where N is the number of design variables x. This leads to
very poor scaling as the number of design variables increases.
Complex-Step Method: The complex-step method [57, 102, 65, 80] is analagous to the
finite-difference method, but uses a complex perturbation rather than a real perturbation. This
leads to a formulation of the form:
df (x)
Im [f (x + ih)]
=
.
dx
h
(3.2)
15
The use of a complex perturbation eliminates the subtractive cancellation issues inherent in
finite-difference methods and allows for extremely high numerical accuracy. Unfortunately, like
finite-difference methods, the cost of this method scales with the number of design variables.
Analytic Methods: For the purposes of this discussion, this class of methods consists of
the adjoint method and the direct method. This class of methods is primarily for improving the
efficiency and accuracy of the computation of derivatives for problems that use iterative solvers
to satisfy a set of residual equations. These methods involve rewriting the total derivative of the
function as a series of partial derivatives and linear solutions. A full derivation of the adjoint
method is shown in Section 3.3.2, but the essence of the method comes down to the following
equation:

df (x)
f (x) f (x) R 1 R
=
.
dx
x
(3.3)
In this expression, total derivatives, denoted by the d operator, involve a solution of the residual
system R, while partial derivatives, denoted by the operator are evaluated at the current value
of the states. By rewriting the total derivative as shown above, the necessity of solving the
full system of residual equations for each derivative component is eliminated, leaving instead
a set of four partial derivatives, and a series of linear solutions. The adjoint method involves
solving the set of linear systems formed by the second and third terms, which requires a linear
solution for each f . The direct method involves solving the set of linear systems formed by the
last two terms, which requires a linear solution for each design variable x. These methods can
be extremely efficient if either the number of functions, f , or the number of design variables, x,
are small. However, as should be evident even from this brief discussion, this class of methods
is significantly more complicated than either of the previous two methods. Further, this class
of methods requires access to the underlying code in order to be able to compute the necessary
partial derivatives, which are typically hand differentiated versions of the underlying evaluation
code. A small variation on these methods, called semi-analytic methods, consists in evaluating
the necessary partial derivatives using numerical techniques.
Automatic Differentiation: Automatic differentiation is a method that involves using
computational tools to perform a systematic application of the chain rule to a piece of code.
In essence, almost every computational algorithm can be broken down into a set of very simple
operations, each of which is simple enough to have a simple analytic derivative. By combining
these simple derivatives together, using the appropriate application of the chain rule, the correct
derivative can be determined for the entire algorithm [37]. If the operations in a given algorithm
are defined as:
ti = fi (t1 , t2 , . . . ti1 ) ,
i = n + 1, n + 2, . . . , m.
(3.4)
16
Then the differentiation of the algorithm, using the chain rule can be written as:
i1
X fi tk
ti
=
,
tj
tk tj
j = 1, 2, . . . , n.
(3.5)
k=1
Note that the propagation of the derivative can be performed either from the design variables
forward or from the objective function backwards, a property that this method shares with
analytic methods. Many tools exist to implement this technique in a variety of programming
languages [17, 84, 10, 38, 9]. Once again this method requires access to the underlying analysis
code.
The ADjoint sensitivity method, mentioned frequently throughout this thesis, is a combination of the last two methods. Automatic differentiation techniques are used to compute the
partial derivatives in the adjoint equations. The combination of these two techniques produces
a method which has the computational efficiency of the adjoint method, and the numerical
accuracy and relative ease of implementation that come with automatic differentiation. A more
detailed discussion of this method is presented in Section 3.3.2.
3.3
ADjoint Stability Derivative Method
The ADjoint approach to stability derivatives presented below builds on the work of Limache
and Cliff [54], and Park et al. [83, 82]. The ADjoint sensitivity method, developed by Mader et
al. [60] is applied to a moving grid CFD formulation which is equivalent to the non-inertial
formulation used by both Limache and Cliff [54] and Park et al. [83, 82] to compute the stability derivatives. As in previous work, this CFD formulation allows both the static and dynamic
derivatives to be computed from a single, steady, flow solution. The main advantage of this
approach is that it combines the computational efficiency of analytic sensitivity methods with
the relatively straightforward implementation of automatic differentiation. This has enabled
the rapid development of an efficient method for the computation of stability derivatives.
3.3.1
Theory
The stability derivative formulation described in this section is based on two key methods. The
first is a CFD code that can compute solutions for rotating geometries. This can be accomplished with either a non-inertial reference frame formulation or a moving grid formulation. The
non-inertial formulation is well described by Limache and Cliff [54], and Park and Green [82],
so no further discussion on this topic is presented here. The other formulation, the moving
grid formulation, is used in this thesis and is discussed later on in this section. The second
key method is an efficient, robust and accurate sensitivity analysis method for the CFD code.
17
For the current work, this is accomplished with the ADjoint method. A brief summary of
this method is provided in Section 3.3.2, with more details available in previous work by the
author [60].
CFD for Rotating Geometries
In order to compute derivatives with respect to both the static (, , V ) and dynamic (p, q, r)
parameters for a given configuration from a single steady-state flow solution, one needs a flow
solver that can compute steady-state solutions for constant, nonzero values of each of the
parameters. For the case of the static parameters, this is accomplished in most standard CFD
solvers. However, in the case of the dynamic parameters, this is usually not the case. The
method used to handle this motion in the following work is a moving grid formulation. The
flow solution is computed using the global velocities as the states. These velocities are expressed
in terms of the moving grid base vectors. This transformation introduces additional terms into
the equations that account for the moving coordinates of the grid, but eliminates the need to
add the centripetal and Coriolis forces as source terms to the momentum equations, as required
in the non-inertial formulation. The moving grid formulation is derived below.
Moving Grid Formulation
To begin, three velocities, u, v and w, are defined such that
u = v + w,
(3.6)
where u is the velocity of the fluid in the fixed frame, v is the velocity of the fluid with respect
to the moving grid and w is the velocity of the moving grid. Using the approach of Warsi [112],
a general form of the conservation law can be written for moving coordinates as
A
(A) w + F = C,
(3.7)
Where, for the conservation of mass, momentum and energy, F takes the form F = Au+B, and
A, B, C and can represent various quantities, depending on what quantity is being conserved.
In addition, Warsi [112] uses the identity:
(Aw) = (A) w + A ( w) ,
(3.8)
which can be derived by applying the product rule to the left hand side and rearranging the
components. This identity can then be rearranged as follows,
(A) w = (Aw) A ( w) .
(3.9)
18

Using this relationship in the conservation law (3.7), one can write
A
(Aw) + A ( w) + F = C.
(3.10)
Substituting F as defined above yields,

A
(Aw) + A ( w) + (Au + B) = C.
(3.11)
Rearranging the above equation, one obtains:

A
+ (Au Aw) + A ( w) + (B) = C.
(3.12)
Since ( w) = 0 i.e., the grid is incompressible this equation simplifies to

A
+ (Au Aw) + (B) = C.
(3.13)
Following the work of Warsi [112] and Ghosh [31], the conservation equations for mass, momentum and energy can be written as follows.
Mass Conservation:
For mass conservation, A = , B = 0, C = 0 and = t. This yields
+ (u w) + (0) = 0,
t
(3.14)
+ ((u w)) = 0.
t
(3.15)
which can be simplified to
Momentum Conservation:
In this case, A = u, B = pI , C = 0 and = t, which
yields
u
+ (u u u w) + (pI ) = 0.
t
Rearranging this equation, yields
u
+ (u u u w + pI ) = 0.
t
(3.16)
(3.17)
Up to this point, the derivation has been general. To obtain the formulation used in this work
one rewrites the equations in terms of the moving grid base vectors to obtain
ui
ai
ai + ui
+ (u u u w + pI ) = 0.
t
t
(3.18)
Since it can be shown that ai /t w/xi = 0, one can write

ui
w
ai + ui i + (u u u w + pI ) = 0.
t
x
(3.19)
Furthermore, letting w = x, it can be shown that ui w/xi = ( u) and therefore in

the moving grid coordinate system the momentum equations are
ui
ai + (u u u w + pI ) + ( u) = 0.
t
(3.20)

Energy Conservation:
19
To obtain the energy conservation equations, one sets A = et ,
B = (pI ) u + kT , C = 0 and = t, which yields

et
+ (et u et w) + ((pI ) u + kT ) = 0.
t
(3.21)
Rearranging this equation, one obtains

et
+ (et u et w + (pI ) u + kT ) = 0,
t
(3.22)
where et = p/( 1) + 1/2|u|2 .

Final Formulation:
Combining Equations (3.15), (3.20) and (3.22) yields the following set
of equations:
+ ((u w)) = 0
t
u
+ [u u u w + pI ] + [ u] = 0
t
et
+ [et u et w + pu u + kT ] = 0.
t
(3.23)
(3.24)
(3.25)
Note that the body forces have been neglected in this derivation. Inclusion of body forces is
accomplished by setting a nonzero value of C in each case.
Flux Vector Form:
Now, restricting the deriviation to just the inviscid portion of the equa-
tions, the flux vector form of the governing equations can be written as,
fi
+
+ s = 0,
t
xi
(3.26)
where xi are the coordinates in the ith direction, and the state and the fluxes for each cell are
ui wi
0
u u
u
u u w u + p
i 1
1
2 3
3 2
i 1
i1
(3.27)
s=
fi =
=
3 u1 1 u3 ,
ui u2 wi u2 + pi2 ,
u2 ,
1 u2 2 u1
ui u3 wi u3 + pi3
u3
0
et
ui (et + p) wi et
and where w = w0 + x.
This is the formulation implemented in SUmb [107] and used in this thesis.
Constant Parameter Motions
Having developed the ability to compute solutions for rotating configurations, one must now
consider how to compute the required derivatives. To accomplish this, one must develop a series
20
of motions that allow for the variation of a single parameter while holding all other parameters
constant. Consider pitch rate, q. As described by Limache and Cliff [54], a constant loop
at a given q generates a steady solution. The radius of the loop is inversely proportional to q.
Thus, as q reduces to zero, the radius approaches infinity and steady level flight is recovered.
Similar ideas can be applied to roll rate, p, and yaw rate, r. However, in these two other
cases, achieving a steady rotating flow is more complicated. When the body axis of the configuration coincides with the wind axis, the logic is the same as for the pitching motion, but if we
incline the body axes at an angle of attack, , relative to the wind, rotation about the body
yaw and roll axes no longer yields a steady state solution. In this case, rotation about the wind
axis is required to generate a steady solution. For small angles of attack, this distinction is not
of great importance. On the other hand, for large angles of attack, this should be taken into
account.
Grid Motion Considerations

One further distinction must be drawn between the moving grid formulation and the noninertial reference frame formulation. To calculate the correct derivatives, the grid motion must
be specified in such a way that the magnitude of the grid velocity is independent of the rotational
rate. More specifically, the grid motion must be specified such that the velocity of the flow at
the nominal center of rotation is that of the desired free stream velocity. This falls out naturally
with the non-inertial reference frame formulation, because the velocity is specified in terms of
the local grid. For the moving grid formulation, this is not so obvious. To implement this, one
needs to specify a grid velocity that is linked to the rotational velocity such that the velocity
of the grids center of rotation is independent of the rotational speed.
Verification of Moving Grid Formulation

To verify the implementation of the moving grid formulation, results are compared for a NACA
0012 airfoil rotating at a finite q to those produced by Limache [55]. In this comparison, inviscid
flow is simulated around a NACA 0012 airfoil at Mach = 0.2 with q = 0.0, 0.01, 0.03, 0.05. The
comparison is performed using a pseudo two-dimensional mesh, which has two cells in the
spanwise direction, with symmetry planes at the ends of the wing segment to reproduce twodimensional flow. Each slice of the mesh contains 65,536 cells for a total mesh size of 131,072
cells. Comparisons are made for the Cp distribution of the solution near the airfoil, as well for
the streamlines of relative velocity in the solution.
The Cp distributions shown in Figs. 3.1a3.1d provide visual verification that the moving grid formulation implemented in SUmb is consistent with the non-inertial reference frame
21
1
Z
pressure
0.9
0.7
0.5
0.3
0.1
0.05
0.01
-0.01
-0.05
-0.1
-0.2
-0.3
-0.4
-0.6
CoefPressure
0.9
0.7
0.5
0.3
0.1
0.05
0.01
-0.01
-0.05
-0.1
-0.2
-0.3
-0.4
-0.6
0.5
-0.5
-1
(a) q=0.0: SUmb
(b) q=0.0: Limache [55]

Y
1
Z
pressure
0.9
0.7
0.5
0.3
0.1
0.05
0.01
-0.01
-0.05
-0.1
-0.2
-0.3
-0.4
-0.6
CoefPressure
0.9
0.7
0.5
0.3
0.1
0.05
0.01
-0.01
-0.05
-0.1
-0.2
-0.3
-0.4
-0.6
0.5
-0.5
-1
(c) q=0.05: SUmb
(d) q=0.05: Limache [55]
Figure 3.1: Cp contour and streamline comparison, rotating NACA 0012: M=0.2,=0.00
22
30
20
20
Cp
10
-10
pressure
0.9
0.3
0.1
0.05
0.01
0.005
-0.005
-0.01
-0.05
-0.1
-0.3
-0.6
10
0.9
0.7
0.5
0.3
0.1
0.05
0.01
-0.01
-0.05
-0.1
-0.2
-0.3
-0.4
-0.6
-10
-20
-20
-30
-30
-20
-10
10
20
30
-30
-20
-10
10
20
(a) q=0.05: SUmb
(b) q=0.05: Limache [55]
30
Figure 3.2: Flowfield streamline comparison, rotating NACA 0012: M=0.2, =0.00
formulation used by Limache [55]. For each value of q, the Cp distribution matches the one
presented by Limache. In addition, the relative velocity contours also match. In particular,
note that for the q = 0.05 case, in Figs. 3.2a and 3.2b, the point of zero relative velocity occurs
at the expected coordinates, (0, 20), in both cases.
Mach
Coefficient
q = 0.00
q = 0.01
q = 0.03
q = 0.05
0.2
Cl
0.001 (0.000)
0.053 (0.053)
0.156 (0.157)
0.260 (0.262)
Cm
0.000 (0.000)
0.018 (0.018)
0.053 (0.053)
0.088 (0.088)
Cl
0.000 (0.000)
0.060 (0.060)
0.179 (0.180)
0.298 (0.299)
Cm
0.000 (0.000)
0.020 (0.020)
0.060 (0.060)
0.100 (0.100)
Cl
0.000 (0.000)
0.107 (0.108)
0.310 (0.316)
0.487 (0.498)
Cm
0.000 (0.000)
0.041 (0.042)
0.121 (0.124)
0.195 (0.201)
0.5
0.8
Table 3.1: Comparison of lift and moment coefficients for the NACA 0012 at = 0 for various
values of q (results from Limache [55] in parentheses)
Finally, in Table 3.1 values of Cl and Cm are compared to reference values from Limache [55].
As can be seen in this table, the two implementations match very well over a range of q values.
The largest discrepancy is 0.011 in Cl at Mach = 0.8 and q = 0.05. This close correlation is
further confirmation that the formulation is correct.
23
3.3.2
ADjoint Approach
Having modified the CFD code to handle rotating geometries, the code can now be differentiated
to get stability derivatives. To do this efficiently, the ADjoint method, previously outlined by
Mader et al. [60], is used. In this approach, automatic differentiation techniques are combined
with the adjoint method to generate the sensitivities for the CFD equations. The application
of this method to stability derivatives is as follows.
Start by considering the functions of interest, I, which may be either the forces (CL , CD , CY )
or moments (Cl , Cm , Cn ) acting on the aircraft. These are functions of both the states of the
system () and the values of the design variables (x), which in this case would be the states of
the dynamic model (, , V, p, q, r, h, i ,etc.). The function of interest can be expressed as
I = I(x, (x)).
(3.28)
The solution of the CFD equations can be represented as a set of governing equations that
are also functions of the states () and the design variables (x). These equations can be written
as
R (x, (x)) = 0.
(3.29)
To generate the adjoint equations for this system, one first writes the total derivative for
both the function of interest (3.28) and the residuals (3.29), which yields,
I
I d
dI
=
+
dx
x dx
(3.30)
dR
R R d
=
+
= 0.
dx
x
dx
(3.31)
and
Where Eq. (3.31) must equal zero, because R = 0 when the governing equations are satisfied.
Note that a distinction is made between total and partial derivatives. In the context of this
thesis, a partial derivative is defined as a derivative evaluated for a constant set of states and
a total derivative is a derivative evaluated including a solution of the governing equations to
determine a new set of states, , that satisfy R = 0. In the above equations, the derivatives
dI/ dx is the total derivative that one would obtain by performing a standard finite-difference
calculation over the entire flow solver. I/x and I/ are partial derivative vectors of size Nx
the number of design variables and NT the total number of flow states respectively.
These derivatives are evaluated for a fixed set of states, . d/ dx is another total derivative,
in this case representing the total derivative of the states with respect to the design variables.
This is a matrix term and is of size NT Nx . Similarly, in Equation (3.31), dR/ dx is the
total derivative of the residuals, including the solution of the governing equations. R/ is
simply the flux Jacobian, a partial derivative of size NT NT . R/x is the partial derivative
24
of the residuals of the governing equations with respect to the design variables, x and is of size
N T N x .
Therefore, the total derivative from Eq. (3.31) can be rewritten

d
R 1 R
=
.
dx
x
Combining Equations (3.30) and (3.32) one obtains

dI
I
I R 1 R
=
.
dx
x
x
(3.32)
(3.33)
Now, the total derivative dI/ dx is expressed in terms of four partial derivatives that do
not require a solution of the residual equations in their computation. Instead, to compute the
total derivative dI/ dx, a series of linear solutions must be performed.
In this work, the adjoint system is solved,

I
R T
= ,
(3.34)
which requires a separate linear solution for each component of I. The other option, the direct
method, involves solving:

d
R
=
,
dx
x
(3.35)
which requires a separate linear solution for each component of x. The relative efficiency of the
two approaches depends on the relative number of components in I and x. In this case, the
number of components in I is six and the number of components in x is six or more, depending
on the number of control derivatives required. Therefore, it is slightly favorable to use the
adjoint method, but the relative numbers are close enough that it is not overly important. In
this thesis, the adjoint approach is used because it has already been implemented to compute
the derivatives for design optimization [58]. In that case, the number of input variables generally
exceeds the number of output variables, making the adjoint approach the more efficient of the
two approaches.
The other consideration associated with Eq. (3.33) is how to calculate the four partial
derivatives that make up the expression. This is where automatic differentiation is used. One
of the most significant drawbacks of the adjoint method outlined above is that the calculation
of the partial derivatives making up the expression can be very complex. In many cases,
such as those involved in complex CFD schemes, the effort required to differentiate the code
used to compute the residuals is significant. By using automatic differentiation to compute
these derivatives, the amount of effort required to complete the differentiation is significantly
reduced. Also, no approximations are made in the differentiation, and as a result, the derivatives
computed with the ADjoint method are extremely accurate.
25
3.3.3
Verification of the ADjoint for Stability Derivatives
To verify the ADjoint implementation for the computation of stability derivatives, two comparisons are performed. In Table 3.2, the values of Cl , Cm , Clq and Cmq are compared
against derivatives computed using the complex-step derivative approximation [65]. Using the
complex-step approach, the sensitivity is computed by
dI(x)
Im [I(x + ih)]
=
+ O(h2 ),
dx
h
where i =
(3.36)
1. As mentioned in Section 3.2, this approximation avoids the subtractive
cancellation issues inherent in finite differences. Therefore, with a sufficiently small step size
in this case 1020 the method is able to produce derivatives with the same accuracy as
the flow solution, making it an excellent benchmark for verifying derivatives.
Derivative
ADjoint
Complex Step
Cl
7.961756758205
7.961756758114
Cm
2.068623684859
2.068623684834
Clq
11.921373826019
11.921373823280
Cmq
3.999949643166
3.999949642440
Table 3.2: Sensitivity verification: NACA 0012 test case, Mach = 0.5, 131,072 cells: 1012
relative convergence
Mach
Derivative
0.1
0.5
0.8
ADjoint
Limache and Cliff. [54]
% Difference
Clq
10.385
10.377
0.08%
Cmq
3.498
3.489
0.26%
Clq
11.921
11.847
0.62%
Cmq
4.000
3.968
0.81%
Clq
21.782
21.889
0.49%
Cmq
8.438
8.884
5.02%
Table 3.3: ADjoint stability derivatives for a NACA 0012 airfoil at = 0.0 degrees
Table 3.2 shows that the ADjoint and complex-step results match to between 9 and 11
digits. This is extremely accurate, far beyond the accuracy of the underlying physical model
used. Furthermore, given the iterative nature of the solvers used, the accuracy is consistent
with the 1012 relative convergence tolerance that was achieved.
26
In Table 3.3, the values of Cmq and Clq are compared against those computed by Limache
and Cliff [54] for the same case. This comparison is done to show that the dependencies of the
coefficients on q are accurately captured with this method. The results in this table show that
the code is very accurate relative to the reference implementation of Limache and Cliff [54]. For
the Mach = 0.1 case and the Mach = 0.5 case, the difference between the two implementations
is less than one percent. In the Mach = 0.8 case, Clq is within one percent of the reference
results, while Cmq differs by approximately five percent. This difference can be attributed
to the fact that there is a shock wave in the solution of the Mach = 0.8 case. The precise
location of the shock has a significant impact on the value of the moment coefficient and hence
the moment coefficient derivatives. Given that SUmb is a structured, multi-block code, and
that the reference results were computed with an unstructured code, slight variations in the
prediction of the shock location are not unexpected. Based on these results, it can be concluded
that, for the NACA 0012 airfoil case, the stability derivatives predicted by the code are correct.
A complete demonstration of the three dimensional capabilities of this method can be found in
Mader and Martins [59].
However, despite the efficiency of this method for computing the static and dynamic derivatives, it has two inherent drawbacks. First of all, it is based on a steady-state flow solution.
While this enhances its desireability from a performance perspective, it means that the flow
solution contains no knowledge of the time history of the flow and is therefore unable to provide
any insight towards the dot derivatives for a given configuration. Since these dot derivatives are necessary to complete the linear flight dynamic model, this is a significant drawback.
Secondly, since the adjoint method was used to generate the stability derivatives, a second
derivative method would be necessary in order to include these derivatives in an optimization.
Possible methods for achieving this with a certain amount of efficiency have been shown by
Ghate and Giles [30] and by Rumpfkeil and Mavriplis [97]. However, even these methods require the solution of an adjoint system for all of the output variables as well as a direct method
solution for each design variable in the problem. For simple problems, this would work, but for
aerodynamic shape optimization problems with large numbers of design variables, such as the
ones demonstrated in this work, the cost of this method would quickly become prohibitive.
3.4
Time-Spectral Stability Derivative Method
The time-spectral stability derivative formulation presented in the following section is similar
to the methods presented by Murman [74] and DaRonch et al. [94] and is motivated by the
complex number derivative methodology outlined by Etkin [27]. There is also a similar
method, outlined in the context of analyzing experimental results outlined in Rohlf et al. [93].
27
The time-spectral stability derivative method is essentially a forced oscillation technique. A

time-spectral CFD solver is used to generate the solution for a prescribed oscillatory motion.
This periodic solution is then used, with a linear regression technique, to generate estimates for
the functionals of interest the force or moment coefficient, the derivative of that coefficient
with respect to the oscillating parameter and the derivative of that coefficient with respect
to the time derivative of the oscillating parameter. While this technique does not have the
numerical accuracy of the previously described ADjoint method, its simple algebraic nature
allows it to be used in conjunction with an adjoint method to compute the gradients necessary
for optimization. The following section outlines the basics of the approach.
3.4.1
Time-spectral CFD
The underlying core of this approach to computing stability derivatives is the time-spectral
CFD method. This method is one of a family of methods that discretize the time derivative in
the unsteady flow equations in a spectral form for periodic problems. These methods exploit
the periodic nature of the problem by expressing the states of the system as a Fourier series in
time. The entire periodic solution can then be recovered from a small number of state instances
spanning the time period, or frequency spectrum, of the problem. These state instances can
then be solved for directly, using standard steady-state convergence acceleration techniques,
thereby eliminating the need to iterate through the startup transients typical of unsteady CFD
problems.
These spectral method computations can be performed in either the time-domain, the frequency domain, or a combination of the two. In the time-domain, the state instances represent
discrete snapshots of the solution in time, while in the frequency domain, the state instances
represent distinct frequencies present in the solution. In each case, the spectral solution is
capable of representing a fundamental frequency as well as a number of higher harmonics. The
number of resolved harmonics is related to the number of time or frequency instances present
in the solution.
Early work on time nonlinear spectral solution techniques was conducted by Hall et al. [41],
who derived a spectral formulation for the two-dimensional NavierStokes equations. This
derivation of the spectral equations is conducted in the frequency-domain. However, to facilitate
computation, the flow equations are transformed back to the time-domain. This allows a
typical, time-domain residual formulation to be used for the computation of the solution in
each of the spectral instances. This residual is augmented by a spectral term that couples the
various solution instances. In this case, while the residuals are computed in the time-domain,
both the spectral operator and the boundary conditions are applied in the frequency domain,
yielding a mixed time-domain/frequency-domain approach. In an extension of this work, Ekici
28
and Hall [24] apply this technique, known as the harmonic balance technique, to multistage
turbomachinery applications where a variety of frequencies may be present. A mixed timedomain/frequency domain approach is also used in this extension.
Another nonlinear spectral solution technique is the nonlinear frequency domain (NLFD)
method developed by McMullen et al. [69, 68, 70]. In this technique, the solution process takes
place primarily in the frequency domain. The states of the system are stored as frequency
domain Fourier coefficients and the solution steps are generated from the frequency domain
residual and spectral operator. To simplify the implementation, the residual is evaluated in the
time domain, where the states are transformed from the frequency domain to the time domain.
Then, the residual is transformed from the time-domain to the frequency domain using fast
Fourier transform (FFT) techniques.
The time-spectral method introduced by Gopinath and Jameson [34, 35] is similar to the
harmonic balance method of Hall et al. [41]. However, the time spectral method is derived
completely in the time domain. This yields a purely real spectral operator, and allows for the
use of the time-domain residual operator in its original form, including boundary conditions.
The time-spectral method is the method that has been used in this thesis. While the original
derivation of this method is due to Gopinath [35], the basics of the method have been included
here to put the remainder of the discussion in context.
To derive the time-spectral equations, one starts by writing the governing equations for
unsteady flow. These can be written as:
V
fi
+
= 0,
t
xi
(3.37)
where xi are the coordinates in the ith direction. Once again, as in Section 3.3, grid motion
will need to be modeled, therefore the moving grid formulation, derived before, is used again
here. However, in this case, since the motions are more general and not intended to produce
a steady solution, the general form of the momentum equations shown in Equation (3.17) is
used. Based on these equations, for inviscid flow, the states and the fluxes for each cell are:
ui wi
u u w u + p
u
i 1
1
i 1
i1
(3.38)
fi = ui u2 wi u2 + pi2
cell = u2 ,
.
ui u3 wi u3 + pi3
u3
ui (et + p) wi et
et
Writing this equation in a concise semi-discrete form yields,
V
d
+ R() = 0.
dt
(3.39)
29
where R represents the spatially discretized residual operator implemented in the flow solver.
For SUmb [107], the flow solver used in this thesis, this is a second order, cell-centered finitevolume scheme. This operator includes all of the boundary conditions and artificial dissipation
operators in the flow solver.
Now, the spectral time-derivative operator can be derived as follows:
One assumes that the states of the system, , can be expressed as a Fourier series. Thus, a
Fourier transform of the system states can be written as,
N 1
1 X n ikxn
k =
e
,
N
(3.40)
n=0
and the inverse transform of the states can be written as:

N 1
2
n =
k eikxn .
(3.41)
k= N 21
Note that in Equations (3.40) and (3.41) there are two representations of the state vector, and
These vectors represent the time-domain and frequency domain representations of the state
.
vector respectively. The time interval for the series is xn = 2n/N , where N is the number
of time intervals, and n is the index of the current time interval. In the frequency domain, k
represents the frequency component index of the state vector.
Combining Equations (3.40) and (3.41) to express n explicitly in the time domain yields:
N 1
2
k= N 21
N 1
1 X l ikxl ikxn
.
e
e
N
(3.42)
l=0
Rearranging slightly gives:

N 1
1 X l
n =
N
l=0
N 1
2
eikxl eikxn ,
(3.43)
k= N 21
or, letting xln = xn xl :

N 1
1 X l
n
=
N
l=0
N 1
2
eikxln .
(3.44)
1 eiN xln
,
1 eixln
(3.45)
k= N 21
Considering the inner sum as a geometric series yields:

N 1
2
X
k= N 21
eikxln = ei
N 1
xln
2
30
which, after some algebraic manipulation, can be expressed as:

N 1
2
eikxln =
k= N 21
sin( N x2ln )
.
sin( x2ln )
(3.46)
Now, returning to Equation (3.44),one gets:

N 1
1 X l sin( N x2ln )
=
.
N
sin( x2ln )
n
(3.47)
l=0
Noting that xln is a function of time, Equation (3.47) can be differentiated with respect to time
to give:
"
#
N 1
1 X l N cos( N2 xln ) sin( N x2ln ) cos( x2ln ) dxln
Dt =
N
2 sin( x2ln )
dtln
2 sin2 ( x2ln )
l=0
n
(3.48)
since N xln /2 is an integer multiple of ,

N xln
)=0
2
(3.49)
N xln
) = (1)(nl) .
2
(3.50)
sin(
and
cos(
Using these relationships in Equation (3.48),

"
#
N
1
(nl)
X
1
N
(1)
dxln
Dt n =
l
.
xln
N
2 sin( 2 )
dtln
(3.51)
l=0
Now consider the derivative dxln / dtln from Equation (3.51). Substituting in the value of xln
and evaluating gives
xln =
2(n l)
2 T (n l)
2
2
2
=
=
t(n l) =
(tn tl ) =
tln .
N
T
N
T
T
T
(3.52)
Therefore, the derivative of xln with respect to time is

dxln
2
=
dtln
T
and this relationship can now be used in Equation (3.51) to get
"
#
N
1
(nl)
X
(1)
Dt n =
l
.
T
sin( x2ln )
(3.53)
(3.54)
l=0
Finally, this expression can be simplified as

Dt n =
N 1
X
dln l
T
l=0
(3.55)
31

where dln is a matrix operator expressed as
(nl)
(1)
if l 6= n
(nl)
sin( N )
dln =
0 if l = n
(3.56)
Thus, Dt is an operator that spans all of the time instances in the solution. By solving the N
coupled time instances represented in the equation,
V Dt n + R( n ) = 0,
(3.57)
where n represents each of the N time instances, one ends up with a coupled set of solutions
that represent the periodic, steady-state solution to the given problem.
Note that while the Dt n operator can be thought of as a matrix vector product, it is
implemented on a cell by cell basis. Thus, for a single state in a single cell, the matrix operator
can be expressed as
d1,0
d0,1
0
Dt = .
.
..
..
..
.
d0,N 1 d1,N 1
dN 1,0
dN 1,1
..
.
(3.58)
This operator is applied across the spectral instances of each state on a cell by cell basis to
form a time-spectral residual
RT S = V Dt n + R( n ) = 0.
(3.59)
This time-spectral residual is then iterated to convergence using a pseudo-time approach

V
wn
+ RT S = 0.
t
(3.60)
Where the pseudo-time iterations are solved using a mixed Runge-Kutta, Newton-Krylov approach. The startup iterations are solved using a full multigrid, five stage Runge-Kutta method.
After two or three orders of magnitude of convergence, the iterations switch over to a preconditioned Newton-Krylov approach. The configuration of the preconditioner for this solver is the
same as described for the ADjoint in Chapter 4.
3.4.2
Linearized Aerodynamic Forces
Having described the time-spectral formulation for the flow solver, the time-spectral approach
to the computation of stability derivatives can now be outlined. To start, consider the linear air
reaction theory outlined by Etkin [27]. This theory states that, for a general motion, the force
32
and moment coefficients of an aircraft for example, the lift coefficient can be approximated
as:
CL = CL0 + CL + CL + CL
+ ... + CL + ...,
(3.61)
where all motion states , , V , p, q, r, ze and their associated time derivatives are included.
CL0 is the value of the coefficient for the steady-state reference flight condition about which
the motion occurs. However, if a simple motion is specified that consists of a single dynamic
state for example the value of the remaining motion states are zero and they, and their
associated derivatives, drop out of the equation. In the case of a pure motion this leads to:
CL CL0 + CL + CL + CL
+ ....
(3.62)
One can then make the further assumption, justified further on in this section, that the higher
order derivatives are small and neglect them, which gives:
CL CL0 + CL + CL ,
(3.63)
Ci Ci0 + Cij j + Cij j.
(3.64)
or in a more general form:
where i = L, D, Fy , Mx , Mz , My and j = , , V, p, q, r. Based on the simplified Equation (3.64),

one is now left with a simple equation with three unknowns, Ci0 , Cij , and Cij .
Diagrams illustrating pure and q motions are shown in Figures 3.3 and 3.4, respectively.
In these diagrams, the grid is moving with respect to a stationary air mass and the arrows in
the diagrams indicate the relative motion of the incoming flow with respect to the grid. The
motion, shown in Figure 3.3, is an oscillating vertical motion with the forward grid velocity
modified in sync with the oscillations such that the incoming velocity maintains a constant
magnitude, but changes direction. Time instances 2 and 3 are at the top and bottom of the
oscillations, respectively, illustrating the points in the cycle where the grid experiences pure
forward motion. Time instance 1 illustrates a point midway through the oscillation where the
grid is experiencing both forward and vertical motion. Note that it is important that the change
in angle of attack comes from the change in grid motion, not a change in the farfield boundary
conditions. This eliminates the phase lag associated with the change in the flow propagating
from the farfield into the aircraft before its effects are felt. The diagram in Figure 3.4 illustrates
a pure q motion. In this case, the grid rotates with an oscillating value of q and the magnitude
and direction of the velocity are modified such that the incoming is constant.
When one of these pure motions is simulated with the time-spectral method, the result is a
solution that consists of force coefficient (CL ,CD ,Cy ) and moment coefficient (Cl ,Cm ,Cn ) values
at N time instances in a periodic solution. A set of solutions corresponding to a pure motion
33
Figure 3.3: Pure angle of attack motion
Figure 3.4: Pure pitching motion
6.0e-02
CL
CL
CL
:
:
4.0e-02
: 3 Point Solution
: 5 Point Solution
3 Point Solution
5 Point Solution
Value
2.0e-02
0.0e+00
-2.0e-02
-4.0e-02
-6.0e-02
0.0e+00
5.0e-01
1.0e+00
Time (s)
1.5e+00
2.0e+00
Figure 3.5: Time-spectral solution: motion: Mach = 0.1
34
6.0e-02
3 point Solution
5 point Solution
7 point Solution
Derivative estimate
solution
4.0e-02
CL
2.0e-02
0.0e+00
-2.0e-02
-4.0e-02
-6.0e-02
-1.0e-02
-5.0e-03
0.0e+00
(rad/s)
5.0e-03
1.0e-02
Figure 3.6: Time-spectral solution: CL vs : Mach = 0.1

are shown in Figure 3.5. To compute stability derivatives from this solution, one computes a
linear least-squares fit of the output coefficient (e.g., CL ) with respect to the primary motion
variable from the solution (e.g., ). The slope of the line resulting from this fit is the stability
derivative (CL ) while the y-intercept of the line is the value of the coefficient (CL ) at the zero
value if the motion perturbation = 0. To demonstrate this process, Figure 3.6 shows a
time-spectral solution plotted against the corresponding motion. As can be seen from the
figure, the general trend of the relationship between the coefficient and the motion parameter
is linear. The linear regression generates the line shown in the figure which quantifies the slope
and intercept of the trend. These two values represent the steady-state value of the coefficient
and its derivative with repect to the motion variable. For example,
y = Cij x + Ci0 .
(3.65)
However, there is also a distinct hysteresis in the solution, as demonstrated by the gap between
the coefficient on the upstroke and downstroke of the oscillation. This hysterisis relates to
the dot derivatives. As discussed by Etkin [27] with respect to the derivatives, the dot
derivatives represent the time lag in the development of the coefficient resulting from a change
in the flow. To estimate this quantity, one subtracts the value of the linear regression line from
the solution. This process eliminates the bulk dependence of the solution on the main motion
35
1.0e-02
3 point solution
5 point solution
7 point solution
Derivative Estimate
Residual CL
5.0e-03
0.0e+00
-5.0e-03
-1.0e-02
-3.0e-02
-2.0e-02
-1.0e-02
0.0e+00
(rad/s2 )
1.0e-02
2.0e-02
3.0e-02
Figure 3.7: Time spectral solution: CL hysteresis vs :

Mach = 0.5
variable, leaving just the variation associated with the hysteresis. This yields another strongly
linear relationship, as shown in Figure 3.7. Taking the slope of this trend yields the value of
Ci . Further, as this plot shows, the linear approximation is a very good representation of the
trend, indicating that any dependence on the higher order time derivatives is small. This is
sufficient to justify the assumption made for Equation (3.63).
Note that in Figure 3.6, all three solutions 3, 5 and 7 time instance produce the same
elliptical solution. This indicates that for the small amplitude (< 0.5 deg oscillations), inviscid
flow solutions used in this work three time instances are sufficient to accurately represent the
periodic solution. Therefore all of the subsequent time-spectral results in this thesis are base
on three time instance solutions.
Also note that, generally speaking, linear regressions will be biased towards large clusters
of data. Thus, if a random set of points is selected on the ellipse shown in Figure 3.6, the linear
regression analysis would produce inconsistent results. However, because the time-spectral CFD
approach generates solution points that are equally spaced around the ellipse, the solution turns
out to be independent of which three evenly space points are chosen. Data demonstrating this
fact is shown in Table 3.4.
36
Time Instance
Set 1
Set 2
Set 3
0.0
0.007558
-0.007558
Cl
-0.009895
0.050868
-0.043125
0.005130
0.003550
-0.008679
Cl
0.023756
0.029739
-0.055646
0.008300
-0.001814
-0.006485
Cl
0.048058
-0.003023
-0.047187
Cl0
Cl
6.218
-0.000717
6.218
-0.000717
6.218
-0.000717
Table 3.4: Effect of point location on time-spectral derivative location
3.4.3
Verification and Validation
To verify and validate this approach, two test cases will be examined. The first, a NACA 0012
airfoil undergoing an oscillating plunging motion, will be compared against a theoretical thin
airfoil theory result. The second, the Stability And Control CONfiguration (SACCON) UCAV,
will be compared against wind tunnel results.
NACA 0012: Test Case
The NACA 0012 case is a pseudo three-dimensional case. The meshes used are two cells thick,
with symmetry planes on both sides to create two dimensional flow. The mesh is a C-mesh
topology split into 16 blocks for efficient parallel balancing. Samples of the meshes are shown
in Figure 3.8. To assess the numerical accuracy of the solutions, a mesh convergence study
was conducted with meshes containing 4,096, 16,384, and 65,536 cells per slice. Plots of these
results are shown in Figure 3.9. The values of the coefficients (CL0 , Cm0 ) and their derivatives
with respect to show excellent numerical accuracy, with relative errors less than 1% on the
finest mesh for both sets of coefficients and derivatives. The accuracy of the derivatives with
respect to are not quite as good, with relative errors on the order of 5% on the finest mesh.
However, that level of error is acceptable for the purposes of this work.
NACA 0012: Validation
Etkin [27] presents theoretical results for an oscillating plunging airfoil based on the work of
Theodorsen. Using the Theodorsen function as a basis, Etkin presents the theoretical values of
the lift and moment coefficients as:
CL = 2F (k)
(3.66)
37
X
Z
(a) Mesh topology
(b) Near field mesh: 65536 cells per slice
Figure 3.8: NACA 0012 meshes
100
101
CL/CL
CL /CL
CL /CL
101
Cm/Cm
Cm /Cm
Cm /Cm
100
101
10
102
103
103
104 3
10
102
N 1/2
(a) CL convergence
101
104 3
10
102
N 1/2
(b) Cm convergence
Figure 3.9: NACA 0012 mesh convergence
101

G(k)
k
1
Cm = 2F (k)(h )
4
G(k)
1
1
(h )
= (h ) + 2
2
k
4
CL = + 2
Cm
38
(3.67)
(3.68)
(3.69)
where F (k) and G(k) are the real and imaginary parts of the Theodorsen function,
C(k) = F (k) + iG(k)
(3.70)
Since the theoretical result varies with frequency, the following comparison is performed over
a variety of reduced frequencies. In addition, because the theoretical result is for an infinitely
thin airfoil, numerical results are computed for the full thickness airfoil as well as half and
quarter thickness airfoils. As shown in Figure 3.10, the numerical results match the theoretical
results well over the full range of reduced frequencies. There is a small offset between the
numerical results and the theoretical results. However, for the derivatives, this discrepancy
reduces significantly for the half and quarter thickness airfoils, indicating that the bulk of this
discrepancy for the derivatives comes from the thin airfoil theory assumptions of the reference
results. For the derivatives, the trend is less conclusive. At lower frequencies, the thin
results match the theoretical result more closely than the full thickness results, while at higher
frequencies the offset is larger for the thin airfoils than the thick airfoils. However, the shape
of the curve for the thinner results matches the theoretical result more closely than the shape
of the thicker airfoils. Note that the theoretical results for the derivatives become undefined
as the reduced frequency tends to zero. This singularity is clearly apparent in Equations (3.67)
and (3.69), and prevents the calculation of a definitive value for the derivative. However, given
the excellent agreement between the theoretical and numerical results, this comparison confirms
that the method is valid for this simple two-dimensional case.
SACCON UCAV: Test Case
The second test case considered is a full, flying-wing, UCAV configuration. The mesh is constructed with an H-H topology and 32 blocks. The geometry of the case is based on the
information provided in Sch
utte et al. [99] and the surfaces of the wing are modeled as inviscid
walls with a symmetry plane imposed at the root. The off wall spacing for the 1,179,000 cell
mesh is 1 103 root chords. Sample meshes are shown in Figure 3.11
Once again, solutions were computed on a series of meshes to assess numerical accuracy.
Specifically, meshes with 147,000, 1,179,000 and 9,430,000 cells were used to compute the solutions. Assessments of the error associated with CL and its derivatives are shown for both an
motion and a q motion in Figure 3.12.
39
6.5
Theoretical
Time-Spectral:Full Thickness
Time-Spectral:Half Thickness
Time-Spectral:Quarter Thickness
6.0
5.0
10
C L
CL
5.5
4.5
15
4.0
20
3.5
25
3.0
0.0
0.2
0.4
0.6
Reduced Frequency (k)
0.8
1.0
30
0.0
(a) M=0.1, =0.00: CL
Theoretical
0.2
0.4
0.6
0.8
1.0
(b) M=0.1, =0.00, CL
6
Theoretical
5
1.0
Cm
C m
3
1.2
1
1.4
0
Theoretical
1.6
0.0
0.2
0.4
0.6
(c) M=0.1, =0.00, Cm
0.8
1.0
2
0.0
0.2
0.4
0.6
0.8
(d) M=0.1, =0.00, Cm
Figure 3.10: Plunging NACA 0012: time-spectral stability derivative verification
1.0
40
(a) Surface mesh: 147,000 cell mesh
(b) Near field mesh: 147,000 cell mesh
Figure 3.11: SACCON meshes
101
100
CL/CL
CL /CL
CL /CL
100
101
101
102
102
103
103 3
10
102
1/N 1/3
101
(a) CL and derivatives: motion
CL/CL
CLq /CLq
104 3
10
102
1/N 1/3
(b) CL and derivatives: q motion
Figure 3.12: SACCON mesh convergence
101
41
The meshes, again, show useful mesh convergence as the grids are refined. The numerical
accuracy of CL and CL is better than 1% for the motion on the finest mesh. The value of
CL for the q motion shows similar accuracy, better than 1%, while the value of CLq is slightly
less accurate, on the order of 5% for the finest mesh.
SACCON UCAV: Validation

Using the 1,179,000 cell grid, a series of comparisons were run against derivatives computed from
experimental data by Rohlf et al. [93]. The goal of this comparison is to place the results from
the time-spectral computation in context with physical, real world results, thereby validating
the method. The flow conditions used are shown in Table 3.5. These conditions are taken
from Tormalm and Schmidt [105] and are intended to match the conditions of the experimental
results from Loeser et al. [56]. The results of these comparisons are show in Figures 3.13
and 3.14.
Parameter
Value
Static pressure (Pa)
97767
Static temperature (K)
291.0
Density (kg/m3 )
1.185
Mach number
0.149
0 - 10 deg.
Half body reference area (m)
0.385
Reference half span (m)
0.769
Reference chord (m)

