Frisbee Aerodynamics

Download as pdf or txt
Download as pdf or txt
You are on page 1of 220
At a glance
Powered by AI
The document discusses the aerodynamics of disc-wing aircraft and frisbees through wind tunnel experiments, computational simulations, and analysis of historical designs.

It discusses the theoretical background of disc-wing aerodynamics and reviews previous literature on the topic. It then describes the experimental facilities and techniques used to study scale models.

The Goldstein Laboratory, Project Tunnel, and 9' x 7' wind tunnels were used to collect load data, surface pressure measurements, and perform flow visualization.

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/268559957

Frisbee(TM) Aerodynamics

Conference Paper · June 2002


DOI: 10.2514/6.2002-3150

CITATIONS READS

9 5,092

2 authors:

Jonny Potts William James Crowther


Sheffield Hallam University The University of Manchester
18 PUBLICATIONS   62 CITATIONS    88 PUBLICATIONS   787 CITATIONS   

SEE PROFILE SEE PROFILE

All content following this page was uploaded by William James Crowther on 16 April 2015.

The user has requested enhancement of the downloaded file.


Disc-wing Aerodynamics

A thesis submitted to The University of Manchester for the degree of


Doctor of Philosophy
in the Faculty of Engineering & Physical Sciences.

2005

Jonathan Potts
School of Mechanical, Aerospace & Civil Engineering

-1-
Contents Page

1
Title Page
2
Contents
6
List of Figures
19
Abstract
21
Declaration & Copyright
22
Acknowledgements
24
The Author
25
Nomenclature

Chapter 1 Introduction 28

1.1 Opening Statements 28

1.2 Definitions 28

1.3 Thesis Aims & Objectives 29

1.4 Thesis Outline 30

Chapter 2 Theoretical Background 31

2.1 Disc-wing Geometric Parameters 31

2.2 Aerodynamic Coefficients 32

2.3 Lift & Drag Characteristics of Low Aspect Ratio Wings 34

2.4 Trim & Stability of Tailless Flight Vehicles 36

2.5 Flight Dynamics of a Spin-stabilised Disc-wing 39

2.5.1 Non-dimensional Roll Time Constant 41

Chapter 3 Literature Review 44

3.1 Opening Statements 44

3.2 Aerodynamics of the Circular Planform 44

3.2.1 Frisbee 44

3.2.2 Discus 47

3.2.3 Coin-like Cylinders 48

3.2.4 Analysis on Span Effectiveness Factor 49

3.3 Flight Dynamics of Spin-stabilised Disc-wings 50

3.3.1 Frisbee 50

-2-
3.3.2 Wind Tunnels 55

Chapter 4 Experimental Facilities & Techniques 57


4.1 Wind Tunnels 57
4.1.1 Goldstein Laboratory 57
4.1.2 Project Tunnel 57
4.1.3 9’ × 7’ Tunnel 57
4.2 Apparatus 57
4.2.1 Rigs 58
4.2.2 Disc-wing Models 58
4.3 Data Acquisition 59
4.3.1 Load data 59
4.3.2 Surface Pressure Data 59
4.3.3 Flow Visualisation 60
4.3.3.1 Surface paint 60
4.3.3.2 Smoke wire 60
4.4 Wind Tunnel Corrections 61

Chapter 5 Experimental Results 63


5.1 Load Data 63
5.1.1 Non-spinning Case 63
5.1.1.1 Baseline force & moment data 63
5.1.1.2 Reynolds number effects 64
5.1.1.3 Comparison with theory 64
5.1.1.4 Comparison with data from the literature 65
5.1.2 Spinning Disc 68
5.1.3 Basic Geometry Comparisons 69
5.1.3.1 Camber 69
5.1.3.2 Cavity 70
5.2 Surface Pressure Data 70
5.2.1 Overview 70
5.2.2 Discussion of Results at Typical Flight AoA (5o) 71
5.2.3 Effect of AoA on Pressure Distributions 72
5.2.4 Effect of Reynolds number 74

-3-
5.2.5 Loads Comparison: Integrated Pressure Data 75
5.3 Surface Flow Visualisation 76
5.3.1 Surface Flow Patterns 76
5.3.1.1 Typical flight angle of attack (5o) 76
5.3.1.2 Range of attack angles (0o to 30o) 77
5.3.2 Comparison with Pressure Data 79
5.4 Smoke Wire Flow Visualisation 81
5.4.1 Centre-line Visualisation 81
5.4.2 Wake Visualisation 82
5.4.3 Visualisation of the Effect of Spin 83

Chapter 6 Flow Topology 86


6.1 Typical Flight Angle of Attack (10o) 86
o o
6.2 Range of Attack Angles (0 to 45 ) 87

Chapter 7 Disc-wing Flightpath Simulation 89


7.1 Flight Equations of Motion 89
7.2 Simulation Implementation 93
7.3 Results from Numerical Experiments 94
7.3.1 Comparison with Experimental Data 94
7.3.2 Analysis of a Typical Frisbee Trajectory 94
7.3.3 Effect of Varying Launch Condition 96
7.3.4 Frisbee Flight Control 97

Chapter 8 Conclusions 99

References 102
Figures 113

Appendices 208
Appendix A History of the Frisbee 209
A.1 Introduction 209
A.2 Wham-O’s FrisbeeTM Disc 209
A.3 Professional Sport Disc Development 210

-4-
Appendix B Disc-wing Aircraft R&D 211
B.1 Introduction 211
B.2 Literature Review: In Brief 211
B.3 History of the Circular Planform 213
B.3.1 Chance-Vought XF-5U-1 213
B.3.2 Avro’s Special Projects Group 213
B.3.3 Lenticular Re-entry Vehicle 214
B.3.4 Self-suspended Flare for Special Ordnance 215

Appendix C Blockage Corrections & Analysis 216


C.1 Maskell’s Blockage Correction 216
C.2 Analysis 218

-5-
List of Figures Page

Figure 2.1 Cross-sectional frisbee-like disc-wing profile. 114

Figure 2.2 Relationship between the Lift & Drag and the Axial X & Normal Z 115
forces. The Side force (not shown here) is identical for both systems, directed
out of the page.

Figure 2.3 Definition of reference locations for longitudinal forces and moments 116
acting on a tailless flight vehicle.

Figure 2.4 Illustration of longitudinal stability and trim characteristics of a 117


tailless vehicle for permutations of camber and static margin.

Figure 2.5 Comparison of the aerodynamic centre location as a function of angle 118
of attack for a Frisbee and a circular planform wing with a non-axisymmetric
airfoil cross-section.

Figure 2.6 (a) Disc-wing flight dynamics. (b) Schematic of body fixed axes. 119

Figure 3.1 Comparison of aerodynamic load data for circular planform wings, 120
derived from various sources found in the literature, with chordwise cross-
sections as seen to the right of the figure.

Figure 4.1 Wind tunnel support strut fixed to balance beam above wind tunnel 121
working section (left), pressure lines (capillary tubing) connected to pressure
transducer (right middle) via a scanivalve (right).

Figure 4.2 Wind tunnel control room. 122

Figure 4.3 L-shaped support strut and disc-wing mounted vertically at 0o 123
incidence.

-6-
Figure 4.4 Surface paint flow visualisation rig configuration, the disc is 124
horizontal at zero incidence. The disc is mounted on a central axle, which is
fixed to a horizontal crossbar, two vertical struts either side of the balance beam
support the crossbar, the incidence arm is connected to control the angle of
attack. Rig designed by Ali (1998).

Figure 4.5 Smoke wire flow visualisation rig configurations: (a) Motor driven 125
axle connected to centre of vertical test model allows visualisation of spinning
disc, up to 24Hz. (b) Geared down motor driven axle connected to rim of test
model allows visualisation of the full central flow field cross-section of a non-
spinning disc.

Figure 4.6 Cross-sectional disc profiles. (a) Frisbee-like (b) Intermediate (c) 126
Flatplate.

Figure 4.7 Wind tunnel configuration to measure interference and tare effects, 127
the disc was mounted on a dummy support (left). The dummy strut was a mirror
image of the measuring strut (right) and held the disc in the correct position (on
the balance centre) but not connected to the balance itself. The measuring strut
was set with the axle tip (centre) at around 3mm from the disc, without touching,
and was fixed to the balance.

Figure 4.8 Comparison of the disc test model and the support strut at zero spin 128
o o 5
rate, AoA = −10 to 30 , V∞ = 20m/s, Re = 3.78×10 .

Figure 4.9 L-shaped rig configuration, to measure the surface pressure 129
distribution. The capillary tubes were carefully wound around the support strut.

Figure 4.10 (a) The pressure tappings in the surface of the disc-wing test models 130
were positioned on a spiral curve. (b) The spiral curve of points was stepped at
12° increments, by yawing the model, to achieve full coverage of the disc-wing
surface. That is a polar grid array of 571 points on each surface, upper and
cavity.

-7-
Figure 4.11 Schematic of (a) smoke wire and oil supply system, (b) the oil 131
feeder device.

Figure 5.1 Load Characteristics at zero spin rate, AoA = −10o to 30o, V∞ = 132
20m/s, Re = 3.78×105.

Figure 5.2 Load Characteristics at zero spin rate, AoA = −10o to 50o, V∞ = 133
20m/s, Re = 3.78×105.

Figure 5.3 Load Characteristics at zero spin rate, AoA = −100o to 100o, V∞ = 134
20m/s, Re = 3.78×105.

Figure 5.4 Effect of Reynolds number at zero spin rate, AoA = −10o to 30o. 135
Tunnel speed varied to achieve different Re (V∞ = 6, 10, 15, 20, 25m/s).

Figure 5.5 Further load characteristics at zero spin rate, AoA = −10o to 30o, V∞ = 136
20m/s, Re = 3.78×105.

Figure 5.6 Comparison of the present experimental data to circular planform 137
wings, both theoretical and experimental.

Figure 5.7 Comparison of the present experimental data to Frisbee-like circular 138
planform wings from the literature.

Figure 5.8 Comparison of the present experimental data to a similar Frisbee-like 139
circular planform wing.

Figure 5.9 Comparison of the present experimental data to a variety of circular 140
planform wings, with chordwise cross-sections as seen below.

Figure 5.10 Lift coefficient: Comparison of the present experimental data to that 141
found in the literature for similar circular planform wings, with chordwise cross-
sections as seen to the right of the figure.

-8-
Figure 5.11 Drag coefficient: Comparison of the present experimental data to 142
that found in the literature for similar circular planform wings, with chordwise
cross-sections as seen to the right of the figure.

Figure 5.12 Pitching moment coefficient: Comparison of the present 143


experimental data to that found in the literature for similar circular planform
wings, with chordwise cross-sections as seen to the right of the figure.

Figure 5.13 Load characteristics at various advance ratios, AdvR = 0 to 1, AoA 144
= −10o to 30o, V∞ = 20m/s, Re = 3.78×105.

Figure 5.14 Load characteristics (detail) at various advance ratios, AdvR = 0 to 145
1, AoA = −10o to 30o, V∞ = 20m/s, Re = 3.78×105.

Figure 5.15 Effect of spin on the Lift Coefficient: Comparison set of spin rates 146
(0, 4, 8, 16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15, 20m/s), AoA =
-10o to 30o.

Figure 5.16 Effect of spin on the Drag Coefficient: Comparison set of spin rates 147
(0, 4, 8, 16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15, 20m/s), AoA =
-10o to 30o.

Figure 5.17 Effect of spin on the Side Force Coefficient: Comparison set of spin 148
rates (0, 4, 8, 16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15, 20m/s),
AoA = -10o to 30o.

Figure 5.18 Effect of spin on the Pitching Moment Coefficient: Comparison set 149
of spin rates (0, 4, 8, 16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15,
20m/s), AoA = -10o to 30o.

Figure 5.19 Effect of spin on the Rolling Moment Coefficient: Comparison set 150
of spin rates (0, 4, 8, 16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15,
20m/s), AoA = -10o to 30o.

-9-
Figure 5.20 Effect of Advance Ratio at 0o AoA for various Reynolds numbers, 151
tunnel speed varied to achieve different Re (V∞ = 6, 10, 15, 20m/s), AdvR = 0
up to 3.5, each curve utilises a consistent range of spin rates (0 to 31Hz).

Figure 5.21 Comparison of disc-wings with various thicknesses achieved by 152


modifying the rim height to increase the cavity depth, with chordwise cross-
sections as seen below, th / c = 0.01 to 0.16.

Figure 5.22 Comparison of the baseline (hollow) Frisbee configuration to a 153


cavity filled (solid) disc-wing geometry.

Figure 5.23 2D flat surface colour weighted contour plot of pressure coefficients 154
Cp derived from measurements of the 3D upper surface pressure distribution of a
non-spinning disc-wing, taken over the entire surface, not over half the disc and
mirrored to the other side, flow direction from top to bottom, AoA = 5º, V∞ =
15m/s, Re = 2.84×105.

Figure 5.24 2D flat surface colour weighted contour plot of pressure coefficients 155
Cp derived from measurements of the 3D cavity surface pressure distribution of
a non-spinning disc-wing, taken over the entire surface, not over half the disc
and mirrored to the other side, flow direction from top to bottom, AoA = 5º, V∞
= 15m/s, Re = 2.84×105.

Figure 5.25 2D flat surface colour weighted contour plot of pressure coefficients 156
Cp derived from measurements of the 3D upper (a) & cavity (b) surface pressure
distribution of a non-spinning disc-wing, taken over the entire surface, not over
half the disc and mirrored to the other side, flow direction from top to bottom,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

Figure 5.26 Central cross-sectional pressure profile at the half span station. 157
Pressure coefficients Cp derived from measurements of the 3D surface pressure
distribution of a non-spinning disc-wing. Non-dimensional disc geometry shown
below pressure plot, leading edge on the left of the figure. Unbroken line - upper
surface, dashed line - lower surface, AoA = 5°, V∞ = 15m/s, Re = 2.84×105.

- 10 -
Figure 5.27 Central cross-sectional pressure profile at the half chord station, 158
pressure coefficients Cp derived from measurements of the 3D surface pressure
distribution of a non-spinning disc-wing. Non-dimensional disc geometry shown
below pressure plot, port wing tip on the left of the figure. Unbroken line - upper
surface, dashed line - lower surface, AoA = 5°, V∞ = 15m/s, Re = 2.84×105.

Figure 5.28 2D flat surface colour weighted contour plots of pressure 159
coefficients Cp for a range of AoA, derived from measurements of the 3D upper
surface pressure distribution of a non-spinning disc-wing, taken over the entire
surface, not over half the disc and mirrored to the other side, flow direction from
top to bottom, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.

Figure 5.29 2D flat surface colour weighted contour plots of pressure 160
coefficients Cp for a range of AoA, derived from measurements of the 3D cavity
surface pressure distribution of a non-spinning disc-wing, taken over the entire
surface, not over half the disc and mirrored to the other side, flow direction from
top to bottom, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.

Figure 5.30 Central cross-sectional pressure profiles at the half span station, for 161
a range of AoA. Pressure coefficients Cp were derived from measurements of the
3D surface pressure distribution of a non-spinning disc-wing. Non-dimensional
disc geometry shown below pressure plot, leading edge on the left of the figure.
Unbroken line - upper surface, dashed line - lower surface, AoA = -10º to 30º,
V∞ = 20m/s, Re = 3.78×105.

Figure 5.31 Central cross-sectional pressure profiles at the half span station, for 162
a range of AoA. Pressure coefficients Cp were derived from measurements of the
3D surface pressure distribution of a non-spinning disc-wing. Non-dimensional
disc geometry shown below pressure plot, leading edge on the left of the figure.
Unbroken line - upper surface, dashed line - lower surface, AoA = -10º to 30º,
V∞ = 20m/s, Re = 3.78×105.

- 11 -
Figure 5.32 Cross-sectional pressure profiles at various span stations (a) 1/2 b, 163
(b) 3/8 b, (c) 1/4 b (d) 1/8 b. Pressure coefficients Cp were derived from
measurements of the 3D surface pressure distribution of a non-spinning disc-
wing. Non-dimensional disc cross-section shown below pressure plot, leading
edge on the left of the figure. Unbroken line - upper surface, dashed line - lower
surface, AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

Figure 5.33 Cross-sectional pressure profiles at various span/chord stations (i) 164
1/2 b, (ii) 3/8 b, (iii) 1/4 b, (iv) 1/8 b, (v) 1/2 c and a range of AoA. Pressure
coefficients Cp were derived from measurements of the 3D surface pressure
distribution of a non-spinning disc-wing. Non-dimensional disc cross-section
shown below pressure plot (e), leading edge (i, ii, iii, iv) and port wing-tip (v) on
the left of each plot. Unbroken line - upper surface, dashed line - lower surface,
AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.

Figure 5.34 2D flat surface colour weighted contour plots of pressure 165
coefficients Cp for a range of Re, derived from measurements of the 3D upper
surface pressure distribution of a non-spinning disc-wing, taken over the entire
surface, not over half the disc and mirrored to the other side, flow direction from
top to bottom, AoA = 5º, V∞ = 6 to 20m/s, Re = 1.13 to 3.78 ×105.

Figure 5.35 2D flat surface colour weighted contour plots of pressure 166
coefficients Cp for a range of Re, derived from measurements of the 3D cavity
surface pressure distribution of a non-spinning disc-wing, taken over the entire
surface, not over half the disc and mirrored to the other side, flow direction from
top to bottom, AoA = 5º, V∞ = 6 to 20m/s, Re = 1.13 to 3.78 ×105.

Figure 5.36 Central cross-sectional pressure profiles at the half span station, for 167
a range of Re. Pressure coefficients Cp were derived from measurements of the
3D surface pressure distribution of a non-spinning disc-wing. Non-dimensional
disc geometry shown below pressure plot, leading edge on the left of the figure.
Unbroken line - upper surface, dashed line - lower surface, AoA = 5º, V∞ = 6 to
20m/s, Re = 1.13 to 3.78 ×105.

- 12 -
Figure 5.37 Pressure load characteristics, integrated from the pressure data over 168
the 3D disc-wing geometry, and a comparison to the load measurements, taken
from the wind tunnel balance, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.

Figure 5.38 Upper surface paint patterns for a non-spinning disc-wing, flow 169
direction from top to bottom, AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

Figure 5.39 Lower (cavity) surface paint patterns for a non-spinning disc-wing, 170
flow direction from top to bottom, AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

Figure 5.40 Upper (a) and Lower (cavity) (b) surface paint patterns for a non- 171
spinning disc-wing, including superimposed labels to aid the explanation of
surface flow features, flow direction from top to bottom, AoA = 5º, V∞ = 15m/s,
Re = 2.84×105.

Figure 5.41 Upper surface paint patterns for a non-spinning disc-wing, over a 172
range of AoA, flow direction from top to bottom, AoA = 0º to 30º, V∞ = 15m/s,
Re = 2.84×105.

Figure 5.42 Lower (cavity) surface paint patterns for a non-spinning disc-wing, 173
over a range of AoA, flow direction from top to bottom, AoA = 0º to 30º, V∞ =
15m/s, Re = 2.84×105.

Figure 5.43 Upper surface paint patterns (detail) for a non-spinning disc-wing, 174
depicting boundary layer reattachment features, flow direction from top to
bottom, (a) Straight line reattachment (half disc) at 20° AoA, (b) Nodal point
reattachment at 30° AoA. V∞ = 15m/s, Re = 2.84×105.

Figure 5.44 Half surface comparison plot: 3D pressure surface colour weighted 175
contour plots of pressure coefficients Cp derived from measurements of the
upper (a) & lower (cavity) (b) surface pressure distribution over a non-spinning
disc-wing, superimposed onto the half surface flow visualisation images, flow
direction from top to bottom, AoA = 5°, V∞ = 15m/s, Re = 2.84×105.

- 13 -
Figure 5.45 Comparison plot: Central cross-sectional, upper surface pressure 176
profile at the half span station, superimposed onto a half surface paint flow
visualisation image. Leading edge on the left of the figure. Pressure
coefficients Cp derived from measurements of the upper surface pressure
distribution over a non-spinning disc-wing. Separation (S1, S2) and
reattachment (R1) lines are marked on for comparison, flow direction from left
to right, AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

Figure 5.46 Half surface comparison plot: 3D pressure surface colour weighted 177
contour plots of pressure coefficients Cp derived from measurements of the
upper surface pressure distribution over a non-spinning disc-wing, superimposed
onto the half surface flow visualisation images, over a range of AoA, flow
direction from top to bottom, AoA = 0º to 30º, V∞ = 15m/s (flow vis) & 20m/s
(pressure), Re = 2.84×105 & 3.78×105.

Figure 5.47 Half surface comparison plot: 3D pressure surface colour weighted 178
contour plots of pressure coefficients Cp derived from measurements of the
lower surface pressure distribution over a non-spinning disc-wing, superimposed
onto the half surface flow visualisation images, over a range of AoA, flow
direction from top to bottom, AoA = 0º to 30º, V∞ = 15m/s (flow vis) & 20m/s
(pressure), Re = 2.84×105 & 3.78×105.

Figure 5.48 Central cross-section of the flow field over a non-spinning disc- 179
wing, for a range of AoA, flow direction from right to left. An electrically
heated wire vaporised oil to create smoke, visualising the flow field when
illuminated. AoA = 0° to 50°, V∞ = 3m/s, Re = 5.67×104.

Figure 5.49 Central cross-section of the leading edge separation bubble over a 180
disc-wing, for a range of AoA, flow direction from right to left. An electrically
heated wire vaporised oil to create smoke, visualising the separation bubble
when illuminated by a laser light sheet. AoA = 0° to 30°, AdvR = 0.7, V∞ =
3m/s, Re = 5.67×104.

- 14 -
Figure 5.50 Central cross-section of the separated shear layer and cavity flow 181
over a non-spinning disc-wing, for a range of AoA, flow direction from left to
right. An electrically heated wire vaporised oil to create smoke, visualising the
cavity flow when illuminated by a laser light sheet. AoA = 0° to 30°, V∞ = 3m/s,
Re = 5.67×104.

Figure 5.51 Visualisation of the wake, downwash and trailing vortex structures 182
downstream of a non-spinning disc-wing, for a range of AoA. An electrically
heated wire vaporised oil to create smoke, visualising the flow field when
illuminated by ambient lighting (i) or a laser light sheet (ii). The camera was
placed downstream of the disc model, flow direction is out of the page. The
wake cross-sections (ii) were taken at a distance of one diameter from the
trailing edge, AoA = 0° to 10°, V∞ = 3m/s, Re = 5.67×104.

Figure 5.52 Visualisation of the wake, downwash and trailing vortex structures 183
extending to just over five diameters downstream form the trailing edge of a
non-spinning disc-wing, for a range of AoA. An electrically heated wire
vaporised oil to create smoke, visualising the flow field when illuminated, flow
direction from right to left, AoA = 0°, 5° & 10°, V∞ = 3m/s, Re = 5.67×104.

Figure 5.53 Cross-sectional slices through the wake, downwash and trailing 184
vortex structures. Each slice was located aft of a non-spinning disc-wing at a
range of positions, up to two chord lengths downstream from the trailing edge.
An electrically heated wire vaporised oil to create smoke, visualising the flow
field when illuminated by a laser light sheet, flow direction is out of the page.
x/c = 0 to 2, AoA = 10°, V∞ = 3m/s, Re = 5.67×104.

Figure 5.54 Visualisation of flow structures over the upper surface planform 185
including the separation bubble and wake over a disc-wing, for various AoA and
advance ratios. An electrically heated wire vaporised oil to create smoke,
visualising the flow field when illuminated, flow direction from top to bottom.
i) AdvR = 0 & 0.9, AoA = 0°, V∞ = 1.5m/s, Re = 2.84×104,
ii) AdvR = 0 & 0.7, AoA = 5°, V∞ = 3m/s, Re = 5.67×104.
b) Spinning Disc: Advancing side on left of figure, retreating side on the right.

- 15 -
Figure 5.55 Cross-sectional slices through the wake, downwash and trailing 186
vortex structures at various spin rates. Each slice was located aft of a disc-wing
model at a couple of positions, up to two chord lengths downstream of the
trailing edge. An electrically heated wire vaporised oil to create smoke,
visualising the flow field when illuminated by a laser light sheet, flow direction
is out of the page. Advancing side on the left of the figure, retreating side on the
right. x/c = 1 & 2, AdvR = 0, 0.7 & 1.6, AoA = 10°, V∞ = 3m/s, Re = 5.67×104.

Figure 5.56 Cross-sectional slices through the wake, downwash and trailing 187
vortex structures at various advance ratios including opposite spin directions.
Each slice was located aft of a disc-wing model at a range of positions, up to two
chord lengths downstream from the trailing edge. An electrically heated wire
vaporised oil to create smoke, visualising the flow field using laser light sheet
illumination, flow direction is out of the page. x/c = 0 to 2, AdvR = −0.7, 0 &
0.7, AoA = 5°, V∞ = 3m/s, Re = 5.67×104. i) Advancing side on the left of the
figure, retreating side on the right. iii) Advancing side on the right of the figure,
retreating side on the left.

