Dynamics Re Entry

Download as pdf or txt
Download as pdf or txt
You are on page 1of 176

Reentry Dynamics and Handling Qualities of a

Generic Hypersonic Vehicle


by

John Jun Araki


Submitted to the Department of Aeronautics and Astronautics
in partial fulfillment of the requirements for the degree of
Master of Science in Aeronautics and Astronautics
at the
MASSACHUSETTS INSTITUTE OF TECHNOLOGY
February 1992

@ Massachusetts Institute of Technology 1992. All rights reserved.

Author ...........
f Aeronautics and Astronautics
January 24,1992

Departm/

Certified by...................
Professor Rudrapatna V. Ramnath
Adjunct Professor of Aeronautics and Astronautics
Thesis Supervisor

A ccepted by .........

....

..

..... ............
.........
Professor Harold Y. Wachman

Chairman, Departmental Committee on Graduate Students


fA(i~J.~5~HUSES ~VINSTiTU
'
OF TI';.%,
:76Y

FEU3 2 0 1992
LIBRA~Lc8

0a~

Reentry Dynamics and Handling Qualities of a Generic


Hypersonic Vehicle
by
John Jun Araki
Submitted to the Department of Aeronautics and Astronautics
on January 24,1992, in partial fulfillment of the
requirements for the degree of
Master of Science in Aeronautics and Astronautics

Abstract
The reentry dynamics and handling qualities of the Generic Hypersonic Aerodynamic Model Example(GHAME) vehicle were studied employing Generalized Multiple Scales theory. The reentry dynamics were examined for an optimal trajectory
designed for the Space Shuttle Orbiter 049 vehicle. Both longitudinal and lateral
directional motions were modeled as time-varying linear differential equations. Generalized Multiple Scales solutions to vehicle reentry dynamics compared accurately
with numerical integration approximations. Second order angle-of-attack perturbations behaved as damped oscillations with increasng frequency. Lateral-directional
reentry dynamics were found to be unstable due to instability in the spiral divergence mode. The existence of non-continuous 'turhing' points in the phugoid mode,
prevented attempts to apply asymptotic methods to the fourth order longitudinal
model. Sensitivity analysis showed the angle-of-attack perturbations to be most affected by changes in C,,, during reentry. Also, angle-of-attack perturbation motions
were found to be, in general, most sensitive at approximately 50,000 to 60,000 vehicle
lengths into the trajectory. The lateral-directional Ireentry dynamics were shown to
be most affected by changes in the stability derivatives N, and L,. Of the lateraldirectional modes, spiral divergence motions were most sensitive to variations in the
stability derivatives. Finally, handling qualities of the GHAME vehicle along shuttle
reentry trajectory were determined to be inadequate. Effects of simple variations in
the behavior of characteristic roots on the handling qualities were determined for a
generic second order system.
Thesis Supervisor: Rudrapatna V. Ramnath
Title: Adjunct Professor of Aeronautics and Astronautics

Dedication
Dedicated to my grandparents.

Yoshiaki and Koharu Araki.


Kazunobu and Naragiku Kanda.

Acknowledgments
I would like to express my deepest gratitude to my parents without whom none of
this would have been possible. Their love and support goes beyond my ability to
express through words.
Also my sincerest gratitude to Professor Rudrapatna V. Ramnath for his invaluable
guidance and wisdom through the completion of this work.
Partial support is acknowledged from NASA Dryden Flight Research Facility through
Vimanic Systems and Professor Rudrapatna V. Ramnath.
I hereby assign my copyright of this thesis to Vimanic Systems.

The gang at Baker these two years:


Nuc, Dinger, Lump, Psycho, Danks, Karl. Thanks for letting me play. It isn't that
you win championships, it's with whom you win them.
Jina and Liz. Smile. You deserve it.
Sally. No, I will never get off your bed. You're still a dummy, but you're ok for a
sister.
The W's. Redman, Rithcie, Ollie, Carvin, Mas, Econ, Bronco, JJ, T, Eileraas, and
Edge. The spirit lives. Thanks for being there. Love, in a Weenie way, WOTY '87.
Jennifer L. Going on 10 years now. Do you realize what fraction of my life that is?!
Kill G.F. SFDPH.
The gang at the Greenhouse. The blackened lake! Homeless! Ben and Andy; Thanks
for everything. The bod will never forget your countless hours of instruction in th.e
art of clam digging.
Finally, to the Poor Fella, Bungy, Gato, Asparagus etc...
You have brightened my life in ways I thought impossible. I will never forget you.

Maybe there aren't any such things as good friends or bad friends - maybe
there are just friends. People who stand by you when you're hurt and who
help you feel not so lonely. Maybe they're always worth feeling scared for,

and hoping for, and living for. Maybe worth dying for too, if that's what
has to be. No good friends. No bad friends. Only people who you want,
need to be with; people who build their houses in your heart... -S. King

Contents
1

Introduction
1.1

Background .............

1.2

Approach

...

. . . .

..........

. .. . .

..............................

GHAME Vehicle and Trajectory


2.1

GHAME Vehicle

2.2

Trajectory.................

21

............

21

23

Generalized Multiple Scales Theory

26

3.1

Theory ..................

26

3.2

Second Order GMS Solution ......

28

3.3

Fourth Order GMS Solution ......

30

Second Order Longitudinal Dynamics


4.1

Overview .................

4.2

Equations of Motion

4.3

32

32

32

GMS Solutions to Dynamics ......

34

4.4

Sensitivity Analysis ...........

41

4.5

Stability Analysis and Displays

55

..........

...

5 Fourth Order Longitudinal Dynamics


5.1

Overview .................

5.2

Equations of Motion

5.3

Longitudinal Stability Derivatives . . .

62

..........

62
62
65

5.4

GMS Solutions to Dynamics .......................

6 Fourth Order Lateral-Directional Dynamics

85

6.1

Overview ..................................

6.2

Equations of Motion

6.3

Lateral-Directional Stability Derivatives

6.4

GMS Solutions to Dynamics ..............................

6.5

Sensitivity Analysis .....

85
............

.........
. ..........

85

. . . . .

97

.....................

7 Handling Qualities

88

110
143

7.1

Overview ..................................

.....

7.2

GHAME Vehicle Reentry Handling Qualities . .............

7.3

Handling Qualities of a Generic Second Order System . .......

143
143
.

149

Summary and Conclusions

162

8.1

Summary and Conclusions ........................

162

8.2

Suggestions for Further Work ......................

166

A Stability Derivative Estimation


A.1 Longitudinal Stability Derivatives

168
. . . . . . . ..

A.2 Lateral-Directional Stability Derivatives

...........

. . . . . . .

168

. . .

169

B Fourth Order Lateral-Directional Sensitivity Differentiation

170

List of Figures
2-1

GHAME Vehicle Configuration

2-2

Trajectory Parameters ..........................

3-1

Concept of Extension . .. . . . . . . . . .............

4-1

Variation of Coefficient w, Along Trajectory . .............

36

4-2

Variation of Coefficient w0 Along Trajectory . .............

37

4-3

Characteristic Root Variations Along Trajectory ........

4-4

Sine-like Solutions to Angle-of-Attack Perturbations . .........

39

4-5

Cosine-like Solutions to Angle-of-Attack Perturbations

40

4-6

GMS Slow Solution Sensitivity to CL

4-7

GMS Sine-like Solution Sensitivity to CL,

4-8

GMS Cosine-like Solution Sensitivity to CL, . .......

4-9

GMS Slow Solution Sensitivity to C,,

..

...................

24
25
. . .

. . . .

........

................

38

46

. ............

.
......

47
48

.................

4-10 GMS Sine-like Solution Sensitivity to C,.

27

49

. . . . ........

. . .

50

4-11 GMS Cosine-like Solution Sensitivity to C,,G. . ..............

51

4-12 GMS Slow Solution Sensitivity to C,. ... .

52

4-13 GMS Sine-like Solution Sensitivity to C,

..

.............
..

.......

...

53

4-14 GMS Cosine-like Solution Sensitivity to C,, . . . . . . . . . . . . . .

54

4-15 Stability Parameter vs. Vehicle Lengths Along Trajectory .......

56

4-16 Stability Parameter vs. Elapsed Time Along Trajectory ........

57

4-17 Stability Parameter Display at 30 Seconds into Trajectory

59

......

4-18 Stability Parameter Display at 200 Seconds into Trajectory ......

4-19 Stability Parameter Display at 400 Seconds into Trajectory . . . . . .

60
61

5-1

Drag Angle-of-Attack Derivative vs. Vehicle Lengths Along Trajectory

5-2

Drag Damping Derivative vs. Vehicle Lengths Along Trajectory

5-3

Vertical Damping Derivative vs. Vehicle Lengths Along Trajectory . .

5-4

Lift Velocity Derivative vs. Vehicle Lengths Along Trajectory

5-5

Angle-of Attack Static Stability Derivative vs. Vehicle Lengths Along

. ..

. . ..

67
68

69
70

Trajectory .................................

71

5-6

Speed Stability Derivative vs. Vehicle Lengths Along Trajectory . . .

72

5-7

Pitch Damping Derivative vs. Vehicle Lengths Along Trajectory . . .

73

5-8

Short Period Roots Along Trajectory . .................

77

5-9

Phugoid Mode Roots Along Trajectory . ................

78

5-10 Sine-like GMS Solutions to Short Period Mode . ............

80

5-11 Cosine-like GMS Solutions to Short Period Mode

81

. ..........
. . . .

82

5-13 Cosine-like Numerical and GMS Solutions to Short Period Mode . . .

83

6-1

Side Force Due to Sideslip vs. Vehicle Lengths Along Trajectory . . .

89

6-2

Dihederal Effect Derivative vs. Vehicle Lengths Along Trajectory

. .

90

6-3

Rolling Moment Due to Yaw Rate vs. Vehicle Lengths Along Trajectory 91

6-4

Roll Damping Derivative vs. Vehicle Lengths Along Trajectory . . ..

92

6-5

Directional Stability Derivative vs. Vehicle Lengths Along Trajectory

93

6-6

Yaw Damping Derivative vs. Vehicle Lengths Along Trajectory.

6-7

Yawing Moment Due to Roll Rate vs. Vehicle Lengths Along Trajectory 95

6-8

Lateral-Directional Roots Along Trajectory . . ..............

6-9

Roll Convergence Root vs. Time into Trajectory . ...........

100

6-10 Spiral Divergence Root vs. Time into Trajectory . ...........

101

5-12 Sine-like Numerical and GMS Solutions to Short Period Mode

6-11 GMS Solution to Roll Convergence Dynamics

.......

6-12 GMS Solution to Spiral Divergence Dynamics

. ............
. ..

6-14 GMS Solution to Cosine-like Dutch Roll Dynamics

. .........

94

99

. ..

6-13 GMS Solution to Sine-like Dutch Roll Dynamics ...

. . .

6-15 GMS Solution to Lateral-Directional Reentry Dynamics ........

. . .

104
105

......

106
107
108

6-16 GMS Solution to Lateral-Directional Reentry Dynamics .


6-17 Sensitivity of Roll Convergence Dynamics to Y,

.........

109
115

. . . . . . . . .

116

6-18 Sensitivity of Spiral Divergence Dynamics to Y, .....

.. . . .

6-19 Sensitivity of Sine-like Dutch Roll Dynamics to Y, ............

117

6-20 Sensitivity of Cosine-like Dutch Roll Dynamics to Y,

118

...........

6-21 Sensitivity of Roll Convergence Dynamics to L, .........

. . . 119

6-22 Sensitivity of Spiral Divergence Dynamics to L, . . . . . . . . . . . .

120

6-23 Sensitivity of Sine-like Dutch Roll Dynamics to L,...........

121

6-24 Sensitivity of Cosine-like Dutch Roll Dynamics to L,

..

.........

122

6-25 Sensitivity of Roll Convergence Dynamics to Lp .........

. . . 123

6-26 Sensitivity of Spiral Divergence Dynamics to Lp .........

. . . 124

6-27 Sensitivity of Sine-like Dutch Roll Dynamics to Lp ...........

125

6-28 Sensitivity of Cosine-like Dutch Roll Dynamics to L

. . . . . . . . .

126

. . . . .

127

6-29 Sensitivity of Roll Convergence Dynamics to L,

.........

6-30 Sensitivity of Spiral Divergence Dynamics to L, ............

128

6-31 Sensitivity of Sine-like Dutch Roll Dynamics to L, ............

129

6-32 Sensitivity of Cosine-like Dutch Roll Dynamics to L,

.........

6-33 Sensitivity of Roll Convergence Dynamics to N, .........

130

. . .

131

6-34 Sensitivity of Spiral Divergence Dynamics to N, .............


6-35 Sensitivity of Sine-like Dutch Roll Dynamics to N,

.........

132
. .

133

6-36 Sensitivity of Cosine-like Dutch Roll Dynamics to N, . . . . . . . . .

134

6-37 Sensitivity of Roll Convergence Dynamics to Np ..............

135

6-38 Sensitivity of Spiral Divergence Dynamics to Np .........


6-39 Sensitivity of Sine-like Dutch Roll Dynamics to Np

. . .

136

...........

137

6-40 Sensitivity of Cosine-like Dutch Roll Dynamics to Np .........

138

6-41 Sensitivity of Roll Convergence Dynamics to N, .............

139

6-42 Sensitivity of Spiral Divergence Dynamics to N, .............

140

6-43 Sensitivity of Sine-like Dutch Roll Dynamics to N,

.......

6-44 Sensitivity of Cosine-like Dutch Roll Dynamics to N,

. . . 141

. . . . . . . .

142

7-1

Short Period Natural Frequency vs. Damping Ratio Along Trajectory

147

7-2

Dutch Roll Natural Frequency vs. Damping Ratio Along Trajectory .

148

7-3

Straight Line Root Trajectory

7-4

GMS Solutions to Cases 5,6,7

7-5

GMS Solutions to Cases 1,8,9 .....................

156

7-6

GMS Solutions to Cases 2,3,4 ......................

157

7-7

Straight Line Root Trajectory of Accelerating System .........

158

7-8

GMS Solutions to Cases 5,10 .......................

159

7-9

Non-Straight Line Root Trajectories of Cases 11,12

154

.....................

155

......................

7-10 GMS Solutions to Cases 6,11,12 .....................

. .........

160
161

List of Tables
2.1

GHAME Vehicle Parameters .........................

7.1

Short Period Requirements for Handling Qualities . ..........

145

7.2

Dutch Roll Requirements for Handling Qualities . ...........

145

7.3

Roll Convergence and Spiral Divergence Requirements for Handling


Qualities ......

..............

7.4

Straight Root Trajectory Systems ............

7.5

Miscellaneous Systems ..........................

23

..............

145
......

151
152

Nomenclature
...................................

Vehicle lengths along trajectory

V .........................................................

Speed along trajectory

h ........................................................................

Altitude

R ........................................

Radial distance from center of the earth

g ........................................

Acceleration due to gravity

Mn ................................................................

Mach number

p..................................................

Air mass density

m ....................................................

Vehicle mass

S ................................................................

Reference area

I ,,,,,Iz,Iz ............................................... ....

M oments of inertia

ka,ky,kz .....................................................
b ......................

Radii of curvature

............................................

...............................

........

Reference span

............................

A .....................................................
1..................................

Reference chord
Aspect ratio

.................................

Vehicle length

o............................. Inverse nondimensional pitching moment of inertia


v....................................................

Ratio of moments of inertia

6 ................................................. nondimensional mass of atmosphere


e...................................................
T ........................................................................
D ...................................................................
W .......................................................................
M .................................................... ...................

Span efficiency
Thrust
...... Drag
Weight
Mom ent

L..................................................
L ..............................

Lift in longitudinal equations


Rolling moment in lateral-directional equations

Y ......................................................................

Side force

N ...............................................................

Yawing moment

CD .............................................................

Drag coefficient

CL ................................................................

Lift coefficient

Cm .......................................................

Moment coefficient

Cv ...........................................................
C

Side force coefficient


Rolling moment coefficient

. .. .. .. . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Yawing moment coefficient

CN ....................................................
a0 . . .

. . .. .. ..

. . ..

.........................
-= a - a ......

. . . ......
.
. ... ... ..

. . . . ... . . . . .

...

......................................

Angle-of-attack

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . .

Angle-of-attack perturbation
Flight path angle

7...............................................................

0......................................

Pitch angle

q...................................................

Pitch angle rate

w.....................................................
S..................................

Initial angle-of-attack

Side velocity

..................................

Roll angle

p....................................................

Roll angle rate

r ..............................................................

Yaw angl, rate

Dv ...................................................

Drag damping derivative

DO ...............................................

Drag angle-of attack derivative

Tv .......................................
Lv ......................

Thrust velocity derivative


...........................
o

Lift velocity derivative

La ................................................

Vertical damping derivative

Mvy ........................................
Ma .........................................

Angle-of-attack static stability derivative

M&...... o....................................
Mi ....

Speed stability derivative

Angle-of-attack damping derivative

...............................................

Yv .........................................

Pitch damping derivative

Side force due to sideslip derivative


14

Lv .....................................................
L, ...............................

Dihederal effect derivative

Rolling moment due to yaw rate derivative

L, ........................................

Roll damping derivative

N , .......................................

Directional stability derivative

N, .......................................

Yaw damping derivative

N ..................................
r1 ,T0

. .. . . .. .. .. . .
. . . . . . . ..

. . . . . . .. . . . . . . . .

Yawing moment due to roll rate derivative


... ..

Independent variables after extension

e........................................... Slowness of variations in coefficients


y......................................

Generic variable representing dynamics

P ........................................

Stability parameter

wn, ..................................................

Natural frequency

.................................................................. Dam ping ratio


T......................................

Total characteristic root path time

Subscripts
-a .......................................................... per angle-of-attack
-q ..............

..........

..........................

per pitch angle rate

- 0..................................................................

Initial term

-T ..................................................................
-

-f

-f

......................................................