Frequency
Reference moment center (m)
Reference rotational center (m)
0.479
6.28 rad/s (1 Hz)
0.6
0.8554
Table 3.5: SACCON test case conditions
The derivatives in Figures 3.13 and 3.14 match the experimental results surprisingly well,
especially considering that the numerical results are computed using an Euler solver. The
results capture the correct order of magnitude for each of the derivatives. As one might expect,
the CL derivatives are the most accurate, giving results within 10% of the experimental values.
The CLq+ and Cmq+ derivatives have errors of 30% and 40% respectively. Given the lack of
viscous effects in the numerical solution, this is amount of error is unsurprising. The error in
Cm is on the order of 80%. This is again most likely due to the missing physics in the Euler
model used for the computation. This is consistent with the accuracy of the coefficients CL
and Cm shown in Figure 3.15 where the lift prediction is very good and the moment
42
3.0
0.0
2.5
0.5
2.0
1.0
1.5
1.5
1.0
2.0
2.5
0.5
CL Exp.
CL TS
0.0
CLq+ Exp.
CLq+ TS
10
3.0
(a) CL derivatives
10
(b) CLq+ derivatives
Figure 3.13: SACCON UCAV: CL derivatives
Cm Exp.
Cm TS
0.0
Cmq+ Exp.
Cmq+ TS
0.3
0.5
0.2
1.0
0.1
0.0
1.5
0.1
0.2
(a) Cm derivatives
10
2.0
(b) Cmq+ derivatives
Figure 3.14: SACCON UCAV: Cm derivatives
10
43
0.5
0.045
CL Exp.
CL TS
CL TS q
0.4
Cm Exp.
Cm TS
Cm TS q
0.040
0.035
0.3
0.030
0.2
0.025
0.1
0.020
0.0
0.1
0.015
10
(a) CL
0.010
10
(b) Cm
Figure 3.15: SACCON UCAV: CL and Cm comparison

prediction is poor. That said, the values predicted by the method are accurate enough to give
a sense of what the stability characteristics of the aircraft are and are sufficiently accurate to
be useful in demonstrating the effect of stability constraints on an aerodynamic optimization.
SACCON UCAV: Frequency Dependence
As a final consideration, one needs to consider the frequency dependence of the derivatives. As
was shown in the two dimensional test case, there can be a significant frequency dependence
in the derivatives. To examine these effects in the three-dimensional case, a frequency sweep
was conducted on the 1,179,000 cell mesh, the results of which are shown in Figures 3.16
through 3.18.
Considering Figure 3.16 first, one can see that while there is a definite frequency dependence
in the derivatives, the values tend to a constant value as the frequency is reduced. The same
can not be said for the and q derivatives. As shown in Figures 3.17 and 3.18 both of these
quantities vary continuously as the frequency is varied. This variation is most pronounced in
the CL derivative, which varies almost linearly over the range of frequencies tested, although
there does appear to be a slight flattening of the trend at the lower end of the frequency range.
The remainder of the and q derivatives are relatively constant over a range of frequencies
between k = 0.03 and k = 0.3. Therefore, frequencies in this range should be used for the
computation of derivatives using this method unless other considerations dictate otherwise. Of
note, the 1 Hz frequency (k = 0.03) used in the experimental comparisons falls at the lower end
of this range.
44
0.20
3.0
2.5
0.15
2.0
0.10
1.5
1.0
0.05
0.5
CL TS
CL Steady
0.0
0.00
0.05
0.10
0.15
k
0.20
0.25
0.30
0.00
0.00
0.05
0.10
(a) CL
0.15
k
0.20
0.25
0.30
0.25
0.30
(b) Cm
Figure 3.16: SACCON UCAV: derivatives: frequency comparison
1.0
1.0
0.5
0.5
0.0
0.0
0.5
0.5
1.0
0.00
0.05
0.10
0.15
k
(a) CL
0.20
0.25
0.30
1.0
0.00
0.05
0.10
0.15
k
0.20
(b) Cm
Figure 3.17: SACCON UCAV: derivatives: frequency comparison
45
0.0
0.0
CLq TS
CLq Steady
0.2
0.4
0.4
0.6
0.6
0.8
0.8
1.0
1.0
1.2
1.2
1.4
1.4
0.00
0.05
0.10
0.15
k
(a) CLq
0.20
0.25
Cmq TS
Cmq Steady
0.2
0.30
0.00
0.05
0.10
0.15
k
0.20
(b) Cmq
Figure 3.18: SACCON UCAV: q and derivatives: frequency comparison
0.25
0.30
Chapter 4
Time-Spectral ADjoint Method

The ultimate goal of this work is to be able to perform gradient-based optimization of a given
aircraft configuration with stability considerations included. In the previous chapter, a timespectral method for the computation of stability derivatives was presented. This method is
capable of computing all of the necessary stability derivatives for the longitudinal stability
constraints described in Chapter 5. However, as time-spectral CFD solutions are relatively
expensive to compute, there is a need for an efficient method for computing the gradient of
the CFD solution with respect to the large number of design variables typical of aerodynamic
shape optimization. In the field of steady-state aerodynamic optimization, adjoint methods
are known to provide sensitivities at a cost that is practically independent of the number of
design variables. Based on this known characteristic, the ADjoint method of Mader et al. [60]
is extended to compute the time-spectral adjoint of the aerodynamic parameters needed for the
optimizations. The details of this computation are presented in this chapter.
4.1
Background
Adjoint methods have become well known in the aerodynamic shape optimization research
community. The method was introduced in the context of fluid dynamics by Pironneau [89]
in the mid 1970s, where it was used to demonstrate how to minimize the drag over bodies
immersed in laminar viscous flows. Angrand [4] introduced the use of an adjoint for the optimization of airfoils in potential flow in the early 1980s. It was then applied in the context of
aerodynamic shape optimization of airfoils and wings with the Euler equations by Jameson [46]
in the late 1980s. Since then it has been applied to the optimization of airfoils including
viscous effects [78, 3, 101], laminar-turbulent transition prediction [23] and multi-point optimizations [101, 79] as well as three-dimensional wings for inviscid flows [48, 44], viscous
flows [47, 81] and multi-point problems [16] and full wing-body configurations [91, 92]. It has
46
Chapter 4. Time-Spectral ADjoint Method
47
also been applied to noise related applications such as sonic-boom reduction [77], hypersonic
flows including magneto-hydrodynamics [63, 62] and has been generalized to multidisciplinary
systems by Martins et al. [64, 66]. In each of these cases, the adjoint method allowed for
the efficient optimization of the design in question with respect to a large numbers of design
variables.
However, while the use of adjoint techniques has become relatively commonplace in the context of steady-state optimizations at least in the research community its use is still relatively uncommon in time-dependent problems. Examples in two-dimensions have been demonstrated by Rumphkeil and Zingg [98], Nadarajah and Jameson [75], Mani and Mavriplis [61],
and Wang et al. [111] and a three-dimensional example has been demonstrated by Mavriplis [67].
These adjoint methods are a significant improvement over finite difference methods, but they
still have a high computational cost. The unsteady adjoint computation requires a reverse
integration in time from the final solution back to the initial condition [61, 75]. Thus, a time
dependent adjoint requires the full forward solution of the unsteady problem, storing the flow
states for each time step along the way, followed by a reverse sweep of the solution process
to find the adjoint solution. While this process is still more efficient than computing a full
unsteady solution for every design variable in the problem, it is still an expensive proposition.
Various methods for reducing the computational resources required for the computation have
been suggested; for example, writing the solution history to disk rather than storing the solution in memory [61] or evaluating only a periodic portion of the time history for the adjoint
problem [75, 98]. However, even with these additions, the computational cost of full unsteady
adjoint methods is still significant.
In the same way that the time-spectral CFD method has a lower computational cost than a
full, unsteady flow solution for periodic problems, the time-spectral adjoint method offers a lower
computational cost than an equivalent unsteady adjoint approach for the same problem. Just
as the spectral solution technique modifies a single unsteady CFD problem into a set of coupled
steady CFD problems, the time-spectral adjoint technique converts a full unsteady adjoint
problem into a single, large, steady adjoint problem. Coupling this with the efficient solution
of large, sparse linear systems provided by modern software packages such as PETSc [6]
allows for the rapid implementation of an adjoint technique for periodic unsteady problems.
It is worth noting that these potential advantages apply to all spectral methods, not just
the time-spectral method. To date, adjoint methods have been developed for each of the
spectral methods mentioned in Section 3.4.1. An adjoint for the two dimensional, viscous
harmonic balance equations was developed by Thomas et al. [104]. In this work, the authors
use a combination of forward and reverse mode automatic differentiation to generate the terms
necessary for the adjoint. Further, they use the adjoint to compute mesh sensitivities for
48
an airfoil and verify their implementation using finite-difference sensitivities as a benchmark.

An adjoint implementation was also developed for the NLFD equations by Nadarajah and
Jameson [76]. In this work, Nadarajah and Jameson use analytic techniques to derive a discrete
adjoint operator for the NLFD solver. The resulting adjoint technique is used to perform the
optimization of an oscillating transonic wing.
In direct relevance to this work, an adjoint implementation for the time-spectral equations
was published by Choi et al. [20]. In this work, a manually-coded adjoint method was applied
to a time-spectral flow solver and used to calculate the gradients required for a helicopter rotor
blade optimization. The method was successful in improving the blades, but the adjoint implementation did not achieve the full numerical accuracy that is theoretically possible with a
discrete adjoint method. This limited accuracy was due to approximations made in differentiating the routines related to the spectral radius and artificial dissipation. In the current work,
the ADjoint approach of Mader et al. [60] is used to generate the adjoint terms. This approach
eliminates the need for these approximations and allows for numerically exact derivatives. Note
that the use of the time-spectral method, which consists of all real number evaluations, is particularly advantageous in the context of this work, as it has allowed for the verification of the
ADjoint sensitivities against those computed using the complex-step method [65]. This verification is not possible for codes that use frequency domain analysis, since they use complex
arithmetic in the solution process.
4.2
Theory
To derive the time-spectral adjoint, as with the steady case, one starts by writing the vectorvalued function of interest, I, as:
I = I(x, n (x)),
(4.1)
where x represents the vector of design variables and n is the state variable vector for the
nth time instance and n = 1, ..., N , with N representing the number of time instances in the
solution.
For a given vector x, the solution of the governing equations of the system yields a vector
n.
As shown in Section 3.3.2, when deriving the adjoint equations, one generally expresses the
governing equations as:

R (x, (x)) = 0.
(4.2)
However, since the flow solution is time-spectral in nature, from van der Weide et al. [108], the
governing equations are redefined as:
RT S = V Dt n + R (x, n (x)) = 0,
(4.3)
49
where R (x, n (x)) is a normal steady-state residual for the nth time instance, n = 1, ..., N , and
Dt is the spectral operator defined in Equation (3.55). This modified residual:
RT S (x, n (x)) = 0,
(4.4)
can now be treated in the same fashion as the steady-state residual would be treated in a
normal adjoint formulation. Thus, as before, the total sensitivity of the vector-valued function
of interest, I, is written as:
dI
I
I d n
=
+ n
.
dx
x dx
(4.5)
Once again, because the governing equations must always be satisfied at a converged solution, the total derivative of the residuals in Equation (4.4) with respect to any design variable
must also be zero. Writing the total derivative of the governing equations with respect to the
design variables gives:
dRT S
RT S
RT S d n
=
+
= 0.
dx
x
n dx
(4.6)
This expression provides the means of eliminating the total derivative d n / dx from the total
sensitivity computation for I. Moving the first term of this equation to the right-hand side
gives:
RT S d n
RT S
=
.
n
dx
x
(4.7)
Substituting the solution of this system into Equation (4.5) yields:

dI
I
I RT S 1 RT S
=
.
dx
x n n
x
(4.8)
The adjoint approach consists in factorizing the RT S / n matrix with the term to its left,
yielding the adjoint system:

RT S
T
=
I
.
(4.9)
This solution is then used in Equation (4.8) to obtain the total sensitivity:
dI
I
RT S
=
T
.
dx
x
x
(4.10)
Just as with the steady-state case, this yields a sensitivity method with a computational cost
that is essentially independent of the number of design variables. Note that because the timespectral system is N times the size of the steady-state system, the adjoint system is also N
times larger than the equivalent steady-state system.
50
4.3
Implementation
Having discussed the theory behind the time-spectral adjoint, it is now possible to discuss the
details of the implementation. Much of the implementation is the same between the steadystate and time-spectral adjoint methods, therefore the implementation of the two methods will
be discussed at the same time.
4.3.1
Single-Cell Routine
The basis for the residual derivatives in the ADjoint approach is a single cell residual routine,
developed from the original residual routines. This single cell residual routine containes all
of the functionality of the original block based routines including dissipation terms and
boundary conditions but is designed to operate on a 5 5 5 cell cube, which is the smallest
block of cells that encloses the second order inviscid flux stencil, shown in Figure 4.1. This
2.5
2
1.5
1
0.5
0
0.5
1
1.5
2
2.5
3
j
i
Figure 4.1: Single cell stencil

single cell residual routine is generated by cutting and pasting the residual computation from
the original residual code, modifying the indices to limit the evaluation to a single stencil. This
process, while not ideal, is less onerous than coding a derivative scheme from scratch. Further,
by reducing the residual computation to that associated with a single cell at a time, the size
of the differentiation problem is reduced significantly, making reverse mode differentiation less
memory intensive. The ability to use the reverse mode is advantageous in this context, as is
shown in Section 4.3.2, because of the large ratio of inputs to outputs in the residual stencil.
An outline of the single cell routine used in this work is shown in Figure 4.2. Note that the
routine includes several boundary conditions, multiple dissipation schemes, the metric terms
51
necessary for the mesh derivatives, as well as the loops required for the time-spectral implementation. Figure 4.2 also shows the locations of the time instance loops needed to compute the
Outer Time Spectral Loop
ComputeResidual
xHalo
AdjustInflowAngle
CheckInputParameters
Metric
VolPym
GridVelocitiesFineLevel
Inner Time Spectral Loop
ReferenceState
SetFlowInfinityState
NormalVelocitiesAllLevels
ComputePressure
BC Symmetry
ApplyAllBCs
BC Farfield
TimeStep
BC EulerWall
Residual
Inviscid Central Flux
Inviscid Scalar Dissipation Flux
InitializeResidual
Inviscid Upwind Flux

Left/Right State
VolumeNormalization
Riemann Flux
Note: Only one of either the scalar dissipation flux or
the upwind flux is computed for a given case.
Figure 4.2: Call-graph: current residual computation: steady and time-spectral cases
residual for all time instances in the solution. The actual spectral operator from Equation (3.55)
is implemented inside the initializeResidual subroutine.
4.3.2
R/ and R/x
All of the derivatives used in the ADjoint formulation are computed using reverse mode AD.
More specifically, the Tapenade [84, 42] AD tool is used to perform reverse mode source transformation AD on the necessary routines. For the residual derivatives R/ and R/x
reverse mode differentiation is performed on the single cell residual routine described above.
To properly understand this choice, it is necessary to discuss, in some detail, the structure of
the single cell residual routine.
For the steady case, the residual computation shown in Figure 4.2 computes the value of the
residual in a single cell. This evaluation produces a result of length N the number of states
per cell one value for each of the governing equations in that cell. The number of inputs
required to generate that result is much higher than this. The variables of interest here are the
system states, , and the mesh coordinates, x. For the second order, finite volume discretization
of the Euler equations used here, this requires the states in the nearest neighbor cells for the
central flux and the nearest neighbor states, as well as the next nearest neighbor states, for the
52
dissipation fluxes. Therefore a total of 13 N states are required for the evaluation of the
residual. The residual evaluation also depends on the coordinate locations of the corner nodes
of each of the nearest neighbor cells. This includes a total of 32 3 independent spatial degrees
of freedom. Therefore, for the five states in the Euler equations, we have an overall ratio of
13 5 + 32 3 = 161 : 5 more than 32 : 1 input variables to output variables. Even
considering the fact that a single reverse mode calculation is about 4.5 times more costly than
the equivalent forward evaluation of the single cell routine, the reverse mode calculation is still
about seven times faster than a forward computation for the single cell routines.
In the time-spectral case, it is necessary to consider the temporal dependence of the operator
in addition to the its spatial dependence. In describing this dependence, this discussion refers
to both on-time-instance states and off-time-instance states. On-time-instance states are those
states that exist in the same time instance as the current residual evaluation. Off-time-instance
states are those states that exist on a time instance other than the one associated with the
current residual evaluation. In the case of a time-spectral solution, the residual is dependent
on the same on-time-instance states and coordinates as described above. In addition to that, it
is dependent on the states and coordinates of the current cell on each of the off-time-instances.
Thus, there are now N N residual values and (13 N + 32 3)N +(N + 8 3)(N 1)
input states and coordinates. This leads to a ratio of 36:1 input variables to output variables
for the Euler equations with three time instances, a ratio that increases as the number of time
instances increases.
Note that both of these ratios include the state variable derivatives of R/ n as well as the
coordinate derivatives of R/x. This fact is made possible because reverse mode AD has been
for the derivative calculation. The reverse accumulation, shown in Equation (3.5), allows the
computation to start with a single residual and, accumulating backwards through the routines,
calculate the derivative of all of the inputs at once. This turns out to be a significant advantage
in this case.
An important aspect of the expanded routine in the context of the time-spectral adjoint is
the need for additional loops to account for the extra time instances. As is shown in Figure 4.2,
there are two time instance loops in the spectral computation. The outer loop accounts for
the time instances in the residual, while the inner loop accounts for the time instances of the
states and coordinates. Therefore, we can see that the inner part of the computation scales
with the number of time instances squared, while the outer part of the computation scales
only with the number of time instances. However, this is a somewhat naive implementation
of the single cell routine. Examining Equation (3.57) more closely, one can see that it is only
the spectral operator, V Dt n , that contains states from all N time instances at one time.
Therefore it is possible to reduce the number of terms inside the inner time-spectral loop to
53
only those necessary for this term. This simplified routine is diagramed in Figure 4.3. With
Outer Time Spectral Loop
ComputeResidual
AdjustInflowAngle
CheckInputParameters
ReferenceState
SetFlowInfinityState
xHalo
Metric
VolPym
GridVelocitiesFineLevel
NormalVelocitiesAllLevels
ComputePressure
BC Symmetry
ApplyAllBCs
BC Farfield
TimeStep
BC EulerWall
InitializeResidual
Inner Time Spectral Loop
Residual
Inviscid Central Flux
Inviscid Scalar Dissipation Flux
Inviscid Upwind Flux
VolumeNormalization
Left/Right State
Riemann Flux
Note: Only one of either the scalar dissipation flux or

the upwind flux is computed for a given case.
Figure 4.3: Call-graph: improved residual computation: time-spectral case
54
this improved implementation, the number of computations inside the inner time-spectral loop
is significantly reduced and the derivative computation now scales, essentially, with the number
of time instances, N , rather than the number of time instances squared, N 2 . Timing results
demonstrating this fact are presented in Section 4.4.3.
4.3.3
I/ and I/x
For most aerodynamic shape optimization problems, the objectives of interest are the forces
and moments or the corresponding coefficients acting on the body being optimized.
Computing the partial derivatives of these quantities with respect to the states and mesh
coordinates, I/ and I/x, is significantly simpler than the computation of the residual
partial derivatives. This becomes evident when comparing the routines required to compute the
residual (Figure 4.3) to the routines required to compute the forces and moments (Figure 4.4).
As these figures show, for the inviscid equations considered here, the force computation simply
ComputeForces
GetSurfaceCoordinates
ComputeSurfacePressures
ComputeSurfaceNormals
ApplyAllBCs
BC EulerWall
ComputeForcesAndMoments
AdjustInflowAngle
ComputeCoefficients
Figure 4.4: Call-graph: force and moment computation

requires an integration of the pressure over the surface of the body in question. This requires
an application of the boundary conditions as well as a surface normal computation, but this is
significantly simpler than the complicated flux computation required for the residual. Further,
the ratio of input variables to output variables strongly favors a reverse mode technique for
this computations. For a typical optimization problem, one might be interested in a few, force
55
and moment coefficients for example, CL and CD for a lift constrained drag minimization
while the surface needed to compute those coefficients may require hundreds, thousands
or even tens of thousands of surface cells to accurately discretize the surface. This yields an
extremely high ratio of input variables to output variables and thus strongly favors the reverse
mode approach. As with the residual routines, the force routines have been differentiated using
the reverse mode, source transformation capabilities of Tapenade [84, 42]. This differentiation
yields a routine that computes all of the state and coordinate derivatives of a specified force or
moment coefficient in a single pass.
The extension of this concept to the time-spectral case is relatively straightforward. In this
thesis, simple spectral objectives, such as the average lift, drag and moment coefficients are
considered. In these cases, the spectral objective is based on simple algebraic combinations of
the corresponding coefficient values at each of the discrete time instances in the solution. For
example, the average drag may be computed as:
CD =
1
1
1
CD1 + CD2 + + CDN
N
N
N
(4.11)
where each of the coefficients, CDi is computed directly from the states of time instance i. While
the time instances of the residual computation are coupled through the spectral time derivative
of Equation (3.57), once the solution is computed, the computation of the coefficients at each
instance in time are independent. Therefore, so long as the objective function is a function
of these independent coefficients, the time instance based coefficients and their derivatives can
be used to form the time spectral objective derivatives. Using the chain rule, this idea can be
expressed as:
I
I CDi
=
.
n
CDi n
(4.12)
In this case, the derivative of interest, I/ n , a vector of size Ni [N Ncells ], is computed

as a matrix-matrix product of two matrices of size Ni N and N [N Ncells ] respectively.
However, a reverse mode computation for the second term, CDi / n already exists from
the steady-state case. Because the coefficient computation on each of the time instances is
independent, once the first derivative matrix, I/CDi , is computed it can be multiplied by the
result of the CDi / n computation, time instance by time instance, to generate the necessary
derivatives without ever storing the second matrix. This approach yields identically the same
ratio of inputs to outputs as the general steady case discussed above. The only distinction is
that in the spectral computation, the coefficient derivatives are run N times, once for each time
instance, while for the steady case, the derivatives are run once.
4.3.4
56
Solution of the Adjoint System
Based on the characteristics of the adjoint equations as well as the fact that both of the residual
sensitivity matrices can be computed at the same time, the matrices from Equation (4.8) are
both stored explicitly in memory for the solution of the adjoint equations. The matrices are
sparse for both the steady case and the time-spectral case. However, the time-spectral solution
has a slightly different sparsity pattern than the steady case. The sparsity patterns of the steady
and time-spectral cases for a 24 block H-H mesh are shown in Figures 4.5 and 4.6 respectively.
The two figures show that the overall sparsity pattern for the two cases are similar. Further,
Figure 4.5: Steady case sparsity pattern: 24 block H-H mesh

the close up views show that what was a single block in the steady case has become a grouping
of N blocks in the spectral case. Note that there is an additional set of off diagonal entries
present in each row for each time instance in the solution. These off diagonal entries come
from the spectral operator that couples the time instances together. The grouping apparent in
Figure 4.6 is indicative of the decision to order the matrix such that all time instances of a given
block are adjacent in the matrix. For SUmb, this ordering provides much better performance
than the alternative, where all blocks of a given time instance are adjacent. This stems from
the fact that SUmb is parallelized by block, not by time instance. Thus, the derivatives for
all time instances of a given block are computed on the same processor. Grouping the cells in
the matrix by block instead of by time instance requires an immense amount of communication
57
Figure 4.6: Time-spectral sparsity pattern: 24 block H-H mesh: 3 time instances
during the assembly of the matrix.
Once the various terms are generated, the Portable, Extensible Toolkit for Scientific Computation (PETSc) [6] is used to store the sparse derivatives and solve the linear system of
equations. More specifically, the linear solution is computed using a restarted GMRES iterative
solver restarted after 150 subspace vectors with an additive-Schwarz global preconditioner
and ILU(1) local preconditioning. The preconditioning matrix is computed using the lumped
dissipation technique of Hicken and Zingg [43]. This approach reduces the bandwidth of the preconditioning matrix, thereby reducing the cost of generating the matrix as well as reducing the
memory requirements for storing and factoring the matrix. In addition, the off-time-instance
entries of the matrix are ignored when forming the preconditioner. This further reduces the
fill-in of the ILU preconditioner and contributes to a reduction in the memory required to solve
the system. Note that since the preconditioning matrix contains derivatives with respect to the
states only, the preconditioning matrix is computed with a reduced residual computation that
does not include the metric terms.
4.4
58
Verification
To verify the implementation of the time-spectral ADjoint, derivatives from the ADjoint are
compared to derivatives computed using the complex-step method [65]. For the complex-step
method, the sensitivity of a function, I(x), is computed as:
dI
Im(I(x + ih))
=
+ O(h2 )
dx
h
where i =
(4.13)
1 and h is an extremely small step in this case 1020 . Because the perturbation
is carried through the code in the complex portion of the variable, the normal subtractive
cancellation issues associated with the finite difference derivative technique are not present.
Thus, h can be made very small, reducing the O(h2 ) truncation error to negligible levels and
yielding numerically exact benchmark derivatives.
4.4.1
Test case
To benchmark the time-spectral adjoint for accuracy, derivatives are shown for a 917,000 cell
ONERA M6 wing mesh. The wing is simulated with Euler wall surfaces and a symmetry plane
at the root. The mesh has an H-H topology with a nominal off wall spacing of 0.002. The near
field symmetry plane of the mesh and a sample solution at Mach=0.8395 and = 3.06 deg.
are shown in Figure 4.7.
4.4.2
Accuracy
The design variables considered for this case are those for a simple planform optimization,
specifically a simple linear twist distribution, a single value of sweep, as well as angle of attack.
Derivatives are considered for both pitching and plunging motions. Table 4.1 shows the values
for a pitching motion and Table 4.2 shows the values for a plunging motion.
As can be seen from Tables 4.1 and 4.2, the time-spectral ADjoint implementation is extremely accurate. For both the pitching and plunging motions, the derivatives of the coefficients
CL , CD , Cm match to between 8 and 12 digits, which is consistent with the accuracy of
the flow solution and adjoint solution used to generate them. The derivatives of the derivative
quantities Cmq , Cm , and Cm are slightly less accurate, showing between 7 and 9 digits
of agreement. This is likely attributable to the fact that they are derivatives of derivatives.
None the less they are certainly accurate enough for the numerical optimizations conducted
in this work, which are typically converged to O 106 . Note also that the very high accuracy
of the results helps to ensure that there are no errors or omissions in the linearization of the
CFD code. This is important for gradient-based optimizations as errors in the gradients of the
59
CoefPressure: -1.2 -1 -0.8 -0.6 -0.4 -0.2 0
(a) ONERA M6 mesh: root slice, near field
0.2 0.4 0.6 0.8
(b) ONERA M6 solution: M = 0.8395, = 3.06 deg.,

917,000 cell mesh
Figure 4.7: ONERA M6 test case
Design Variable
Adjoint
Complex Step
9.630674822394102
9.630674822417102
7.800095886613103
7.800095886603103
Cm
4.840706880027102
4.840706880028102
Cmq
-4.420105136652101
-4.420105097664101
CL
-1.216072002894103
-1.216072052039103
-2.50954656804104
-2.50954662672104
5.594081595955103
5.594081590141103
-3.31317766211796101
-3.31317768040761101
2.9912916269936102
2.9912916277759102
2.754766857997103
2.754766858921103
1.7946758914073102
1.7946758914994102
-2.20548652926130101
-2.20548652147293101
CL
CD
CD
Sweep
Cm
Cmq
CL
CD
Cm
Cmq
Twist
Table 4.1: Sensitivity verification: ONERA M6, pitching motion, 917,000 cells: 1012 relative
convergence
60
Adjoint
Complex Step
CL
9.6528123612840102
9.6528123639616102
CD
8.484790352376103
8.484790352748103
5.0358984234680102
5.0358984256455102
Cm
2.13562444901338101
2.13562443734326101
Cm
-9.500031568715116100
-9.500031661976111100
CL
-1.207345004567101
-1.207345059580103
CD
-3.04006061953104
-3.04006068084104
5.490542451699103
5.490542444506103
Cm
9.7207475211477102
9.7207475784541102
Cm
1.68212115913168101
1.68212116263489101
CL
2.9974067104863102
2.9974067113723102
CD
2.950611281951103
2.950611282771103
1.8579734812958102
1.8579734813456102
Cm
9.2896566888685102
9.2896566793109102
Cm
-3.325725018549023100
-3.325725019779438100
Cm
Cm
Cm
Design Variable
Sweep
Twist
Table 4.2: Sensitivity verification: ONERA M6, plunging motion, 917,000 cells: 1012 relative
convergence
61
objective and constraints can have a significant impact on the performance of gradient-based
optimization techniques.
4.4.3
Performance
To assess the computational cost of the method, the cost of the time-spectral ADjoint is compared against the cost of the time-spectral flow solver, as well as the steady-state ADjoint. The
flow solver in this case is a NewtonKrylov solver implemented using the nonlinear solvers in
PETSc. The steady-state ADjoint is also solved using PETSc and implemented with the same
methods as the time-spectral ADjoint. The results of these comparisons are shown in Table 4.3.
Time-Spectral
Steady-State
N =3
N =5
N =7
N =9
Number of processors
24
80
112
176
224
Number of flow states
4587520
13762560
22937600
32112640
41287680
191147
172032
204800
182458
184320
Percent load imbalance
7.14
7.15
17.86
Flow solution time: (1010 convergence)
1.08
1.56
1.87
2.46
2.34
Jacobian matrix assembly time
0.26
0.46
0.58
0.68
0.66
Preconditioner assembly time
0.03
0.04
0.08
0.1
0.13
Algebraic volume/surface sensitivity time
0.14
0.14
0.14
0.14
0.14
1.89
2.89
2.53
2.98
2.75
Total sensitivity time
0.23
0.23
0.23
0.23
0.23
Total adjoint time
2.54
3.76
3.56
4.12
3.92
4.82
8.71
14.82
21.82
1.76
8.95
11.81
17.93
25.71
Average # of flow states per processor
Adjoint solution time: (10
10
convergence)
Flow solution: normalized total cost

Adjoint solution: normalized total cost
Table 4.3: Time-spectral ADjoint computational cost breakdown for ONERA M6 (normalized
with respect to a total flow solution cost of 160.3 seconds)
Table 4.3 shows that the time-spectral ADjoint implementation is cost competitive with the
steady-state case and that the overall cost of the ADjoint method is reasonable. Comparing
on a case by case basis, the adjoint solution ranges between 1.6 and 2.4 times the cost of a
flow solution. Further the ratio of the steady case and the three time instance case are the
highest ratios, indicating that the implementation scales well. To quantify this more exactly,
the log of the total computational cost (normalized time multiplied by the number of flow
states) is plotted versus the log of the number of flow states in the problem (Ncell N N ) in
62
Figure 4.8. In an ideal case, this plot would have a slope of one, indicating that the amount of
work required to solve the problem is scaling exactly with the problem size. Figure 4.8 shows
that the value of this slope for the current implementation of the flow solver is 1.38 and the
slope for the implementation of the time-spectral adjoint is 1.21. While not perfect, these values
show that the problem is scaling well. Looking at the performance from an absolute sense, the
Log of Total Computational Cost
4.5
4
3.5
Slope =1.21
Slope = 1.38
2.5
2
1.5
1
Flow Solution
Adjoint Solution
0.5
0
6.8
7.2
7.4
7.6
Log of number of flow states
Figure 4.8: Time-spectral ADjoint scaling

implementation of the time spectral ADjoint allows for the solution of problems including
both the flow solution and the adjoint solution with over 41 million flow states in under
30 minutes on 224 Intel Nehalem processors. As will be shown in Chapter 6, at this level of
computational cost, optimization of periodic unsteady problems are manageable.
4.5
Summary
In this chapter, the time spectral ADjoint has been introduced, developed and verified. The theory has been demonstrated on a meaningful sized case and has been shown to be both extremely
accurate and cost effective. Having demonstrated this technique for computing the gradients
neccesary for aerodynamic shape optimization with stability derivatives, the discussion can now
63
move to the remaining disciplines neccesary for the various stability-constrained optimizations
such as center of gravity and moment of inertia calculations and to a discussion of the
optimization results.
Chapter 5
Auxilliary Analyses
Aside from the CFD-based stability derivative calculations, which form the major component
of the development in this thesis, there are some additional analyses which are necessary for
the various stability constraints. The sections in this chapter outline each of these, why they
are necessary, and what is involved in each of the calculations.
5.1
Center of Gravity Calculation
The center of gravity (CG) location is fundamental for most of the calculations in this thesis.
Firstly, it is the point about which all aerodynamic moments are calculated. Therefore, the
moment coefficient, Cm , and all of its derivatives in this case Cm , Cm and Cmq are
strongly dependent on the CG location. Secondly, the mass moment of inertia calculation uses
the CG location as a reference point. All of these quantities are necessary for the computation
of the stability parameters used as constraints. Therefore, having an appropriate CG location
is necessary to produce meaningful design optimization results.
For this work, a relatively simple wing CG calculation was implemented. The method is
derived from the work of Chai et al. [18]. In that work, the authors state that the wing CG
for a normal transport wing is located between the fore and aft spars along the wing mean
aerodynamic chord (MAC). Since flying wings are being examined in this study, it is assumed
that this estimate of the wing CG is a reasonable estimate of the CG for the entire aircraft.
The starting point for the calculation is the computation of the wing MAC and the location
of its quarter chord. These calculations are based on the methods presented in ESDU item
76003 [25]. One then determines the location of the intersections of the spars with the MAC. In
this work, the fore and aft spars are assumed to be at 25% and 75% of the MAC, respectively.
The longitudinal location of the CG can then be determined as a percentage CG%M AC
of the distance between the spars at the MAC xspar . Thus, a value of zero for CG% would
64
65
Chapter 5. Auxilliary Analyses

Wing Root
Mean Aerodynamic Chord
Forward Spar
deltaX spar
Rear Spar
CG Range:
% of Dist. Between spars
Wing Tip
Figure 5.1: Diagram of CG calculation

place the CG at the intersection of the forward spar and the MAC, a value of one for CG%
would place the CG at the intersection of the rear spar and the MAC, and a value of negative
one for CG% would place the CG a distance of xspar in front of the intersection of the forward
spar and the MAC. Figure 5.1 illustrates the key parameters in the approach.
5.2
Moment of Inertia Calculations
The moment of inertia plays a key role in the dynamic stability constraints. Just as mass relates
force to acceleration, moment of inertia relates moment to rotational acceleration. As a result,
the pitch moment of inertia Izz shows up in the normalization of all of the moment terms
in the short-period approximation described in Chapter 2. Thus, an estimate for the moment
of inertia is required for the dynamic-stability constrained optimizations.
The calculations for the moments of inertia are based on first principles. The definition of
moment of inertia is:
Z
I=
r2 dm
(5.1)
To compute this value, the domain of integration, , is set to be the surface of the aircraft.
In this case, the surface is defined by a B-spline surface within the pyGeo geometry engine of
66
Y
X
Z
Figure 5.2: Discrete surface mesh with lumped masses at the cell centroids
Kenway et al. [49]. Having defined the surface in this fashion, a discrete surface mesh can be
created by evaluating the spline over a uniform distribution of values in u,v parameter space. A
sample wing mesh is shown in Figure 5.2. These discrete surface cells are then given a thickness
(t) and a density (), and the area of each cell is computed using the following:
v1 = X[j + 1, k + 1] X[j, k]
(5.2)
v2 = X[j, k + 1] X[j + 1, k]
1
s =
(v1 v2 )
2
sTotal = sx + sy + sz
(5.3)
(5.4)
(5.5)
where v1 and v2 are the diagonal vectors of the cell, s is the projected area of the cell in
each coordinate direction and sTotal is the total area of the cell. The key variables in this
computation are illustrated in Figure 5.3. The area and thickness of the cell are multiplied
to get the volume of the cell, which when multiplied by the density gives the total mass for
67
S_Total
i,j = 2,2
i,j = 2,1
v1
v2
i,j = 1,2
i,j =1,1
Figure 5.3: Cell area calculation
each of the discrete cells on the surface mesh. These lumped masses are illustrated as spheres
in Figure 5.2. Considering this mass to be located at the centroid of the cell, the moment of
inertia can the be computed as:
I=
mr2 .
(5.6)
In this case, r2 is the distance from the lumped mass to the CG, perpendicular to the rotational
axis of interest. Therefore, for Ixx :
r2 = (ymass yCG )2 + (zmass zCG )2 ,
(5.7)
r2 = (xmass xCG )2 + (zmass zCG )2 ,
(5.8)
r2 = (ymass yCG )2 + (xmass xCG )2 .
(5.9)
for Iyy :
and for Izz :
In this thesis, the coordinate system used has x in the flow direction and z out the wing.
Therefore the moment of inertia of interest for the pitching calculations is Izz or:
I=

m (ymass yCG )2 + (xmass xCG )2 .
(5.10)
68
Because this value approximates the moment of inertia of the wing itself, an additional multiplier
is added to the computation to account for the remaining mass distribution in the aircraft. This
parameter is given to the optimizer as a variable to allow it to satisfy the constraints in the
dynamic stability constrained optimization.
5.3
Root Bending Moment Constraint
The root bending moment constraint is included to give some consideration to the structural
implications of the varying wing shapes. There are well known trade-offs between aerodynamic
and structural performance for wings, the most prominent of which has to do with the effects
of span. Increasing the span increases the aspect ratio of the wing, thereby reducing induced
drag. However, the same span extension also increases the bending moment at the root of the
wing, requiring a heavier structure to support the aerodynamic load. A similar effect is caused
by wing sweep. For shearing sweep where individual sections of the wing are translated
in the flow-wise direction to sweep the wing the effective structural length of the wing
increases as the wing is swept. This can be counteracted by reducing the span of the wing,
which introduces a trade-off between induced drag and wave drag at transonic Mach numbers.
Therefore, if span and sweep are used as design variables, it is necessary to have some method
to account for these trade-offs between aerodynamic performance and structural weight. In
this thesis, the root bending moment is used as a proxy for the structural performance of the
wing. The assumption implicit in this approach is that two wings with the same root bending
moment will require the same amount of material to support the load on the wing and thus
have the same weight. This is a relatively simplistic assumption, but serves as a useful metric
for including the effects of structural considerations in the optimization.
To compute the bending moment coefficient, the pressure is integrated over the aircraft to
get the force and moment coefficients about a reference point, xref . This computation yields
the values of Cfx , Cfy , Cfz , Cmx , Cmy and Cmz , which are the force and moment coefficients in
the three principle Cartesian axes. Note that x is the flow direction, y is the vertical direction
and z is the span-wise coordinate. A bending reference point, xbending , is then specified at the
root of the wing. This is the point about which the net bending moment is calculated. The
bending moment is then calculated as:
xbendingy xrefy
xbendingz xrefz
Cfz
cref
cref
x
xbendingx xrefx
bendingy xrefy
Cfy
+ Cfx
cref
cref
Cbendingx
= Cmx + Cfy
Cbendingz
= Cmz
q
2
2
=
Cbending
+ Cbending
Cb
(5.11)
69

x
z
Wing Tip
Wing Root
X_ref
Cmz
X_bending
Flow Direction
Cmx
C_bending_z
C_bending_x
C_b
Assumed support orientation
Figure 5.4: Bending moment calculation

where cref is the reference chord length, Cb is the total root bending moment coefficient and
Cbendingx and Cbendingz are the Cartesian components of the root bending moment about the
bending reference point a the wing root.
A diagram illustrating the various components of this calculation is shown in Figure 5.4.
Note that since the torsional component of the wing moment is non-zero, the effective bending
moment is not necessarily aligned with the wing. Also note that as the wing is swept, this
torsional component increases, causing the combined effective bending moment to sweep with
the wing. Therefore, it is assumed that the wing can be supported normal to the effective
total bending moment, regardless of the wing sweep and pressure distribution, Thus, only the
total magnitude of the bending coefficient, Cb , is considered. The reference value of Cb used to
constrain the wing can have a significant effect on the optimal design. In this study, the value
from the elliptical optimum computed at M = 0.5 is used as the reference value. This
value is then scaled with the root cross section to ensure that as the optimizer increases the
aspect ratio of the wing, the allowable bending moment is reduced to account for the reduced
second moment of area at the wing root. The reasoning behind this scaling is shown below.
The definition of bending stress is:
=
My
I
(5.12)
I
y
(5.13)
Rearranging for the moment gives:

M=
In this work, scaling the chord of the wing also scales its thickness, so both the thickness
and chord effects can be considered together. Therefore, doing a dimensional analysis on the
moment we get:
M
L4
= L3 .
L
(5.14)
70

Expressing this in terms of coefficients we get
1 2
V A
cCbending L3
2
(5.15)
qA
cCbending L3
(5.16)
or,
However, because the wing is assumed to have a constant t/c ratio, c is also of dimension L.
Also, A is constrained to be constant in the optimizations, therefore
Cbending
L3
qAL
(5.17)
Cbending
L2
qA
(5.18)
or,
Thus, with q and A kept constant, the allowable bending moment must be scaled with L2 to
enforce a constant allowable bending stress at the wing root. L is the scaling factor applied to
the root chord of the wing.
5.4
Stability Constraints
As stated previously, the purpose of this work is to develop a practicable means of including
stability constraints based on three-dimensional CFD methods into an aerodynamic optimization problem. The approach adopted here is to model the stability characteristics of the aircraft
with a linear flight dynamic model, which allows the aerodynamic portion of the problem to be
modeled with linear stability derivatives. The remainder of this section outlines, more specifically, how these stability derivatives, and the resulting linear flight dynamics model, are used
to formulate stability constraints for the optimization problem. Three stability constraints are
considered, two based on static stability theory and one based on dynamic stability theory.
A discussion of other stability criteria applied to a flying wing configuration can be found in
Agenbag [1]
5.4.1
Static Longitudinal Stability Definition
The first static constraint is based on the definition of longitudinal pitch stiffness. To develop
this criteria, consider first the definition of equilibrium. For an aircraft to be in equilibrium,
the sum of the forces acting on the CG and the sum of the moments about the CG must be
zero. That means that lift must equal weight requiring a CL > 0 and CmCG must also be
zero. Further, the aircraft should return to this equilibrium condition when subject to minor
perturbations. Making the assumption that significant perturbations occur only in angle of
71
attack, , and that a nose up pitching moment is positive; then to return to equilibrium, a nose
down or negative pitching moment must be developed when increases. This requires
that the derivative CmCG / be negative. Therefore, the conditions for static stability in pitch
are:
CL > 0
CmCG
CmCG
= 0
(5.19)
< 0
This set of conditions enforces the limiting case of static pitch stability and is implemented as
shown in Section 6.3.
5.4.2
Static Margin
As will become evident from the results in Chapter 6, the limiting case of the definition does
not always yield a useful result. If the constraint is active, the design ends up on the limiting
bound of the stable envelope, which is not generally a desirable quality for an aircraft. This
raises the question of how to quantify a desirable value for Cm . For this, one can turn to the
definition of static margin. First, consider the definition of the moment coefficient,
CmCG = CmNP + (hCG hNP )CL
(5.20)
Where hCG and hNP are the streamwise locations of the CG and neutral point respectively,
normalized by the MAC. Differentiating with respect to yields,
CmCG
CL
= (hCG hNP )
(5.21)
or, using stability derivative notation,

Cm = (hCG hNP )CL
(5.22)
Defining the static margin, Kn , as the distance between the CG and the neutral point normalized
by the MAC,
Kn = hNP hCG
(5.23)
and substituting this relationship back into Equation (5.22) results in,
Cm = Kn CL
(5.24)
which can be rearranged as,

Kn =
Cm
.
CL
(5.25)
72

Thus, by specifying a static margin, one can determine a meaningful value for Cm .
To this point the discussion has been completely general and applicable to any aircraft. By
examining Equations (5.20) and (5.22), one can gain insight as to how the various stablility
conditions may be satisfied for a flying wing. Equation (5.20) shows that the moment coefficient
is dependent on the moment about the neutral point which does not change with angle of
attack and the location of the neutral point with respect to the CG. Thus, the optimizer
can trim the aircraft by altering one of three parameters. It can alter the moment coefficient
about the neutral point, accomplished primarily with airfoil shape adding reflex to the airfoil
or by twisting down the wing tips on a swept wing. It can alter the neutral point of the
aircraft, accomplished primarily by sweeping the wing. Finally, it can alter the CG location
of the aircraft, which is also affected by wing sweep, as well as payload location. However,
Equation (5.22) shows that these last two quantities also impact Cm and hence the static
margin. Thus, conceptually, the optimizer must first adjust the sweep and payload location to
create an acceptable static margin, then it must trim the aircraft using airfoil shape and section
twist. Practically speaking the optimizer will adjust the parameters simultaneously to satisfy
all of the necessary constraints. These various elements will be discussed along with the results
in Chapter 6.
5.4.3
Control Anticipation Parameter
The control anticipation parameter (CAP) is a method used to quantify the handling qualities
of an aircraft based on its short-period characteristics. The fundamental idea behind this
approach is that a pilots ability to fly an aircraft precisely along a given flight path is related
to the pilots ability to anticipate response of the aircraft. In this approach, Bihrle [7] relates
the pilots ability to anticipate the aircrafts response to a ratio between the instantaneous
pitch acceleration of the aircraft, q,
and the steady-state normal acceleration of the aircraft,
n. This can be expressed as:
CAP =
q
n
(5.26)
However, these quantities are not necessarily simple to evaluate for an aircraft. Therefore, with
a little rearranging, Bihrle [7] provides the following expression:
CAP =
n2
n
(5.27)
where n2 is the natural frequency of aircraft in pitch and n is simply,

n =
1
2
2 V ACL
or the change in normal acceleration of the aircraft with angle of attack.
(5.28)
73
The United States Military has specified acceptable limits for CAP and damping ratio for
various combinations of aircraft and flight conditions [73]. For an aircraft at cruise (Category
B), those limits are listed in Table 5.1.
Damping Ratio Limits
CAP Limits
Lower Bound
Upper Bound
Lower Bound
Upper Bound
Level 1
0.30
2.0
0.085
3.6
Level 2
0.20
2.0
0.038
10.0
Level 3
0.15
0.038
Table 5.1: Stability bounds for damping ratio and CAP parameter
5.5
Geometry Handling
The geometry for the optimizations detailed in Chapter 6 is manipulated using a multi-layered
approach. The top level geometry definition is handled by pyACDT [86]. This geometry definition
is based on a conceptual design methodology, so the wing is defined in terms of individual linear
segments, each with a defined area, span, taper ratio, sweep and twist. The design variables
presented in Section 6.2 are configured such that each section comprises one ninth of the area
and span specified by the optimizer. The sweep of all nine sections are defined to be the
same such that the leading edge of the wing is always straight. Finally, the twist variables are
configured such that the tip twist of one section is the same as the root twist of the section
immediately outboard of it, yielding nine independent values of section twist. The root twist
angle of the wing is fixed at zero degrees, to avoid duplicating the design degree of freedom
of the angle of attack. Note that the conceptual planform parameters from this layer are also
used in the CG computation outlined in Section 5.1.
The construction of the detailed geometry is handled by pyPSG as presented by Kenway et
al. [49]. This tool is used to create a watertight surface representation of the aircraft, a set of
spline volumes enveloping the aircraft and a set of reference axes inside the aircraft. The set
of volumes enveloping the aircraft are used in conjunction with a free form deformation (FFD)
technique to handle the detailed geometry manipulations during the optimization [49]. The
surface points on the CFD mesh are embedded parametrically in these spline volumes, such
that when the volumes are moved or deformed, the CFD surface mesh is modified as well. The
reference axes are used to tie together the control points of the volumes enveloping the aircraft.
By tying the control points together with a reference axis, the effects of physical design variables
such as sweep, twist, taper, area, span and chord can be recreated. To implement these
74
Figure 5.5: Geometry manipulation setup

design variable groupings, the pyACDT layer has been modified to compute the appropriate
coefficients required by the reference axis to transfer the motion from the conceptual geometry
to the FFD volumes, and onto the CFD surface. A graphic depicting the surface, FFD and
reference axes is shown in Figure 5.5. Note that the lines between the spheres represent the links
between the FFD volume control points and the reference axis, the large rectangular volume is
the FFD volume and the line along the trailing edge marked by the small cubes is the reference
axis.
Figure 5.6 further demonstrates the approach, showing a deformed geometry. This deformation includes a span extension from 3m to 3.5m, a sweep of two degrees, a twist of negative
ten degrees near mid-span and a tip twist of positive ten degrees.
Mesh Deformation
Once the CFD surface mesh is deformed, those deformations need to be propagated to the
remainder of the CFD volume mesh. This is handled using a hybrid, finite element algebraic
mesh deformation technique, as outlined in Kenway et al. [49]. In this technique, the CFD
mesh is represented by a very coarse finite element mesh. This mesh is deformed using the
perturbations of the coarse mesh surface nodes as input values. When the finite element system
is solved, the large surface deformations are propagated to the remainder of the volume mesh.
Once the large deformations are handled, an algebraic mesh deformation scheme based on
trans-finite interpolation is used to handle the detailed surface deformations on the blocks
immediately adjacent to the surface. This completes the two stage mesh deformation process
and provides a fully updated volume mesh to the CFD Solver.
Figure 5.6: Deformed geometry
75
Chapter 6
Design Optimization Results