Figure 6.1 Two-dimensional central cross-sectional flow topology for a non- 188
rotating disc-wing at 10° AoA. The proposed topology is based upon the flow
visualisation, depicting many of the flow structures, namely the separation
bubble, cavity flow and turbulent wake. Flow direction from right to left.

Figure 6.2 Three-dimensional flow topology for a non-rotating disc-wing at 10° 189
AoA. The proposed topology is based upon the flow visualisation, depicting
many of the flow structures, namely the separation bubble and trailing vortices.
Flow direction from right to left.

Figure 6.3 Two-dimensional central cross-sectional flow topology for a non- 190
rotating disc-wing for a range of AoA. Each proposed topology is based upon
the flow visualisation, depicting many of the flow structures, namely the
separation bubble, cavity flow and turbulent wake. Flow direction from right to
left, AoA = 0º to 30º.

- 16 -
Figure 6.4 Two-dimensional central cross-sectional flow topology for a non- 191
rotating disc-wing at 45º AoA, pre- (a) and post-stall (b). Each proposed
topology is based upon the flow visualisation, depicting many of the flow
structures, namely the separation bubble and turbulent wake. Flow direction
from right to left, AoA = 0º to 30º.

Figure 7.1 Additional axes systems used for the disc-wing simulation. 192
(a) Zero sideslip body axes (xyz)3. (b) Relative wind axes (xyz)4.

Figure 7.2 Comparison of simulated results (no dots) with experimental time 193
histories (dots) from flight tests (flight f2302; Hummel, 2003), for an identical
set of initial conditions, given below.

Figure 7.3 Trajectory plots and parameter time histories for a simulated Frisbee 194
flight path. Solid lines correspond to results from using the full non-linear
aerodynamic model. Grey lines correspond to results from using a linear
approximation.

Figure 7.4 Comparison of the Frisbee-like disc-wing wind tunnel data to the 195
linear approximations for use in the simulation. The alpha dependent
aerodynamic coefficients are governed by the equations beneath each respective
figure, linear derivatives given also. Angle of attack range given in radians, AoA
= −0.2 to 0.6 rads.

Figure 7.5 Effect of launch conditions (advance ratio) on disc-wing XZ and XY 196
trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o , θ L = 0o , ψ L = 0o , α L = 0o , VL = 19 m/s, AdvR = 0 to 8.

Figure 7.6 Effect of launch conditions (pitch angle) on disc-wing XZ and XY 197
trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o , θ L = 0o to 90°, ψ L = 0o , α L = 0o , VL = 19 m/s, AdvR = 1.1.

Figure 7.7 Range as a function of launch pitch angle. 198

Figure 7.8 Flight duration as a function of launch pitch angle. 199


- 17 -
Figure 7.9 Effect of launch conditions (angle of attack) on disc-wing XZ and XY 200
trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o , θ L = 0o , ψ L = 0o , α L = 0o to 90°, VL = 19 m/s, AdvR = 0.5.

Figure 7.10 Effect of launch conditions (roll angle) on disc-wing XZ and XY 201
trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o to 90°, θ L = 0o , ψ L = 0o , α L = 0o , VL = 19 m/s, AdvR = 0.5.

Figure 7.11 Disc landing position (XY location at which Z=0) as a function of 202
launch roll angle. Re-plotted from data in Fig. 57.8a.

Figure 7.12 Trajectory plots and parameter time histories for a simulated spiral 203
turn manoeuvre obtained using a constant rolling moment control input.
φ L = −45o, θL = −30o, ψL = 0o, αL = 5o, VL = 13.5 m/s, CLcontrol = 0.1, AdvR = 0.5.

Figure 7.13 Trajectory plots and parameter time histories for a simulated spiral 204
roll manoeuvre obtained using a constant pitching moment control input.
φ L = −0o, θL = −20o, ψL = 0o, αL = -3o, VL = 19 m/s, CMcontrol = 0.2, AdvR = 0.34.

Figure C.1 (a) Uncorrected lift data taken in the wind tunnel CLu2 plotted against 205
uncorrected drag data CDu, equation of the best fit line given.
(b) Disc-wing drag analysis for a lifting wing, after Maskell (Fig. 8, 1965).
(c) Uncorrected wind tunnel data CLu2 plotted against CDu, for the negative AoA
range, equation of the best fit line given.

Figure C.2 Corrected and uncorrected drag results for various AoA ranges. 206
o o o o o o
(a) AoA –10 to 30 , (b) AoA –10 to 50 , (c) AoA –100 to 100 .

Figure C.3 Experimentally measured open face bluff body drag coefficients for 207
various disc planform configurations (Potts & Crowther, 2002), compared to
similar shapes detailed in White (1999) and Stilley & Carstens (1971), AoA =
+90o or –90o.

- 18 -
Abstract

The University of Manchester

Abstract of Thesis submitted by Jonathan Potts for the Degree of Doctor of


Philosophy and entitled Disc-wing Aerodynamics
Month and Year of submission: March 2005

Disc-wings are a class of un-powered, axi-symmetric flight vehicles that use spin to
achieve acceptably stable flight characteristics. Examples of commonly encountered
disc-wings include the Frisbee sports disc, the athletics discus and the clay pigeon.
Historically, it appears that most disc-wing designs have been based on trial and error
approaches. The main aim of the present work is to develop a theory of flight for spin-
stabilised disc-wings that can be used to inform the process of their design. This theory
of flight is based both on theoretical analysis and experimental data.

It is shown from a simple trim and stability analysis that a disc-wing with positive
camber will trim at a positive angle of attack. However, for most axi-symmetric cross-
sectional shapes, the aerodynamic centre is ahead of the centre of the disc (which by
definition is the disc centre of gravity). Hence, the static margin is negative and the disc
is unstable in pitch.

In practice, a disc-wing must be spun in order to fly successfully. The imparted angular
momentum due to the spin means that, through precessional effects, the destabilising
pitching moments tend to result in a rolling motion rather than a pitching motion. Thus,
without spin, a disc-wing would tumble soon after release. With spin however, the disc-
wing will not tumble, instead it tends to exhibit a relatively benign roll to the left or
right, depending on the spin direction.

The aerodynamic characteristics of various disc-wing geometries based around a


Frisbee sports disc are investigated through a series of wind tunnel experiments on a
spinning and non-spinning disc. It is shown that the basic lift and drag characteristics
are consistent with those expected for a finite wing of the same aspect ratio. The
pitching moment characteristic is key to understanding the resulting disc dynamics. A
comparison of pitching moment curves is given, for a number of different cross-
- 19 -
sectional profiles, some tested as part of the present work and some taken from data
found in the literature. It is shown that the Frisbee cross-section is unique in that the
pitching moment is zero at around 9o angle of attack, approximately coincident with the
angle attack for best lift to drag ratio, and that the disc is approximately neutrally stable
in this region. It is these characteristics that enable a typical Frisbee to fly successfully.
Spin has almost negligible effect on aerodynamic forces and moments.

Force and moment data is supported by surface pressure data, and by on and off surface
flow visualisation. Surface pressure data shows that the aerodynamic centre of the
Frisbee cross-section is shifted aft by the presence of an aft pressure peak that is not
present on other cross-section shapes. The aft pressure peak is a function of both the
upper surface geometry and the presence of the cavity on the under surface of the disc.
Flow visualisation and pressure data are used to propose a model of disc-wing flow
topology that is dependant on the angle of attack and includes leading edge separation
and reattachment, recirculating cavity flow and a pair of trailing vortices.

To understand further disc-wing flight dynamics and the effect of aerodynamic


characteristics, a six-degree of freedom disc-wing simulation model was developed
using Matlab. The simulation is validated against published Frisbee trajectory data
obtained from free-flight experiments. Flight profiles are also discussed for a number of
different launch conditions consistent with a range of typical Frisbee throws. The
simulation is also used to demonstrate that with control moments from suitable control
effectors, it is possible to generate a number of proscribed manoeuvres, including a
spiral turn and a spiral roll.

- 20 -
Declaration

No portion of the work referred to in the thesis has been submitted in support of an
application for another degree or qualification of this or any other university or other
institute of learning.

Copyright

Copyright in text of this thesis rests with the author. Copies (by any process) either in
full, or of extracts, may be made only in accordance with instructions given by the
author and lodged in the John Rylands University Library of Manchester. Details may
be obtained from the Librarian. This page must form part of any such copies made.
Further copies (by any process) of copies made in accordance with such instructions
may not be made without the permission (in writing) of the Author.

The ownership of any intellectual property rights which may be described in this thesis
is vested in the University of Manchester, subject to any prior agreement to the contrary,
and may be made available for use by third parties without the written permission of the
University, which will prescribe the terms and conditions of any such agreement.

Further information on the conditions under which disclosures and exploitation may
take place is available from the Head of the School of Mechanical, Aerospace & Civil
Engineering.

- 21 -
Acknowledgements

Bill Crowther, PhD Supervisor. Thankyou so much for pondering over FrisbeeTM flight
in the first place, thus providing the opportunity for me to enjoy contemplating with
you. Supporting me in frequent international conference visits, even against your better
judgement, has made my research programme a much more exciting and challenging
experience. Your patience and perseverance have truly been tested to the end.

My time at the Goldstein Lab. has been a wonderful experience, the friendly working
environment far outweighed the daily noise pollution. David Smith, apart from being an
all round electronics genius, has offered me his ear whenever I needed it, his pastoral
care has been spot on, you are such a good man! His wife Carole is an extension to the
support the Smiths have collectively provided to students and staff at the Lab. and
beyond. Andrew Kennaugh is the font of all knowledge and all that is City, if you can’t
find a reference get Andrew on the case. The dynamic speed with which the technicians
fabricate equipment, constantly foregoing tea-breaks to meet tight deadlines, will
forever astound. Dave Mould constructed the marvellously versatile wind tunnel rig and
disc-wing model, it could not have worked out any better. Thanks for putting up with,
‘Is it ready yet?’ first thing in the morning.

The breadth of knowledge I have acquired during these few years can primarily be
attributed to the mentoring of Andrew Crook. His unselfish servant attitude is worked
out practically through his brief you, show you, leave you approach. I miss many things
about him, particularly his constant banging on about how nature has been doing fluid
mechanics for thousands of years but none more so than his friendship. Just one more
Ahmedia before I leave…

I have had the pleasure of making many friends on my travels in the USA. In particular
I would like to thank Peter Lissaman for making himself available throughout my
research, you always seem to be two steps ahead of me. My American friends have
always made me feel extremely welcome, none more so than at UC Davis. Mont
Hubbard & Sarah Hummel have cleared their schedule more than once to accommodate
my somewhat impromptu visits.

- 22 -
I must acknowledge the disc-wing related experimental work of Wajid Ali, Martin
Stone and Carlos Martin-Henry, undergraduate students at the University of
Manchester. All have now successfully graduated on the Aerospace Engineering course.

This research was funded by the Engineering and Physical Sciences Research Council
(EPSRC), award reference number 98317373. International Travel Grants, primarily to
present papers at various international conferences in the USA, were secured through
multiple applications and an extensive search for relevant awarding bodies: EPSRC,
RAcEng, RAeS & IoP.

____________________________________________

I cannot capture the gratitude I have for my parents John & Maxine Potts in brief. Their
unceasing support in every way possible has ensured that I can work this thing out to the
end. Along with my sister Kate, they have provided the immovable standard on which
my life is based.

How could I forget the constant banter, “Eeh-ya Jonneh, put d’keckle on an’ make me a
brew before ah brake yer jaw, d’yer get meh” from the infamous Manc brothers Phil
an’ar kid Paul Stapleton. You have livened up my evenings and weekends with never-
ending inner city comedy. Mary & Adge Bennett have treated me like one of their own.

Thanks to Graham & Veronica Dawson I have been able to find refuge from my work in
Staly-vegas, claiming them there hills as my own. I have especially enjoyed the many
late nights I’ve spent with your son David, over a wee dram.

- 23 -
The Author

Master of Mathematics (MMath) in Combined Mathematics, awarded 2:1 with honours.


Jonathan Potts studied mathematics as an undergraduate at the University of
Manchester, U.K. (1994-98) gaining experience in a varied cross-section of the
discipline, with special interest in applied mathematics. A grounding in fluid mechanics
followed including research project work on Ocean Waves, Boomerangs and Tornado-
like Vortices. Written reports contain a review of the scientific literature base, original
work includes an experiment on the bath-tub vortex and an analytical model of
triangular boomerang dynamics.

He chose to continue his education at the University of Manchester as a postgraduate in


the Division of Aerospace Engineering (1998-2002), based at their experimental fluids
and aerodynamics facility, the Goldstein Research Laboratory. He investigated the
aerodynamics of a spin-stabilised axi-symmetric circular wing configuration i.e. disc-
wing and conducted a feasibility assessment of appropriate flow control methodologies.
Dissemination activities include the presentation of state of the art knowledge, from the
disc-wing research initiative, at international conferences including AIAA ASM&E
(Reno, Nevada, USA), aerospace engineering institutions including invited seminars at
USC & UC Davis, CA and briefed interested professionals at AeroVironment Inc.

- 24 -
Nomenclature

a Lift curve slope, finite wing


a0 Lift curve slope, infinite wing
AdvR Advance ratio ( r c / 2V∞)
AoA Angle of attack (°)
AR Aspect ratio
c Disc-wing chord & diameter (m)
CDrag Drag coefficient
CDo Profile drag coefficient
cg Centre of gravity
CLift Lift coefficient
CM Pitching moment coefficient (about c/2)
cp Centre of pressure
Cp Pressure coefficient
CL Rolling moment coefficient
CY Side force coefficient
H Angular momentum (kg m2 s)
I Moment of inertia (kg m2)
e Span efficiency factor
Lift Lift (N)
LE Leading edge
L Rolling moment (Nm)
M Pitching moment (Nm)
m Mass of disc-wing (kg)
N Yawing moment (Nm)
Re Reynolds number
S Surface planform area (m2)
th Disc thickness (m)
TE Trailing edge
x Displacement vector, along roll axis (m)
y Displacement vector, along pitch axis (m)
z Displacement vector, along yaw axis (m)
g Acceleration due to gravity (ms-2)

- 25 -
p Rate of roll (rad s-1)
q Rate of pitch (rad s-1)
r Rate of yaw (rad s-1)
q∞ Dynamic pressure (½ ρ∞V∞2)
V∞ Freestream/wind velocity (ms-1)
α Geometric angle of attack (°)
αo Zero lift angle of attack (°)
π Circle circumference/diameter ratio (~ 3.14)
ρ∞ Density of air (kg m-3)

- 26 -
One night I dreamed I was walking
Along the beach with the Lord.
Many scenes from my life flashed across the sky.
In each scene I noticed footprints in the sand.
Sometimes there were two sets of footprints.
Other times there were one set of footprints.
This bothered me because I noticed that
During the low periods of my life when I was
Suffering from anguish, sorrow, or defeat,
I could see only one set of footprints,
So I said to the Lord, "You promised me,
Lord, that if I followed You,
You would walk with me always.
But I noticed that during the most trying periods
Of my life there have only been
One set of prints in the sand.
Why, When I have needed You most,
You have not been there for me?"
The Lord replied,
"The times when you have seen only one set of footprints
Is when I carried you."

Footprints in the Sand


by Mary Stevenson
081122 – 060199
- 27 -
Chapter 1 Introduction

1.1 Opening Statements


It is the Frisbee1 sports disc that has inspired this study. The ability of such a simple
shape to glide so successfully over short distances has amazed many over its short
history. With a minimum of practice a novice can extract maximum enjoyment from
throwing, catching and observing the simple glider design, oblivious to the complex
aerodynamics securing such stable flight characteristics.

The flying disc is the more generic term used to describe what is most famously known
as the Frisbee disc. That is a throwing implement with circular planform, typically
launched with a specific backhand motion to impart velocity and spin to the disc in
flight. The sports disc is therefore a flying gyroscope, with similar dynamics to a
boomerang but vastly different aerodynamics. A history of the Frisbee disc is outlined
in Appendix A.

The original goal of the work presented in this thesis was to develop a controllable
flight vehicle based on the principles of a Frisbee sports disc. Flow control technologies
were to be used to provide control forces and moments and a suitable flight control
system developed. However, soon after starting the work, it became clear that, firstly,
the basic aerodynamics and flight dynamics of a disc wing were more complex than
initially thought, and that, secondly, the available literature on disc wing aerodynamics
was sparse and rather inconclusive. In light of this, the original study was refocused on
obtaining a comprehensive aerodynamic data set for the Frisbee sports disc and
developing a sound theoretical framework for understanding how the aerodynamics
influences the vehicle flight dynamics.

1.2 Definitions
For the present work, a ‘disc-wing’ is defined as wing with circular planform and axial
symmetry. Axial symmetry is important because it means that the disc-wing has no
preferred angular orientation for flight, and hence can be spin-stabilised. Note that this
_____________________________
1
Frisbee is used to define the axisymmetric aerodynamic shape of the generic flying
disc-wing model tested in this study, for ease of description and understanding. Frisbee,
frisbee-like and disc are used interchangeably throughout the thesis and should be taken
to have the same meaning. FrisbeeTM is a registered trademark of Wham-O Inc.
- 28 -
disc-wing definition includes objects such as the Frisbee and discus, but excludes
aircraft with non-axisymmetric circular planforms such as the Black Widow MAV or
Chance-Vought XF-5U-1 (see Appendix B) that fly at a fixed angular orientation to the
free stream.

Aerodynamics is principally the study of the motion of air around a body and the forces
and moments that arise on the body as result of the air motion. Within the present study,
of particular interest are the lift, drag and pitching moments acting on a disc-wing at
various angles of attack to the oncoming wind. Furthermore, to understand the
mechanism by which forces and moments are generated, it is necessary to measure the
pressure distribution over the surface of the disc-wing and to identify the associated
flow topology through on and off surface flow visualisation experiments.

Flight dynamics is concerned with understanding how a body moves under the influence
of aerodynamic, gravitational and propulsive forces. Spin-stabilised disc-wings offer an
interesting challenge to flight dynamics in that the response of the disc to moments is
dominated by gyroscopic precession. It is the effect of precession that causes the
familiar roll of a sports disc along its flight path. The cross-sectional profile of a disc-
wing has a strong influence on the rate at which the disc rolls. It turns out that the shape
of Frisbee is such that the roll rate is minimised, however there was no experimental
evidence at the outset of the present work to show why this was the case.

The principle tool of study in the present work is the wind tunnel. The wind tunnel
works on the simple premise that the aerodynamic forces and moments acting on a body
moving through still air are the same as the aerodynamic forces and moments acting on
a stationary body in an air stream moving at the same velocity as the original body.
However, from an experimental point of view, it is far more convenient to move the air
and keep the body stationary than it is to move the body and keep the air stationary.

1.3 Thesis Aims & Objectives


Aim
To develop a theory of flight for spin-stabilised disc-wings based on theoretical analysis
and experimental data.

- 29 -
Objectives
1. To provide context and motivation for experimental work through introductory
remarks and a review the existing disc-wing literature.
2. To provide a basic theoretical context for understanding the aerodynamic and flight
dynamic characteristics of a disc-wing.
3. To present, describe and discuss wind tunnel results and to propose a flow topology
consistent with the experimental data.
4. To describe the development of a disc-wing dynamic simulation programme based
on wind tunnel data and to present and describe simulated trajectory results
obtained.

1.4 Thesis Outline


Chapter 2 presents a theoretical analysis of the geometric, aerodynamic and flight
dynamic parameters relevant to disc-wings and Chapter 3 provides a review of the disc-
wing literature. The apparatus and techniques used to perform the experimental work
are described in Chapter 4, with experimental work and proposed flow topology
presented and discussed in Chapters 5 and 6, respectively. Chapter 7 introduces the
disc-wing dynamic simulation programme and presents results for a range of simulated
trajectories. Finally, Chapter 8 summarises the overall conclusions from the work.

- 30 -
Chapter 2 Theoretical background
Summary
This chapter presents a theoretical analysis of key aspects of disc wing geometry,
aerodynamics and flight dynamics. Section 2.1 provides a derivation of
aerodynamically relevant disc-wing geometric parameters. Section 2.2 develops a
simple aerodynamic model for the lift and drag characteristics of a disc wing based on
finite wing theory and basic drag estimation. Section 2.3 considers the impact of
pitching moment characteristics on the trim and stability of generic tailless flight
vehicles and establishes the pitching moment design requirements for a successful disc-
wing. Finally, section 2.4 examines the fundamental dynamics of disc wing flight and
identifies a non-dimensional roll rate parameter that characterises disc-wing
trajectories.

2.1 Disc-wing Geometric Parameters


The basic disc-wing geometry is defined by its diameter. The root chord c, located at the
mid-span station, is identical to the diameter d of the circular planform and is therefore
taken as the characteristic length for a disc-wing. This allows the definition of the disc-
wing reference area,
2
d
S =π . (2.1)
4

By definition, the aspect ratio of a circular planform is given by,

2
b
AR = (2.2)
S

where b is the wing span at the mid-chord station, such that b = c = d. Therefore for the
circular planform,
4
AR = ~ 1.27 . (2.3)
π

The thickness to chord ratio t h c is defined using the root chord c. The thickness th is
defined as the perpendicular distance of the disc-wing rim tip above the flat central
plate, see Fig. 2.1. This is really a non-dimensional measure of the amount of maximum
camber, used here for ease of comparison between disc-wing shapes within the
literature. Typically a Frisbee has thickness to chord ratio of around,

- 31 -
th
~ 0.15 . (2.4)
c

The mean chord c for the disc planform is given by,

πc
c= (2.5)
4

The mean aerodynamic chord mac, which is the chord weighted chord, is calculated
using,
2 b2 2
S ∫0
mac = c y dy . (2.6)

where cy is the chord length at the relevant span station y and b is the wing span. Using
cylindrical polar coordinates to define the chord at each span station the integral in
equation (2.2) becomes,
4b 0
mac = −
π ∫π
2
sin 3 θ dθ . (2.7)

Therefore the mean aerodynamic chord for the circular planform is,

8b
mac = ~ 0.85c . (2.8)

2.2 Aerodynamic coefficients


It is necessary to define the aerodynamics forces in terms of dimensionless coefficients.
This is achieved by normalising the aerodynamic forces by the dynamic pressure q∞ and
the surface planform area S. The aerodynamic Drag, Side Force and Lift are therefore
defined as,
Drag
C Drag =
q∞ S
Side
C Side = (2.9)
q∞ S
Lift
C Lift =
q∞ S

- 32 -
respectively. Similarly the aerodynamic moments are normalised by the characteristic
length c also such that the aerodynamic Rolling, Pitching and Yawing moments are
defined as,
L
CL =
q∞ S c
M
CM = (2.10)
q∞ S c
N
CN =
q∞ S c

respectively. The Axial, Side and Normal forces are defined here also given by,

X
CX =
q∞ S
Y
CY = (2.11)
q∞ S
Z
CZ =
q∞ S

respectively. The Axial, Side and Normal forces are the aerodynamic force components
along the body fixed axes x, y, z (see Fig. 2.2) and are related to the Lift, Drag and Side
force by,

X = Lift . sin α − Drag . cos α


Y = Side (2.12)
Z = − Lift . cos α − Drag . sin α

where α is the geometric angle of attack.

Pressure is made dimensionless using pressure coefficient Cp defined as,

p w − p∞
Cp = (2.13)
q∞

where pw is the pressure on the disc-wing surface, p∞ is the total free-stream pressure
and q∞ the dynamic pressure defined by,

- 33 -
q∞ = 1
2 ρ ∞V∞ 2 (2.14)

where ρ∞ is the air density and V∞ the free-stream flow speed.

2.3 Lift & Drag Characteristics of Low Aspect Ratio Wings


The lift and drag characteristics of a disc-wing can be determined theoretically using
finite wing theory or lifting line theory (Anderson, 1991). However, it is noted that
finite wing theory is conventionally restricted to aspect ratios greater than 2, whereas
disc-wings have an aspect ratio of 1.27, see equation 2.3. The application of this theory
to the disc-wing is intended therefore only as a first approximation, offering an
insightful starting point.

As a first approximation, it is reasonable to use the lift curve slope for an infinite wing
defined by,
ao = 2 π . (2.15)

It can be shown using an analysis of induced velocity (Anderson, 1991) that the lift
curve slope of a finite span wing is reduced compared to that of an infinite wing with
the same cross section,
ao
a= (2.16)
1 + (ao π eAR)

where a o is the lift curve slope of the infinite wing, a is the lift curve slope of the finite

wing and e is the span effectiveness factor.