........................................

.......................................
.......................

. Slowly varying quantity

Rapidly varying quantity

........................

.................

Trim value

Final value

per sideslip angle

-r ............................................................

per yaw angle rate

-p.........................................

per roll angle rate

-sh ........................................................

of short period m ode

-rc ....................................................

of roll convergence mode

-sp ...................................................

of spiral divergence mode

-dr ...........................................................

of dutch roll m ode

Chapter 1
Introduction
1.1

Background

With the emergence of new technologies, there has been recent interest in the development of hypersonic reentry vehicles. Unlike the Space Shuttle, these hypersonic
vehicles would perform a conventional horizontal take-off and achieve a low-Earth
orbit powered by a multimode propulsion system. Several vehicles of this nature such
as the X-30 National Aerospace Plane have been proposed for both commercial and
military purposes. It is clear that this class of hypersonic vehicles represents the next
generation of aeronautical development.
One of the many issues encountered in the development of such hypersonic vehicles is the problem of predicting the dynamics of the aircraft during its reentry into
the Earth's atmosphere. The task of analyzing the reentry dynamics of space vehicles
is one which has already generated considerable research interest. For example, much
work has been done to predict the reentry behavior of ballistic missiles. The equations
of motion describing the dynamics of space vehicles during reentry are in general, nonlinear. An exact analytical solution to these equations has not been discovered, and
many approximate solutions have been proposed. In the past, such approximations
to reentry dynamics have been developed by making restricting assumptions on the
nature of the vehicle and reentry trajectory as well as the forces affecting the motion. For example, after neglecting gravity in favor of aerodynamic forces , Allen[5]

showed that reentry dynamics can be represented in the form of Bessel functions.
Etkin[6] developed approximations to reentry dynamics after limiting the trajectory
to those with small flight path angles. In one particular approach, a unified linear
time-varying differential equation was developed by Vinh and Laitone[4] to describe
longitudinal angle-of-attack perturbation for any reentry trajectory. However, due to
the variable coefficients, solutions could only be obtained by limiting the trajectories
to two special cases. A straight line reentry trajectory reduced the equation to Kummer's equation while angle-of-attack perturbations were shown to be in the form of
Mathieu's functions for a trajectory with shallow flight path angles.[3]
Similar to Vinh and Laitone's equation for angle-of-attack perturbations, in general, it is possible to accurately represent both the longitudinal and lateral-directional
reentry dynamics of a vehicle with time-varying linear differential equations. The Generalized Multiple Scales(GMS) theory[7, 8] developed by Ramnath offers asymptotic
approximations to such equation as well as other complex problems. The longitudinal dynamics of the Space Shuttle were predicted by Ramnath employing GMS
theory[3]. By separating the inherent dynamics of the solution into 'fast' and 'slow'
parts, Ramnath developed approximations to linear time-varying differential equations while offering complete generality. Unlike the solutions of Vinh and Laitone,
only mild restrictions upon the trajectory or the nature of the vehicle are necessary
to predict the reentry dynamics with these asymptotic approximations. As shown
by Ramnath, the only assumption made by the GMS theory is that the coefficients
of the differential equation vary slowly when compared to the time constant of the
reentry motion. During entry into the earth's atmosphere, the main variations in the
coefficients are due to changes in air density, aerodynamic forces, and moments. Experience with ballistic missiles has shown that these parameters vary relatively slowly
when compare to the time constants of the dynamic motions. Aside from generality,
another advantage that the GMS theory presents is that the asymptotic approximations to the reentry dynamics come in the form of simple elementary functions.
This allows these approximations to be employed for further analysis regarding the
stability and sensitivity of reentry dynamics. Such analysis is not possible with the

approximations developed by Vinh, Loh, and Laitone which result in representations


of the dynamics in the form of non-elementary functions such as those of Bessel and
Mathieu. Ramnath's approach is followed in this work and his technique is briefly
presented in Chapter 3.

1.2

Approach

In this work, the reentry dynamics and handling qualities of the Generic Hypersonic
Aerodynamic Model Example(GHAME) vehicle along an optimal Space Shuttle trajectory are studied employing GMS theory. The GHAME is a computer simulation
model designed to provide accurate aerodynamic data for generic vehicles in the hypersonic flight regime. The reentry trajectory along which the the dynamics and
handling qualities of the GHAME vehicle are studied is one which was originally
designed to minimize the weight of the thermal protection system of the Space Shuttle orbiter 049 vehicle. The GHAME vehicle and the optimal shuttle trajectory are
detailed in Chapter 2.
GMS theory is employed to predict both the longitudinal and lateral-directional
reentry motions of the GHAME vehicle. The conceptual foundations of the GMS
method as well as the second and fourth order GMS approximations used to study
the GHAME vehicle dynamics are detailed in Chapter 3. The longitudinal reentry behavior of the aircraft is explored in two different ways. First, the behavior of
angle-of-attack perturbation dynamics is studied in Chapter 4 by applying Ramnath's
solutions to the second order time-varying linear differential equation developed by
Vinh and Laitone. Sensitivity of the angle-of-attack perturbations to several different aerodynamic coefficients is examined through partial differentiation of the GMS
approximation which is justified by Ramnath and Radovsky[16]. Also, second order
longitudinal stability is assessed through a stability criterion derived by Ramnath's
GMS method. A possible manner in which this stability information can be displayed to a flight crew is considered. The full fourth order longitudinal dynamics
of the GHAME vehicle as it travels along the optimal shuttle reentry trajectory are

studied in Chapter 5. A fourth order linear differential equation describing the longitudinal motions is derived, and the GMS approximations are employed once again
in an attempt to predict reentry dynamics.
Similarly, the lateral-directional dynamics of the GHAME vehicle along the Shuttle reentry trajectory are studied in Chapter 6. Again, following Ramnath's theory
on parameter sensitivity of variable systems, a sensitivity analysis is conducted in
order to determine the effects of variations in the stability derivatives on the reentry
dynamics. Finally, the handling qualities of the GHAME vehicle are investigated
in Chapter 7 through comparison of reentry parameters with set of flying quality
specifications. Also, GMS theory is employed to study the relative differences in the
handling qualities of generic time-varying second order systems whose characteristic
roots exhibit different behaviors in time.

Chapter 2
GHAME Vehicle and Trajectory
2.1

GHAME Vehicle

The recent interest in the development of hypersonic vehicles has led to a need for
accurate aerodynamic data in this flight regime. Much of the existing data is not
available to general users. The Generic Hypersonic Aerodynamic Model Example[1]
was developed at the Dryden Flight Research Facility in order to satisfy this demand
for realistic aerodynamic data in hypersonic flight. The data included in the GHAME
provides a model with which it is possible to conduct simulations for the design of
control and guidance systems as well as trajectory optimization. The generic nature
of the GHAME data enables it to be a starting point for working designs of both
commercial and military aircraft.
The aerodynamic data contained in the GHAME model are for a particular generic
vehicle geometry and were developed as a combinaton of existing aircraft and theories. Actual data from vehicles such as the Space Shuttle Orbiter, lifting body type
aircraft, as well as theories such as the modified Newtonian impact flow method were
employed in developing the final GHAME aerodynamic data. In order to maintain the
realistic nature of the data, certain aerodynamic anomalies contained in some of the
sources were retained. Lateral-directional derivatives for Mach numbers above 8 were
taken exclusively from Space Shuttle data. Below this speed, the lateral-directional
derivatives are an equal combination of the Space Shuttle and a swept double delta

configuration. The effect of the tip fins from the swept double delta configuration
was included with the justification that the larger vertical tail of the GHAME vehicle
would have a larger effect on the aerodynamics than the small vertical tail of the
Space Shuttle Orbiter. The data at the higher Mach numbers was also adjusted in
order to insure a smooth transition in the derivatives. The longitudinal aerodynamic
coefficients are an equal mix of various sources at all Mach numbers. The drag coefficients were modified through multiplication factors and biases in order to provide
L/D numbers which are realistic when compared with the Space Shuttle. Both the
lateral-directional and longitudinal sets of data were adjusted for reference span and
reference area.
The GHAME data was developed for a flight regime typical of a single stage-toorbit mission. Such a mission would entail a powered horizontal take-off from conventional runways, and accelerating to orbital velocities with air-breathing engines until
achieving a low-Earth-orbit. Upon completing its mission in orbit, the GHAME vehicle would reenter the Earth's atmosphere and maneuver to an unpowered horizontal
landing.
The sources employed to model the aerodynamic data result in a physical configuration of the GHAME vehicle which is analogous to the X-24B or the X-24C with a
more slender fuselage. The GHAME configuration is a delta wing vehicle with mixed
elevons serving both as ailerons and elevators. Provisions were made in the daba for
a single rudder, however no aerodynamic considerations were given to control jets,
speed brake, landing gear, and other variable physical elements.
Mass properties of the GHAME vehicle were estimated by approximating the
vehicle geometry through the use of simple shapes. The fuselage was modeled as
a cylinder 20 ft. in diameter and 120 ft. in length. This allows enough volume for
the liquid hydrogen fuel. Two 10 degree half cones were attached to the cylinder
to complete the fuselage structure. Both the delta wings and the vertical tail were
modeled as thin triangular plates, with the wings containing no dihedral. The engine
module was wrapped around the bottom surface of the fuselage. The complete vehicle
configuration is shown in Fig.2-1. The reference area of the vehicle is 6000 ft 2 and

Length, 1
Ref. Area, S
Ref. Span, b
Ref. Chord, c
Mass, m
I,,
_
I_
I,,
Iz,

233.4 ft.
6000 ft. 2
80 ft.
75 ft.
120,000 lbs.
.87 x 106 slugs-ft.
14.2 x 106 slugs-ft. 2
14.9 x 106 slugs-ft. 2
.28 x 106 slugs-ft. 2

Table 2.1: GHAME Vehicle Parameters


the reference span is 80 ft. Overall vehicle length is 233.4 ft. while the reference chord
is 70 ft.
The mass properties of the GHAME vehicle which were estimated employing
the simple configuration shown in Fig.2-1 were assumed to be on the same order
of magnitude as current existing supersonic cruise aircraft. Most of the GHAME
vehicle estimates were derived from the XB-70 aircraft. The gross take-off weight
was estimated to be approximately 300,000 lbs, of which 60% was assumed to be
the liquid hydrogen fuel. For the purposes of this study of reentry dynamics, the
GHAME vehicle is considered to be at the fuel burnout mass of 120,000 lbs. Also,
the moments of inertia are taken at fuel burnout values. All of the critical GHAME
vehicle parameters employed in this study are shown in Table 2.1.

2.2

Trajectory

In this study, the dynamics of the Generic Hypersonic Aerodynamic Model Example
vehicle are examined as it traverses a prescribed trajectory returning it into the
Earth's atmosphere. The trajectory employed is one which was originally designed
to minimize the thermal-protection-system(TPS) weight of the Space Shuttle Orbiter
049 vehicle[2]. The TPS of the Shuttle Orbiter consists of a collection of 22 metallic
panels of varying composition and thickness. In order to obtain the optimal trajectory,
the method of steepest descent was applied iteratively to minimize the total heat load

50

Figure 2-1: GHAME Vehicle Configuration

24

__

I.a
12

4.5
4'

It

Figure 2-2: Trajectory Parameters


at the stagnation point. The optimal trajectory produced a minimum TPS weight of
30,700 lbs.
Following Ramnath[3], the optimal Shuttle trajectory is detailed in Fig.2-2 where
angle of attack, velocity, altitude, and flight path angle are all shown as functions
of the non-dimensional variable (. This non-dimensional variable is the number of
vehicle lengths traversed along the trajectory and is discussed further in Chapter 3.
The Space Shuttle reentry trajectory covers a range of 0 to 290,000 vehicle lengths
traversed and a descent from 400,000 ft. to 100,000 ft. The terminal down range
distance is 5400 nautical miles, and the cross track distance is 1100 nautical miles.
The maximum acceleration does not exceed 3 g's. Angle-of-attack values are very
large along the trajectory while flight path angles vary from 00 to -40. It should be
noted that the variation of real time with respect to ( is non-linear as seen in Fig.2-2.
The total time required to fly the optimal trajectory is approximately 1900 seconds.

Chapter 3
Generalized Multiple Scales
Theory
3.1

Theory

This chapter and the techniques contained within closely follow the development of
the Generalized Multiple Scales(GMS) theory by Ramnath[7, 8]. The GMS method is
an asymptotic approach for approximating solutions to a variety of complex systems.
The concept of asymptotic solutions is based on the original work of Poincark and has
been employed to obtain engineering approximations in fields such as mechanics and
astrodynamics. The complete generalization of the multiple scales asymptotic analysis
was achieved by Ramnath in his development of GMS theory, and his approach has
been successfully applied to a large number of problems including the investigation
of the behavior of vehicles such as VTOL aircraft and the Space Shuttle.
One particular application of GMS theory is the approximation of solutions to
a class of linear ordinary differential equations having variable coefficients. While
first order linear equations of such kind are solved in the form of an exponential,
higher order equations cannot be resolved in this way. Often, the exact solutions to
higher order equations only exist in the form of transcendental functions such as those
of Bessel which cannot be expressed as simple analytic functions of the coefficients.
However, it is very often desirable to have approximate solutions in such analytic

Mixed

2ehavior

Extension

slow
Motions

Sast

a " Ih1abAr~
A
Allh~A
ft aA AMotions

Figure 3-1: Concept of Extension


forms which allow for simple manipulations and sensitivity analysis.
As developed by Ramnath, the Generalized Multiple Scales method is based on
the premise that the variable coefficients of the linear differential equations contain
a small parameter.

Often, direct expansion in powers of this parameter leads to

a misrepresentation of the true solution over a particular range of the independent


variable. Such a misrepresentation is called a nonuniformity in the perturbation expansion and is avoided by the GMS approximations through the concept of extension.
The fundamental idea behind the concept of extension is to enlarge the domain of the
independent variable to a space of higher dimension. This is achieved by the introduction of general scale functions which reparameterize the independent variable into

a set of new independent clocks. The scale functions can in general be both complex
and nonlinear. If the original independent variable is time, qualitatively, extension is
similar to having a particular motion be recorded by a set of independent observers
each with time pieces running at different rates. In this way, the general solution
is separated into characteristic motions which occur at different rates or on different scales. The scaling functions are chosen such that the non-uniformities of direct
perturbation theory are eliminated. The complete generalization of this technique
was achieved by Ramnath[7, 8] and has been fruitfully applied to a large number of
complex problems. The concept of extension is illustrated in Fig.3-1.
Generalized Multiple Scales theory employs the concept of extension in order to
provide asymptotic approximations to ordinary differential equations. Extension allows the dynamics to be separated into motions on different scales after which they
are combined to produce an approximation to the full solution.

Mathematically,

applying extension to ordinary differential equations results in sets of partial differential equations with a new independent variables. These equations are solved, and
the approximation is completed by returning the problem to its original independent
variable. Ramnath's Generalized Multiple Scales theory is applied and detailed in
the next two sections.

3.2

Second Order GMS Solution

In this section, the Generalized Multiple Scales theory is demonstrated on a second


order linear differential equation with varying coefficients[3]. The GMS theory approximates the full solution through separation of natural motions.

Consider the

following equation.

d2X

dx

+ wl(t)-

+ wo(t)z = 0

(3.1)

In order to employ the GMS approximation, it is necessary to assume that the


coefficients of the above equation vary slowly. That is to say the variations in the

coefficients are slow in comparison to the time constant of the dynamic motion. Mathematically, this is equivalent to stating that the coefficients of Eq.3.1 vary on a new
slow variable f defined as
S=
=t

(3.2)

The small parameter e in the above equation is a measure of the ratio of time constants
of the solution and coefficient variation. Asymptotic solutions are obtained as e-- 0.
As developed by Ramnath, asymptotic solutions of the form z = z,(ro)zf(Tr) are
sought for Eq.3.1 by the GMS method. The final GMS solution is given by

z = z.(t)z (t)

(3.3)

= I(w(t) 2 - 4wo(t)l-

(3.4)

where the slow solution 2, is

X,(t)

and the fast solution is given by

Xf(t)

COep

k,(t )dt) in (

C2exp tk,(t)dt cos

tk(t)dt +

ki(t)dt

(3.5)

k, and kcare respectively the real and imaginary parts of the characteristic roots of
Eq.3.1. C1 and C2 are arbitrary constants which depend on the initial conditions of
the original differential equation. The above approach was developed by Ramnath in
order to solve such systems of any order. This general theory is applied to a fourth
order system in the next section.

3.3

Fourth Order GMS Solution

The GMS approximation to a fourth order linear differential equation with varying coefficients is detailed in this section.

The full GMS solution is obtained by

first obtaining approximations to the dynamics associated with each of the modes of
motion.[7, 8] Consider the fourth order equation
d4 z

. d3
d2 zx
+ w2(t)-- + wt)
+ w3(t)-

+ wo(t)z = 0

(3.6)

The characteristic roots which describe the solution of the equation are them given
by the fourth order algebraic equation

s4 + W3 s3 +W4s2 +ws +Wo = 0

(3.7)

Since the coefficients of this equation vary with time, it is clear that the characteristic
roots of this system also change with time. Depending on the nature of the coefficients,
the four roots of the algebraic equation are comprised of pairs of complex conjugates
or real roots. A particular mode of motion is represented by either a pair of complex
conjugates roots or a single real root. As shown by Ramnath, the GMS solution
to Eq.3.6 is obtained by first approximating the motion associated with each of the
modes. If a mode is represented by a single real root, k, then the GMS appr1il:.ation
to its characteristic motion is given by

x(t) = emp ('

k(t)dt)

(3.8)

If a particular mode is represented by a complex conjugate pair of roots given by


k(t) = k, + iki then the GMS fast and slow solutions to the characteristic motion are
given by
X(t)= X(t)xz(t)

(3.9)

where the slow solution is

,(t)= ep

2ik(t)

(3.10)

and the fast solution is given by

Xf(t)

Cexp

k(kt)dt) sin

C2exp(

k,(t)dt cos

k1(t)dt) +
ki(t)dt)

(3.11)

C1 and C 2 are arbitrary constants to be determined by initial conditions. The full


GMS solution to the fourth order equation shown in Eq.3.6 is obtained by a linear
combination of the approximations to each of the mode motions. For example, consider a system which contains three modes consisting of a pair of complex roots and
two real roots. If the complex roots are given by k, + iki and the real roots are kI
and k 2 respectively, then the full GMS solution to the system is given by

x(t)

C=exp

irtf kAt)

C2 exp

Csexp (j'

dt) exp

2ik(t) dt) exp

ki(t)dt) + C4 exp (j'

(ft

k((t)dt) sin (j
kr(t)dt) cos

k2(t)dt)

ki(t)dt) +
' k1 (t)dt) +

(3.12)

The constants C 1, C2 , C3 , and C4 are determined by initial conditions of the original


differential equation.