To demonstrate the effectiveness of the methods developed in this thesis, a series of design
optimization problems is solved with a simple rectangular wing as the baseline design. The wing
has a half span of three meters and a chord of one meter, giving the wing an aspect ratio of six.
The wing has a taper ratio of one and a leading edge sweep angle of zero degrees. The initial
airfoil section has a NACA 0012 profile. Details of the geometry are summarized in Table 6.1
and a perspective view of the wing is shown in Figure 6.1. This test case is based on the induced
drag validation case proposed by Hicken and Zingg [44]. Hicken and Zingg demonstrate that
using this wing, with sections twisted about the trailing edge, one can reproduce the elliptical
distribution outlined by lifting line theory. In particular, they highlight the use of a straight
trailing edge, with spanwise sections twisted about the trailing edge, to minimize the impact
of nonplanar effects in the wake. This configuration is used as the initial geometry in this
set of optimizations so that the twist-only optimization outlined in Section 6.3.1 reproduces
the elliptical result and can be used as a point of comparison for the remaining optimizations.
This will provide a means of quantifying the effect of the various stability constraints on the
optimal solution. All of the optimization results presented are computed on an 1,105,920 cell
mesh. The mesh has a C-O topology and is split into 32 blocks. The farfield boundary is
approximately 15 chords from the wing and the off wall spacing of the mesh is 1 103 m at
the leading edge and 5 104 m at the trailing edge. Note that all optimizations conducted in
this thesis use the pyOpt [87] interface to SNOPT [32]. pyOpt is a Python based optimization
framework that wraps a variety of commercial and open source optimization packages with a
common Python interface [87], allowing for a straightforward interchange of various optimizers
for a given problem. SNOPT is an optimization package based on the SQP method for the
solution of large-scale non-linear optimization problems [32]
76
77
Chapter 6. Design Optimization Results

Parameter
Value
Half wing area (m)
3.0
Half wing span (m)
3.0
Chord (m)
1.0
Leading edge sweep (deg.)
0.0
Taper ratio
1.0
Wing tip washout (deg.)
0.0
Wing dihedral
0.0
Table 6.1: Baseline wing: geometry specifications
Y
X
Z
Figure 6.1: Initial geometry of wing test case
6.1
Multidisciplinary Optimization Overview
In order to better understand the optimization problems presented in this chapter and the choice
of variables used herein, it is necessary to provide a brief overview of multidisciplinary design
optimization (MDO) architectures. Generally speaking, an MDO architecture is a method used
to handle the coupling between disciplines in a multidisciplinary optimization problem. Numerous architectures have been developed to handle this coupling and the various methods can be
divided into two broad categories: monolithic architectures and decompositional architectures.
Monolithic architectures such as the multidisciplinary feasible (MDF) architecture [21], the
individual discipline feasible (IDF) architecture [21] and the simultaneous analysis and design
(SAND) [39] architecture set up the problem as a single level optimization and handle the
objective function optimization and interdisciplinary coupling in one optimization problem.
78
Decompositional architectures such as collaborative optimization (CO) [14], concurrent subspace optimization (CSSO) [11], bi-level integrated system synthesis (BLISS) [100] and analytical target cascading [50] decompose the optimization into multiple optimization problems
where each secondary optimization problem handles a portion of the overall optimization and a
global coordination problem ensures that the various secondary optimization problems produce
the correct multidisciplinary optimum.
As an aside, note that all optimization diagrams in this thesis, including the MDF and
IDF diagrams in this section, are presented in the extended design structure matrix (XDSM)
format of Lambe and Martins [51]. These diagrams are designed to show the data connectivity
between the disciplines as well as the process flow of the solution algorithm. The large blocks
on the diagonal represent the major analyses in the optimization process. These include the
discipline analyses, the evaluation of any constraints from the converged disciplines as well as the
optimization iterations themselves. The off-diagonal nodes represent data connectivity between
the disciplines, with the boxes above the diagonal representing feed-forward connections and
the boxes below the diagonal representing feed-back connections. The thick gray lines represent
data flow paths between the various disciplines, and the thin black line represents the process
flow through the solution algorithm. The case specific diagrams included in later sections have
been updated to include the actual variables used in each optimization formulation.
Returning to the discussion of architectures, two MDO architectures are relevant to this
thesis, the MDF architecture and the IDF architecture. In the MDF architecture, as shown
in Figure 6.2, all of the disciplines are tightly coupled. Each discipline passes the relevant
portion of its solution the coupling variables, y to the other disciplines directly. This
coupled system is then iterated to convergence at each optimization iteration before taking a
step in the design space. As a result of this tight coupling, the MDF approach also requires
a complete, coupled-sensitivity analysis of all the disciplines. In the IDF architecture, shown
in Figure 6.3, the disciplines are completely decoupled. Each discipline is solved once per
optimization iteration and is responsible for its own sensitivities. In this case, the optimizer
passes target coupling variables, y t , to each discipline to facilitate the solution of the individual
disciplines. To ensure that the system is multidisciplinary feasible at the optimum solution,
an additional set of compatibility constraints, y y t = 0, is added to the problem to ensure
that the optimizer specified coupling variables match the actual computed value of the various
coupling variables. As will become evident in the following sections, the optimization problems
solved in this thesis are actually a hybrid of both MDF and IDF approaches. The analyses used
herein can be decomposed into four primary disciplines: geometry, aerodynamics, structures and
stability. The geometry discipline handles the wing surface manipulations and the computation
of the CG, MAC and moment of inertia, the aerodynamics discipline handles the computation
79
x
x
yt
0, 71:
Optimization
2 : x0 , x1
3 : x0 , x2
4 : x0 , x3
6:x
1, 52:
MDA
2 : y2t , y3t
3 : y3t
y1
5 : y1
2:
Analysis 1
3 : y1
4 : y1
6 : y1
y2
5 : y2
3:
Analysis 2
4 : y2
6 : y2
y3
5 : y3
4:
Analysis 3
6 : y3
6:
Functions
7 : f, c
Figure 6.2: Generic MDF architecture [51]
x, y t
x
0,31:
Optimization
2 : x, y t
3 : f, c, cc
2:
Functions
y1
2 : y1
y2
2 : y2
y3
2 : y3
1 : x0 , x1 , y2t , y3t
1 : x0 , x2 , y1t , y3t
1 : x0 , x3 , y1t , y2t
1:
Analysis 1
1:
Analysis 2
Figure 6.3: Generic IDF architecture [51]
1:
Analysis 3
80
of the force and moment coefficients, as well as the stability derivatives for the aircraft, the
structures discipline computes the bending coefficient and the stability discipline computes the
static margin, CAP and damping ratio for the aircraft. In this thesis, the task of coupling
the disciplines is simplified greatly by the fact that, in this case, there is no feedback between
disciplines. Thus, the tightly coupled MDF architecture amounts to a sequenced evaluation of
the disciplines. However, when using the adjoint methods described in this work, coupling the
derivatives of the various disciplines is not as straightforward. This motivates the use of the IDF
architecture. In this context, the IDF architecture allows the individual discipline sensitivity
analyses to function unmodified and greatly simplifies computation of accurate gradients.
Despite this apparent advantage, there are some disciplines for which tight coupling makes
more sense. First of all, the coupling between the geometry and the aerodynamics requires a
large number of coupling variables. The number of mesh points on the surface can be in the
order thousands or more. In an IDF architecture, this would add thousands of variables and
constraints to the optimization problem, which would increase the complexity of the optimization problem unnecessarily. By tightly coupling the geometry surface to the aerodynamics,
those coupling variables do not become design variables and the optimization is simplified significantly. The other section where tight coupling makes sense is between the aerodynamics and
structures. The bending coefficient calculation is based on six force and moment coefficients, of
which only one , Cmz , is already present in the optimization. Separating these two disciplines
would require the solution of five additional adjoint problems per optimization iteration. By
tightly coupling the aerodynamics and structures, this can be reduced to a single additional
adjoint problem for the bending coefficient. A detailed overview of each optimization problem
is presented with the results for each formulation in the following sections.
6.2
Optimization Study Outline
The study presented in this chapter involves optimizing the baseline wing design by solving
different optimization problems to demonstrate the usefulness of the proposed approach. The
various optimization problems outlined in Section 6.3 are studied at three different Mach
numbers: 0.5, 0.7 and 0.85. This allows the effects of the various constraints to be considered
in both the subsonic and transonic regimes.
6.2.1
Design Variables
Two sets of design variables are used: one involving only planform variables and one involving
planform variables as well as 280 surface shape design variables. These surface shape design
variables the control points of the FFD volume modify the detailed surface shape of the
81
wing and affect both the stream-wise and span-wise profile of the wing. The planform-only
results were run at the two lower Mach numbers, while the optimizations including the surface
shape variables were run at all three Mach numbers. The design variables and their limits are
outlined in Tables 6.2 and 6.3. Table 6.2 shows the primary design variables that alter shape
and planform of the wing as well as the variables that affect the flow condition of the test case,
while Table 6.3 shows the compatibility design variables introduced as a result of the hybrid
IDF/MDF architecture used to solve the problems.
Design variable
Lower bound
Upper bound
Angle of attack (deg.)
Symbol
15
15
Section twist (deg.) (9 sections)
10
10
Area (m)
A1
2.9
3.2
Span (m)
2.0
3.2
Sweep (deg.)
42.5
Center of gravity variable
CG%
Izz Modifier
PIzz
0.5
10,20,30
0.05
0.05
FFD control points: y-offset (m)
FFD
Table 6.2: Primary design variables and their bounds
Design variable
Lower bound
Upper bound
Moment reference location (m)
xtCG
20
20
Rotation point (m)
xtCG
20
20
Chord (m)
ctroot
0.5
1.4
MACt
Target drag coefficient
t
CD
Target lift curve slope
CLt
20
Target moment curve slope
t
Cm
20
Target derivative
t
Cm
20
20
Target pitch derivative
t
Cm
q
20
Target MAC (m)
Symbol
Table 6.3: Compatibility design variables and their bounds

The relationship between each variable and the optimization problems can be described as
follows:
Primary Variables:
Angle of attack, : A variable to modify the free-stream flow direction

and/or the mean grid velocity direction of the test case. This is the primary
variable used to satisfy the CL constraint.
82
Section twist, i : Variables to modify the twist values for individual sections
along the wing. The sections rotate about the reference axis, which for this
study is located at the wing trailing edge. These variables primarily affect the
spanwise lift distribution, which in turn affects the induced drag created by
the wing.
Area, A1 : The planform area of half the wing. This value is divided equally
among the sections specified in pyACDT. Note that this value remains essentially
constant. It varies to allow the calculated value of area, A2 , meet the constraint.
This is necessary because A2 varies slightly with wing twist and sweep.
Span, b: The span of the half wing. This value is divided equally among the
sections specified in pyACDT. Since the area is essentially fixed this variable also
determines the chord of the wing. The span has a strong impact on both the
induced drag and the root bending moment coefficient of the wing.
Sweep, : The leading edge sweep of the wing. This variable affects both the
wave drag and the root bending moment coefficient of the wing.
Center of gravity variable, CG% : A variable that controls the x location of
the aircraft center of gravity. This value represents the value of the location of
the center of gravity in terms of a percentage of the distance between the front
and rear spars at the MAC. Modifying this variable is equivalent to modifying
xCG and is used to provide a nicely scaled range of values for the optimizer to
work with.
Izz modifier, PIzz : A factor that multiplies the pitch moment of inertia. The
base value of moment of inertia is calculated using the method described in
Section 5.2. This variable is a direct multiplier of that calculated value.
FFD control points: y-offset, FFD : Variables controlling the vertical
motion of the FFD control points. These variables modify the shape of the
wing surface and have a significant impact on the pitching moment of the wing
as well as the wave drag caused by the wing.
Compatibility design variables:
Moment reference location, xtCG : An IDF target value that provides the
x location of the reference point used to calculate the moment coefficient, Cm ,
and all related quantities. This value is constrained to be equal to xCG at the
optimal solution.
83
Rotation point, xtCG : An IDF target value that provides the x location of
the rotation point for the flow solutions that involve a pitching motion. This
value is constrained to be equal to the xCG at the optimal solution.
Chord, ctroot : An IDF target variable for the root chord length. This variable
is used to scale the allowable root bending moment coefficient, Cbref , and is
constrained to be equal to the value of root chord calculated in pyACDT.
Target MAC, MACt : An IDF target variable for the reference chord length
used in the CFD solver. Constrained to be equal to the MAC computed from
the conceptual geometry definition at the optimal solution.
t
Target drag coefficient, CD
: A variable resulting from the IDF architecture.
The variable is used as an input to the various stability disciplines. The value
is constrained to be equal to the drag coefficient predicted by the CFD solver
at the optimal solution.
Target lift curve slope, CLt : A variable resulting from the IDF architecture.
is constrained to be equal to the CL value predicted by the CFD solver at the
optimal solution.
t
Target moment curve slope, Cm
: A variable resulting from the IDF ar
chitecture. The variable is used as an input to the various stability disciplines.

The value is constrained to be equal to the Cm value predicted by the CFD
solver at the optimal solution.
t
Target derivative, Cm
: A variable resulting from the IDF architecture.
is constrained to be equal to the Cm value predicted by the CFD solver at the
optimal solution.
t
Target pitch derivative, Cm
: A variable resulting from the IDF architecq
ture. The variable is used as an input to the various stability disciplines. The
value is constrained to be equal to the Cmq value predicted by the CFD solver
at the optimal solution.
Note that not all design variables are used for all cases. A number of the variables listed
are present as a function of the IDF architecture mentioned previously and are, therefore, only
present when the associated disciplines are included in the problem statement. The specific sets
of variables used in each problem are shown in the problem statements presented in Section 6.3.
84
Figure 6.4: Thickness constraints: constraint locations and initial values represented by the
vertical line segments marked by the black spheres
6.2.2
Additional Constraints for Shape Variables
There are two additional constraints imposed for cases that include shape design variables.
The first constraint is a thickness constraint, included to keep the optimizer from making the
wing too thin. This is implemented by computing thicknesses at a variety of locations around
the aircraft and constraining them. At the optimal solution, they must be no smaller than
their initial value. The locations where these thicknesses are evaluated are defined by the
user. The user specifies an enclosed portion of the planform, inside which the thicknesses
are to be constrained. The user then defines the number of spanwise and chordwise sections
to be considered inside this area, which generates a grid of points at which the thicknesses
are constrained. Figure 6.4 demonstrates this process for the current test case. The vertical
line segments marked by the black spheres indicate the location and initial thicknesses of the
thickness constraints.
The second constraint concerns the control points of the FFD at the leading and trailing
edges of the wing. If, at either of these locations, the control points are allowed to move in
the same direction vertically, they have the same impact as a twist variable, leading to an
ill-defined problem. This is demonstrated in Figure 6.5. In this figure, the top left image shows
the unperturbed geometry, the top right image shows a geometry with the tip twisted upwards,
the bottom left image shows the geometry with the leading edge control points perturbed in the
same direction, and the bottom right image shows the geometry with the leading edge control
points perturbed in opposite directions as well as a second set of control points perturbed
upwards. As can be seen from the second and third images, both twisting the wing and moving
85
the leading edge control points in tandem produce a change in the effective twist angle of the
section. In the fourth figure, where the leading edge points are moved in equal and opposite
directions and the interior control points are moved together, there is a significant change in
the section shape, but the effective twist of the section does not change. Therefore, the control
points at the leading edge and trailing edge are constrained to move in equal and opposite
directions.
6.3
Optimization Problem Statements and Qualitative Results
To assess the effects of stability constraints on the aerodynamic shape optimization results,
a series of optimization problems is considered. The following section describes the various
formulations. The three optimization problems in Section 6.3.1 are a set of reference optimizations, designed to provide a basis for comparison with the stability constrained cases. The
two optimization problems in Section 6.3.2 are those based on static longitudinal stability constraints and are designed to show the impact that even basic stability constraints can have on
the optimal design. The final optimization, shown in Section 6.3.3 is based on dynamic longitudinal stability constraints and shows the full impact of the methods developed in this work.
The problem statements are interspersed with figures showing the results of the optimizations.
These figures are used to demonstrate the effects of the different formulations for the different
combinations of design variables and Mach number. More detailed quantitative results and the
associated discussion are presented in Section 6.4.
6.3.1
Reference Problems
Baseline Problem
As a first step in the study, a problem is formulated to find the angle of attack, , and CG
location, xCG , that yield a trimmed aircraft at the target value of CL for each Mach number.
These problems are formulated as:
minimize
CD
w.r.t. x = , xCG
subject to
(6.1)
CLref CL 0
Cm = 0
or in graphical form as shown in Figure 6.6. Note that the lift coefficient constraint is formulated as an inequality constraint rather than an equality constraint. Because induced drag is
86
Figure 6.5: FFD control point constraints: initial configuration (upper left), perturbation in
section twist (upper right), unconstrained perturbation of leading edge control points (lower
left), constrained perturbation of leading edge control points as well as a perturbation of midchord control points (lower left)
87
x : , xCG
0,31:
Optimization
, xCG
3 : f, c
CL , CD
, Cm
1 : , xCG
2:
f : CD
c : CLref CL
Cmz
2 : CL , CD , Cm
1:
Aerodynamics
Figure 6.6: XDSM for the baseline problem
proportional to CL2 , minimizing CL also minimizes drag, therefore the upper bound on the CL
constraint is unnecessary.
The solutions for Mach = 0.5, 0.7 and 0.85 are shown in Figures 6.7 through 6.9. These
solutions show that the solutions at the three different Mach numbers each have different
characteristics. The solution at M = 0.5 is fully subsonic, with a correspondingly peaky Cp
distribution on the airfoil. Both the neutral point of the wing and the required center of gravity
to trim are just slightly forward of the wing quarter chord. The M = 0.7 case is just starting
into the transonic regime and the inner half of the wing is exhibiting a weak shockwave near
the leading edge. The location of the center of gravity to trim the aircraft is slightly further
forward than in the M=0.5 case, but is still near the quarter chord. The neutral point has
also shifted slightly forward. The M = 0.85 case is fully in the transonic regime and exhibits a
strong shockwave on the aft half of the airfoil. As a result, the neutral point and the center of
gravity location to trim are significantly farther back, near the half chord.
Twist Optimization
As a second metric for comparison, a simple, twist-only optimization has been completed for
the subsonic Mach = 0.5 case. This problem is designed to reproduce the elliptical result
Figure 6.7: Baseline problem: M=0.5, e = 0.964
Figure 6.8: Baseline problem: M=0.7
88
89
Figure 6.9: Baseline problem: M=0.85
from lifting line theory and is formulated as:

minimize
CD
w.r.t. x = , i , CG% , xtCG

subject to
(6.2)
CLref CL 0
Cm = 0
xtCG xCG = 0
The data dependencies and solution process are shown in Figure 6.10. In this optimization,
twist variables i are added to control the lift distribution, and the full CG calculation is added
to control the location of the CG. Note that the aircraft is trimmed during the optimization
by forcing the pitch moment coefficient about the CG, Cm , to be zero. However, since the
location of the CG is allowed to move through the variable CG% the aircraft can be
trimmed without affecting the aerodynamic shape. This formulation yields the solution shown
in Figure 6.11.
The lift distribution achieved with this solution clearly shows an optimal solution that
closely approximates the elliptical solution. The exception to this is the area near the tip. This
location consists of complex three-dimensional flows that violate the assumptions used to derive
90
x : , i , CG% , xtCG ,MACt

, i , CG%
0,41:
Optimization
4 : f, c, c
3 : xtCG ,MACt
1 : i , CG%
2 : , xtCG
1:
Geometry
2: Surface
3:
f : CD
c : CLref CL
Cmz
cc : xCG xtCG
MAC MACt
xCG ,MAC
3 : xCG ,MAC
CL , CD
, Cm
z
3 : CL , CD , Cmz
Figure 6.10: XDSM for the twist optimization problem
Figure 6.11: Twist optimization: M=0.5, e = 0.977
2:
Aerodynamics
91
the elliptical result; therefore, it is expected that the lift distribution in this region does not
match the elliptical distribution.
Bending Moment Constrained Optimizations
As a third and final benchmark, a simple aerostructural optimization problem is considered. In
this problem, span (b), sweep () and area (A1 ) are added as additional design variables and a
root bending moment constraint, Cb , is added to keep these variables within sensible bounds.
The value of the bending coefficient at the root is that of the elliptical optimal solution at
Mach = 0.5. Additional compatibility constraints for the root chord, croot , and the calculated
area, A2 , have also been added to account for the new flexibility added to the problem. This
problem is formulated as:
minimize
CD
w.r.t. x = , i , A1 , b, , CG% , FFD , ctroot , xtCG , MACt

subject to
(6.3)
CLref CL 0
Cm = 0
Cb Cbref = 0
A2 Aref = 0
xCG xtCG = 0
MAC MACt = 0
croot ctroot = 0
Figure 6.12 depicts the data and process flow for this formulation in graphical form. The
root bending moment constrained optimizations have the same design variable flexibility as the
stability constrained cases described in Sections 6.3.2 and 6.3.3, but are constrained only by the
root bending moment. Therefore they are able to produce shapes with lower drag than those
cases with stability constraints. These optimizations provide the best indication of how much
performance is sacrificed to ensure that the stability constraints are satisfied. The results for
the planform-only optimizations are shown in Figures 6.13 and 6.14, while the results including
shape variables are shown in Figures 6.15 and 6.16. Note that there is no shape optimization
result for the M=0.5 case as the airfoil shape has very little impact on the drag of a subsonic
wing in inviscid flow.
Comparing the elliptical optimum solution shown in Figure 6.11 to the bending constrained
case at M=0.5, shown in Figure 6.13, there are some subtle yet important differences. In the
bending constrained case, the optimizer has taken advantage of the span variable to reduce the
induced drag. There is a visible extension in the span of the wing as compared to the original
92
x : , i , A1 , b,
, CG% , FFD ,
MACt , xtCG ,
ctroot
, i , A1 , b ,
, CG% , FFD
0,51:
Optimization
4 : MACt , xtCG , ctroot
1 : i , A1 , b, ,
CG% , FFD
2 : , xtCG , MACt
3 : xtCG , MACt
4:
f : CD
5 : f, c, cc
c : CLref CL
cc : xCG xtCG
Cmz
MAC MACt
Cb Cbref (ctroot )
croot ctroot
A2 Aref
xCG ,M AC ,ctroot
4 : xCG ,MAC,croot
CL , CD
, Cm
, A2
z
4 : CL , CD , Cmz , A2
Cb
4 : Cb
1:
Geometry
2: Surface
2:
Aerodynamics
3 : Cfx , Cfy , Cfz ,

Cmx , Cmy , Cmz
3:
Bending
Analysis
Figure 6.12: XDSM for the root bending moment constrained problem
Figure 6.13: Bending moment constrained optimization: planform variables only: M=0.5,
e = 0.964
93
Figure 6.14: Bending moment constrained optimization: planform variables only: M=0.7
Figure 6.15: Bending moment constrained optimization: planform and shape variables: M=0.7
94
Figure 6.16: Bending moment constrained optimization: planform and shape variables: M=0.85
wing, which is shown as a black outline. This increase in span is limited by the allowable
root bending moment and comes with a corresponding movement away from an elliptical lift
distribution.
Looking at the M=0.7 case, shown in Figure 6.14, the optimizer has added a significant
amount of sweep to the wing. This causes a trade-off between induced drag and wave drag.
The added sweep reduces the effective Mach number normal to the wing leading edge, thereby
reducing the strength of the leading edge shock. With sufficient sweep, the Mach number
normal to the leading edge of the wing is reduced below the critical Mach number, eliminating
the shock wave and the corresponding wave drag. However, adding sweep increases the effective
root bending moment. Thus, as sweep is added, the span must be reduced to meet the bending
constraint. This is evident in the difference in span between the Mach = 0.5 case shown in
Figure 6.13 and the Mach = 0.7 case shown in Figure 6.14.
When shape variables are added, the need for sweep to reduce wave drag is eliminated and,
as shown in Figure 6.15, the optimizer does not sweep the wing, allowing for a larger span.
Further, there is now a significant change in the Cp distribution on the wing. The individual
sections now show a rooftop pressure profile at the leading edge, allowing the maximum section
Cp to stay below the critical Cp for the wing, thereby eliminating the wave drag on the wing.
Note that the planform is now similar to the optimum M = 0.5 planform in Figure 6.13.
95
At M = 0.85, the optimizer has produced the features typical of supercritical transonic
airfoils for the individual airfoil sections. The sections show a rooftop Cp profile and are highly
aft loaded. However, even with these transonic airfoils, the wave drag is not entirely eliminated,
so the optimizer has introduced some sweep in the design to mitigate this effect. This added
sweep has caused a corresponding reduction in span to maintain the required bending moment.
Note that because of the aft loaded nature of these airfoils, the neutral point of the wing is
significantly forward of the required CG location for trimmed flight, leading to a statically
unstable configuration.
6.3.2
Static Stability Constrained Problems
Cm Constrained Optimizations
The first stability constrained case is based on the traditional definition of static-stability. As
described in Section 5.4, this definition requires the aircraft to have positive pitch stiffness, at
a trimmed state, with positive lift. The formulation for this case is shown in Figure 6.17 an
can be written mathematically as:
minimize
CD
w.r.t. x = , i , A1 , b, , CG% , FFD , ctroot , xtCG , MACt

subject to
(6.4)
CLref CL 0
Cm = 0
Cm 0
Cb Cbref = 0
A2 Aref = 0
xCG xtCG = 0
MAC MACt = 0
croot ctroot = 0
In this problem, the previous trim constrained case is made stricter by requiring that Cm < 0.
As a result, the center of gravity variable is no longer as free to move to trim the aircraft. It
now has to satisfy both the Cm constraint and the Cm constraint. As is shown in the results,
this creates a trade-off for the optimizer to deal with and frequently leads to changes in the
optimal configuration.
As shown in Table 6.4, the optimal wing at Mach = 0.5 is very slightly unstable. In order
to satisfy the Cm constraint, the optimizer adds sweep to the wing, resulting in the wing
shown in Figure 6.18. As discussed in Section 5.4, stabilizing the wing requires manipulating
96

x : , i , A1 , b,
, CG% , FFD ,
MACt , xtCG ,
ctroot
, i , A1 , b ,
, CG% , FFD
0,51:
Optimization
4 : MACt , xtCG , ctroot
1 : i , A1 , b, ,
CG% , FFD
2 : , xtCG , MACt
3 : xtCG , MACt
4:
f : CD
5 : f, c, cc
xCG , M AC ,croot
CL , CD
, Cm
,
z
Cm
, A2
Cb
c : CLref CL
Cmz
cc : xCG xtCG
Cb Cbref (ctroot ) MAC MACt
A2 Aref
croot ctroot
Cm < 0
4 : xCG ,MAC,croot
4 : CL , CD , Cmz ,
Cm , A2
1:
Geometry
2: Surface
2:
Aerodynamics
4 : Cb

Cmx , Cmy , Cmz
3:
Bending
Analysis
Figure 6.17: XDSM for the Cm constrained problem
the relative locations of the CG and the neutral point. Sweeping the wing shifts the neutral
point of the wing further aft. This will nominally increase the stability of the aircraft, if the
CG stays fixed. However, sweeping the wing also shifts the CG further aft. In this thesis, the
extra degree of freedom provided by the CG% variable allows the optimizer to shift the CG
forward to compensate for shift caused by the sweep. This simulates moving the positions of
the internal systems and payload to provide a feasible CG range. Unfortunately, as shown in
Equation (5.20) shifting the distance between the CG and the neutral point also affects the
trim of the aircraft. However, the addition of sweep to the wing also allows the optimizer to
use washout of the wing tip to provide the pitch moment needed to trim the aircraft in its new
state. Once shape variables are added, this additional sweep is not necessary to stabilize the
aircraft. As shown in Figure 6.19, the optimizer is able to use the shape variables to modify the
Cp to achieve the same effects. In this case, the Cp distribution, especially near mid-span, has
developed negative lift at the trailing edge, which alters the CmNP of the wing. This allows the
optimizer to shift the CG forward stabilizing the wing using CG% while still maintaining
a trimmed state. Because the wing is in a stable trimmed state without sweep, there is no
increase in the bending moment associated with sweep, therefore the optimizer is again able to
increase the span from the baseline value of three meters, thereby increasing the aspect ratio
of the wing and reducing the resulting induced drag.
Moving to the Mach = 0.7 case, the results in Table 6.6 show that for the planform only
case, the Cm constraint is not active. Because of the amount of sweep necessary to counteract
the wave drag present at the transonic Mach number, the optimal wing from the bending
97
Figure 6.18: Cm constrained optimization: planform variables only: M=0.5, e = 0.963
Figure 6.19: Cm constrained optimization: planform and shape variables: M=0.5, e = 0.968
98
Figure 6.20: Cm constrained optimization: planform variables only: M=0.7
constrained case has a positive static margin. Thus, the solution with the Cm constraint,
shown in Figure 6.20, is essentially the same as the bending constrained solution shown in
Figure 6.14. The same is not true of the shape optimization cases at Mach = 0.7. The bending
constrained case from Figure 6.15 has a negative static margin. Thus, Cm constrained case
shown in Figure 6.21, the optimizer uses the shape variables to alter the Cp distribution of the
wing to trim the wing in a stable state. In this case, in addition to adding the negative lift
at the trailing edge, the optimizer has flattened the Cp distribution at the leading edge of the
wing, eliminating the pressure peak and thereby reducing the wave drag. As a result, the wing
is once again unswept, allowing for the maximum span extension possible.
The Mach = 0.85 case, shown in Figure 6.22, is physically a more challenging problem for
the optimizer to solve. As in the previous bending constrained case, the optimizer is attempting
to reduce the wave drag with the airfoil shape. However, in this case, the optimizer is unable
to produce the heavily aft loaded airfoils that are optimal for this case because they lead to an
unstable design. As a result, the optimizer is forced to compromise between the reducing the
drag and maintaining Cm < 0. These compromises show up in different ways at the different
sections of the wing. At the wing root, the Cp profile is spread over the entire chord of the
section with significant lift generated at both the leading and trailing edges with relatively
little lift generated mid chord. The mid-wing section exhibits a rooftop Cp distribution over
99
Figure 6.21: Cm constrained optimization: planform and shape variables: M=0.7

the front half of the foil, helping to reduce the wave drag of the wing. However, the addition of
the Cm constraint prevents the optimizer from extending this trend over the entire foil. As a
result there is a fairly constant amount of lift generated over the forward two thirds of the foil
with negative lift generated at the trailing edge. The optimizer has also added more sweep to
the wing than in the comparable bending constrained case. As discussed earlier, this reduces
the effective Mach number the wing sees and allows tip washout to contribute to wing trim.
In addition to this general discussion, there are two specific secondary characteristics that are
worth highlighting. First, the optimizer is adding a significant loading to the bottom surface
of the leading edge of the wing at the root. This forward loading helps reduce the moment of
the root section. Also, the optimizer has developed an interesting inflection in the camber of
the mid-wing airfoil. At the trailing edge, the airfoil starts to develop the high camber shape
typical of transonic airfoils, but partway to the trailing edge the foil develops reflex to help
reduce the moment of the wing.
Static Margin Constrained Optimizations
As discussed in the previous chapter, the Cm constrained formulation does not accomplish
exactly what is desired. This formulation of the constraint is only able to guarantee a neutrally
stable result. As shown in the Mach = 0.5 planform-only optimization case, when the constraint
100
Figure 6.22: Cm constrained optimization: planform and shape variables: M=0.85
is active, it produces a Cm of zero, giving a neutrally stable aircraft. To generate designs that
have positive static-stability, a formulation based on static margin is used. That formulation is
101
as follows:
minimize
CD
w.r.t. x = , i , A1 , b, , CG% , FFD

t
ctroot , xtCG , MACt , CLt , Cm
subject to
(6.5)
CLref CL 0
Cm = 0
Knref Kn 0
Cb Cbref = 0
A2 Aref = 0
xCG xtCG = 0
MAC MACt = 0
croot ctroot = 0
CL CLt = 0
t
Cm Cm
=0
Note that in the above formulation, Kn is the static margin and that additional constraints
t
have been added for CLt and Cm
corresponding to the additional IDF target variables that
have been added for the computation of the static margin constraint. A graphical depiction of
solution process for this formulation is shown in Figure 6.23.
The planform-only, static margin constrained optimization at Mach = 0.5, shown in Figure 6.24, shows an increase in sweep, just as the Cm case did. In this case, however, the
increase in sweep is much more significant. Just as before, the optimizer is adding sweep and
wash-out to trim the aircraft for a more forward CG position. In this case, the magnitude of the
changes is larger to compensate for the correspondingly large separation between the neutral
point and CG for a 5% static margin. As with the Cm case, adding shape variables to the
problem allows the optimizer to satisfy the stability constraint without sweeping the wing. As
a result, the static margin constrained optimal solution is similar to the optimal solution for
the Cm constrained case.
The solution for the planform-only optimization at Mach = 0.7 continues the trend of the
Cm constrained case. The optimizer adds sweep and twist to create separation between the
neutral point of the wing and the CG location of the aircraft while maintaining a trimmed
state. As can be seen in Figure 6.26 there is now visible separation between the neutral point
and the CG location.
Again, the addition of shape variables allows the optimizer to un-sweep the wing without
102
x : , i , A1 , b,
, CG% , FFD ,
MACt , xtCG ,
t
ctroot , CLt , Cm
, i , A1 , b ,
, CG% , FFD
0,61:
Optimization
t
5 : ctroot , MACt , xtCG ,CLt ,Cm
1 : i , A1 , b, ,
CG% , FFD
2 : , xtCG , MACt
3 : xtCG , MACt
t
4 : CLt , Cm
5:
f : CD
6 : f, c, cc
xCG , M AC ,croot
CL , CD
, Cm
,
z
CL , Cm
, A2
cc : xCG xtCG
c : CLref CL
MAC MACt
Cmz
croot ctroot
Knref Kn
CL CLt
Cb Cbref (ctroot )
t
Cm Cm
A2 Aref
5 : xCG ,MAC,croot
5 : CL , CD , Cmz ,
CL , Cm , A2
Cb
5 : Cb
Kn
5 : Kn
1:
Geometry
4: MAC
2: Surface
2:
Aerodynamics

Cmx , Cmy , Cmz
3:
Bending
Analysis
4:
Stability
and
Control
Figure 6.23: XDSM for the Kn constrained problem
Figure 6.24: Kn constrained optimization: planform variables only: M=0.5, e = 0.944
103
Figure 6.25: Kn constrained optimization: planform and shape variables: M=0.5, e = 0.968
Figure 6.26: Kn constrained optimization: planform variables only: M=0.7
104
Figure 6.27: Kn constrained optimization: planform and shape variables: M=0.7
sacrificing performance. In this case the optimal solution has a very small amount of sweep.
This is possibly due to the increased difficulty of satisfying the trim constraint at the higher
Mach number.
The Mach = 0.85 case shows similar characteristics to the previously discussed Cm constrained case. Once again the root section has a distributed Cp profile with lift split between
the fore and aft portions of the airfoil. The mid-section foils are show a rooftop profile on the
forward portion of the foil and exhibit some reflex at the trailing edge, while the tip foil also
exhibits some reflex. Note that once again the optimizer has added a significant amount of
sweep to the wing and reduced the span accordingly.
6.3.3
Dynamic Stability Constrained Problems
The dynamic stability formulation used in this work is based on the CAP parameter described
in Section 5.4.3. This formulation builds on the previously described static margin formulation
105
Figure 6.28: Kn constrained optimization: planform and shape variables: M=0.85

and can be written as:
minimize
CD
w.r.t. x = , i , A1 , b, , CG% , FFD , ctroot , xtCG , MACt ,

t
t
t
t
CD
, CLt , Cm
, Cm
, Cm
subject to
CLref CL 0
Cm = 0
Knref Kn 0
0.085 CAP 3.6
0.3 sp 2.0
A2 Aref = 0
xCG xtCG = 0
MAC MACt = 0
croot ctroot = 0
t
=0
CD CD
CL CLt = 0
t
Cm Cm
=0
t
Cm Cm
=0
t
Cmq Cm
=0
q
(6.6)
106

x : , i , A1 , b,
, CG% , FFD ,
MACt , xtCG ,
t
ctroot , CD
, CLt ,
t
t
t
Cm
, Cm
, Cm
q
, i , A1 , b ,
, CG% , FFD
0,61:
Optimization
t
, CLt ,
5 : MACt , xtCG , ctroot , CD
t
t
t
, Cm
, Cm
Cm
q
1 : i , A1 , b, ,
CG% , FFD
2 : , xtCG , MACt
3 : xtCG , MACt
t
, CLt ,
4 : CD
t
t
t
Cm
, Cm
, Cm
5:
f : CD
6 : f, c, c
xCG , M AC ,croot
c : CLref CL
Cmz
Knref Kn
Cb Cbref (ctroot )
A2 Aref
CAP
sp
cc : xCG xtCG
MAC MACt
croot ctroot
t
CD CD
CL CLt
t
Cm Cm
t
Cm Cm
t
Cmq Cm
q
5 : xCG ,MAC,croot
5 : CL , CD , Cmz ,
CL , Cm , Cm ,
Cmq , A2
CL , CD
, Cm
,
z
CL , Cm
, A2
Cb
5 : Cb
Kn , CAP , sp
5 : Kn , CAP, sp
1:
Geometry
4: MAC
2: Surface
2:
Aerodynamics

Cmx , Cmy , Cmz
3:
Bending
Analysis
4:
Stability
and
Control
Figure 6.29: XDSM for the CAP constrained problem
Relative to the static margin optimization, two primary constraints have been added, a CAP
constraint and a sp constraint. These constraints define the appropriate limits for the dynamic
stability parameters. The range of allowable CAP values constrains the allowable short-period
frequency, and the damping ratio, sp , is constrained directly. Also, IDF constraints have been
t , C t , and C t to reflect the addition of these variables for the computation of the
added for CD
m
mq
dynamic stability parameters. A graphical depiction of the solution process for this formulation
is shown in Figure 6.29.
The CAP constrained optimizations add the dynamic stability constraints to the problem.
This requires the consideration of extra stability derivatives as well as the mass moment of
inertia of the aircraft during the optimization. The optimal solution for the CAP constrained
planform-only optimization at Mach = 0.5 shown in Figure 6.30 is essentially the same
as the static margin result from Figure 6.24. This results from the fact that the static margin
constraint is still active and that moment of inertia multiplier is able to raise the moment
of inertia sufficiently to satisfy the CAP and damping constraints. If the moment of inertia
multiplier were limited to a smaller value, the optimal solution would likely be more highly
swept with a larger static margin.
An interesting result from the addition of the dynamic constraints is that the addition of
the shape variables no longer produces an unswept wing. As shown in Figure 6.31, the optimal
solution now has almost 20 degrees of sweep. This results largely from the need to maintain a
107
Figure 6.30: CAP constrained optimization: planform variables only: M=0.5, e = 0.942
high enough moment of inertia to satisfy the damping requirements. Another interesting side
effect of this result is that the section profiles no longer exhibit any reflex at the trailing edge.
This comes from the fact that the optimizer is now able to use sweep and tip washout to trim
the aircraft without any additional penalty, so the Cp distribution is not required to add reflex
to do so.
The planform-only Mach = 0.7 CAP constrained case produces a result similar to static
margin constrained case. As with the Mach = 0.5 case, the moment of inertia multiplier is
able to increase the moment of inertia to the point where the CAP constraint and damping
constraints are satisfied.
As with the previous CAP constraint cases, at Mach = 0.7 the shape variables are not sufficient to allow the optimizer to reduce the sweep of the wing to zero. Again the optimal results
from the bending constrained case and the static-stability constrained cases have insufficient
damping to satisfy the dynamic constraints. In this case, however, the added sweep causes the
static margin constraint to be inactive, though the mid-span section still develops a significant
amount of reflex.
At Mach = 0.85 the CAP constrained result is essentially the same as the static margin
constrained result. Because the shape variables were not able to completely eliminate the sweep
at this higher Mach number, the static margin case has enough sweep to ensure dynamic stability
108
Figure 6.31: CAP constrained optimization: planform and shape variables: M=0.5, e = 0.970
Figure 6.32: CAP constrained optimization: planform variables only: M=0.7
109
Figure 6.33: CAP constrained optimization: planform and shape variables: M=0.7
given a sufficient moment of inertia. Thus, the optimal result exhibits the same combinations
of Cp distributions and planform variables as the static margin constrained case.
6.4
Results Tables
To compare the results in a more quantitative fashion, the results of the optimizations are
presented in tabular form below. For the simpler cases where the various stability parameters
were not included in the optimizations, the remaining parameters have been calculated for
comparison purposes. A short summary of the most significant parameters is shown in Tables 6.4
through 6.8 to support the following discussion. A full compilation of the results for each
test case is shown in Appendix B. Note that the drag values presented are from steady-state
simulations of the optimal shapes for each case. The remainder of the values are computed
using the time-spectral methods used in the optimizations.
6.4.1
Mach = 0.5 Cases
The key results for the Mach = 0.5 cases are shown in Tables 6.4 and 6.5. The first parameter
to consider is the drag for each of these cases. The drag for the planform-only cases shown
in Table 6.4 show precisely what one would expect. The optimal elliptical solution from the
110
Figure 6.34: CAP constrained optimization: planform and shape variables: M=0.85
twist-only optimization has a lower drag than the baseline case. The predicted value of drag for
this case is 0.00489, within 2.5% of the theoretical value of 0.00477, leading to a span efficiency
of 0.977. The remaining error in this case can be attributed to numerical errors artificial
dissipation for example as well as the three-dimensional flow at the wing tip.
The added flexibility of the span variable, introduced in the bending moment constrained
case, allows the optimizer to further reduce the drag in the baseline aerostructural case. Considering the problem from this perspective allows an additional 1.4% reduction in drag. Once
the stability constraints are included in the optimization, the drag increases. This increase in
drag comes from the changes necessary to make the aircraft stable. In the case of the Cm
constraint, the changes are relatively minor, so the increase in drag is correspondingly small.
However, for the static margin and CAP constrained cases, the necessary increase in sweep is
much more significant, causing a much larger increase in the drag.
Also of note are the span efficiency values for the various optimal solutions. In each case,
the span efficiency is lower than for the elliptic solution. This is unsurprising, as the lift
distributions in these case deviate from elliptical. While the span efficiencies are lower for the
bending moment and Cm constrained cases, the increased aspect ratio of these designs makes
up for the loss of span efficiency, allowing for a slightly lower overall drag coefficient. In the case
of the static margin and CAP constrained cases, both the aspect ratio and the span efficiency
111

are lower than in the elliptical case, leading to a higher drag coefficient.
Looking at the values of the various stability constraints, the results show that the staticstability constraints are active in each of the stability constrained cases. This leads to a neutrally
stable result in the Cm constrained case and results with the desired static margin in the last
two cases. Also note that the three cases without the stability constraints are statically unstable.
Finally, the CAP constrained case is essentially identical to the static margin constrained case.
This is primarily because the inertia modifier is able to increase the moment of inertia sufficiently
to satisfy the damping constraint without modifying the geometry.
Parameter
Baseline
Elliptic
Bending
Cm
Kn
CAP
CD
0.004954
0.004888
0.004817
0.004844
0.005123
0.005133
Cm
Kn (%)
e (%)
0.002242
0.013867
0.029788
-0.000000
-0.246456
-0.246417
-0.044805
-0.277744
-0.595908
0.000000
4.999788
5.000118
0.963761
0.976806
0.963537
0.962759
0.943660
0.941934
CAP
0.000258
-0.015125
-0.037377
0.003236
0.167942
0.168177
sp
8.047202
-0.000000
-0.000000
2.314511
0.301925
0.301636
n (rad/s)
0.065557
-0.000000
-0.000000
0.233514
1.651115
1.652217
Izz (kg m )
244.664280
242.153023
228.835160
247.707360
494.760972
493.932604
(deg.)
3.434484
4.042761
7.545133
7.665806
9.065110
8.899279
(deg.)
0.000000
0.000000
0.000000
3.291240
19.280100
19.250614
b (m)
PIzz
3.000000
3.000000
3.042917
3.035421
2.981338
2.981104
10.000000
10.000000
10.000000
10.000000
10.000000
10.000000
Table 6.4: NACA 0012 wing: planform only optimization results: 1107k cells, M = 0.5
The results for the shape optimization cases in Table 6.5 show a very different picture. In
these cases, both of the static-stability constrained cases have essentially the same drag as the
planform-only bending constrained case. This comes from the fact that both the aspect ratio
and span efficiency are similar to the planform-only bending constrained case. Further, the
results show that neither of the static stability constraints are active for this case. However,
the Cp distributions from Figures 6.18, 6.20, 6.24 and 6.26 clearly show that the solution
is different for the planform-only cases and the shape optimization cases. This leads to the
conclusion that, for subsonic cases, airfoil shape can be used to trim the aircraft for a wide
variety of CG locations with little penalty in terms of drag. This supports the idea that a
model with more stringent limits on CG could be used in the context of this case and still allow
the optimizer to find feasible solutions, a fact that is not guaranteed for the planform-only
cases.
The same is not true for the dynamic stability constrained case. The results for this case
show that both the damping constraint and the static margin constraint are active. The addition
112
of the dynamic constraint pushes the optimizer to increase the moment of inertia of the aircraft,
which causes an increase in sweep of the wing. Because of the root bending moment constraint,
a reduction of the span ensues, causing a 7% increase in drag even though the span efficiency is
similar to the two static-stability constrained cases. Note that the moment of inertia multiplier
is also at its upper bound. Therefore, an increase in this limit may reduce the required sweep
and thus the amount of drag increase. This simply emphasizes the importance of developing
an improved moment of inertia prediction technique for any future work in this area.
Cm
Kn
CAP
CD
0.004807
0.004801
0.005149
Cm
-0.665371
-0.752988
-0.246065
Kn (%)
13.345331
15.092049
5.000230
0.967616
0.968178
0.970392
Parameter
e (%)
CAP
0.550552
0.586254
0.164005
sp
0.166841
0.161038
0.299986
n (rad/s)
3.043741
3.142237
1.621730
Izz (kg m )
380.015557
403.325544
517.368619
(deg.)
3.064103
3.057671
3.051991
(deg.)
0.000000
0.000000
19.888040
b (m)
PIzz
3.039381
3.040392
2.932521
10.000000
10.000000
10.000000
Table 6.5: NACA 0012 wing: shape optimization results: 1107k cells, M = 0.5
6.4.2
Mach = 0.7 Cases
The planform-only cases at Mach = 0.7 shown in Table 6.6 are not as interesting as the
Mach = 0.5 cases. The optimal bending constrained result has a static margin of roughly 0.7%,
satisfying the Cm constraint. As a result, the optimal results for the bending constrained case
and the Cm constrained case are essentially the same. The use of the static margin constraint
requires an increase in sweep to provide the necessary static margin. However, this additional
sweep, along with a relatively large moment of inertia are sufficient to satisfy the CAP and
damping constraints. Thus, the CAP constrained case has essentially the same optimal solution
as the static margin constrained case.
Examining the shape optimization cases at Mach = 0.7 shown in Table 6.7 , the results
once again show what one would expect. The bending constrained case produces a drag result
that is within one count of the optimal solutions at Mach = 0.5. This highlights the ability
of the shape optimizations to modify the airfoil shapes to eliminate the wave drag caused
113