By definition, the aspect ratio of a circular planform is 4/π, see equation 2.3. Defining
a o as 2π from equation (2.15) and e as 1 as a first approximation, substitution of these
values into (2.10) yields a theoretical value of the lift curve slope of a disc-wing,


a= ~ 2.44 . (2.17)
1+π 2

Due to the positive camber of a disc-wing the theoretical lift curve is given by,

- 34 -
CL = a (α − α o ) (2.18)

where αo is the zero lift angle of attack. From experiment the value for αo was taken to
be –2o such that,
C L = 2.44α + 0.085 . (2.19)

The aerodynamic drag generated by the disc-wing can be modelled as the sum of profile
drag CDo and induced drag using the familiar drag polar equation,

2
C
C D = C Do + L . (2.20)
π eAR

Due to the bluff nature of practical disc-wing cross-sectional shapes, an estimate of the
disc profile drag coefficient can be made by modelling the disc as a sphere of the same
frontal area as the disc. The profile drag coefficient can then be obtained by scaling
according to the ratio of frontal (subscript f) to planform (subscript p) areas,

Sf
CDo = CDp = CDf . (2.21)
S

The frontal area of the disc can be approximated from the disc centre-line thickness to
chord ratio,
S f = thc (2.22)

and the planform area S is given in equation (2.1). Substituting (2.1) & (2.22) into (2.21)
then gives,
4t h
C Do = C Df . (2.23)
πc

The drag coefficient of a sphere in turbulent flow CDf based on the frontal area is
approximately 0.5 and the centre line thickness to chord ratio of a typical Frisbee is
approximately 0.15, see equation (2.4). Substitution of these values in equation (2.23)
gives,
C Do = 0.0955 ≈ 0.1 . (2.24)

- 35 -
The induced drag component at a given angle of attack can be estimated from known
geometric parameters and the lift coefficient from equation (2.19). Substituting these
values into equation (2.20) gives a drag polar equation,

2
CD = 0.1 + 0.25 CL . (2.25)

Finally, the drag curve can be written in terms of α by substituting equation (2.19) into
(2.25) such that,
C D = 0.1 + 1.5(α + 0.035) 2 . (2.26)

Theoretical results derived in this section are compared to experimental results in


Chapter 5.

2.4 Trim & Stability of Tailless Flight Vehicles


A number of requirements need to be met in order for a vehicle to achieve recognisably
successful flight. The two most fundamental of these are that:

1. The vehicle should be able to generate a lift force at a least equal to its weight.
2. The vehicle should be able to achieve and vary aerodynamic balance such that
orientation of the vehicle with respect to the free stream can be controlled.

If sustained flight is required, then a further condition is that:

3. The vehicle should be able to generate a thrust force at least equal to the drag force.

Lastly, for ‘passive’ flight vehicles without an automatic flight control system or human
pilot:

4. The vehicle should be aerodynamically stable such that when disturbed, it returns to
its original balanced state, set by condition 2 above.

Most conventional flight vehicles satisfy condition 1 using a primary aerodynamic


surface for generating lift (a wing), and conditions 2 and 4 using a secondary
aerodynamic surface (horizontal stabiliser). Whilst the provision of a horizontal

- 36 -
stabiliser, in the form of a tailplane or foreplane, is often beneficial to the overall design,
it is not a necessary condition for balance and stability.

The disc-wing is an example of a wider class of tailless flight vehicles including hang
gliders and delta winged aircraft that use a single aerodynamic surface to both generate
lift, and provide balance and stability.

The longitudinal forces and moments acting on a generic tailless aircraft are shown in
Fig. 2.3. In keeping with linear airfoil theory (Anderson, 1991), the forces and moments
are modelled as a lift force, Lift, and zero lift pitching moment, Mo, acting at the
aerodynamic centre of the section.

For a 2D airfoil section at low speeds and small angles of attack (attached,
incompressible flow), the zero lift pitching moment is constant and the aerodynamic
centre is located at approximately a quarter chord back from the leading edge. The zero
lift pitching moment is negative for positively cambered sections and positive for
negatively cambered sections.

Taking moments about the centre of gravity of the tailless configuration the following
moment balance equation is obtained,

M cg = M o − Lift ( x ac − xcg ) . (2.27)

Non-dimensionalising, (2.27) becomes,

C M cg = C M o − C Lift k n (2.28)

where k n is the static margin, defined as,

⎛ x ac − x cg ⎞
k n = ⎜⎜ ⎟⎟ . (2.29)
⎝ c ⎠

The system is balanced (trimmed) when C M cg = 0 . The lift coefficient for trim is

therefore given by,


- 37 -
CMo
C Lifttrim = . (2.30)
kn

In order to meet the first condition for flight above, C Lifttrim must be greater than zero.

This is achieved if both C M o and k n are positive or if both C M o and k n are negative.

For stability (fourth condition for flight), the change in moment about the centre of
gravity with increase in lift coefficient must be negative i.e.,

∂C M cg
<0 . (2.31)
∂C Lift

Differentiating (2.28) gives,


∂C M cg
= − kn , (2.32)
∂C Lift

thus for stability, k n must be positive.

Combining results from (2.30) and (2.32), it can be concluded that a stable
configuration requires a positive C M o for balance, whereas an unstable configuration

requires a negative C M o for balance. This implies that a stable configuration should

have negative camber and an unstable configuration positive camber.

To illustrate the effects of camber and static margin on the stability and trim of a tailless
flight vehicle, Fig. 2.4 compares the pitching moment characteristics with angle of
attack for the four possible permutations of camber and static margin. It is assumed that
lift is directly proportional to angle of attack.

For a stable tailless aircraft C M o must be positive to provide balance and hence the wing

must have negative camber. This configuration is shown in Fig. 2.4b. For a disc-wing,
the centre of gravity must be at the centre of the disc and in practice the aerodynamic
centre will always be ahead of this point, therefore the static margin will always be

- 38 -
negative. Thus to balance a disc-wing C M o must also be negative, i.e. the camber must

be positive. This configuration is shown in Fig. 2.4c.

Fig. 2.5 shows the aerodynamic centre location as a function of angle of attack for both
a Frisbee disc and a non-axisymmetric circular planform wing with an aerofoil cross
section (Zimmerman, 1935). Over the angle of attack range 0o to 10o the aerodynamic
centre of the Frisbee is just ahead of the half chord point of the disc, giving a static
margin of approximately zero and thus neutral pitch stability. For the circular planform
wing with an airfoil cross-section, however, the aerodynamic centre is at approximately
quarter chord, resulting in a large negative static margin and hence negative stability.

An explanation of the reason why the Frisbee cross-sectional shape results in a


favourable pitching moment characteristic compared to other cross-sectional profiles is
provided in Chapter 6, and forms an important outcome of the work presented in this
thesis.

2.5 Flight Dynamics of a Spin-stabilised Disc-wing


The most distinctive feature of disc-wing flight compared to the flight of other
projectiles or aircraft is the tendency for disc-wings to roll about their direction of
motion. As an illustration consider Fig. 2.6a. For a frisbee-like shape at typical flight
angles of attack, if the centre of pressure (cp) of the disc-wing is ahead of the centre of
the disc i.e. ahead of the cg. This results in an untrimmed nose up pitching moment. If
the disc is rotating, gyroscopic effects dictate that this pitching moment M results in a
precessional rolling rate, p. Using the conventional body fixed axes definition (Fig.
2.6b), for a disc rotating in the direction of positive yaw r then a positive pitching
moment will generate roll rate.

The purpose of this section is to derive a non-dimensional roll rate parameter based on
the disc’s physical properties and launch conditions that can be used to predict the
approximate flight behaviour of the disc. The starting point for the analysis is the
standard set of equations describing the motion of an aircraft with six degrees of
freedom (Nelson, 1998). Assuming a horizontal flightpath, equilibrium of forces and
negligible yawing moment (a reasonable assumption based on experimental results), the
motion of the disc is governed by the rolling moment and pitching moment equations,

- 39 -
L = I x p& − I xz r& + qr (I z − I y ) − I xz pq (2.33)

(
M = I y q& − rp (I x − I z ) − I xz p 2 − r 2 . ) (2.34)

Now for a uniform axisymmetric circular disc,

I x = I y = 12 I z and I xz = 0 (2.35)

reducing equations (2.33) and (2.34) to,

L = I x p& + 12 qrI z (2.36)

M = I y q& − 12 rpI z . (2.37)

Due to the rapid spin of the disc (Hess, 1975) it can be assumed that the angular motion

of the disc is unaccelerated thus,

p& << qr and q& << rp . (2.38)

Therefore equations (2.36) and (2.37) can be written,

2L
q= (2.39)
rI z
2M
p=− . (2.40)
rI z

These equations express the fundamental effect of gyroscopic precession on the


dynamics of a disc-wing. i.e. for a positive spin rate, a positive rolling moment causes
pitch up and a positive pitching moment causes a roll left wing down. Since the
aerodynamic rolling moment on a symmetrical disc is very small (see Chapter 5), the
precessional pitch rate is negligible. However, the pitching moment acting on a disc-
wing may be quite large and hence the roll rate can be significant. The tendency of a
disc to roll in flight is anecdotally known as the “turnover effect” (Schuurmans, 1990).

- 40 -
2.5.1 Non-dimensional Roll Time Constant
It is proposed that non-dimensionalised disc-wing trajectories from similar starting
conditions will be nominally similar if the non-dimensional roll rate is similar.

For the present analysis, the non-dimensional time and non-dimensional distance are
given by equations (2.41) and (2.42) respectively, where t is dimensional time and l is
dimensional length,
tg
tˆ = (2.41)
V∞
lg
lˆ = 2 . (2.42)
V∞

The non-dimensional disc roll rate is thus obtained as follows,

pV 2MV∞
pˆ = =− . (2.43)
g rI z g

Using equation (2.28) and assuming that M o for the disc is zero, the disc pitching
moment can be approximated as,

M = −k n C Lift q∞ S c . (2.44)

The mass moment of inertia Iz of the disc about the z-axis will depend on the mass
distribution. To put the moment of inertia into context, it is useful to consider the
equation,
2
I z = m k zz (2.45)

where kzz is the radius of gyration.

Values of kzz for various disc-wing shapes become larger when the weight is removed
towards the circumference. Firstly, consider a circular cylinder with negligible height
which has radius of gyration kzz = (c/2)√2 ~ 0.71(c/2). Secondly, as the mass is
redistributed towards the circumference kzz increases, a typical value for a Frisbee-like

- 41 -
disc-wing is kzz = 0.86(c/2) (Hubbard & Hummel, 2002). Lastly, for the theoretical case
of a flying ring with the total mass distributed evenly at the outside radius kzz = 1(c/2).

For the present analysis, however, it is assumed that the disc is a uniform circular
cylinder of negligible height for which,

1
I z = mc2 (2.46)
8

where m is the overall mass of the disc.

Substituting equations (2.44) and (2.46) into (2.43), the following result is obtained,

16 k n C Lift q ∞ S V∞
pˆ = . (2.47)
mg r c

Given that for the condition Lift = Weight,

mg
C Lift = (2.48)
q∞ S
and that,
rc
AdvR = (2.49)
2 V∞
equation (2.47) simplifies to,
8k n
pˆ = . (2.50)
AdvR

Equation (2.50) states that the non-dimensional roll rate of a disc-wing in unaccelerated
horizontal flight is proportional to the ratio of static margin to the advance ratio. Thus
roll rate is minimised when:

1. The aerodynamic centre of the disc is closest to the centre of the disc (minimum
k n ) i.e. a well designed disc.
2. The disc is launched with a high spin rate compared to the forward velocity (high
advance ratio).

- 42 -
Both these conditions above are consistent with everyday experience of Frisbee
throwing. The effect of varying p̂ on the flight path of a disc-wing is investigated
numerically in Chapter 7.

- 43 -
Chapter 3 Literature Review

Summary
This chapter provides a review of the disc-wing aerodynamics and flight dynamics
literature. The review is split into two sections: the first on the aerodynamic
characteristics of circular planform wings and the second on the flight dynamic
characteristics. The first section includes aerodynamic data from both spin-stabilised
(axisymmetric) and non-spin-stabilised (axisymmetric) disc-wings. The second section
discusses the dynamics of spin-stabilised discs only, focusing on the Frisbee and discus.

3.1 Opening Statements


The flying sports disc is of fundamental interest to the aerodynamicist, however the
peripheral nature of such a shape to mainstream aeronautical applications has ensured it
has largely escaped scientific scrutiny. As such, the literature base detailing the
aerodynamics of circular planform wings is patchy, with much of the key work often in
obscure sources and unpublished reports.

3.2 Aerodynamics of Circular Planform Wings


This section provides a descriptive review of the aerodynamics literature for circular
planform wings. A quantitative comparison of results from the various sources is
included later as part of the discussion of experimental results from the present work in
Chapter 5.

3.2.1 Frisbee
Arguably the first scientific study into disc-wing aerodynamics resulted from a U.S.
Navy project considering the development of a self-suspended flare. The proposed flare
was essentially a spin-stabilised axi-symmetric flying disc (Stilley, 1972; Stilley &
Carstens, 1972). Familiar shapes such as the Frisbee, clay pigeon & right circular
cylinders were tested in the wind tunnel alongside various flare configurations. Test
results were obtained for a non-spinning frisbee-like model (hollow) and published as
typical plots of axial, normal and pitching moment coefficients against angle of attack.
The load measurements were far from rigorous, data points were few, enough just to
capture the general trends. The drawback of barely sufficient data is that it leaves the
reader wondering what happens between data points, particularly in the vicinity of the

- 44 -
aerodynamic stall. The normal curve for the frisbee-like model is linear below the stall
at around 35° AoA and the pitching moment, taken about the semi-chord position, is
linear with positive gradient trimmed just above 10° AoA.

The effect of the cavity provided a large change in nose down pitching moment, when
comparing the solid (cavity filled) and hollow flares. The effect of spin on the
aerodynamic loads was investigated on a right circular cylinder (th<<c) and found to be
negligible, for the purposes of their work. Their technique for the measurement of
‘Magnus’ aerodynamic loads was a complicated test model support structure. The
present author offers the suggestion that their side sting mount approach introduces
large interference effects particularly at high angles of attack, to account for the
premature onset of aerodynamic stall. They concluded that the accurate measurement of
aerodynamic loads due to spin was problematic, recommended as an area for further
investigation. Aerodynamic damping measurements for the Frisbee (hollow) were
considered very small, in comparison to other conventional and unconventional bodies,
with a pitch damping coefficient of around 0.5 for the positive angle of attack range.

Lazzara et al (1980) described a project to develop a wind tunnel balance and measure
the aerodynamic loads acting on a Frisbee-like flying disc. Lift and drag results were
presented for various flow speeds and spin rates over a narrow angle of attack range (0°
to 10°). They concluded that spin generates a small lift component, although this may
be a simple result of experimental accuracies.

Ali (1998), in an undergraduate project at the University of Manchester, measured lift,


drag and pitching moment of a spinning Frisbee over a range of angles of attack and
spin rates. This study was a precursor to the present study, providing a useful starting
point for the design of a wind tunnel mounting rig for a spinning disc, see Fig. 4.4.

Mitchell (1999), as part of his Masters thesis at the University of Nevada, measured lift
and drag for three non-spinning disc-wing configurations at various flow speeds. The
results were plotted for each AoA (–20° to 20°) separately in the form of L/D ratio
against Re. The vastly different curves were attributed to the varying camber and
contour of each disc-wing. These graphs suggest that all three discs were strongly
Reynolds number dependent, over the entire range of angles and flow speeds tested.
The present research findings however are quite the opposite, the aerodynamic loads
- 45 -
were found to be Reynolds number independent, for the (more comprehensive) range of
flow speeds tested. Flow visualisation using tuft and smoke techniques enabled the
observation of the upwash ahead of the leading edge, the downwash aft of the trailing
edge and the existence of trailing edge vortices, although the images included in the
report are merely clear enough to see their existence and nothing more. The flow over
the upper surface was described to be completely attached throughout whereas the lower
(concaved) surface was entirely detached and separated. The effect of spin was deemed
to have no effect on the development or location of boundary layer flow structures.
Although the above statements regarding flow visulisation are perhaps correct for low
angles of attack the present research offers discussion and evidence to suggest
otherwise. The upper surface flow includes a separation bubble, the cavity shear layer
reattaches to the cavity at low angles of attack (5o) and although the broad flow
structures over the spinning disc are similar they are far from identical to the non-
spinning case. Most worthy of note here, is that the near surface flow direction within
the separation bubble, on the spinning disc, is dominated by the movement of the
surface at that locality.

Higuchi et al (2000) investigated the flow over a similar disc-wing (golf disc) to that
tested in the present study, using smoke wire flow visualisation and PIV (particle image
velocimetry) measurements. A laser light sheet was used to illuminate various flow
structures including trailing vortices, the separation bubble and upper & lower (cavity)
surface streamwise flow cross-sections, both on spinning and non-spinning discs. At
high angles of attack (30o) the separated shear layer was reported to be forced
downwards by the strong trailing vortex pair, which is consistent with the findings of
the present study. Vortex strength was calculated from PIV results. When compared to
the non-spinning baseline case, the vortex strength remained unchanged for a spinning
disc at a low angle of attack (5o) but decreased at a higher AoA (15o). A reduction in
circulation was stated to cause the loss of vortex strength, attributed to the separation
bubble becoming larger with spin. Enhanced lift at low AoA (5o) and reduced lift at
higher AoA (15o) is described as being due to the effect of spin, based on observations
of the wake. Load measurements from the present research, for an equivalent Reynolds
number (10m/s), confirm that the lift enhancement, as a result of spin, decreases with
increased AoA, Fig 5.15b. At no point is the lift reduced however, although the shape of
the golf disc is slightly different to the disc tested in the present study and as such, it is
possible that the two shapes exhibit contrasting aerodynamic loads. The reduced

- 46 -
strength of the two trailing vortices with spin is attributed quite rightly to the change in
effective camber on advancing and receding surfaces. This paper also incorporated flow
visualisation results, proposing a flow topology for a flat disc at incidence. Their wind
tunnel model comprised a right circular cylinder with rounded edges, similar to the
‘coin-like’ cylinders studied by Zdravkovich et al (1998).

Yasuda (1999) measured the lift and drag characteristics of a recreational sports disc,
analogous to the one tested in the present study, and a flat plate disc also. The wind
tunnel results were taken for various flow speeds & spin rates but were consistently
lower than the measurements taken in the present study, across the board, for both the
cambered disc (see Figs 5.10, 5.11) & the flat plate also. Systematic balance errors or
coefficient calculation errors could account for this. Both spin and roughness strips
(ridges) were found to have no effect on the load measurements.

Nakamura & Fukamachi (1991) visualised the flow past a Frisbee using the smoke wire
method. The observations are as follows: A horizontal plane of smoke filaments,
aligned with the upper surface of the disc, shows a symmetric wake for the non-
spinning case. A vertical smoke grid downstream of the disc shows a pair of trailing
vortices formed behind the disc, the effect of spin causes an asymmetric wake. They
concluded that the spinning disc strengthened the trailing vortex pair thereby enhancing
the downwash. Although an asymmetric downwash is commonly observed in the wake
of a spinning disc, the Advance Ratio (AdvR = 2.26) was not matched to that which
would be typical in flight, i.e. AdvR < 1. The present study has shown that the lift
increases with high AdvR (>> 1) at low Re numbers (see Fig. 5.15a), which confirms
this conclusion. However, with increased Re number, equivalent to typical free-flight
conditions, the AdvR is more weakly dependent upon spin. This suggests that the vortex
strength enhancement is dependent upon AdvR. However this was overlooked due to
the study being limited to a single test case for a spinning disc.

3.2.2 Discus
Independent studies by Ganslen (1964) and Tutjowitsch (1976) measured the lift & drag
characteristics for the discus from field athletics. Ganslen’s (1964) load data (lift and
drag) shows variation with velocity i.e. Reynolds number change, especially at high
AoA (30o to 50o). His smoke flow visualisation (flow speed 18 m/s) depicts the discus

- 47 -
flow field both in planform and cross-section at various AoA including the trailing
vortices, separation bubble and deep stall flow structures.

Kentzer & Hromas (1958) measured the pitching moment, in addition to the
aerodynamic forces. The effect of spin on the aerodynamic loads has also measured
although the presented spin rate (2.5 rev/sec) is well below that for typical discus
throws, 5 to 8 rev/sec (Ganslen, 1964). The effect of spin does not change the lift and
drag curves except to prolong the stall (24°) by at least 4°.

Much of the load data in the available reports present measurements which agree only in
their general characteristics. The researchers disagree on stall AoA, there are
discrepancies in lift curve slope and the form of the drag curve also. It is interesting to
see the decrease in drag at stall measured by both Kentzer & Hromas (1958) and
Tutjowitsch (1976) however Ganslen (1964) does not report any such thing. None of the
above investigators have presented a thorough study of the effect of spin on the
aerodynamic loads over a range of typical flight speeds. The only offering being from
Kentzer & Hromas (1958) who presented the lift, drag and pitching moment for one
speed and one non-zero spin rate only.

No-one to the knowledge of the author, has measured the spin dependent aerodynamic
rolling moment, yawing moment and side force. There is a need for a more definitive
aerodynamic study of the discus to corroborate the present experimental work currently
available on the subject.

3.2.3 Coin-like Cylinders


In related work, Zdravkovich et al (1998) studied the aerodynamics of what he called
‘coin-like cylinders’. These have a thickness to chord ratio th/c << 1, where the
thickness th is the cylinder height and the chord c is the diameter of the cylinder. A
talcum powder-paraffin film was applied to the cylinder surface to visualise the surface
flow. With its planform parallel to the free stream, the flow over the cylinder revealed
surface patterns which are similar to those observed during the present experimental
work, namely a semi-circular separation line and straight line reattachment. This gave
an insight into the interaction of flow structures over the cylinder and allowed the
specification of the flow topology. Unfortunately the study was limited to 0o angle of
attack and as such has limited relevance in the context of the present study.

- 48 -
3.2.4 Analysis on Span Effectiveness factor
As a way of evaluating the relevant aerodynamic load data, within the literature, for
circular planform wings, a graph of CD against CL2 was plotted in Figure 3.1. The
aerodynamic load data for various circular wing shapes are compared to the present
disc-wing measurements in Figures 5.5 to 5.12 and discussed in full in Chapter 5.
However, the graph in Figure 3.1 is considered here as a way of analysing the Span
Effectiveness factor e for circular planforms. The Span Effectiveness or Span Efficiency
factor e is a measure of the drag induced by a finite wing, the theoretical maximum
being e = 1 for an elliptical lift distribution which yields minimum induced drag,
Anderson (1991). The curves plotted in Figure 3.1 are for a variety of different
geometries with chordwise cross-sections as seen in the key next to the figure. These
geometries are sports implements (reviewed in section 3.2) and circular wings from
aerospace applications (reviewed in appendix B).

To understand the similarities between various circular wing shapes, the gradient of
respective curves are compared by means of the span efficiency factor. Applying
equation 2.20, the gradient of each curve in Figure 3.1 is given by (π e AR)-1 and the AR
is identical for every wing with circular planform. Therefore the AR is 4/π and the
gradient is given by 0.25 e-1. Taking the gradient of the experimental curves in Figure
3.1 we can calculate the span effectiveness e.

The values of e for the circular planform geometries compared in Figure 3.1 vary
between 0.5 and 0.8. The graph shows Frisbee-like wings with approximately the same
span effectiveness e ~ 0.8, namely Stilley & Carstens (1972) and Ali (1998), however
Zimmermans (1935) planar wing also has similar gradient or in other words e ~ 0.8. The
graph also shows Frisbee-like wings with contrasting gradients, namely Stilley &
Carstens (1972) and Yasuda (1999) have span effectiveness e ~ 0.8 and 0.6,
respectively. However, recall that the measurements Yasuda (1999) took are expected to
have systematic errors, which would account for the lower span efficiency here, e ~ 0.6.
There are also examples of similar planar wings with approximately the same span
effectiveness e ~ 0.5, namely Greif & Tolhurst (1963) and Tutjowitsch (1976).

In general, the curves for Frisbee-like wings have similar magnitudes and similar span
effectiveness e ~ 0.8, see Figure 3.1. However, circular planar wings with thin cross-
section also have similar span effectiveness to the Frisbee-like wings, e ~ 0.8. Circular

- 49 -
planar wings with thick cross-section have much lower span effectiveness, e ~ 0.5. The
Frisbee-like shape therefore has large span effectiveness, relative to other circular wings
of identical aspect ratio, within the literature. The Frisbee-like shape therefore has
similar induced drag to a thin circular wing, such as the Zimmerman (1935) Clark Y,
but lower induced drag than thick solid wings, such as the discus. This is an indicator of
the inherent flow structures created over a circular wing, namely trailing vortices in
close proximity which induce a strong central downwash with associated spanwise lift
distribution and thus span efficiency factor. However, the fact that thicker circular
wings have lower span efficiency suggests that the presence of the cavity on the
Frisbee-like shape, returns lower induced drag back down to a value similar to planar
wings. However, even though they both have a similar span efficiency factor, the
spanwise lift distribution for Frisbee-like and planar circular wings could well be
different.