Chapter 4
Second Order Longitudinal
Dynamics
4.1

Overview

The second order longitudinal dynamics of the GHAME vehicle flying along the Space
Shuttle trajectory are studied through the use of second order GMS solutions developed in section 3.1. The GMS solutions to angle-of-attack perturbations during
reentry are compared to numerical approximations. The sensitivity of longitudinal
motions to various aerodynamic coefficients is obtained by the differentiation of the
analytical approximations provided by GMS theory. This is based on Ramnath's
sensitivity theory of variable systems. Second order longitudinal stability is assessed
through a GMS stability criterion developed by Ramnath. One possible way to display such stability information to a flight crew is considered.

4.2

Equations of Motion

The following equations describing the longitudinal motions of an aircraft are developed under the assumption that the vehicle experiences lift, but does not exhibit a
rolling or yawing motion. The coordinate system is such that the x-axis is always
tangential to the instantaneous flight path. Under such conditions, the equations of

motion in the plane of symmetry are described by[4, 3, 10, 11]

fV = -pSCDV'1(2m) - gainV
pSICV

2m=
2

2IC,

- g _3g (I

2R

cosy

-I
I

(4.1)
(4.2)

.sin28

(4.3)

and the kinematic relations


= q + (V/R)cosy

(4.4)

R = Vin7

(4.5)

S=7+a

(4.6)

The dot in the equations above denotes differentiation with respect to time.
Assuming that the slope of the lift curve is approximately independent of flight
speed and Mach number at high supersonic speeds, the aerodynamic coefficients in
Eqs.4.1-4.3 are linearized through a Taylor Series expansion about the nominal trajectory. After eliminating 0 and V from the equations, a change of variable is made. The
independent variable time is replaced by a non-dimensional parameter 4 according to
the relationship

S=

V(t)dt

(4.7)

The new independent variable is the non-dimensional number of vehicle lengths


traversed along the trajectory. This variable transformation leads to the following
general equation for transient angle-of-attack perturbation.
a" + wi(,)a' + wo()a = 0

(4.8)

where
w1i()

(4.9)

+ Cq,)] + V'/V

= 6[CL, - 0(CmC

= -6 (Cm. + gCo.cos) + 8'CLo +

6V'VCL, - 62[CL,(,C,, + Co0)D+ CoC0ce


31 (l)

+ ao)

+
(4.10)

/vcos2(-7

The non-dimensional parameters are defined by


6

pSI
2m

IP, - I,,
I,

ml 2
-PS
I,

(4.11)

The primes in Eqs.4.8-4.10 represent differentiation with respect to the new independent variable . It can be seen from Eqs.4.9 and 4.10 that both the coefficients wl
and w0 are functions of parameters which depend on instantaneous flight conditions,
and can be calculated only if the trajectory is explicitly known. It is clear that Eq.4.8
is a second order linear differential equation containing coefficients which vary with
the independent variable. In its most general form, Eq.4.8 cannot be solved exactly.
In the next section, Ramnath's Generalized Multiple Scales theory is employed to
obtain approximate solutions to Eq 4.8 with only mild restrictions on the vehicle or
trajectory.

4.3

GMS Solutions to Dynamics

Approximate solutions to angle-of-attack perturbations of the GHAME vehicle flying


along the shuttle trajectory are developed by applying Ramnath's GMS solutions
detailed in section 3.1. The GMS solutions are employed under the assumption that
the coefficients of Eq.4.8 vary slowly. During reentry from an altitude of 400,000 ft.,
the variations in the coefficients of Eq.4.8 are primarily caused by the changes in
density, velocity, and moment parameters. These changes are slow when compared

to the time constant of vehicle motion.[3, 8] From the equations developed in section
3.2, the GMS solution to Eq.4.8 is given by

()

= (4wo - w )
+C 2 exp

(j

[ex(ep
,
k,()d)

k,()d) sin(
cos (j tc(e)d)

,ki(0)d)
(4.12)

The coefficients of Eq.4.8 as defined in Eqs.4.9 and 4.10 are calculated along the
Shuttle trajectory employing the aerodynamic data included in the GHAME model.
Since nominal angle-of-attack values of the prescribed trajectory are so large, a leastsquares analysis is required to extrapolate aerodynamic data for the GHAME vehicle
at the flight conditions desired. The values of the coefficients

wl and wo are plotted

along the trajectory in Figs.4-1 and 4-2. Again, the independent variable ( is the
number of vehicle lengths traversed along the trajectory. From the coefficients shown
in Figs.4-1 and 4-2, the characteristic roots of Eq.3.1 are calculated. Variations of
these roots as the GHAME vehicle progresses along the reentry trajectory are shown
in Fig.4-3. Finally, this information is substituted into Eq.4.12 in order to obtain
final GMS approximations to the angle-of-attack perturbation dynamics.
Two sets of solutions are developed for different sets of initial conditions. In order
to obtain a solution of a sine wave nature, the initial conditions are set at &(~o) = 0
and &'(ao) = ki(o). A solution of a cosine wave form is obtained with the initial
conditions &(o) = 1 and &'(o) = 0. For the purposes of comparison, a numerical
approximation to Eq.4.8 is also generated by means of a Runge-Kutta integration
scheme. In addition to the numerical approximation, an exact solution to a 'frozen'
system is obtained for further comparison. The coefficients of Eq.4.8 are assumed
to be constant at their initial reentry values, and a constant coefficient analysis is
employed to generate solutions to the 'frozen' system. The GMS approximations
are plotted for the two sets of initial conditions in Fig.4-4 and 4-5 along with their
respective numerical and 'frozen' counterparts. The fast scale solution of the GMS
approximation is also included.
As seen in Figs.4-4 and 4-5, the angle-of-attack perturbations of the GHAME

0.18
0.16
0.14
0.12
0.1
0.08

0.06
0.04
0.02
0

0.5

1.5

2.5

3
x10 5

Figure 4-1: Variation of Coefficient w, Along Trajectory

xO10- 3

wo

n
0

0.5

1.5

2.5

3
x10 5

Figure 4-2: Variation of Coefficient wo Along Trajectory

n n00

0.02

ml

0.0 15 .................
0.01
0.005
0

-0.005
-0.01 .
-0.015

-0.02 ............
-0.02

-0.015

-0.01

-0.005
k,

Figure 4-3: Clock Function Variations Along Trajectory

0.

-0.

10
x10 4

Figure 4-4: Sine-like Solutions to Angle-of-Attack Perturbations

0.5

-0.5

-1
0

10

x10 4

Figure 4-5: Cosine-like Solutions to Angle-of-Attack Perturbations

vehicle during reentry behave as damped oscillations with increasing frequency. Upon
comparison to the numerical solutions, the GMS fast scale solution predicts these
frequency changes of the angle-of-attack perturbation dynamics quite accurately. In
both the sine and cosine-like cases, the zero-crossings of the fast scale solutions and
the numerical solutions occur virtually at the same instant. When the slow scale
solution is incorporated to form the full GMS approximation, the complete dynamics
are predicted with great accuracy. For both sets of initial conditions, the fast scale
solutions predict the frequency changes yet consistently overshoot the magnitude
variations. When the slow scale solution is included, the magnitude of the dynamics is
also predicted accurately. In the sine-like dynamics, the full GMS solution overshoots
the amplitude of the numerical solution by a negligible margin, and matches the
frequency completely. In the cosine-like case, both the GMS and numerical solutions
exhibit the same amplitude behavior, while the frequency appears to have a phase
shift towards the beginning of the trajectory. This may be due to the proximity of

the clock function roots to the real axis early in the trajectory.
The 'frozen' approximation is one that is often employed in the engineering analysis of slowly varying systems. Despite its use, it can clearly be seen that by freezing
the system at the initial point, the dynamics of the vehicle are totally misrepresented.
In both sets of initial conditions, the frozen approximation becomes quite invalid after
leaarly
approximately half a cycle. The dynamics of angle-of-attack perturbations
better predicted by GMS theory.

4.4

Sensitivity Analysis

In the study of vehicle dynamics, it is often useful to determine how certain physical
parameters of the aircraft can affect its motion. Since the GMS method provides
solutions to vehicle dynamics in simple analytical forms, a sensitivity analysis can be
performed by simple partial differentiation with respect to various physical parameters. Sensitivity of the GHAME vehicle second order angle-of-attack perturbations
to certain aerodynamic coefficients is studied in this manner. The aerodynamic coef-

ficients CL., C,,o,

and C,,,, are chosen due to their importance in the determination

of longitudinal dynamics.
It should be noted that partial differentiation of the GMS angle-of-attack solutions with respect to the aerodynamic coefficients does not provide a true sensitivity
analysis. Partial differentiation with respect to the aerodynamic coefficients assumes
that these parameters are constant. In reality, these aerodynamic coefficients also
vary along the trajectory, and in order to conduct a true sensitivity study, variational
principles should be utilized. However, Ramnath has shown that the partial differentiation is a suitable approximation to variational methods, and vehicle sensitivity
to physical parameters can be studied by treating the aerodynamic coefficients as if
they are constant.
The GMS approximation to angle-of-attack perturbation can be written in the
form
d(O) = a,(s)

(4.13)

af()

where the fast and slow solutions are as shown in Eq.4.12. Partial differentiation with
respect to the aerodynamic coefficient CL, is now carried out. The GMS solution
shown above is not only a function of the variable , but also a function of the the
aerodynamic coefficients in question. Thus, differentiating Eq.4.13 with respect to
CL, gives
-

OCL

S Ca,
ao CL,,

ifa
a

OCL,

(4.14)

It now must be determined how the fast and slow GMS solutions each vary with
respect to changes in CL,. From Eq.4.12, the fast solution, af, can be written in the
form

af = CieAsin(B) +C2eAcos(B)

(4.15)

where A and B are defined by


A = fk,( )d,= -f
0o

wi(,)d(
o

(4.16)

B =

k(()d

- w ()2

[4wo()

d(

(4.17)

Without considering the dependence of initial or boundry conditions on the parameter, differentiating Eq.4.15 with respect CL, leads to the equation

(a dA

OCL,

SC

DCL,

C2 OB
J \Ae i(B)+ C eA+
oB

CL.

L.

DB
Cco(B)

OC

(4.18)

where from Eqs.4.16 and 4.17


dA

-1 [ 4wl d~

(4.19)

dCL,
2 Ja

CL.,

and

OB

dC

80s,

=-

1 f"

4'.

[erwo_w.
-

)L
4
10(
OCL, -

2wL,

, )I d

(4.20)

Now the slow GMS solution is differentiated with respect to the aerodynamic coefficient. The slow solution is given by
n,(() = (4wo - wi)-r

(4.21)

Partial differentiation with respect to CL, leads to the following.

C(4wo

_1 (4a
0,
_w
-_1
4

DCL,
-

4 wo - 2w, CL,)
--

(4.22)

Inorder to calculate the sensitivity of the full GMS angle-of-attack perturbations to


CL., itisnow required to determine the sensitivity of the coefficients wl and w0 to
the aerodynamic parameter. Once this isobtained, the information is substituted
into the above equations, and the effect of CL, on vehicle dynamics is obtained. The
effect of other aerodynamic coefficients such as C,,. and C,, on vehicle dynamics

can also be determined in a similar matter. All of the sensitivity equations developed
above are rendered valid for any vehicle parameter by simply replacing CLa with the
desired aerodynamic coefficient. Sensitivity solutions for C,, and C,, can also be
obtained by determining their respective effects on wl and w2 and substituting the
information into Eqs.4.14-4.22.
Differentiating Eqs.4.9 and 4.10 with respect to the aerodynamic parameters men-

tioned above leads to the following.


-=

OCLa

5' + 6v'/v - 6 2'C,,

(4.23)

dwo

(4.24)

Ow1

OWo

- 6

o=

(4.25)

O -60

(4.26)

-6a

(4.27)

Owo

80

= -6 2 0CL

(4.28)

By substituting the above expressions and the trajectory information into Eqs.4.144.22, the sensitivity of the GHAME vehicle angle-of-attack perturbations to CL,.
C,m, and C,, is determined.
Sensitivity solutions are calculated for the same initial conditions employed in
section 4.3. Sensitivity of a cosine-like solution is obtained with the initial conditions
a(o) = 1, a'(o) = 0 while sine-like solution sensitivity is studied with the initial
conditions a(eo) = 0, &'(o) = ki(eo). GHAME vehicle sensitivity to the three aero-

'dynamic parameters along the Shuttle trajectory for both sets of initial conditions is

shown in Figs.4-6 through 4-14. Sensitivity of the slow GMS solution to the various
parameters is also included.
It can be seen from these plots that the angle-of-attack perturbation sensitivity

to the three aerodynamic parameters all exhibit a similar behavior. For both sets
of initial conditions, the reentry sensitivity to CL., GCm,

and C,

oscillates with

the same frequency that was evident in the solutions to the actual angle-of-attack
dynamics. Also, the amplitude of vehicle sensitivity behaves in a similar manner for
all cases. Initially, the values are relatively small, after which they
increase to a maximum at approximately 50,000 to 60,000 vehicle lengths into the
trajectory. This is followed by an amplitude decay until sensitivity values reach zero
at approximately 90,000 vehicle lengths. Since the sensitivity of the angle-of-attack
perturbations reaches a maximum for all three aerodynamic parameters at virtually
the same time, this particular section of the trajectory is crucial to the longitudinal
dynamics.

When the GHAME vehicle reaches the point in the trajectory where

it has traveled 50,000 to 60,000 vehicle lengths, changes in the three aerodynamic
parameters could have profound effects on the longitudinal dynamics.
It is important to note that the three aerodynamic parameters do not all affect
the vehicle dynamics to the same degree. Although each parameter's effect on the
dynamics reaches a maximum at approximately the same time, the actual sensitivity
of angle-of-attack perturbations to each differs greatly. As can be seen from the
plots, angle-of-attack perturbations are least sensitive to changes in the moment due
to pitch rate parameter, C,,, while they are most sensitive to C,,.

In fact, angle-

of-attack dynamics are approximately 2000 times more sensitive to C,,. than they
are to C,,. Similarly, a change in C,,,. will have about 500 times a greater affect on
the longitudinal dynamics than the same change in CL. might produce. The relative
effects that each of the aerodynamic parameters has on the longitudinal dynamics of
the GHAME vehicle is result of the GMS slow solution sensitivity to the parameters.
The plots of the GMS slow solution sensitivity show that C,, has a much greater
impact on the full GMS solution than either of the other two parameters. It is clearly
evident that when concerned with the effect of changes in aerodynamic coefficients
on GHAME vehicle longitudinal dynamics, greater consideration should be given to
G,,, than the other aerodynamic parameters.

-0.
-0.
-0.
-0.
-0.
DCLO

-0.'
-0.
-0.
-0.'
0

10

x104

Figure 4-6: GMS Slow Solution Sensitivity to CLo

OCL

10

x 10'

Figure 4-7: GMS Sine-like Solution Sensitivity to CL,

I\

0.1

0.08
0.06
0.04
0.02
OCL.

0
-0.02
-0.04
-0.06
-0.08
-0.1
0

10
x10 4

Figure 4-8: GMS Cosine-like Solution Sensitivity to CL,

aC,
aCmat

10
x10 4

Figure 4-9: GMS Slow Solution Sensitivity to C,.

20

10

0o
0Cm

-10

-20

3-A

10
x10 4

Figure 4-10: GMS Sine-like Solution Sensitivity to C,.

30
20
10
0
-10
-20
-30
-W0

10
x 10

Figure 4-11: GMS Cosine-like Solution Sensitivity to C,.