Parameter
Baseline
Bending
Cm
Kn
CAP
CD
0.005570
0.005207
0.005207
0.005325
0.005325
Cm
0.087088
-0.043966
-0.043519
-0.284135
-0.284114
Kn (%)
-1.491942
0.745596
0.738342
4.999817
4.999817
CAP
-0.055895
0.020482
0.020155
0.087320
0.087312
sp
-0.000000
0.672371
0.677448
0.310233
0.310233
n (rad/s)
-0.000000
0.631212
0.626015
1.272602
1.272483
417.494607
623.660078
627.885797
963.347405
963.470201
2.933511
7.613268
7.615328
8.590863
8.557310
(deg.)
0.000000
14.106302
14.119718
22.861602
22.864088
b (m)
3.000000
2.980515
2.980963
2.953919
2.953687
16.184566
16.184566
16.184566
16.184566
16.184566
Izz (kg m2 )
(deg.)
PIzz
Table 6.6: NACA 0012 wing: planform only optimization results: 1107k cells, M = 0.7
by transonic flow. While the bending constrained case is statically unstable, the addition of
the static stability constraints does not cause any significant change in the drag. Thus, the
optimizer is able to find airfoil profiles that are able to both eliminate the transonic wave drag
and satisfy the moment criteria necessary for static stability at the same time. As with the
Mach = 0.5 case, there is enough flexibility in the airfoil shape to allow a wide variety of CG
locations.
As with the Mach = 0.5 optimization cases, the CAP constraint forces the optimizer to
add sweep to the wing to increase the damping ratio of the aircraft. As a result, there is a
corresponding increase in the amount of drag produced by the wing. Note also that the inertia
modifier is double what it was for the Mach = 0.5 case. Further, the fact that it is at its upper
limit once again highlights the need for more investigation into methods for estimating this
parameter.
6.4.3
Mach = 0.85 Cases
The Mach = 0.85 results, shown in Table 6.8, exhibit some interesting characteristics. Once
again, the bending constrained optimal solution is massively unstable, with a static margin
of -23.5%. The addition of the static stability constraints remedies this deficiency, with both
constraints producing optimal solutions at the limiting value of the constraints. However, in
this case the drag does not increase as the additional stability constraints are added. In fact,
the most constrained case produces the solution with the lowest drag. This might be explained
by the larger spans obtained in the Kn and CAP optimal solutions. Finally, as with some of the
previous cases, the moment of inertia multiplier is sufficient to satisfy the CAP and damping
114

Parameter
Bending
Cm
Kn
CAP
CD
0.004973
0.004935
0.004939
0.005090
Cm
0.539667
-0.275911
-0.298319
-0.302163
Kn (%)
-9.112122
4.634752
4.999844
5.121750
CAP
-0.378485
0.121639
0.129099
0.099087
sp
-0.000000
0.286944
0.277583
0.299995
n (rad/s)
Izz (kg m2 )
(deg.)
-0.000000
1.561267
1.609959
1.389344
380.807451
601.407689
611.432529
847.763119
2.987315
3.032540
3.008049
3.282016
(deg.)
0.000000
0.084255
0.529970
13.921081
b (m)
3.010364
3.030450
3.029077
2.972135
20.000000
20.000000
20.000000
20.000000
PIzz
constraints based on the static margin constrained optimal solution, therefore the static margin
constrained optimum and the CAP constrained optimum are essentially the same.
Parameter
Baseline
Bending
Cm
Kn
CAP
CD
0.039898
0.005792
0.005733
0.005640
0.005619
Cm
0.055392
2.566048
0.000007
-0.484800
-0.489347
Kn (%)
-0.729501
-23.522592
-0.000079
4.999816
5.051372
CAP
-0.036492
-0.975527
0.000523
0.086524
0.086135
sp
-0.000000
-0.000000
3.979094
0.308502
0.309052
n (rad/s)
-0.000000
-0.000000
0.121288
1.644602
1.639253
316.346371
416.989558
1137.655818
993.139864
1010.020857
(deg.)
2.169667
3.437140
3.009218
3.643078
3.665365
(deg.)
0.000000
7.609797
24.783732
20.849705
21.144092
b (m)
3.000000
2.881879
2.857234
2.901532
2.898588
21.988981
21.988981
21.988981
21.988981
21.988981
Izz (kg m2 )
PIzz
6.5
Summary
This chapter presented an optimization study exploring the effects of static and dynamic stability constraints on the optimal shape of flying wings. The results showed that the stability
constraints can have a significant impact on the optimal shape of the wing, usually causing a corresponding increase in drag. However, for cases in the subsonic regime and the lower end of the
115
transonic regime, the study showed that airfoil shape can be used to satisfy the static-stability
constraints without significant degradation in performance. As a result, in these flow regimes,
the optimizer tends to prefer using airfoil shape rather than wing sweep and twist to satisfy the
stability requirements. At higher transonic Mach numbers, the study showed that degradation
in performance seems to be unavoidable and that the addition of sweep is necessary, regardless
of the airfoil shape, to achieve satisfactory results. The same can not be said for the dynamic
stability constrained cases. In these cases, the study showed that for the subsonic and low-end
transonic cases, the optimizer was forced to add sweep to the wing, regardless of the airfoil
shape, in order to raise the damping ratio of the aircraft to acceptable levels. At the higher
end of the transonic regime, the static stability constrained results required sufficient sweep to
be feasible for both the static and dynamic stability constraints. Thus, the results of this study
underline the importance of considering both static and dynamic stability considerations in the
design of flying-wing aircraft.
Chapter 7
Conclusions
The work described in this thesis demonstrated a method for conducting stability-constrained
aerodynamic shape optimization of aircraft configurations using three-dimensional computational fluid dynamics techniques. To facilitate this method, two approaches for computing
stability derivatives using CFD were developed. The first approach, based on the ADjoint sensitivity analysis method, was shown to be efficient for accurately computing full sets of static
and dynamic stability derivatives. While this technique is very effective for analyzing the stability derivatives of a given configuration, the use of the ADjoint method in the computation of the
stability derivatives makes it prohibitively expensive to compute the gradients of these stability
derivatives for use in a gradient based optimization context. The second approach was developed
using simple linear regression techniques together with a time-spectral CFD solution. Because
this technique does not require an adjoint method to compute the stability derivative values, it
is possible to apply an adjoint method to the combined stability derivative/time-spectral flow
solution process, facilitating efficient gradient calculations for optimization. Further, because
the time-spectral CFD solution includes unsteady flow information, it is possible to estimate
the transient, or dot, stability derivatives, which can play an important role in assessing dynamic stability. Based on these last two qualities, the time-spectral stability derivative method
was selected for further development and was integrated into an aerodynamic optimization
framework.
To facilitate efficient optimization with this framework, a time-spectral adjoint method for
sensitivity analysis was developed and verified using the complex-step method. The computed
derivatives match reference values to between 7 and 12 digits, giving more than enough accuracy
for the optimizations performed in this work. The efficiency of the sensitivity analysis method
was demonstrated on test cases up to roughly 40 million flow states. The largest adjoint test
case analyzed for the scaling study was solved in under 30 minutes on 224 processors.
Using these techniques, a study was conducted to explore the effects of a series of static and
116
Chapter 7. Conclusions
117
dynamic stability constraints on the optimal design of a flying wing configuration. The study
showed that the various constraints can have a significant effect on the optimal solution. Most
optimizations conducted without the stability constraints resulted in statically unstable designs
that would incur significant drag penalties in order to be made statically stable. However, in the
cases with static stability constraints, the addition of shape variables to the optimization was
able to counteract the adverse effects of the stability constraints by generating reflex airfoils,
virtually eliminating the drag penalty associated with the static-stability constraints in most
cases. The same can not be said for the dynamic stability constrained cases. Increases in sweep
and the corresponding increases in drag were required in most of these cases to satisfy
the damping ratio constraint associated with dynamic stability.
Given the relatively simplistic nature of the center of gravity and moment of inertia calculations used in this work, the specific results shown here must be taken with a certain degree of
caution. However, the fact that optimal wing designs tend to violate the stability constraints if
they are not enforced clearly shows that static and dynamic stability constraints have an important impact on aerodynamic performance. Thus, while the work presented here is by no means
the final answer to the problem, the results shown demonstrate the importance of including
stability constraints in aerodynamic optimization, specifically when looking at stability-critical
aircraft such as flying wings.
Chapter 8
Future Work
The results of the work presented in this thesis suggest several areas for future exploration. An
overview of some of these topics is included below.
Viscous Stability Derivative Computations: The work in this thesis has been conducted
entirely with an inviscid Euler flow solver. The predictive capabilities of the stability
derivative methods presented in this work would be significantly improved if combined
with an efficient RANS flow solver. The use of a RANS solver would allow more complete
validation against experimental data, which would allow for more concrete assessment
of the accuracy and validity of the methods. By using a RANS solver, the range of
flow conditions which could be accurately modeled would be significantly increased. This
would help in evaluating the extreme flow conditions which are often critical in the design
process. In addition to considering the analysis with a RANS flow solver, an efficient
adjoint will be needed in order to optimize flows in the RANS regime.
Mass and Moment of Inertia Models: As discussed in the optimization study, there are
several cases where the estimated values of CG location and moment of inertia affect the
optimal result. This is particularly true when the dynamic stability constraints are considered. Therefore, it is important that further exploration in this area include improved
estimates for these quantities.
Dynamic Stability Constraints: While this thesis has shown that it is possible to optimize
an aircraft using high-fidelity CFD and dynamic stability constraints, this work has really
only scratched the surface of possible formulations for these constraints. An interesting
avenue of exploration would be to investigate the effects of various stability formulations
on both the optimization process and its results. This would likely also necessitate the
consideration of lateral stability derivatives and the impact that they have on design.
118
Chapter 8. Future Work
119
Multi-point optimization: While this work has demonstrated that stability-constrained

optimization is possible, the current work only considers a single design point. In real
world design, many design and off design points must be evaluated to properly assess
the performance of a given design. Therefore, extending this framework to include multipoint design would be a worthwhile endeavor. One major addition that would be necessary
to allow this is the consideration of control effectors. This would be needed to allow the
aircraft to be properly trimmed at various design points. Note that the addition of control
effectors would also add additional possibilities in terms of controller design.
Extended applications: The study presented here is applied to a relatively simple case.
While the broad strokes of the conclusions developed here are applicable to a variety
of cases, it would be interesting to explore the effects of stability-constraints on a more
realistic design, where more is known about the aircraft.
References
[1] Daniel Sarel Agenbag. Longitudinal handing characteristics of a tailless gull-wing aircraft.
Masters thesis, University of Pretoria, Pretoria, South Africa, July 2008.
[2] Airbus. Global market forecast: 2000 - 2019. Technical report, Airbus Industrie, France,
2000.
[3] W. Kyle Anderson and Daryl L. Bonhaus. Airfoil design on unstructured grids for turbulent flows. AIAA Journal, 37(2):185191, 1999.
[4] F. Angrand. Optimum design for potential flows. International Journal for Numerical
Methods in Fluids, 3(3):265282, 1983.
[5] Deric A. Babcock and Dr. Andrew S. Arena Jr. Estimating aircraft stability derivatives
through finite element analysis. AIAA Paper 2004-5174, August 2004.
[6] Satish Balay, Kris Buschelman, Victor Eijkhout, William D. Gropp, Dinesh Kaushik,
Matthew G. Knepley, Lois Curfman McInnes, Barry F. Smith, and Hong Zhang. PETSc
users manual. Technical Report ANL-95/11 - Revision 2.1.5, Argonne National Laboratory, 2004.
[7] William Bihrle. A handling qualities theory for precise flight path control. AFFDL
Technical Report AFFDL-TR-65-198, June 1966.
[8] C. Bischof, A. Carle, G. Corliss, A. Grienwank, and P. Hoveland. ADIFOR: Generating
derivative codes from Fortran programs. Scientific Programming, 1(1):1129, 1992.
[9] Christian H. Bischof, H. Martin B
ucker, Bruno Lang, A. Rasch, and Andre Vehreschild.
Combining source transformation and operator overloading techniques to compute derivatives for MATLAB programs. In Proceedings of the Second IEEE International Workshop
on Source Code Analysis and Manipulation (SCAM 2002), pages 6572, Los Alamitos,
CA, USA, 2002. IEEE Computer Society.
120
References
121
[10] Christian H. Bischof, Lucas Roh, and Andrew Mauer. ADIC An extensible automatic
differentiation tool for ANSI-C. SoftwarePractice and Experience, 27(12):14271456,
1997.
[11] C. L. Bloebaum, P. Hajela, and J. Sobieszczanski-Sobieski. Non-hierarchic system decomposition in structural optimization. Engineering Optimization, 19(3):171186, 1992.
[12] Boeing. Current market outlook: 2009 - 2028. Technical report, Boeing Commercial
Airplanes, Seattle, WA, 2009.
[13] A. L. Bolsunovsky, N. P. Buzoverya, B. I. Gurevich, V. E. Denisov, A. I. Dunaevsky, L. M.
Shkadov, O. V. Sonin amd A. J. Udzhuhu, and J. P. Zhurihin. Flying wing problems
and decisions. Aircraft Design, 4(4):193219, 2001.
[14] Robert D. Braun and Ilan M. Kroo. Development and application of the collaborative
optimization architecture in a multidisciplinary design environment. In N. Alexandrov
and M. Y. Hussaini, editors, Multidisciplinary Design Optimization: State-of-the-Art,
pages 98116. SIAM, 1997.
[15] Patrick Brezillon, Jean-Francois Staub, Anne-Marie Perault-Staub, and Gerard Milhaud.
Numerical estimation of the first order derivative: approximate evaluation of an optimal
step. Computers and Mathematics with Applications, 7(4):333347, 1981.
[16] Howard P. Buckley, Beckett Y. Zhou, and David. W. Zingg. Airfoil optimization using
practical aerodynamic design requirements. Journal of Aircraft, 47(5):17071719, 2010.
[17] Alan Carle and Mike Fagan. ADIFOR 3.0 overview. Technical Report CAAM-TR-00-02,
Rice University, 2000.
[18] S. Chai, P. Crisafulli, and W.H.Mason. Aircraft center of gravity estimation in conceptual/preliminary design. In Proceedings of the 1st AIAA Aircraft Engineering, Technology,
and Operations Congress, Los Angeles, CA, 1995. AIAA 95-3882.
[19] Eric F. Charlton. Numerical stability and control analysis towards falling-leaf prediction
capabilities of SPLITFLOW for two generic high-performance aircraft models. NASA
Contractor Report CR-1998-208730, September 1998.
[20] Seongim Choi, Mark Potsdam, Kihwan Lee, Gianluca Iaccarino, and Juan J. Alonso.
Helicopter rotor design using a time-spectral and adjoint-based method. AIAA Paper
2008-5810, September 2008. Proceedings of the 12th AIAA/ISSMO Multidisciplinary
Analysis and Optimization Conference, British Columbia, CA.
References
122
[21] E. J. Cramer, J. E. Dennis, P. D. Frank, R. M. Lewis, and G. R. Shubin. Problem

formulation for multidisciplinary optimization. SIAM Journal on Optimization, 4(4):754
776, 1994.
[22] Russell M. Cummings, Adam Jirasek, Kristian Petterson, and Stefan Schmidt. SACCON
static and dynamic motion flow physics simulation using cobalt. In Proceedings of the
28th AIAA Applied Aerodynamics Conference, Chicago, IL, 2010. AIAA 2010-4691.
[23] J. Driver and D. W. Zingg. Numerical aerodynamic optimization incorporating laminarturbulent transition prediction. AIAA Journal, 45(8):18101818, 2007.
[24] Kivanc Ekici and Kenneth C. Hall. Nonlinear analysis of unsteady flows in multistage
turbomachines using harmonic balance. AIAA Journal, 45(5):10471057, 2007.
[25] ESDU. Geometrical properties of cranked and straight tapered wing planforms. Technical
Report Item No. 76003, ESDU International plc, London, UK, January 1976.
[26] ESDU. ESDU dynamics series. Technical Report Various Items, ESDU International plc,
London, UK, 2010.
[27] Bernard Etkin. Dynamics of Atmospheric Flight. Dover Publications, Mineola, New York,
2000.
[28] John W. Gallman, Stephen C. Smith, and Ilan M. Kroo. Optimization of joined-wing
aircraft. Journal of Aircraft, 30:897905, 1993.
[29] Frank H. Gern, Amir H. Naghshineh-Pour, Erwin Sulaeman, Rakesh K. Kapania, and
Raphael T. Haftka. Structural wing sizing for multidisiplinary design optimization of a
strut-braced wing. Journal of Aircraft, 38:154163, 2001.
[30] Davendra P. Ghate and Michael B. Giles. Effcient Hessian calculation using automatic
differentiation. AIAA Paper 2007-4059, June 2007. Proceedings of the 18th AIAA Applied
Aerodynamics Conference, Miami, FL.
[31] Amrit Raj Ghosh. Solution to three-dimensional thin-layer NavierStokes equations in
rotating coordinates for flow through turbomachinery. Masters thesis, Mississippi State
University, 1996.
[32] Philip E. Gill, Walter Murray, and Michael A. Saunders. Snopt: An sqp algorithm for
large-scale constrained optimization. SIAM Review, 47(1):99131, 2005.
[33] Andrew G. Godfrey and Eugene M. Cliff. Direct calculation of aerodynamic force derivatives: A sensitivity-equation approach. AIAA Paper 98-0393, AIAA, 1998.
References
123
[34] Arathi K. Gopinath and Antony Jameson. Time spectral method for periodic unsteady
computations over two- three dimensional bodies. AIAA Paper 2005-1220, January 2005.
[35] Arathi Kamath Gopinath. Efficient Fourier-Based Algorithms for Time-Periodic Unsteady Problems. PhD thesis, Stanford University, Stanford, CA 94305, April 2007.
[36] Joel M. Grasmeyer. Multidisciplinary design optimization of a strut-braced wing aircraft.
Masters thesis, Virginia Polytechnic Institue and State University, Blacksburg, VA, April
1998.
[37] Andreas Griewank. Evaluating Derivatives. SIAM, Philadelphia, 2000.
[38] Andreas Griewank, David Juedes, and Jean Utke. Algorithm 755: ADOL-C: A package
for the automatic differentiation of algorithms written in C/C++. ACM Transactions on
Mathematical Software, 22(2):131167, 1996.
[39] Raphael T. Haftka. Simultaneous analysis and design. AIAA Journal, 23(7):10991103,
July 1985.
[40] Sohrab Haghighat, Hugh H. T. Liu, and Joaquim R. R. A. Martins. Aeroservoelastic
design optimization of a flexible wing. Journal of Aircraft, 2011. In Press.
[41] Kenneth C. Hall, Jeffery P. Thomas, and W.S.Clark. Computation of unsteady nonlinear
flows in cascades using a harmonic balance technique. AIAA Journal, 40(5):879886,
2002.
[42] L. Hascoet and V Pascual. Tapenade 2.1 users guide. Technical report 300, INRIA, 2004.
[43] Jason E. Hicken and David. W. Zingg. Parallel NewtonKrylov solver for the Euler equations discretized using simultaneous-approximation terms. AIAA Journal, 46(11):2773
2786, 2008.
[44] Jason E. Hicken and David. W. Zingg. Induced-drag minimization of nonplanar geometries
based on the Euler equations. AIAA Journal, 48(11):25642575, 2010.
[45] ICAO. Aviation emissions in context. Technical report, International Civil Aviation
Organization (ICAO), Montreal, CA, 2009.
[46] A. Jameson. Aerodynamic design via control theory. Journal of Scientific Computing,
3(3):233260, September 1988.
[47] A. Jameson, L. Martinelli, and N. A. Pierce. Optimum aerodynamic design using the
Navier-Stokes equations. Theoretical and Computational Fluid Dynamics, 10:213237,
1998.
References
124
[48] Antony Jameson. Optimum aerodynamic design using control theory. Computational
Fluid Dynamics Review, pages 495528, 1995.
[49] Gaetan K.W. Kenway, Graeme J. Kennedy, and Joaquim R. R. A. Martins. A CADfree approach to high-fidelity aerostructural optimization. In Proceedings of the 13th
AIAA/ISSMO Multidisciplinary Analysis and Optimization Conference, Fort Worth, TX,
2010. AIAA 2010-9231.
[50] Hyung Min Kim, Nestor F. Michelena, Panos Y. Papalambros, and Tao Jian. Target cascading in optimal system design. Journal of Mechanical Design, 125(3):474480, September 2003.
[51] Andrew B. Lambe and Joaquim R. R. A. Martins. Extensions to the design structure matrix for the description of multidisciplinary design, analysis, and optimization processes.
Structural and Multidisciplinary Optimization, 2011.
[52] Joosung Joseph Lee. Historical and future trends in aircraft perfomance, cost, and emissions. Masters thesis, Massachusetts Institute of Technology, Cambridge, MA, September
2000.
[53] R. H. Liebeck. Design of the blended wing body subsonic transport. Journal of Aircraft,
41:1025, 2004.
[54] A.C. Limache and E.M. Cliff. Aerodynamic sensitivity theory for rotary stability derivatives. Journal of Aircraft, 37:676 683, 2000.
[55] Alejandro Cesar Limache. Aerodynamic Modeling Using Computational Fluid Dynamics
and Sensitivity Equations. PhD thesis, Virginia Polytechnic Institute and State University,
Blacksburg, VA 24061, April 2000.
[56] Thomas D. Loeser, Dan D. Vicroy, and Andreas Sch
utte. SACCON static wind tunnel
tests at DNW-NWB and 14x22 NASA LaRC. In Proceedings of the 28th AIAA Applied
Aerodynamics Conference, Chicago, IL, 2010. AIAA 2010-4393.
[57] J. N. Lyness and C. B. Moler. Numerical differentiation of analytic functions. SIAM
Journal on Numerical Analysis, 4(2):202210, 1967.
[58] Charles A. Mader and Joaquim R. R. A. Martins. Stability-constrained aerodynamic
shape optimization of a flying wing configuration. In Proceedings of the 13th AIAA/ISSMO
Multidisciplinary Analysis and Optimization Conference, Fort Worth, TX, 2010. AIAA
2010-9199.
References
125
[59] Charles A. Mader and Joaquim R. R. A. Martins. Computation of aircraft stability derivatives using an automatic differentiation adjoint approach. AIAA Journal, 49(12):2737
2750, December 2011.
[60] Charles A. Mader, Joaquim R. R. A. Martins, Juan J. Alonso, and Edwin van der Weide.
ADjoint: An approach for the rapid development of discrete adjoint solvers. AIAA Journal, 46(4):863873, 2008.
[61] Karthik Mani and Dimitri J. Mavriplis. Unsteady discrete adjoint formulation for twodimensional flow problems with deforming meshes. AIAA Journal, 46(6):13511364, 2008.
[62] Andre C. Marta and Juan J. Alonso. High-speed MHD flow control using adjoint-based
sensitivities. In Proceedings of the 14th AIAA/AHI Space Planes and Hypersonic Systems
and Technologies Conference, Canberra, Australia, 2006. AIAA 2006-8009.
[63] Andre C. Marta, Charles A. Mader, Joaquim R. R. A. Martins, Edwin van der Weide,
and Juan J. Alonso. A methodology for the development of discrete adjoint solvers using
automatic differentiation tools. International Journal of Computational Fluid Dynamics,
21(9):307327, October 2007.
[64] Joaquim R. R. A. Martins, Juan J. Alonso, and James J. Reuther. High-fidelity aerostructural design optimization of a supersonic business jet. Journal of Aircraft, 41(3):523530,
2004.
[65] Joaquim R. R. A. Martins, Peter Sturdza, and Juan J. Alonso. The complex-step derivative approximation. ACM Transactions on Mathematical Software, 29(3):245262, 2003.
[66] Joaquim R.R.A. Martins, Juan J. Alonso, and James J. Reuther. A coupled-adjoint
sensitivity analysis method for high-fidelity aero-structural design. Optimization and
Engineering, 6(1):3362, 2005.
[67] Dimitri J. Mavriplis. Solution of the unsteady discrete adjoint for three-dimensional
problems on dynamically deforming unstructured meshes. In Proceedings of the 46th
AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, January 2008. AIAA 2008727.
[68] Matthew McMullen, Antony Jameson, and Juan Alonso. Demonstration of nonlinear
frequency domain methods. AIAA Journal, 44(7):14281435, 2006.
[69] Matthew McMullen, Antony Jameson, and Juan J. Alonso. Application of a nonlinear
frequency domain solver to the Euler and Navier-Stokes equations. In Proceedings of the
References
126
40th AIAA Aerospace Sciences Meeting and Exhibit, Reno, NV, January 2002. AIAA
2002-0120.
[70] M.S. McMullen and A. Jameson. The computational efficiency of non-linear frequency
domain methods. Journal of Computational Physics, 212(2):637661, 2006.
[71] Duane McRuer, Irving Ashkenas, and Dunstan Graham. Aircraft Dynamics and Automatic Control. Princeton University Press, Princeton, New Jersey, 1973.
[72] Bruno Mialon, Alex Khrabov, Andrea Da Ronch, Luca Cavagna, Mengmeng Zhang, and
Sergio Ricci. Benchmarking the prediction of dynamic derivatives: Wind tunnel tests,
validation, acceleration methods. In Proceedings of the 28th AIAA Applied Aerodynamics
Conference, Chicago, IL, 2010. AIAA 2010-8244.
[73] United States Military. Flying qualities of piloted airplanes. Military Specification MILF-8785c, November 1980.
[74] Scott M. Murman. Reduced-frequency approach for calculating dynamic derivatives.
AIAA Journal, 45(6):11611168, 2007.
[75] Siva K. Nadarajah and Antony Jameson. Optimum shape design for unsteady flows with
time-accurate continuous and discrete adjoint methods. AIAA Journal, 45(7):14781491,
2007.
[76] Siva K. Nadarajah and Antony Jameson. Optimum shape design for unsteady threedimensional viscous flows using a nonlinear frequency-domain method. Journal of Aircraft, 44(5):15131527, 2007.
[77] Siva K. Nadarajah, Antony Jameson, and Juan J. Alonso. Sonic boom reduction using an
adjoint method for wing-body configurations in supersonic flow. AIAA Paper 2002-5547,
2002.
[78] Marian Nemec and David W. Zingg. NewtonKrylov algorithm for aerodynamic design
using the NavierStokes equations. AIAA Journal, 40(6):11461154, June 2002.
[79] Marian Nemec, David. W. Zingg, and Thomas H. Pulliam. Multipoint and multi-objective
aerodynamic shape optimization. AIAA Journal, 42(6):10571065, 2004.
[80] James C. Newman, W. K. Anderson, and L. Whitfield, D. Multidisciplinary sensitivity
derivatives using complex variables. Technical Report MSSU-COE-ERC-98-08, Computational Fluid Dynamics Laboratory, 1998.
References
127
[81] Eric J. Nielsen and W. Kyle Anderson. Aerodynamic design optimization on unstructured
meshes using the NavierStokes equations. AIAA Journal, 37(11):14111419, 1999.
[82] Michael A. Park and Lawrence L. Green. Steady-state computation of constant rotational
rate dynamic stability derivatives. AIAA Paper 2000-4321, August 2000.
[83] Michael A. Park, Lawrence L. Green, Raymond C. Montgomery, and David L. Raney.
Determination of stability and control derivatives using computational fluid dynamics and
automatic differentiation. AIAA Paper 1999-3136, June 1999.
[84] V. Pascual and L. Hascoet. Extension of TAPENADE towards Fortran 95. In H. M.
B
ucker, G. Corliss, P. Hovland, U. Naumann, and B. Norris, editors, Automatic Differentiation: Applications, Theory, and Tools, Lecture Notes in Computational Science and
Engineering. Springer, 2005.
[85] Sergey Peigin and Boris Epstein. Computational fluid dynamics driven optimization of
blended wing body aircraft. AIAA Journal, 44(11):27362745, 2006.
[86] R. Perez and J. Martins. pyACDT: An object-oriented framework for aircraft design
modelling and multidisciplinary optimization. In Proceedings of the 12th AIAA/ISSMO
Multidisciplinary Analysis and Optimization Conference, Victoria, BC, September 10 12, 2008.
[87] Ruben E. Perez, Peter W. Jansen, and Joaquim R. R. A. Martins. pyOpt: A Pythonbased object-oriented framework for nonlinear constrained optimization. Structures and
Multidisciplinary Optimization, 2011.
[88] Ruben E. Perez, Hugh H.T. Liu, and Kamran Behdinan. Multidisciplinary optimization
framework for control-configuration integration in aircraft conceptual design. Journal of
Aircraft, 43(6):19371948, 2006.
[89] O. Pironneau. On optimum design in fluid mechanics. Journal of Fluid Mechanics,
64:97110, 1974.
[90] N. Qin, A. Vavalle, A. Le Moigne, M. Laban, K. Hackett, and P. Weinerfelt. Aerodynamic
considerations of blended wing body aircraft. Progress in Aerospace Sciences, 40:321343,
2004.
[91] James J. Reuther, Antony Jameson, Juan J. Alonso, Mark J. Rimlinger, and David
Saunders. Constrained multipoint aerodynamic shape optimization using an adjoint formulation and parallel computers, part 1. Journal of Aircraft, 36(1):5160, 1999.
128
References
[92] James J. Reuther, Antony Jameson, Juan J. Alonso, Mark J. Rimlinger, and David
Saunders. Constrained multipoint aerodynamic shape optimization using an adjoint formulation and parallel computers, part 2. Journal of Aircraft, 36(1):6174, 1999.
[93] Detlef Rohlf, Stefan Schmidt, and Jonathan Irving. SACCON stability and control analysis applying system identification techniques. In Proceedings of the 28th AIAA Applied
Aerodynamics Conference, Chicago, IL, 2010. AIAA 2010-4399.
[94] A. Da Ronch, M. Ghoreyshi, K.J. Badcock, S. Gortz, M. Widhalm, R.P. Dwight, and
M.S. Campobasso. Linear frequency domain and harmonic balance predictions of dynamic derivatives. In Proceedings of the 28th AIAA Applied Aerodynamics Conference,
Chicago, IL, 2010. AIAA 2010-4699.
[95] A. Da Ronch, D. Vallespin, M. Ghoreyshi, and K.J. Badcock. Computation of dynamic
derivatives using CFD. In Proceedings of the 28th AIAA Applied Aerodynamics Conference, Chicago, IL, 2010. AIAA 2010-4817.
[96] J.-F. Le Roy and S. Morgand. SACCON CFD static and dynamic derivatives using elsA.
In Proceedings of the 28th AIAA Applied Aerodynamics Conference, Chicago, IL, 2010.
AIAA 2010-4562.
[97] Markus P. Rumpfkeil and Dimitri J. Mavriplis.
Efficient Hessian calculations using
automatic differentiation and the adjoint method with applications. AIAA Journal,
48(10):24062417, October 2010.
[98] Markus P. Rumpfkeil and David. W. Zingg. The optimal control of unsteady flows with
a discrete adjoint method. Optimization and Engineering, 11(1):522, 2010.
[99] Andreas Sch
utte, Dietrich Hummel, and Stephan M. Hitzel. Numerical and experimental
analyses of the vortical flow around the SACCON configuration. In Proceedings of the
28th AIAA Applied Aerodynamics Conference, Chicago, IL, 2010. AIAA 2010-4690.
[100] Jaroslaw Sobieszczanski-Sobieski, Jeremy S. Agte, and Robert R. Sandusky. Bi-level
integrated system synthesis. AIAA Journal, 38(1):164172, January 2000.
[101] B. I. Soemarwoto and Th. E. Labrujere. Airfoil design and optimization methods: Recent
progress at NLR. International Journal for Numerical Methods in Fluids, 30(2):217228,
1999.
[102] William Squire and George Trapp. Using complex variables to estimate derivatives of real
functions. SIAM Review, 40(1):110112, 1998.
129
References
[103] Martin Stettner and Ralph Voss. Aeroelastics, flight mechanics and handling qualities of
the MOB BWB configuration. In Proceedings of the 9th AIAA/ISSMO Multidisciplinary
Analysis and Optimization Conference, Atlanta, Georgia, September 4 - 6, 2002.
[104] Jeffery P. Thomas, Kenneth C. Hall, and Earl H. Dowell. Discrete adjoint approach for
modeling unsteady aerodynamic design sensitivities. AIAA Journal, 43(9):19311936,
2005.
[105] Magnus Tormalm and Stefan Schmidt. Computational study of static and dynamic vortical flow over the delta wing SACCON configuration using the FOI flow solver Edge.
In Proceedings of the 28th AIAA Applied Aerodynamics Conference, Chicago, IL, 2010.
AIAA 2010-4561.
[106] USAF.
The USAF stability and control digital datcom.
AFFDL Technical Report
AFFDL-TR-79-3032, February 2003.

[107] E. van der Weide, G. Kalitzin, J. Schluter, and J. J. Alonso. Unsteady turbomachinery computations using massively parallel platforms. In Proceedings of the 44th AIAA
Aerospace Sciences Meeting and Exhibit, Reno, NV, 2006. AIAA 2006-0421.
[108] Edwin van der Weide, Arathi K. Gopinath, and Antony Jameson. Turbomachinery applications with the time spectral method. AIAA Paper 2005-4905, June 2005.
[109] Dan D. Vicroy. Blended-wing-body low-speed flight dynamics: Summary of ground tests
and sample results. AIAA Paper 2009-933, January 2009. Proceedings of the 47th AIAA
Aerospace Sciences Meeting Including The New Horizons Forum and Aerospace Exposition, Orlando, Florida.
[110] Sean Wakayama and Ilan Kroo. The challenge and promise of blended-wing-body optimization. AIAA Paper 1998-4736, 1998.
[111] Li Wang, Dimitri J. Mavriplis, and W. Kyle Anderson. Adjoint sensitivity formulation for
discontinuous galerkin discretizations in unsteady inviscid flow problems. AIAA Journal,
48(12):28672883, 2010.
[112] Z.U.A. Warsi. Fluid Dynamics: Theoretical and Computational Approaches. Taylor &
Francis, 3rd edition, 2006.
[113] Julian Wolkovitch. The joined wing: An overiew. Journal of Aircraft, 23:161178, 1986.
Appendix A
Linear Flight Dynamics

The derivation in the following appendix describes how to obtain the linear flight dynamic
model discussed in Chapter 2.
Coupled Longitudinal-Lateral Equations

To start the derivation, consider the equations of motion from Etkin [27].

(u + wq rv)
sin
X

Y + mg sin cos = m (v + ru pw)

(w + pv qu)
cos cos
Z
Ix

Ix
Ixy Izx
p
0 r q
p

Iy
Iyz q
0 p Ixy
Iyz q + r
r
Izx Iyz
Iz
r
q p
0
Iz
(A.1)
Ixy Izx

M = Ixy
Iy

Izx Iyz
N

1 sin tan cos tan
p

= 0
cos
sin q
0 sin sec cos sec
(A.2)
(A.3)
These equations assume that the aircraft is rigid, that the effect of spinning rotors is negligible,
that the mass of the aircraft is constant and that the aircraft is flying through a still air mass.
First, using small disturbance theory, Equations (A.1) through (A.3) can be rewritten as:
X0 + X
sin 0 cos 0
u + w0 q v0 r
Y0 + Y + mg
= m v + u0 r w0 p
cos
Z0 + Z
cos 0 0
w + v0 p u0 q
130
(A.4)
131
Appendix A. Linear Flight Dynamics
L0 + L
M0 + M = Ixy

Iy
Izx Iyz
N0 + N

p

Iyz
q
r
Iz
Ixy Izx
Ix
1 0 tan 0
p

= 0 1

0

q .
0 0 sec 0
r
(A.5)
(A.6)
These equations require the following additional assumptions:

Small angle theory applies - i.e. sin = ,cos = 1,etc.
All 2 terms are sufficiently small that they can be neglected
There is no initial bank angle - i.e 0 = 0
There are no initial rotations or accelerations - i.e. p0 = q0 = r0 = u 0 = v 0 = w 0 = p0 =
q0 = r0 = 0
Looking in more detail at the condition that there are no initial rotations or accelerations, one
can determine the values of the initial forces (X0 ,Y0 ,Z0 ) and moments (L0 ,M0 ,N0 ). Substituting
zero accelerations and rotations into Equations (A.1) and (A.2) yields the following:

X0
sin 0
0

Y0 + mg 0 = 0

Z0
cos 0
0

0

M0 = 0

N0
0
L0
(A.7)
(A.8)
Substituting these values back into Equations (A.4) and (A.5) yields:
u + w0 q v0 r
X
cos 0
Y + mg cos 0 = m v + u0 r w0 p
0
w + v0 p u0 q
Z
Ix
Iyz q
Iz
r
(A.9)
Ixy Izx

M = Ixy
Iy

N
Izx Iyz
(A.10)
132
Given the preceding equations, one will have the following state variables: u, v, w, p, q, r, , ,
,
.
On this basis, the linear aerodynamic reactions
and their time derivatives, u,
v,
w,
p,
q,
r,
,
can be expressed as follows:
X
Xi i
Xj j
XControl Control
Y Yi i Yj j YControl Control
Z Zi i Zj j ZControl Control
=
+
+
L Li i Lj j LControl Control
M M i M j M
Control
i
j
Control
N
Ni i
Nj j
NControl Control
(A.11)
where i = u, v, w, p, q, r, , , and j = di/dt. The subscribed quantities are derivatives of

each force or moment quantity with respect to the corresponding subscript. Note that since
the atmosphere is assumed to be constant this allows the location and altitude variables
xE , yE , zE to be ignored in the force equations. Now, by combining Equation (A.12) with
Equations (A.6) (A.9) and (A.10) we can get a matrix version of the equations as follows:
(Xu m)
Xv
Xw
Xp
Xq
Xr
Yu
(Yv m)
Yw
Yp
Yq
Yr
Zu
Zv
(Zw m)
Zp
Zq
Zr
Lu
Lv
Lw
(Lp Ix )
(Ixy + Lq )
(Ixz + Lr )
Mu
Mv
Mw
(Ixy + Mp )
(Iy + Mq )
(Izy + Mr )
Nu
Nv
Nw
(Ixz + Np )
(Izy + Nq )
(Iz + Nr )
v

w

p

q +
r

Xu
Xv
Xw
Xp
(Xq mw0 )
Xr + mv0
(X mg cos 0 )
Yu
Yv
Yw
Yp + mw0
Yq
Yr muo
Y + mg cos 0
Zu
Zv
Zw
Zp mv0
(muo + Zq )
Zr
Z mg sin 0
Lu
Lv
Lw
Lp
Lq
Lr
Mu
Mv
Mw
Mp
Mq
Mr
Nu
Nv
Nw
Np
Nq
Nr
tan 0
v

w

p

q +
r

sec o
XControl
YControl
ZControl
LControl
MControl Control = 0
N
Control
0
0
0
(A.12)
133
Further, because we have assumed a constant atmosphere, the orientation of the aircraft
has no impact on the forces, therefore the derivatives of the forces with respect to the change
in the Euler angles , , are zero, which simplifies the equations to:

(Xu m)
Xv
Xw
Xp
Xq
Xr
Yu
(Yv m)
Yw
Yp
Yq
Yr
Zu
Zv
(Zw m)
Zp
Zq
Zr
Lu
Lv
Lw
(Lp Ix )
(Ixy + Lq )
(Ixz + Lr )
Mu
Mv
Mw
(Ixy + Mp )
(Iy + Mq )
(Izy + Mr )
Nu
Nv
Nw
(Ixz + Np )
(Izy + Nq )
(Iz + Nr )
v

w

p

q +
r

Xu
Xv
Xw
Xp
(Xq mw0 )
Xr + mv0
mg cos 0
Yu
Yv
Yw
Yp + mw0
Yq
Yr muo
mg cos 0
Zu
Zv
Zw
Zp mv0
(muo + Zq )
Zr
mg sin 0
Lu
Lv
Lw
Lp
Lq
Lr
Mu
Mv
Mw
Mp
Mq
Mr
Nu
Nv
Nw
Np
Nq
Nr
tan 0
v

w

p

q +
r

sec o
XControl
YControl
ZControl
LControl
MControl Control = 0
N
Control
0
(A.13)
Solving for the time derivative terms yields a differential equation of the form:
h i
x =
ih i
h ih i
x + B
(A.14)
which is in a standard state-space matrix equation and can be solved using a normal differential
equation solvers. The inversion of the matrix analytically yields a very messy matrix, so for
actual computation it is probably better to invert the matrix on the fly.
Appendix B
Results Data Tables

Planform Only Optimization Cases
Parameter
Baseline
Elliptic
Bending
Cm
Kn
CAP
CLtarget
0.300000
0.300000
0.300000
0.300000
0.300000
0.300000
CL
0.300000
0.300000
0.300000
0.300000
0.300000
0.299992
CD
0.004954
0.004888
0.004817
0.004844
0.005123
0.005133
Cm
-0.000010
0.000010
-0.000010
-0.000010
-0.000010
-0.000008
Cm
0.002242
0.013867
0.029788
-0.000000
-0.246456
-0.246417
CL
5.003494
4.992613
5.052105
5.055258
4.869859
4.869545
Cm
0.620236
0.687028
0.889591
0.879304
0.472464
0.464135
Cmq
-0.865683
-0.856633
-0.860454
-0.906214
-1.455247
-1.453889
Kn (%)
-0.044805
-0.277744
-0.595908
0.000000
4.999788
5.000118
e (%)
0.963761
0.976806
0.963537
0.962759
0.943660
0.941934
xCG (m)
0.243780
0.244786
0.239811
0.311960
0.663416
0.663022
MAC (m)
1.000000
1.000000
0.989442
0.992048
1.012213
1.012048
xc4MAC (m)
0.250000
0.250000
0.247361
0.335290
0.774497
0.773554
CAP
0.000258
-0.015125
-0.037377
0.003236
0.167942
0.168177
sp
8.047202
-0.000000
-0.000000
2.314511
0.301925
0.301636
n (rad/s)
0.065557
-0.000000
-0.000000
0.233514
1.651115
1.652217
244.664280
242.153023
228.835160
247.707360
494.760972
493.932604
Izz (kg m )
1595.868750
1595.868750
1595.868750
1595.868750
1595.868750
1595.868750
(deg.)
W (kg)
3.434484
4.042761
7.545133
7.665806
9.065110
8.899279
(deg.)
0.000000
0.000000
0.000000
3.291240
19.280100
19.250614
b (m)
3.000000
3.000000
3.042917
3.035421
2.981338
2.981104
c (m)
1.000000
1.000000
0.902535
0.914259
1.005006
1.012058
10.000000
10.000000
10.000000
10.000000
10.000000
10.000000
PIzz
Table B.1: NACA 0012 wing: planform only optimization results summary: 1107k cells, M =
0.5
134
135
Appendix B. Results Data Tables
Shape Optimization Cases