3.3 Flight Dynamics of Spin-stabilised Disc-wings


3.3.1 Frisbee
Katz (1968) developed a simplified mathematical model of a rotating flying disc
treating the spin as an independent parameter. Firstly, considering linear aerodynamic
moments alone, the derivation of classical stability criteria was given, including steady
state results. A non-linear model was derived by adding lift, drag and gravity forces, so
that typical flight trajectory solutions could be calculated numerically. From the stability
analysis, the increased spin rate was shown to [spin-] stabilise the motion towards a
constant precession rate condition. This work is sketchily written and difficult to follow
with no formalised conclusions. The flight trajectories are largely two-dimensional and
difficult to visualise with scant information on what is being presented in the figures.

Lissaman (1994) considered the dynamics of a spinning oblate spheroid including


mathematical stability analysis. The development of a characteristic dynamics matrix
from the equations of motion and subsequent decomposition into dynamic groups
enabled the modes of the dynamic motion to be determined from the roots of the
characteristic equation. A formal dynamic analysis approach enabled the interpretation
of the dynamic modes with specific reference to the flying sports disc. Numerical results
for a Frisbee disc, with launch parameters chosen carefully to match a realistic throw
release, were analysed and compared to Stilley & Carstens (1972) data. The dynamic

- 50 -
modes for both discs were found to be quite similar, their behaviour supported by actual
flight tests (Stilley & Carstens, 1972).

Revisiting disc flight dynamics, Lissaman (1998) presented the linearised equations of
motion and discussed various numerical codes used to solve them. However problems
arose when attempting to integrate the flight equations, it was noted that very small time
steps were required to maintain convergence which suggests more computational power
was needed. Lissaman (1999) also described the aerodynamics at zero spin rate & non-
zero spin, the gyroscopic dynamics and both the longitudinal & lateral flight trajectory.
Much of the discussion was based on wind tunnel results from the early data collected
for present study. An approximate solution of the longitudinal and lateral trajectories
was described but not presented, aerodynamic coefficients were taken from the present
research. The spin dependent rolling moment was described to be caused by the delayed
separation on the retreating side and earlier separation on the advancing side which
makes the spanwise lift distribution asymmetric. However, the local surface moves
across the flow on the leading edge of a spinning disc, directly along the arced
separation rather then advancing or retreating. Therefore it is not possible to treat the
fluidic generation of the rolling moment so simplistically, the rolling moment is
generated by a unique combination of interacting flow structures over an axisymmetric
body with a rotational slip condition. Latterly, Lissaman (2001; 2003a; 2003b)
considered the maximum range of a flying disc compared to other well understood
projectiles, mainly spheres.

Hummel & Hubbard (2000; 2001; 2002; 2003; 2004) analysed both the Frisbee throw
and flight, from throw biomechanics through the development of a numerical flightpath
simulation, even to the extent of identifying aerodynamic coefficients from free-flight
tests. Hubbard & Hummel (2000) developed a 5dof simulation and demonstrated basic
Frisbee flight trajectories successfully. Hummel (2003) compared simulated trajectories
to experimental flight path data (f2302), as in the present study (Fig. 7.3), to validate the
simulation.

The simulation included a (linearised) dynamic model for the identification of


parameters from flight tests (Hubbard & Hummel, 2002). Free-flight experiments
accurately tracked the position of a Frisbee in three dimensions using high-speed video
cameras. Linear approximations to aerodynamic coefficients were derived from the

- 51 -
flight data and compared reasonably well to wind tunnel data from the present study, a
comparison is given in Figure 5.7.

Estimation of parameters from flight tests is significantly more challenging than wind
tunnel testing and there is considerable scatter in the results obtained. Furthermore, it is
only possible to obtain parameters linearised around a nominal flight condition. That
said, the only experimental data currently available on the rate damping derivatives
(Hummel & Hubbard, 2000; 2002) is from flight tests, meaning the technique provides
an important input for simulation work.

The biomechanics of Frisbee throwing (Hubbard & Hummel, 2001) provides useful
information on the range of launch conditions (attitude, velocity and spin rate)
physically possible and, coupled with the use of simulation, allows their optimisation
for specific performance goals.

With maturity this study has the potential to accurately simulate the Frisbee throw and
resulting flight trajectory, with the power to yield the optimal throwing technique for
various purposes, such as maximum range. It could also be used to analyse the throwing
technique of specific athletes. In presented work however, the simulation did not
progress much further than the validation and the biomechanics model was not taken far
enough to couple the two.

Pozzy (2001) investigated the correlation of throw speed with range by clocking the
disc release speed at launch, from a professional field of disc golf players, with a radar
gun. The range was plotted against launch speed on a scatter graph and a best fit line
drawn to show the correlation. Although it is no great surprise that the range increases
with higher release speeds, this is a useful resource for the comparison of simulated
predictions. Lissaman (2003b) recently considered the 2D flight dynamics to find the
upper bounds for maximum range via an optimisation procedure. The solutions were
plotted alongside field results (Pozzy, 2001) and found to have reasonable agreement. In
both cases the dependence of range on velocity was modelled with a straight line, in
spite of the quadratic dependence of the launch speed on the range. Lissaman (2003b)
noted that this was due to energy dissipation by drag at high velocities which reduced
the favourable effects of lift. It is helpful to bare in mind that the golf throw data was
collected in various locations, wind conditions and topology. This could explain the

- 52 -
cluster of scattered points that represented long range achieved for moderate launch
speed, which was attributed to throwing technique (Pozzy, 2001).

Cotroneo (1980) analysed biomechanic and aerodynamic aspects of disc flight from
throw observations. The study was focussed on the comparison of back-hand and side-
arm (fore-hand) throwing technique for maximum distance, including a biomechanic
analysis of the athlete subjects with contrasting throw actions yet exceptional range.
Minimal wind tunnel test results for a Frisbee without the concentric rings [that forced
boundary layer transition] on the upper disc surface, reduced the optimum release angle
of attack and improved the lift to drag ratio. Release velocities were correlated with
range to confirm that initial velocity is the most important factor affecting the maximum
distance thrown. Both back-hand and sidearm throwers could achieve approximately the
same range. Pozzy (2002) also investigated the throwing technique, analysing high-
speed camera footage of disc golf professionals’ launch strategy.

Recently, Lorenz (2004) began to investigate the free-flight dynamics of a Frisbee via
measurements from onboard instrumentation. A varied array of real-time data was
recorded in flight from pressure sensors, accelerometers, an infra-red sensor, and a
magnetometer mounted within the cavity of a Frisbee and uploaded to a computer post
flight. The flight trajectory was recorded using a conventional video recorder, digitised
and converted into physical distance. The body attitude was calculated from data taken
with the various sensors and velocity computed from the video record. The aerodynamic
force coefficients could then be calculated from the instantaneous accelerations and the
pitching moment from the precessional roll rate, which was derived from the attitude
record. The aerodynamic loads were compared to the wind tunnel data of the present
study and were in agreement. However, only two points were plotted on each graph of
lift, drag and pitching moment and as such were merely only verification that the
instrumentation was providing realistic data. The discrepancy between the free-flight
load data and the wind tunnel data, of the present study, was at times large in drag and
pitching moment but still well within the vicinity of the wind tunnel data to profess a
successful correlation. The ongoing research by Lorenz, using this onboard
instrumentation methodology, is potentially an extremely useful way of gathering
aerodynamic data in free-flight or even within the wind tunnel environment. Future
work is proposed to determine the effect of spin on the separation and measure the
surface pressure distribution on a spinning disc.

- 53 -
Danowsky & Cohanim (2002) sought to develop a computer model that predicted
aerodynamic parameters and use these to simulate free-flight trajectories, which was a
similar study to that of Hubbard & Hummel (2000). Instead of deriving parameters from
free-flight experiments, they used potential flow theory that was slightly modified to
account for the spin.

A wind tunnel balance was constructed from a combination of load cells to give 3dof,
set in two positions one which picked up Lift, Drag & Pitching Moment and the other
Lift, Side Force & Rolling Moment. A golf disc (Frisbee) was fixed to the balance via a
motor driven axle, at 0o angle of attack, to test at various flow and spin rate
combinations. The flexibility of the plastic disc particularly at the centre meant that it
deformed during high wing loading, observed most dramatically at zero spin but also
causing a nose-down orientation for the spinning disc. The solitary angle of attack test
case, deformity of the test model and the omission of matched advance aatios over the
flow rate test range, limit the accuracy of the data. Smoke wire flow visualisation was
used to record the port and starboard sidewash angles from the spinning disc.

Potential flow theory was applied to a virtual disc geometry, similar to that tested in the
wind tunnel, to predict aerodynamic parameters. The vortex lattice method was applied
to the disc, accounting for spin by the use of a slip boundary condition. The sidewash
angles were plotted against advance ratio and were deemed to be dependent parameters,
however their plots show two scattered arrays which suggested independence. These
angles were required by the computer code to specify the trailing edge. Steady state
solutions were obtained iteratively using a wake evolution methodology from which
aerodynamic forces could be computed. Simulated flow visualisations based on the
wake solutions show the presence of trailing vortices.

The inputs to the flightpath model were derived from the zero angle of attack case alone
and as such the accuracy of the trajectories is somewhat dubious. Never-the-less, for
high spin rates the plotted trajectories modelled the gyroscopic roll stiffness, the disc
held its orientation in a much straighter lateral flightpath. The effect of the bank angle
launch condition was also illustrated.

- 54 -
The accuracy of the results obtained during this research is questionable, however this is
due to the sheer breadth of the subject matter and analysis techniques employed here, in
a short period of time. The wind tunnel tests and the potential flow solutions were
limited to the zero angle of attack, which restricted the derivation of aerodynamic inputs
solely from this single orientation. The application of potential flow and wake evolution
to the Frisbee has, to the knowledge of the author, never been done before. It is a shame
that they didn’t have time go into more depth.

Tuck & Lazauskas (2004) applied a general lifting surface computer code to the circular
planform, in order to compare lift and pitching moment to semi-analytical solutions.
Expanded solutions to include axisymmetric discs modelled a simple Frisbee-like
profile closely. A combination of these solutions were then used to outline another set
of axisymmetric cross-sectional profile solutions that eliminated the pitching moment
entirely but as a result drastically reduced the lift too. Eliminating the pitching moment
is a highly desirable property of flying discs as it eliminates precessional roll. It is good
to keep in mind however, that these solutions are theoretical and as such do not model
turbulent and separated flow regimes which dominate the flow over a flying disc in air.

3.3.2 Discus
The earliest study of discus flight (Taylor, 1932) was initiated by the Intercollegiate
Associations of Amateur Athletics of America (ICAAAA) in response to a puzzling
question posed by discus throwers who noticed that they could achieve greater distances
when throwing into a headwind. Taylor (1932) investigated this observation using wind
tunnel tests and a flight path simulation to offer practical guidelines for intelligent
throwing in a prevailing wind. The wind tunnel data and details of the mathematical
analysis were not published however, only results of predicted (2DOF) longitudinal
flight-paths based on a variation in ambient wind velocity. The model predicted the
optimum attitude (35° to the horizontal) and angle of attack (0°) given to the discus at
launch to achieve maximum range, for still (no wind) conditions. Using the no wind
case as a baseline, Taylor’s (1932) calculations suggested that a headwind increased the
range whereas a tailwind decreased the range.

As the discus is released for an efficient throw in still air, it is oriented to maximise lift
and minimise drag in the climb portion of the flight (Frohlich, 1981). Investigators
roughly agree that optimal release angles are around 30° attitude and -10° AoA
- 55 -
(Bartlett’s Table 1, 1992). Although this results in negative lift during the initial portion
of flight (Ganslen, 1964; Frohlich, 1981), the discus makes use of the low drag, negative
AoA range to minimise deceleration over the climb portion of the flight. The initial
upward momentum on release, projects the discus away from the ground and it isn’t
until the discus reaches the zero lift AoA (0°) that there is any aerodynamic lift
contribution. The discus generates lift throughout the middle portion of the flight at
highest elevation, with a higher rate of deceleration due to the increased drag at high
AoA. As the discus descends during the latter portion of the flight, the drag approaches
a maximum further retarding the speed as the discus flight path approaches vertical back
to ground. A negative AoA at launch is arguably impractical (Samozwetow, 1960) as it
limits the speed and stability given to the discus on release. However, Terauds (1978)
reported a launch AoA range of –10.5° to –27.5° for elite male discus throwing athletes,
which supports the calculated optimum AoA values reported in the literature.

Contrary to the common assumption that the discus attitude is constant throughout the
flight, small aerodynamic moments will cause the disc to pitch and roll. Samozwetow
(1960) reported the roll left wing tip down and subsequent banked left turn towards the
end of the discus flight, caused by the (nose up) pitching moment. As the discus rolls
the AoA is reduced, Hubbard (1989) suggested that this may cause a favourable change
in pitch attitude gaining aerodynamic advantage in flight. Tutjowitsch (1976) claimed
the discus bank angle could be ignored given high spin rates. However Soong’s (1976)
simulation predicted the range to be spin dependent (Soong’s figure 5a, 1976) for
typical and atypical discus throw spin rates, from 4 up to 740 rev/sec. Voigt (1972) also
reported an increase in range with increased spin rate, based on flight test observations.

- 56 -
Chapter 4 Experimental Facilities and Techniques

Summary
This chapter presents the experimental facilities, apparatus, data acquisition techniques
and wind tunnel corrections. Section 4.1 outlines the wind tunnel facilities. Section 4.2
describes the apparatus constructed to mount the disc-wing models in the wind tunnel
and the various models themselves. Section 4.3 outlines the techniques used to acquire
aerodynamic load & pressure data, and also includes the flow visualisation
methodologies. Section 4.4 discusses the blockage corrections applied to the drag data.

4.1 Wind Tunnels


4.1.1 Goldstein Laboratory
The Disc-wing wind tunnel research was carried out at the Goldstein Laboratory, a self-
contained experimental fluids and aeronautics facility located at Barton Aerodrome,
approximately 8 miles from the primary University of Manchester Campus. Established
in 1946 by Sidney Goldstein as part of the Mathematics Department the Fluid Motion
Laboratory later became the Goldstein Aeronautical Engineering Laboratory, now as
part of the Department of Aeronautical Engineering.

4.1.2 Project Tunnel


The primary low speed wind tunnel used for this research was an open-circuit blow
down tunnel with a test section of 0.9×1.1m, a top speed of 50m/s and turbulence level
of 0.5%. The tunnel is fitted with a six-component balance above the working section
and a pressure transducer with scanivalve measurement system, see Fig. 4.1. Outputs
are fed via a digital converter to a computer located in the control room (Fig. 4.2), all
measurement equipment was operated from here. The balance had 360o yaw control and
an incidence arm with ±30o range.

4.1.3 9’ × 7’ Tunnel
A second low speed wind tunnel had a closed-circuit with a test section of 2.1×2.7m (9’
× 7’), a top speed of 70m/s and a turbulence level of 0.1%. This was used for the smoke
wire experiments due to the superior flow quality at low speeds.

- 57 -
4.2 Apparatus
4.2.1 Rigs
A number of metal frames were used to mount the disc-wing in the wind tunnel in
various configurations. The first (Fig. 4.3) was an L shaped arm with the disc mounted
vertically on a horizontal axle supported by a vertical strut. The disc was mounted, with
its planform vertical, on a motor driven axle to test at various spin rates, the disc’s
centre of mass remained at the balance centre at all times.

The second rig was used for surface paint flow visualisation and held the disc in the
horizontal plane (Fig. 4.4). It consisted of two vertical struts connected to a horizontal
crosspiece. In the centre of the horizontal bar, an axle was mounted vertically, at zero
incidence, on which the disc could be spun with its planform horizontal. Incidence was
adjusted using an incidence arm.

The first support strut was also adapted for use in the 9’ × 7’ tunnel (Fig. 4.5a). This
was necessary for the smoke wire flow visualisation work as the flow quality in this
wind tunnel was superb. Modified parts were designed to fit the versatile L-shaped strut,
including a geared down motor (for fine incidence control) and side sting mounted disc
(Fig. 4.5b), for the purpose of visualising the full central flow field cross-section of a
non-spinning disc.

4.2.2 Disc-wing Models


The geometry of the disc-wing used for the majority of experiments was based on that
of a FloaterTM sports disc manufactured by Dynamic DiscsTM. The half cross-sectional
profile can be seen in Fig. 2.1, with the axis of symmetry vertical and on the left of the
figure.

The model used for spinning and non-spinning load measurements was machined from
aluminium and dynamically balanced by hand. Commercially available plastic discs
were adapted for taking surface pressure measurements and flow visualisation studies.

Other test models were a flat plate disc and a semi flat plate disc i.e. an intermediate
disc-wing model between the flat plate and the Frisbee. Both were formed from flat
aluminium sheet to the profiles shown in Fig. 4.6.

- 58 -
4.3 Data Acquisition
4.3.1 Load Data
The load measurements for the disc-wing were taken using a six component overhead
balance in the 0.9×1.1m wind tunnel. The disc-wing was tested over a range of
Reynolds numbers from 1.13×105 to 3.78×105, corresponding to a speed range of 6m/s
to 20m/s, with an angle of attack range from –100° to as much as 100° and spin rates up
to 24Hz or 1440rpm.

The aerodynamic forces acting on the rig and disc induced mechanical moments due to
the off-centre lone strut configuration. This had a significant effect on the pitching and
rolling moment measurements. Taking measurements of the moments caused by the
static loading of the strut at the centre of the balance, these mechanical components
were removed.

Interference and tare effects due to the strut were measured with the disc mounted on a
dummy support, Fig. 4.7. The dummy strut was a mirror image of the measuring strut
and held the disc in the correct position, on the balance centre. The aerodynamic loads
acting on the wind tunnel support strut (including interference effects from the dummy
strut mounted disc) are presented in Fig. 4.8 alongside results for the test model itself.

4.3.2 Surface Pressure Data


Two pressure tapped disc models were fabricated, one for upper surface pressures, the
other for lower surface pressures. The necessary pressure tubing used for surface
pressure measurement, connected the pressure tappings to the scanivalve. The capillary
tubes were carefully wound around the L shaped rig, as shown in Fig. 4.9, to limit
interference. The tappings were arranged in a curved line (Fig. 4.10a) to both facilitate
the manufacturing process and reduce boundary layer interference effects on down
stream tappings. The pressure across the entire disc surface was measured by yawing
the model at 12° increments to achieve effective coverage (Fig. 4.10b) providing 1200
measurements in total, for each set of conditions. Tests were performed at 20m/s
through an AoA range of -10° to 30° and at 5° for velocities 6, 10, 15 & 20m/s.

- 59 -
4.3.3 Flow Visualisation
4.3.3.1 Surface paint
A film of fluorescent paint, made up of a mix of two parts kerosene to one part
fluorescent powder, was applied liberally to the surface of the disc. The disc was then
mounted in the wind tunnel at the required angle of attack and the tunnel run until the
kerosene had evaporated. Streak lines in the resulting surface patterns indicate the time-
averaged orientation of the surface flow direction and reveal important surface flow
structures, such as separation lines. Flow patterns were photographed outside the wind
tunnel under ultra-violet rich light. Note that for the flow visualisation studies, the disc
was mounted in the horizontal plane to minimise the effects of gravity on the surface
flow patterns. The surface oil flow technique is, by its nature, only suitable for the non-
spinning disc.

4.3.3.2 Smoke wire


There are many methods for generating smoke filaments such as the smoke wire, smoke
wand and smoke rake. The smoke wire technique was chosen for this investigation as it
could provide many thin smoke filaments, ideal for the visualisation of a complex 3D
flow such as this. The smoke is generated by the resistive heating of a wire causing the
vaporisation of oil droplets. As the flow passes the wire it creates a smoke filled wake
behind it. With a small diameter wire (0.1mm) and at low speeds (<2m/s) the wake is
laminar, producing clearly defined smoke filaments. The most rewarding results from
the smoke wire technique are achieved by a careful balance of flow speed, electrical
power and oil supply. As with any other flow visualisation technique the captured
images are only as good as the camera and lighting equipment.

A vertical wire was mounted upstream of the disc with a pressurised oil reservoir
connected allowing continuous feed (Fig. 4.11). The nichrome flat wire (0.1×0.4mm
rectangular cross-section) was electrically heated causing the oil to vaporise producing
smoke filaments. The wire was aligned with the flow so that the narrow edge pointed
upstream, this limited the turbulence generated behind the wire. A number of models
were constructed so that the disc-wing could be mounted in both the vertical and
horizontal planes. This allowed the wire to remain vertical and therefore continuous
smoke filaments could be generated due to the gravity oil feed system. The disc was
tested at a maximum flow speed of 3m/s and 0° to 50° incidence. At speeds greater than
3m/s smoke filaments from the wire became turbulent.

- 60 -
An oil feed system (Fig. 4.11a) was set up on the wind tunnel to provide a continuous
supply to the smoke wire. A regulated cylinder filled with compressed nitrogen
pressurised an oil reservoir to 10psi, this was held above the tunnel. The reservoir
pressure forced the oil to flow along the capillary tubing and dripped oil onto the smoke
wire through the feeder device (Fig. 4.11b). The oil feeder was a simple and effective
device that dripped oil onto the wire through a narrow bore steel tube, this also provided
a contact for the power supply. The wire was electrically heated by a power supply
connected at each end and the base of the wire was weighted to keep it taut.

The flow field cross-sections were illuminated using 1000 watt halogen spot lamps
positioned on opposite sides of the disc. The cross-sections of the separation and vortex
structures were illuminated using a laser light sheet. The laser beam was transmitted
through a fibre optic cable and a cylindrical lens split the beam into a light sheet of
4mm thickness. The air-cooled, argon-ion laser was operated at 200mW intensity.
Footage of the smoke was captured by a video camera operating at various shutter
speeds and individual images were transferred to a computer via a frame grabber card.
For optimum image definition and contrast, Ondina oil was chosen as it generates
clearly visible smoke filaments.

4.4 Wind Tunnel Corrections


Many methods of correcting for blockage effects were considered but the one which
seemed most applicable and gave reasonable results is Maskell’s separated flow
blockage correction. The complexity of the flow over a disc-wing has demanded a
careful analysis to realistically represent the true drag of the disc. The two definitive
flow ‘states’ for a disc-wing at incidence are pre- and post-stall which are dominated by
induced and separated drag respectively. Physically this corresponds to the transition
from a pre-stall flow state dominated by large trailing vortices to a post-stall state with a
large closed separation bubble aft of the disc (Fail et al, Fig. 6a, 1957).

The unconventional shape of the disc-wing has meant that the classical blockage
correction methods proposed by Maskell among others must be applied intelligently to
represent realistically the true drag appropriately.

- 61 -
The experiments were conducted in a wind tunnel with an octagonal test section with
height 0.9m, width 1.1m and cross sectional area Swt = 0.9m2. The disc-wing model and
wake cause the flow to accelerate around the obstruction due to the confinement of an
enclosed working section. This tends to introduce an error in the drag measurements but
can be removed by applying a blockage correction. The blockage is dependent upon the
flow regime, in the case of the disc-wing the blockage is negligible at low AoA (around
0°) and significant at higher AoA (above 30°, say).

Full details of how Maskell’s blockage correction was applied to the drag results taken
in the wind tunnel is outlined in Appendix C, both corrected and uncorrected CD curves
for the full range of AoA are shown in Figure C.2.

It was deemed more reasonable to leave the lift and pitching moment data uncorrected
than to misrepresent the lift and pitching moment curves through the inappropriate
application of standardised wind tunnel correction methods, developed primarily for
higher aspect ratio wings. The drag correction however was considered to represent the
flow physics appropriately and the resulting post-stall drag data agrees closely with high
AoA results found in the literature (Stilley & Carstens, 1971; 1972; White, 1999).

- 62 -
Chapter 5 Experimental Results

Summary
This chapter presents the wind tunnel based experimental work, including force &
moment characteristics, surface pressure data and flow visualisation results. Section
5.1 is split into three main sections: section 5.1.1 presents the baseline load
characteristics for the non-spinning disc-wing configuration and outlines test results of
basic geometry variations to illustrate why the Frisbee-like shape is so stable. Section
5.1.2 compares and contrasts the present load data with a representative low aspect
ratio (disc) wing based on finite wing theory and similar disc-wings found in the
literature. Section 5.1.3 considers the effect of spin on the aerodynamics for various
Reynolds numbers and advance ratios. Section 5.2 outlines the surface pressure data for
the non-spinning disc, including colour contour plots, cross-sectional profiles, effect of
Reynolds number and a comparison of aerodynamic coefficients integrated from the
pressure data to those directly measured in the wind tunnel. Section 5.3 presents flow
visualisation images taken using surface paint and smoke wire techniques, to provide
information for a description of the flow physics.

5.1 Load Data


5.1.1 Non-spinning Case

5.1.1.1 Baseline force and moment data

Force and moment coefficient data for the non-spinning disc for a range of angle of
attack scales is shown in Figs. 5.1 to 5.5. A number of important observations can be
made from these data:

• The angle of attack for zero lift is approximately –3o.