-0.005

-0.01

49Cm

-0.015

-0.02

-0.025

-0.03

10
x104

Figure 4-12: GMS Slow Solution Sensitivity to C,,

U.U2

0.015
0.01
0.005

0Cm 1

-0.005
-0.01
.n n I

Figure 4-1:
Sensitivity
GMS Sine-liketo10
Solution

Figure 4-13: GMS Sine-like Solution Sensitivity to Cm,

10

0.02
0.015
0.01
0.005

~m

O
-0.005
-0.01
-0.015
ML0

0
x10 4

Figure 4-14: GMS Cosine-like Solution Sensitivity to C,

4.5

Stability Analysis and Displays

The stability of the second order longitudinal GHAME vehicle dynamics is investigated in this section. The stability of variable systems such as the GHAME vehicle
is, in general, very difficult to predict. Simple stability criteria applicable to this case
have been developed by Ramnath. This approach is used to predict the longitudinal
stability of the GHAME vehicle as it traverses the Space Shuttle reentry trajectory.
The effectiveness of such a criterion to provide useful information to the GHAME
vehicle flight crew is explored by presenting the stability information in the form of
a flight display.
An GMS criterion for longitudinal dynamics of an aircraft was developed by Ramnath as

P = CL, - CDT - 9Cm,

(4.29)

where CD, is the trim drag coefficient value and o is as defined in Eq.4.11. If the
stability parameter, P, is greater than zero, the vehicle's second order longitudinal
motions can be considered stable. If the expression in the Eq.4.29 is of negative
value, than the aircraft is longitudinally unstable. Substituting the GHAME vehicle
data into the above expression, longitudinal second order stability along the shuttle
trajectory is predicted. The stability parameter for the GHAME vehicle is plotted
versus vehicle lengths into the trajectory as well as time elapsed in Figs.4-15 and 4-16.
It can be seen from these plots that the stability parameter never becomes negative as the GHAME vehicle travels along the Space Shuttle trajectory. Therefore,
the second order longitudinal dynamics remain stable for the entire reentry. Figs.4-15
and 4-16 show that at approximately 60,000 vehicle lengths or 225 seconds into the
trajectory, the stability parameter is at a minimum and the GHAME vehicle is close
to becoming longitudinally unstable. It should be noted that this particular section
of the trajectory is also the time at which the GHAME vehicle was found to be most
sensitive to the aerodynamic parameters Cm,,, CL., and Cm,. The second order longitudinal dynamics are most affected by changes in the three aerodynamic coefficients
at the approximately the same time that the GHAME vehicle is closest to becoming

AR

".J

4
3.5
3
P

2.5
2
1.5
1

0.5

1.5

2.5

3
x10 5

Figure 4-15: Stability Parameter vs. Vehicle Lengths Along Trajectory

At
at.J

3.5
3
P

2.5
2
1.5
1
0.5

200

400

600

800

1000

1200

1400

1600

t(sec)

Figure 4-16: Stability Parameter vs. Elapsed Time Along Trajectory

i800

longitudinally unstable. It is obvious that particular attention should be paid when


traversing this particular area of the reentry trajectory.
In order to alert the flight crew of such critical sections in the trajectory, it may
be desirable to display stability information in the form of a display in the cockpit.
If it is assumed that all of the trajectory and vehicle information is known prior to
the flight, than the stability parameter, P, can be displayed in a manner such that
the flight crew is presented with its immediate past, present, and future values. As
suggested by Ramnath[15], a possible display of such a nature is shown in Fig.4-17
through 4-19. As seen in the figures, a bar graph concept is employed to show the
values of the stability parameter in the neighborhood of a particular instant along
the trajectory. Stability values are displayed for up to 60 seconds into the immediate
future and for 30 seconds of the immediate past. The display is updated continuously
and the elapsed time readout provides the flight crew with their relative location along
the entire trajectory. The displays in Figs.4-17 through 4-19 are the actual values
of the stability parameter, P, for the GHAME vehicle as it flies along the Shuttle
trajectory. The display in Fig.4-17 in which the vehicle has been flying along the
trajectory for 200 seconds, shows the area discussed above where the longitudinal
stability parameter is at its minimum. Using displays as the ones shown, it may
be possible to effectively inform the flight crew as to when the vehicle is traversing
sensitive and critical sections of the reentry trajectory.

~\\\\\\~

mrrnii

--

--

_i
'Csx

1
---

-30

-15

+15

I
+30

I
+4s

+0o

ELAPSED TIME: 30 sec

Figure 4-17: Stability Parameter Display at 30 Seconds into Trajectory

SEC.

-1

-30

q q9 9
U

~
I
-15

L
0

+15

+30

+45

Z
+60

SEC.

ELAPSED TIME: 200 sec

Figure 4-18: Stability Parameter Display at 200 Seconds into Trajectory

~c~

P
1

s
o0

K8%

hq
L
U

~r

K.

U=1

-1

-30

-15

+15

+30

+45

+60

ELAPSED TIME: 400 sec

Figure 4-19: Stability Parameter Display at 400 Seconds into Trajectory

SEC.

Chapter 5
Fourth Order Longitudinal
Dynamics
5.1

Overview

The fourth order longitudinal dynamics of the GHAME vehicle flying along the Space
Shuttle reentry trajectory are studied in this section. Attempts to approximate the
solutions to the longitudinal equations of motion are made through the use of the
fourth order Generalized Multiple Scales solutions as developed by Ramnath. These
approximations are again compared to numerical integration solutions in order to
determine the accuracy of GMS methods.

5.2

Equations of Motion

The longitudinal equations of motion for a vehicle in flight are in general non-linear
and time varying. The longitudinal reentry dynamics of the GHAME vehicle are
investigated in this chapter by developing approximate solutions to the equations of
motion linearized about a steady flight condition. This is justified by the PoincarkLyapunov Theorem which states that in most cases, the local stability of a non-linear
system is described exactly in the same manner as the behavior of its linearized form.
Insight into the longitudinal dynamics of the GHAME vehicle can be obtained by

simply developing solutions to the equations of motion linearized about some nominal
steady flight condition.
The general longitudinal dynamics of a flight vehicle are described by the following
three equations which are obtained through balancing the lift and drag forces as well
as the moments acting on the aircraft[9-11].
- mV + (T - D) - W( - a)=

(5.1)

L - W - mV(6 - &) = O

(5.2)

My - I4,d = 0

(5.3)

Eq.5.1 is the drag equation and represents a balancing of forces in the direction of
flight, while the lift equation shown in Eq.5.2 describes the sum of forces in a direction
perpendicular to the flight path. The moment equation shown in Eq.5.3 balances the
moments experienced by the aircraft around its center of mass. In order to linearize
these equations of motion, the flight parameters a, V, and 0 are represented as
perturbations about some steady flight value. They are given by
a = ao + Aa

(5.4)

v = Vo + av

(5.5)

0= 00 + AO

(5.6)

where ao, Vo, and Oo are the equilibrium steady state values. The forces of thrust,
lift, and drag as well as the aerodynamic moment are expressed about some nominal
steady state value through expansion in a Taylor series. Taylor series expansion allows
the forces and moment to be written as
L = Lo +

OL
OL
Aa +
AV + ...
8a
dV

(5.7)

OT

5VAV + ...

T = To +

OD
OD
D = Do + -OAV + -Aa+
OV
Oa

M = Mo +

aM
av

OM

AV + -

8a

aM

Aa + -

oe

(5.8)
...

OM

+ -- A& + ...
8,

(5.9)

(5.10)

Lo, To, Do, and Mo are nominal values of the aerodynamic forces and moment which
produce the equilibrium flight condition given by ao, Vo, and 0o. Eqs.5.4 through
5.10 are substituted into the the general non-linear equations of motion and upon
manipulation results in the following.
AM + AV(Dv - Tv) + Aa(Da - g) + gaO = 0

(
-

)AV +A& + ()

a-A

= 0

M,AV - M&Ad - MaAa + AM - MjAO = 0

(5.11)

(5.12)
(5.13)

The parameters Dv, D,, Tv, Lv/Vo, L,/Vo, Mv, Ma, Ma, and Mj appearing in the
equations above are the longitudinal stability derivatives of the vehicle and vary with
time as the flight conditions change along the reentry trajectory. The drag damping
term Dv is defined as
Dv =

1 0D
m aV

(5.14)

and all of the other stability derivatives relating the change in flight parameters to
drag and lift are defined in the same manner. They are defined as the partial derivative
of the lift or drag with respect to the flight parameter in question normalized by the
inverse of vehicle mass. Similarly, the stability derivatives involving the moment of
the vehicle are defined as the partial derivative of the moment with respect to the
flight parameter in question normalized by the inverse of the moment of inertia I,.

Hence, the speed stability term My is defined by the following.

Myv -

1 0M

(5.15)

The longitudinal stability derivatives of the GHAME vehicle and their approximation
along the reentry trajectory are detailed in the following section.
The linearized longitudinal equations of motion shown in Eqs.5.11 through 5.13
are simplified and written into the final state space form[9]

s + Dv-

Tv

Lv/Vo
-My

D -g

9
g9

a+ Ll/Vo

-sa

-(M& +Ms)as( - Mi)

AV

= 0
Ao

(5.16)

where s is the derivative operator A. The stability derivatives appearing in the state
equation are functions of different flight parameters such as air density and flight
velocity which vary with time as the GHAME vehicle reenters the Earth's atmosphere.
It is clear that the state equation describing the fourth order longitudinal motions is a
time varying system which cannot be solved employing traditional constant coefficient
methods. Attempts to develop approximate solution to Eq.5.16 are made using GMS
methods in section 5.4.

5.3

Longitudinal Stability Derivatives

From the equations of longitudinal motion developed in the previous section, it is


clear that the dynamics of the GHAME vehicle during reentry along the Space Shuttle
trajectory are very much dependent on the stability derivatives. The GHAME vehicle
stability derivatives along the shuttle reentry trajectory are detailed in this section.
For the purposes of dynamics and stability, the thrust velocity derivative, Tv, and
the angle-of-attack damping derivative, Ma often have little effect, and are therefore
ignored.

The longitudinal stability derivatives of the GHAME vehicle are approximated by


substituting vehicle aerodynamic data and other parameters of the Shuttle reentry
trajectory into equations available for estimating stability derivatives. The actual
equations used for the approximations are left to be detailed in Appendix A. The
longitudinal stability derivatives of the GHAME vehicle along the Shuttle reentry
trajectory are shown in Figs.5-1 through 5-7.

Each of the parameters is plotted

against the (, the number of vehicle lengths traversed along the trajectory.
There are certain tendencies exhibited by longitudinal stability derivatives which
are a result of the inherent nature of aircraft and the conventions of definitions.
For example, in a typical aircraft, the parameters D,, Dv, L,/Vo,

and Lv/Vo

are usually of positive value, and it can be seen that this is in fact the case for
the GHAME vehicle. Similarly, due to convention, the pitch damping derivative,
MO, is usually negative for conventional aircraft. Again, as can be seen in Fig.5-7, the
GHAME vehicle pitch damping derivative conforms and is always negative during the
reentry flight. It is clear that the longitudinal stability derivatives of the GHAME
vehicle during reentry do not exhibit strange anomalies.

It should be noted that

certain stability derivatives exhibit similar behavior as the vehicle progresses along
the trajectory. The derivatives Dv, L,,/Vo, and Lv/Vo increase in in a very similar
manner along the trajectory, while the drag angle-of-attack derivative, D,, and the
speed stability term, My exhibit the same behavior. This may be due to the fact
that these stability derivatives are functions of trajectory characteristics such as air
density in similar ways.
Two of the longitudinal stability derivatives involved in the equations of motion
have significant roles in determining the longitudinal stability of the aircraft. The
angle-of-attack stability parameter, M,,, determines the static stability of longitudinal
motions. If Ma is negative than the vehicle is statically stable. As seen in Fig.5-5,
throughout the entire reentry trajectory, the GHAME vehicle angle-of-attack stability
parameter remains negative, and the aircraft becomes statically more stable as it
progresses further. M, is closer to zero at initial parts of the reentry, and the vehicle
is less statically stable at these sections.

100
90
80
70
60
Dc( 1 )50
40
30
20
10
A

0.5

1.5

2.5

3
x10 5

Figure 5-1: Drag Angle-of-Attack Derivative vs. Vehicle Lengths Along Trajectory

x10-3

3.

Dv (

2.

1.

0.5

1.5

2.5

3
x10 5

Figure 5-2: Drag Damping Derivative vs. Vehicle Lengths Along Trajectory

'10
U. 10

0.16
0.14
0.12
0.1

La/Vo

S0.08
0.06
0.04
0.02
0

0.5

1.5

2.5

x10 5

Figure 5-3: Vertical Damping Derivative vs. Vehicle Lengths Along Trajectory

x10-6
3

2.5

Lv /Vo
1.5

(f)
1

0.5

0.5

1.5

2.5

x10 s

Figure 5-4: Lift Velocity Derivative vs. Vehicle Lengths Along Trajectory

-U.U4

-0.06
-0.08'
-0.1

Ma

-0.12

-0.14
-0.16
-0.18
-0.2
.02
-U.".

0.5

1.5

2.5

3
x10 5

Figure 5-5: Angle-of-Attack Static Stability Derivative vs. Vehicle Lengths Along
Trajectory

C,

x 10-6

-1

-1.5

My

ft sec
-2

-2.5

-3

-~

0.5

1.5

2.5

x10 5

Figure 5-6: Speed Stability Derivative vs. Vehicle Lengths Along Trajectory

-0.005
-0.01
-0.015

-0.02
Mei

-0.025

secI
-0.03

-0.035
-0.04
E

-A

nAc

1
0.5

2.5

3
x10 s

Figure 5-7: Pitch Damping Derivative vs. Vehicle Lengths Along Trajectory

Another important stability derivative with regards to dynamics is the speed stability term, My. If My is positive than it has a dynamically destablizing effect on
the aircraft while if it is negative than the tendency is to statically destabalize the
vehicle. Therefore, it is usually desirable to maintain My as close to zero as possible, and as seen in Fig.5-6, the speed stability parameter values for the GHAME
vehicle are extremely small. The speed stability term is statically most destablizing
at approximately 60,000 vehicle lengths into the trajectory and is dynamically most
destablizing after traveling 160,000 vehicle lengths.
As shown, it is possible to obtain some insight into the longitudinal dynamics of the
GHAME vehicle by analyzing the stability derivatives. The fourth order longitudinal
reentry dynamics of the GHAME vehicle described by these stability derivatives is
studied through the use of GMS theory in the next section.

5.4

GMS Solutions to Dynamics

The Generalized Multiple Scales theory is utilized in this section to study the fourth
order longitudinal dynamics of the GHAME vehicle as it reenters the atmosphere
along the optimal Shuttle trajectory. The linearized state equations which describe
the longitudinal vehicle motions are given in Eq.5.16. Since the stability derivatives
contained in the state equations vary along the trajectory as detailed in section 5.2,
it is clear that the fourth order systems is time-varying. In order to apply the GMS
solutions shown in section 3.3 to this time-varying system, the equations of motion
must be written as a fourth order linear differential equation. The transformation of
the linearized state equations into such a form is accomplished by making a dominant
approximation to the actual longitudinal equations of motion.
For the purposes of transforming the state equations into a fourth order linear
differential equation, it is assumed for a moment that the stability derivatives contained in Eq.5.16 are all of constant value along the trajectory. If this is the case,
than solutions to the dynamics of the three independent variables, A V, Aa, and
AO, are identical, and their response is described by solving an equation obtained

by calculating the determinant of the main matrix in Eq.5.16. Ignoring the stability
derivatives Tv and Ma, we have

s + Dv
det

Lv/Vo

-Mv

D, - g
a+

9
o

-M,

-M 8 4 +W

-L

+ W25

+ W1 + Wo

(5.17)

s(s - Mj)

where
W3

= La/Vo - Mi + Dv

(5.18)

= Dv(La/Vo) - DvMi - Mi(La/Vo) - Ma - Da(La/Vo) + g(Lv/Vo)

(5.19)

WO= MvD: - MaDv - DMi(Lm/Vo) + DM4(Lv/Vo) - gMi(Lv/Vo)

(5.20)

W2

WO = g[Mv(Lc/Vo)-

Ma(Lv/Vo)]

(5.21)

and s again is the derivative operator d. Under the assumption that the longitudinal
stability derivatives in the above equations are constant, the full longitudinal response
of the vehicle is described by setting the above determinant equal to zero and replacing
the higher order s terms with their respective higher order derivatives. Since the
independent variables, AV, AO, and Aa have the same response, all three will be
replaced by a generic variable y, and the longitudinal dynamics of the GHAME vehicle
are represented by the expression
d~y
d4y +
dt4

day + W d2y + W1 dy + WOY = 0


dt3

dt2

dt

(5.22)

As shown by Ramnath, this equation is a dominant approximation to the actual


longitudinal behavior and is made under the assumption that the stability derivatives
in Eqs.5.18-21 are constant. However, it is recognized that the coefficients in Eq.5.22
and the stability derivatives actually do vary with time, and they are now allowed to
do so. This finally results in the longitudinal dynamics of the GHAME vehicle being

represented by the fourth order linear time-varying equation

d4t

S+

(t)

d3t

+2(t)

d2t

()+

dy

w(t)y = 0

(5.23)

where the coefficients w3 , w2 , w 1 , and w0 are as defined in Eqs.5.18 through 5.21.


The fourth order GMS solutions shown in section 3.3 require that the each mode
of motion and its corresponding roots be identified before an approximation to the
dynamics can be made. The roots associated with the longitudinal equation of motion
are determined by solving the algebraic equation

a4

3
38 s

+ W2 2 +

s+w = 0

(5.24)

Since the coefficients of the above expression vary with time, the roots of the equation
will move as the vehicle traverses along the trajectory. The roots associated with the
longitudinal motions of the GHAME vehicle and their movement with time are shown
in Figs.5-8 and 5-9. Fig. 5-8 clearly shows the roots of the short period mode and
the way they vary as the vehicle travels 1657 seconds into the Shuttle trajectory. As
expected, the short period mode is a complex conjugate pair and remains in the lefthalf plane throughout the entire trajectory. As the GHAME vehicle progresses further
into the atmosphere, the short period damping as well as the frequency increases. Due
to the fact that the phugoid mode and the short period mode of the vehicle occur on
such differing frequency scales, in Fig.5-8, the roots representing the phugoid mode
appear as a set of unresolvable points near the origin. The movement of the phugoid
roots along the trajectory is plotted in detail in Fig.5-9.
It is clear that the roots of the phugoid mode do not behave in a conventional
manner. As the GHAME vehicle begins its reentry, the phugoid roots are a pair
of complex conjugates in the right-half plane. As the vehicle progresses along the
trajectory, the pair of roots move into the left-half plane and then once again return
toward the right-half plane. At approximately 400 seconds into the reentry trajectory,
the complex phugoid roots abruptly become a pair of real roots, one negative and
the other positive. These real roots move towards the origin until approximately 600

__

11

V.J

0.4

..
t=778

t=57

0.3
0.2

. .

. ....
. . . . . . . . . . . . . . . . . . . . . . . ... .

0.1

............ ............ ...

..