Cm
Kn
CAP
CLtarget
0.300000
0.300000
0.300000
CL
0.299996
0.300000
0.300000
CD
0.004807
0.004801
0.005149
Parameter
Cm
0.000011
0.000010
0.000010
Cm
-0.665371
-0.752988
-0.246065
CL
5.048222
5.052577
4.810858
Cm
-0.519543
-0.699890
0.129640
Cmq
-1.687359
-1.822218
-1.441170
Kn (%)
13.345331
15.092049
5.000230
e (%)
0.967616
0.968178
0.970392
xCG (m)
0.111642
0.094569
0.685657
MAC (m)
0.987633
0.987476
1.022911
xc4MAC (m)
0.246908
0.246869
0.786160
CAP
0.550552
0.586254
0.164005
sp
0.166841
0.161038
0.299986
n (rad/s)
3.043741
3.142237
1.621730
380.015557
403.325544
517.368619
Izz (kg m )
1595.868750
1595.868750
1595.868750
(deg.)
W (kg)
3.064103
3.057671
3.051991
(deg.)
0.000000
0.000000
19.888040
b (m)
3.039381
3.040392
2.932521
c (m)
0.987643
0.987486
1.022920
10.000000
10.000000
10.000000
PIzz
Table B.2: NACA 0012 wing: shape optimization results summary: 1107k cells, M = 0.5
136
Parameter
Baseline
Bending
Cm
Kn
CAP
CLtarget
0.300000
0.300000
0.300000
0.300000
0.300000
CL
0.300000
0.300000
0.300000
0.300000
0.300000
CD
0.005438
0.005211
0.005210
0.005337
0.005337
Cm
-0.000010
0.000010
-0.000010
0.000010
-0.000010
Cm
0.087088
-0.043966
-0.043519
-0.284135
-0.284114
CL
5.837208
5.835770
5.833275
5.564044
5.563520
Cm
1.773076
1.346408
1.349035
0.382728
0.382836
Cmq
-1.033108
-1.334847
-1.334505
-1.847488
-1.847325
Kn (%)
-1.491942
0.745596
0.738342
4.999817
4.999817
xCG (m)
0.228396
0.547026
0.547261
0.744416
0.744515
MAC (m)
1.000000
1.010461
1.010438
1.021364
1.021382
0.250000
0.627116
0.627538
0.878067
0.878098
-0.055895
0.020482
0.020155
0.087320
0.087312
xc4MAC (m)
CAP
sp
-0.000000
0.672371
0.677448
0.310233
0.310233
n (rad/s)
-0.000000
0.631212
0.626015
1.272602
1.272483
417.494607
623.660078
627.885797
963.347405
963.470201
1595.868750
1595.868750
1595.868750
1595.868750
1595.868750
Izz (kg m )
W (kg)
(deg.)
2.933511
7.613268
7.615328
8.590863
8.557310
(deg.)
0.000000
14.106302
14.119718
22.861602
22.864088
b (m)
3.000000
2.980515
2.980963
2.953919
2.953687
c (m)
1.000000
1.010471
1.010448
1.021374
1.021392
16.184566
16.184566
16.184566
16.184566
16.184566
PIzz
Table B.3: NACA 0012 wing: planform only optimization results summary: 1107k cells, M =
0.7
137
Shape Optimization Cases

Parameter
Bending
Cm
Kn
CAP
CLtarget
0.300000
0.300000
0.300000
0.300000
CL
0.299998
0.299999
0.300000
0.300000
CD
0.004973
0.004935
0.004939
0.005090
Cm
0.000010
-0.000010
0.000010
0.000010
Cm
0.539667
-0.275911
-0.298319
-0.302163
CL
5.942244
6.011765
6.023204
5.844173
Cm
4.575228
2.564250
2.393675
0.909075
Cmq
-0.710714
-1.394649
-1.417798
-1.515923
Kn (%)
-9.112122
4.634752
4.999844
5.121750
xCG (m)
0.315935
0.181118
0.188897
0.521108
MAC (m)
0.996680
0.990240
0.990598
1.009484
xc4MAC (m)
0.249170
0.249788
0.261659
0.620716
CAP
-0.378485
0.121639
0.129099
0.099087
sp
-0.000000
0.286944
0.277583
0.299995
n (rad/s)
Izz (kg m2 )
W (kg)
-0.000000
1.561267
1.609959
1.389344
380.807451
601.407689
611.432529
847.763119
1595.868750
1595.868750
1595.868750
1595.868750
(deg.)
2.987315
3.032540
3.008049
3.282016
(deg.)
0.000000
0.084255
0.529970
13.921081
b (m)
3.010364
3.030450
3.029077
2.972135
c (m)
PIzz
0.996690
0.990250
0.990608
1.009494
20.000000
20.000000
20.000000
20.000000
138
Parameter
Baseline
Bending
Cm
Kn
CAP
CLtarget
0.300000
0.300000
0.300000
0.300000
0.300000
CL
0.299999
0.300000
0.300000
0.300000
0.300000
CD
0.039898
0.005792
0.005733
0.005640
0.005619
Cm
0.000006
0.000018
0.000011
0.000010
-0.000003
Cm
0.055392
2.566048
0.000007
-0.484800
-0.489347
CL
7.593097
10.474287
8.431867
9.377942
9.359115
Cm
-2.673158
17.959887
-0.205315
-6.295536
-6.294362
Cmq
-0.313080
-1.189516
-1.743179
-1.893410
-1.891502
Kn (%)
-0.729501
-23.522592
-0.000079
4.999816
5.051372
xCG (m)
0.437910
0.827035
0.918967
0.789222
0.795677
MAC (m)
1.000000
1.041490
1.049455
1.033954
1.035078
xc4MAC (m)
0.250000
0.452886
0.921985
0.811024
0.819287
CAP
-0.036492
-0.975527
0.000523
0.086524
0.086135
sp
-0.000000
-0.000000
3.979094
0.308502
0.309052
n (rad/s)
Izz (kg m2 )
W (kg)
-0.000000
-0.000000
0.121288
1.644602
1.639253
316.346371
416.989558
1137.655818
993.139864
1010.020857
1595.868750
1595.868750
1595.868750
1595.868750
1595.868750
(deg.)
2.169667
3.437140
3.009218
3.643078
3.665365
(deg.)
0.000000
7.609797
24.783732
20.849705
21.144092
b (m)
3.000000
2.881879
2.857234
2.901532
2.898588
c (m)
PIzz
1.000000
1.041500
1.049465
1.033964
1.035088
21.988981
21.988981
21.988981
21.988981
21.988981
Appendix C
Relevant Optimization Parameters
139
140
Appendix C. Relevant Optimization Parameters

CFD Parameters
Parameter
Time-spectral
Time-spectral amplitude
Value
units
6.28
rad./s
0.034
rad.
k2 dissipation
k4 dissipation
0.006
Time-spectral instances
Dissipation lumping parameter
Multigrid restriction relaxation
0.8
Dissipation scaling exponent
0.1
Startup CFL number
0.9
Multigrid cycle
4w
CFD Convergence tolerance
1010
Runge-Kutta to Newton-Krylove switch tolerance
0.001
Newton-Krylov global preconditioner

Newton-Krylov local preconditioner
ILU fill level
Additive Schwarz
ILU
2
Preconditioner reordering
RCM
ASM overlap
GMRES subspace size
100
Newton-Krylov Jacobian reset lag
10
ADjoint Parameters
ADjoint convergence tolerance
1e-8
ADjoint global preconditioner
Additive Schwarz
ADjoint local preconditioner
ILU
ILU fill level
RCM
GMRES subspace size
100
ASM overlap
2
SNOPT Parameters
Major feasibility tolerance
106
Major optimality tolerance
106
Major step limit

Linesearch
0.1
Non-derivative
Table C.1: Relevant parameters: M=0.5 cases, planform only
141

CFD Parameters
Parameter
Time-spectral
Value
units
6.28
rad./s
0.034
rad.
k2 dissipation
0.25
k4 dissipation
0.01562
0.8
0.1
Startup CFL number
0.9
Multigrid cycle
4w
1010
0.001

ILU fill level
Additive Schwarz
ILU
2
RCM
ASM overlap
GMRES subspace size
100
10
ADjoint Parameters
1e-8
Additive Schwarz
ILU
ILU fill level
RCM
GMRES subspace size
100
ASM overlap
2
SNOPT Parameters

Major step limit
Linesearch
106
5 105
0.1
Non-derivative
Table C.2: Relevant parameters: M=0.5 cases, with shape variables
142

CFD Parameters
Parameter
Time-spectral
Value
units
6.28
rad./s
0.0017
rad.
k2 dissipation
0.25
k4 dissipation
0.01562
0.8
0.1
Startup CFL number
0.9
Multigrid cycle
4w
1010
0.001

ILU fill level
Additive Schwarz
ILU
2
RCM
ASM overlap
GMRES subspace size
100
10
ADjoint Parameters
1e-8
Additive Schwarz
ILU
ILU fill level
RCM
GMRES subspace size
100
ASM overlap
2
SNOPT Parameters
106
106
Major step limit

Linesearch
0.5
Non-derivative
Table C.3: Relevant parameters: M=0.7 cases, planform only
143

CFD Parameters
Parameter
Time-spectral
Value
units
6.28
rad./s
0.034
rad.
k2 dissipation
0.25
k4 dissipation
0.01562
0.8
0.1
Startup CFL number
0.9
Multigrid cycle
4w
1010
0.001

ILU fill level
Additive Schwarz
ILU
2
RCM
ASM overlap
GMRES subspace size
100
10
ADjoint Parameters
1e-8
Additive Schwarz
ILU
ILU fill level
RCM
GMRES subspace size
100
ASM overlap
2
SNOPT Parameters

Major step limit
Linesearch
106
5 105
0.1
Non-derivative
144

CFD Parameters
Parameter
Time-spectral
Value
units
6.28
rad./s
0.0017
rad.
k2 dissipation
0.5
k4 dissipation
0.01562
0.8
0.1
Startup CFL number
0.9
Multigrid cycle
4w
1010
0.001

ILU fill level
Additive Schwarz
ILU
2
RCM
ASM overlap
GMRES subspace size
150
10
ADjoint Parameters
1e-8
Additive Schwarz
ILU
ILU fill level
RCM
GMRES subspace size
150
ASM overlap
2
SNOPT Parameters

Major step limit
Linesearch
106
5 105
0.5
Non-derivative
Appendix D
FFD Coordinate Tables