• The best lift to drag ratio is approximately 3, and occurs at around 7o AoA.

• The lift curve is approximately linear (CLα ~ 0.05 per degree) up to the stall, which
occurs at around 45o AoA. The flow is separated even at zero AoA and the term
‘stall’ is therefore somewhat unconventionally used here to describe the change
from partially separated to fully separated flow.

- 63 -
• The stall is associated with a steep decrease in lift, drag and pitching moment.

• The minimum profile drag CDo is 0.085 at the zero lift AoA, –3o.

• The side force and rolling moment are negligible over the whole AoA range, as
would be expected for a rotationally symmetric disc.

• The slope of the pitching moment (about the centre of gravity) is generally
positive, indicating negative pitch stability. However, between 0° to 10o AoA, the
positive gradient, and hence instability, is much reduced.

• A (unstable) trim point exists at approximately 8o AoA, meaning the angle at which
the pitching moment is zero.

• Since the disc is asymmetric in the xy plane, force and moment characteristics are
dissimilar at positive and negative incidences. In particular, a less clearly defined
stall is not present at negative incidences. This is due to the differing leading edge
geometry presented to the flow.

• The span efficiency e was found to be 0.83.

5.1.1.2 Reynolds number effects


Lift, drag and pitching moment data from –10o to 30o AoA at a range of Reynolds
numbers corresponding to a speed range from 6 to 25m/s are shown in Fig. 5.4. For the
speed range tested the force and moment coefficients are approximately independent of
Reynolds number, however a slightly more significant change is noticeable in the
pitching moment compared to the force coefficients.

5.1.1.3 Comparison with theory


A comparison of experimental lift and drag data for a sports disc with that predicted
from finite wing theory as outlined in section 2.3 is shown in Fig. 5.6. The figure also
includes data for a non-axisymmetric circular planform wing with a Clark Y cross
section. The following observations have been made:

Lift curves

• The lift curve slope of the Zimmerman wing is almost identical to that predicted
from theory (~ 0.041 per degree).

- 64 -
• The sports disc lift curve slope is on average slightly greater than expected from
theory (~ 0.049 per degree).

Drag polars

• The zero lift drag coefficient calculated for the disc-wing (~ 0.1) is almost identical
to the experimental value.

• The zero lift drag coefficient of the Zimmerman wing is approximately a factor of
5 less than for the Frisbee-like sports disc. This performance increase is due to the
streamlining of the trailing edge, which is impractical for an axisymmetric cross
section.

• The second derivatives of the Zimmerman and theoretical drag curves are similar,
indicating that the lift dependent drag coefficients are similar.

• The second derivative of the sports disc drag curves is greater than that for the
theoretical curve, indicating that the lift dependent drag coefficients for the sports
disc are greater than that predicted from theory. Note, however, that the theoretical
drag polar is calculated from the theoretical lift curve slope. If the theoretical drag
polar is calculated from the actual lift curve slope the agreement in lift dependent
drag coefficients is much closer. Also, note that using a span efficiency value e (=
0.83) derived from experiment (Fig. 5.5d) did not make any appreciable difference,
theoretical curves were therefore calculated using the first approximation e = 1.

5.1.1.4 Comparison with data from the literature


Comparison with other Frisbee-like geometries

A comparison of the aerodynamic loads to data from the literature is shown in Fig. 5.7,
for Frisbee-like disc-wing geometries with cross-sectional profiles similar to that of Fig.
2.1. The data is derived from three wind tunnel experiments (Stilley & Carstens, 1972;
Ali, 1998; Yasuda, 1999) and a free-flight experiment (Hubbard & Hummel, 2002).
Hubbard & Hummel’s coefficients were estimated from multiple short flights of a flying
sports disc (<20m range), similar to the disc-wing used in the present study.

The lift curves (Fig. 5.7a) all have a negative zero lift AoA and similar gradients. The
drag curves are all approximately parabolic (Fig. 5.7b) with good agreement for the

- 65 -
profile drag but some disparity for the lift dependent drag coefficient. This is to be
expected as a result of the small variation in the lift curve slopes.

The pitching moment curves (Fig. 5.7c) are all of similar form, with a region of reduced
slope between 0 and 10o AoA. However, there is a negative pitching moment offset in
the data of Stilley & Carstens (1972) at low AoA and a positive offset in the data of Ali
(1998) at a higher AoA. It is not clear whether these are genuine differences or
discrepancies due to experimental methods used.

A comparison of the data from the present study to that of Stilley & Carstens over an
AoA range of –30 to +90 is shown in Fig. 5.8. The most noticeable difference is that the
Stilley & Carstens disc stalls at around 30o AoA whereas the disc in the present study
stalls at 45o, leading to large differences in forces and moments between 30o and 45o
AoA. Notice also that the severity of the stall for the Potts & Crowther disc is much
greater than that for the Stilley & Carstens disc. Once both discs have stalled, however,
the data is once again in reasonable agreement. It is believed that the differences in the
stall data between the two discs is attributable to difference in leading edge profile and
differences in tunnel mounting arrangement (Stilley & Carstens mounted their disc from
the side, which would interfere with the rollup of the leading edge vortex on one side).

Comparison of circular planforms with non Frisbee-like profiles

A comparison of the present data to that found in the literature for non-Frisbee-like
circular planform geometries is shown in Fig. 5.9. The data set plotted here is derived
from many experiments working at very different Reynolds numbers up to 3.3×106
(Demele & Brownson, 1961). Note that these disc-wings are all axi-symmetric apart
from the Zimmerman disc.

The Discus (Tutjowitsch, 1976), from field athletics, has zero camber and therefore zero
lift at zero AoA (Fig. 5.9a). The Avrocar shape (Greif & Tolhurst, 1963) and the Clark
Y disc-wing (Zimmerman, 1935) are both positively cambered and the lift curves are
offset compared to the Discus (Fig. 5.9a).

The sharp rim streamlined geometry (Greif, Kelly & Tolhurst, 1960) was shown to have
negative lift up to around 2° AoA. The central raised platform geometry suggests that
this shape should have positive camber and positive lift at zero AoA. Their wind tunnel
apparatus (Greif, Kelly & Tolhurst, 1960) included a single L-shaped strut, similar to
that used in the present study, centrally attached through the raised platform. The strut

- 66 -
diameter was around half that of the raised platform diameter. Therefore the most likely
explanation for a zero lift measurement at 2° AoA is strut interference over the upper
surface raised platform.

The re-entry geometry (Demele & Brownson, 1961) has negative camber and a lift
curve with positive zero lift AoA, at around 3°.

The Discus has half the profile drag (CDo ~ 0.04) of the Frisbee-like shape (Fig. 5.9b).
However, the more streamlined sharp-rimmed disc (Greif, Kelly & Tolhurst, 1960) is
unexpectedly much larger (CDo ~ 0.12), again the most likely situation is that the strut
interference has added an error into the drag measurements, as previously reasoned. It is
also surprising that the drag measurements for the Avrocar shape (Greif & Tolhurst,
1963) and the re-entry geometry (Demele & Brownson, 1961) are so different, even the
profile drag (CDo ~ 0.06 & CDo < 0.01, respectively).

The pitching moment curves (Fig. 5.9c) fall into three sets: Firstly, the Frisbee-like disc
has minimum gradient (0.001 per degree) over the typical flight AoA range (0° to 10°)
and trim AoA at 9°. Secondly, Greif & Tolhurst’s (1963), Greif, Kelly & Tolhurst’s
(1960) and Zimmerman’s (1935) disc geometries overlay one another and have steeper
gradient (0.008 per degree) and trim AoA at around 3°. Thirdly, Kentzer & Hromas’
(1958) Discus and Demele & Brownson’s (1961) geometry overlay each other with
even steeper gradient (0.01 per degree) and a trim AoA at 0° or just above.

For the typical flight angle of attack range (0° to 10°), the negatively cambered re-entry
disc (Demele & Brownson, 1961) has the highest nose up pitching moment, followed by
the Discus (Kentzer & Hromas, 1958) but has approximately the same gradient and
magnitude. For the positively cambered discs (Zimmermen, 1935; Greif, Kelly &
Tolhurst, 1960; Greif & Tolhurst, 1963) the nose up (overlaying) pitching moment
curves decrease in magnitude and in gradient slightly also, all have a zero pitching
moment trim condition at 3° AoA. As the cavity is introduced i.e. Frisbee-like disc-
wing, the pitching moment gradient decreases, becoming almost zero but slightly
positive with a higher trim condition at 9° AoA. So from negative (Demele &
Brownson, 1961), through zero (Kentzer & Hromas, 1958) to positive camber
(Zimmerman, 1935; Greif, Kelly & Tolhurst, 1960; Greif & Tolhurst, 1963) and the
introduction of a cavity (Frisbee-like shape), the pitching moment curve has decreasing
gradient & magnitude and increasing trim AoA (Fig. 5.9c).

- 67 -
All the data described so far, found in the literature, is plotted on large-scale graphs for
easy comparison (Figs 5.10 to 5.12).

5.1.2 Force and Moment Data for a Spinning Disc

Force and moment data for a disc at various spin rates (0, 4, 8, 16 & 24Hz) and free
stream speeds (6, 10, 15 & 20m/s) is shown in Figs 5.13 to 5.19. Note that an advance
ratio of 1 corresponds to the maximum value expected for a typical hand launch of a
sports disc. Spin direction is clockwise when viewed from above.

The first figure (5.13) shows data for a speed of 20m/s plotted at a similar scale to that
shown for previous non-spinning cases. A number of conclusions can be drawn from
this data:

• For advance ratios up to 1, the change in lift and drag is small compared to the
changes due to variations in AoA.

• Increasing advance ratio generates an increasing side force increment. This


increment is independent of angle of attack over the range presented. The sign of
the side force increment is consistent with the sign of the Magnus force generated
on a spinning cylinder with cross flow.

• There are small changes in pitching moment and rolling moment between 0o and
10o AoA due to the effects of spin. The changes are consistent with negative
changes in CMo and CRo, however it is not clear why the spin causes these changes,
although some evidence is provided by the off surface flow visualisation of the
spinning disc discussed later.

For reference, the side force, rolling moment and pitching moment results discussed
above are shown at increased y-axis scale in Fig. 5.14.

To investigate the effect of higher advance ratios, the tests presented in Fig. 5.13 were
repeated at the same spin rates but reduced free stream speeds, Figs 5.15 to 5.19. The
most important result from these tests is that at sufficiently high advance ratios (>1),
there is a significant increase lift (e.g. ∆CL of 0.5). It is speculated that this is due to
spin-induced radial flow components on the upper and cavity surfaces of the disc.

- 68 -
From a dimensional perspective, it is expected that force and moment coefficient data
taken at different spin rates and free stream speeds should collapse if plotted against the
advance ratio. To investigate this, data was obtained at a fixed angle of attack (0o) by
varying advance ratio for a number of different speeds, Fig. 5.20. From this figure it is
apparent that the correlation with advance ratio is in good agreement for the lift, drag,
pitching moment and rolling moment. Although the correlation for side force is not
apparent here (Fig. 5.20c), re-plotting side force against spin rate clearly shows that the
curves overlay one another. This suggests that the side force is independent of Re
number and dependent upon spin rate, for reasons that are at present unclear.

The spin dependent rolling moment (Fig. 5.20e) curves show a good correlation with
advance ratio. This suggests that the asymmetric lift generation over the spinning disc is
consistent across appropriately matched flow conditions. The errors observed on the
lowest Reynolds number (Re = 1.13×105) CR curve at rapid spin rates and low flow
speeds, are likely to be caused by the vibration transmitted to the balance at resonant
(spin) frequencies, particularly at AdvR ~ 2 & 3.

5.1.3 Basic Geometry Comparisons

5.1.3.1 Camber

Force and moment data for various discs with increased camber (or cavity depth) are
presented in Fig. 5.21, cross-sectional profiles of the various axisymmetric discs are
provided in a key below the figure. The dimensions of three of the discs are given on
Fig. 4.6, the thickness of the fourth test model was extended using a thin annular ring
fixed inside the rim of the disc, protruding at a fixed height above the lip.

Using the frisbee-like configuration as the baseline case (th /c = 0.14), the reduced drag
and increased lift of the intermediate shape (th /c = 0.06) suggests superior performance
but the larger (nose up) pitching moment would give stronger roll divergence, above 3°
AoA. This would cause the disc to roll, bank and sideslip away from the initial flight
direction more strongly than the frisbee-like disc. The minimum thickness of the flat
plate disc (th /c = 0.01) ensures that the profile drag is very low, CDo = 0.015 at 0° AoA,
but the pitching moment is even larger (stronger roll), than the more cambered shapes
tested. The disc with increased rim height (th /c = 0.16) was shown to have a relatively
moderate (but not small) nose up pitching moment over the typical flight angle of attack

- 69 -
range. The frisbee-like geometry therefore, has a unique small nose down pitching
moment for typical flight angles of attack that is lost for both increased and reduced
camber.

5.1.3.2 Cavity

The aerodynamic loads acting on the frisbee-like configuration (hollow cavity) are
compared to the cavity filled variant (solid) in Fig. 5.22. The upper surface is identical
for both discs, the only difference being the cavity geometry. The lift curve slope of the
solid disc is much lower (~ 0.03 per degree) than for the hollow shape. The drag of the
solid shape is roughly half that of the frisbee-like disc, the profile drag is much lower,
CDo ~ 0.047. The pitching moment gradient of the solid geometry is much steeper and
nose up, above 2o AoA. The large contrast in load measurements is primarily caused by
the inside cavity rim, of the hollow configuration, being presented to the incident flow.
This increases both the lift & drag and provides a large nose down pitching moment
increment, reducing the pitching moment slope close to zero. The frisbee-like (hollow)
geometry therefore, has a unique small nose down pitching moment for typical flight
angles of attack that is lost for the cavity filled (solid) variant.

5.2 Surface Pressure Data


5.2.1 Overview

The surface pressure distribution on a non-rotating disc-wing was investigated over an


angle of attack range and at various flow speeds. The data is presented as colour
weighted contour plots of pressure coefficients Cp and cross-sectional profiles for both
the upper and lower (cavity) surfaces. The colour contour plots are generated directly
from the pressure data, the pressure profiles are taken from a fine grid of points 1/2 mm
apart interpolated from the pressure data.

The following outlines the surface pressure distribution for a typical angle of attack and
contrasts the variation over an AoA range. The effect of Reynolds number on the
pressure distribution is discussed and the load data integrated from the pressure data is
compared to the directly measured aerodynamic load data.

- 70 -
The pressure data plotted in Fig. 5.23 and thereafter takes the 3D surface pressure
distribution and creates a 2D surface, with the 2D radial distance equal to the 3D surface
distance. This has the effect of stretching the data at the edges of the disc compared to
the true view of the surface when viewed from above. Note also that whilst all pressure
distributions are nominally symmetric about the xz plane, data was taken over the entire
surface of disc, not over half the disc and mirrored.

5.2.2 Discussion of Results at Typical Flight Angle of Attack (5°)

The upper & lower surface pressure distributions at Re = 2.84×105, equivalent to a flow
speed of 15m/s and AoA = 5° are shown in Figs 5.23 to 5.27.

The highest pressure region (A), Fig. 5.25a, occurs on the leading edge upper surface
with a low pressure crescent region (B) aft of this region. The central part of the disc
shows pressure recovery (C). The low pressure band (D) on the trailing edge rim is
associated with trailing vortices and bluff body effects. Low pressure (F) is seen within
the cavity (Fig. 5.25b) except inside the trailing edge rim (G).

The central cross-sectional pressure profile at the half span station is shown in Fig. 5.26,
leading edge on the left of the figure. The upper surface is shown by the unbroken line,
the lower surface curve is dashed. The cross-sectional disc geometry is also shown to
relate the location of pressure peaks to the disc surface. The high (positive) pressure
coefficient on the leading edge is shown by the spike at zero chord (Cp –> 1), positive
pressure axis down. The lowest pressure region at around tenth chord (Cp ~ -1.25) and
secondary peak thereafter produce a large suction on the leading edge. The pressure
recovery on the upper surface is the same pressure as in the cavity (Cp ~ -0.25) through
the central portion (0.3c to 0.6c). The high pressure trough in the cavity (Cp ~ 0.3) and
the upper surface low pressure peak (Cp ~ -0.8) are shown on the trailing edge. It is clear
that the main contributions to the lift are from the low pressure peak on the leading edge
and the large pressure difference on the trailing edge. The spikes in the cavity surface
data at x/c ~ 0.05 & 0.95 are due to the presence of the vertical surface at this location.

The pressure profile at the half chord station is shown in Fig. 5.27, port wing tip on the
left of the figure. The symmetry of the profile is evident, the only pressure difference
visible at the wing tips. Much of the upper and lower surface is set at a consistent low
pressure (Cp = -0.3).

- 71 -
5.2.3 Effect of Angle of Attack on Pressure Distributions

The variation in surface pressure distribution with angle of attack is shown in Figs 5.28
to 5.31, for Re = 2.84×105 and AoA = -10° to 30°.

The pressure of the upper surface distribution (Fig. 5.28) can be seen to decrease, as a
whole, with increased angle of attack. The high pressure region (red) on the leading
edge becomes narrower and moves progressively forward with increased AoA. The low
pressure region aft of the leading edge high pressure region (yellow, Fig. 5.28a;
blue/green, Fig. 5.28e) grows with AoA, extending downstream and decreasing in
pressure magnitude (Cp ~ -1, Fig. 5.28c; Cp ~ -2.5, Fig. 5.28i). The central pressure
recovery region (orange, Fig. 5.28a; yellow, Fig. 5.28e) decreases in pressure slightly
with AoA. Above 10° AoA the pressure recovery region reduces in size and moves
downstream locating aft of the disc centre at around x/c = 0.7. The low pressure band on
the trailing edge (green, Fig. 5.28a; green, Fig. 5.28e) curves upstream near the wing
tips at -10° AoA, becoming a figure of eight formation at 0°. Low pressure spots
develop on the wing tips (blue, Fig. 5.28e) by 10° AoA, growing in size and moving
upstream with increasing AoA. A pressure recovery region is visible on the trailing edge
rim, which decreases in pressure with AoA (orange, Fig. 5.28a; yellow, Fig. 5.28i).

The pressure of the cavity surface distribution (Fig. 5.29) can be seen to increase, as a
whole, with increased angle of attack. The disc cavity exhibits low pressure over much
of the lower surface (green, Fig. 5.29a; yellow, Fig. 5.29c) for low AoA. The high
pressure region on the trailing edge rim (orange, Fig. 5.29a; red, Fig. 5.29e) grows in
size as the cavity becomes more exposed at increased AoA. At high AoA a higher
pressure region develops just ahead of the disc centre (red, Fig. 5.29h; dark red, Fig.
5.29i). At 30° the isolated high pressure region (dark red, Fig. 5.29i) just ahead of the
centre approaches a Cp of 1, it decreases thereafter until the approach to the trailing edge
where the steep rim surface forces compression and the Cp approaches 1 again.

The central cross-sectional pressure profiles at the half span station are shown in Fig.
5.30, for various AoA at 10° intervals. Considering firstly the upper surface, from an
AoA of -10° the low pressure peak on the leading edge increases in magnitude with
AoA (Cp ~ -0.5 at -10° AoA; Cp ~ -2.5 at 30°) and moves towards the leading edge.
However, at 20° AoA the initial low pressure peak decreases in magnitude slightly from

- 72 -
10° (Cp ~ -1.5 at 10° AoA; Cp ~ -1.3 at 20°) but the secondary peak becomes elongated
(20° AoA) moving further downstream. The central higher pressure trough increases in
magnitude slightly with AoA (Cp ~ -0.1 at -10° AoA; Cp ~ -0.3 at 10°) and shifts
downstream aft of the disc centre (x/c ~ 0.7, 20° AoA) for higher AoA. The low
pressure peak on the trailing edge increases in magnitude with AoA (Cp ~ -0.7 at -10°
AoA; Cp ~ -1.3 at 30°) but remains located at around x/c = 0.9.

Secondly, considering the lower surface (Fig. 5.30), the low pressure over much of the
surface becomes higher with increased AoA (Cp ~ -0.7 at -10° AoA; Cp ~ 0.6 at 30°, x/c
= 0.5). The high pressure trough within the trailing edge cavity becomes higher with
AoA (Cp ~ -0.2 at -10° AoA, Cp ~ 1 at 30°; x/c = 0.95) and another high pressure trough
develops at around x/c = 0.3 for high AoA (Cp ~ 0.9 at 30° AoA).

A high pressure spike is seen on the leading edge, for both the upper and lower surfaces,
at all AoA tested (Fig. 5.30 & 5.31).

The central cross-sectional pressure profiles at the half span station are shown in Fig.
5.31, for various AoA at 5° intervals. The data presented here (Fig. 5.31) allows the
comparison of a greater number of attack angles and the gradual transition with
increasing AoA is more evident. The lower surface curve shows lower pressure
(overall) than the upper surface at -10° AoA. Overall, the upper and lower surface
pressure becomes lower and higher, respectively, with increased AoA.

The cross-sectional pressure profiles at various span stations are shown in Fig. 5.32, for
5° AoA. The cross-sectional slice through the disc, shown below each plot, is at the
respective span position to relate the location of pressure peaks to the disc surface. A
full description of the pressure profile at the half span station (Fig. 5.32a) is given
previously with reference to Fig. 5.26. At further outboard stations the profile is broadly
similar, almost identical at the 3/8 span position (Fig. 5.32b). The magnitude of the low
pressure (double) peak on the leading edge decreases in magnitude towards the wing
tips (Cp ~ -1.2 at 3/8 b station; Cp ~ -0.8 at 1/8 b station). The low pressure peak on the
trailing edge increases in magnitude (Cp ~ -0.7 at 1/2 b station; Cp ~ -1.3 at 1/8 b
station) at outboard stations becoming more of a spike at the 1/8 span position (Fig.
5.32d). The lower surface remains at a constantly low pressure (Cp ~ -0.25), for all span
stations, aft of the high pressure spike except towards the trailing edge. The high

- 73 -
pressure trough just upstream of the trailing edge, inside the cavity rim, is at a consistent
maximum pressure for all span stations (Cp ~ 0.3) except 1/8 b (Cp ~ -0.2).

A grid of cross-sectional pressure profiles at various span/chord stations and various


AoA are shown in Fig. 5.33, to give a more complete set of pressure profiles for
comparison. Many of the profiles and trends have previously been described with
reference to Figs 5.30 to 5.32. Just to comment on the variation with AoA at the 1/2
chord station (Fig. 5.33v), the symmetry is evident at low AoA (-10° to 10°) but the
upper surface pressure profile becomes asymmetric at higher AoA (20° & 30°). The
constant low pressure over the central portion of the upper surface increases in
magnitude slightly for increasing AoA (-10° to 10°), increasing further for higher AoA
(20°, Fig. 5.33d,v) and developing a central trough at 30° (Fig. 5.33e,v). The low
pressure (upper surface) increases in magnitude towards the wing tips (x/b = 0 & 1) for
all AoA (Cp ~ -0.5, Fig. 5.33a,v; Cp ~ -1.5, Fig. 5.33d,v). The constant low pressure
over the central portion of the lower surface becomes high pressure for increasing AoA
(Cp ~ -0.6 at x/c = 0.5, Fig. 5.33a,v; Cp ~ 0.6 at x/c = 0.5, Fig. 5.33e,v). The pressure
drops sharply at the wing tips (lower surface) for high AoA (10° to 30°, Fig. 5.33v).

5.2.4 Effect of Reynolds number on pressure distributions


The variation in surface pressure distribution with Reynolds number is shown in Figs
5.34 to 5.36, for AoA = 5º, Re = 1.13 to 3.78 ×105 equivalent to V∞ = 6 to 20m/s.

The pressure surface distributions presented in Figs 5.34 & 5.35 are at first glance
identical but there are small differences. The low pressure crescent aft of the leading
edge (blue, Fig. 5.34) narrows with increased Re, the downstream edge moves
upstream. The magnitude of this low pressure region increases, as does that of the low
pressure band on the trailing edge (blue, Fig. 5.34a; dark blue, Fig. 5.34d). The pressure
distribution within the cavity is unaffected by Re (Fig. 5.35).

The central cross-section profiles (Fig. 5.36) show that the upper surface low pressure
double peak aft of the leading edge narrows but increases in magnitude with increased
Re. The primary low pressure peak location remains at x/c ~ 0.1 for all Re tested,
whereas the secondary peak moves upstream (Re = 1.13 ×105 at x/c ~ 0.3; Re = 2.84
×105 at x/c ~ 0.2). The low pressure peak on the trailing edge increases in magnitude

- 74 -
(Cp ~ -0.6, Re = 1.13 ×105; Cp ~ -0.9, Re = 3.78 ×105) but remains at x/c ~ 0.9. The
lower surface pressure profiles overlay one another for all Re tested.