--_
................
................
...
..
..........................
. . . ....
... ..
. . . . . . ... . . . . . . . ...

. ... . .. . . ... .

.........................

. . . . . . . . ..

. . . . . . :..
.. .... ..

.. .. . ..

. . . .. . . . . . . . . . . . . . . .

......
............................. I
...

....................

..

,.................
..............
..... : ................
.............
........
..............
....
..........
....
.....
.......
~~~~~"
....
....
.. '~'

-0.1

. ............. .... .. ..................................

_n ,

... .........................

................. .......... .. . .......................... .. .. .... ............ ...... .

. . . .. . . . .. . .
.
. . . . . . . . . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . . ....

. . . . . . .. . . . .. . . . . . . . . . . . . .. .. . . . . . . . . . . .. . . . . . . .. . . . . . . . . . . . .. . . . . . . . . . . . . .. . . .

. .

. .

. .

. .

......
.................

..
..
..
..
....
..
..
..
,...
I.

-01

-0.05

-0.04

-0.03

-0.02
k,

-0.01

Figure 5-8: Short Period Roots Along Trajectory

0.01

xlO-3

t=670
t=0

tt=1 57
t=569:

t386

k, 0

...............

t=1657
/

t=569

t=386

..... ..
... .
.............................

.....................
3
t--302O

t=670
.... ... ... ... ... ... ... .... .... .

-2
-2
_1

-1.5

-1

-0.5

0.5

1.5

2
x10-3

Figure 5-9: Phugoid Mode Roots Along Trajectory

78

seconds into the trajectory at which point they once again become complex.
The peculiar behavior of the phugoid mode roots requires greater rigor and a more
careful use of Generalized Multiple Scales to study the GHAME vehicle dynamics.
The points at which the phugoid roots change from complex conjugates to real roots
and vice versa are known as 'turning points' and represent a change in the nature
of the mode between oscillatory and non-oscillatory behavior. In regard to the GMS
theory, these 'turning points' present additional mathematical difficulties. When a
'turning point' is present, greater care is needed in generating the GMS approximations. Because of this, the study of GHAME vehicle longitudinal dynamics as it
travels along the Shuttle reentry trajectory is restricted to predicting only the short
period behavior of the vehicle.
The GHAME vehicle root information shown in Fig.5-8 is substituted into the
fourth order GMS solutions given in section 3.3 in order to approximate the nature of
the short period mode motions. Once again, two sets of initial conditions are chosen
to produce solutions of different forms. In order to simulate a sine-like solution, the
initial conditions are set at y(O) = 0, y'(0) = .2181, while a cosine-like solution is
obtained with the initial conditions y(O) = 1, y'(0) = 0. The fast GMS solution as
well as the full GMS solution to the short period behavior

for both sets of initial

conditions are shown in Fig.5-10 and 5-11. Unlike the GMS solutions to the second order angle-of-attack perturbations obtained in Chapter 4, these approximations
exhibit virtually no difference between the fast solution and the complete solution.
The fast GMS solution contains all of the frequency and magnitude information to
be obtained by GMS theory and the addition of the slow solution has virtually no
effect on the approximation. Once again, to determine the degree of accuracy with
which the GMS theory predicts the dynamics of the GHAME vehicle, a Runge-Kutta
integration scheme was employed to obtain numerical solutions to the short period
mode. The numerical solutions to the short period behavior along with the full GMS
approximations for both sets of initial conditions are plotted in Figs.5-12 and 5-13.
As with the angle-of-attack perturbations in Chapter 4, the GMS approximations
coincide extremely well with their numerical counterparts. In both the sine-like and

1
0.8
0.6
0.4
0.2
YpI 0
-0.2
-0.4
-0.6
-0.8
_1

100

200

300

400

500

600

t(sec)

Figure 5-10: Sine-like GMS Solutions to Short Period Mode

700

800

Yap

100

--

200

---

300

400

---

500

---

600

700

t(sec)

Figure 5-11: Cosine-like GMS Solutions to Short Period Mode

---

800

0.8
0.6
0.4
0.2
MYp

-0.2
-0.4

-0.6
-0.8
-1

100

200

300

400

500

600

700

t(sec)

Figure 5-12: Sine-like Numerical and GMS Solutions to Short Period Mode

800

0.8
0.6
0.4
0.2
Yis

-0.2
-0.4
-0.6
-0.8
0

100

200

300

400

500

600

700

t(sec)

Figure 5-13: Cosine-like Numerical and GMS Solutions to Short Period Mode

800

cosine-like solutions, the amplitude of the short period behavior is overshot by the
GMS approximations, however, the discrepancy is small and disappears as the vehicle
progresses further into the trajectory. Unfortunately, due to the presence of the turning points in the phugoid mode, a complete study of the fourth order longitudinal
reentry dynamics of the GHAME vehicle could not be accomplished in this study.
However, the accuracy of the asymptotic method is again demonstrated in predicting
the short period behavior of the vehicle.

Chapter 6
Fourth Order Lateral-Directional
Dynamics
6.1

Overview

In this chapter, the lateral-directional dynamics of the GHAME vehicle as it traverses


the Space Shuttle reentry trajectory are studied. As before, the motions of the hypersonic vehicle are investigated through the use of Generalized Multiple Scales theory
which provides approximation to non-autonomous differential equations. The GMS
solutions obtained to the lateral-directional dynamics of the GHAME vehicle are once
again compared to numerical solutions in order to determine the accuracy of such approximations. Finally, a sensitivity analysis of the lateral-directional motions is conducted by differentiating the GMS solutions with respect to the lateral-directional
stability derivatives.

6.2

Equations of Motion

Similar to their longitudinal counterparts, the equations of motion describing the


lateral-directional dynamics of a flight vehicle are generally non-linear. The PoincarkLyapunov theorem is again invoked, and in order to gain insight into the behavior of
the GHAME vehicle, GMS theory is employed to approximate solutions to the lateral-

directional equations of motion linearized about a nominal steady flight condition.


The non-linear equations describing the lateral-directional motions of the GHAME
vehicle are given by[9, 10, 11]
Y - mi - mVr + mge = 0

(6.1)

L - I,,i + Iz, = 0

(6.2)

N - I,z + Iz = 0

(6.3)

where Y, L, and N are respectively the aerodynamic side force, rolling moment, and
yawing moment on the vehicle. The variable v represents the component of velocity
perpendicular to the flight path while r and are the yaw rate and roll angle of the
vehicle. The three equations shown above are a result of balancing aerodynamic and
inertial forces as well as moments which affect lateral-directional motions. Eq.6.1 is
developed through the equating of forces in the direction perpendicular to the flight
path. Eqs.6.2 and 6.3 are the result of balancing the rolling and yawing moments of
the aircraft. These three equations are now linearized using the exact same process
employed to linearize the longitudinal equations of motion. The three parameters v, r,
and q are represented as perturbations about some nominal steady state value, while
the aerodynamic force and moments are rewritten in the form a Taylor expansion
given by
Y = YO +

L = Lo +
N = No +

v + ...

(6.4)

9L

OL
OL
Av
+
Ar
+
Ap + ...
v
Or

Op

ON

ov

ON
Or

Av + j

ON

r + -

Op

p + ...

(6.5)
(6.6)

where p is the roll rate and the other variables are as defined before. These Taylor
series representations and the perturbation forms of v, r, and 4 are substituted into
the nonlinear equations of motion given in Eq.6.1 through 6.3. Upon manipulation,
the resulting linearized lateral-directional equations of motion given in state space

form are[9]
[ -Y

-L,

-L,

-N,

a - N,

s2

-g

Av

- Ljs

Ar

- Np

where s is once again the derivative operator .

(6.7)

The parameters Yt, L,, L,, L,,

N,, N,, and N, appearing in the linearized state equation are the lateral-directional
stability derivatives of the aircraft and vary with time as the GHAME vehicle travels
along the reentry trajectory. The stability derivative Y, is defined as

Y,

1 oY
m dv

(6.8)

The stability derivatives involving the rolling and yawing moments are defined as the
partial derivative of the moment with respect to the particular parameter normalized
by the inverse of the relevent moment of inertia. For example, the dihederal term L,,
is given by
L, =

L
I,, 0v

(6.9)

All of the other derivatives involving rolling moment are defined in the same manner.
Similarly, the yawing moment derivatives are normalized by the inverse of I,, and
can be written in the same form as N, which is defined as

N, =

1 8N

(6.10)

The lateral-directional stability derivatives of the GHAME vehicle during reentry


along the optimal trajectory are detailed in the next section. As with the longitudinal
derivatives, these parameters vary along the trajectory as the flight conditions change,
and the linearized equations of motion in Eq.6.7 are time-varying. GMS theory is
employed in section 6.3 to develop approximate solutions to this system and gain
insight into the lateral-directional reentry behavior of the GHAME vehicle.

6.3

Lateral-Directional Stability Derivatives

It is clear by examining the lateral-directional equations of motion shown in Eq.


6.7 that the stability derivatives of the GHAME vehicle primarily determine its
lateral-directional behavior. Similar to their longitudinal counterparts, these lateraldirectional stability derivatives are functions of a variety of parameters and vary with
the changing flight conditions during reentry. The lateral-directional stability derivatives of the GHAME vehicle along the Shuttle trajectory are shown in Figs.6-1 through
6-7. The derivatives are plotted against , the number of vehicle lengths traversed
along the trajectory by the GHAME vehicle. The equations employed to approximate
the lateral-directional stability derivatives are left to be detailed in Appendix A.
Although the values of the lateral-directional stability derivatives largely determine the dynamics, insights into the nature of the vehicle and trajectory can be
gained by studying the derivatives themselves. For example, due to the dihederal of
the main wings, sideslip components of velocity causes most aircraft to experience
a change in local angle-of-attack values between the two wings. Hence, the vehicle
experiences a rolling moment due to sideslip velocity. The stability derivative L,,
represents this tendency, and its values for the GHAME vehicle along the optimal
reentry trajectory are plotted in Fig.6-2. Since the GHAME vehicle has no dihederal angle in the main wings, the values of L. are relatively small. Although main
contributions to the dihederal effect are due to the main wings, the fuselage and tail
can also be factors. They are likely responsible for the small positive values of L,
exhibited by the GHAME vehicle. For stability purposes, it is desirable to keep the
dihederal derivative within narrow limits, and as seen in Fig.6-2, L,, remains almost
at a constant value through most of the reentry trajectory.
Another important lateral-directional stability derivative is the directional stability term N,. This parameter represents the tendency of flight vehicles to yaw into
the relative wind and is mainly caused by the change in oncoming airflow angle experienced by the vertical tail. The values of N,, for the GHAME vehicle are plotted
versus vehicle lengths into the reentry trajectory in Fig.6-5. It can be seen that early

-0.002
-0.004
-0.006

(.e) -0.008
-0.01
-0.012
-

lA
1

0.5

1.5

2.5

3
x10 3

Figure 6-1: Side Force Due to Sideslip vs. Vehicle Lengths Along Trajectory

x104
3

2.5

L,

1.5

ft see
1

0.5

0.5

1.5

2.5

x10 5

Figure 6-2: Dihederal Effect Derivative vs. Vehicle Lengths Along Trajectory

0.014
0.012
0.01

-C,

0.008
0.006
0.004
0.002
A

0.5

1.5

2.5

3
x10 s

Figure 6-3: Rolling Moment Due to Yaw Rate vs. Vehicle Lengths Along Trajectory

-0.01

LP -0.02

-0.03

-
0

0.5

1.5

2.5

3
x105

Figure 6-4: Roll Damping Derivative vs. Vehicle Lengths Along Trajectory

x102

1.8
1.6
1.4
1.2
N,

ftsc)0.8
0.6
0.4
0.2
0

0.5

1.5

2.5

x105

Figure 6-5: Directional Stability Derivative vs. Vehicle Lengths Along Trajectory

x10-3
0

-1
-2
-3
N,

-5
-6
-.7

0.5

1.5

2.5

3
x10 5

Figure 6-6: Yaw Damping Derivative vs. Vehicle Lengths Along Trajectory

94

x 10-4

2
1

0
-1
-2
p

-3
-4

-5
-6
-7

8n

0.5

1.5

2.5

x10 5

Figure 6-7: Yawing Moment Due to Roll Rate vs. Vehicle Lengths Along Trajectory

in the trajectory, the values of N, are not much different from those of L,, and a
sideslip velocity will approximately cause equal rolling and yawing moments on the
vehicle. However, as the GHAME vehicle progresses further into the atmosphere, the
absolute value of N, increases, and the directional stability term has a much greater
effect on the vehicle than the dihederal term. In fact, with the exception of of L,,
all of the lateral-directional stability derivatives increase their effects on the vehicle
as it travels further into the atmosphere. Since the stability derivatives represent
aerodynamic tendencies of the vehicle, in all probability, the increased effects on the
vehicle are due to the increase in air density further into the atmosphere.
The stability derivatives again with the exception of the dihederal term, do not
exhibit abnormal behavior with regard to sign behavior at the hypersonic speeds
encountered during reentry. The parameters remain as they would for conventional
subsonic aircraft except for L,. In conventional aircraft with positive dihederal angles,
the definition of the standard coordinate system implies that the dihederal term is
negative. If the aircraft sideslips to the right, the result is for the vehicle to roll
right wing up. However, Fig.6-2 shows that the GHAME vehicle dihederal term is
positive throughout the trajectory. If the GHAME vehicle sideslips to the right, it
will experience a rolling moment causing the right wing to go down. If it is assumed
that L, is only affected by wing dihederal, then aerodynamically this makes no sense.
However, since the GHAME vehicle has no wing dihederal, it is conceivable that the
positive values of Lv are caused by the effects of the fuselage and tail.
The lateral-directional stability derivatives of the GHAME vehicle are employed
in the next section to approximate solutions to reentry dynamics. Due to the fact
that these parameters vary along the reentry trajectory, simple constant coefficient
methods are not very useful, and GMS theory is employed. The sensitivity of the
lateral-directional dynamics to each of these stability derivatives is considered in
section 6.4.

6.4

GMS Solutions to Dynamics

Ramnath's GMS analysis outlined in Chapter 3 is utilized in this section to approximate the lateral-directional dynamics of the GHAME vehicle. The linearized
lateral-directional equations of motion shown in Eq.6.7 are time-varying, and the
asymptotic method is required to provide solutions to the system. As was the case
with longitudinal dynamics, in order to apply GMS theory, it is convenient to express
the linearized equations of motion in the form of a fourth order linear differential
equation. This transformation is accomplished in exactly the same manner which
was employed with the longitudinal equations of motion. First, it is assumed that the
stability derivatives described in the previous section are of constant value through
out the trajectory. Under such an assumption, the three independent variables of
the lateral-directional equations of motion Av, Ar, and AOb have the same response.
They are replaced by the generic variable y whose dynamics are described by the
determinant of the main matrix in Eq.6.7. Following the same procedure and reasoning employed in transforming the longitudinal equations of motion, a dominant
approximation to the lateral-directional motions is developed in the form desired. It
is given by the fourth order linear time-varying differential equation
d
dd4+

,ydy+
3
t

(t-)

d2y

+ w(t)

dy

+ Wo(t)

=0

(6.11)

where
w3 (t) = -Lp - N, - Yv

(6.12)

w2(t) = VN, - L,N, + YLp + N,(L, + Y,)

(6.13)

wi(t) = Y,(L,Np - N,L,) - gL, + VNpL - VLN,

(6.14)

wo(t) = g(L,N,. - NL,)

(6.15)

Solutions to this time-varying differential equation are approximated by the GMS


method to determine the lateral-directional reentry behavior of the GHAME vehicle.
The fourth order GMS theory detailed in Chapter 3 approximates the solution

of a differential equation by first approximating the dynamics to each of the system's modes of motion. These approximations require that the characteristic roots
associated with each of the GHAME vehicle's modes be determined along the entire
trajectory. This is accomplished by solving the algebraic equation

s4 +

W3 s3 + W2s

+ w1S + W0 = 0

(6.16)

where the coefficients are as defined in Eqs.6.12 through 6.15. Since, these coefficients
vary with time, it is expected that the roots associated with the modes of motion will
not remain stationary. The roots representing the lateral-directional motions of the
GHAME vehicle are plotted in Fig.6-8 for up to 1657 seconds into the trajectory.
It is clear that the GHAME vehicle possesses the three modes of motion which are
typical of lateral-directional behavior in conventional aircraft. The dutch roll mode is
represented by the complex conjugate pair of roots while the spiral divergence and roll
convergence modes are identified by the two roots on the real axis. The root which
remains in the left-half plane represents the roll convergence mode, while the other is
the spiral divergence root. As the GHAME vehicle travel further into the atmosphere,
the dutch roll mode experiences increase in both the frequency and damping terms.
The changes in the frequency term are greater than those of the damping term which
varies very little. Fig.6-8 shows that the roll convergence root initially starts close
to the origin than travels away from it as time progresses. The spiral root generally
remains in the same area. Figs.6-9 and 6-10 are plots of the roll convergence and
spiral divergence roots versus time, and better show the way in which these roots
vary along the trajectory.
The fourth order GMS equations shown in section 3.3 are now employed in order
to approximate the characteristic motion associated with each of the GHAME vehicle's three lateral-directional modes of motion. Since the roll convergence and spiral