Note that the baseline FFD is the undeformed FFD that is the starting point for all designs. To
acheive the final design, embed the geometry in the baseline FFD and move its control points
to the values for the desired case. Note that the FFD volumes used in this thesis are cubic
spline volumes.
145
146
Appendix D. FFD Coordinate Tables
Control Point
x (m)
y (m)
z (m)
-0.030023
-0.089964
0.000000
-0.030023
-0.089964
0.224443
-0.030023
-0.089964
0.561108
-0.030023
-0.089964
1.009994
-0.030023
-0.089964
1.346659
-0.030023
-0.089964
1.683323
-0.030023
-0.089964
2.019988
-0.030023
-0.089964
2.468874
-0.030023
-0.089964
2.805539
-0.030023
-0.089964
3.029982
10
-0.030023
0.089964
0.000000
11
-0.030023
0.089964
0.224443
12
-0.030023
0.089964
0.561108
13
-0.030023
0.089964
1.009994
14
-0.030023
0.089964
1.346659
15
-0.030023
0.089964
1.683323
16
-0.030023
0.089964
2.019988
17
-0.030023
0.089964
2.468874
18
-0.030023
0.089964
2.805539
19
-0.030023
0.089964
3.029982
20
1.030000
-0.089964
0.000000
21
1.030000
-0.089964
0.224443
22
1.030000
-0.089964
0.561108
23
1.030000
-0.089964
1.009994
24
1.030000
-0.089964
1.346659
25
1.030000
-0.089964
1.683323
26
1.030000
-0.089964
2.019988
27
1.030000
-0.089964
2.468874
28
1.030000
-0.089964
2.805539
29
1.030000
-0.089964
3.029982
30
1.030000
0.089964
0.000000
31
1.030000
0.089964
0.224443
32
1.030000
0.089964
0.561108
33
1.030000
0.089964
1.009994
34
1.030000
0.089964
1.346659
35
1.030000
0.089964
1.683323
36
1.030000
0.089964
2.019988
37
1.030000
0.089964
2.468874
38
1.030000
0.089964
2.805539
39
1.030000
0.089964
3.029982
Table D.1: FFD coordinates for baseline case, M = 0.5
147
Control Point
x (m)
y (m)
z (m)
-0.030023
-0.089964
0.000000
-0.030134
-0.088686
0.224443
-0.030187
-0.088066
0.561108
-0.029823
-0.092231
1.009994
-0.029681
-0.093806
1.346659
-0.029371
-0.097148
1.683323
-0.029132
-0.099640
2.019988
-0.028820
-0.102813
2.468874
-0.027433
-0.115856
2.805539
-0.025212
-0.134100
3.029982
10
-0.030023
0.089964
0.000000
11
-0.029911
0.091242
0.224443
12
-0.029856
0.091863
0.561108
13
-0.030219
0.087697
1.009994
14
-0.030352
0.086121
1.346659
15
-0.030626
0.082776
1.683323
16
-0.030823
0.080280
2.019988
17
-0.031066
0.077101
2.468874
18
-0.031961
0.064015
2.805539
19
-0.032936
0.045663
3.029982
20
1.030000
-0.089964
0.000000
21
1.029888
-0.090001
0.224443
22
1.029834
-0.090019
0.561108
23
1.030198
-0.089898
1.009994
24
1.030335
-0.089852
1.346659
25
1.030627
-0.089753
1.683323
26
1.030844
-0.089678
2.019988
27
1.031121
-0.089583
2.468874
28
1.032254
-0.089181
2.805539
29
1.033834
-0.088593
3.029982
30
1.030000
0.089964
0.000000
31
1.030112
0.089927
0.224443
32
1.030166
0.089909
0.561108
33
1.029802
0.090030
1.009994
34
1.029664
0.090076
1.346659
35
1.029372
0.090171
1.683323
36
1.029153
0.090242
2.019988
37
1.028875
0.090332
2.468874
38
1.027727
0.090691
2.805539
39
1.026110
0.091169
3.029982
Table D.2: FFD coordinates for twist-only case, M = 0.5
148
Control Point
x (m)
y (m)
z (m)
-0.029706
-0.089014
0.000000
-0.012241
-0.207891
0.227654
-0.025961
-0.124677
0.569134
-0.019994
-0.166213
1.024442
-0.022051
-0.153194
1.365923
-0.018324
-0.176055
1.707403
-0.020865
-0.160829
2.048884
-0.013987
-0.199294
2.504192
-0.020956
-0.160255
2.845672
-0.004947
-0.240368
3.072582
10
-0.029706
0.089014
0.000000
11
-0.033114
-0.031090
0.227654
12
-0.032200
0.053243
0.569134
13
-0.033524
0.011301
1.024442
14
-0.033293
0.024479
1.365923
15
-0.033585
0.001319
1.707403
16
-0.033449
0.016755
2.048884
17
-0.033343
-0.022320
2.504192
18
-0.033439
0.017335
2.845672
19
-0.031561
-0.064340
3.072582
20
1.019125
-0.089014
0.000000
21
1.029357
-0.084920
0.227654
22
1.022227
-0.087919
0.569134
23
1.025805
-0.086501
1.024442
24
1.024687
-0.086962
1.365923
25
1.026647
-0.086142
1.707403
26
1.025343
-0.086694
2.048884
27
1.028628
-0.085259
2.504192
28
1.025294
-0.086714
2.845672
29
1.032099
-0.083577
3.072582
30
1.019125
0.089014
0.000000
31
1.008484
0.091881
0.227654
32
1.015988
0.090000
0.569134
33
1.012274
0.091013
1.024442
34
1.013445
0.090711
1.365923
35
1.011385
0.091231
1.707403
36
1.012759
0.090890
2.048884
37
1.009271
0.091714
2.504192
38
1.012811
0.090877
2.845672
39
1.005485
0.092452
3.072582
Table D.3: FFD coordinates for bending moment constrained case, M = 0.5
149
Control Point
x (m)
y (m)
z (m)
-0.029785
-0.089249
0.000000
0.000786
-0.208439
0.227093
0.006569
-0.124604
0.567733
0.039208
-0.169594
1.021919
0.056635
-0.156152
1.362558
0.080244
-0.180328
1.703198
0.097277
-0.165384
2.043837
0.130484
-0.204094
2.498024
0.143118
-0.166206
2.838663
0.169224
-0.239461
3.065165
10
-0.029785
0.089249
0.000000
11
-0.020142
-0.031172
0.227093
12
0.000384
0.053786
0.567733
13
0.025125
0.008348
1.021919
14
0.044915
0.021961
1.362558
15
0.064271
-0.002547
1.703198
16
0.083934
0.012614
2.043837
17
0.110323
-0.026738
2.498024
18
0.129631
0.011782
2.838663
19
0.142813
-0.062928
3.065165
20
1.021809
-0.089249
0.000000
21
1.045127
-0.085144
0.227093
22
1.057531
-0.088164
0.567733
23
1.087524
-0.086622
1.021919
24
1.105959
-0.087102
1.362558
25
1.127619
-0.086228
1.703198
26
1.145929
-0.086774
2.043837
27
1.175348
-0.085316
2.498024
28
1.191706
-0.086745
2.838663
29
1.209243
-0.083863
3.065165
30
1.021809
0.089249
0.000000
31
1.024199
0.092123
0.227093
32
1.051346
0.090226
0.567733
33
1.073440
0.091319
1.021919
34
1.094239
0.091010
1.362558
35
1.111646
0.091554
1.703198
36
1.132585
0.091224
2.043837
37
1.155187
0.092039
2.498024
38
1.178218
0.091242
2.838663
39
1.182831
0.092670
3.065165
Table D.4: FFD coordinates for Cm constrained case, M = 0.5
150
Control Point
x (m)
y (m)
z (m)
-0.030390
-0.091063
0.000000
0.065499
-0.212676
0.223047
0.172441
-0.156364
0.557617
0.334167
-0.190209
1.003712
0.451384
-0.191205
1.338282
0.572255
-0.210816
1.672852
0.688648
-0.207685
2.007423
0.851301
-0.238010
2.453517
0.965092
-0.223681
2.788087
1.044150
-0.270726
3.011686
10
-0.030390
0.091063
0.000000
11
0.044146
-0.031806
0.223047
12
0.161003
0.025403
0.557617
13
0.316775
-0.008915
1.003712
14
0.433817
-0.009928
1.338282
15
0.551229
-0.029908
1.672852
16
0.668175
-0.026713
2.007423
17
0.825470
-0.057725
2.453517
18
0.941795
-0.043051
2.788087
19
1.012524
-0.091367
3.011686
20
1.042580
-0.091063
0.000000
21
1.131069
-0.086875
0.223047
22
1.243293
-0.088976
0.557617
23
1.402234
-0.087747
1.003712
24
1.519351
-0.087709
1.338282
25
1.638051
-0.086949
1.672852
26
1.754817
-0.087072
2.007423
27
1.913424
-0.085836
2.453517
28
2.029248
-0.086431
2.788087
29
2.100819
-0.084406
3.011686
30
1.042580
0.091063
0.000000
31
1.109715
0.093995
0.223047
32
1.231854
0.092790
0.557617
33
1.384842
0.093547
1.003712
34
1.501784
0.093567
1.338282
35
1.617025
0.093960
1.672852
36
1.734344
0.093899
2.007423
37
1.887594
0.094449
2.453517
38
2.005951
0.094199
2.788087
39
2.069193
0.094953
3.011686
Table D.5: FFD coordinates for Kn constrained case, M = 0.5
151
Control Point
x (m)
y (m)
z (m)
-0.030385
-0.091048
0.000000
0.065366
-0.212641
0.223030
0.170069
-0.142195
0.557574
0.334662
-0.196033
1.003633
0.448765
-0.181150
1.338177
0.571230
-0.210708
1.672721
0.686769
-0.204331
2.007266
0.849381
-0.236095
2.453325
0.962597
-0.219908
2.787869
1.041692
-0.268228
3.011447
10
-0.030385
0.091048
0.000000
11
0.044016
-0.031801
0.223030
12
0.161115
0.039681
0.557574
13
0.316242
-0.014871
1.003633
14
0.432966
0.000259
1.338177
15
0.550221
-0.029827
1.672721
16
0.666886
-0.023323
2.007266
17
0.823887
-0.055792
2.453325
18
0.939964
-0.039223
2.787869
19
1.010507
-0.088822
3.011447
20
1.042410
-0.091048
0.000000
21
1.130762
-0.086860
0.223030
22
1.241566
-0.089445
0.557574
23
1.401954
-0.087510
1.003633
24
1.517514
-0.088070
1.338177
25
1.636861
-0.086937
1.672721
26
1.753149
-0.087188
2.007266
27
1.911610
-0.085901
2.453325
28
2.027073
-0.086568
2.787869
29
2.098638
-0.084503
3.011447
30
1.042410
0.091048
0.000000
31
1.109412
0.093980
0.223030
32
1.232612
0.092431
0.557574
33
1.383533
0.093652
1.003633
34
1.501715
0.093339
1.338177
35
1.615852
0.093943
1.672721
36
1.733265
0.093819
2.007266
37
1.886116
0.094402
2.453325
38
2.004440
0.094116
2.787869
39
2.067452
0.094903
3.011447
Table D.6: FFD coordinates for CAP constrained case, M = 0.5
152
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.030023
-0.089964
-0.000000
CP
50
0.096170
0.089964
-0.000000
100
0.348556
-0.089964
0.000000
-0.030023
-0.089964
0.224443
51
0.096170
0.089964
0.224443
101
0.348556
-0.089964
0.224443
-0.030023
-0.089964
0.561108
52
0.096170
0.089964
0.561108
102
0.348556
-0.089964
0.561108
-0.030023
-0.089964
1.009994
53
0.096170
0.089964
1.009994
103
0.348556
-0.089964
1.009994
-0.030023
-0.089964
1.346659
54
0.096170
0.089964
1.346659
104
0.348556
-0.089964
1.346659
-0.030023
-0.089964
1.683323
55
0.096170
0.089964
1.683323
105
0.348556
-0.089964
1.683323
-0.030023
-0.089964
2.019988
56
0.096170
0.089964
2.019988
106
0.348556
-0.089964
2.019988
-0.030023
-0.089964
2.468874
57
0.096170
0.089964
2.468874
107
0.348556
-0.089964
2.468874
-0.030023
-0.089964
2.805539
58
0.096170
0.089964
2.805539
108
0.348556
-0.089964
2.805539
-0.030023
-0.089964
3.029982
59
0.096170
0.089964
3.029982
109
0.348556
-0.089964
3.029982
10
-0.030023
0.089964
-0.000000
60
0.197125
-0.089964
0.000000
110
0.348556
0.089964
0.000000
11
-0.030023
0.089964
0.224443
61
0.197125
-0.089964
0.224443
111
0.348556
0.089964
0.224443
12
-0.030023
0.089964
0.561108
62
0.197125
-0.089964
0.561108
112
0.348556
0.089964
0.561108
13
-0.030023
0.089964
1.009994
63
0.197125
-0.089964
1.009994
113
0.348556
0.089964
1.009994
14
-0.030023
0.089964
1.346659
64
0.197125
-0.089964
1.346659
114
0.348556
0.089964
1.346659
15
-0.030023
0.089964
1.683323
65
0.197125
-0.089964
1.683323
115
0.348556
0.089964
1.683323
16
-0.030023
0.089964
2.019988
66
0.197125
-0.089964
2.019988
116
0.348556
0.089964
2.019988
17
-0.030023
0.089964
2.468874
67
0.197125
-0.089964
2.468874
117
0.348556
0.089964
2.468874
18
-0.030023
0.089964
2.805539
68
0.197125
-0.089964
2.805539
118
0.348556
0.089964
2.805539
19
-0.030023
0.089964
3.029982
69
0.197125
-0.089964
3.029982
119
0.348556
0.089964
3.029982
20
0.020454
-0.089964
0.000000
70
0.197125
0.089964
0.000000
120
0.424272
-0.089964
-0.000000
21
0.020454
-0.089964
0.224443
71
0.197125
0.089964
0.224443
121
0.424272
-0.089964
0.224443
22
0.020454
-0.089964
0.561108
72
0.197125
0.089964
0.561108
122
0.424272
-0.089964
0.561108
23
0.020454
-0.089964
1.009994
73
0.197125
0.089964
1.009994
123
0.424272
-0.089964
1.009994
24
0.020454
-0.089964
1.346659
74
0.197125
0.089964
1.346659
124
0.424272
-0.089964
1.346659
25
0.020454
-0.089964
1.683323
75
0.197125
0.089964
1.683323
125
0.424272
-0.089964
1.683323
26
0.020454
-0.089964
2.019988
76
0.197125
0.089964
2.019988
126
0.424272
-0.089964
2.019988
27
0.020454
-0.089964
2.468874
77
0.197125
0.089964
2.468874
127
0.424272
-0.089964
2.468874
28
0.020454
-0.089964
2.805539
78
0.197125
0.089964
2.805539
128
0.424272
-0.089964
2.805539
29
0.020454
-0.089964
3.029982
79
0.197125
0.089964
3.029982
129
0.424272
-0.089964
3.029982
30
0.020454
0.089964
0.000000
80
0.272841
-0.089964
-0.000000
130
0.424272
0.089964
-0.000000
31
0.020454
0.089964
0.224443
81
0.272841
-0.089964
0.224443
131
0.424272
0.089964
0.224443
32
0.020454
0.089964
0.561108
82
0.272841
-0.089964
0.561108
132
0.424272
0.089964
0.561108
33
0.020454
0.089964
1.009994
83
0.272841
-0.089964
1.009994
133
0.424272
0.089964
1.009994
34
0.020454
0.089964
1.346659
84
0.272841
-0.089964
1.346659
134
0.424272
0.089964
1.346659
35
0.020454
0.089964
1.683323
85
0.272841
-0.089964
1.683323
135
0.424272
0.089964
1.683323
36
0.020454
0.089964
2.019988
86
0.272841
-0.089964
2.019988
136
0.424272
0.089964
2.019988
37
0.020454
0.089964
2.468874
87
0.272841
-0.089964
2.468874
137
0.424272
0.089964
2.468874
38
0.020454
0.089964
2.805539
88
0.272841
-0.089964
2.805539
138
0.424272
0.089964
2.805539
39
0.020454
0.089964
3.029982
89
0.272841
-0.089964
3.029982
139
0.424272
0.089964
3.029982
40
0.096170
-0.089964
-0.000000
90
0.272841
0.089964
-0.000000
140
0.499988
-0.089964
0.000000
41
0.096170
-0.089964
0.224443
91
0.272841
0.089964
0.224443
141
0.499988
-0.089964
0.224443
42
0.096170
-0.089964
0.561108
92
0.272841
0.089964
0.561108
142
0.499988
-0.089964
0.561108
43
0.096170
-0.089964
1.009994
93
0.272841
0.089964
1.009994
143
0.499988
-0.089964
1.009994
44
0.096170
-0.089964
1.346659
94
0.272841
0.089964
1.346659
144
0.499988
-0.089964
1.346659
45
0.096170
-0.089964
1.683323
95
0.272841
0.089964
1.683323
145
0.499988
-0.089964
1.683323
46
0.096170
-0.089964
2.019988
96
0.272841
0.089964
2.019988
146
0.499988
-0.089964
2.019988
47
0.096170
-0.089964
2.468874
97
0.272841
0.089964
2.468874
147
0.499988
-0.089964
2.468874
48
0.096170
-0.089964
2.805539
98
0.272841
0.089964
2.805539
148
0.499988
-0.089964
2.805539
49
0.096170
-0.089964
3.029982
99
0.272841
0.089964
3.029982
149
0.499988
-0.089964
3.029982
Table D.7: FFD coordinates for baseline case with shape variables, M = 0.5, part 1
153
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.499988
0.089964
0.000000
200
0.727136
-0.089964
-0.000000
250
0.903807
0.089964
0.000000
151
0.499988
0.089964
0.224443
201
0.727136
-0.089964
0.224443
251
0.903807
0.089964
0.224443
152
0.499988
0.089964
0.561108
202
0.727136
-0.089964
0.561108
252
0.903807
0.089964
0.561108
153
0.499988
0.089964
1.009994
203
0.727136
-0.089964
1.009994
253
0.903807
0.089964
1.009994
154
0.499988
0.089964
1.346659
204
0.727136
-0.089964
1.346659
254
0.903807
0.089964
1.346659
155
0.499988
0.089964
1.683323
205
0.727136
-0.089964
1.683323
255
0.903807
0.089964
1.683323
156
0.499988
0.089964
2.019988
206
0.727136
-0.089964
2.019988
256
0.903807
0.089964
2.019988
157
0.499988
0.089964
2.468874
207
0.727136
-0.089964
2.468874
257
0.903807
0.089964
2.468874
158
0.499988
0.089964
2.805539
208
0.727136
-0.089964
2.805539
258
0.903807
0.089964
2.805539
159
0.499988
0.089964
3.029982
209
0.727136
-0.089964
3.029982
259
0.903807
0.089964
3.029982
160
0.575704
-0.089964
-0.000000
210
0.727136
0.089964
0.000000
260
0.979523
-0.089964
0.000000
161
0.575704
-0.089964
0.224443
211
0.727136
0.089964
0.224443
261
0.979523
-0.089964
0.224443
162
0.575704
-0.089964
0.561108
212
0.727136
0.089964
0.561108
262
0.979523
-0.089964
0.561108
163
0.575704
-0.089964
1.009994
213
0.727136
0.089964
1.009994
263
0.979523
-0.089964
1.009994
164
0.575704
-0.089964
1.346659
214
0.727136
0.089964
1.346659
264
0.979523
-0.089964
1.346659
165
0.575704
-0.089964
1.683323
215
0.727136
0.089964
1.683323
265
0.979523
-0.089964
1.683323
166
0.575704
-0.089964
2.019988
216
0.727136
0.089964
2.019988
266
0.979523
-0.089964
2.019988
167
0.575704
-0.089964
2.468874
217
0.727136
0.089964
2.468874
267
0.979523
-0.089964
2.468874
168
0.575704
-0.089964
2.805539
218
0.727136
0.089964
2.805539
268
0.979523
-0.089964
2.805539
169
0.575704
-0.089964
3.029982
219
0.727136
0.089964
3.029982
269
0.979523
-0.089964
3.029982
170
0.575704
0.089964
-0.000000
220
0.802852
-0.089964
0.000000
270
0.979523
0.089964
-0.000000
171
0.575704
0.089964
0.224443
221
0.802852
-0.089964
0.224443
271
0.979523
0.089964
0.224443
172
0.575704
0.089964
0.561108
222
0.802852
-0.089964
0.561108
272
0.979523
0.089964
0.561108
173
0.575704
0.089964
1.009994
223
0.802852
-0.089964
1.009994
273
0.979523
0.089964
1.009994
174
0.575704
0.089964
1.346659
224
0.802852
-0.089964
1.346659
274
0.979523
0.089964
1.346659
175
0.575704
0.089964
1.683323
225
0.802852
-0.089964
1.683323
275
0.979523
0.089964
1.683323
176
0.575704
0.089964
2.019988
226
0.802852
-0.089964
2.019988
276
0.979523
0.089964
2.019988
177
0.575704
0.089964
2.468874
227
0.802852
-0.089964
2.468874
277
0.979523
0.089964
2.468874
178
0.575704
0.089964
2.805539
228
0.802852
-0.089964
2.805539
278
0.979523
0.089964
2.805539
179
0.575704
0.089964
3.029982
229
0.802852
-0.089964
3.029982
279
0.979523
0.089964
3.029982
180
0.651420
-0.089964
0.000000
230
0.802852
0.089964
-0.000000
280
1.030000
-0.089964
-0.000000
181
0.651420
-0.089964
0.224443
231
0.802852
0.089964
0.224443
281
1.030000
-0.089964
0.224443
182
0.651420
-0.089964
0.561108
232
0.802852
0.089964
0.561108
282
1.030000
-0.089964
0.561108
183
0.651420
-0.089964
1.009994
233
0.802852
0.089964
1.009994
283
1.030000
-0.089964
1.009994
184
0.651420
-0.089964
1.346659
234
0.802852
0.089964
1.346659
284
1.030000
-0.089964
1.346659
185
0.651420
-0.089964
1.683323
235
0.802852
0.089964
1.683323
285
1.030000
-0.089964
1.683323
186
0.651420
-0.089964
2.019988
236
0.802852
0.089964
2.019988
286
1.030000
-0.089964
2.019988
187
0.651420
-0.089964
2.468874
237
0.802852
0.089964
2.468874
287
1.030000
-0.089964
2.468874
188
0.651420
-0.089964
2.805539
238
0.802852
0.089964
2.805539
288
1.030000
-0.089964
2.805539
189
0.651420
-0.089964
3.029982
239
0.802852
0.089964
3.029982
289
1.030000
-0.089964
3.029982
190
0.651420
0.089964
-0.000000
240
0.903807
-0.089964
-0.000000
290
1.030000
0.089964
0.000000
191
0.651420
0.089964
0.224443
241
0.903807
-0.089964
0.224443
291
1.030000
0.089964
0.224443
192
0.651420
0.089964
0.561108
242
0.903807
-0.089964
0.561108
292
1.030000
0.089964
0.561108
193
0.651420
0.089964
1.009994
243
0.903807
-0.089964
1.009994
293
1.030000
0.089964
1.009994
194
0.651420
0.089964
1.346659
244
0.903807
-0.089964
1.346659
294
1.030000
0.089964
1.346659
195
0.651420
0.089964
1.683323
245
0.903807
-0.089964
1.683323
295
1.030000
0.089964
1.683323
196
0.651420
0.089964
2.019988
246
0.903807
-0.089964
2.019988
296
1.030000
0.089964
2.019988
197
0.651420
0.089964
2.468874
247
0.903807
-0.089964
2.468874
297
1.030000
0.089964
2.468874
198
0.651420
0.089964
2.805539
248
0.903807
-0.089964
2.805539
298
1.030000
0.089964
2.805539
199
0.651420
0.089964
3.029982
249
0.903807
-0.089964
3.029982
299
1.030000
0.089964
3.029982
Table D.8: FFD coordinates for baseline case with shape variables, M = 0.5, part 2
154
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.029652
-0.091344
-0.000000
CP
50
0.094981
0.091148
-0.000000
100
0.344246
-0.088478
0.000000
-0.031836
-0.060642
0.227389
51
0.098001
0.116197
0.227389
101
0.341898
-0.068313
0.227389
-0.032023
-0.056972
0.568473
52
0.098384
0.118963
0.568473
102
0.341673
-0.065627
0.568473
-0.031557
-0.065196
1.023252
53
0.097483
0.112390
1.023252
103
0.342224
-0.070271
1.023252
-0.031694
-0.063573
1.364336
54
0.097730
0.113538
1.364336
104
0.342066
-0.069805
1.364336
-0.030387
-0.082700
1.705420
55
0.095788
0.097862
1.705420
105
0.343497
-0.081812
1.705420
-0.032837
-0.040124
2.046504
56
0.100591
0.136086
2.046504
106
0.340586
-0.054955
2.046504
-0.029778
-0.091016
2.501282
57
0.095109
0.090566
2.501282
107
0.344120
-0.087154
2.501282
-0.028335
-0.106039
2.842366
58
0.093844
0.075652
2.842366
108
0.345527
-0.097190
2.842366
-0.027368
-0.114336
3.068993
59
0.093191
0.067583
3.068993
109
0.346433
-0.103345
3.068993
10
-0.029652
0.091344
-0.000000
60
0.194687
-0.086548
0.000000
110
0.344246
0.090302
0.000000
11
-0.026577
0.120859
0.227389
61
0.192404
-0.061375
0.227389
111
0.347158
0.109595
0.227389
12
-0.026181
0.123863
0.568473
62
0.192194
-0.057377
0.568473
112
0.347515
0.113485
0.568473
13
-0.027110
0.116102
1.023252
63
0.192712
-0.062458
1.023252
113
0.346671
0.109699
1.023252
14
-0.026857
0.118955
1.364336
64
0.192562
-0.062303
1.364336
114
0.346903
0.109360
1.364336
15
-0.028840
0.100421
1.705420
65
0.193943
-0.077417
1.705420
115
0.345044
0.096227
1.705420
16
-0.023883
0.142637
2.046504
66
0.191217
-0.043890
2.046504
116
0.349540
0.123014
2.046504
17
-0.029524
0.093925
2.501282
67
0.194561
-0.083772
2.501282
117
0.344374
0.090395
2.501282
18
-0.030777
0.078087
2.842366
68
0.195982
-0.096898
2.842366
118
0.343086
0.079440
2.842366
19
-0.031410
0.068066
3.068993
69
0.196913
-0.107238
3.068993
119
0.342392
0.074164
3.068993
20
0.020201
-0.086924
0.000000
70
0.194687
0.090267
0.000000
120
0.419026
-0.089724
-0.000000
21
0.017995
-0.064155
0.227389
71
0.197664
0.114183
0.227389
121
0.416645
-0.072603
0.227389
22
0.017803
-0.061694
0.568473
72
0.198037
0.118199
0.568473
122
0.416412
-0.070751
0.568473
23
0.018281
-0.070956
1.023252
73
0.197158
0.112554
1.023252
123
0.416980
-0.075231
1.023252
24
0.018141
-0.066163
1.364336
74
0.197399
0.112646
1.364336
124
0.416818
-0.073808
1.364336
25
0.019464
-0.085046
1.705420
75
0.195491
0.098158
1.705420
125
0.418273
-0.085313
1.705420
26
0.016953
-0.042065
2.046504
76
0.200171
0.131574
2.046504
126
0.415271
-0.061144
2.046504
27
0.020075
-0.090030
2.501282
77
0.194815
0.090239
2.501282
127
0.418899
-0.089485
2.501282
28
0.021513
-0.103925
2.842366
78
0.193541
0.077048
2.842366
128
0.420300
-0.098079
2.842366
29
0.022472
-0.112224
3.068993
79
0.192871
0.070573
3.068993
129
0.421193
-0.102684
3.068993
30
0.020201
0.093367
0.000000
80
0.269466
-0.087390
-0.000000
130
0.419026
0.090126
-0.000000
31
0.023255
0.116010
0.227389
81
0.267151
-0.064636
0.227389
131
0.421904
0.107542
0.227389
32
0.023645
0.119624
0.568473
82
0.266933
-0.060899
0.568473
132
0.422254
0.110474
0.568473
33
0.022727
0.112097
1.023252
83
0.267468
-0.065800
1.023252
133
0.421427
0.106470
1.023252
34
0.022978
0.115428
1.364336
84
0.267314
-0.065698
1.364336
134
0.421655
0.106943
1.364336
35
0.021012
0.096843
1.705420
85
0.268720
-0.079149
1.705420
135
0.419821
0.095524
1.705420
36
0.025907
0.136968
2.046504
86
0.265902
-0.048856
2.046504
136
0.424225
0.118477
2.046504
37
0.020329
0.090339
2.501282
87
0.269340
-0.084990
2.501282
137
0.419153
0.090676
2.501282
38
0.019072
0.076196
2.842366
88
0.270755
-0.096515
2.842366
138
0.417859
0.081073
2.842366
39
0.018431
0.068798
3.068993
89
0.271673
-0.104856
3.068993
139
0.417152
0.075944
3.068993
40
0.094981
-0.086795
-0.000000
90
0.269466
0.089924
-0.000000
140
0.493805
-0.090897
0.000000
41
0.092742
-0.060739
0.227389
91
0.272411
0.111833
0.227389
141
0.491392
-0.075834
0.227389
42
0.092542
-0.056243
0.568473
92
0.272776
0.115882
0.568473
142
0.491151
-0.074667
0.568473
43
0.093037
-0.062623
1.023252
93
0.271914
0.110622
1.023252
143
0.491737
-0.078824
1.023252
44
0.092893
-0.061491
1.364336
94
0.272151
0.111139
1.364336
144
0.491570
-0.077361
1.364336
45
0.094241
-0.077757
1.705420
95
0.270268
0.097382
1.705420
145
0.493050
-0.086493
1.705420
46
0.091638
-0.039706
2.046504
96
0.274855
0.128019
2.046504
146
0.489956
-0.066713
2.046504
47
0.094854
-0.084357
2.501282
97
0.269594
0.090412
2.501282
147
0.493679
-0.090957
2.501282
48
0.096286
-0.099041
2.842366
98
0.268313
0.078389
2.842366
148
0.495072
-0.098077
2.842366
49
0.097232
-0.109945
3.068993
99
0.267631
0.072531
3.068993
149
0.495954
-0.102894
3.068993
Table D.9: FFD coordinates for Cm constrained case with shape variables, M = 0.5, part 1
155
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.493805
0.089535
0.000000
200
0.718144
-0.088577
-0.000000
250
0.892630
0.091369
0.000000
151
0.496651
0.104089
0.227389
201
0.715632
-0.082269
0.227389
251
0.895301
0.087966
0.227389
152
0.496993
0.105508
0.568473
202
0.715368
-0.083020
0.568473
252
0.895602
0.084614
0.568473
153
0.496183
0.102458
1.023252
203
0.716005
-0.086810
1.023252
253
0.894883
0.081595
1.023252
154
0.496407
0.103124
1.364336
204
0.715825
-0.086217
1.364336
254
0.895083
0.083074
1.364336
155
0.494598
0.093250
1.705420
205
0.717380
-0.091609
1.705420
255
0.893407
0.080654
1.705420
156
0.498909
0.113925
2.046504
206
0.714009
-0.084065
2.046504
256
0.897227
0.074812
2.046504
157
0.493933
0.089724
2.501282
207
0.718017
-0.093912
2.501282
257
0.892757
0.079360
2.501282
158
0.492631
0.082323
2.842366
208
0.719390
-0.095423
2.842366
258
0.891418
0.084086
2.842366
159
0.491912
0.077996
3.068993
209
0.720234
-0.095431
3.068993
259
0.890634
0.074688
3.068993
160
0.568585
-0.090159
-0.000000
210
0.718144
0.088416
0.000000
260
0.967409
-0.091100
0.000000
161
0.566139
-0.078020
0.227389
211
0.720892
0.094197
0.227389
261
0.964788
-0.093444
0.227389
162
0.565890
-0.076752
0.568473
212
0.721211
0.092963
0.568473
262
0.964499
-0.094738
0.568473
163
0.566493
-0.080094
1.023252
213
0.720451
0.089073
1.023252
263
0.965193
-0.096159
1.023252
164
0.566321
-0.079101
1.364336
214
0.720662
0.090013
1.364336
264
0.964998
-0.096387
1.364336
165
0.567827
-0.087108
1.705420
215
0.718928
0.084658
1.705420
265
0.966636
-0.097446
1.705420
166
0.564640
-0.070917
2.046504
216
0.722963
0.091025
2.046504
266
0.962958
-0.100649
2.046504
167
0.568458
-0.090810
2.501282
217
0.718271
0.080835
2.501282
267
0.967282
-0.096390
2.501282
168
0.569845
-0.096136
2.842366
218
0.716949
0.078965
2.842366
268
0.968632
-0.087938
2.842366
169
0.570714
-0.099488
3.068993
219
0.716193
0.080070
3.068993
269
0.969435
-0.090901
3.068993
170
0.568585
0.088501
-0.000000
220
0.792924
-0.087797
0.000000
270
0.967409
0.090867
-0.000000
171
0.571398
0.100506
0.227389
221
0.790379
-0.086296
0.227389
271
0.970048
0.088440
0.227389
172
0.571732
0.101879
0.568473
222
0.790108
-0.086816
0.568473
272
0.970341
0.086999
0.568473
173
0.570939
0.097488
1.023252
223
0.790761
-0.090966
1.023252
273
0.969639
0.084351
1.023252
174
0.571159
0.098585
1.364336
224
0.790577
-0.090978
1.364336
274
0.969835
0.085573
1.364336
175
0.569375
0.091008
1.705420
225
0.792157
-0.095194
1.705420
275
0.968184
0.084419
1.705420
176
0.573594
0.107071
2.046504
226
0.788694
-0.093099
2.046504
276
0.971912
0.080831
2.046504
177
0.568712
0.087203
2.501282
227
0.792797
-0.096589
2.501282
277
0.967536
0.086092
2.501282
178
0.567404
0.081333
2.842366
228
0.794163
-0.093966
2.842366
278
0.966191
0.092498
2.842366
179
0.566672
0.078117
3.068993
229
0.794995
-0.096589
3.068993
279
0.965394
0.088578
3.068993
180
0.643364
-0.089445
0.000000
230
0.792924
0.089890
-0.000000
280
1.017262
-0.090718
-0.000000
181
0.640885
-0.079705
0.227389
231
0.795639
0.091245
0.227389
281
1.014620
-0.091481
0.227389
182
0.640629
-0.079157
0.568473
232
0.795950
0.089524
0.568473
282
1.014325
-0.091466
0.568473
183
0.641249
-0.083134
1.023252
233
0.795208
0.086321
1.023252
283
1.015030
-0.091204
1.023252
184
0.641073
-0.082418
1.364336
234
0.795414
0.086486
1.364336
284
1.014833
-0.091315
1.364336
185
0.642604
-0.088654
1.705420
235
0.793705
0.083618
1.705420
285
1.016487
-0.090820
1.705420
186
0.639325
-0.076460
2.046504
236
0.797648
0.084320
2.046504
286
1.012748
-0.091850
2.046504
187
0.643238
-0.091350
2.501282
237
0.793051
0.079621
2.501282
287
1.017135
-0.090606
2.501282
188
0.644618
-0.095201
2.842366
238
0.791721
0.079688
2.842366
288
1.018480
-0.090178
2.842366
189
0.645474
-0.096938
3.068993
239
0.790953
0.077986
3.068993
289
1.019275
-0.090482
3.068993
190
0.643364
0.087935
-0.000000
240
0.892630
-0.086875
-0.000000
290
1.017262
0.090718
0.000000
191
0.646145
0.097469
0.227389
241
0.890042
-0.089918
0.227389
291
1.019879
0.089727
0.227389
192
0.646472
0.096737
0.568473
242
0.889760
-0.093090
0.568473
292
1.020168
0.089518
0.568473
193
0.645695
0.093324
1.023252
243
0.890436
-0.094918
1.023252
293
1.019476
0.089721
1.023252
194
0.645910
0.093929
1.364336
244
0.890246
-0.094651
1.364336
294
1.019670
0.089702
1.364336
195
0.644151
0.087741
1.705420
245
0.891859
-0.097276
1.705420
295
1.018035
0.090304
1.705420
196
0.648278
0.099333
2.046504
246
0.888274
-0.101900
2.046504
296
1.021702
0.088864
2.046504
197
0.643492
0.083437
2.501282
247
0.892503
-0.097011
2.501282
297
1.017389
0.090522
2.501282
198
0.642176
0.080168
2.842366
248
0.893859
-0.090607
2.842366
298
1.016039
0.090992
2.842366
199
0.641433
0.079167
3.068993
249
0.894675
-0.099143
3.068993
299
1.015234
0.091829
3.068993
156
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.029647
-0.096516
-0.000000
CP
50
0.094966
0.089974
-0.000000
100
0.344191
-0.088613
0.000000
-0.032319
-0.055697
0.227465
51
0.099061
0.122590
0.227465
101
0.341241
-0.063475
0.227465
-0.032765
-0.044993
0.568662
52
0.100346
0.132003
0.568662
102
0.340630
-0.055349
0.568662
-0.031702
-0.067444
1.023592
53
0.097739
0.112291
1.023592
103
0.341996
-0.068271
1.023592
-0.032708
-0.047763
1.364789
54
0.100158
0.130490
1.364789
104
0.340713
-0.057238
1.364789
-0.030332
-0.088843
1.705987
55
0.095712
0.095048
1.705987
105
0.343495
-0.082170
1.705987
-0.032646
-0.051180
2.047184
56
0.099961
0.128214
2.047184
106
0.340802
-0.057000
2.047184
-0.029056
-0.104602
2.502114
57
0.094413
0.081494
2.502114
107
0.344775
-0.092504
2.502114
-0.027842
-0.115873
2.843312
58
0.093483
0.071206
2.843312
108
0.345933
-0.100030
2.843312
-0.027091
-0.120964
3.069999
59
0.093013
0.065344
3.069999
109
0.346629
-0.105188
3.069999
10
-0.029647
0.096516
-0.000000
60
0.194656
-0.085439
0.000000
110
0.344191
0.091361
0.000000
11
-0.025459
0.134245
0.227465
61
0.191817
-0.053488
0.227465
111
0.348101
0.116812
0.227465
12
-0.024120
0.143980
0.568662
62
0.191272
-0.044822
0.568662
112
0.349276
0.124420
0.568662
13
-0.026827
0.123261
1.023592
63
0.192517
-0.060798
1.023592
113
0.346871
0.112335
1.023592
14
-0.024315
0.143858
1.364789
64
0.191344
-0.046641
1.364789
114
0.349106
0.123080
1.364789
15
-0.028897
0.105275
1.705987
65
0.193964
-0.077320
1.705987
115
0.344930
0.097579
1.705987
16
-0.024521
0.144195
2.047184
66
0.191423
-0.046929
2.047184
116
0.348926
0.121578
2.047184
17
-0.030197
0.091535
2.502114
67
0.195243
-0.089061
2.502114
117
0.343634
0.086766
2.502114
18
-0.031109
0.078471
2.843312
68
0.196423
-0.099811
2.843312
118
0.342667
0.077399
2.843312
19
-0.031560
0.069799
3.069999
69
0.197141
-0.109331
3.069999
119
0.342160
0.073153
3.069999
20
0.020198
-0.096986
0.000000
70
0.194656
0.089431
0.000000
120
0.418959
-0.091312
-0.000000
21
0.017489
-0.062681
0.227465
71
0.198677
0.120101
0.227465
121
0.415953
-0.069607
0.227465
22
0.017021
-0.053938
0.568662
72
0.199918
0.129490
0.568662
122
0.415310
-0.062118
0.568662
23
0.018124
-0.076565
1.023592
73
0.197392
0.112730
1.023592
123
0.416735
-0.073758
1.023592
24
0.017081
-0.052739
1.364789
74
0.199737
0.127328
1.364789
124
0.415397
-0.063076
1.364789
25
0.019512
-0.092740
1.705987
75
0.195399
0.095830
1.705987
125
0.418260
-0.086022
1.705987
26
0.017147
-0.054339
2.047184
76
0.199547
0.125596
2.047184
126
0.415491
-0.063189
2.047184
27
0.020788
-0.103787
2.502114
77
0.194102
0.082894
2.502114
127
0.419541
-0.095657
2.502114
28
0.021995
-0.112689
2.843312
78
0.193156
0.071834
2.843312
128
0.420688
-0.100085
2.843312
29
0.022738
-0.118077
3.069999
79
0.192672
0.068561
3.069999
129
0.421373
-0.106863
3.069999
30
0.020198
0.093974
0.000000
80
0.269424
-0.086064
-0.000000
130
0.418959
0.091831
-0.000000
31
0.024349
0.124748
0.227465
81
0.266529
-0.057159
0.227465
131
0.422813
0.113762
0.227465
32
0.025667
0.133967
0.568662
82
0.265951
-0.048885
0.568662
132
0.423955
0.120210
0.568662
33
0.023000
0.112926
1.023592
83
0.267256
-0.063162
1.023592
133
0.421611
0.109811
1.023592
34
0.025474
0.134799
1.364789
84
0.266029
-0.050872
1.364789
134
0.423790
0.119407
1.364789
35
0.020947
0.096275
1.705987
85
0.268729
-0.078636
1.705987
135
0.419695
0.097172
1.705987
36
0.025272
0.132110
2.047184
86
0.266112
-0.050914
2.047184
136
0.423616
0.118023
2.047184
37
0.019647
0.083322
2.502114
87
0.270009
-0.089736
2.502114
137
0.418400
0.088156
2.502114
38
0.018728
0.073110
2.843312
88
0.271178
-0.099024
2.843312
138
0.417422
0.079637
2.843312
39
0.018269
0.068334
3.069999
89
0.271885
-0.106456
3.069999
139
0.416904
0.076174
3.069999
40
0.094966
-0.088573
-0.000000
90
0.269424
0.090142
-0.000000
140
0.493727
-0.092019
0.000000
41
0.092201
-0.054262
0.227465
91
0.273389
0.118566
0.227465
141
0.490665
-0.071989
0.227465
42
0.091700
-0.044804
0.568662
92
0.274597
0.127489
0.568662
142
0.489989
-0.066951
0.568662
43
0.092864
-0.063471
1.023592
93
0.272131
0.113091
1.023592
143
0.491475
-0.076779
1.023592
44
0.091765
-0.045499
1.364789
94
0.274422
0.125477
1.364789
144
0.490081
-0.067353
1.364789
45
0.094277
-0.080234
1.705987
95
0.270165
0.097108
1.705987
145
0.493025
-0.086740
1.705987
46
0.091837
-0.046149
2.047184
96
0.274237
0.123931
2.047184
146
0.490181
-0.068702
2.047184
47
0.095555
-0.092732
2.502114
97
0.268868
0.084975
2.502114
147
0.494307
-0.095050
2.502114
48
0.096750
-0.104436
2.843312
98
0.267911
0.074143
2.843312
148
0.495443
-0.099593
2.843312
49
0.097482
-0.113127
3.069999
99
0.267416
0.070817
3.069999
149
0.496117
-0.105929
3.069999
Table D.11: FFD coordinates for Kn constrained case with shape variables, M = 0.5, part 1
157
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.493727
0.089454
0.000000
200
0.718030
-0.089720
-0.000000
250
0.892488
0.089940
0.000000
151
0.497525
0.106845
0.227465
201
0.714800
-0.080802
0.227465
251
0.895989
0.084223
0.227465
152
0.498634
0.113432
0.568662
202
0.714026
-0.080911
0.568662
252
0.896923
0.081888
0.568662
153
0.496350
0.104236
1.023592
203
0.715694
-0.088075
1.023592
253
0.894961
0.081252
1.023592
154
0.498474
0.112410
1.364789
204
0.714134
-0.082100
1.364789
254
0.896790
0.077257
1.364789
155
0.494460
0.093616
1.705987
205
0.717321
-0.091635
1.705987
255
0.893208
0.080750
1.705987
156
0.498305
0.111786
2.047184
206
0.714249
-0.083867
2.047184
256
0.896649
0.077672
2.047184
157
0.493166
0.086277
2.502114
207
0.718606
-0.094944
2.502114
257
0.891919
0.080526
2.502114
158
0.492177
0.078682
2.843312
208
0.719709
-0.094736
2.843312
258
0.890870
0.087788
2.843312
159
0.491648
0.077375
3.069999
209
0.720349
-0.095919
3.069999
259
0.890283
0.073043
3.069999
160
0.568494
-0.092329
-0.000000
210
0.718030
0.086616
0.000000
260
0.967255
-0.094745
0.000000
161
0.565376
-0.076338
0.227465
211
0.721661
0.094388
0.227465
261
0.963840
-0.099423
0.227465
162
0.564668
-0.071371
0.568662
212
0.722672
0.094057
0.568662
262
0.962957
-0.102392
0.568662
163
0.566214
-0.079662
1.023592
213
0.720569
0.086498
1.023592
263
0.964825
-0.104176
1.023592
164
0.564765
-0.071849
1.364789
214
0.722527
0.092930
1.364789
264
0.963081
-0.104608
1.364789
165
0.567790
-0.087967
1.705987
215
0.718756
0.082976
1.705987
265
0.966538
-0.103465
1.705987
166
0.564870
-0.072923
2.047184
216
0.722374
0.089445
2.047184
266
0.963215
-0.105480
2.047184
167
0.569074
-0.094922
2.502114
217
0.717465
0.078376
2.502114
267
0.967826
-0.100336
2.502114
168
0.570199
-0.100218
2.843312
218
0.716442
0.075528
2.843312
268
0.968892
-0.086237
2.843312
169
0.570861
-0.101477
3.069999
219
0.715881
0.078859
3.069999
269
0.969496
-0.090966
3.069999
170
0.568494
0.088450
-0.000000
220
0.792797
-0.090187
0.000000
270
0.967255
0.096919
-0.000000
171
0.572237
0.102709
0.227465
221
0.789512
-0.086287
0.227465
271
0.970701
0.091997
0.227465
172
0.573313
0.107442
0.568662
222
0.788705
-0.088746
0.568662
272
0.971602
0.087823
0.568662
173
0.571090
0.098005
1.023592
223
0.790433
-0.092992
1.023592
273
0.969701
0.086627
1.023592
174
0.573159
0.105950
1.364789
224
0.788818
-0.090173
1.364789
274
0.971475
0.085975
1.364789
175
0.569226
0.090404
1.705987
225
0.792086
-0.095362
1.705987
275
0.967974
0.088041
1.705987
176
0.572995
0.103835
2.047184
226
0.788939
-0.091413
2.047184
276
0.971339
0.084404
2.047184
177
0.567932
0.083151
2.502114
227
0.793372
-0.095997
2.502114
277
0.966685
0.090540
2.502114
178
0.566932
0.078650
2.843312
228
0.794464
-0.092257
2.843312
278
0.965625
0.096923
2.843312
179
0.566392
0.076992
3.069999
229
0.795093
-0.095661
3.069999
279
0.965027
0.091978
3.069999
180
0.643262
-0.090822
0.000000
230
0.792797
0.089100
-0.000000
280
1.017101
-0.094567
-0.000000
181
0.640088
-0.078023
0.227465
231
0.796373
0.090811
0.227465
281
1.013648
-0.095568
0.227465
182
0.639347
-0.076044
0.568662
232
0.797351
0.087997
0.568662
282
1.012743
-0.095362
0.568662
183
0.640954
-0.083940
1.023592
233
0.795308
0.083871
1.023592
283
1.014652
-0.094943
1.023592
184
0.639450
-0.076402
1.364789
234
0.797211
0.086447
1.364789
284
1.012871
-0.095445
1.364789
185
0.642556
-0.089375
1.705987
235
0.793521
0.081206
1.705987
285
1.016382
-0.094671
1.705987
186
0.639560
-0.078491
2.047184
236
0.797063
0.084187
2.047184
286
1.013008
-0.095023
2.047184
187
0.643840
-0.095262
2.502114
237
0.792231
0.078191
2.502114
287
1.017671
-0.094188
2.502114
188
0.644954
-0.096661
2.843312
238
0.791197
0.079656
2.843312
288
1.018729
-0.094796
2.843312
189
0.645605
-0.097748
3.069999
239
0.790625
0.076575
3.069999
289
1.019326
-0.094568
3.069999
190
0.643262
0.086843
-0.000000
240
0.892488
-0.090276
-0.000000
290
1.017101
0.094567
0.000000
191
0.646949
0.098077
0.227465
241
0.889128
-0.095540
0.227465
291
1.020509
0.093281
0.227465
192
0.647993
0.101593
0.568662
242
0.888277
-0.097290
0.568662
292
1.021388
0.092479
0.568662
193
0.645829
0.092023
1.023592
243
0.890086
-0.098095
1.023592
293
1.019527
0.093317
1.023592
194
0.647843
0.099737
1.364789
244
0.888397
-0.099152
1.364789
294
1.021264
0.092646
1.364789
195
0.643991
0.086480
1.705987
245
0.891773
-0.099346
1.705987
295
1.017817
0.094192
1.705987
196
0.647684
0.097143
2.047184
246
0.888525
-0.100963
2.047184
296
1.021132
0.092314
2.047184
197
0.642698
0.080136
2.502114
247
0.893060
-0.096893
2.502114
297
1.016529
0.094569
2.502114
198
0.641687
0.076126
2.843312
248
0.894137
-0.086165
2.843312
298
1.015462
0.095885
2.843312
199
0.641136
0.077563
3.069999
249
0.894752
-0.099014
3.069999
299
1.014857
0.096058
3.069999
158
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.030711
-0.090225
-0.000000
CP
50
0.098373
0.095174
-0.000000
100
0.356542
-0.101787
0.000000
0.050789
-0.044189
0.219395
51
0.187622
0.136146
0.219395
101
0.437689
-0.066283
0.219395
0.178519
-0.058827
0.548487
52
0.312097
0.118021
0.548487
102
0.565654
-0.076760
0.548487
0.348189
-0.074481
0.987277
53
0.479915
0.108320
0.987277
103
0.735402
-0.090098
0.987277
0.474852
-0.074419
1.316369
54
0.606681
0.109353
1.316369
104
0.862062
-0.084138
1.316369
0.602781
-0.089235
1.645462
55
0.731942
0.095217
1.645462
105
0.990034
-0.093688
1.645462
0.728862
-0.082412
1.974554
56
0.859329
0.100311
1.974554
106
1.116105
-0.088038
1.974554
0.899720
-0.106693
2.413344
57
1.026458
0.079695
2.413344
107
1.286942
-0.100499
2.413344
1.030826
-0.144265
2.742436
58
1.150899
0.040755
2.742436
108
1.417630
-0.124734
2.742436
1.103222
-0.137600
2.963190
59
1.224398
0.048631
2.963190
109
1.490128
-0.124112
2.963190
10
-0.030711
0.090225
-0.000000
60
0.201641
-0.107527
0.000000
110
0.356542
0.093702
0.000000
11
0.058655
0.134250
0.219395
61
0.282929
-0.054299
0.219395
111
0.445555
0.131841
0.219395
12
0.183052
0.110731
0.548487
62
0.410800
-0.075404
0.548487
112
0.570188
0.116358
0.548487
13
0.350844
0.104880
0.987277
63
0.580517
-0.086829
0.987277
113
0.738057
0.108445
0.987277
14
0.477611
0.106007
1.316369
64
0.707178
-0.082665
1.316369
114
0.864821
0.110219
1.316369
15
0.602858
0.090118
1.645462
65
0.835133
-0.092096
1.645462
115
0.990111
0.102578
1.645462
16
0.730248
0.098281
1.974554
66
0.961208
-0.084894
1.974554
116
1.117491
0.105224
1.974554
17
0.897384
0.079948
2.413344
67
1.132053
-0.099525
2.413344
117
1.284606
0.090217
2.413344
18
1.021964
0.042801
2.742436
68
1.262909
-0.127250
2.742436
118
1.408768
0.062104
2.742436
19
1.095429
0.048375
2.963190
69
1.335366
-0.129928
2.963190
119
1.482335
0.063781
2.963190
20
0.020923
-0.088304
0.000000
70
0.201641
0.093659
0.000000
120
0.433993
-0.097140
-0.000000
21
0.102376
-0.061831
0.219395
71
0.290795
0.137029
0.219395
121
0.515068
-0.070714
0.219395
22
0.230137
-0.078028
0.548487
72
0.415334
0.118169
0.548487
122
0.643082
-0.079298
0.548487
23
0.399818
-0.092518
0.987277
73
0.583172
0.109565
0.987277
123
0.812845
-0.091522
0.987277
24
0.526480
-0.085585
1.316369
74
0.709937
0.110694
1.316369
124
0.939504
-0.085560
1.316369
25
0.654415
-0.098231
1.645462
75
0.835210
0.099251
1.645462
125
1.067485
-0.094195
1.645462
26
0.780495
-0.087841
1.974554
76
0.962594
0.103645
1.974554
126
1.193553
-0.089808
1.974554
27
0.951349
-0.106834
2.413344
77
1.129717
0.083871
2.413344
127
1.364386
-0.101491
2.413344
28
1.082400
-0.135313
2.742436
78
1.254047
0.051987
2.742436
128
1.494991
-0.121978
2.742436
29
1.154810
-0.125755
2.963190
79
1.327573
0.051903
2.963190
129
1.567510
-0.120351
2.963190
30
0.020923
0.096874
0.000000
80
0.279091
-0.105802
-0.000000
130
0.433993
0.094436
-0.000000
31
0.110242
0.131690
0.219395
81
0.360309
-0.060337
0.219395
131
0.522935
0.125879
0.219395
32
0.234670
0.121070
0.548487
82
0.488227
-0.075172
0.548487
132
0.647615
0.113323
0.548487
33
0.402473
0.107625
0.987277
83
0.657960
-0.088358
0.987277
133
0.815500
0.105664
0.987277
34
0.529239
0.109408
1.316369
84
0.784620
-0.082981
1.316369
134
0.942263
0.107740
1.316369
35
0.654492
0.092447
1.645462
85
0.912584
-0.092875
1.645462
135
1.067562
0.101403
1.645462
36
0.781881
0.097772
1.974554
86
1.038656
-0.086088
1.974554
136
1.194939
0.103554
1.974554
37
0.949013
0.077444
2.413344
87
1.209497
-0.099684
2.413344
137
1.362050
0.090984
2.413344
38
1.073538
0.030157
2.742436
88
1.340269
-0.126145
2.742436
138
1.486129
0.065262
2.742436
39
1.147017
0.050823
2.963190
89
1.412747
-0.127558
2.963190
139
1.559717
0.068749
2.963190
40
0.098373
-0.101339
-0.000000
90
0.279091
0.093077
-0.000000
140
0.511443
-0.093476
0.000000
41
0.179756
-0.053274
0.219395
91
0.368175
0.135957
0.219395
141
0.592448
-0.071198
0.219395
42
0.307564
-0.074288
0.548487
92
0.492761
0.117508
0.548487
142
0.720509
-0.081731
0.548487
43
0.477260
-0.086472
0.987277
93
0.660615
0.109744
0.987277
143
0.890287
-0.090776
0.987277
44
0.603922
-0.082773
1.316369
94
0.787379
0.111100
1.316369
144
1.016946
-0.086682
1.316369
45
0.731865
-0.092085
1.645462
95
0.912661
0.101558
1.645462
145
1.144935
-0.093614
1.645462
46
0.857943
-0.084952
1.974554
96
1.040042
0.105270
1.974554
146
1.271001
-0.090152
1.974554
47
1.028794
-0.099981
2.413344
97
1.207162
0.087201
2.413344
147
1.441831
-0.100715
2.413344
48
1.159761
-0.130841
2.742436
98
1.331407
0.058291
2.742436
148
1.572352
-0.117526
2.742436
49
1.232191
-0.129657
2.963190
99
1.404954
0.057219
2.963190
149
1.644891
-0.115664
2.963190
Table D.13: FFD coordinates for CAP constrained case with shape variables, M = 0.5, part 1
159
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.511443
0.094382
0.000000
200
0.743795
-0.085277
-0.000000
250
0.924513
0.101057
0.000000
151
0.600314
0.116931
0.219395
201
0.824588
-0.078736
0.219395
251
1.013007
0.092798
0.219395
152
0.725042
0.107306
0.548487
202
0.952790
-0.085656
0.548487
252
1.137987
0.096036
0.548487
153
0.892942
0.099469
0.987277
203
1.122615
-0.087325
0.987277
253
1.305970
0.093312
0.987277
154
1.019705
0.102440
1.316369
204
1.249272
-0.088294
1.316369
254
1.432728
0.089719
1.316369
155
1.145012
0.096349
1.645462
205
1.377287
-0.092770
1.645462
255
1.558083
0.089010
1.645462
156
1.272388
0.099180
1.974554
206
1.503347
-0.089479
1.974554
256
1.685446
0.082090
1.974554
157
1.439495
0.087421
2.413344
207
1.674164
-0.092217
2.413344
257
1.852532
0.088455
2.413344
158
1.563490
0.067952
2.742436
208
1.804434
-0.103794
2.742436
258
1.976081
0.088106
2.742436
159
1.637098
0.069877
2.963190
209
1.877034
-0.104106
2.963190
259
2.049797
0.086184
2.963190
160
0.588894
-0.090329
-0.000000
210
0.743795
0.098656
0.000000
260
1.001964
-0.082248
0.000000
161
0.669828
-0.072276
0.219395
211
0.832454
0.102992
0.219395
261
1.082521
-0.095781
0.219395
162
0.797936
-0.082658
0.548487
212
0.957324
0.095264
0.548487
262
1.210881
-0.098585
0.548487
163
0.967730
-0.088866
0.987277
213
1.125270
0.089956
0.987277
263
1.380757
-0.096521
0.987277
164
1.094388
-0.086487
1.316369
214
1.252031
0.092267
1.316369
264
1.507411
-0.094891
1.316369
165
1.222386
-0.093126
1.645462
215
1.377364
0.086398
1.645462
265
1.635456
-0.096713
1.645462
166
1.348450
-0.089207
1.974554
216
1.504733
0.087642
1.974554
266
1.761508
-0.093893
1.974554
167
1.519275
-0.097727
2.413344
217
1.671828
0.084132
2.413344
267
1.932312
-0.093864
2.413344
168
1.649713
-0.112829
2.742436
218
1.795572
0.076263
2.742436
268
2.062304
-0.094526
2.742436
169
1.722272
-0.111658
2.963190
219
1.869241
0.078821
2.963190
269
2.134972
-0.096850
2.963190
170
0.588894
0.094835
-0.000000
220
0.821246
-0.084478
0.000000
270
1.001964
0.096187
-0.000000
171
0.677694
0.111054
0.219395
221
0.901968
-0.083201
0.219395
271
1.090387
0.094912
0.219395
172
0.802469
0.100731
0.548487
222
1.030217
-0.089041
0.548487
272
1.215414
0.095009
0.548487
173
0.970385
0.094173
0.987277
223
1.200058
-0.088823
0.987277
273
1.383412
0.096286
0.987277
174
1.097147
0.097096
1.316369
224
1.326714
-0.091300
1.316369
274
1.510170
0.094659
1.316369
175
1.222463
0.090881
1.645462
225
1.454738
-0.093201
1.645462
275
1.635533
0.091150
1.645462
176
1.349836
0.094229
1.974554
226
1.580795
-0.091935
1.974554
276
1.762894
0.090639
1.974554
177
1.516939
0.083851
2.413344
227
1.751608
-0.091211
2.413344
277
1.929976
0.094392
2.413344
178
1.640851
0.069545
2.742436
228
1.881795
-0.098753
2.742436
278
2.053442
0.098854
2.742436
179
1.714479
0.072472
2.963190
229
1.954416
-0.102390
2.963190
279
2.127179
0.095234
2.963190
180
0.666345
-0.087582
0.000000
230
0.821246
0.101271
-0.000000
280
1.053598
-0.090268
-0.000000
181
0.747208
-0.074730
0.219395
231
0.909834
0.098997
0.219395
281
1.134107
-0.097247
0.219395
182
0.875363
-0.084035
0.548487
232
1.034751
0.096914
0.548487
282
1.262499
-0.096928
0.548487
183
1.045172
-0.087497
0.987277
233
1.202713
0.091060
0.987277
283
1.432385
-0.097806
0.987277
184
1.171830
-0.086873
1.316369
234
1.329472
0.092901
1.316369
284
1.559039
-0.095565
1.316369
185
1.299837
-0.093085
1.645462
235
1.454815
0.087033
1.645462
285
1.687090
-0.093928
1.645462
186
1.425898
-0.088595
1.974554
236
1.582181
0.086238
1.974554
286
1.813141
-0.095991
1.974554
187
1.596719
-0.095194
2.413344
237
1.749272
0.087331
2.413344
287
1.983942
-0.096006
2.413344
188
1.727074
-0.107816
2.742436
238
1.872933
0.080743
2.742436
288
2.113878
-0.094164
2.742436
189
1.799653
-0.108234
2.963190
239
1.946623
0.084085
2.963190
289
2.186559
-0.093596
2.963190
190
0.666345
0.096394
-0.000000
240
0.924513
-0.081820
-0.000000
290
1.053598
0.090268
0.000000
191
0.755074
0.106709
0.219395
241
1.005141
-0.088841
0.219395
291
1.141974
0.094624
0.219395
192
0.879896
0.096928
0.548487
242
1.133453
-0.089831
0.548487
292
1.267032
0.095416
0.548487
193
1.047828
0.090888
0.987277
243
1.303314
-0.091824
0.987277
293
1.435041
0.096921
0.987277
194
1.174589
0.093967
1.316369
244
1.429969
-0.091638
1.316369
294
1.561798
0.094645
1.316369
195
1.299914
0.087728
1.645462
245
1.558005
-0.093051
1.645462
295
1.687167
0.093902
1.645462
196
1.427284
0.090350
1.974554
246
1.684060
-0.091739
1.974554
296
1.814527
0.095529
1.974554
197
1.594384
0.083035
2.413344
247
1.854868
-0.087942
2.413344
297
1.981606
0.096785
2.413344
198
1.718211
0.072578
2.742436
248
1.984943
-0.091854
2.742436
298
2.105015
0.097120
2.742436
199
1.791860
0.075149
2.963190
249
2.057590
-0.099053
2.963190
299
2.178766
0.096195
2.963190
160
Control Point
x (m)
y (m)
z (m)
-0.030337
-0.090905
0.000000
0.043534
-0.212307
0.222985
0.115132
-0.139416
0.557464
0.235126
-0.189165
1.003434
0.315229
-0.166290
1.337912
0.403360
-0.189863
1.672390
0.485928
-0.181646
2.006869
0.599726
-0.191176
2.452839
0.688464
-0.214829
2.787318
0.734313
-0.201477
3.010810
10
-0.030337
0.090905
0.000000
11
0.022218
-0.031751
0.222985
12
0.106640
0.042196
0.557464
13
0.217890
-0.008173
1.003434
14
0.302019
0.015040
1.337912
15
0.386001
-0.008883
1.672390
16
0.470016
-0.000533
2.006869
17
0.582136
-0.010218
2.452839
18
0.666703
-0.034325
2.787318
19
0.714907
-0.020705
3.010810
20
1.040775
-0.090905
0.000000
21
1.107259
-0.086724
0.222985
22
1.185075
-0.089390
0.557464
23
1.301414
-0.087622
1.003434
24
1.383510
-0.088462
1.337912
25
1.469579
-0.087596
1.672390
26
1.552930
-0.087903
2.006869
27
1.665814
-0.087546
2.452839
28
1.751876
-0.086623
2.787318
29
1.799307
-0.087150
3.010810
30
1.040775
0.090905
0.000000
31
1.085943
0.093833
0.222985
32
1.176584
0.092222
0.557464
33
1.284178
0.093370
1.003434
34
1.370300
0.092868
1.337912
35
1.452220
0.093384
1.672390
36
1.537018
0.093210
2.006869
37
1.648223
0.093412
2.452839
38
1.730115
0.093880
2.787318
39
1.779901
0.093622
3.010810
Table D.15: FFD coordinates for bending moment constrained case, M = 0.7
161
Control Point
x (m)
y (m)
z (m)
-0.030337
-0.090903
0.000000
0.043599
-0.212303
0.223019
0.115224
-0.138909
0.557547
0.235625
-0.190305
1.003585
0.315173
-0.163532
1.338113
0.404956
-0.195804
1.672642
0.484692
-0.171064
2.007170
0.605041
-0.214472
2.453208
0.682179
-0.177796
2.787736
0.743848
-0.241476
3.011258
10
-0.030337
0.090903
0.000000
11
0.022283
-0.031750
0.223019
12
0.106822
0.042703
0.557547
13
0.218188
-0.009337
1.003585
14
0.302447
0.017828
1.338113
15
0.386550
-0.014932
1.672642
16
0.470642
0.010198
2.007170
17
0.583342
-0.033965
2.453208
18
0.666944
0.003371
2.787736
19
0.717376
-0.061607
3.011258
20
1.040751
-0.090903
0.000000
21
1.107299
-0.086722
0.223019
22
1.185167
-0.089405
0.557547
23
1.301775
-0.087577
1.003585
24
1.383634
-0.088558
1.338113
25
1.470541
-0.087367
1.672642
26
1.552577
-0.088289
2.007170
27
1.668473
-0.086636
2.453208
28
1.749499
-0.088043
2.787736
29
1.803521
-0.085521
3.011258
30
1.040751
0.090903
0.000000
31
1.085983
0.093830
0.223019
32
1.176765
0.092207
0.557547
33
1.284338
0.093392
1.003585
34
1.370908
0.092802
1.338113
35
1.452135
0.093505
1.672642
36
1.538526
0.092974
2.007170
37
1.646774
0.093871
2.453208
38
1.734265
0.093124
2.787736
39
1.777049
0.094348
3.011258
Table D.16: FFD coordinates for Cm constrained case, M = 0.7
162
Control Point
x (m)
y (m)
z (m)
-0.030665
-0.091886
0.000000
0.080540
-0.214598
0.220996
0.209437
-0.153167
0.552489
0.404249
-0.202655
0.994481
0.541031
-0.186756
1.325975
0.686386
-0.215570
1.657468
0.824429
-0.207143
1.988962
1.017099
-0.236544
2.430954
1.154180
-0.224554
2.762447
1.244084
-0.263110
2.984542
10
-0.030665
0.091886
0.000000
11
0.058994
-0.032093
0.220996
12
0.198705
0.030292
0.552489
13
0.384810
-0.019914
0.994481
14
0.524393
-0.003739
1.325975
15
0.664668
-0.033085
1.657468
16
0.804198
-0.024488
1.988962
17
0.991675
-0.054539
2.430954
18
1.130876
-0.042265
2.762447
19
1.213957
-0.081824
2.984542
20
1.052005
-0.091886
0.000000
21
1.155743
-0.087660
0.220996
22
1.290259
-0.089940
0.552489
23
1.480845
-0.088129
0.994481
24
1.619254
-0.088735
1.325975
25
1.761469
-0.087621
1.657468
26
1.900518
-0.087955
1.988962
27
2.089358
-0.086764
2.430954
28
2.228109
-0.087259
2.762447
29
2.312106
-0.085620
2.984542
30
1.052005
0.091886
0.000000
31
1.134196
0.094845
0.220996
32
1.279526
0.093519
0.552489
33
1.461405
0.094612
0.994481
34
1.602616
0.094283
1.325975
35
1.739751
0.094863
1.657468
36
1.880287
0.094701
1.988962
37
2.063934
0.095242
2.430954
38
2.204805
0.095030
2.762447
39
2.281979
0.095666
2.984542
Table D.17: FFD coordinates for Kn constrained case, M = 0.7
163
Control Point
x (m)
y (m)
z (m)
-0.030665
-0.091888
0.000000
0.080544
-0.214602
0.220978
0.209242
-0.151770
0.552446
0.404110
-0.201854
0.994403
0.540951
-0.186186
1.325870
0.686312
-0.215073
1.657338
0.824276
-0.206205
1.988806
1.017236
-0.236957
2.430763
1.153892
-0.222999
2.762230
1.244105
-0.262989
2.984310
10
-0.030665
0.091888
0.000000
11
0.058997
-0.032094
0.220978
12
0.198755
0.031706
0.552446
13
0.384812
-0.019094
0.994403
14
0.524414
-0.003156
1.325870
15
0.664682
-0.032574
1.657338
16
0.804211
-0.023528
1.988806
17
0.991740
-0.054958
2.430763
18
1.130862
-0.040671
2.762230
19
1.213999
-0.081696
2.984310
20
1.052024
-0.091888
0.000000
21
1.155766
-0.087662
0.220978
22
1.290167
-0.089990
0.552446
23
1.480813
-0.088162
0.994403
24
1.619248
-0.088758
1.325870
25
1.761475
-0.087643
1.657338
26
1.900493
-0.087993
1.988806
27
2.089455
-0.086748
2.430763
28
2.228046
-0.087324
2.762230
29
2.312168
-0.085627
2.984310
30
1.052024
0.091888
0.000000
31
1.134219
0.094847
0.220978
32
1.279681
0.093487
0.552446
33
1.461515
0.094597
0.994403
34
1.602710
0.094272
1.325870
35
1.739845
0.094856
1.657338
36
1.880428
0.094684
1.988806
37
2.063959
0.095250
2.430763
38
2.205017
0.095003
2.762230
39
2.282062
0.095666
2.984310
Table D.18: FFD coordinates for CAP constrained case, M = 0.7
164
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.029924
-0.091807
-0.000000
CP
50
0.095851
0.083121
-0.000000
100
0.347399
-0.088165
0.000000
-0.029725
-0.094611
0.225218
51
0.095656
0.080326
0.225218
101
0.347598
-0.088969
0.225218
-0.029466
-0.097838
0.563046
52
0.095417
0.077823
0.563046
102
0.347852
-0.090367
0.563046
-0.028637
-0.106407
1.013483
53
0.094736
0.070098
1.013483
103
0.348652
-0.095490
1.013483
-0.029207
-0.099903
1.351311
54
0.095190
0.076751
1.351311
104
0.348105
-0.092948
1.351311
-0.028932
-0.102592
1.689138
55
0.094964
0.074667
1.689138
105
0.348370
-0.095236
1.689138
-0.028935
-0.102263
2.026966
56
0.094967
0.075294
2.026966
106
0.348367
-0.095320
2.026966
-0.028667
-0.104596
2.477403
57
0.094758
0.075222
2.477403
107
0.348624
-0.097740
2.477403
-0.028091
-0.109918
2.815231
58
0.094347
0.072023
2.815231
108
0.349167
-0.101591
2.815231
-0.027935
-0.110532
3.040246
59
0.094244
0.071660
3.040246
109
0.349313
-0.102505
3.040246
10
-0.029924
0.091807
-0.000000
60
0.196470
-0.089765
0.000000
110
0.347399
0.094561
0.000000
11
-0.030118
0.090109
0.225218
61
0.196669
-0.091732
0.225218
111
0.347204
0.093652
0.225218
12
-0.030356
0.087651
0.563046
62
0.196925
-0.093734
0.563046
112
0.346963
0.091858
0.563046
13
-0.031028
0.079039
1.013483
63
0.197736
-0.100290
1.013483
113
0.346262
0.085688
1.013483
14
-0.030581
0.084172
1.351311
64
0.197181
-0.095631
1.351311
114
0.346731
0.088896
1.351311
15
-0.030803
0.081174
1.689138
65
0.197449
-0.097707
1.689138
115
0.346499
0.086537
1.689138
16
-0.030801
0.080905
2.026966
66
0.197446
-0.097139
2.026966
116
0.346502
0.087200
2.026966
17
-0.031006
0.077811
2.477403
67
0.197708
-0.099979
2.477403
117
0.346285
0.083797
2.477403
18
-0.031405
0.071973
2.815231
68
0.198264
-0.104301
2.815231
118
0.345853
0.078334
2.815231
19
-0.031505
0.069655
3.040246
69
0.198414
-0.105415
3.040246
119
0.345743
0.076794
3.040246
20
0.020386
-0.093040
0.000000
70
0.196470
0.088007
0.000000
120
0.422864
-0.088010
-0.000000
21
0.020585
-0.096245
0.225218
71
0.196275
0.085464
0.225218
121
0.423062
-0.088516
0.225218
22
0.020843
-0.099416
0.563046
72
0.196035
0.082742
0.563046
122
0.423316
-0.089789
0.563046
23
0.021668
-0.107348
1.013483
73
0.195346
0.075776
1.013483
123
0.424110
-0.094358
1.013483
24
0.021102
-0.101309
1.351311
74
0.195807
0.081296
1.351311
124
0.423568
-0.092416
1.351311
25
0.021375
-0.104278
1.689138
75
0.195578
0.079396
1.689138
125
0.423831
-0.094502
1.689138
26
0.021372
-0.103708
2.026966
76
0.195581
0.080107
2.026966
126
0.423828
-0.095177
2.026966
27
0.021639
-0.105219
2.477403
77
0.195369
0.079995
2.477403
127
0.424083
-0.096961
2.477403
28
0.022210
-0.108908
2.815231
78
0.194950
0.074886
2.815231
128
0.424619
-0.100331
2.815231
29
0.022365
-0.109790
3.040246
79
0.194844
0.073752
3.040246
129
0.424763
-0.101221
3.040246
30
0.020386
0.084941
0.000000
80
0.271935
-0.088849
-0.000000
130
0.422864
0.095195
-0.000000
31
0.020192
0.080541
0.225218
81
0.272133
-0.090300
0.225218
131
0.422669
0.095010
0.225218
32
0.019953
0.077482
0.563046
82
0.272389
-0.091980
0.563046
132
0.422426
0.094372
0.563046
33
0.019278
0.069332
1.013483
83
0.273194
-0.097837
1.013483
133
0.421720
0.089383
1.013483
34
0.019728
0.076273
1.351311
84
0.272643
-0.094302
1.351311
134
0.422194
0.090538
1.351311
35
0.019504
0.073728
1.689138
85
0.272910
-0.096181
1.689138
135
0.421960
0.088354
1.689138
36
0.019506
0.074511
2.026966
86
0.272907
-0.096132
2.026966
136
0.421962
0.088260
2.026966
37
0.019299
0.073494
2.477403
87
0.273166
-0.098898
2.477403
137
0.421743
0.083906
2.477403
38
0.018896
0.069749
2.815231
88
0.273716
-0.102890
2.815231
138
0.421305
0.079738
2.815231
39
0.018794
0.070008
3.040246
89
0.273864
-0.104091
3.040246
139
0.421193
0.078455
3.040246
40
0.095851
-0.092275
-0.000000
90
0.271935
0.091591
-0.000000
140
0.498329
-0.088587
0.000000
41
0.096049
-0.094939
0.225218
91
0.271740
0.090083
0.225218
141
0.498526
-0.089196
0.225218
42
0.096307
-0.097401
0.563046
92
0.271499
0.087523
0.563046
142
0.498780
-0.089910
0.563046
43
0.097126
-0.104125
1.013483
93
0.270804
0.080974
1.013483
143
0.499568
-0.093949
1.013483
44
0.096564
-0.098864
1.351311
94
0.271269
0.085618
1.351311
144
0.499030
-0.092290
1.351311
45
0.096835
-0.100781
1.689138
95
0.271039
0.083599
1.689138
145
0.499291
-0.093813
1.689138
46
0.096832
-0.099914
2.026966
96
0.271041
0.084217
2.026966
146
0.499288
-0.094226
2.026966
47
0.097097
-0.101753
2.477403
97
0.270827
0.082656
2.477403
147
0.499541
-0.096337
2.477403
48
0.097662
-0.106419
2.815231
98
0.270401
0.076585
2.815231
148
0.500071
-0.099259
2.815231
49
0.097814
-0.107605
3.040246
99
0.270293
0.075436
3.040246
149
0.500213
-0.100089
3.040246
Table D.19: FFD coordinates for bending moment constrained case with shape variables, M =
0.7, part 1
165
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.498329
0.092793
0.000000
200
0.724722
-0.088116
-0.000000
250
0.900807
0.095777
0.000000
151
0.498133
0.093368
0.225218
201
0.724920
-0.088163
0.225218
251
0.900610
0.096417
0.225218
152
0.497890
0.093334
0.563046
202
0.725171
-0.088866
0.563046
252
0.900363
0.095963
0.563046
153
0.497178
0.089582
1.013483
203
0.725942
-0.090541
1.013483
253
0.899620
0.095075
1.013483
154
0.497656
0.089614
1.351311
204
0.725417
-0.090252
1.351311
254
0.900122
0.093688
1.351311
155
0.497420
0.087021
1.689138
205
0.725673
-0.091214
1.689138
255
0.899876
0.092417
1.689138
156
0.497423
0.086277
2.026966
206
0.725670
-0.091574
2.026966
256
0.899879
0.091383
2.026966
157
0.497201
0.083654
2.477403
207
0.725915
-0.092901
2.477403
257
0.899645
0.088791
2.477403
158
0.496757
0.080811
2.815231
208
0.726426
-0.094881
2.815231
258
0.899166
0.086502
2.815231
159
0.496642
0.080012
3.040246
209
0.726561
-0.095102
3.040246
259
0.899040
0.085044
3.040246
160
0.573793
-0.089324
-0.000000
210
0.724722
0.091188
0.000000
260
0.976271
-0.089354
0.000000
161
0.573991
-0.089590
0.225218
211
0.724526
0.091094
0.225218
261
0.976468
-0.089371
0.225218
162
0.574243
-0.090459
0.563046
212
0.724281
0.091209
0.563046
262
0.976716
-0.089428
0.563046
163
0.575026
-0.093825
1.013483
213
0.723551
0.088833
1.013483
263
0.977468
-0.089507
1.013483
164
0.574493
-0.092056
1.351311
214
0.724043
0.089043
1.351311
264
0.976959
-0.089583
1.351311
165
0.574752
-0.093497
1.689138
215
0.723802
0.088063
1.689138
265
0.977208
-0.089830
1.689138
166
0.574749
-0.093599
2.026966
216
0.723804
0.087659
2.026966
266
0.977205
-0.089798
2.026966
167
0.574999
-0.095207
2.477403
217
0.723576
0.086228
2.477403
267
0.977443
-0.090079
2.477403
168
0.575522
-0.097553
2.815231
218
0.723112
0.084078
2.815231
268
0.977932
-0.089954
2.815231
169
0.575662
-0.098204
3.040246
219
0.722991
0.083947
3.040246
269
0.978060
-0.090786
3.040246
170
0.573793
0.089992
-0.000000
220
0.800187
-0.086430
0.000000
270
0.976271
0.090671
-0.000000
171
0.573598
0.090109
0.225218
221
0.800384
-0.086145
0.225218
271
0.976075
0.090858
0.225218
172
0.573354
0.089406
0.563046
222
0.800634
-0.086440
0.563046
272
0.975827
0.090699
0.563046
173
0.572636
0.086180
1.013483
223
0.801400
-0.087909
1.013483
273
0.975078
0.090714
1.013483
174
0.573119
0.087288
1.351311
224
0.800880
-0.088335
1.351311
274
0.975585
0.090607
1.351311
175
0.572881
0.085641
1.689138
225
0.801133
-0.089512
1.689138
275
0.975337
0.090417
1.689138
176
0.572883
0.085433
2.026966
226
0.801130
-0.090338
2.026966
276
0.975340
0.090334
2.026966
177
0.572659
0.084028
2.477403
227
0.801374
-0.091783
2.477403
277
0.975103
0.090204
2.477403
178
0.572208
0.081686
2.815231
228
0.801878
-0.093740
2.815231
278
0.974618
0.089943
2.815231
179
0.572092
0.081224
3.040246
229
0.802011
-0.094541
3.040246
279
0.974490
0.089271
3.040246
180
0.649258
-0.089261
0.000000
230
0.800187
0.093169
-0.000000
280
1.026581
-0.090077
-0.000000
181
0.649455
-0.089461
0.225218
231
0.799991
0.093593
0.225218
281
1.026777
-0.089972
0.225218
182
0.649707
-0.090222
0.563046
232
0.799745
0.093430
0.563046
282
1.027025
-0.089866
0.563046
183
0.650484
-0.092910
1.013483
233
0.799009
0.091995
1.013483
283
1.027773
-0.089605
1.013483
184
0.649955
-0.091525
1.351311
234
0.799506
0.091346
1.351311
284
1.027267
-0.089759
1.351311
185
0.650212
-0.092692
1.689138
235
0.799262
0.090083
1.689138
285
1.027515
-0.089697
1.689138
186
0.650209
-0.092689
2.026966
236
0.799265
0.089495
2.026966
286
1.027512
-0.089687
2.026966
187
0.650457
-0.094167
2.477403
237
0.799034
0.087513
2.477403
287
1.027748
-0.089609
2.477403
188
0.650974
-0.096023
2.815231
238
0.798564
0.085323
2.815231
288
1.028233
-0.089517
2.815231
189
0.651112
-0.096506
3.040246
239
0.798441
0.084438
3.040246
289
1.028360
-0.089459
3.040246
190
0.649258
0.089939
-0.000000
240
0.900807
-0.083782
-0.000000
290
1.026581
0.090077
0.000000
191
0.649062
0.089731
0.225218
241
0.901003
-0.083055
0.225218
291
1.026384
0.090103
0.225218
192
0.648817
0.089347
0.563046
242
0.901253
-0.083416
0.563046
292
1.026136
0.090163
0.563046
193
0.648093
0.086577
1.013483
243
0.902010
-0.084308
1.013483
293
1.025383
0.090402
1.013483
194
0.648581
0.087553
1.351311
244
0.901496
-0.085738
1.351311
294
1.025893
0.090218
1.351311
195
0.648341
0.086396
1.689138
245
0.901747
-0.087035
1.689138
295
1.025644
0.090320
1.689138
196
0.648344
0.086276
2.026966
246
0.901744
-0.088038
2.026966
296
1.025647
0.090309
2.026966
197
0.648118
0.085105
2.477403
247
0.901984
-0.090520
2.477403
297
1.025409
0.090389
2.477403
198
0.647660
0.082910
2.815231
248
0.902480
-0.092621
2.815231
298
1.024919
0.090622
2.815231
199
0.647542
0.082566
3.040246
249
0.902611
-0.093871
3.040246
299
1.024790
0.090650
3.040246
0.7, part 2
166
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.029730
-0.093868
-0.000000
CP
50
0.095231
0.081877
-0.000000
100
0.345154
-0.086666
0.000000
-0.029611
-0.091086
0.226721
51
0.095784
0.080475
0.226721
101
0.345273
-0.083576
0.226721
-0.029232
-0.089027
0.566803
52
0.096414
0.082557
0.566803
102
0.345650
-0.081413
0.566803
-0.028200
-0.093504
1.020245
53
0.096702
0.078580
1.020245
103
0.346685
-0.083220
1.020245
-0.028110
-0.088782
1.360327
54
0.097630
0.084793
1.360327
104
0.346771
-0.081710
1.360327
-0.026797
-0.098055
1.700409
55
0.097320
0.075692
1.700409
105
0.348084
-0.088199
1.700409
-0.029336
-0.054701
2.040490
56
0.102161
0.116647
2.040490
106
0.345290
-0.060962
2.040490
-0.025715
-0.096716
2.493933
57
0.098565
0.080443
2.493933
107
0.349167
-0.089182
2.493933
-0.024142
-0.107400
2.834014
58
0.098185
0.073134
2.834014
108
0.350702
-0.097848
2.834014
-0.022949
-0.114790
3.060140
59
0.097872
0.067657
3.060140
109
0.351836
-0.103708
3.060140
10
-0.029730
0.093868
-0.000000
60
0.195201
-0.087031
0.000000
110
0.345154
0.093933
0.000000
11
-0.029177
0.096056
0.226721
61
0.195319
-0.084187
0.226721
111
0.345707
0.096498
0.226721
12
-0.028547
0.096870
0.566803
62
0.195697
-0.082444
0.566803
112
0.346335
0.097551
0.566803
13
-0.028260
0.092818
1.020245
63
0.196731
-0.084783
1.020245
113
0.346625
0.094915
1.020245
14
-0.027330
0.097711
1.360327
64
0.196819
-0.081895
1.360327
114
0.347551
0.097491
1.360327
15
-0.027640
0.088406
1.700409
65
0.198131
-0.088801
1.700409
115
0.347241
0.092644
1.700409
16
-0.022714
0.130513
2.040490
66
0.195440
-0.054004
2.040490
116
0.351911
0.119984
2.040490
17
-0.026396
0.088928
2.493933
67
0.199214
-0.087898
2.493933
117
0.348486
0.092002
2.493933
18
-0.026763
0.077389
2.834014
68
0.200765
-0.097527
2.834014
118
0.348081
0.080540
2.834014
19
-0.027057
0.067761
3.060140
69
0.201922
-0.107990
3.060140
119
0.347728
0.074034
3.060140
20
0.020254
-0.092759
0.000000
70
0.195201
0.086507
0.000000
120
0.420131
-0.088052
-0.000000
21
0.020374
-0.096510
0.226721
71
0.195753
0.086154
0.226721
121
0.420249
-0.085252
0.226721
22
0.020753
-0.094904
0.566803
72
0.196382
0.087704
0.566803
122
0.420627
-0.083007
0.566803
23
0.021785
-0.099357
1.020245
73
0.196671
0.084729
1.020245
123
0.421662
-0.084876
1.020245
24
0.021874
-0.091537
1.360327
74
0.197598
0.089184
1.360327
124
0.421747
-0.083401
1.360327
25
0.023187
-0.100951
1.700409
75
0.197289
0.082502
1.700409
125
0.423060
-0.089332
1.700409
26
0.020614
-0.061062
2.040490
76
0.202061
0.118317
2.040490
126