5.2.5 Loads Comparison: Integrated Pressure Data and Balance


Measurements

Pressure load characteristics were integrated from the pressure data over the 3D disc-
wing geometry. The disc was divided into a 3D grid of flat surfaces, each was a
trapezium (isosceles trapezoid) with a central pressure tapping. The grid was generated
by splitting the disc into 60 axisymmetric segments 6° apart, each segment divided
further into 39 trapezium surfaces, from the centre of the upper surface (tapping 1)
along the profile around the rim and lip (tapping 20) onto the lower surface and back to
the disc centre (tapping 39) within the cavity. The pressure over each trapezium surface
was assumed constant and equal to the pressure at each corresponding tapping. The
incident pressure produced a small force (10-3 N) at the tapping position, acting in the
direction perpendicular to the surface. Each force was resolved to the vertical and
horizontal directions and appropriate contributions were summed to calculate total
values for the aerodynamic pressure normal and axial forces. The normal and axial
forces were then resolved to account for AoA and therefore calculate the pressure lift
and drag (Fig. 5.37a&b). The pitching moment calculation took account of the
perpendicular distance of each vertical and horizontal force from the y-axis (side force
axis, see Fig. 2.6b), which was located at (x/c, z/t) = (1/2, 1/2). Appropriate moment
contributions were summed to calculate the aerodynamic pressure pitching moment
(Fig. 5.37c).

The integrated pressure data is compared to the load measurements, taken from the wind
tunnel balance, in Fig. 5.37 for AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105. The
pressure lift curve closely follows the total lift curve (Fig. 5.37a), but has slightly less
lift for the low AoA range, the zero pressure lift AoA at around -1º. The profile pressure
drag (0.035) is less than half that of the total profile drag (0.085) at -3º AoA, the drag
curves (Fig. 5.37b) converge at higher AoA approaching 30º. The pressure pitching
moment curve is similar in form to the total pitching moment curve (Fig. 5.37c),
negative (nose down) pitching moment at low AoA (<15º), zero pitching moment from
15º to 20º and positive (nose up) pitching moment for higher AoA (above 20º). The

- 75 -
entire pressure moment curve is well below (nose down increment) the total pitching
moment curve for all AoA.

The aerodynamic loads are a combined result of the pressures and shear stresses, acting
locally on the disc model surface, generated by a crossflow. The integrated pressure
data does not account for the shear stress loading component, causing discrepancies
between the curves presented in Fig. 5.34. The shear stress acting on the leading edge
upper surface, due to accelerating flow, causes local lift & drag which in turn drives a
nose up pitching moment. This explains the slightly higher lift and much larger drag &
pitching moment curves plotted from the balance data. The reattachment of the shear
layer to the upper surface will also contribute a drag force due to the downstream shear
stress. Added to the aforementioned sources of shear stress is inside the trailing edge
cavity rim at higher angles of attack, which provides nose down pitching moment. This
explains the large pitching moment discrepancy above 15o AoA.

5.3 Surface Flow Visualisation

5.3.1 Surface Flow Patterns

5.3.1.1 Typical flight angle of attack (5°)

The surface paint patterns for a non-spinning disc-wing are shown for both the upper
(Fig. 5.38) & cavity (lower) surfaces (Fig. 5.39) at a typical flight angle of attack (5°)
and a flow speed of 15m/s, equivalent to a Reynolds number of 2.84×105.

A description of the surface flow is given with reference to Fig. 5.40, which
superimposes labels onto the surface flow patterns taken of Figs 5.38 & 5.39 to aid the
explanation of the surface flow features.

The upper surface pattern (Fig. 5.40a) shows that the surface flow direction is from the
leading edge towards L1 . This is indicated by the streak lines within region A. The
boundary layer separates from the surface at an arc of near constant radius (L1),
followed by reattachment at a line of similar geometry (L2). The boundary layer remains
attached throughout region C and then separates at L3. The initial separation and
subsequent reattachment at the leading edge forms a separation bubble in region B. At
the trailing edge, a similar reversed surface flow pattern is observed within region D, as
fluid is drawn towards the stagnation line at L3. The two points V1,V2 at symmetrical
- 76 -
positions are separation nodes formed by the detachment of the trailing vortices from
the bluff body surface. Streak lines in the vicinity of V1 and V2 suggest that fluid from
the separation bubble and from beneath the cavity feeds into these vortices.

The cavity surface pattern (Fig. 5.40b) indicates that the boundary layer separates off
the leading edge lip and reattaches on the inside of the trailing edge rim leaving a
stagnation line (which cannot be seen on Fig. 5.40b). The shear layer encloses a
reversed flow region (F) from the trailing edge towards the stagnation line L4 and a
‘dead air’ (or weakly circulating) pocket (E) aft of the leading edge rim. It is worth
noting that the line L4 which defines the boundary between the reversed flow (F) and
dead air pocket (E) is straight.

5.3.1.2 Range of angles of attack (0° to 30°)


A description of the variation in surface paint patterns with AoA follows with reference
to Figs 5.41 & 5.42. Firstly a polar coordinate system is defined for quantitative
demarcation of surface flow features; if the centre of the disc-wing is (0,θp) using polar
coordinates (rp,θp), rp defined as the radial position and θp the angular component. Take
positive θp in the positive yaw direction (0 ≤ θp < 2π) and the point (c/2,0) central on the
leading edge, where c is the chord length i.e. c/2 the disc radius.

On the upper surface (Fig. 5.41) at 0° AoA, the flow separates at a line (arc) of constant
radius, which crosses the body fixed roll axis at (0.8c/2,0). For increased AoA, at 5° the
flow separates at the same position (0.8c/2,0), again with reference to the roll axis,
gradually moving upstream with increased AoA to (0.95c/2,0) at 30°. The separated
boundary layer can be seen to reattach for the full AoA range 0° to 30° depicted in Fig.
5.41.

The surface patterns indicate that the reattachment line, is an arc of constant radius from
0° to 10° AoA moving downstream from (0.5c/2,0) at 0° to (0.4c/2,0) at 10°. Between
10° and 20° AoA the reattachment line transitions to a straight line (perpendicular to the
roll axis) but it is difficult to see this on Fig. 5.41. A close up of the reattachment is
given for 20° AoA on Fig 5.43a, crossing the roll axis at (0.1c/2,0). The arrows show
the flow direction away from the stagnation line.

- 77 -
The reattachment line shifts further downstream to a position of (0.2c/2,π) at 25°, before
returning upstream to a central position of (0,π) at 30° AoA. For 25° to 30° AoA the
straight reattachment line narrows, transitioning to a nodal point, see Fig. 5.41f&g. A
close up of the nodal reattachment is shown for 30° AoA in Fig 5.43b. The radial
outflow on the surface, from this point, has left an uneven pattern suggesting unsteady
effects as the reattachment of the boundary layer dictates the surface flow in this
locality.

For increasing angle of attack therefore the reattachment line is initially an arc of
constant radius (0° to 10°) which transitions to a straight line (10° to 20°) and then
narrows in width to become a nodal point or focus of reattachment (20° to 30°).

The separation line on the trailing edge of the disc-wing crosses the roll axis at a
position of (0.8c/2,π) at 0° AoA. For increased angle of attack the separation line does
not move much, merely shifting slightly downstream to (0.9c/2,π) at 30° AoA. At low
AoA (0° to 10°) the trailing edge separation line exists between two nodal points which
are clearly visible in Figs 5.41a,b&c, also labeled V1 & V2 on Fig. 5.40a. The flow in
close vicinity to these nodal points is a complex interaction of the bluff body wake,
separated boundary layer and, at high enough AoA, trailing vortices also. As the AoA
increases further, these nodes change in form and location moving upstream towards the
wing tips. As the reattachment line transitions from an arc to a straight line (10° to 20°)
the separation nodes become increasingly dominated by the detachment of trailing
vortices. The positions of these nodes are (0.8c/2,5π/6) & (0.8c/2,7π/6) at 0° AoA,
moving through (c/2,2π/3) & (c/2,4π/3) at 15° to (c/2,π/2) & (c/2,3π/2) at 30°.

Within the cavity at low AoA (Fig. 5.42), the shear layer that separates off the leading
edge reattaches to the surface inside the trailing edge rim. At 0° AoA, the reversed flow
beneath the shear layer generates two spiral nodes which are located at (0.7c/2,π/5) and
(0.7c/2,9π/5). A half annulus shaped dead air region can be seen ahead of the spiral
nodes (Fig. 5.42a).

Thereafter the straight-line stagnation (as labelled L4 in Fig. 5.40b) becomes the most
obvious feature within the cavity paint patterns, caused by the reversed flow beneath the
shear layer and the dead air enclosed behind the leading edge rim, for 5° to 10° AoA.
The straight stagnation line is perpendicular to the roll axis, crossing the axis at
(0.2c/2,0). Between 10° & 15°, the shear layer reattachment line moves upstream, off

- 78 -
the inside surface of the trailing edge rim. The reattachment is located centrally (0,0) at
15° and is probably better described as a nodal point than a line. The streak line patterns
indicate that the near surface flow is towards the trailing edge from the reattachment.
The paint patterns also suggest that reversed flow occurs from the reattachment towards
the dead air stagnation line. This encloses a recirculating bubble, beneath the shear
layer. As the AoA increases the reattachment line moves further upstream, crossing the
roll axis at (0.5c/2,0) for 30° (Fig. 5.42g). Evidence for reversed flow ahead of the
reattachment is present in the paint patterns up to the highest AoA tested using surface
paints, 30°. This reversed flow suggests that a recirculating bubble is present at 15° and
above. As the reattachment moves upstream the bubble moves with it. This pushes the
dead air pocket stagnation line towards the leading edge until the bubble is directly
behind the leading edge and the dead air pocket disappears.

It is also worth noting that on the downstream half of the cavity, particularly for higher
AoA (20° to 30°), the streak lines aft of the reattachment can be seen to converge on the
centre line. This is due to the convergence of fluid as the rim narrows towards the
trailing edge.

5.3.2 Comparison of Surface Flow Visualisation with Pressure Data

The comparison images of Fig. 5.44 are the half surface 3D pressure distribution,
viewed perpendicular to the surface planform, superimposed onto the half surface paint
flow visualisation images, at 5° AoA. The separation line, on the upper surface (Fig.
5.44a), is shown to be just aft of the lowest pressure (dark blue) region on the leading
edge. The boundary layer reattachment is responsible for the pressure recovery
thereafter (green). The trailing edge separation and nodes correspond to the low
pressure (blue) band and are similar in form. The high pressure region (orange) on the
trailing edge, within the cavity (Fig. 5.44b, corresponds to where the separated
boundary layer reattaches inside the rim. The rest of the pressure cavity shows suction
(green). The straight surface paint line (Fig. 5.44b) defining the dead air boundary is not
visible in the pressure data.

The central cross-sectional pressure profile for the upper surface is compared to the
surface paint flow distribution in Fig. 5.45, with the chord wise positions of the
separation and reattachment lines marked on, for 5° AoA. It is clear that the separation

- 79 -
bubble corresponds to the double low pressure peak close to the leading edge. The
separation line (S1) is positioned just after the highest (low pressure) peak, and the flow
separates due to an unfavourable (adverse) pressure gradient. The pressure recovery at
the reattachment line (R1) and throughout the central portion of the upper surface is due
to the reattachment of the boundary layer. Another suction peak occurs approaching the
trailing edge separation line (S2), again the boundary layer separates under the influence
of an adverse pressure gradient.

The comparison images of Fig. 5.46 & 5.47 are the half surface 3D pressure
distributions, viewed perpendicular to the surface planform, superimposed onto the half
surface paint flow visualisation images, for a range of AoA (0° to 30°). The direct
comparison of paint patterns with the pressure distribution (Figs 5.46 & 5.47) in this
case is over two flow speeds. The surface paint flow visualisation patterns were taken
for a flow speed of 15m/s whereas the pressure data was taken at 20m/s. Although the
flow features have been shown to have slightly different boundaries at various Reynolds
numbers (see Fig. 5.34 to 5.36), this is perfectly valid for the purpose of deriving
relevant information to further define the flow characteristics over an AoA range (0° to
30°).

It can be seen that the upper surface paint patterns are similar in form to the pressure
colour contour plots, see Fig. 5.46. At low AoA (0° to 10°) the low pressure region
(green / blue) on the leading edge coincides well with the crescent shaped surface
pattern formed by the separation bubble. The leading edge separation line is consistently
just aft of the highest pressure peak for all AoA, similarly for the trailing edge
separation line, the boundary layer always separating from the surface under the
influence of an adverse pressure gradient. As the boundary layer reattachment line
moves downstream with increased AoA, the pressure recovery region (yellow / orange)
narrows in the streamwise direction. For high AoA (25° to 30°), note that the nodal
point of reattachment coincides with the upstream boundary of the pressure recovery
region, located aft of the disc centre. The lower pressure band on the trailing edge exists
for all AoA (0° to 30°) and corresponds to the trailing edge separation line. The low
pressure regions (dark blue) on the wing-tips correspond to the complex surface paint
patterns formed by bluff body effects and the detachment of trailing vortices. These low
pressure regions grow larger with increased AoA, merging into the low pressure region
generated by the separation bubble, at higher AoA (25° to 30°). It is difficult to see from
Fig. 5.46 how these low pressure regions relate to the bluff body effects on the suction
- 80 -
side of the disc. The detachment of trailing vortices with increased AoA is accompanied
by low pressure, which becomes increasingly dominant at the wing-tips for increased
AoA. Note that for 30° AoA the clearly visible surface paint node of trailing vortex
detachment (Fig. 5.46g) on the wing-tip coincides with low pressure.

The similarities between the lower surface paint patterns and the pressure colour
contour plots (Fig. 5.47) are less obvious than for the upper surface. The high pressure
at the trailing edge (0°, orange) moves gradually upstream with increased AoA but there
is no obvious pressure feature, which coincides with the stagnation line, and spiral node
features in the surface paint patterns for low AoA (0° to 10°). At 15° the shear layer
reattaches just ahead of the centre, pressure recovery is observed aft of the nodal point
of reattachment. The pressure recovery remains bounded by the location of the
reattachment as it moves upstream with increasing AoA (15° to 30°). At 30° the isolated
high pressure region (dark red) directly aft of the reattachment approaches a Cp of 1, it
decreases thereafter until the approach to the trailing edge where the steep rim surface
forces compression and the Cp approaches 1 again.

5.4 Smoke Wire Flow Visualisation


5.4.1 Centre-line Visualisation

The central cross-sectional profiles of the disc-wing flow field are shown in Figs 5.48 to
5.50 at various AoA (0° to 50°) and a flow speed of 3m/s, equivalent to a Reynolds
number of 5.67×104.

The flow field over a non-rotating disc-wing at various AoA (0° to 50°) are shown in
Fig. 5.48. At 0° AoA (Fig. 5.48a) the flow field shows straight smoke filaments, which
are relatively undisturbed from their horizontal free-stream position, except for the
turbulent bluff body wake. As the AoA increases to 10° & 20° the filaments show
upward deflection upstream of the leading edge and subsequent downwash aft of the
trailing edge. At 30° & 40° AoA the steep curvature of the filaments above the upper
surface are particularly apparent (Fig. 5.48d,e) as is the vortex shedding off the trailing
edge in the wake of the disc. The separation bubble is visible as a dark region close to
the leading edge, the smoke was not entrained into the structure but the separated shear
layer outlines the shape in shadow. The shear layer reattaches to the surface on the
centre line throughout the range 0° to 40° AoA. Stall occurs at 45° (see Fig. 5.2)
- 81 -
creating a much broader wake and shedding larger vortex structures, see Fig. 5.48f with
the wing in deep stall.

The central cross-section of the separation bubble at various AoA (0° to 30°) is shown
in Fig. 5.49. The laminar separation forms a distinct shear layer and the unsteady
reattachment entrains smoke into the bubble. The separated shear layer is seen to
become unstable due to a similar process to the Kelvin-Helmholtz instability for
inviscid flows, according to linear stability theory. The shear layer rolls up into one or
more laminar structures before becoming turbulent. The separation bubble forms even at
0° AoA (Fig. 5.49a), the thickness of the bubble structure enlarges with increased AoA.
It is possible to see the reattachment of the shear layer in Fig. 5.49a where a slug
flattens out on the surface and reaches into the separation bubble. Whilst observing the
separation bubble in the wind tunnel the reattachment was seen to dance on the surface
in an unsteady manner. Note the upwash angle of the fluid upstream of the leading edge.

The central cross-section of the cavity flow for various AoA (0° to 30°) are shown in
Fig. 5.50, illuminated by a laser light sheet. The laminar separation on the leading edge
lip forms a distinct shear layer (Fig. 5.50a&b), which becomes unstable (Kelvin-
Helmholtz instability) rolling up into one or more laminar structures before becoming
turbulent. The turbulent recirculation within the cavity stems from the trailing edge rim
at low AoA (0° & 10°) with reversed flow observed beneath the shear layer (Fig. 5.42).
For increased AoA (20° to 30°) the shear layer reattaches ahead of the disc centre (Fig.
5.50d). Unstable turbulent structures are shed with regular frequency from the
separation bubble (Fig. 5.50c).

5.4.2 Wake Visualisation

The bluff body wake, downwash and trailing vortex structures downstream of a non-
spinning disc-wing are shown Fig. 5.51, for the low AoA range (0° to 10°). The ambient
illumination in Fig. 5.51i shows the downwash generated in the wake of the disc-wing.
The strong central downwash and the form of the trailing vortices becomes apparent
with increased AoA from 0° through 10° (Fig. 5.51a,b&c). The laser light sheet
illumination in Fig. 5.51ii shows the wake cross-section at one chord length downstream
of the disc trailing edge. Fig. 5.51aii shows the unsteady bluff body wake at 0° AoA.
The structure of the trailing vortices has begun to develop at 5° AoA (Fig. 5.51bii), the

- 82 -
turbulence generated by the bluff body still visible. Note the close vicinity of the trailing
vortex pair due to the low aspect ratio, as the AoA increases the central downwash
develops and stronger rotation is given to the trailing vortices. At 10° AoA the vortex
structures are more clearly defined (Fig. 5.51cii). The comparison of these two methods
of illumination enable the reader to better understand how the wake cross-sections relate
to the disc-wing model.

The bluff body wake, downwash and trailing vortex structures extending downstream
from a non-spinning disc-wing are shown Fig. 5.52, for the low AoA range (0° to 10°).
Note the symmetry in the wake. The ambient illumination used for Fig. 5.52a shows the
turbulent bluff body wake at 0° AoA to just over five chord lengths downstream from
the disc trailing edge. The trailing vortices become more clearly defined with increased
AoA, from 0° through 10° (Fig. 5.52a,b&c). Fig. 5.52b is a transitional stage of trailing
vortex development, the bluff body wake is still evident but the trailing vortices begin to
take shape (see Fig. 5.51b also). The trailing vortex pair is established at 10° AoA (Fig.
5.52c). The inboard edge of the trailing vortices is clearly defined in Fig. 5.52c by two
symmetrical white lines, which narrow with distance downstream. The strong central
downwash exists between these two lines with the trailing vortices either side. Close
behind the trailing edge the smoke is entrained into the trailing vortex structures and the
wake seems narrower than the disc itself (Fig. 5.52c). This is deceptive however: At the
trailing edge, these white lines define the inboard edge of the trailing vortex detachment
from the disc surface. The outboard edge of the wake and vortex detachment is bounded
at the wing tips.

Cross-sectional slices through the wake at a range of positions downstream of a non-


spinning disc-wing at 10° AoA are shown in Fig. 5.53. The laser light sheet illuminates
turbulence at the trailing edge (x/c = 0, Fig. 5.53a) and the curvature on either side of
the disc, indicates central downwash. With distance downstream, the central downwash
forces the turbulent bluff body wake lower and the form of the trailing vortices becomes
clearer as more smoke is entrained into the vortices.

5.4.3 Visualisation of the Effect of Spin

The flow structures on the upper surface of a spinning disc-wing (AdvR = 0, 0.7 & 0.9)
at 0° & 5° AoA are shown in Fig. 5.54. At 0° AoA, the separation bubble is clearly

- 83 -
identifiable for the non-spinning case (Fig. 5.54ai), note the symmetry particularly in
the wake. For the spinning case (Fig. 5.54bii) the form of the separation bubble is
largely unchanged, however the bubble shifts slightly towards the leading edge on the
advancing side (left of the figure) but remains unchanged on the retreating side. Also,
the fluid within the bubble is curved around the disc with the local surface and is mostly
shed from the retreating side. The wake becomes asymmetric and is deflected slightly
towards the advancing side. At 5° AoA, the turbulence in the near surface reattached
flow is clearly visible at this higher Reynolds number (Fig. 5.54ii, Re = 5.67×104). For
the non-spinning case (Fig. 5.54bi), the smoke is entrained into the separation bubble
within the central corridor (inboard) only. However, for the spinning case (Fig. 5.54bii),
the boundary reattachment is visualised further outboard on the advancing disc surface
(Fig. 5.54bii, left hand side) forces the smoke layer to feed into the separation bubble.
The moving surface (due to spin) can also be seen to deflect the smoke filaments within
the separated shear layer (Fig. 5.54b) in the spin direction, when compared to the non-
spinning case (Fig. 5.54a). For 5° AoA also, the wake becomes asymmetric and is
deflected slightly towards the advancing side (Fig. 5.54bii).

Cross-sectional slices through the wake, central downwash and trailing vortex structures
downstream of a spinning disc-wing (AdvR = 0, 0.7 & 1.6) at 10° AoA, are shown in
Fig. 5.55. The symmetrical vortices (Fig. 5.55i), illuminated by a laser light sheet, for
the non-spinning disc become asymmetric with increased rotation (Fig. 5.55ii&iii). The
advancing side of the disc, on the left side of each image, inhibits the near surface flow
(for non-zero advance ratio, Fig. 5.55ii&iii) and retards the downwash on that side
inhibiting the vorticity given to the trailing vortex. Whereas the retreating side of the
disc, on the right hand side of each image, aides the near surface flow (for non-zero
advance ratio, Fig. 5.55ii&iii) and augments the downwash on that side enhancing the
vorticity given to the trailing vortex. This deforms the trailing vortex detached from the
advancing side into a more unsteady bluff body structure, which is also raised slightly
higher than the undisturbed trailing vortex detached from the retreating side. Note the
similarity between the deformation of the trailing vortex detached from the advancing
side at the two downstream positions shown in Fig. 5.55a&b.

Cross-sectional slices through the wake, central downwash and trailing vortex structures
at a range of positions downstream of a spinning disc-wing (including opposite spin
directions, AdvR = 0.7, 0 & −0.7) at 5° AoA, are shown in Fig. 5.56. At the trailing
edge, a marked downwash cusp is observed aft of the retreating side of the disc (Fig.
- 84 -
5.56ai, right hand side), which is not present for the non-spinning case (Fig. 5.56aii). If
this localised downwash is tracked downstream (Fig. 5.56ia-g), the initially small
downwash cusp (Fig. 5.56ai) develops into enhanced downwash aft of the retreating
side (Fig. 5.56ib-d, right hand side), compared to that downstream of the advancing
side. The form of the trailing vortices starts to become apparent at one chord length
downstream (Fig. 5.56e). The enhanced downwash aft of the retreating side sweeps into
the trailing vortex thus enhancing its rotation (Fig. 5.56ie&f). Two chord lengths
downstream, the trailing vortices at 5° (Fig. 5.56gi) show asymmetry similar to that at
10° (Fig. 5.55bii). Namely, the trailing vortex aft of the advancing side is raised slightly
higher than the trailing vortex aft of the retreating side (Fig. 5.56gi). The retarded
downwash aft of the advancing side is also observed in Fig. 5.56i, compared to the non-
spinning case (Fig. 5.56ii), at each respective downstream station. The slightly raised
trailing vortex aft of the advancing side is attributed to this retarded downwash, which
also inhibits the rotation given to the vortex on that side. The wake downstream of a
disc with reversed rotation (AdvR = −0.7) is shown in Fig. 5.56iii. The same wake
characteristics are depicted here, the enhanced downwash now on the left of each
image, due to the opposite direction of disc rotation. There is a good level of symmetry
between Figs 5.56i and 5.56iii, taking into account the turbulent and unsteady nature of
the bluff body wake at 5° AoA.

- 85 -
Chapter 6 Flow Topology

Summary
This chapter proposes a flow topology to define the non-spinning disc-wing flow physics
in both two and three dimensions. Section 6.1 presents the 2D mid-span flow field cross-
section and 3D flow structures topologically for a typical flight angle of attack (10o).
Section 6.2 illustrates the variation in mid-span flow field with angle of attack (0o to
30o) including pre- and post-stall topologies (+/− 45o).