divergence modes are each represented by a single real root, their respective dynamics

0.8

-.................................

.....
................
....

0.6

...................

0.4 i-

.. :.......

0.2 -

...........
;657
''
t,..

0
-0.2 .
-0.4

&

"

....................
657.............

.-....... ... . . . . . . . .

1[

"

..........................

..

.............
;................-

-0.6
-0.8

. .......
..

-1

-0.035

-0.03

-0.025

-0.02

-0.015

-0.01

-0.005

Figure 6-8: Lateral-Directional Roots Along Trajectory

99

0.005

0.01

-0.005

-0.01

k,, -0.015

-0.02

-0.025

n Wn
0

200

400

600

800

1000

1200

1400

160U

Figure 6-9: Roll Convergence Root vs. Time into Trajectory

100

1800

x10-3
2.6
2.4
2.2
2
1.8
ka,

1.6

1.4
1.2
1

0.8
nI

AM

200

400

600

800

1000

1200

1400

1600

Figure 6-10: Spiral Divergence Root vs. Time into Trajectory

101

1800

are predicted by the expressions

y,,(t) = ezp (fotk,(t)dt)

yoP(t) = exp (

(6.17)
(6.18)

k(t)dt)
kf

k,, and k,P are the roots of the roll convergence and spiral divergence modes respectively and are plotted in Figs.6-9 and 6-10. The characteristic dynamics of the
remaining dutch roll mode are represented by a pair of complex conjugate roots as
seen in Fig.6-8. If these roots are defined as kd, = kdr,, + iked,i, then the dutch roll
response is approximated by the equation

(6.19)

Ydr(t) = C3Ydrl(t) + C4Yd2(t)


where C3 and C 4 are arbitrary constants. yd,1 and yd,2 are given by

Yd

(t) = exp (ft I2k~


' (t)

yd,2(t)

exp (

kd,,(t)dt ain

= ezp (f ild (t ) dt) ep (f kd,(t)dt) cos (

k(t)dt)
k(

t kdi(t)dt

(6.20)

(6.21)

The roots of the GHAME vehicle's three modes along the Shuttle reentry trajectory
are substituted into the above equations, and the characteristic motions of the lateraldirectional behavior are predicted. The GMS approximations to the characteristic
reentry dynamics associated with the roll convergence and spiral divergence modes
are shown in Fig.6-11 and 6-12. Since the roll convergence roots remain in the lefthalf plane, it is expected that its response is stable. As seen in Fig.6-11, this is in fact
the case. Similarly, the response of the spiral mode is unstable due to the fact that
its roots remain in the right-half of the complex plane through out the entire reentry
trajectory. Also, the dynamics of the spiral mode diverges faster than the roll mode
converges. The characteristic motions of the dutch roll mode during reentry are shown
in Figs.6-13 and 6-14. The first of these figures represents the sine wave component

102

of the dutch roll motions. The cosine component, Ydr2, is shown in Fig.6-14. Both
plots show that the dutch roll reentry dynamics are damped oscillations with high
frequency and little damping. There is little change in either as the GHAME vehicle
progresses along the trajectory.
The responses shown in Figs.6-11 through 6-14 are approximations to the characteristic motions associated with each of the GHAME vehicle's three lateral-directional
modes. In order to predict the complete lateral-directional behavior during reentry,
the approximations to each of the modes are combined in a linear fashion. The full
GMS approximation to the solution of Eq.6.11 is given by
y(t) = Clye(t) + C

ys(t)
+ C3Ydl(t)+ C4ydr2(t)

(6.22)

where C1, C2,Cs, and C4 are arbitrary constants dependent on initial conditions of
the original differential equation. Using this linear combination of the characteristic
modal motions, the lateral-directional reentry behavior of the GHAME vehicle is
predicted for two different sets of initial conditions. In order to assess the accuracy of
the GMS approximations, numerical solutions are also obtained for both sets of initial
conditions. Fig.6-15 shows both GMS and numerical approximations to the reentry
lateral-directional dynamics for the initial conditions y"'(0) = 1, y"(0) = 0, y'(0) = 0,
and y(O) = 0. The GMS and numerical solutions for the initial conditions y"'(0) = 0,
y"(O) = 0, y'(0) = 1, and y(O) = 0 are plotted in Fig.6-16. It can be seen for both sets
of initial conditions that the GMS approximations are virtually indistinguishable from
the numerical solutions. Since the characteristic motion associated with the spiral
divergence mode is unstable, the full lateral-directional behavior of the GHAME
vehicle during reentry is also unstable. The first set of initial conditions produces
a response containing enough of the dutch roll motions to exhibit some oscillatory
behavior. The dynamics for the second set of initial conditions contain very little
of the dutch roll motions and therefore produces a response which appears to be
non-oscillatory.
In this section, the lateral-directional dynamics of the GHAME vehicle along

103

0.9
0.8
0.7
y,e

0.6

0.5
0.4
0.3

01
0

100

200

300

400

500

600

t(sec)

Figure 6-11: GMS Solution to Roll Convergence Dynamics

104

700

~10

4.5
4
3.5
3
Yap

2.5

2
1.5

'A

100

200

400

300

500

600

700

Figure
6-12:
Spiral
GMS
Solution
to Divergence Dynamiec)

Figure 6-12: GMS Solution to Spiral Divergence Dynamics

105

8UU

0.8
0.6
0.4
0.2
adrl

-0.2
-0.4
-0.6
-0.8
-1

20

40

60

100

80

120

140

160

t(sec)

Figure 6-13: GMS Solution to Sine-like Dutch Roll Dynamics

106

180

200

0.8
0.6
0.4
0.2
Ydr

-0.2
-0.4
-0.6
-0.8
-1
0

20

40

60

80

100

120

140

160

t(sec)

Figure 6-14: GMS Solution to Cosine-like Dutch Roll Dynamics

107

180

200

AA

2O0

- Numerical
M
---GMS

180
160
140
120
100
80

60-

40-

y(O)

V(o)

y"(O)

y"'(0)

20n

.._.

10

30

20

40

50

t(sec)

Figure 6-15: GMS Solution to Lateral-Directional Reentry Dynamics

108

60

Numerical

-Numerical
--- GMS

500

400

300

200

100

n
U

50

100

200

150

y'(O)

yt"(0)

250

300

350

t(sec)

Figure 6-16: GMS Solution to Lateral-Directional Reentry Dynamics

109

400

the Space Shuttle reentry trajectory were predicted by employing fourth order GMS

approximations. The GMS solutions were indistinguishable from numerical solutions,


and once again proved themselves to be accurate. The lateral-directional reentry
behavior of the GHAME vehicle was shown to be unstable due to the instability
of the characteristic motion associated with the spiral divergence mode. Since the
lateral-directional variable y is a dominant approximation to the actual variables Av,
Ar and Aq0, these parameters should possess the same unstable behavior found in
the above analysis.

6.5

Sensitivity Analysis

In this section, the sensitivity of the GHAME vehicle's lateral-directional reentry dynamics to the stability derivatives detailed in section 3.3 is studied. The effects of
changing these stability derivatives on the dynamics of the vehicle are determined
by employing the GMS approximations to the motions associated with each of the
lateral-directional modes. Once again, due to the analytical forms of these GMS approximations, it is possible to conduct a sensitivity analysis by partial differentiation
with respect to the stability derivatives. As noted in section 4.4, partial differentiation does not provide a true sensitivity analysis due to the fact that the stability
derivatives themselves are functions of time. In order to determine the exact sensitivity of the dynamics to these parameters, variational calculus is necessary. However,
partial differentiation is justified by Ramnath's asymptotic sensitivity theory and is
employed in this study.
As with the approximation of the actual lateral-directional dynamics, the sensitivity of vehicle motions to the stability derivatives is developed by first considering
the motions associated with each of the modes. The sensitivity of the characteristic
modal motions is determined by partially differentiating Eqs.6.17 through 6.21 with
respect to the stability derivatives. For example, the sensitivity of the roll convergence and spiral divergence motions to the directional stability derivative is given
by partially differentiating Eqs.6.17 and 6.18 with respect to N. . After, shifting the
110

differentiation inside the integral, this lead to

exp j=kod ) t
=yp
ezp

_N (o

'

Ok,

kt

yrc

(6.23)

dt

(6.24)

kdt, i
N
0 a6.24

k, dt)

In order to determine the sensitivity of the dutch roll motions, only the fast GMS
solution is considered. The slow solution has little effect on the approximation, and for
the purposes of simplicity, it is disregarded. The sensitivity of the sine-like and cosinelike dutch roll motions to the directional stability term is determined by differentiating
Again, after shifting the

the fast GMS solution contained in Eqs.6.20 and 6.21.


differentiation inside the integral, this results in
0

Ydr

(t

kdrr dt

dO(t kdri t

NYdr

exp

exp

rr dt exp

kdrirdtt sinl

(6.25)

kdrd) cos (

kdridt

kdrrdt COS

kdridt) -

(6.26)

(ft O0'.i dt) exp (ft kdrdt)

sin

'

(ft kdridt

The differentiation can be accomplished in the same manner for the other stability
derivatives, and their sensitivities can be obtained simply by replacing N,, with the
parameter of choice in the above equations. In Eqs 6.23 through 6.26, the partial
derivatives

, and -9-k

O' _
ON,

ON,'

ON,

still need to be determined before the sensitivity

ON,

of each characteristic motion is obtained. It is known that the coefficients of the


algebraic equation which determine the mode roots are functions of the stability
derivatives as defined by Eq.6.12 through 6.15.

Also, the roots can be explicitly

written as functions of the coefficients. Therefore, in order to obtain partial derivatives


of the roots with respect to the stability derivative, the chain rule is employed. Since
the process relating the roots of Eq.6.22 to its coefficients is rather tedious, the details
111

of the complete differentiation are left for Appendix B. The results of the chain

rule differentiation and root information is substituted into the above equations to
complete the analysis. The sensitivities of the GHAME vehicle's lateral-directional
reentry modal motions with respect to the seven stability derivatives are shown in
Figs.6-17 through 6-44. The results of the dutch roll mode are separated into the
sensitivities of the sine and cosine components.
In order to obtain the sensitivity of the full lateral-directional reentry dynamics,
a linear combination is employed once more. Without considering the dependence
of initial and boundary conditions on the parameter, differentiating the linear GMS

approximation to the full dynamics given in Eq.6.22 leads to

-y

Oyre

S O,N, +
ON,

VO! J P d l
y, ++
+ O,8NN,
ON,
8

C4

2
dr+
2
N,

(6.27)

where C1 , C2 , C3 , and C4 are constants previously determined by the initial conditions of the original differential equation. Justification for this procedure is based on
Ramnath's work on GMS sensitivity theory. It is evident from section 6.4 that the
full lateral-directional response of the GHAME vehicle is very much dependent on
the initial conditions. The initial conditions determine the amount of influence that
each of the modes of motion has on the full response. Since initial conditions can be
chosen arbitrarily and the presence of each mode can vary, it is meaningless to study
the sensitivity of the GHAME vehicle lateral-directional dynamics for a specific set
of initial conditions. Much more insight is gained by considering the sensitivities of
each mode of motion which are combined to obtain the full sensitivity.
It is clear from studying Figs.6-17 through 6-44 that the sensitivities of the modal
motions associated with respect to each of the stability derivatives exhibit similar
patterns of behavior. First, all of the stability derivatives have the most influence on
the spiral divergence mode. A change in any of the seven stability derivatives will
have a greater effect on the dynamics of the spiral mode than on either of the roll
convergence or dutch roll modes. The roll convergence dynamics are almost as sensitive as the spiral divergence motions, however, are not quite as affected by variations

112

in the stability derivatives. Variations in any stability derivative has the least effect
on the motions associated with the dutch roll mode. Its sensitivity is several orders
of magnitude less than the sensitivities of the other two modes. Therefore, lateraldirectional responses which contain relatively greater portions of the spiral divergence
mode are more sensitive to changes in the stability derivatives than responses which
may contain relatively greater portions of either the roll convergence or dutch roll
modes.
Although variations in any of the seven stability derivatives will affect the spiral
mode more than the other two, each of the stability derivatives does not affect a
particular motion in the same manner. It is evident from Figs.6-17 through 6-44 that
each of the modal motions is more sensitive to the directional derivative, N, than any
of the other stability derivatives. The roll convergence, spiral divergence, and dutch
roll motions are each more affected by changes in N, than any of the other six stability
derivatives. An equal change in the dihederal term, L, has almost as much affect,
however, the other derivatives have sensitivities which are several orders of magnitude
less than the sensitivity with respect to the directional derivative. This behavior of the
GHAME vehicle is similar to that of a conventional aircraft whose lateral-directional
motions are also most sensitive to N, and L,,. Since the directional stability term
is very much a function of the vertical tail size of the aircraft, the GHAME vehicle
vertical tail size has great implications on its reentry lateral-directional dynamics. Of
the other stability derivatives, the plots reveal that L,, Y,, and Np have the least
effect on the dynamics.

Particularly, the sensitivity of each modal motion to Lp

remains relatively very small for up to approximately 500 seconds into the trajectory.
Compared to the sensitivities of modal motions to N,, and L,, variations in Lp, Y,,
and Np have virtually no effect on the lateral-directional reentry dynamics of the
GHAME vehicle.
It is evident from the plots in Figs.6-17 through 6-44 that the sensitivities of the
characteristic spiral divergence motion with respect to all the stability derivatives
grow unbounded. This is to be expected since the actual spiral mode dynamics are
unstable. The sensitivities of the roll convergence and dutch roll modes both grow
113

rapidly at approximately 600 seconds into the trajectory. The time histories of the
roll convergence and dutch roll sensitivity variations are exhibited for the specific
values of the vehicle parameters and trajectory flown in this investigation. As seen
in Appendix B, this is due to the nature of the partial differentiation which renders
the sensitivities proportional to the time integrals of various parameters. A possible
explanation is that the increasing behavior of these integrals forces the sensitivities
of the roll convergence and dutch roll motions to exhibit rapidly increasing behavior
before the convergence of the true dynamics can take effect. However, the sensitivities of roll convergence and dutch roll motions with respect to the derivatives, LP, L,,
and Np all have an instant which their values become zero before they experience the
rapid increase and subsequent convergence. The sensitivities of the roll convergence
and dutch roll motions with respect to L, and Np are zero at approximately 600
seconds into the shuttle trajectory. The sensitivity of the motions to Lp is zero at
approximately 500 seconds into the trajectory. Although the effects of changing Lp
and Np on the lateral-directional reentry dynamics are minimal, there are moments in
the trajectory where such variations will have absolutely no effect on roll convergence
and dutch roll motions. This is significant for the stability derivative L" which has
relatively great effect on the dynamics throughout the trajectory except for this particular section where it becomes zero before increasing rapidly and then converging..

114

x10 8
2

1.8
1.6
1.4
1.2

S1
0.8
0.6
0.4
0.2
n

100

200

400

300

500

600

t(sec)

Figure 6-17: Sensitivity of Roll Convergence Dynamics to I-Y

115

700

x10l
U

-2
-4
-6
Oyp

-8

aYV

-10
-12
-14
-I

100

200

300

400

500

600

t(sec)

Figure 6-18: Sensitivity of Spiral Divergence Dynamics to Y,

116

700

x10 5
2.5
2
1.5
1
0.5

OYdrl

aYV
-0.5
-1
-1.5
-2

2I
0

100

200

400

300

500

600

t(sec)

Figure 6-19: Sensitivity of Sine-like Dutch Roll Dynamics to Y',

117

700

x10 5

PIC

OYdr2

49
Y

100

200

300

400

500

600

tFigure
6-20:
Sensitivity
of
Cosine-lik(sec)

Figure 6-20: Sensitivity of Cosine-like Dutch Roll Dynamics to Y,

118

700

x 100

4
By,
OL,

3
2
1
0
-1
2-

100

200

300

400

500

600

t(sec)

Figure 6-21: Sensitivity of Roll Convergence Dynamics to L,

119

700

xlOI"
1

0
-1
-2
oyp

6L,

-3
-4
-5
-6

100

200

400

300

500

t(sec)

Figure 6-22: Sensitivity of Spiral Divergence Dynamics to L,

120

600

700

x10 8
I

AO0
J.O

0.6
0.4
0.2
"Ydrl

OLV
-0.2
-0.4
-0.6
-0.8
.1

100

200

300

400

500

600

t(sec)

Figure 6-23: Sensitivity of Sine-like Dutch Roll Dynamics to L,

121

700

x 108
1
i

0.8
0.6
0.4
0.2
9Ydr2
/

aLV

8L,

-0.2
-0.4
-0.6
-0.8
_1

--

100

200

300

400

500

600

t(sec)

Figure 6-24: Sensitivity of Cosine-like Dutch Roll Dynamics to L,

122

700

x10 S
I ,

IL,

10

i y,

OLP
4
2
0
-9

100

200

300

400

500

600

t(sec)

Figure 6-25: Sensitivity of Roll Convergence Dynamics to Lp

123

700

x10 9

-2

Y--

OLP

-4

-6

-8

-in
0

100

200

400

300

500

600

t(sec)

Figure 6-26: Sensitivity of Spiral Divergence Dynamics to L,

124

700

ai0t

x10 5

6
4

-vwW
2
-

OLp
-2
-4

-6
8
-O

100

200

300

400

500

600

t(sec)

Figure 6-27: Sensitivity of Sine-like Dutch Roll Dynamics to Lp

125

700

x 1O
aY
6
4
2
Ydr2 0
OLP

~'JWVV~AMIWVVV\AMIWWWWVWV\

-2
-4
-6
8

100

200

300

400

500

600

t(sec)

Figure 6-28: Sensitivity of Cosine-like Dutch Roll Dynamics to Lp

126

700

x10 S

aYL,
Ur,

100

200

400

300

500

600

t(sec)

Figure 6-29: Sensitivity of Roll Convergence Dynamics to L,

127

700

x10 9
10
A

9
8
7
6
OdL,

5
4
3
2
1
0

100

200

300

400

500

600

t(sec)

Figure 6-30: Sensitivity of Spiral Divergence Dynamics to L,

128

700

x106
1

'.3

L
I

TI

0.5

9Ydrl

OL,.

\~u

-0.5

-1

1- 5

100

200

300

400

500

600

t(sec)

Figure 6-31: Sensitivity of Sine-like Dutch Roll Dynamics to L,

129

700

x10 6

II

0.
BYdr2

W-d'2

UVL,

-0.5

-1

-1.5
0
t(sec)

Figure 6-32: Sensitivity of