0.420215
-0.066418
2.040490
27
0.024269
-0.097518
2.493933
77
0.198534
0.086171
2.493933
127
0.424143
-0.090907
2.493933
28
0.025837
-0.105957
2.834014
78
0.198143
0.076528
2.834014
128
0.425671
-0.098158
2.834014
29
0.027022
-0.112646
3.060140
79
0.197814
0.070118
3.060140
129
0.426793
-0.103071
3.060140
30
0.020254
0.083215
0.000000
80
0.270178
-0.086480
-0.000000
130
0.420131
0.096653
-0.000000
31
0.020808
0.080520
0.226721
81
0.270296
-0.083230
0.226721
131
0.420684
0.099809
0.226721
32
0.021438
0.083722
0.566803
82
0.270674
-0.081126
0.566803
132
0.421312
0.101788
0.566803
33
0.021725
0.079415
1.020245
83
0.271708
-0.083316
1.020245
133
0.421602
0.098815
1.020245
34
0.022654
0.085582
1.360327
84
0.271795
-0.081461
1.360327
134
0.422527
0.100462
1.360327
35
0.022344
0.076380
1.700409
85
0.273107
-0.087758
1.700409
135
0.422217
0.095266
1.700409
36
0.027236
0.117061
2.040490
86
0.270365
-0.056344
2.040490
136
0.426837
0.118301
2.040490
37
0.023589
0.080207
2.493933
87
0.274190
-0.088123
2.493933
137
0.423463
0.091714
2.493933
38
0.023216
0.071381
2.834014
88
0.275733
-0.097420
2.834014
138
0.423050
0.082072
2.834014
39
0.022915
0.066721
3.060140
89
0.276879
-0.105579
3.060140
139
0.422685
0.076060
3.060140
40
0.095231
-0.089638
-0.000000
90
0.270178
0.090565
-0.000000
140
0.495108
-0.089302
0.000000
41
0.095350
-0.088057
0.226721
91
0.270730
0.091980
0.226721
141
0.495226
-0.086831
0.226721
42
0.095729
-0.085583
0.566803
92
0.271359
0.092494
0.566803
142
0.495603
-0.085410
0.566803
43
0.096762
-0.088377
1.020245
93
0.271648
0.090130
1.020245
143
0.496639
-0.086795
1.020245
44
0.096850
-0.084437
1.360327
94
0.272575
0.093331
1.360327
144
0.496723
-0.085476
1.360327
45
0.098163
-0.092561
1.700409
95
0.272265
0.088187
1.700409
145
0.498036
-0.090629
1.700409
46
0.095539
-0.053632
2.040490
96
0.276986
0.119452
2.040490
146
0.495140
-0.071352
2.040490
47
0.099245
-0.090108
2.493933
97
0.273510
0.090261
2.493933
147
0.499119
-0.092363
2.493933
48
0.100806
-0.099588
2.834014
98
0.273112
0.078532
2.834014
148
0.500640
-0.097663
2.834014
49
0.101979
-0.110391
3.060140
99
0.272771
0.072150
3.060140
149
0.501750
-0.102905
3.060140
167
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.495108
0.095495
0.000000
200
0.720039
-0.087736
-0.000000
250
0.894985
0.091845
0.000000
151
0.495660
0.098850
0.226721
201
0.720156
-0.087567
0.226721
251
0.895536
0.089783
0.226721
152
0.496288
0.102059
0.566803
202
0.720532
-0.087367
0.566803
252
0.896162
0.089746
0.566803
153
0.496579
0.100016
1.020245
203
0.721569
-0.089614
1.020245
253
0.896456
0.087955
1.020245
154
0.497503
0.099996
1.360327
204
0.721652
-0.089764
1.360327
254
0.897377
0.086170
1.360327
155
0.497193
0.094194
1.700409
205
0.722964
-0.094110
1.700409
255
0.897066
0.083803
1.700409
156
0.501762
0.112550
2.040490
206
0.719916
-0.086253
2.040490
256
0.901363
0.077834
2.040490
157
0.498439
0.088972
2.493933
207
0.724049
-0.094924
2.493933
257
0.898313
0.080171
2.493933
158
0.498019
0.081761
2.834014
208
0.725546
-0.095542
2.834014
258
0.897853
0.085146
2.834014
159
0.497642
0.077848
3.060140
209
0.726621
-0.095749
3.060140
259
0.897413
0.075210
3.060140
160
0.570085
-0.088595
-0.000000
210
0.720039
0.089459
0.000000
260
0.969962
-0.091656
0.000000
161
0.570203
-0.087027
0.226721
211
0.720590
0.089066
0.226721
261
0.970079
-0.092986
0.226721
162
0.570579
-0.086093
0.566803
212
0.721217
0.088692
0.566803
262
0.970453
-0.094521
0.566803
163
0.571616
-0.087800
1.020245
213
0.721510
0.086578
1.020245
263
0.971493
-0.096202
1.020245
164
0.571700
-0.086768
1.360327
214
0.722432
0.086796
1.360327
264
0.971573
-0.095845
1.360327
165
0.573012
-0.091026
1.700409
215
0.722121
0.082474
1.700409
265
0.972884
-0.096835
1.700409
166
0.570065
-0.075261
2.040490
216
0.726537
0.089435
2.040490
266
0.969666
-0.100379
2.040490
167
0.574096
-0.092849
2.493933
217
0.723368
0.080228
2.493933
267
0.973970
-0.096446
2.493933
168
0.575609
-0.096888
2.834014
218
0.722925
0.079066
2.834014
268
0.975443
-0.087048
2.834014
169
0.576707
-0.100057
3.060140
219
0.722514
0.079723
3.060140
269
0.976478
-0.090033
3.060140
170
0.570085
0.092157
-0.000000
220
0.795016
-0.087643
0.000000
270
0.969962
0.090933
-0.000000
171
0.570637
0.094552
0.226721
221
0.795133
-0.088573
0.226721
271
0.970513
0.089032
0.226721
172
0.571264
0.096674
0.566803
222
0.795509
-0.089149
0.566803
272
0.971138
0.087589
0.566803
173
0.571556
0.095684
1.020245
223
0.796546
-0.091636
1.020245
273
0.971433
0.086273
1.020245
174
0.572480
0.094874
1.360327
224
0.796628
-0.091781
1.360327
274
0.972353
0.086324
1.360327
175
0.572169
0.089234
1.700409
225
0.797940
-0.095284
1.700409
275
0.972042
0.085407
1.700409
176
0.576687
0.103894
2.040490
226
0.794841
-0.093040
2.040490
276
0.976288
0.081541
2.040490
177
0.573416
0.085208
2.493933
227
0.799025
-0.096430
2.493933
277
0.973290
0.085527
2.493933
178
0.572988
0.080721
2.834014
228
0.800515
-0.094038
2.834014
278
0.972821
0.093567
2.834014
179
0.572599
0.077683
3.060140
229
0.801578
-0.096381
3.060140
279
0.972370
0.089677
3.060140
180
0.645062
-0.088206
0.000000
230
0.795016
0.090573
-0.000000
280
1.019947
-0.090938
-0.000000
181
0.645180
-0.087338
0.226721
231
0.795567
0.089446
0.226721
281
1.020063
-0.090907
0.226721
182
0.645556
-0.086856
0.566803
232
0.796194
0.088627
0.566803
282
1.020438
-0.090832
0.566803
183
0.646592
-0.088447
1.020245
233
0.796486
0.086520
1.020245
283
1.021477
-0.090646
1.020245
184
0.646676
-0.088075
1.360327
234
0.797408
0.086073
1.360327
284
1.021557
-0.090848
1.360327
185
0.647988
-0.092535
1.700409
235
0.797097
0.082724
1.700409
285
1.022868
-0.090608
1.700409
186
0.644990
-0.080171
2.040490
236
0.801462
0.084572
2.040490
286
1.019616
-0.091688
2.040490
187
0.649072
-0.093859
2.493933
237
0.798345
0.079763
2.493933
287
1.023954
-0.090630
2.493933
188
0.650578
-0.095999
2.834014
238
0.797894
0.080879
2.834014
288
1.025422
-0.090326
2.834014
189
0.651664
-0.097517
3.060140
239
0.797471
0.078276
3.060140
289
1.026449
-0.090244
3.060140
190
0.645062
0.089841
-0.000000
240
0.894985
-0.086861
-0.000000
290
1.019947
0.090938
0.000000
191
0.645614
0.090834
0.226721
241
0.895102
-0.088637
0.226721
291
1.020497
0.090762
0.226721
192
0.646241
0.091732
0.566803
242
0.895477
-0.088419
0.566803
292
1.021123
0.090604
0.566803
193
0.646533
0.090539
1.020245
243
0.896516
-0.090168
1.020245
293
1.021417
0.090666
1.020245
194
0.647456
0.090167
1.360327
244
0.896597
-0.092172
1.360327
294
1.022337
0.090588
1.360327
195
0.647145
0.084757
1.700409
245
0.897908
-0.094247
1.700409
295
1.022026
0.090889
1.700409
196
0.651612
0.096099
2.040490
246
0.894741
-0.099192
2.040490
296
1.026238
0.089480
2.040490
197
0.648392
0.082146
2.493933
247
0.898994
-0.096527
2.493933
297
1.023274
0.090857
2.493933
198
0.647956
0.079605
2.834014
248
0.900474
-0.089246
2.834014
298
1.022801
0.091200
2.834014
199
0.647557
0.078695
3.060140
249
0.901521
-0.098328
3.060140
299
1.022342
0.091614
3.060140
168
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.029741
-0.092024
-0.000000
CP
50
0.095266
0.081621
-0.000000
100
0.345279
-0.084356
0.000000
-0.028170
-0.086875
0.226618
51
0.097922
0.086604
0.226618
101
0.346844
-0.080746
0.226618
-0.025047
-0.087510
0.566546
52
0.101091
0.086092
0.566546
102
0.349966
-0.079096
0.566546
-0.020511
-0.091470
1.019783
53
0.104906
0.082738
1.019783
103
0.354508
-0.082495
1.019783
-0.017835
-0.085113
1.359711
54
0.108573
0.090670
1.359711
104
0.357174
-0.080646
1.359711
-0.013959
-0.093829
1.699638
55
0.110927
0.081830
1.699638
105
0.361061
-0.082731
1.699638
-0.012020
-0.078714
2.039566
56
0.115463
0.096573
2.039566
106
0.362963
-0.074201
2.039566
-0.006680
-0.092792
2.492803
57
0.118320
0.084741
2.492803
107
0.368340
-0.084685
2.492803
-0.002471
-0.103213
2.832731
58
0.120521
0.075769
2.832731
108
0.372526
-0.094063
2.832731
0.000441
-0.112603
3.058777
59
0.121569
0.072042
3.058777
109
0.375373
-0.107161
3.058777
10
-0.029741
0.092024
-0.000000
60
0.195271
-0.087789
0.000000
110
0.345279
0.094672
0.000000
11
-0.027082
0.099327
0.226618
61
0.196838
-0.081355
0.226618
111
0.347931
0.100896
0.226618
12
-0.023913
0.100489
0.566546
62
0.199961
-0.079142
0.566546
112
0.351100
0.101520
0.566546
13
-0.020100
0.096177
1.019783
63
0.204500
-0.080950
1.019783
113
0.354919
0.099701
1.019783
14
-0.016429
0.101206
1.359711
64
0.207170
-0.078312
1.359711
114
0.358579
0.103119
1.359711
15
-0.014080
0.092447
1.699638
65
0.211053
-0.085773
1.699638
115
0.360940
0.097144
1.699638
16
-0.009532
0.107208
2.039566
66
0.212970
-0.073525
2.039566
116
0.365452
0.106187
2.039566
17
-0.006686
0.092719
2.492803
67
0.218332
-0.084460
2.492803
117
0.368334
0.094030
2.492803
18
-0.004477
0.080240
2.832731
68
0.222527
-0.092628
2.832731
118
0.370519
0.083409
2.832731
19
-0.003409
0.068529
3.058777
69
0.225400
-0.112630
3.058777
119
0.371524
0.076622
3.058777
20
0.020262
-0.094092
0.000000
70
0.195271
0.086882
0.000000
120
0.420283
-0.083788
-0.000000
21
0.021832
-0.089693
0.226618
71
0.197926
0.091716
0.226618
121
0.421846
-0.080220
0.226618
22
0.024955
-0.089911
0.566546
72
0.201094
0.091910
0.566546
122
0.424968
-0.079742
0.566546
23
0.029492
-0.094128
1.019783
73
0.204912
0.088249
1.019783
123
0.429512
-0.083090
1.019783
24
0.032166
-0.086060
1.359711
74
0.208576
0.095436
1.359711
124
0.432175
-0.084000
1.359711
25
0.036044
-0.095066
1.699638
75
0.210932
0.088148
1.699638
125
0.436065
-0.080981
1.699638
26
0.037978
-0.080654
2.039566
76
0.215459
0.101320
2.039566
126
0.437960
-0.077115
2.039566
27
0.043323
-0.093521
2.492803
77
0.218326
0.088668
2.492803
127
0.443344
-0.086708
2.492803
28
0.047529
-0.103660
2.832731
78
0.220520
0.081189
2.832731
128
0.447525
-0.095901
2.832731
29
0.050432
-0.110905
3.058777
79
0.221551
0.072031
3.058777
129
0.450360
-0.103198
3.058777
30
0.020262
0.084945
0.000000
80
0.270275
-0.086044
-0.000000
130
0.420283
0.097632
-0.000000
31
0.022920
0.086981
0.226618
81
0.271841
-0.081200
0.226618
131
0.422934
0.103373
0.226618
32
0.026089
0.086815
0.566546
82
0.274963
-0.078814
0.566546
132
0.426102
0.105018
0.566546
33
0.029903
0.082863
1.019783
83
0.279504
-0.081675
1.019783
133
0.429923
0.104035
1.019783
34
0.033572
0.090773
1.359711
84
0.282172
-0.078661
1.359711
134
0.433581
0.106081
1.359711
35
0.035923
0.082477
1.699638
85
0.286057
-0.083930
1.699638
135
0.435944
0.098467
1.699638
36
0.040466
0.096558
2.039566
86
0.287967
-0.073296
2.039566
136
0.440449
0.106509
2.039566
37
0.043316
0.084434
2.492803
87
0.293336
-0.084130
2.492803
137
0.443338
0.094108
2.492803
38
0.045522
0.075336
2.832731
88
0.297526
-0.094349
2.832731
138
0.445518
0.084576
2.832731
39
0.046582
0.067732
3.058777
89
0.300387
-0.108883
3.058777
139
0.446510
0.077432
3.058777
40
0.095266
-0.089161
-0.000000
90
0.270275
0.090991
-0.000000
140
0.495287
-0.085147
0.000000
41
0.096835
-0.083405
0.226618
91
0.272929
0.097315
0.226618
141
0.496849
-0.081561
0.226618
42
0.099957
-0.081800
0.566546
92
0.276097
0.096460
0.566546
142
0.499971
-0.082201
0.566546
43
0.104495
-0.083773
1.019783
93
0.279915
0.094750
1.019783
143
0.504516
-0.084419
1.019783
44
0.107168
-0.080087
1.359711
94
0.283577
0.099477
1.359711
144
0.507177
-0.082569
1.359711
45
0.111048
-0.088192
1.699638
95
0.285936
0.092973
1.699638
145
0.511069
-0.085571
1.699638
46
0.112974
-0.072719
2.039566
96
0.290455
0.104156
2.039566
146
0.512957
-0.078514
2.039566
47
0.118327
-0.085724
2.492803
97
0.293330
0.092160
2.492803
147
0.518348
-0.086981
2.492803
48
0.122528
-0.097435
2.832731
98
0.295520
0.083341
2.832731
148
0.522524
-0.097021
2.832731
49
0.125418
-0.108311
3.058777
99
0.296537
0.073066
3.058777
149
0.525346
-0.102081
3.058777
169
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.495287
0.097268
0.000000
200
0.720299
-0.087492
-0.000000
250
0.895309
0.089766
0.000000
151
0.497936
0.102311
0.226618
201
0.721857
-0.088436
0.226618
251
0.897951
0.088649
0.226618
152
0.501105
0.104858
0.566546
202
0.724978
-0.088179
0.566546
252
0.901118
0.087385
0.566546
153
0.504927
0.105034
1.019783
203
0.729527
-0.089680
1.019783
253
0.904947
0.085772
1.019783
154
0.508583
0.103113
1.359711
204
0.732182
-0.089053
1.359711
254
0.908592
0.084582
1.359711
155
0.510948
0.097774
1.699638
205
0.736081
-0.091966
1.699638
255
0.910970
0.082974
1.699638
156
0.515446
0.102499
2.039566
206
0.737947
-0.090465
2.039566
256
0.915428
0.081426
2.039566
157
0.518342
0.091015
2.492803
207
0.743361
-0.093316
2.492803
257
0.918364
0.081805
2.492803
158
0.520518
0.084561
2.832731
208
0.747522
-0.096075
2.832731
258
0.920514
0.083106
2.832731
159
0.521497
0.078293
3.058777
209
0.750306
-0.097606
3.058777
259
0.921425
0.079641
3.058777
160
0.570291
-0.086279
-0.000000
210
0.720299
0.089767
0.000000
260
0.970313
-0.096733
0.000000
161
0.571851
-0.083571
0.226618
211
0.722944
0.091073
0.226618
261
0.971866
-0.095659
0.226618
162
0.574973
-0.083679
0.566546
212
0.726112
0.091015
0.566546
262
0.974987
-0.093858
0.566546
163
0.579520
-0.084822
1.019783
213
0.729938
0.088452
1.019783
263
0.979540
-0.094332
1.019783
164
0.582179
-0.083433
1.359711
214
0.733588
0.088470
1.359711
264
0.982188
-0.094975
1.359711
165
0.586073
-0.087521
1.699638
215
0.735960
0.085197
1.699638
265
0.986095
-0.095077
1.699638
166
0.587954
-0.082127
2.039566
216
0.740436
0.087337
2.039566
266
0.987936
-0.095918
2.039566
167
0.593352
-0.089526
2.492803
217
0.743354
0.085051
2.492803
267
0.993374
-0.094696
2.492803
168
0.597523
-0.095952
2.832731
218
0.745516
0.082160
2.832731
268
0.997520
-0.090486
2.832731
169
0.600333
-0.100154
3.058777
219
0.746457
0.078848
3.058777
269
1.000261
-0.088474
3.058777
170
0.570291
0.092964
-0.000000
220
0.795303
-0.088626
0.000000
270
0.970313
0.089666
-0.000000
171
0.572939
0.097151
0.226618
221
0.796859
-0.088792
0.226618
271
0.972953
0.085084
0.226618
172
0.576107
0.098922
0.566546
222
0.799981
-0.090060
0.566546
272
0.976121
0.086019
0.566546
173
0.579931
0.098625
1.019783
223
0.804531
-0.092153
1.019783
273
0.979951
0.086431
1.019783
174
0.583584
0.098001
1.359711
224
0.807184
-0.091972
1.359711
274
0.983593
0.085494
1.359711
175
0.585952
0.092576
1.699638
225
0.811085
-0.094871
1.699638
275
0.985974
0.085910
1.699638
176
0.590442
0.096316
2.039566
226
0.812944
-0.094246
2.039566
276
0.990425
0.084713
2.039566
177
0.593346
0.089178
2.492803
227
0.818365
-0.095555
2.492803
277
0.993368
0.086401
2.492803
178
0.595517
0.083377
2.832731
228
0.822521
-0.095881
2.832731
278
0.995513
0.089981
2.832731
179
0.596483
0.076353
3.058777
229
0.825293
-0.096318
3.058777
279
0.996412
0.092439
3.058777
180
0.645295
-0.088224
0.000000
230
0.795303
0.089718
-0.000000
280
1.020316
-0.089658
-0.000000
181
0.646854
-0.085364
0.226618
231
0.797947
0.089541
0.226618
281
1.021867
-0.089659
0.226618
182
0.649976
-0.086074
0.566546
232
0.801115
0.088214
0.566546
282
1.024989
-0.090190
0.566546
183
0.654524
-0.087375
1.019783
233
0.804942
0.086218
1.019783
283
1.029543
-0.089994
1.019783
184
0.657180
-0.085698
1.359711
234
0.808589
0.086539
1.359711
284
1.032189
-0.089867
1.359711
185
0.661077
-0.089508
1.699638
235
0.810964
0.084474
1.699638
285
1.036097
-0.089757
1.699638
186
0.662950
-0.086398
2.039566
236
0.815432
0.084572
2.039566
286
1.037934
-0.089974
2.039566
187
0.668356
-0.092448
2.492803
237
0.818358
0.082940
2.492803
287
1.043377
-0.089767
2.492803
188
0.672523
-0.091577
2.832731
238
0.820515
0.081771
2.832731
288
1.047519
-0.089596
2.832731
189
0.675320
-0.104699
3.058777
239
0.821443
0.080841
3.058777
289
1.050252
-0.089248
3.058777
190
0.645295
0.091446
-0.000000
240
0.895309
-0.088616
-0.000000
290
1.020316
0.089658
0.000000
191
0.647942
0.093031
0.226618
241
0.896863
-0.089437
0.226618
291
1.022955
0.089297
0.226618
192
0.651110
0.094168
0.566546
242
0.899984
-0.090951
0.566546
292
1.026122
0.089811
0.566546
193
0.654935
0.093066
1.019783
243
0.904536
-0.093008
1.019783
293
1.029954
0.089857
1.019783
194
0.658586
0.092786
1.359711
244
0.907186
-0.093854
1.359711
294
1.033595
0.089399
1.359711
195
0.660956
0.088500
1.699638
245
0.911091
-0.095783
1.699638
295
1.035977
0.089797
1.699638
196
0.665439
0.091569
2.039566
246
0.912939
-0.096910
2.039566
296
1.040423
0.089144
2.039566
197
0.668350
0.086484
2.492803
247
0.918370
-0.096526
2.492803
297
1.043370
0.089770
2.492803
198
0.670516
0.083570
2.832731
248
0.922520
-0.093491
2.832731
298
1.045513
0.090265
2.832731
199
0.671470
0.076272
3.058777
249
0.925275
-0.096640
3.058777
299
1.046403
0.090531
3.058777
170
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.030308
-0.090394
-0.000000
CP
50
0.097082
0.091122
-0.000000
100
0.351862
-0.094735
0.000000
0.023903
-0.081391
0.222358
51
0.153208
0.094976
0.222358
101
0.406052
-0.072383
0.222358
0.107600
-0.093869
0.555896
52
0.234746
0.085130
0.555896
102
0.489770
-0.076288
0.555896
0.217899
-0.096595
1.000613
53
0.344904
0.090128
1.000613
103
0.600069
-0.076374
1.000613
0.301666
-0.105908
1.334151
54
0.426646
0.078816
1.334151
104
0.683803
-0.083915
1.334151
0.383740
-0.103578
1.667688
55
0.509795
0.081243
1.667688
105
0.765900
-0.080609
1.667688
0.467896
-0.117331
2.001226
56
0.591375
0.068907
2.001226
106
0.849979
-0.087860
2.001226
0.578155
-0.124653
2.445943
57
0.701583
0.069121
2.445943
107
0.960236
-0.094475
2.445943
0.666936
-0.171458
2.779481
58
0.781990
0.020368
2.779481
108
1.048264
-0.130132
2.779481
0.707436
-0.107970
3.002401
59
0.832834
0.081507
3.002401
109
1.089583
-0.100812
3.002401
10
-0.030308
0.090394
-0.000000
60
0.198994
-0.097373
0.000000
110
0.351862
0.086481
0.000000
11
0.025825
0.103388
0.222358
61
0.253193
-0.065366
0.222358
111
0.407974
0.105447
0.222358
12
0.107356
0.091076
0.555896
62
0.336902
-0.076834
0.555896
112
0.489526
0.099680
0.555896
13
0.217515
0.092188
1.000613
63
0.447201
-0.075447
1.000613
113
0.599684
0.113629
1.000613
14
0.299267
0.078443
1.334151
64
0.530948
-0.084828
1.334151
114
0.681404
0.103023
1.334151
15
0.382409
0.088332
1.667688
65
0.613036
-0.079488
1.667688
115
0.764569
0.102196
1.667688
16
0.464014
0.072883
2.001226
66
0.697146
-0.087464
2.001226
116
0.846097
0.095397
2.001226
17
0.574223
0.079631
2.445943
67
0.807403
-0.091689
2.445943
117
0.956303
0.087477
2.445943
18
0.654881
0.033441
2.779481
68
0.895732
-0.131047
2.779481
118
1.036209
0.052313
2.779481
19
0.705451
0.085249
3.002401
69
0.936724
-0.101710
3.002401
119
1.087599
0.087776
3.002401
20
0.020648
-0.084173
0.000000
70
0.198994
0.088181
0.000000
120
0.428296
-0.091571
-0.000000
21
0.074857
-0.087699
0.222358
71
0.255114
0.100594
0.222358
121
0.482482
-0.077626
0.222358
22
0.158556
-0.092209
0.555896
72
0.336658
0.090808
0.555896
122
0.566204
-0.079109
0.555896
23
0.268855
-0.092526
1.000613
73
0.446816
0.099768
1.000613
123
0.676503
-0.078858
1.000613
24
0.352617
-0.099853
1.334151
74
0.528549
0.088758
1.334151
124
0.760230
-0.085217
1.334151
25
0.434695
-0.099536
1.667688
75
0.611705
0.089651
1.667688
125
0.842332
-0.083402
1.667688
26
0.518840
-0.107602
2.001226
76
0.693264
0.080564
2.001226
126
0.926396
-0.089620
2.001226
27
0.629099
-0.102680
2.445943
77
0.803471
0.076514
2.445943
127
1.036652
-0.097124
2.445943
28
0.717780
-0.147156
2.779481
78
0.883678
0.034242
2.779481
128
1.124529
-0.128953
2.779481
29
0.758389
-0.094781
3.002401
79
0.934740
0.084592
3.002401
129
1.166013
-0.099860
3.002401
30
0.020648
0.093677
0.000000
80
0.275428
-0.097153
-0.000000
130
0.428296
0.087742
-0.000000
31
0.076778
0.094245
0.222358
81
0.329623
-0.067481
0.222358
131
0.484403
0.104634
0.222358
32
0.158312
0.083554
0.555896
82
0.413336
-0.075474
0.555896
132
0.565960
0.102907
0.555896
33
0.268470
0.085921
1.000613
83
0.523635
-0.075148
1.000613
133
0.676118
0.118874
1.000613
34
0.350218
0.074761
1.334151
84
0.607375
-0.083378
1.334151
134
0.757831
0.108746
1.334151
35
0.433363
0.079163
1.667688
85
0.689468
-0.079163
1.667688
135
0.841001
0.107442
1.667688
36
0.514958
0.064374
2.001226
86
0.773562
-0.086609
2.001226
136
0.922513
0.099585
2.001226
37
0.625167
0.065775
2.445943
87
0.883820
-0.092322
2.445943
137
1.032720
0.091912
2.445943
38
0.705725
0.013706
2.779481
88
0.971998
-0.130057
2.779481
138
1.112474
0.058055
2.779481
39
0.756404
0.074558
3.002401
89
1.013154
-0.101976
3.002401
139
1.164028
0.089707
3.002401
40
0.097082
-0.094209
-0.000000
90
0.275428
0.086656
-0.000000
140
0.504730
-0.090406
0.000000
41
0.151286
-0.069138
0.222358
91
0.331544
0.103800
0.222358
141
0.558912
-0.079290
0.222358
42
0.234990
-0.081613
0.555896
92
0.413092
0.095510
0.555896
142
0.642638
-0.083443
0.555896
43
0.345289
-0.079181
1.000613
93
0.523250
0.107124
1.000613
143
0.752937
-0.081533
1.000613
44
0.429045
-0.089424
1.334151
94
0.604976
0.096254
1.334151
144
0.836657
-0.087332
1.334151
45
0.511127
-0.083366
1.667688
95
0.688137
0.095970
1.667688
145
0.918764
-0.085851
1.667688
46
0.595257
-0.092927
2.001226
96
0.769680
0.088802
2.001226
146
1.002812
-0.091482
2.001226
47
0.705515
-0.093271
2.445943
97
0.879887
0.082116
2.445943
147
1.113068
-0.098235
2.445943
48
0.794045
-0.136124
2.779481
98
0.959943
0.044128
2.779481
148
1.200795
-0.125461
2.779481
49
0.834818
-0.098847
3.002401
99
1.011169
0.086042
3.002401
149
1.242442
-0.098242
3.002401
171
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.504730
0.090300
0.000000
200
0.734033
-0.085409
-0.000000
250
0.912379
0.087876
0.000000
151
0.560833
0.100462
0.222358
201
0.788201
-0.083553
0.222358
251
0.968459
0.086396
0.222358
152
0.642394
0.101964
0.555896
202
0.871940
-0.088814
0.555896
252
1.050042
0.083919
0.555896
153
0.752552
0.120576
1.000613
203
0.982238
-0.089053
1.000613
253
1.160199
0.077851
1.000613
154
0.834259
0.111452
1.334151
204
1.065939
-0.094257
1.334151
254
1.241871
0.076668
1.334151
155
0.917433
0.109125
1.667688
205
1.148060
-0.092445
1.667688
255
1.325070
0.082146
1.667688
156
0.998930
0.100030
2.001226
206
1.232062
-0.096283
2.001226
256
1.406485
0.080233
2.001226
157
1.109136
0.092939
2.445943
207
1.342316
-0.094479
2.445943
257
1.516688
0.085976
2.445943
158
1.188740
0.060853
2.779481
208
1.429591
-0.108498
2.779481
258
1.595490
0.084761
2.779481
159
1.240458
0.089528
3.002401
209
1.471731
-0.095074
3.002401
259
1.648082
0.082890
3.002401
160
0.581164
-0.086889
-0.000000
210
0.734033
0.091702
0.000000
260
0.988813
-0.091673
0.000000
161
0.635342
-0.080406
0.222358
211
0.790122
0.087111
0.222358
261
1.042967
-0.094136
0.222358
162
0.719072
-0.085419
0.555896
212
0.871696
0.087005
0.555896
262
1.126720
-0.100586
0.555896
163
0.829370
-0.083535
1.000613
213
0.981853
0.096500
1.000613
263
1.237018
-0.101832
1.000613
164
0.913085
-0.089083
1.334151
214
1.063541
0.094212
1.334151
264
1.320697
-0.103265
1.334151
165
0.995196
-0.087049
1.667688
215
1.146729
0.093133
1.667688
265
1.402834
-0.102318
1.667688
166
1.079229
-0.092164
2.001226
216
1.228180
0.090648
2.001226
266
1.486784
-0.098143
2.001226
167
1.189484
-0.097451
2.445943
217
1.338384
0.090830
2.445943
267
1.597037
-0.095025
2.445943
168
1.277060
-0.119903
2.779481
218
1.417537
0.070817
2.779481
268
1.683810
-0.086829
2.779481
169
1.318872
-0.097032
3.002401
219
1.469746
0.089171
3.002401
269
1.726496
-0.102022
3.002401
170
0.581164
0.090583
-0.000000
220
0.810467
-0.086987
0.000000
270
0.988813
0.092087
-0.000000
171
0.637263
0.095265
0.222358
221
0.864631
-0.086591
0.222358
271
1.044888
0.085972
0.222358
172
0.718828
0.096208
0.555896
222
0.948374
-0.091265
0.555896
272
1.126476
0.084937
0.555896
173
0.828986
0.114773
1.000613
223
1.058672
-0.091571
1.000613
273
1.236633
0.078045
1.000613
174
0.910686
0.107323
1.334151
224
1.142367
-0.096763
1.334151
274
1.318299
0.080747
1.334151
175
0.993865
0.104999
1.667688
225
1.224492
-0.095233
1.667688
275
1.401502
0.079498
1.667688
176
1.075346
0.096575
2.001226
226
1.308479
-0.098641
2.001226
276
1.482902
0.075595
2.001226
177
1.185552
0.092497
2.445943
227
1.418733
-0.093287
2.445943
277
1.593105
0.090382
2.445943
178
1.265006
0.063117
2.779481
228
1.505857
-0.101658
2.779481
278
1.671755
0.095539
2.779481
179
1.316887
0.088728
3.002401
229
1.548160
-0.094857
3.002401
279
1.724511
0.080838
3.002401
180
0.657598
-0.085250
0.000000
230
0.810467
0.091751
-0.000000
280
1.039769
-0.088288
-0.000000
181
0.711771
-0.081694
0.222358
231
0.866552
0.085780
0.222358
281
1.093920
-0.093039
0.222358
182
0.795506
-0.086885
0.555896
232
0.948130
0.085089
0.555896
282
1.177676
-0.087442
0.555896
183
0.905804
-0.086228
1.000613
233
1.058287
0.087894
1.000613
283
1.287974
-0.093629
1.000613
184
0.989512
-0.091010
1.334151
234
1.139968
0.087870
1.334151
284
1.371649
-0.086158
1.334151
185
1.071628
-0.089474
1.667688
235
1.223161
0.088243
1.667688
285
1.453788
-0.092061
1.667688
186
1.155645
-0.093829
2.001226
236
1.304596
0.088245
2.001226
286
1.537728
-0.090301
2.001226
187
1.265900
-0.096184
2.445943
237
1.414800
0.089526
2.445943
287
1.647981
-0.091576
2.445943
188
1.353326
-0.114115
2.779481
238
1.493802
0.075925
2.779481
288
1.734653
-0.082786
2.779481
189
1.395301
-0.096135
3.002401
239
1.546176
0.089046
3.002401
289
1.777449
-0.088654
3.002401
190
0.657598
0.091343
-0.000000
240
0.912379
-0.087317
-0.000000
290
1.039769
0.088288
0.000000
191
0.713693
0.090511
0.222358
241
0.966537
-0.091318
0.222358
291
1.095842
0.092398
0.222358
192
0.795262
0.091279
0.555896
242
1.050286
-0.092856
0.555896
292
1.177432
0.087523
0.555896
193
0.905419
0.106383
1.000613
243
1.160584
-0.093368
1.000613
293
1.287589
0.093757
1.000613
194
0.987113
0.101598
1.334151
244
1.244270
-0.095407
1.334151
294
1.369250
0.086958
1.334151
195
1.070297
0.099136
1.667688
245
1.326402
-0.099266
1.667688
295
1.452457
0.092505
1.667688
196
1.151763
0.093489
2.001226
246
1.410367
-0.097746
2.001226
296
1.533846
0.091595
2.001226
197
1.261968
0.092027
2.445943
247
1.520621
-0.091337
2.445943
297
1.644049
0.092887
2.445943
198
1.341271
0.066347
2.779481
248
1.607544
-0.090679
2.779481
298
1.722599
0.086806
2.779481
199
1.393317
0.088892
3.002401
249
1.650066
-0.095550
3.002401
299
1.775464
0.089316
3.002401
172
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.031269
-0.095548
-0.000000
CP
50
0.100160
0.085002
-0.000000
100
0.363018
-0.090367
0.000000
0.001239
-0.132689
0.215606
51
0.126373
0.045883
0.215606
101
0.395309
-0.109996
0.215606
0.045873
-0.145272
0.539016
52
0.168973
0.032026
0.539016
102
0.439782
-0.117142
0.539016
0.106225
-0.168888
0.970229
53
0.225785
0.012534
0.970229
103
0.499753
-0.127027
0.970229
0.148479
-0.160433
1.293638
54
0.269223
0.019376
1.293638
104
0.542149
-0.123692
1.293638
0.192330
-0.166136
1.617047
55
0.312270
0.014943
1.617047
105
0.585906
-0.125880
1.617047
0.235112
-0.165981
1.940457
56
0.355581
0.015693
1.940457
106
0.628751
-0.121726
1.940457
0.290004
-0.144093
2.371670
57
0.414099
0.036083
2.371670
107
0.683997
-0.110890
2.371670
0.332959
-0.139717
2.695079
58
0.457416
0.039989
2.695079
108
0.726980
-0.113524
2.695079
0.357288
-0.129396
2.913105
59
0.482679
0.057474
2.913105
109
0.751375
-0.113791
2.913105
10
-0.031269
0.095548
-0.000000
60
0.205303
-0.091541
0.000000
110
0.363018
0.081346
0.000000
11
-0.004984
0.061442
0.215606
61
0.237681
-0.116606
0.215606
111
0.389086
0.055723
0.215606
12
0.037670
0.051352
0.539016
62
0.282219
-0.121201
0.539016
112
0.431579
0.045778
0.539016
13
0.094608
0.035890
0.970229
63
0.342342
-0.132608
0.970229
113
0.488137
0.030798
0.970229
14
0.138000
0.040454
1.293638
64
0.384681
-0.128974
1.293638
114
0.531670
0.037519
1.293638
15
0.181078
0.037305
1.617047
65
0.428475
-0.131573
1.617047
115
0.574653
0.034631
1.617047
16
0.224368
0.042974
1.940457
66
0.471295
-0.126762
1.940457
116
0.618007
0.035443
1.940457
17
0.282768
0.061248
2.371670
67
0.526399
-0.113331
2.371670
117
0.676761
0.053548
2.371670
18
0.326075
0.060907
2.695079
68
0.569371
-0.114273
2.695079
118
0.720096
0.064461
2.695079
19
0.351316
0.061026
2.913105
69
0.593740
-0.116844
2.913105
119
0.745403
0.071235
2.913105
20
0.021303
-0.091306
0.000000
70
0.205303
0.080413
0.000000
120
0.441876
-0.090628
-0.000000
21
0.053782
-0.123995
0.215606
71
0.231458
0.047789
0.215606
121
0.474123
-0.109461
0.215606
22
0.098395
-0.132152
0.539016
72
0.274016
0.033835
0.539016
122
0.518564
-0.116148
0.539016
23
0.158695
-0.149253
0.970229
73
0.330726
0.016613
0.970229
123
0.578459
-0.126133
0.970229
24
0.200968
-0.144602
1.293638
74
0.374202
0.023649
1.293638
124
0.620883
-0.122568
1.293638
25
0.244807
-0.149235
1.617047
75
0.417223
0.019977
1.617047
125
0.664621
-0.124456
1.617047
26
0.287597
-0.143618
1.940457
76
0.460551
0.020580
1.940457
126
0.707478
-0.121004
1.940457
27
0.342536
-0.128354
2.371670
77
0.519164
0.039938
2.371670
127
0.762795
-0.111152
2.371670
28
0.385495
-0.126068
2.695079
78
0.562488
0.048517
2.695079
128
0.805784
-0.113247
2.695079
29
0.409833
-0.121145
2.913105
79
0.587768
0.063186
2.913105
129
0.830192
-0.112331
2.913105
30
0.021303
0.090999
0.000000
80
0.284161
-0.090953
-0.000000
130
0.441876
0.084079
-0.000000
31
0.047559
0.051991
0.215606
81
0.316495
-0.112555
0.215606
131
0.467900
0.061641
0.215606
32
0.090191
0.037863
0.539016
82
0.361001
-0.118797
0.539016
132
0.510361
0.053931
0.539016
33
0.147079
0.017088
0.970229
83
0.421048
-0.129125
0.970229
133
0.566843
0.040800
0.970229
34
0.190489
0.022554
1.293638
84
0.463415
-0.125826
1.293638
134
0.610404
0.046582
1.293638
35
0.233555
0.018391
1.617047
85
0.507191
-0.128127
1.617047
135
0.653369
0.044263
1.617047
36
0.276854
0.018262
1.940457
86
0.550023
-0.123571
1.940457
136
0.696735
0.045624
1.940457
37
0.335300
0.037937
2.371670
87
0.605198
-0.111432
2.371670
137
0.755559
0.062857
2.371670
38
0.378611
0.041106
2.695079
88
0.648176
-0.113674
2.695079
138
0.798900
0.071293
2.695079
39
0.403861
0.050925
2.913105
89
0.672557
-0.115060
2.913105
139
0.824221
0.075090
2.913105
40
0.100160
-0.092114
-0.000000
90
0.284161
0.079892
-0.000000
140
0.520733
-0.091949
0.000000
41
0.132596
-0.121692
0.215606
91
0.310272
0.051094
0.215606
141
0.552936
-0.110301
0.215606
42
0.177176
-0.126047
0.539016
92
0.352797
0.038800
0.539016
142
0.597346
-0.116216
0.539016
43
0.237401
-0.139533
0.970229
93
0.409431
0.022638
0.970229
143
0.657165
-0.126269
0.970229
44
0.279702
-0.135357
1.293638
94
0.452936
0.029762
1.293638
144
0.699617
-0.122591
1.293638
45
0.323522
-0.138334
1.617047
95
0.495938
0.026412
1.617047
145
0.743336
-0.123796
1.617047
46
0.366325
-0.133559
1.940457
96
0.539279
0.027084
1.940457
146
0.786206
-0.121905
1.940457
47
0.421335
-0.118097
2.371670
97
0.597962
0.045724
2.371670
147
0.841594
-0.111215
2.371670
48
0.464299
-0.117037
2.695079
98
0.641292
0.056770
2.695079
148
0.884588
-0.111833
2.695079
49
0.488650
-0.118804
2.913105
99
0.666586
0.067268
2.913105
149
0.909009
-0.110508
2.913105
0.85, part 1
173
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.520733
0.087845
0.000000
200
0.757305
-0.097138
-0.000000
250
0.941306
0.105271
0.000000
151
0.546714
0.069274
0.215606
201
0.789378
-0.107347
0.215606
251
0.967055
0.108694
0.215606
152
0.589143
0.063441
0.539016
202
0.833692
-0.110955
0.539016
252
1.009313
0.106953
0.539016
153
0.645549
0.052893
0.970229
203
0.893282
-0.116423
0.970229
253
1.065313
0.107202
0.970229
154
0.689138
0.057022
1.293638
204
0.935819
-0.111004
1.293638
254
1.109053
0.107178
1.293638
155
0.732084
0.055315
1.617047
205
0.979481
-0.111628
1.617047
255
1.151898
0.107141
1.617047
156
0.775462
0.057640
1.940457
206
1.022389
-0.107955
1.940457
256
1.195343
0.102678
1.940457
157
0.834358
0.072724
2.371670
207
1.077990
-0.103024
2.371670
257
1.254617
0.095620
2.371670
158
0.877705
0.077328
2.695079
208
1.121001
-0.104636
2.695079
258
1.297994
0.092849
2.695079
159
0.903038
0.077227
2.913105
209
1.145462
-0.102970
2.913105
259
1.323397
0.086626
2.913105
160
0.599590
-0.093702
-0.000000
210
0.757305
0.102536
0.000000
260
1.020163
-0.079781
0.000000
161
0.631750
-0.111256
0.215606
211
0.783155
0.096955
0.215606
261
1.052091
-0.078013
0.215606
162
0.676128
-0.116964
0.539016
212
0.825488
0.097189
0.539016
262
1.096298
-0.075365
0.539016
163
0.735871
-0.126502
0.970229
213
0.881666
0.093085
0.970229
263
1.155635
-0.074001
0.970229
164
0.778351
-0.122945
1.293638
214
0.925340
0.092143
1.293638
264
1.198266
-0.079636
1.293638
165
0.822051
-0.123094
1.617047
215
0.968229
0.091410
1.617047
265
1.241865
-0.085684
1.617047
166
0.864934
-0.122175
1.940457
216
1.011645
0.096133
1.940457
266
1.284815
-0.091783
1.940457
167
0.920392
-0.108564
2.371670
217
1.070754
0.095211
2.371670
267
1.340652
-0.093058
2.371670
168
0.963392
-0.109815
2.695079
218
1.114117
0.086778
2.695079
268
1.383681
-0.090947
2.695079
169
0.987827
-0.108183
2.913105
219
1.139490
0.084693
2.913105
269
1.408186
-0.093033
2.913105
170
0.599590
0.092000
-0.000000
220
0.836163
-0.092625
0.000000
270
1.020163
0.109788
-0.000000
171
0.625528
0.077685
0.215606
221
0.868192
-0.097895
0.215606
271
1.045869
0.113871
0.215606
172
0.667925
0.073945
0.539016
222
0.912473
-0.096322
0.539016
272
1.088095
0.115564
0.539016
173
0.724254
0.066146
0.970229
223
0.971988
-0.099102
0.970229
273
1.144018
0.116095
0.970229
174
0.767872
0.068788
1.293638
224
1.014553
-0.097595
1.293638
274
1.187787
0.111591
1.293638
175
0.810799
0.067440
1.617047
225
1.058196
-0.096780
1.617047
275
1.230613
0.105141
1.617047
176
0.854190
0.071102
1.940457
226
1.101117
-0.097509
1.940457
276
1.274071
0.096950
1.940457
177
0.913157
0.082678
2.371670
227
1.156788
-0.097736
2.371670
277
1.333416
0.095284
2.371670
178
0.956509
0.080896
2.695079
228
1.199805
-0.101530
2.695079
278
1.376798
0.095743
2.695079
179
0.981855
0.079400
2.913105
229
1.224279
-0.101266
2.913105
279
1.402215
0.093314
2.913105
180
0.678448
-0.096319
0.000000
230
0.836163
0.106524
-0.000000
280
1.072735
-0.093837
-0.000000
181
0.710564
-0.111539
0.215606
231
0.861969
0.104769
0.215606
281
1.104634
-0.092759
0.215606
182
0.754910
-0.116827
0.539016
232
0.904270
0.104068
0.539016
282
1.148819
-0.092443
0.539016
183
0.814576
-0.124502
0.970229
233
0.960372
0.101490
0.970229
283
1.208105
-0.091783
0.970229
184
0.857085
-0.119718
1.293638
234
1.004074
0.100239
1.293638
284
1.250755
-0.091873
1.293638
185
0.900766
-0.119429
1.617047
235
1.046944
0.099334
1.617047
285
1.294342
-0.091630
1.617047
186
0.943661
-0.116976
1.940457
236
1.090373
0.103212
1.940457
286
1.337300
-0.091675
1.940457
187
0.999191
-0.106256
2.371670
237
1.149552
0.096488
2.371670
287
1.393184
-0.092336
2.371670
188
1.042196
-0.107060
2.695079
238
1.192921
0.090566
2.695079
288
1.436217
-0.092380
2.695079
189
1.066644
-0.105916
2.913105
239
1.218307
0.086642
2.913105
289
1.460731
-0.092562
2.913105
190
0.678448
0.097351
-0.000000
240
0.941306
-0.081321
-0.000000
290
1.072735
0.093837
0.000000
191
0.704342
0.087645
0.215606
241
0.973277
-0.079379
0.215606
291
1.098411
0.094834
0.215606
192
0.746707
0.086154
0.539016
242
1.017516
-0.076985
0.539016
292
1.140616
0.095179
0.539016
193
0.802960
0.080312
0.970229
243
1.076929
-0.076143
0.970229
293
1.196489
0.095657
0.970229
194
0.846606
0.081181
1.293638
244
1.119532
-0.079157
1.293638
294
1.240276
0.095367
1.293638
195
0.889514
0.079582
1.617047
245
1.163150
-0.081986
1.617047
295
1.283089
0.095383
1.617047
196
0.932918
0.084581
1.940457
246
1.206087
-0.085867
1.940457
296
1.326556
0.095258
1.940457
197
0.991955
0.090471
2.371670
247
1.261853
-0.091040
2.371670
297
1.385948
0.094749
2.371670
198
1.035313
0.083318
2.695079
248
1.304877
-0.093139
2.695079
298
1.429334
0.094676
2.695079
199
1.060673
0.082625
2.913105
249
1.329369
-0.099524
2.913105
299
1.454759
0.094553
2.913105
0.85, part 2
174
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.031508
-0.095538
-0.000000
CP
50
0.100926
0.101049
-0.000000
100
0.365794
-0.092779
0.000000
0.065687
-0.076982
0.213763
51
0.201441
0.106358
0.213763
101
0.462928
-0.078114
0.213763
0.214615
-0.090447
0.534406
52
0.348332
0.084297
0.534406
102
0.611908
-0.083122
0.534406
0.416800
-0.140829
0.961932
53
0.542243
0.037239
0.961932
103
0.813837
-0.112545
0.961932
0.563829
-0.128995
1.282575
54
0.690785
0.048689
1.282575
104
0.960967
-0.109106
1.282575
0.712523
-0.135454
1.603219
55
0.838501
0.044279
1.603219
105
1.109598
-0.113058
1.603219
0.859294
-0.124676
1.923863
56
0.987228
0.052519
1.923863
106
1.256485
-0.104564
1.923863
1.056535
-0.124840
2.351388
57
1.184713
0.054541
2.351388
107
1.453738
-0.103184
2.351388
1.205891
-0.136784
2.672032
58
1.332039
0.040569
2.672032
108
1.602978
-0.116720
2.672032
1.292015
-0.138724
2.888699
59
1.417305
0.053430
2.888699
109
1.689040
-0.121241
2.888699
10
-0.031508
0.095538
-0.000000
60
0.206873
-0.088652
0.000000
110
0.365794
0.084816
0.000000
11
0.069028
0.115225
0.213763
61
0.304032
-0.072036
0.213763
111
0.466268
0.095856
0.213763
12
0.215901
0.105176
0.534406
62
0.452991
-0.078785
0.534406
112
0.613195
0.086938
0.534406
13
0.409897
0.061790
0.961932
63
0.655023
-0.113952
0.961932
113
0.806934
0.055541
0.961932
14
0.558405
0.066898
1.282575
64
0.802112
-0.109814
1.282575
114
0.955544
0.063769
1.282575
15
0.706143
0.062410
1.603219
65
0.950768
-0.114424
1.603219
115
1.103218
0.060709
1.603219
16
0.854831
0.073580
1.923863
66
1.097609
-0.103452
1.923863
116
1.252022
0.069149
1.923863
17
1.052312
0.076493
2.351388
67
1.294857
-0.101174
2.351388
117
1.449515
0.072676
2.351388
18
1.199677
0.065637
2.672032
68
1.444143
-0.112736
2.672032
118
1.596764
0.077185
2.672032
19
1.284963
0.057983
2.888699
69
1.530230
-0.126412
2.888699
119
1.681988
0.078997
2.888699
20
0.021466
-0.087778
0.000000
70
0.206873
0.092732
0.000000
120
0.445255
-0.094403
-0.000000
21
0.118653
-0.067585
0.213763
71
0.307372
0.100348
0.213763
121
0.542376
-0.080822
0.213763
22
0.267587
-0.081249
0.534406
72
0.454277
0.083889
0.534406
122
0.691367
-0.086152
0.534406
23
0.469739
-0.128005
0.961932
73
0.648119
0.041666
0.961932
123
0.893245
-0.112037
0.961932
24
0.616780
-0.119232
1.282575
74
0.796688
0.052456
1.282575
124
1.040395
-0.109482
1.282575
25
0.765466
-0.125412
1.603219
75
0.944388
0.048445
1.603219
125
1.189013
-0.112591
1.603219
26
0.912253
-0.112765
1.923863
76
1.093146
0.056773
1.923863
126
1.335923
-0.105160
1.923863
27
1.109495
-0.115304
2.351388
77
1.290634
0.059938
2.351388
127
1.533179
-0.106050
2.351388
28
1.258836
-0.123822
2.672032
78
1.437929
0.051579
2.672032
128
1.682395
-0.117055
2.672032
29
1.344952
-0.127244
2.888699
79
1.523178
0.066452
2.888699
129
1.768445
-0.118769
2.888699
30
0.021466
0.103361
0.000000
80
0.286334
-0.090547
-0.000000
130
0.445255
0.084289
-0.000000
31
0.121993
0.114138
0.213763
81
0.383480
-0.075094
0.213763
131
0.545716
0.094796
0.213763
32
0.268874
0.087676
0.534406
82
0.532450
-0.080642
0.534406
132
0.692653
0.089554
0.534406
33
0.462835
0.041051
0.961932
83
0.734430
-0.112882
0.961932
133
0.886341
0.064441
0.961932
34
0.611357
0.049602
1.282575
84
0.881539
-0.109223
1.282575
134
1.034971
0.071862
1.282575
35
0.759086
0.046778
1.603219
85
1.030183
-0.113374
1.603219
135
1.182634
0.068417
1.603219
36
0.907790
0.055162
1.923863
86
1.177047
-0.103884
1.923863
136
1.331461
0.077497
1.923863
37
1.105273
0.054966
2.351388
87
1.374297
-0.101542
2.351388
137
1.528956
0.080483
2.351388
38
1.252622
0.041588
2.672032
88
1.523560
-0.114328
2.672032
138
1.676181
0.089906
2.672032
39
1.337900
0.038195
2.888699
89
1.609635
-0.123881
2.888699
139
1.761393
0.080457
2.888699
40
0.100926
-0.088351
-0.000000
90
0.286334
0.087367
-0.000000
140
0.524715
-0.094667
0.000000
41
0.198101
-0.069718
0.213763
91
0.386820
0.097382
0.213763
141
0.621824
-0.083652
0.213763
42
0.347046
-0.077333
0.534406
92
0.533736
0.085187
0.534406
142
0.770825
-0.089230
0.534406
43
0.549146
-0.116799
0.961932
93
0.727527
0.047625
0.961932
143
0.972652
-0.110975
0.961932
44
0.696208
-0.112085
1.282575
94
0.876116
0.057192
1.282575
144
1.119823
-0.107992
1.282575
45
0.844881
-0.117166
1.603219
95
1.023803
0.054071
1.603219
145
1.268428
-0.111480
1.603219
46
0.991691
-0.104768
1.923863
96
1.172584
0.062382
1.923863
146
1.415362
-0.104182
1.923863
47
1.188936
-0.103490
2.351388
97
1.370075
0.065567
2.351388
147
1.612619
-0.108882
2.351388
48
1.338253
-0.112842
2.672032
98
1.517346
0.064383
2.672032
148
1.761813
-0.113234
2.672032
49
1.424357
-0.128725
2.888699
99
1.602583
0.074027
2.888699
149
1.847851
-0.116361
2.888699
175
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.524715
0.085479
0.000000
200
0.763097
-0.091958
-0.000000
250
0.948504
0.102543
0.000000
151
0.625164
0.094727
0.213763
201
0.860168
-0.088910
0.213763
251
1.048887
0.100880
0.213763
152
0.772112
0.092204
0.534406
202
1.009201
-0.091708
0.534406
252
1.195891
0.098855
0.534406
153
0.965749
0.073718
0.961932
203
1.210874
-0.100408
0.961932
253
1.389254
0.097058
0.961932
154
1.114399
0.080551
1.282575
204
1.358106
-0.097746
1.282575
254
1.538013
0.093442
1.282575
155
1.262049
0.077224
1.603219
205
1.506674
-0.100299
1.603219
255
1.685596
0.089417
1.603219
156
1.410899
0.085209
1.923863
206
1.653676
-0.100068
1.923863
256
1.834570
0.085207
1.923863
157
1.608397
0.091594
2.351388
207
1.850941
-0.099816
2.351388
257
2.032080
0.085149
2.351388
158
1.755599
0.097473
2.672032
208
2.000065
-0.105024
2.672032
258
2.179158
0.088484
2.672032
159
1.840798
0.077449
2.888699
209
2.086066
-0.105481
2.888699
259
2.264292
0.088076
2.888699
160
0.604176
-0.093075
-0.000000
210
0.763097
0.094242
0.000000
260
1.027965
-0.083031
0.000000
161
0.701272
-0.085757
0.213763
211
0.863508
0.098413
0.213763
261
1.124995
-0.092652
0.213763
162
0.850284
-0.090468
0.534406
212
1.010488
0.102216
0.534406
262
1.274063
-0.095927
0.534406
163
1.052059
-0.108631
0.961932
213
1.203971
0.097006
0.961932
263
1.475565
-0.098639
0.961932
164
1.199250
-0.104772
1.282575
214
1.352682
0.099467
1.282575
264
1.622865
-0.102594
1.282575
165
1.347844
-0.107464
1.603219
215
1.500294
0.097544
1.603219
265
1.771391
-0.104648
1.603219
166
1.494800
-0.103891
1.923863
216
1.649214
0.100555
1.923863
266
1.918471
-0.106397
1.923863
167
1.692060
-0.106212
2.351388
217
1.846718
0.099778
2.351388
267
2.115743
-0.102457
2.351388
168
1.841230
-0.110817
2.672032
218
1.993851
0.085417
2.672032
268
2.264790
-0.097392
2.672032
169
1.927256
-0.111944
2.888699
219
2.079014
0.083910
2.888699
269
2.350749
-0.097097
2.888699
170
0.604176
0.087354
-0.000000
220
0.842557
-0.090752
0.000000
270
1.027965
0.107617
-0.000000
171
0.704612
0.095502
0.213763
221
0.939616
-0.092100
0.213763
271
1.128335
0.097466
0.213763
172
0.851570
0.095030
0.534406
222
1.088660
-0.091619
0.534406
272
1.275350
0.092227
0.534406
173
1.045156
0.083076
0.961932
223
1.290281
-0.096094
0.961932
273
1.468662
0.090355
0.961932
174
1.193827
0.088683
1.282575
224
1.437533
-0.093843
1.282575
274
1.617441
0.086800
1.282575
175
1.341464
0.086384
1.603219
225
1.586089
-0.099873
1.603219
275
1.765011
0.084577
1.603219
176
1.490337
0.092102
1.923863
226
1.733115
-0.098474
1.923863
276
1.914008
0.081847
1.923863
177
1.687837
0.103182
2.351388
227
1.930382
-0.100222
2.351388
277
2.111520
0.086713
2.351388
178
1.835016
0.091910
2.672032
228
2.079482
-0.102625
2.672032
278
2.258576
0.090973
2.672032
179
1.920203
0.078947
2.888699
229
2.165471
-0.102981
2.888699
279
2.343697
0.091498
2.888699
180
0.683636
-0.092902
0.000000
230
0.842557
0.099387
-0.000000
280
1.080939
-0.094478
-0.000000
181
0.780720
-0.087070
0.213763
231
0.942956
0.101145
0.213763
281
1.177960
-0.094943
0.213763
182
0.929743
-0.090867
0.534406
232
1.089946
0.104174
0.534406
282
1.327036
-0.094564
0.534406
183
1.131467
-0.104701
0.961932
233
1.283378
0.099050
0.961932
283
1.528503
-0.093105
0.961932
184
1.278678
-0.101407
1.282575
234
1.432110
0.098869
1.282575
284
1.675816
-0.093405
1.282575
185
1.427259
-0.103268
1.603219
235
1.579709
0.096266
1.603219
285
1.824334
-0.093269
1.603219
186
1.574238
-0.103678
1.923863
236
1.728652
0.099016
1.923863
286
1.971430
-0.093575
1.923863
187
1.771501
-0.101330
2.351388
237
1.926159
0.089474
2.351388
287
2.168704
-0.093648
2.351388
188
1.920648
-0.108097
2.672032
238
2.073268
0.086463
2.672032
288
2.317735
-0.093287
2.672032
189
2.006661
-0.108428
2.888699
239
2.158419
0.086643
2.888699
289
2.403686
-0.093111
2.888699
190
0.683636
0.090171
-0.000000
240
0.948504
-0.084760
-0.000000
290
1.080939
0.094478
0.000000
191
0.784060
0.096568
0.213763
241
1.045547
-0.089547
0.213763
291
1.181300
0.093829
0.213763
192
0.931029
0.098668
0.534406
242
1.194605
-0.091126
0.534406
292
1.328322
0.094135
0.534406
193
1.124563
0.090736
0.961932
243
1.396158
-0.090539
0.961932
293
1.521600
0.095407
0.961932
194
1.273254
0.095569
1.282575
244
1.543437
-0.095767
1.282575
294
1.670393
0.095214
1.282575
195
1.420879
0.093820
1.603219
245
1.691976
-0.100548
1.603219
295
1.817954
0.095396
1.603219
196
1.569775
0.098238
1.923863
246
1.839033
-0.103474
1.923863
296
1.966967
0.095063
1.923863
197
1.767278
0.105514
2.351388
247
2.036303
-0.102083
2.351388
297
2.164481
0.095056
2.351388
198
1.914433
0.086552
2.672032
248
2.185372
-0.099421
2.672032
298
2.311521
0.095359
2.672032
199
1.999609
0.082197
2.888699
249
2.271344
-0.100562
2.888699
299
2.396634
0.095463
2.888699
176
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.031043
-0.102065
-0.000000
CP
50
0.099435
0.091523
-0.000000
100
0.360391
-0.094760
0.000000
0.051582
-0.105314
0.217077
51
0.182158
0.078518
0.217077
101
0.443016
-0.089598
0.217077
0.177191
-0.118544
0.542691
52
0.304804
0.054853
0.542691
102
0.568579
-0.097213
0.542691
0.346227
-0.159081
0.976845
53
0.468237
0.023012
0.976845
103
0.737268
-0.110832
0.976845
0.468495
-0.133908
1.302460
54
0.592991
0.040091
1.302460
104
0.859731
-0.102688
1.302460
0.594792
-0.163977
1.628074
55
0.716069
0.024033
1.628074
105
0.985763
-0.114198
1.628074
0.715396
-0.125857
1.953689
56
0.841632
0.046385
1.953689
106
1.106730
-0.096122
1.953689
0.881591
-0.140903
2.387843
57
1.006498
0.046504
2.387843
107
1.272854
-0.098571
2.387843
1.008106
-0.159471
2.713457
58
1.129455
0.027207
2.713457
108
1.399084
-0.114274
2.713457
1.080631
-0.155648
2.932532
59
1.200838
0.046506
2.932532
109
1.471490
-0.130787
2.932532
10
-0.031043
0.102065
-0.000000
60
0.203818
-0.080700
0.000000
110
0.360391
0.078189
0.000000
11
0.051680
0.106443
0.217077
61
0.286442
-0.081452
0.217077
111
0.443114
0.079634
0.217077
12
0.174342
0.085926
0.542691
62
0.412023
-0.090187
0.542691
112
0.565730
0.065879
0.542691
13
0.337890
0.063628
0.976845
63
0.580851
-0.110181
0.976845
113
0.728931
0.042518
0.976845
14
0.462579
0.066175
1.302460
64
0.703237
-0.102678
1.302460
114
0.853816
0.055198
1.302460
15
0.585746
0.060407
1.628074
65
0.829374
-0.116778
1.628074
115
0.976717
0.044282
1.628074
16
0.711188
0.077673
1.953689
66
0.950197
-0.093745
1.953689
116
1.102522
0.063778
1.953689
17
0.876077
0.077771
2.387843
67
1.116349
-0.099476
2.387843
117
1.267340
0.064697
2.387843
18
0.999129
0.056695
2.713457
68
1.242693
-0.111367
2.713457
118
1.390107
0.063580
2.713457
19
1.070552
0.040255
2.932532
69
1.315146
-0.137551
2.932532
119
1.461411
0.076558
2.932532
20
0.021148
-0.070995
0.000000
70
0.203818
0.083323
0.000000
120
0.438678
-0.098920
-0.000000
21
0.103773
-0.080465
0.217077
71
0.286541
0.077843
0.217077
121
0.521303
-0.092681
0.217077
22
0.229376
-0.101303
0.542691
72
0.409175
0.057216
0.542691
122
0.646856
-0.100648
0.542691
23
0.398366
-0.134524
0.976845
73
0.572514
0.027017
0.976845
123
0.815476
-0.113351
0.976845
24
0.520660
-0.117585
1.302460
74
0.697321
0.043693
1.302460
124
0.937979
-0.104968
1.302460
25
0.646921
-0.137045
1.628074
75
0.820328
0.028611
1.628074
125
1.063957
-0.113911
1.628074
26
0.767574
-0.109243
1.953689
76
0.945988
0.050849
1.953689
126
1.184997
-0.099047
1.953689
27
0.933760
-0.120487
2.387843
77
1.110835
0.051345
2.387843
127
1.351106
-0.101786
2.387843
28
1.060236
-0.122349
2.713457
78
1.233716
0.038043
2.713457
128
1.477280
-0.115077
2.713457
29
1.132745
-0.139037
2.932532
79
1.305068
0.061692
2.932532
129
1.549662
-0.128719
2.932532
30
0.021148
0.096983
0.000000
80
0.282105
-0.088106
-0.000000
130
0.438678
0.079166
-0.000000
31
0.103871
0.083531
0.217077
81
0.364729
-0.085725
0.217077
131
0.521401
0.081005
0.217077
32
0.226527
0.059667
0.542691
82
0.490301
-0.093074
0.542691
132
0.644008
0.071452
0.542691
33
0.390029
0.027620
0.976845
83
0.659060
-0.109508
0.976845
133
0.807139
0.052654
0.976845
34
0.514744
0.041864
1.302460
84
0.781484
-0.101874
1.302460
134
0.932063
0.063154
1.302460
35
0.637875
0.027059
1.628074
85
0.907569
-0.114698
1.628074
135
1.054911
0.053790
1.628074
36
0.763366
0.049313
1.953689
86
1.028463
-0.094320
1.953689
136
1.180788
0.072673
1.953689
37
0.928246
0.047145
2.387843
87
1.194601
-0.097613
2.387843
137
1.345592
0.072827
2.387843
38
1.051259
0.026588
2.713457
88
1.320888
-0.113172
2.713457
138
1.468303
0.077580
2.713457
39
1.122667
0.026234
2.932532
89
1.393318
-0.133369
2.932532
139
1.539583
0.078518
2.932532
40
0.099435
-0.071896
-0.000000
90
0.282105
0.079442
-0.000000
140
0.516965
-0.098485
0.000000
41
0.182060
-0.077453
0.217077
91
0.364828
0.078492
0.217077
141
0.599589
-0.096213
0.217077
42
0.307653
-0.087732
0.542691
92
0.487452
0.060970
0.542691
142
0.725134
-0.102855
0.542691
43
0.476574
-0.117010
0.976845
93
0.650722
0.033913
0.976845
143
0.893684
-0.115784
0.976845
44
0.598907
-0.106825
1.302460
94
0.775568
0.048805
1.302460
144
1.016226
-0.107910
1.302460
45
0.725115
-0.123105
1.628074
95
0.898523
0.035703
1.628074
145
1.142151
-0.114767
1.628074
46
0.845841
-0.095611
1.953689
96
1.024255
0.056598
1.953689
146
1.263264
-0.101484
1.953689
47
1.012012
-0.106373
2.387843
97
1.189087
0.057573
2.387843
147
1.429358
-0.105677
2.387843
48
1.138432
-0.112966
2.713457
98
1.311912
0.050233
2.713457
148
1.555475
-0.116125
2.713457
49
1.210917
-0.142882
2.932532
99
1.383239
0.071094
2.932532
149
1.627834
-0.125558
2.932532
177
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.516965
0.081366
0.000000
200
0.751826
-0.095278
-0.000000
250
0.934495
0.109723
0.000000
151
0.599688
0.083188
0.217077
201
0.834450
-0.106002
0.217077
251
1.017218
0.113049
0.217077
152
0.722285
0.077339
0.542691
202
0.959967
-0.107635
0.542691
252
1.139766
0.110442
0.542691
153
0.885347
0.063704
0.976845
203
1.128309
-0.116592
0.976845
253
1.302457
0.102762
0.976845
154
1.010310
0.072431
1.302460
204
1.250968
-0.114439
1.302460
254
1.427629
0.102413
1.302460
155
1.133105
0.064431
1.628074
205
1.376734
-0.118344
1.628074
255
1.550141
0.101515
1.628074
156
1.259055
0.081904
1.953689
206
1.498064
-0.118952
1.953689
256
1.676478
0.097083
1.953689
157
1.423844
0.082818
2.387843
207
1.664116
-0.111097
2.387843
257
1.841191
0.084957
2.387843
158
1.546499
0.091123
2.713457
208
1.790062
-0.115967
2.713457
258
1.963542
0.089477
2.713457
159
1.617755
0.074219
2.932532
209
1.862349
-0.112786
2.932532
259
2.034672
0.082462
2.932532
160
0.595252
-0.097334
-0.000000
210
0.751826
0.094935
0.000000
260
1.012782
-0.082406
0.000000
161
0.677876
-0.099926
0.217077
211
0.834549
0.097015
0.217077
261
1.095406
-0.092220
0.217077
162
0.803412
-0.105143
0.542691
212
0.957118
0.101776
0.542691
262
1.220893
-0.090628
0.542691
163
0.971892
-0.116846
0.976845
213
1.119972
0.097059
0.976845
263
1.389003
-0.096049
0.976845
164
1.094473
-0.110308
1.302460
214
1.245052
0.103362
1.302460
264
1.511792
-0.105102
1.302460
165
1.220346
-0.115261
1.628074
215
1.367688
0.100804
1.628074
265
1.637382
-0.105875
1.628074
166
1.341531
-0.105660
1.953689
216
1.493856
0.109783
1.953689
266
1.758954
-0.106719
1.953689
167
1.507611
-0.108394
2.387843
217
1.658602
0.107423
2.387843
267
1.924957
-0.107475
2.387843
168
1.633671
-0.117926
2.713457
218
1.781086
0.086191
2.713457
268
2.050714
-0.087401
2.713457
169
1.706005
-0.120669
2.932532
219
1.852271
0.082110
2.932532
269
2.122922
-0.092880
2.932532
170
0.595252
0.084767
-0.000000
220
0.830113
-0.095693
0.000000
270
1.012782
0.109557
-0.000000
171
0.677975
0.086587
0.217077
221
0.912737
-0.107884
0.217077
271
1.095505
0.104441
0.217077
172
0.800563
0.084322
0.542691
222
1.038245
-0.107579
0.542691
272
1.218044
0.098867
0.542691
173
0.963555
0.075117
0.976845
223
1.206517
-0.114024
0.976845
273
1.380666
0.090631
0.976845
174
1.088557
0.082840
1.302460
224
1.329215
-0.114464
1.302460
274
1.505876
0.084403
1.302460
175
1.211300
0.076474
1.628074
225
1.454928
-0.120345
1.628074
275
1.628336
0.080891
1.628074
176
1.337322
0.091410
1.953689
226
1.576331
-0.124773
1.953689
276
1.754745
0.068873
1.953689
177
1.502097
0.097236
2.387843
227
1.742368
-0.110429
2.387843
277
1.919443
0.070852
2.387843
178
1.624694
0.093179
2.713457
228
1.868258
-0.111142
2.713457
278
2.041738
0.091882
2.713457
179
1.695927
0.072350
2.932532
229
1.940521
-0.109676
2.932532
279
2.112844
0.087012
2.932532
180
0.673539
-0.096025
0.000000
230
0.830113
0.101858
-0.000000
280
1.064973
-0.093329
-0.000000
181
0.756163
-0.103091
0.217077
231
0.912835
0.104980
0.217077
281
1.147597
-0.093241
0.217077
182
0.881689
-0.106818
0.542691
232
1.035396
0.109987
0.542691
282
1.273078
-0.092724
0.542691
183
1.050100
-0.117136
0.976845
233
1.198180
0.105338
0.976845
283
1.441141
-0.091574
0.976845
184
1.172720
-0.112764
1.302460
234
1.323299
0.108152
1.302460
284
1.563957
-0.091911
1.302460
185
1.298540
-0.116087
1.628074
235
1.445882
0.107120
1.628074
285
1.689511
-0.091249
1.628074
186
1.419797
-0.112139
1.953689
236
1.572123
0.111775
1.953689
286
1.811131
-0.092002
1.953689
187
1.585863
-0.110857
2.387843
237
1.736854
0.102205
2.387843
287
1.977126
-0.091634
2.387843
188
1.711867
-0.116745
2.713457
238
1.859281
0.085728
2.713457
288
2.102845
-0.090563
2.713457
189
1.784177
-0.117424
2.932532
239
1.930443
0.086115
2.932532
289
2.175037
-0.089989
2.932532
190
0.673539
0.089254
-0.000000
240
0.934495
-0.093991
-0.000000
290
1.064973
0.093329
0.000000
191
0.756262
0.090856
0.217077
241
1.017119
-0.103794
0.217077
291
1.147696
0.093208
0.217077
192
0.878841
0.092781
0.542691
242
1.142615
-0.099670
0.542691
292
1.270229
0.093674
0.542691
193
1.041763
0.086226
0.976845
243
1.310794
-0.101060
0.976845
293
1.432804
0.094354
0.976845
194
1.166805
0.093644
1.302460
244
1.433545
-0.108001
1.302460
294
1.558041
0.093883
1.302460
195
1.289494
0.088885
1.628074
245
1.559187
-0.117097
1.628074
295
1.680465
0.094266
1.628074
196
1.415589
0.102642
1.953689
246
1.680687
-0.121109
1.953689
296
1.806923
0.093406
1.953689
197
1.580349
0.106417
2.387843
247
1.846705
-0.106640
2.387843
297
1.971612
0.093473
2.387843
198
1.702890
0.087626
2.713457
248
1.972519
-0.094699
2.713457
298
2.093868
0.093557
2.713457
199
1.774099
0.078252
2.932532
249
2.044750
-0.103616
2.932532
299
2.164958
0.093350
2.932532
178
CP
x (m)
y (m)
z (m)
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
-0.031076
-0.102506
-0.000000
CP
50
0.099543
0.091710
-0.000000
100
0.360783
-0.094771
0.000000
0.052396
-0.101501
0.216856
51
0.183826
0.081616
0.216856
101
0.444252
-0.085794
0.216856
0.180026
-0.116485
0.542141
52
0.307984
0.056327
0.542141
102
0.571845
-0.094788
0.542141
0.352262
-0.163690
0.975854
53
0.473482
0.020048
0.975854
103
0.743639
-0.111901
0.975854
0.475435
-0.131730
1.301138
54
0.600363
0.043485
1.301138
104
0.867115
-0.100091
1.301138
0.604313
-0.168353
1.626422
55
0.724949
0.021697
1.626422
105
0.995629
-0.114279
1.626422
0.726218
-0.126121
1.951707
56
0.852426
0.047868
1.951707
106
1.117969
-0.093154
1.951707
0.895132
-0.144492
2.385420
57
1.019525
0.044893
2.385420
107
1.286777
-0.098649
2.385420
1.023468
-0.162738
2.710704
58
1.144344
0.025042
2.710704
108
1.414810
-0.115983
2.710704
1.097201
-0.159834
2.929621
59
1.216741
0.042723
2.929621
109
1.488394
-0.134351
2.929621
10
-0.031076
0.102506
-0.000000
60
0.204039
-0.079400
0.000000
110
0.360783
0.077958
0.000000
11
0.053208
0.110795
0.216856
61
0.287509
-0.076611
0.216856
111
0.445063
0.082295
0.216856
12
0.177377
0.086163
0.542141
62
0.415117
-0.086881
0.542141
112
0.569197
0.066580
0.542141
13
0.343023
0.057909
0.975854
63
0.587088
-0.113546
0.975854
113
0.734400
0.040134
0.975854
14
0.469803
0.067251
1.301138
64
0.710443
-0.099996
1.301138
114
0.861483
0.057661
1.301138
15
0.594510
0.056115
1.626422
65
0.839103
-0.118912
1.626422
115
0.985826
0.042230
1.626422
16
0.721842
0.076020
1.951707
66
0.961268
-0.092209
1.951707
116
1.113593
0.064726
1.951707
17
0.888976
0.074015
2.385420
67
1.130119
-0.100779
2.385420
117
1.280621
0.064149
2.385420
18
1.013897
0.053152
2.710704
68
1.258273
-0.113283
2.710704
118
1.405238
0.062480
2.710704
19
1.086344
0.035530
2.929621
69
1.331916
-0.142033
2.929621
119
1.477537
0.074587
2.929621
20
0.021171
-0.070099
0.000000
70
0.204039
0.083669
0.000000
120
0.439155
-0.099487
-0.000000
21
0.104643
-0.077845
0.216856
71
0.288321
0.080464
0.216856
121
0.522623
-0.090169
0.216856
22
0.232268
-0.100919
0.542141
72
0.412469
0.058025
0.542141
122
0.650209
-0.099100
0.542141
23
0.404445
-0.138054
0.975854
73
0.577849
0.024304
0.975854
123
0.821914
-0.114272
0.975854
24
0.527659
-0.114669
1.301138
74
0.704811
0.046848
1.301138
124
0.945451
-0.102879
1.301138
25
0.656488
-0.139403
1.626422
75
0.829300
0.026398
1.626422
125
1.073892
-0.114208
1.626422
26
0.778451
-0.108305
1.951707
76
0.956893
0.052130
1.951707
126
1.196319
-0.095831
1.951707
27
0.947351
-0.122824
2.385420
77
1.123963
0.050283
2.385420
127
1.365106
-0.101783
2.385420
28
1.075647
-0.124414
2.710704
78
1.248702
0.036137
2.710704
128
1.493078
-0.116704
2.710704
29
1.149360
-0.142913
2.929621
79
1.321059
0.059229
2.929621
129
1.566632
-0.131778
2.929621
30
0.021171
0.096909
0.000000
80
0.282411
-0.087484
-0.000000
130
0.439155
0.078869
-0.000000
31
0.105455
0.086434
0.216856
81
0.365880
-0.081035
0.216856
131
0.523434
0.083551
0.216856
32
0.229620
0.061172
0.542141
82
0.493481
-0.090304
0.542141
132
0.647561
0.072196
0.542141
33
0.395206
0.025074
0.975854
83
0.665363
-0.111262
0.975854
133
0.812675
0.050714
0.975854
34
0.522027
0.044687
1.301138
84
0.788779
-0.099105
1.301138
134
0.939819
0.065415
1.301138
35
0.646685
0.025227
1.626422
85
0.917366
-0.115666
1.626422
135
1.064089
0.051951
1.626422
36
0.774075
0.050279
1.951707
86
1.039619
-0.091853
1.951707
136
1.191943
0.073575
1.951707
37
0.941195
0.045574
2.385420
87
1.208448
-0.098097
2.385420
137
1.358951
0.072338
2.385420
38
1.066075
0.024122
2.710704
88
1.336541
-0.114966
2.710704
138
1.483507
0.077199
2.710704
39
1.138503
0.022601
2.929621
89
1.410155
-0.137766
2.929621
139
1.555775
0.076775
2.929621
40
0.099543
-0.070941
-0.000000
90
0.282411
0.079546
-0.000000
140
0.517527
-0.099014
0.000000
41
0.183014
-0.073146
0.216856
91
0.366692
0.081054
0.216856
141
0.600994
-0.094126
0.216856
42
0.310632
-0.085920
0.542141
92
0.490833
0.061846
0.542141
142
0.728573
-0.102217
0.542141
43
0.482721
-0.121557
0.975854
93
0.656124
0.031135
0.975854
143
0.900189
-0.116820
0.975854
44
0.605995
-0.104486
1.301138
94
0.783147
0.051618
1.301138
144
1.023787
-0.106440
1.301138
45
0.734752
-0.126306
1.626422
95
0.907563
0.033481
1.626422
145
1.152156
-0.115272
1.626422
46
0.856801
-0.095758
1.951707
96
1.035243
0.057661
1.951707
146
1.274669
-0.099534
1.951707
47
1.025680
-0.108803
2.385420
97
1.202292
0.056769
2.385420
147
1.443435
-0.105944
2.385420
48
1.153915
-0.114633
2.710704
98
1.326970
0.048444
2.710704
148
1.571346
-0.117850
2.710704
49
1.227598
-0.147461
2.929621
99
1.399298
0.069069
2.929621
149
1.644871
-0.128004
2.929621
179
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
CP
x (m)
y (m)
z (m)
150
0.517527
0.080863
0.000000
200
0.752642
-0.095184
-0.000000
250
0.935510
0.109590
0.000000
151
0.601806
0.085148
0.216856
201
0.836107
-0.104924
0.216856
251
1.019785
0.113425
0.216856
152
0.725925
0.078488
0.542141
202
0.963665
-0.107228
0.542141
252
1.143866
0.111192
0.542141
153
0.890950
0.062275
0.975854
203
1.135016
-0.117684
0.975854
253
1.308419
0.104506
0.975854
154
1.018156
0.074344
1.301138
204
1.258795
-0.114740
1.301138
254
1.435948
0.103158
1.301138
155
1.142353
0.062874
1.626422
205
1.386945
-0.120493
1.626422
255
1.559757
0.103725
1.626422
156
1.270294
0.082890
1.951707
206
1.509720
-0.120115
1.951707
256
1.688162
0.098755
1.951707
157
1.437280
0.082669
2.385420
207
1.678422
-0.112083
2.385420
257
1.855035
0.085649
2.385420
158
1.561775
0.091643
2.710704
208
1.806151
-0.116460
2.710704
258
1.979206
0.090013
2.710704
159
1.634014
0.072478
2.929621
209
1.879587
-0.113620
2.929621
259
2.051286
0.082549
2.929621
160
0.595899
-0.097484
-0.000000
210
0.752642
0.094686
0.000000
260
1.013882
-0.084280
0.000000
161
0.679365
-0.098313
0.216856
211
0.836919
0.096349
0.216856
261
1.097344
-0.094118
0.216856
162
0.806937
-0.104427
0.542141
212
0.961017
0.102876
0.542141
262
1.224878
-0.091308
0.542141
163
0.978465
-0.118225
0.975854
213
1.125777
0.097526
0.975854
263
1.395934
-0.095960
0.975854
164
1.102123
-0.109348
1.301138
214
1.253164
0.104654
1.301138
264
1.519916
-0.105006
1.301138
165
1.230419
-0.116599
1.626422
215
1.377142
0.100637
1.626422
265
1.647823
-0.106156
1.626422
166
1.353020
-0.104896
1.951707
216
1.505344
0.110654
1.951707
266
1.770888
-0.107264
1.951707
167
1.521764
-0.109899
2.385420
217
1.672267
0.109652
2.385420
267
1.939519
-0.107508
2.385420
168
1.649615
-0.119117
2.710704
218
1.796580
0.086783
2.710704
268
2.067046
-0.086428
2.710704
169
1.723109
-0.122833
2.929621
219
1.868730
0.081998
2.929621
269
2.140382
-0.092976
2.929621
170
0.595899
0.084265
-0.000000
220
0.831014
-0.094686
0.000000
270
1.013882
0.108197
-0.000000
171
0.680177
0.087597
0.216856
221
0.914478
-0.107667
0.216856
271
1.098156
0.103291
0.216856
172
0.804289
0.085499
0.542141
222
1.042029
-0.108166
0.542141
272
1.222230
0.098926
0.542141
173
0.969226
0.074375
0.975854
223
1.213291
-0.115352
0.975854
273
1.386695
0.091500
0.975854
174
1.096492
0.084535
1.301138
224
1.337131
-0.115767
1.301138
274
1.514284
0.085031
1.301138
175
1.220616
0.075348
1.626422
225
1.465209
-0.122101
1.626422
275
1.638020
0.080836
1.626422
176
1.348644
0.092931
1.951707
226
1.588071
-0.126380
1.951707
276
1.766512
0.067991
1.951707
177
1.515609
0.097497
2.385420
227
1.756752
-0.110615
2.385420
277
1.933364
0.070372
2.385420
178
1.640043
0.094586
2.710704
228
1.884420
-0.110653
2.710704
278
2.057474
0.092415
2.710704
179
1.712252
0.071081
2.929621
229
1.957825
-0.110438
2.929621
279
2.129525
0.086610
2.929621
180
0.674271
-0.096103
0.000000
230
0.831014
0.100799
-0.000000
280
1.066130
-0.093430
-0.000000
181
0.757736
-0.102003
0.216856
231
0.915290
0.104674
0.216856
281
1.149592
-0.093460
0.216856
182
0.885301
-0.106010
0.542141
232
1.039381
0.110812
0.542141
282
1.277121
-0.092754
0.542141
183
1.056740
-0.118465
0.975854
233
1.204052
0.106124
0.975854
283
1.448117
-0.091372
0.975854
184
1.180459
-0.112189
1.301138
234
1.331500
0.109745
1.301138
284
1.572140
-0.091942
1.301138
185
1.308682
-0.118116
1.626422
235
1.455406
0.107491
1.626422
285
1.699998
-0.091084
1.626422
186
1.431370
-0.112497
1.951707
236
1.583695
0.113168
1.951707
286
1.823121
-0.091921
1.951707
187
1.600093
-0.112300
2.385420
237
1.750596
0.103759
2.385420
287
1.991739
-0.091499
2.385420
188
1.727883
-0.117850
2.710704
238
1.874848
0.086581
2.710704
288
2.119225
-0.090426
2.710704
189
1.801348
-0.118908
2.929621
239
1.946968
0.086005
2.929621
289
2.192541
-0.089938
2.929621
190
0.674271
0.089020
-0.000000
240
0.935510
-0.095023
-0.000000
290
1.066130
0.093430
0.000000
191
0.758548
0.091078
0.216856
241
1.018973
-0.106734
0.216856
291
1.150404
0.093189
0.216856
192
0.882653
0.093946
0.542141
242
1.146514
-0.102339
0.542141
292
1.274473
0.093637
0.542141
193
1.047501
0.086232
0.975854
243
1.317658
-0.102765
0.975854
293
1.438878
0.094453
0.975854
194
1.174828
0.095308
1.301138
244
1.441579
-0.109845
1.301138
294
1.566508
0.093820
1.301138
195
1.298879
0.088340
1.626422
245
1.569560
-0.117846
1.626422
295
1.690195
0.094353
1.626422
196
1.426994
0.103853
1.951707
246
1.692538
-0.122763
1.951707
296
1.818745
0.093380
1.951707
197
1.593938
0.107394
2.385420
247
1.861190
-0.107073
2.385420
297
1.985583
0.093552
2.385420
198
1.718311
0.089238
2.710704
248
1.988777
-0.094441
2.710704
298
2.109653
0.093618
2.710704
199
1.790491
0.077390
2.929621
249
2.062143
-0.103845
2.929621
299
2.181684
0.093559
2.929621