6.1 Typical Flight Angle of Attack (10°)


The proposed topology of the flow over a non-rotating disc-wing is given based upon
the flow visualisation. The two-dimensional central flow cross-section at 10° AoA
shown in Fig. 6.1 depicts many of the flow structures that have been previously
discussed, namely separation bubble, cavity flow and turbulent wake. With flow from
right to left, first of all note the upwash ahead of the leading edge, then the boundary
layer separates off the upper surface reattaching shortly thereafter creating a
recirculating bubble. Further downstream the reattached boundary layer separates from
the upper surface on the trailing edge, this separated shear layer contributes to the wake
which has downwash angle slightly greater than the AoA. Within the cavity, the shear
layer that separates off the leading edge lip reattaches on the inside of the trailing edge
rim, reversed flow occurs beneath the shear layer towards the leading edge. The fluid
flowing towards the leading edge (reversed flow) within the cavity is restricted from
flowing further forward by the stagnant air pocket in the lee of the leading edge rim.
This stagnant air aft of the leading edge is denoted by the weakly moving arrow heads
on Fig. 6.1. The straight stagnation line in the cavity surface paint patterns (Fig. 5.40b)
marks the boundary between the reversed flow and the stagnant air, within the cavity.
This boundary is depicted in the topology (Fig. 6.1) as a separation line. On the tip of
the trailing edge rim, the boundary layer separates from the surface, contributing to the
vortices shed into the wake.

The three-dimensional flow topology at 10° AoA is shown in Fig. 6.2 and depicts the
structure of the separation bubble and trailing vortices, flow from right to left. The
separation bubble forms a crescent like shape, the arced separation and reattachment
lines are clearly marked. The arrows show the direction of the separated shear layer
- 86 -
flow. Three cross-sectional slices through the structure show the recirculation / roll-up
of fluid within the bubble. The trailing vortices are in close proximity to each other,
these drive a strong central downwash between the pair. Their rotation is in the
conventional sense for a typical wing section at positive AoA, the vortex trailing from
the port wing-tip has clockwise rotation (when viewed from downstream) whereas the
starboard trailing vortex rotates counter clockwise. Note the entrainment of fluid into
the vortices from the upper surface boundary layer and also from within the cavity.

6.2 Range of Attack Angles (0° to 45°)


The two-dimensional central flow cross-sections for a range of AoA are shown in Figs
6.3 & 6.4, flow from right to left. At 0° AoA the separation bubble is small (Fig. 6.3a),
the shear layer beneath the disc covers the entire cavity. Reversed flow above the cavity
shear layer occurs centrally for almost the entire length of the cavity. The leading edge
rim forces the central reversed flow to split, deflected towards the wing tips. This
deflection provided the rotation to generate two stagnation points either side of the mid-
span, these spiral nodes were visualised in the surface paint patterns (Fig. 5.42a).
Separation occurs on the trailing edge, the separated shear layers influence the structure
of the bluff body wake.

With increased AoA the separation bubble on the upper surface becomes enlarged,
reattachment occurs further downstream. Also, the shear layer separating off the leading
edge lip reattaches further upstream and the wake downwash angle increases.

At 10° AoA the separation bubble enlarges slightly (Fig. 6.3b), the shear layer beneath
the disc still covers the entire cavity. Reversed flow above the cavity shear layer occurs
centrally for only half the length of the cavity, as the dead air pocket grows larger.

At 20° AoA the upper surface separation bubble enlarges further (Fig. 6.3c), as the
reattachment line moves further downstream. The cavity shear layer now reattaches at
the half chord position, a recirculating bubble is enclosed beneath the shear layer,
downstream from the dead air pocket. Within the cavity, the reattached boundary layer
remains attached throughout the inside of the trailing edge rim, separating off the lip.

At 30° AoA the upper surface separation bubble enlarges further still (Fig. 6.3d). The
cavity shear layer now reattaches at the quarter chord position enclosing the
- 87 -
recirculating bubble directly behind the leading edge rim. The dead air pocket no longer
exists, eliminated by the bubble.

The flow topology for both pre- and post-stall regimes at 45° AoA are shown in Fig.
6.4. Pre-stall (Fig. 6.4a) the upper surface separation bubble is now enlarged to the
extent of being oversized. It is the strong central downwash, driven by the trailing
vortices, that sustains the separation bubble up to such an extreme upper limit AoA. The
separated shear layer is forced back onto the surface aft of the disc centre. The cavity
recirculating bubble does not exist at such a high AoA, instead the boundary layer is
attached to the cavity surface following the profile. The pre-stall wake is narrow, not
much thicker than the disc itself.

Post-stall (Fig. 6.4b) the separated shear layer on the upper surface no longer reattaches
but becomes unstable rolling up into vortex structures shed into the wake. Similarly, the
shear layer separated from the trailing edge lip sheds large vortex structures. A
representation of the reversed flow towards the disc upper surface, within the wake, is
also added to Fig. 6.4b. This reversed flow will create a central nodal reattachment point
on the upper surface, similar to that illustrated in the surface paint visualisations at 30o
AoA, see Fig. 5.41 & 5.43b. The post-stall wake is much thicker, approximately the
width of the disc chord. The flow within the cavity remains largely unchanged.

Aerodynamic stall appears in the load data (Fig 5.2) and the wake flow visualisations
(Fig. 5.48), at 45o AoA as illustrated in the flow topology (Fig. 6.4). Note that, this is
not consistent with Stilley & Carstens (1972) experiments. The author offers the
suggestion that the complicated side sting support structure (Stilley & Carstens, Fig. 3,
1972) accounts for the premature onset of aerodynamic stall, just above 30o AoA. The
interference effects caused by the sting (connected to the rim at one of the wingtips) on
the test model are particularly intrusive at high angles of attack. The sting directly
blocks the fluidic generation of the natural roll up of the trailing vortex on the strut side
of the disc. This weakens the trailing vortex pair and in turn reduces the strength of
central downwash. The partially separated flow regime therefore breaks down at just
above 30o AoA, the trailing vortex pair unable to drive the shear layer back to the
surface.

- 88 -
Chapter 7 Disc-wing Flightpath Simulation

Summary
This chapter presents a six-degree of freedom disc-wing simulation model developed
using Matlab. Section 7.1 outlines the flight equations of motion of a rigid flight vehicle
in three-dimensions. The computer simulation is validated against published Frisbee
trajectory data obtained from free-flight experiments. Flight profiles are discussed for a
number of different launch conditions consistent with a range of typical Frisbee throws.
The simulation is also used to demonstrate that with control moments from suitable
control effectors, it is possible to generate a number of proscribed flight manoeuvres.

7.1 Flight Equations of Motion


The disc-wing simulation is based on the standard equations of motion of a rigid flight
vehicle in three dimensions (Etkin & Reid, 1982). For convenience, four separate axes
systems are used:

1. Earth axes, (xyz )1

2. Body axes, (xyz )2

3. Zero sideslip body axes, (xyz )3

4. Relative wind axes, (xyz )4

The earth axes and body axes are the conventional axes used for aircraft stability and
control. The zero sideslip body axes are obtained by rotating the body axes through -β2,
Fig. 7.1a. The relative wind axes are then obtained by rotating the zero sideslip body
axes through -α3, Fig. 7.1b.

The transformation matrix required to change forces and velocities between coordinate
systems is defined as,

⎡cos θ cosψ sin φ sin θ cosψ − cos φ sinψ cos φ sin θ cosψ + sin φ sinψ ⎤
Ta (φ , θ ,ψ ) = ⎢⎢ cos θ sinψ sin φ sin θ sinψ + cos φ cosψ cos φ sin θ sinψ − sin φ cosψ ⎥⎥ (7.1)
⎢⎣ − sin θ sin φ cos θ cos φ cos θ ⎥⎦

- 89 -
where φ, θ and ψ are the roll, pitch and yaw angles through which the axes system is
rotated (in the order yaw, pitch, roll). The transformation of angular rates about different
axes systems requires a further matrix,

⎡1 sin φ tan θ cos φ tan θ ⎤


Tr (φ , θ ,ψ ) = ⎢⎢0 cos φ − sin φ ⎥⎥ (7.2)
⎢⎣0 sin φ sec θ cos φ sec θ ⎥⎦

where once again, φ, θ and ψ are the roll, pitch and yaw angles through which the axes
system is rotated.

To simplify presentation, a notation system is adopted whereby,

(xyz )2 = Ta (xyz )1 , (xyz )3 = Ta (xyz )2 , (xyz )4 = Ta (xyz )3


12 23 34
(7.3)

and
Ta12 = Ta (φ , θ ,ψ ) , Ta 23 = Ta (0,0,− β 2 ) , Ta34 = Ta (0,−α 3 ,0 ) . (7.4)

The position and orientation of the disc wing is defined by a position vector x and
attitude vector θ. Using standard aircraft notation,

x1 = [ X Y Z ]T , θ1 = [φ θ ψ ]T (7.5)
and
x& 2 = [u v w]T , θ& 2 = [ p q r ]T . (7.6)

Furthermore, using the transformations defined above,

x& 2 = Ta12 x& 1 , θ& 2 = Tr12 θ& 1 . (7.7)

The dynamics of the system is governed by Newton’s second law, which using the
notation defined by Etkin and Reid can be conveniently expressed as,

F2 ~
&x& 2 = − ω x& 2 (7.8)
m

- 90 -
&θ& = I −1M − ω~ θ& (7.9)
2 2 2

where F2 and M2 are the body axes force and moment vectors acting on the disc, m is
the disc mass, I is the disc moment of inertia matrix defined as,

⎡I x 0 0 ⎤ ⎡ 12 I z 0 0⎤
I = ⎢⎢ 0 Iy 0 ⎥⎥ = ⎢⎢ 0 1
2 Iz 0 ⎥⎥ (7.10)
⎢⎣ 0 0 I z ⎥⎦ ⎢⎣ 0 0 I z ⎥⎦

and
⎡ 0 −r q ⎤
ω = ⎢⎢ r
~ 0 − p ⎥⎥
⎢⎣− q p 0 ⎥⎦
(7.11)

The principle task in solving equations (7.8) & (7.9) is now that of obtaining the
aerodynamic forces and moments acting on the disc wing as a function of its velocity,
aerodynamic attitude and rate. The approach adopted is as follows:

Firstly, calculate the sideslip angle of the disc within the body axes system,

⎛ x& 2 (2) ⎞
β 2 = tan −1 ⎜⎜ ⎟⎟ . (7.12)
x
&
⎝ 2 ⎠ (1)

Next, calculate the disc velocity in the zero sideslip body axes system,

x& 3 = Ta23 x& 2 . (7.13)

Using this result, the true angle of attack of the disc can be calculated as follows,

⎛ x& 3 (3) ⎞
α 3 = tan −1 ⎜⎜ ⎟⎟ (7.14)
⎝ x& 3 (1) ⎠

with the orientation of the relative wind axes given by,

x& 4 = Ta34 x& 3 . (7.15)

- 91 -
The aerodynamic force and moment coefficients can now be obtained either from linear
aerodynamic derivatives (Hubbard & Hummel, 2000), or as in the present case, a
mixture of linear derivatives for the rate terms and look up table results based on wind
tunnel experiments for the steady terms. The complete set of force and moment
coefficients used is as follows,
C Drag = C Drag (α ) (7.16)

C Lift = C Lift (α ) (7.17)

C Side = CYr (r3 , x


&4 ) (7.18)

p4 c
C L = C Lp (7.19)
2x
&4

q4 c
C M = C M (α ) + C M q (7.20)
2x
&4

r4 c
C N = C Nr (7.21)
2x
&4

The dimensional forces and moments acting on the disc are then given by,

F4 = q ∞ S C F (7.22)
and,
M 4 = M 2 = q∞ S c C M (7.23)
where,
C F = [−C Drag C Side − C Lift ] , (7.24)

C M = [C L C M C N ] (7.25)
and,
2
q ∞ = 12 ρ ∞ x& 1 . (7.26)

The forces and moments now have to be resolved back in to body axes. This is achieved
as follows,
F2 ( aero ) = Ta 23 ' Ta34 ' F4 (7.27)

M 2 = Tr23 ' Tr34 ' M 4 . (7.28)

- 92 -
Finally, gravitational forces have to be added,

F2 = F2 ( aero ) + Ta12 mg (7.29)

where,
g = [0 0 g ]
T
. (7.30)

Note that gravitational moments are zero since the moment reference point is the centre
of gravity.

Equations (7.8) & (7.9) can now be integrated to give, via suitable axes transformation,
the earth axes position and orientation as required.

7.2 Simulation Implementation


In implementing the mathematical model developed in section 7.1, a choice has to be
made as to how disc angular momentum is represented in equation (7.9). There are two
options: firstly, the body fixed coordinate system can be fixed to the disc such that it
rotates with disc, or secondly, the body fixed coordinated system can be fixed only to
the disc yaw and pitch axes, i.e. in such a manner that it does change orientation as the
disc spins. In the first case, the angular momentum of the disc is included implicitly in
the angular momentum equation. However, in the second case, the angular momentum
of the disc must be calculated separately from the spin rate, and its effect added as an
extra term in equation (7.9). The advantage of the second method over the first is that it
reduces the computation time by a factor of approximately 5 over the first method,
however, higher order dynamics of any wobble at the beginning of the motion may not
be modelled correctly.

A number of numerical experiments based on typical launch conditions were performed


to evaluate the impact of the angular momentum model on the simulation output. It was
found that results were identical when plotted at reasonable scales (e.g. maximum error
magnitudes less than 0.1% of full scale deflection). A further related issue concerns
modelling of disc yaw damping Nr, which leads to disc spin-down during flight.
Numerical experiments showed that the spin rate reduced by approximately 3% during a
typical 4 second flight, based on a value of Nr from Hummel & Hubbard (2002).
However, the impact of the spin-down on the disc trajectory was, once again, not visible

- 93 -
on graphs plotted at sensible scale. For these reasons, unless otherwise stated, the results
presented in this thesis are for a disc rotating at constant speed (no spin-down), and with
non-rotating body axes.

The disc-wing mathematical model was implemented using Matlab 5.3 running on a
233 mHz Pentium PC. Equations of motion were solved numerically using the ODE23s
solver in Matlab, with relative and absolute error tolerance set to the default values of
1x10-3 and 1x10-6, respectively. Typical simulation run times were of the order of 3
minutes for a 5 second flight (non-rotating body axes). Run time increased with
increasing advance ratio.

The simulation was thoroughly debugged using a series of numerical experiments based
on initial conditions for which the output trajectory could be easily predicted from
simple physics. These experiments were designed to show that the implementation of
the mathematical model was physically sound for all combinations of attitude and angle
of attack. Example tests included horizontal release at zero forward speed,
horizontal/vertical launch at zero angle of attack, launch at +/−90o angle of attack, etc.

7.3 Results from Numerical Experiments


7.3.1 Comparison with experimental data
A comparison between an experimental Frisbee trajectory from Hummel (2003) and a
simulated trajectory from the same initial conditions is presented in Fig. 7.2. The results
are qualitatively similar, however, the velocity magnitude for the simulated data shows
a rapid decrease immediately following launch that is not present in the experimental
data. This deceleration is caused by wobble of the disc following launch, which causes
an oscillation in angle of attack of amplitude 15o and hence increase in drag. The author
of the experimental study reported that there was significant disc wobble during the
initial part of the trajectory presented, however, this was the best data available.
Unfortunately there is a lack of detailed published data on Frisbee trajectories and there
is a clear need for further experimental data to provide a fuller validation of the
simulation.

7.3.2 Analysis of a typical Frisbee trajectory


Trajectory plots and flight parameter time histories for a typical Frisbee trajectory are
shown in Fig. 7.3. The disc is launched at a pitch attitude of 15o in the XZ plane
- 94 -
( φ L = 0 o , θ L = 15 o ,ψ L = 0 o ) with a velocity of magnitude 15m/s aligned with the
attitude vector, such that u = 15, v = 0, w = 0 m/s and the angle of attack is zero. The
disc is spinning at 5 revs/sec in a clockwise direction when viewed from above (r
positive). The total flight time is 4 seconds, with discs shown in the trajectory plots
every 0.2 seconds. Discs are drawn eight times their actual size. Non-spinning body
axes are used, with constant spin rate. Black lines represent results from using the full
non-linear aerodynamic model for the steady derivatives, grey lines represent results
from using linearised derivatives.

The full non-linear aerodynamic model case will be discussed first. The trajectory
planform view (Fig. 7.3a) shows the classic ‘S’ shaped flight profile associated with
Frisbee flight in which the disc first turns one way, then turns the opposite way. This ‘S’
shaped profile arises from the relationship between pitching moment and angle of attack
for the disc (Fig. 7.4c). At the start of the flight, with angle of attack close to zero, the
pitching moment is negative. For a positive spin rate, this leads to a positive roll rate
due to precessional effects (equation 39) and hence an increasingly positive roll attitude
(Fig. 7.3c). As the flight progresses to around 2 seconds, the pitch attitude remains
roughly constant (Fig. 7.3g), however the angle of attack is steadily increasing (Fig.
7.3e) due to the increase vertical velocity of the disc. When the angle of attack becomes
greater than 9o, the pitching moment reverses sign and becomes positive, leading to a
negative roll rate and hence an increasingly negative roll attitude. At around 3 seconds,
the speed has reached a minimum point of around 4.5m/s, the angle of attack has
reached a maximum point of around 40o, and the disc is banked at an angle of around
−15o. The disc then turns sharply to the left with the angle of attack decreasing and the
absolute attitude increasing (Fig. 7.3f). Note that the advance ratio varies from 0.3 at the
beginning of the flight to around 1 at 3 seconds due to changes in the disc speed over
this period.

Focusing now on the linear aerodynamic case (grey lines and symbols), it can be seen
that behaviour similar to the non-linear case is obtained initially, however, the rate of
trajectory divergence increases appreciably after 2 seconds. This is due to larger
differences in the aerodynamic models at increased angles of attack. The linear fit
model diverges considerably from the wind tunnel pitching moment data (Fig. 7.4) for
angles of attack greater than 10o, as previously noted.

- 95 -
7.3.3 Effect of Varying Launch Condition
Figs 7.5 to 7.8 show Frisbee trajectories in the XZ and XY planes for various launch
initial conditions. The default initial condition is a launch attitude of
φ L ,θ L ,ψ L = [0 0 0] o and body axis velocity u, v, w = [19 0 0] m/s. Changes from these
default conditions are indicated on each graph. The time interval between each data
point is 0.2 seconds.

The effect of varying advance ratio at launch (obtained by varying spin rate, initial
velocity held constant at 19m/s) is shown in Fig. 7.5a&b. The non-dimensional analysis
presented in section 2.5.1 predicts that the non-dimensional roll rate of the disc
increases with decreasing advance ratio, and the results in Fig. 7.5a qualitatively support
this. At high advance ratios much greater than 1, the disc follows an almost straight path
in the XY plane along the X axis. As the advance ratio decreases to around 0.4, the disc
increasingly turns to the right, however the vertical motion is not much affected. At an
advance ratio of 0.1, the increasingly positive roll attitude of the disc leads to an
increasing rate of descent as the disc ‘slips’ out of the turn. Finally, at an advance ratio
of 0, i.e. a non-spinning disc, the disc pitches leading edge down immediately after
launch and descends with a tumbling motion in the vertical plane.

The next set of data (Fig. 7.6a&b) shows the effect of varying launch attitude. The XZ
data (Fig. 7.6a) shows the classic ‘boomerang’ effect by which a disc can be made to
return to the thrower if launched with sufficient pitch attitude. For the present case, a
velocity reversal occurs for a launch pitch attitude of greater than 30o. Note that the data
for this test was obtained at a high advance ratio of 1.1, so that the disc roll rate was
minimised, and there is little lateral displacement in the XY plane (Fig. 7.6b). The
relationships between launch pitch angle and maximum range is plotted in Fig. 7.7, and
between launch pitch angle and maximum duration in Fig. 7.8. These results shows that,
based on the present simulation, the launch pitch angle for maximum range is
approximately 10o and the launch pitch angle for maximum duration is approximately
20o. Note that range is much more sensitive to launch angle than is duration.

The effect of launch angle of attack for a horizontally launched disc is shown in Figs.
7.9a&b. At zero angle of attack, the resulting trajectory is very similar to the zero
launch attitude case in Fig. 7.5b, however, there is a slight difference due to the +3o
offset of the aerodynamic angle of attack compared to the geometric angle of attack for
- 96 -
the disc (see Fig. 7.4a). As the launch angle of attack is increased to 60o, the behaviour
is qualitatively similar to the effect of launch attitude, though both the range and
duration are much reduced due to the effects of increased drag in the initial part of the
flight. At a launch angle of attack of 90o, i.e. the disc is pitched to 90o and launched
horizontally, the disc decelerates rapidly, immediately following a descending flight
path, as would be expected. Note that the angle of attack tests were performed with an
advance ratio of 0.5, and hence the lateral dispersion in the results is greater than in Fig.
7.6b.

The effect of launch roll attitude on the trajectory of a disc launched at zero angle of
attack and a pitch attitude of 20o is shown in Figs. 7.10a&b. At zero roll attitude, the
disc proscribes the classic ‘S’ shaped flight path. This is more pronounced than the ‘S’
profile shown in Fig. 4 since the advance ratio in the present case is reduced. By
introducing a negative roll angle at launch, the initial curvature to the right is corrected,
and an approximately straight initial flight path is achieved for a roll angle around -7o.

Of particular interest is the relationship between the landing spot of the disc, i.e. the XY
location of the disc when Z = 0, and the launch roll angle, Fig. 7.11. Notice that the disc
landing point follows an ‘S’ shaped locus starting on the left hand side of the X axis
(from the point of view of a person standing at the launch point) and moving to the right
hand side as the launch roll angle is changed from −30o to +30o. The locus is offset in
that the landing point for zero roll angle is approximately 8m to the right of the X axis.
A landing closest to the X axis is achieved for an initial roll angle of approximately -6o.

7.3.4 Frisbee Flight Control


The original purpose of the work presented in this paper was to investigate the
possibility of developing a steerable projectile based on a spin-stabilised disc-wing. For
completeness, two examples of trajectories with fixed control inputs are discussed:
firstly a downward spiral turn and secondly a spiral roll.

Trajectory plots and parameter time histories for the spiral turn are shown in Fig. 7.12.
The turn is set up by launching the disc at an initial bank angle of -45o and a pitch
attitude of -30o, and by applying a constant control input rolling moment. From Fig.
7.12 it can be seen that after about 2 seconds the turn reaches approximately steady
conditions, with an angle of attack of around 15o, an absolute attitude of around 30o, and

- 97 -
a speed of around 7m/s. Note that if the drag is set to zero, it is possible to launch the
disc with a zero pitch attitude and a constant speed horizontal turn is achieved.

Trajectory plots and parameter time histories for the spiral roll are shown in Fig. 7.13.
In this case, the manoeuvre is set up by launching the disc at a pitch attitude of -20o and
an angle of attack of -3o, and by applying a constant pitching moment control input.
Note that the trajectory is such that the angle of attack remains constant at -3o, which is
the angle of attack for zero lift. Since the angle of attack is constant, the lift to drag ratio
of the disc is constant and hence the angle of the flight path in the XZ plane is constant,
despite the velocity reducing.

- 98 -
Chapter 8 Conclusions

1. The disc-wing is a class of tailless, longitudinally unstable flight vehicles that is


inertially stabilised using spin. The disc-wing is most commonly encountered as the
Frisbee sports disc.

2. Due to precessional effects, out of balance pitching moments result in a rolling


motion of the disc as it flies. This behaviour results in the commonly observed ‘drop
off’ behaviour of sports discs, especially towards the end of its flight.

3. The aerodynamic lift and drag characteristics of a sports disc are in keeping with
what would be expected for a finite wing with low aspect ratio. That is, a relatively
shallow lift curve, linear up to high angles of attack (45o) and a drag characteristic
reasonably approximated by a component dependent on the square of the lift
(induced drag) and non lift dependent component (profile drag at zero lift).

4. The pitching moment characteristic of a successful sports disc is unique in that the
trim angle of attack (~ 9o) is approximately coincident with the angle of attack for
best lift to drag ratio (~ 7o), and that the static margin in this region is least positive
(~ 0), i.e. least unstable. For other cross-sectional profiles, e.g. Frisbee-like but with
reduced rim height, stability is strongly negative for all flight angles of attack.

5. Surface pressure data shows that at typical flight angles of attack, the Frisbee-like
profile has a leading edge suction peak similar to that found on conventional airfoil
sections, but also a second suction peak towards the aft of the section. This, coupled
with increased positive pressure in the cavity at the rear of the section, is consistent
with the aftwards shift of the aerodynamic centre implied by the pitching moment
results (conclusion 4).

6. At flight angles of attack, the flow around a sports disc typically exhibits a crescent
shaped separation bubble a short distance down stream of the leading edge, lower
surface leading edge separation and subsequent recirculation in the forward half of
the cavity, and a turbulent wake like flow from the aft rim. Further aft, the wake
flow is dominated by the presence of the roll up of the trailing vortex pair. At higher
angles of attack, the wake vortex pair is stabilised by the separating leading edge

- 99 -
flow, delaying the stall to approximately 45o. Post-stall, the disc acts effectively as a
bluff body, with an associated turbulent wake.