Cosine-like Dutch Roll Dynamics
to L,

130

xl10

0
-2
-4

-6
Oy,r

-8

-10

-12

-1A

100

200

300

400

500

600

t(sec)

Figure 6-33: Sensitivity of Roll Convergence Dynamics to N,

131

700

x10' 5
1'i

I0

10

Oy,,

M8Nv

6
4

100

200

300

400

500

600

t(sec)

Figure 6-34: Sensitivity of Spiral Divergence Dynamics to N,

132

700

x10' 2
L

1.5
1
0.5
Ydrl

ON

0
0

-0.5
-1
-1.5
-2

100

200

300

400

500

600

t(sec)

Figure 6-35: Sensitivity of Sine-like Dutch Roll Dynamics to N,

133

700

x10' 2

0.
0

U~dt2
Nr

-0.5
-1
-1.5
_"9
-1

4*W

500

600

t(sec)

Figure 6-36: Sensitivity of Cosine-like


Dutch Roll Dynamics to N,

134

700

x10 8
3

Oyrc

-p

-5

-10

-I,

100

200

400

300

500

600

t(sec)

Figure 6-37: Sensitivity of Roll Convergence Dynamics to Np

135

700

x 109
I ~

IL

10
8
6

ONp
4

2
0

100

200

400

300

500

600

t(sec)

Figure 6-38: Sensitivity of Spiral Divergence Dynamics to Np

136

700

x10 6
1

0.8
0.6
0.4
0.2
OYdrl

ONp
-0.2
-0.4
-0.6
-0.8
-i

100

200

400

300

500

600

t(sec)

Figure 6-39: Sensitivity of Sine-like Dutch Roll Dynamics to N,

137

700

x10 6

OYdr2

ONp

-%,.w

100

200

300

400

__

___

500

600

t(sec)

Figure 6-40: Sensitivity of Cosine-like Dutch Roll Dynamics to Np

138

___

700

x 109
-L

1.5
y,,rc

aN,
1

0.5

100

200

300

400

500

600

t(sec)

Figure 6-41: Sensitivity of Roll Convergence Dynamics to N,

139

700

x 10O
U

-0.2
-0.4
-0.6
-0.8
y, -1

ON,

-1.2
-1.4
-1.6
-1.8
-2,

100

200

300

400

500

600

t(sec)

Figure 6-42: Sensitivity of Spiral Divergence Dynamics to N,

140

700

x10 6

1
4ydr l

-N, O
-1

-2

-3
u

100

200

300

400

500

600

t(sec)

Figure 6-43: Sensitivity of Sine-like Dutch Roll Dynamics to N,.

141

700

x106

Ydr2

SN,

-1

-2

-'

100

200

300

400

500

600

t(sec)

Figure 6-44: Sensitivity of Cosine-like Dutch Roll Dynamics to N,

142

700

Chapter 7
Handling Qualities
7.1

Overview

The issue of handling qualities with regard to the GHAME vehicle is considered in
this chapter. First, the handling qualities of the GHAME vehicle during reentry along
the optimal Space Shuttle trajectory are determined. This is accomplished by comparing the reentry values of certain vehicle parameters with specifications designed
to differentiate levels of handling qualities. The specifications are taken from a set of
military standards for flying qualities of piloted vehicles. Also, GMS theory is employed to briefly consider the handling qualities of a generic second order time-varying
system. By comparing the responses of systems with differing characteristic root behavior, general conclusions are drawn regarding the relationship between system root
movement and handling qualities.

7.2

GHAME Vehicle Reentry Handling Qualities

The handling qualities of the GHAME vehicle during reentry along the Shuttle trajectory are studied in this section. The level of handling quality is determined by
comparing GHAME vehicle reentry parameters to a set of military specifications.
These specifications are taken from 'Military Specification: Flying Qualities of Piloted Airplanes'[13] which contains the handling quality requirements of all United
143

States military aircraft. For the purposes of this military document, the GHAME
vehicle is classified as Class III aircraft which is described as a large, heavy, lowto-medium maneuverability aircraft. Also, the optimal Shuttle reentry trajectory is
assumed to be of flight phase category B. This part of a mission is described as a phase
normally accomplished using gradual maneuvers without precision tracking although
accurate flight path control may be required.
The specifications to meet flying quality requirements are presented such that the
handling qualities are separated into three levels. Level 1 represents flying qualities
which are clearly adequate for accomplishing a particular flight phase. If the handling
qualities allow the completion of a flight phase but only after a significant increase
in pilot workload, then they are Level 2. Finally, Level 3 flying qualities represent
dynamics which allow the vehicle to be controlled safely but only with excessive pilot
workload. In terms of a well known subjective rating system, Level 1 corresponds to
Cooper-Harper ratings of 1 through 3, while Level 2 represents Cooper-Harper ratings
of 4 through 6. A Cooper-Harper rating between 6 and 9 corresponds to handling
quality Level 3.[13, 14] Requirements necessary to be classified into one of these
levels are made on each of the longitudinal and lateral-directional modes. As seen in
Chapter 5, the behavior of the GHAME vehicle phugoid mode during reentry is quite
abnormal, and it is not considered in this section. The handling quality requirements
for a Class III vehicle in the Category B flight phase are shown in Tables 7.1 through
7.3.[13]
The damping ratios and the natural frequencies of the dutch roll and short period
modes of the GHAME vehicle during reentry are calculated from their respective
characteristic roots shown in Figs.5-8 and 6-8. Since, the roots of these modes vary
along the reentry trajectory, the damping ratios and natural frequencies are functions
of time. If the roots are defined as k = kI,+ ikl,

then the damping ratio and natural

frequency are given by

c =-k,/(kr

+ k2 )

Wn = (k,2 + kZ )

144

(7.1)
(7.2)

Min. (
Level 1
0.30
Level 2
0.20
Level 3
0.15

Max. (
2.00
2.00
-

Table 7.1: Short Period Requirements for Handling Qualities

Level 1

Min.
0.08

Level 2
Level 3

0.02
0

Min. (w,,
0.15

Min. w,
0.40

0.10
-

0.40
0.40

Table 7.2: Dutch Roll Requirements for Handling Qualities

Roll Convergence
Max. time constant

Spiral Divergence
Min. time to double amplitude

Level 1

1.4 sec.

20.0 sec.

Level 2

3.0 sec.

8.0 sec.

Level 3

10.0 sec.

4.0 sec.

Table 7.3: Roll Convergence and Spiral Divergence Requirements for Handling Qualities

145

The natural frequency of the short period mode is plotted against its damping ratio
in Fig.7-1. Likewise, Fig.7-2 shows the dutch roll natural frequency plotted against
its damping ratio. It is interesting to note that except scaling,the behavior seen in
these two plots is quite similar. This is to be expected since the behavior of the roots
seen in Figs.5-8 and 6-8 also exhibit similar behaviors.
The information shown in Figs.7-1 and 7-2 is compared to the handling quality
requirements of Tables 7-1 through 7-3 to determine whether the short period and
dutch roll modes of the GHAME vehicle insure adequate handling qualities during
reentry. Upon analysis, it is clear that the short period behavior is quite inadequate in
terms of handling qualities. It is not until the vehicle is at the end of the trajectory
that the short period damping ratio satisfies minimum Level 3 requirements. The
dutch roll reentry behavior also has poor implications on the handling qualities of
the GHAME vehicle. Fig.7-2 shows although dutch roll natural frequency by itself
satisfies Level 1 requirements, C and Cw,, only satisfy Level 3 specifications. Therefore,
dutch roll behavior induces a handling quality rating of Level 3 during reentry. It is
obvious that neither of the GHAME vehicle's two oscillatory modes induce adequate
handling qualities in the aircraft during reentry. At Level 3, the dutch roll behavior
allows the vehicle to be controlled safely, but only after excessive workload on the
pilot. The short period reentry behavior does not even qualify for Level 3 status
and renders the GHAME vehicle uncontrollable during reentry along the Shuttle
trajectory.
The handling quality specifications for the roll convergence and spiral divergence
modes are expressed in terms of time parameters concerning the amplitude of their respective responses. The roll convergence requirement is placed on the time constant
which is defined as the time required for the amplitude of a response to decay to
ezp(-1) times its original value. From the characteristic response shown in Fig.6-11,
the time constant of the GHAME vehicle's roll convergence mode is approximately
750 seconds. This does not even satisfy the handling qualities requirements for Level
3. The roll convergence behavior of the GHAME vehicle also renders it uncontrollable
during reentry. The spiral divergence mode, however, does exhibit favorable behavior
146

S-11

0.45
E

0.4

0.35
Wn

t=1900.

0.3

0.25

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

Figure 7-1: Short Period Natural Frequency vs. Damping Ratio Along Trajectory

147

0.95
0.9
0.85
0.8
W07

0.75
0.7
0.65
0.6
n 44

0.005

0.01

0.015

0.02

0.025

Figure 7-2: Dutch Roll Natural Frequency vs. Damping Ratio Along Trajectory

148

with regard to handling qualities. Fig.6-12 shows that the time for the spiral divergence characteristic response to double its amplitude is approximately 300 seconds.
This clearly satisfies the Level 1 handling qualities requirement detailed in Table 73. Although the spiral divergence behavior is favorable to handling qualities, the
other three modes only represent handling quality ratings of Level 3 or poorer. The
GHAME vehicle is clearly uncontrollable during reentry along the Shuttle trajectory.

7.3

Handling Qualities of a Generic Second Order


System

In this section, the handling qualities and of a generic time-varying second order
system is studied in order to determine the general effects of varying characteristic
root behaviors. The responses of the systems are approximated using the second
order GMS solutions developed in section 3.2. A generic time-varying second order
system can be expressed by the equation

+ wi(t)y + wo(t)y = 0

(7.3)

The response of this system is then governed by the roots of the algebraic equation

s2 + W18 +

0o=

(7.4)

If the system is autonomous, then the two coefficient of Eq.7.3 are constant and the
response is described by a pair of roots which remain fixed with time. However, if
the coefficients wl and w0 vary with time, than the characteristic roots do not remain
stationary. The path and speed of the characteristic roots in the complex plane are
determined by the nature of the two coefficients. The effects of simple variations in
the path and speed of the roots on the response and handling qualities is examined
in this section. In order to accomplish this, systems having the same characteristic
roots at the initial and final time are considered. The path and speed at which the

149

roots get from the initial point to the final point are varied and the their responses
are compared.
In order to gain insight with regards to handling qualities, the initial and final
points of the characteristic roots for the systems to be considered are chosen such that
they represent two different levels of handling quality ratings. If the characteristic
roots are a pair of complex conjugates in the form k = k, + ki, then they are set such
that
k,(0) = -. 054

ki(0) = .8984

(7.5)

k,(T) = -. 225

ki(T) = 1.483

(7.6)

where T is the total time the roots of the system take to travel from the initial point
to their final position. It follows from Eq.7.1 and 7.2 that
wn(0) = .9 rads/sec
w,(T) = 1.5 rads/sec

C(0) = .06

(7.7)

((T) = .15

(7.8)

If it is assumed that the generic second order system represents a dutch roll mode,
then upon comparing the above natural frequency and damping ratio values to Table
7-2, the initial point is of Level 2 handling qualities. Similarly, the system exhibits
Level 1 handling qualities at the final time. The roots are now allowed to move from
the Level 2 point to the Level 1 point in varying ways. The responses of these differing
root behaviors are compared to each other as well as to those of two constant systems
where the roots remain fixed at the Level 1 and Level 2 points.
Initially, only systems with roots moving from the Level 2 point to the Level 1
point in a straight line with constant speed are considered. The straight line root
trajectory of these systems is shown in Fig.7-3. The total time allowed for the roots to
travel from initial to final point is varied in order to determine how the speed of root
movement affects the response. The specifications of the systems considered as well
as the two constant systems are shown in Table 7.4. The solutions to the systems
in Table 7.4 are approximated using second order GMS theory and are plotted in
150

Root
Trajectory

I
Case
Case
Case
Case
Case
Case
Case
Case
Case

1
2
3
4
5
6
7
8
9

Straight
Straight
Straight
Straight
Straight
Straight
Straight

Line
Line
Line
Line
Line
Line
Line

Initial Root
Position
Level 1
Level 2
Level 2
Level 2
Level 2
Level 2
Level 2
Level 2
Level 2

Final Root
Position
Level 1
Level 2
Level 1
Level 1
Level 1
Level 1
Level 1
Level 1
Level 1

Total Trajectory
Time, T

Root
Speed

800 sec.
250 sec.
125 sec.
50 sec.
25 sec.
5 sec.
1 sec.

Const.
Const.
Const.
Const.
Const.
Const.
Const.

Table 7.4: Straight Root Trajectory Systems


Figs.7-4 through 7-6. It is evident from Fig.7-4 that as the root speed between initial
and final points increases, the frequency and the settling time of the response both
decrease. As the speed of the roots is increased further, Fig. 7-5 shows that the
responses of the systems approach the solution to the constant system with roots
fixed at the Level 1 point. Conversely, as the speed of the roots is decreased greatly,
the responses of the systems approach the Level 2 constant system as seen in Fig.76. In general, as the speed of the roots increases from slow to fast, the behavior of
the response changes from exhibiting similar behavior to the solution of the system
with fixed roots at the initial point to the solution of the system with root, contant
at the final point. This implies that if a second order time-varying system possesses
characteristic roots which travel in a straight line at constant speed, then the handling
qualities of the system fall in between the behavior of the systems with fixed roots at
the initial and final positions. No matter how fast or slow the roots travel between
the two points shown in Fig.7-3, the resulting behavior never exceeds the handling
quality ratings of Case 1 and 2.
Three different systems are now considered whose characteristic roots do not necessarily travel between the initial and final points in a straight line or with constant
speed. The specifications of these three systems are shown in Table 7.5. The acceleration of the roots along the straight line trajectory of Case 10 is detailed in Fig.7-7.

151

Initial Root

Final Root

Total Trajectory

Root

Trajectory
Case 10 Straight Line

Position
Level 2

Position
Level 1

Time, T
50 sec.

Speed
Accel.

Case 11

See Fig.7-8

Level 2

Level 1

50 sec.

Const.

Case 12

See Fig.7-8

Level 2

Level 1

50 sec.

Const.

Root

Table 7.5: Miscellaneous Systems


The response of this accelerating system is compared to the Case 5 response in Fig.7-8
because they initially share the same root velocity. The non-straight line trajectories
of Case 11 and 12 are shown in Figs. 7-9. These two curved trajectory systems are
compared to the Case 6 system due to the fact that they possess equal trajectory
times. Their responses are shown in Fig.7-10.
Fig.7-8 shows that the response of the Case 10 system initially mirrors the response
of Case 5, but later exhibits greater frequency and damping. Upon examination of
the second order GMS approximations detailed in Chapter 2, this behavior becomes
clear. Initially, the roots of both systems travel with the same velocity along the same
path, and therefore, they experience the same response. Later, when the roots of the
Case 10 system have accelerated to greater speed, they experience smaller values of k,
and greater values of ki than those of the Case 5 system. Hence, the response exhibits
higher frequency and greater damping. It is evident that even if the roots are allowed
to accelerate or decelerate along the straight line trajectory, the responses of such
systems still falls in between the behavior exhibited by the Case 1 and 2 constant
systems.
The responses of the two systems with the curved root trajectories shown in Fig.7-9
also behave as would be expected from the nature of the GMS approximations. Although they possess equal trajectory time, Fig.7-9 shows that the characteristic roots
of the Case 11 system experience greater values of ki and smaller values of k, than
those of the Case 6 system. Hence, the Case 11 response exhibits higher frequency,
but less damping than the Case 6 response.

Similar arguments can be employed

to explain the response of the Case 12 system which exhibits smaller frequency and

152

greater damping than the Case 6 response. It is concluded upon examination of the
GMS equations that no matter what the speed of the characteristic roots, as long as
their path does not stray beyond the rectangular boundaries defined by their initial
and final points, the resulting handling qualities behavior falls in between those of
the constant systems with roots fixed at initial and final points.

153

:tT

1.5
1-

two

0.5
. ...............................

.................................

................... ............... ................................. i. .............................

.................................

..................

0o
.-...............................

...............

,.................................

...............................

-0.5
................................
.
................... ........
.....
....
-.....
lbi ...
......

........... .....................................................................................................
-1 .- ...................
" " " " " " "
-I

-1.5
"

-0.25

I"""

-0.2

-0.15

-0.1

k,

Figure 7-3: Straight Line Root Trajectory

154

-0.05

0.8
0.6
0.4
0.2
y

0
-0.2
-0.4
-0.6
-0.8
-1

10

20

40

30

50

t(sec)

Figure 7-4: GMS Solutions to Cases 5,6,7

155

0)

70

80

0.8
0.6
0.4
0.2
Y 0
-0.2
-0.4
-0.6
-0.8
1

10

15

20

25

30

35

t(sec)

Figure 7-5: GMS Solutions to Cases 1,8,9

156

40

45

50

10

15

20

25

30

135

t(sec)

Figure 7-6: GMS Solutions to Cases 2,3,4

157

111

4U

11

111

3U

...............
50 ...
..
.................................
i..
..
..
..
....
........
..........
...
...........
.....................
i............
...........
.........
-

1.5

50

S-30!
1
0.5

-t=2 :20

................................

0
-0.5

.........................................
...........................
...- .

!-

-0.25

-0.2

-0.15

-0.1

-0.05

Figure 7-7: Straight Line Root Trajectory of Accelerating System

158

10

15

20

25

30

35

t(sec)

Figure 7-8: GMS Solutions to Cases 5,10

159