Mader Charles A 201206 PHD Thesis

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

Mader Charles A 201206 PHD Thesis

Uploaded by

Copyright:

Available Formats

Stability-Constrained Aerodynamic Shape Optimization with

Applications to Flying Wings

Charles Alexander Mader

A thesis submitted in conformity with the requirements

c 2012 by Charles Alexander Mader

Charles Alexander Mader

The Case for Unconventional Configurations . . . . . . . . . . . . . . . . . . . . .

The Need for Stability Constraints in High-Fidelity Optimization . . . . . . . . .

Overview of the Stability Constraint Approach . . . . . . . . . . . . . . . . . . .

2 Linear Flight Dynamics

Decoupled Longitudinal Equations . . . . . . . . . . . . . . . . . . . . . . . . . .

3 CFD-Based Stability Derivatives

Sensitivity Analysis Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

ADjoint Stability Derivative Method . . . . . . . . . . . . . . . . . . . . . . . . . 16

Verification of the ADjoint for Stability Derivatives . . . . . . . . . . . . . 25

Time-Spectral Stability Derivative Method

Linearized Aerodynamic Forces . . . . . . . . . . . . . . . . . . . . . . . . 31

Verification and Validation . . . . . . . . . . . . . . . . . . . . . . . . . . 36

4 Time-Spectral ADjoint Method

Solution of the Adjoint System . . . . . . . . . . . . . . . . . . . . . . . . 56

Center of Gravity Calculation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64

Moment of Inertia Calculations . . . . . . . . . . . . . . . . . . . . . . . . . . . . 65

Root Bending Moment Constraint . . . . . . . . . . . . . . . . . . . . . . . . . . 68

Static Longitudinal Stability Definition . . . . . . . . . . . . . . . . . . . 70

Control Anticipation Parameter . . . . . . . . . . . . . . . . . . . . . . . . 72

6 Design Optimization Results

Multidisciplinary Optimization Overview

Optimization Study Outline . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

Additional Constraints for Shape Variables . . . . . . . . . . . . . . . . . 84

Optimization Problem Statements and Qualitative Results . . . . . . . . . . . . . 85

Static Stability Constrained Problems . . . . . . . . . . . . . . . . . . . . 95

Dynamic Stability Constrained Problems . . . . . . . . . . . . . . . . . . 104

Mach = 0.5 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109

Mach = 0.7 Cases

Mach = 0.85 Cases . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113

A Linear Flight Dynamics

B Results Data Tables

C Relevant Optimization Parameters

D FFD Coordinate Tables

ADjoint stability derivatives for a NACA 0012 airfoil at = 0.0 degrees . . . . . 25

Effect of point location on time-spectral derivative location . . . . . . . . . . . . 36

SACCON test case conditions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

Stability bounds for damping ratio and CAP parameter . . . . . . . . . . . . . . 73

Baseline wing: geometry specifications . . . . . . . . . . . . . . . . . . . . . . . . 77

Primary design variables and their bounds . . . . . . . . . . . . . . . . . . . . . . 81

Compatibility design variables and their bounds . . . . . . . . . . . . . . . . . . . 81

Cp contour and streamline comparison, rotating NACA 0012: M=0.2,=0.00 . . 21

Flowfield streamline comparison, rotating NACA 0012: M=0.2, =0.00 . . . . . 22

Pure angle of attack motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Pure pitching motion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33

Time-spectral solution: motion: Mach = 0.1 . . . . . . . . . . . . . . . . . . 33

Time-spectral solution: CL vs : Mach = 0.1 . . . . . . . . . . . . . . . . . . . . 34

Time spectral solution: CL hysteresis vs :

NACA 0012 meshes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

NACA 0012 mesh convergence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37

3.10 Plunging NACA 0012: time-spectral stability derivative verification . . . . . . . . 39

Single cell stencil . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

Call-graph: current residual computation: steady and time-spectral cases . . . . 51

Call-graph: improved residual computation: time-spectral case . . . . . . . . . . 53

Call-graph: force and moment computation . . . . . . . . . . . . . . . . . . . . . 54

Steady case sparsity pattern: 24 block H-H mesh . . . . . . . . . . . . . . . . . . 56

Time-spectral sparsity pattern: 24 block H-H mesh: 3 time instances . . . . . . . 57

ONERA M6 test case . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59