40

45

50

1.5
tNSOO

0.5
ki

. ............

....

............ ...................

.................................

...................

.....................

.................................

. . ...............................

.................................

...............................

..........

...............................

Case 12

........

.................................

.................................

....

.............

..............................

........

.-

o ...............................

.~
~ ~.................
...............
.............................
-0.5
-1
-1.5
I

-0.25

-0.2

-0.15

-0.1

-0.05

k,

Figure 7-9: Non-Straight Line Root Trajectories of Cases 11,12

160

10

15

20

25

30

35

t(sec)

Figure 7-10: GMS Solutions to Cases 6,11,12

161

40

45

50

Chapter 8
Summary and Conclusions
8.1

Summary and Conclusions

In this work, the reentry dynamics and handling qualities of the Generic Hypersonic
Aerodynamic Model Example vehicle were examined along a Space Shuttle optimal
trajectory. The analysis of both the longitudinal and lateral-directional reentry motions was conducted through asymptotic approximations provided by Generalized
Multiple Scales theory developed by Ramnath.
The reentry angle-of-attack perturbations of the vehicle were predicted by applying second order GMS solutions to the linear differential equation developed by
Vinh and Laitone. GMS approximations showed that the dynamics of the GHAME
vehicle angle-of-attack perturbations along the Shuttle optimal trajectory behave as
damped oscillations with increasing frequency. Comparison with numerical integration approximations showed the GMS solutions to be of considerable accuracy. The
fast part of the GMS solution predicted the frequency variations of the true dynamics
extremely well while the amplitude changes were incorporated by the slow solution.
The GMS approximations to the angle-of-attack perturbations were also compared
with constant coefficient solutions applied after freezing the system at its initial point.
Although such 'frozen' approximations are employed often in engineering analysis, it
was shown that such methods totally misrepresent the angle-of-attack perturbations
after only half a cycle of the dynamics.
162

The sensitivity of the angle-of-attack perturbations to the aerodynamic coefficients CLa, CGa, and Cm, was studied through partial differentiation of the GMS
solutions. It was seen that the sensitivity of the dynamics to these three parameters
oscillate with the same frequency behavior of the actual dynamics. Changes in the
three aerodynamic coefficients have the greatest effect on angle-of-attack perturbations at approximately 50,000 to 60,000 vehicle lengths into the reentry trajectory.
The parameter C,, was shown to have much greater influence on the second order longitudinal dynamics than either of the other two parameters. GHAME vehicle
angle-of-attack perturbations were found to be 2000 times more sensitive to variations
in C,, than to C,, and 500 times more sensitive than to changes in CL,.

Finally,

the second order longitudinal dynamics was investigated using a stability criterion.
The dynamics were shown to be closest to instability at approximately the same time
that the vehicle was most sensitive to changes in the aerodynamic coefficients.
The full fourth order longitudinal dynamics of the GHAME vehicle along the
Shuttle reentry trajectory were examined after deriving a fourth order time-varying
differential equation from the linearized equations of motion. The longitudinal stability derivatives were estimated along the reentry trajectory for use in the fourth
order GMS approximations. It was seen that the stability derivatives of the GHAME
vehicle exhibit the same behavior as those of conventional aircraft with regards to
sign convention. The angle-of-attack stability term, Ma remained negative throughout the entire trajectory to insure static stability of the longitudinal dynamics. It was
also seen that values of the derivative M,, were extremely small as desired for design
purposes. Mv was found to be statically most destablizing at approximately 60,000
vehicle lengths into the trajectory. Since the parameter remained negative throughout
the entire reentry, it had no negative effects on the dynamic stability. The stability
derivatives were employed to determine the characteristic roots associated with each
of the longitudinal modes of motion. The characteristic roots of the phugoid mode
were found to contain several 'turning' points where the roots changed from being
complex to real pairs. Such a behavior represents deep mathematical difficulties was
considered beyond the current scope to predict phugoid mode motions.
163

GMS solutions to the short period mode showed its reentry behavior to be one
of damped oscillations. Unlike the angle-of-attack perturbations, there were no frequency variations of significance. Also, the GMS fast solution to short period motions
were virtually identical with the full GMS approximation. Once again, the GMS approximation was compared to numerical solutions in order to determine accuracy.
Although the asymptotic approximations slightly over shot the amplitude of short
period dynamics, they were still of considerable accuracy. Due to the behavior of
the phugoid mode roots, a complete analysis of the full GHAME vehicle fourth order
longitudinal dynamics could not be accomplished with GMS theory. However, the
accuracy of the asymptotic method was again demonstrated in predicting the reentry
short period behavior of the vehicle.
In order to study the lateral-directional reentry dynamics of the GHAME vehicle,
once again, a fourth order time-varying linear differential equation was derived form
the equation of motions. The lateral-directional stability derivatives were estimated
along the reentry trajectory employing the equations in Appendix A. With the exception of the dihederal derivative LV, the lateral-directional stability derivatives of the
GHAME vehicle exhibited the same sign behavior as those of conventional aircraft.
Since, the GHAME vehicle has no dihederal in the main wings, the values of L, were
relatively small. Again with the exception of L,, all of the derivatives increased their
effect on the vehicle dynamics as the GHAME vehicle progressed further into the
atmosphere.
The characteristic roots associated with each of the vehicles modes of motion
were found and the GMS approximations were employed to predict the dynamics
associated with each of these modes. The the roll convergence mode was found to
be stable, while the spiral divergence mode exhibited unstable behavior. The characteristic motions of the dutch roll mode were shown to be damped oscillations with
little damping and virtually no variations in the frequency. Once again, there was
little difference between the GMS fast solution and the full GMS approximations to
dutch roll motions. The asymptotic solutions to each of the mode dynamics were
linearly combined for two sets of different initial conditions to obtain approximations
164

to the complete lateral-directional reentry motions. Due to the instability of the spiral divergence mode, the full lateral-directional dynamics were found to be unstable
regardless of the initial conditions. Once again comparisons between the GMS solutions and numerical solutions showed the asymptotic approximations to be of great
accuracy.
The sensitivity of the reentry lateral-directional dynamics to each of the stability
derivatives was determined through partial differentiation of the the GMS approximations to each of the mode motions. Spiral mode dynamics were found to be most
sensitive to changes in any of the stability derivatives. Roll convergence motions were
almost as sensitive, however, effects of variations in the stability derivatives were several orders of magnitude less for dutch roll dynamics than either of the other two
modes. Reentry lateral-directional dynamics which contain proportionally greater
amounts of the spiral divergence mode will experience greater sensitivity to the stability derivatives. Also, it was seen that the directional derivative N, has the most
effect on all modal motions. Changes in N, affected the roll convergence, spiral divergence, and dutch roll dynamics more than any of the other derivatives. The sensitivity
of motions to L, were comparable to those of N, , however, effects of changing the
other stability derivatives on the dynamics were orders of magnitude less. Especially
the parameter L, which remained relatively very small for a significant part of the
trajectory. The sensitivities of spiral mode motions with respect to the different stability derivatives grew unbounded due to the behavior of the actual dynamics. The
roll convergence and dutch roll sensitivities exhibited rapidly increasing behavior at
approximately 500 seconds into the trajectory. It was seen that the sensitivity of roll
convergence and dutch roll motions to the derivatives LP, L,, and Np became zero
shortly before this rapid increase and subsequent convergence.
Finally, the handling qualities of the GHAME vehicle along the Shuttle trajectory
were examined by comparing reentry parameters with a set of flying quality specifications. It was found that the short period handling qualities do not qualify for Level
3 status, and render the GHAME vehicle uncontrollable. The dutch roll mode exhibited handling qualities of Level 3 which represents controllable behavior but only after
165

excessive workload on the flight crew. The GMS approximation to the roll convergence motion was employed to determine that the time constant of its response also
did not meet Level 3 specifications. Only the spiral divergence reentry behavior was
found to represent satisfactory handling qualities. In general, the handling qualities
of the GHAME vehicle along the Shuttle reentry trajectory were found to be clearly
inadequate.
The way in which variations in the characteristic root behavior affect the handling
qualities was studied by considering a generic second order system. Through the
nature of the GMS solutions, it was determined that as long as the path of the
characteristic roots do not go beyond the rectangle defined by its initial and final
points, the handling quality behavior falls in between those of the constant systems
with roots fixed at the initial and final points. Also, it was found that the faster
the roots of a system roots move, the handling quality behavior approaches that of
the constant system with fixed roots at the final point. become more Conversely,
the slower the characteristic roots travel along a path, the behavior of the system
approaches that of the constant system defined by the initial point. As long as the
path of roots does not extend beyond the rectangle defined by the initial and final
root positions, the handling quality behavior of the system falls in between those of
the constant system defined by these points.

8.2

Suggestions for Further Work

The following is suggested for further work regarding the GHAME vehicle and GMS
theory:
* The reentry dynamics of the GHAME vehicle be studied for a trajectory which
is designed for an aircraft closer to configuration of the GHAME vehicle.
* GMS control theory be utilized to stabilize the lateral-directional reentry dynamics and improve the handling qualities of the GHAME vehicle.

166

* Human factors issues be considered in determining the optimal manner of presenting a flight crew with the stability information in Chapter 2.
* Further theoretical investigation of the GMS method be conducted to better
determine when only the 'fast' solution is adequate for accurate approximation
of solutions to linear differential equations.
* The implications of characteristic root behavior on the response of a system be
studied further. This may lead to the use of GMS theory in the design process
of systems which exhibit non-autonomous behavior.

167

Appendix A
Stability Derivative Estimation
A.1

Longitudinal Stability Derivatives

The following equation are used to estimate the longitudinal stability derivatives of
the GHAME vehicle along the Shuttle reentry trajectory.[9, 10, 11]
(A.1)

reA

Dv

2g
CL

VCz

(CD

Mn OCD
2 OMno

La/V pVSL
2m

Ly / Vo a2CL
V2 CL
Ma

2gc
VCL k

(A.3)

Mo OCL )

(CL

2 OMnO

gcCM

CL kma

(m
168

(A.2)

M, O8Cm
2 OM.,o

(A.4)

(A.5)

(A.6)

gc 2

A.2

2VCLk2 Cm
2VCLCk,

(A.7)

Lateral-Directional Stability Derivatives

The following equations are employed to approximate the GHAME vehicle lateraldirectional stability derivatives along the Shuttle reentry trajectory.[9, 10, 11]
(A.8)

CY

L,9

gb

V C---

L,

(A.9)

L,. , 2VCk C.

(A.10)

LP M -2V C k C1 ,
2VCLk
'

(A.11)

gb
N,,

N, M

Np

V CLk CNO

(A.12)

CN,

(A.13)

CN,

(A.14)
(A.14)

2VCLk2
22
9b

20CLkz

169

Appendix B
Fourth Order Lateral-Directional
Sensitivity Differentiation
The differentiation is carried out for N, in this section, however, the general process
is the same for the other stability derivatives. We start with the equation
84 + wa;s

+ W2s 2 + Is + wo = 0

(B.1)

where the coefficients are as defined in Eqs6.12 through 6.15. This equation which
defines the roots describing the lateral-directional motion is rewritten as[12]
x3 +pzX

+ qx + r = 0

(B.2)

where
P = -W

q = w3 W2 - 4wo

r = 4w2 Wo

2
WWoO- oW

(B.3)
(B.4)

(B.5)

Now, let
a = 3(3q - p2 )

170

(B.6)

b=-(2p'

- 9pq + 27r)

27

(B.7)

b2b + 4 + T7

+ (

B -

(B.8)

+ a

(B.9)

4 F7

then one root of Eq.B.2 is given by

S= A + B - p/3

(B.10)

Now let,

R= (2
D=

(3

w2 +

(B.11)

4w3w2 - 84

4R

D=4w
E=

-- w +

+22

- w3)

- W1 -

(B.12)

(B.13)

4R

Then, the roots representing the three lateral-directional modes of motion are given
by
w3

k,, -

R
+ -2 + T

kdr, -

8,

(B.15)

R
2

We will now concentrate on finding

D
2

w3 R
4 + 22

kP -

(B.16)
(B.17)

22

but again the same procedure can be followed

for the other roots. We can write that

ON,

Okr,, Owo +
Owo O8N,

ON, + Arc ON,

Owl1 ON,,

171

8Ow ON,

Okr Ow3

8w3 ON,,

(B.18)

Differentiating Eqs.6.12 through 6.15 with respect to N,, we get

Oaw$ = 0

(B.19)

ONv

8w2
O2

ON,

(B.20)

Ow = -vL

(B.21)

O0= -L,

(B.22)

ON,

These four equations are in terms of parameters which are known for the shuttle
trajectory and can be directly substituted into Eq.B.18. We must now determine the
partial derivative of k,, with respect to the w's. Differentiating Eq.B.14, we get

k
Own

= 0,1,2

1 OR

6Okrc,,

Ow3

(B.23)

2 Ow,

2 Ow,

20w 3

1 OD

20w

(B.24)

In order to evaluate the above two expressions, we are now left to determine the
partial derivatives of D and R with respect to the w's. Differentiating Eq.B.11 leaves
BR
OR = 1R-1 (z
Own 2
k8w4

OR

Ow3

1+

-1

R-

n = 0,1

Be

(B.25)

(B.26)

+ O

-s

(B.27)

O83

Also differentiating Eq.B.12, we get

OD O 1- D

Owo
OD

D 1 -2R

1[-2R -

R
SOwl2

2
R

1 (4w

4R 2

-8

- W

OR
o
0

1
OR
(4wos - 8wl - W )
8
4Ra
172

(B.28)

(B.29)

8D

Ow2

1D-1 [ 2R OR W3
2
- Ow2 + R

__D

11
OD D-

1(4sw - 8wl - w
4R 2
)8-2

r_2R 0ROR +

1
1-(4w2 - 3w2)-

1
dR
(4wsw2- 8w - w 3)2
4R
3

(B30)

(B.31)

3
+ -w
s3

Eqs.B.25 through B.31 are substituted into Eqs.B.23 and B.24 which are in turn substituted into Eq.B.18. The expression for 8N,
k is now in terms of trajectory parameters
which are all known except for the partial derivatives of z with respect to the w's.
We differentiate Eq.B.10 which results in
8e S- BA
O + 8B

1 8p

Own

3Own

Ow,

Own

n = 0,1, 2, 3

(B.32)

In order to determine the above expression, we finally differentiate Eqs.B.3 through


B.9 with respect to the w's. This results in the following.
O

= 0

n = 0,1,3

(B.33)

(B.34)

Ow2
q

-4

(B.35)

W3

(B.36)

oq 0

(B.37)

OWi

Ow2
q= W

(B.38)

OW3

Or
Owr
Or

FWl

-w

+ 4W2

- -2wo
173

(B.39)
(B.40)

Or
Or

Ow3

Ob

22

Own

8A

4wo

(B.41)

-2w 3w0

(B.42)

Oq
Oa
---Ow,
Owe

2 Op
p

1 q 8p
3Ow,

2q
3Own

A-2
3

3 Ow,

Or

a
4

Own

-2
SB
54

27

n = 0,1,2,3

OW,

Ob

27On

.- a2

(B.43)

(B.44)

(B.45)

1 1A-2+1 b(b + 3)-

OB

= 0, 1,2, 3

o,n

n = 0, 1,2, 3

(B.46)

1 1 (b2 3 ']
4T 27
Own

n = 0,1,2, 3

We substitute the above equations into Eq.B.32 then subsequently substitute the
resulting expression into Eq.B.18.

This finally results in an expression of '4c


8N, in

terms of the w's which are known along the trajectory. Eq.B.18 is calculated in this
way and then substituted into the final sensitivity equation shown in Eq.6.

174

Bibliography
[1] Bowers, Albion and Iliff, Kenneth ,"A Generic Hypersonic Aerodynamic Model
Example(GHAME) for Computer Simulation", Dryden Flight Research Facility,
August 5, 1988
[2] Deyst, J., Kriegsman, B., and Marcus, F., "Entry- Trajectory Design to Minimize Thermal-Protection-System Weight", Charles Stark Draper Laboratory,
November, 1971
[3] Ramnath, R. V. and Sinha, P., "Dynamics of the Space Shuttle during Entry
into Earth's Atmosphere", AIAA Journal, Vol. 13, No. 3, March, 1975
[4] Vinh, N. X. and Laitone, E. V., "Longitudinal Dymnamic Stability of a Shutttle
Vehicle", Journal of Aeronautial Sciences, Vol.XIX, No. 5, March, 1972
[5] Allen, H. J., "Motion of a Ballistic Missle Angularly Misaligned with the Flight
Path Upon Entering the Atmosphere and its Effect Upon Aerodynamic Heating,
Aerodynamic Loads, and Miss Distance", TN 4048, October, 1957, NACA
[6] Etkin, B., "Longitudinal Dynamics of Lifting Vehicle in Orbital Flight", Journal
of Aeronautical Sciences, Vol. 28, October, 1961
[7] Ramnath, R. V. and Sandri, G., 'A Generalized Multiple Scales Approach to a
Class of Linear Differential Equations", Journal of Mathematical Analysis and
Application", Vol. 28, 1969
[8] Ramnath, R. V., "A Multiple Time Scales Approach to a Class of Linear Systems", Rept. AFFDL-TR-68-60, October, 1968
175

[9] Ramnath, R.V., Flight Dynamics and Control class notes, Fall, 1990
[10] Etkin, B., Dynamics of Atmospheric Flight, Wiley, New York, 1972
[11] Etkin, B., Dynamics of Flight: Stability and Control, Wiley, New York, 1982
[12] Beyer, W., CRC Standard Mathematical Tables, CRC Press, Boca Raton,
Florida, 1979
[13] "Military Specification: Flying Qualities of Piloted Airplanes", MIL-F-87585C,
November 5, 1980
[14] Cooper, G. and Harper, R, Jr., "The Use of Pilot Ratings in Evaluation of
Aircraft Handling Qualities", NASA TN D-5153, April, 1969
[15] Ramnath, R.V., "Presentation to NASA Dryden Flight Research Facility",
1989,1990, and April, 1991
[16] Ramnath, R.V. and Radovsky, S., "Sensitivity Analysis of Variable Systems",
Proc. JACC, Denver, 1978

176

You might also like