7. Spin has very little effect on the aerodynamic forces and moments acting on a disc-
wing at typical flight advance ratios (~ 1). The largest effect is on the rolling
moment, off surface flow visualisation shows that this is consistent with
asymmetries in the leading edge separation bubble and vortex wake.

8. A six degree of freedom mathematical model of a rotating disc-wing has been


derived and a simplified analysis used to make predictions of the disc pitch and roll
rate in response to aerodynamic moments. A non-dimensional analysis of the roll
rate equation has shown that non-dimensional disc trajectories will be fundamentally
similar if a parameter based on the ratio of the disc static margin to the disc advance
ratio is similar.

9. The disc-wing mathematical model has been implemented successfully in Matlab,


and it has been shown that simulation can be simplified (and hence speeded up) with
no significant loss of accuracy by using a body axis coordinate system that does not
rotate with the disc. The simulation has been validated against a single experimental
trajectory and reasonable agreement achieved. However more experimental flight
test data is needed to provide a more comprehensive validation data base.

10. A comparison has been made between results for a typical throw trajectory based on
using a nonlinear aerodynamic model for steady terms and a model based on linear
derivatives. The results are qualitatively similar, though differences increase for
parts of the trajectory where the angle of attack is above around 10o due to
inaccuracies in the linear pitching moment approximation.

11. The effect of a range of launch conditions has been investigated numerically. It is
shown that:
• A high advance ratio (i.e. high disc spin rate) leads to straighter trajectories in the
XY plane, as would be expected.
• The launch pitch angles for maximum range and maximum duration with φ L = 0 are
10o and 20o respectively.
• The effect of launch angle of attack is qualitatively similar to the effect of launch
pitch attitude for angles of attack less than 60o. At higher angles of attack, increased
drag at launch leads to greatly reduced range and duration.
- 100 -
• The locus of disc landing position as a function of launch roll angle for the typical
launch pitch angle of 20o has been shown to be an ‘S’ shape curve, with the locus
crossing the X axis for a roll angle of -6o.

12. Finally, it has been shown that with appropriate initial conditions and appropriate
control moment input, it is feasible to explore hypothetical disc-wing manoeuvres
such as a spiral turn and a spiral roll.

- 101 -
References

- 102 -
Ali W., Aerodynamics of Rotating Disc Wings, Division of Aerospace Engineering,
The Manchester School of Engineering, University of Manchester, UK, April 1998.

Anderson J.D., Fundamentals of Aerodynamics, 2nd Ed., McGraw-Hill, New York,


1991.

Asai M. & Yamamoto M., Frisbee-The Flow About a Frisbee, J. Japan Soc. Flu. Mech.,
Vol. 17, No. 1, 1998 (in Japanese).

Atsumi S., Pressure Distribution on a Wing with Circular Plan Form, J. Aeronautical
Sci., pp. 499-501, 1937.

Author Unknown, The Goldstein Aeronautical Engineering Research Laboratory, Aero.


Report No. 9100, School of Aeronautical Eng., University of Manchester, U.K., 1991.

Barger V. & Olsson M., Classical Mechanics: A Modern Perspective, McGraw-Hill,


New York, 1973.

Barlow J.B., Rae Jr. W.H. & Pope A., Low Speed Wind Tunnel Testing (3rd Ed.), John
Wiley & Sons, New York, USA, 1999.

Bartlett R.M., The Biomechanics of the Discus Throw: A Review, J. Sports Sci., Vol.
10, pp. 467-510, 1992.

Bauer S.X.S., An Aerodynamic Assessment of Micro-Drag Generators (MDGs), AIAA


98-2621, 16th AIAA Applied Aero. Conf., Albuquerque, NM, June 1998.

Borger J., Identified Flying Object, G2 Supplement, The Guardian, UK, 13 Sept. 2001,
pp 1-3,8,9.

Cook N., Review Supplement (Extract from: The Hunt for Zero Point), The Mail on
Sunday, UK, 12 Aug. 2001, pp 43-46.

Cook N., The Hunt for Zero Point, Century, London, U.K. Sept. 2001.

- 103 -
Cook W. L., Summary of Lift and Lift/Cruise Fan Powered Lift Concept Technology,
NASA Ames Research Center, NASA-CR-177619, Aug. 1993.

Cooper I., Dalzell D. & Silverman E., Flight of the Discus, Report: E.Sc. 472, Div. Eng.
Sci., Purdue University, Layfayette, Indiana, USA, May 1959.

Cooper K.R. (Ed.) et al, Closed-Test-Section Wind Tunnel Blockage Corrections for
Road Vehicles, SAE Special Report SP-1176, Soc. Auto. Eng. Inc., Warrendale, PA,
USA, 1996.

Cotroneo P.W., Biomechanical and Aerodynamical Aspects of the Backhand and


Sidearm Frisbee-Disc Throws for Distance, MS Thesis, Faculty of Phys. Ed., Cal. State
Univ., Hayward, June 1980.

Crook A. & Wood N.J., Measurements and Visualisations of Synthetic Jets, AIAA
2001-0145, 39th Aero. Sci. Meet & Exhibit, Reno, NV, Jan. 2001.

Danowsky B. & Cohanim B., Analysis of a Flying Disc, Senior Project Report No. SP-
S02-006-R-BPD, Iowa State University, Iowa, USA, 26th April 2002.

Deckert W. H., The Lift-Fan Powered-Lift Aircraft Concept Lessons Learned, NASA
Ames Research Center, NASA CR-177616, Sept. 1993.

Demele F.A. & Brownson J.J., Subsonic Longitudinal Aerodynamic Characteristics of


Disks with Elliptic Cross Sections and Thickness-Diameter Ratios from 0.225 to 0.325.,
NASA Ames Research Center, NASA TM-X-556, May 1961.

Demele F.A. & Lazzeroni F. A., Effects of Control Surfaces on the Aerodynamic
Characteristics of a Disk Reentry Shape at Large Angles of Attack and a Mach Number
of 3.5, NASA Ames Research Center, NASA TM-X-576, Aug. 1961.

Fail R., Lawford L.A. & Eyre R.C.W., Low-Speed Experiments on the Wake
Characteristics of Flat Plates Normal to an Air Stream, ARC R&M No. 3120, June
1957.

- 104 -
Frohlich C., Aerodynamic Effects on Discus Flight, Am. J. Physics, Vol. 49, pp. 1125-
32, 1981.

Frost J.C.M. & Earl T.D., The Circular Wing in Forward Flight, Section from the paper:
Flow Phenomena of the Focused Annular Jet, Symposium on Ground Effect
Phenomena, Dept. Aero. Eng., Princeton University, Oct. 1959.

Ganslen R.V., Aerodynamic Factors which Influence Discus Flight, Unpublished


Research Report, University of Arkansas, Arkansas, USA, 1958.

Ganslen R.V., Aerodynamic Forces in Discus Flight, Scholastic Coach, Vol. XXVIII,
Pages 44 & 77, 1959.

Ganslen R.V., Aerodynamic and Mechanical Forces in Discus Flight, Athletic J., Vol.
44, Pages 50, 52, 68 & 88-89, 1964.

Gerloff M., The Discus Body and its Application to V/STOL Aircraft and Space
Vehicles, Aero/Space Engineering, Jan. 1960, pp. 51-56.

Glaskin M., Torque It Up (Frontiers - Emerging Technologies), New Scientist, 2 Feb


2002.

Goto Y., Hiramoto R., Higuchi H. & Meisel I., Vortical Flows Past Cambered Flying
Disks, JSASS 38th Aircraft Symposium, Sendai, Japan, Oct. 2000.

Grasmeyer J.M. & Keennon, Development of the Black Widow Micro Air Vehicle,
AIAA 2001-0127, 39th Aero. Sci. Meet & Exhibit, Reno, NV, USA, Jan. 2001.

Greif R.K., Kelly M.W. & Tolhurst Jr. W. H., Wind-tunnel Tests of a Circular Wing
with Annular Nozzle in Proximity to the Ground, NASA Ames Research Center, NASA
TN-D-317, May 1960.

Greif R.K., Tolhurst Jr. W. H., Large-scale Wind-tunnel Tests of a Circular Plan-form
Aircraft with a Peripheral Jet for Lift, Thrust and Control, NASA Ames Research
Center, NASA TN-D-1432, Feb. 1963.

- 105 -
Grosjean C., Lee G., Hong W., Tai Y.C. & Ho C.M., Micro Balloon Actuators for
Aerodynamic Control, 11th Ann. Int. Workshop on MEMS, Heidelberg, Germany, pp.
166-171, Jan. 1998.

Harrison T., Wood N.J. & Zhong S., Turbulent Separation Control Using Vortex
Generators, RAeS Aerodynamics Research Conference Proc., London, UK, Apr. 2001.

Headrick E.E., Flying Saucer, Patent Application, Ser. No. 3359678, Patented 23rd Dec
1976.

Higuchi H., Goto Y., Hiramoto R. & Meisel I., Rotating Flying Disks and Formation of
Trailing Vortices, AIAA 2000-4001, 18th AIAA Applied Aero. Conf., Denver, CO,
USA, Aug. 2000.

Hubbard M., The Throwing Events in Track and Field, Chapter 6: Biomechanics of
Sport (edited by Vaughan C.L.), pp. 213-38, CRC Press, Boca Raton, Florida, USA,
1989.

Hubbard M. & Hummel S., Simulation of Frisbee Flight, 5th Conf. on Mathematics and
Computers in Sport, G. Cohen (Ed.), University of Technology, Sydney, Australia, June
2000.

Hummel S. & Hubbard M., A Musculoskeletal Model for the Backhand Frisbee Throw,
8th International Symposium on Computer Simulation in Biomechanics, Politecnico di
Milano, Milan, Italy, July 2001.

Hummel S. & Hubbard M., Identification of Frisbee Aerodynamic Coefficients using


Flight Data, 4th International Conference on the Engineering of Sport, Kyoto, Japan,
Sept. 2002.

Hummel S.A., Frisbee Flight Simulation and Throw Biomechanics, Masters Thesis,
University of California, Davis, California, USA, 2003.

- 106 -
Hummel S. & Hubbard M., Implications of Frisbee Dynamics and Aerodynamics on
Possible Flight Patterns, 5th International Conference on the Engineering of Sport,
Davis, California, USA, Sept. 2004.

Johnson S.E.D., FRISBEE A Practitioner’s Manual and Definitive Treatise, Workman


Publishing Co., New York, USA, 1975.

Katz P., The Free Flight of a Rotating Disc, pp. 150-155, Isreal J. Tech., Vol. 6, No. 1-
2, 1968.

Kentzer C.P. & Hromas L.A., Letter to: Mr. Willis B.E. (Unpublished Research
Report), School Aero. Eng., Purdue University, Layfayette, Indiana, USA, 2nd July
1958.

Keyes J.W., Aerodynamic Characteristics of Lenticular and Elliptic Configurations at a


Mach Number of 6, NASA Langley Research Center, NASA TN-D-2606, Aug. 1965.

Laslett M., Boomlab, MSc Thesis, Aston University, Birmingham, UK, 1997.

Lazzara S., Schweitzer C. & Toscano, J., Design and Testing of a Frisbee Wind Tunnel
Balance, Unpublished Report, Code: Engin 12, Brown University, Advisor: Prof.
Karlson S.K.F., May 1980.

Lazzeroni F. A., Experimental Investigation of a Disk-shaped Reentry Configuration at


Transonic and Low Supersonic Speeds: NASA Ames Research Center, NASA TM-X-
652, May 1962.

Lissaman P.B.S., Stability and Dynamics of a Spinning Oblate Spheriod, Unpublished


Document, 2nd Dec. 1994.

Lissaman P.B.S., The Meaning of Lift, AIAA 96-0161, 34th Aero. Sci. Meet & Exhibit,
Reno, NV, Jan. 1996.

Lissaman P.B.S., Disc Flight Dynamics, Unpublished Document, 29th Dec. 1998.

- 107 -
Lissaman P.B.S., Disc Flight Description, Unpublished Document, University of
Southern California, 12th Jan. 1999.

Lissaman P.B.S., Private Communication (with Hubbard M.), 2000.

Lissaman P.B.S., Range of a Free Disc, Unpublished Document, 4th Jan. 2001.

Lissaman P.B.S., Physics of the Far Flung Frisbee, Unpublished Document, 11th Dec.
2003.

Lissaman P.B.S., Upper Bounds on Range of a Free Disc, Unpublished Document, 19th
Dec. 2003.

Lorenz R.D., Flying Saucers, New Scientist, pp. 40-1, 19th June 2004.

Lorenz R.D., Flight and Attitude Dynamics Measurements of an Instrumented Frisbee,


pp. 738-48, Meas. Sci. Tech., Vol. 16, 11th Feb. 2005.

Martin-Henry C., Disc-wing Flow Visualisation, Division of Aerospace Engineering,


The Manchester School of Engineering, University of Manchester, UK, April 2000.

Maskell E.C., A Theory of the Blockage Effects on Bluff Bodies and Stalled Wings in a
Closed Wind Tunnel, ARC R&M No. 3400 Nov. 1965.

Meisel I., Struktur der Umstromung einer angestellten Scheibe, Diplomarbeit,


Universitat Karlsruhe, 2000 (in German).

Mitchell T.L., The Aerodynamic Response of Airborne Discs, MS Thesis, University of


Nevada, Las Vegas, NV, USA, 1999.

Murray D.C., The Avro VZ-9 Experimental Aircraft – Lessons Learned, AIAA-90-
3237-CP, 1990.

Nakamura Y. & Fukamachi N., Visualisation of Flow Past a Frisbee, Fluid Dyn. Res.,
v7, pp31-35, 1991

- 108 -
Peterson H., Calculation of Self-Suspended Flare Trajectories, Denver Research
Institute, NAD/Crane RDTR No. 193, AD731866, 16 Sept. 1971.

Potts J., Stabilising the Spin, Unmanned Vehicles Magazine, Shephard Press, Aug.
2001.

Potts J.R. & Crowther W.J., The Flow Over a Rotating Disc-wing, RAeS Aerodynamics
Research Conference, London, UK, April 2000.

Potts J.R. & Crowther W.J., Visualisation of the Flow Over a Disc-wing, Proc. of the
Ninth International Symposium on Flow Visualization, Edinburgh, Scotland, UK, Aug.
2000.

Potts J.R. & Crowther W.J., Flight Control of a Spin Stabilised Axi-symmetric Disc-
wing, AIAA 2001-0253, 39th Aero. Sci. Meet & Exhibit, Reno, NV, USA, Jan. 2001.

Potts J.R. & Crowther W.J., Application of Flow Control to a Disc-wing UAV, 16th
UAV Systems Conference, Bristol, UK, April 2001.

Potts J.R. & Crowther W.J., Disc-wing UAV: A Feasibility Study in Aerodynamics &
Control, CEAS Aerospace Aerodynamics Research Conference, Cambridge, UK, 10th -
13th June 2002.
Also presented orally as:
Potts J.R. & Crowther W.J., Aerodynamics and Control of a Spin-stabilised Disc-wing,
Session 136-APA-16: CEAS Aerospace Aerodynamics Conference Highlights (Invited),
41st AIAA Aero. Sci. Meet & Exhibit, Reno, NV, USA, Jan. 2003.

Potts J.R. & Crowther W.J., FrisbeeTM Aerodynamics, AIAA 2002-3150, 20th Applied
Aero. Conf. & Exhibit, St. Louis, MI, USA, 24th - 26th June 2002.

Pozzy T., Getting More Distance - How Important Is Disc Speed, Disc Golf World
News, No. 57, Spring 2001.

Pozzy T., Getting More Distance - How The Pros Do It, Disc Golf World News, No. 60,
Winter 2002.

- 109 -
Reed R. D., Lister D. & Huntley J. D., Wingless Flight: The Lifting Body Story, NASA
Dryden Flight Research Center, NASA SP-4220, 1997.

Rose B., Flying Saucers – Avro’s Secrets, Air Pictorial, May 2001, pp 342-7.

Saltzman E. J., Wang K.C. & Iliff K. W., Flight-determined Subsonic Lift and Drag
Characteristics of Seven Lifting-body and Wing-body Reentry Vehicle Configurations
with Truncated Bases, AIAA-99-0383, 37th AIAA Aerospace Sciences Meeting and
Exhibit, Reno, Nevada, USA, Jan. 1999.

Samozwetow A., Diskus und Wind, Der Leichtathletik Trainer, Vol. 2, Pages 407-10,
1970.

Schuurmans M., Flight of the Frisbee, New Scientist, 28th July 1990, pp 37-40.

Soong T.C., The Dynamics of the Discus Throw, J. Appl. Mech., Vol. 98, pp. 531-6,
1976.

Spearman M. L. & Braswell D.O., Aerodynamics of a Sphere and an Oblate Spheroid


for Mach Numbers from 0.6 to 10.5 including the Effect of Test Conditions, NASA
Langley Research Center, NASA TM-109016, Aug. 1993.

Spearman M. L., Supersonic Aerodynamic Characteristics of some Reentry Concepts


for Angles of Attack up to 90 degrees, AIAA-85-1795, 12th AIAA Atmospheric Flight
Vehicles Conference, Snowmass, CO, USA Aug. 1985.

Spearman M. L., Supersonic Aerodynamic Characteristics of Some Reentry Concepts


for Angles of Attack up to 90 degrees, NASA Langley Research Center, NASA-TM-
87645, Nov. 1985.

Stilley G.D. & Carstens D.L., Adaptation of Frisbee Flight Principle to Delivery of
Special Ordnance, AIAA Paper No. 72-982, 1972.

Stilley G.D., Aerodynamic Analysis of the Self-Suspended Flare, Honeywell Inc.,


NAD/Crane RDTR No.199, AD740117, 23 Feb. 1972.

- 110 -
Stone M., Compact Disc Frisbee Aerodynamics, Division of Aerospace Engineering,
The Manchester School of Engineering, University of Manchester, UK, April 1999.

Taylor J.A., Behaviour of the Discus in Flight, Athletic J., Vol. 12, pp. 9-10 & 44-7,
1932.

Terauds J., Technical Analysis of the Discus, Scholastic Coach, Vol. 47, Pages 98, 100,
104, 166, 1978.

Tobak M. & Peake D.J., Topology of Three Dimensional Separated Flows, Ann. Rev.
Fluid Mech., v14, pp61-85, 1982.

Tuck E.O. & Lazauskas L., Lifting Surfaces with Circular Planforms, 19th International
Workshop on Water Waves and Floating Bodies, Cortona, Italy, 28-31 April 2004.
(to appear in Journal of Ship Research, 2005).

Tutjowitsch V.N., Theorie der Sportlichen Wurfe Teil 1, Leistungssport, Vol. 7, pp. 1-
161, 1976.

Voigt H., Wirkungen der Luftkrafte auf die Flugweite beim Diskuswerf, Der
Leibeserzeihung, Vol. 21, Pages 319-26, 1972.

Wagner R., American Combat Planes (3rd Ed.), Doubleday & Co., Garden City, New
York, USA, 1982.

Ware G. M., Investigation of the Low-Subsonic Aerodynamic Characteristics of a


Model of a Modified Lenticular Reentry Configuration, NASA Langley Research
Center, NASA TM-X-756, Dec 1962.

Washabaugh P.D., Bernal L.P., Najafi K. et al, An Approach Toward Wafer Integrated
Micro Air Vehicle, 15th International UAV Systems Conf., Bristol, U.K., Apr. 2000.

White F.M., Fluid Mechanics (4th Ed.), McGraw-Hill, Singapore, 1999.

- 111 -
Wilson J., Roswell Plus 50, Popular Mechanics, July 1997.

Wilson J., America’s Nuclear Flying Saucer, Popular Mechanics, Nov. 2000.

Yasuda K., Flight- and Aerodynamic Characteristics of a Flying Disk, Japanese Soc.
Aero. Space Sci., Vol. 47, No. 547, pp 16-22, 1999 (in Japanese).

Zdravkovich M.M., Flaherty A.J., Pahle M.G. & Skelhorne I.A., Some Aerodynamic
Aspects of Coin-like Cylinders, JFM, v360, pp73-84, 1998.

Zimmerman C.H., Characteristics of Clark Y Airfoils of Small Aspect Ratios, NACA


Langley Memorial Aeronautical Laboratory, NACA-TR-431, 1932.

Zimmerman C.H., Aerodynamic Characteristics of Several Airfoils of Low Aspect


Ratio, NACA Langley Memorial Aeronautical Laboratory, NACA-TN-539, Aug. 1935.

- 112 -
Figures

- 113 -
Figure 2.1 Cross-sectional frisbee-like disc-wing profile.

- 114 -
Figure 2.2 Relationship between the Lift & Drag and the Axial X & Normal Z forces.
The Side force (not shown here) is identical for both systems,
directed out of the page.

- 115 -
Lift, L
Centre of
xac
Zero lift gravity (cg)
pitching xcg location
moment, Mo
Aerodynamic
centre (ac)
location

free
stream
direction Weight, W
+ve moment
Static margin, kn = xac - xcg

Figure 2.3 Definition of reference locations for longitudinal forces and moments acting
a tailless flight vehicle.

- 116 -
L L
Mcg Mcg
Mo Mo

free
free W stream
αtrim -ve W dMcg
stream -ve

Mo +ve

α, L α, L
Mo -ve
dMcg Stable but αtrim +ve Stable and
-ve
dα unbalanced balanced

a) Positively camber, positive static margin b) Negative camber, positive static margin

L Mcg L
Mcg
Mo Mo

free
free stream
W stream dMcg W
+ve

Mo +ve

αtrim +ve α, L α, L
Mo -ve
αtrim -ve
Unstable but Unstable and
dMcg balanced unbalanced
+ve

c) Positive camber, negative static margin d) Negative camber, negative static margin

Figure 2.4 Illustration of longitudinal stability and trim characteristics of a


tailless flight vehicle for permutations of camber and static margin.

- 117 -
x ac /c
0.6

0.5

0.4

0.3

0.2

Zimmerman 1935
Potts & Crow ther 2002

0.1
0 5 10 15 20 25 30
AoA

Figure 2.5 Comparison of the aerodynamic centre location as a function of angle


of attack for a Frisbee (Potts & Crowther, 2002) and a circular planform wing
with a non-axisymmetric airfoil cross-section (Zimmerman, 1935).

- 118 -
a)

b)

Figure 2.6 (a) Disc-wing flight dynamics. (b) Schematic diagram of body fixed axes.
N.B. for a conventional aircraft the nose would point in the positive x-direction.

- 119 -
CD
0.5

0.4

0.3
- 120 -

0.2

Zimmerman 1935
Greif, Kelly & Tolhurst 1960
Demele & Brow nson 1961
0.1 Greif & Tolhurst 1963
Stilley & Carstens 1972
Tutjow itsch 1976
Ali 1998
Yasuda 1999

0
0 0.5 1 CL2 1.5

Figure 3.1 Comparison of aerodynamic load data for circular planform wings, derived from various sources found in the literature,
with chordwise cross-sections as seen to the right of the figure.
Figure 4.1 Wind tunnel support strut fixed to balance beam above wind
tunnel working section (left), pressure lines (capillary tubing)
connected to pressure transducer (right middle) via a scanivalve (right).

- 121 -
Figure 4.2 Wind tunnel control room. Equipment listed as follows:
1. Betz pressure manometer (top left) sitting on top of the
2. Yaw & incidence controller (below 1),
3. Power supply for motor (above 2),
4. Flow speed controller (above 3),
5. Scanivalve stepper control (right of 3),
6. Wind tunnel computer including raw data visual display (right of 5),
7. Data processing computer with pressure contour plot displayed (right of 6).

- 122 -
Figure 4.3 L-shaped support strut and disc-wing mounted vertically at 0o incidence.

- 123 -
Figure 4.4 Surface paint flow visualisation rig configuration, the disc is horizontal at
zero incidence. The disc is mounted on a central axle, which is fixed to a horizontal
crossbar, two vertical struts either side of the balance beam support the crossbar, the
incidence arm is connected to control the angle of attack. Rig designed by Ali (1998).

- 124 -
a)

b)

Figure 4.5 Smoke wire flow visualisation rig configurations: (a) Motor driven axle
connected to centre of vertical test model allows visualisation of spinning disc, up to
24Hz. (b) Geared down motor driven axle connected to rim of test model allows
visualisation of the full central flow field cross-section of a non-spinning disc.

- 125 -
Figure 4.6 Cross-sectional disc-wing profiles.
(a) Frisbee-like (b) Intermediate (c) Flat Plate.

- 126 -
Figure 4.7 Wind tunnel configuration to measure interference and tare effects, the disc
was mounted on a dummy support (left). The dummy strut was a mirror image of the
measuring strut (right) and held the disc in the correct position (on the balance centre)
but not connected to the balance itself. The measuring strut was set with the axle tip
(centre) at around 3mm from the disc, without touching, and was fixed to the balance.

- 127 -
CL CD
2 1
Disc Disc
Strut Strut
1.5 0.8

1 0.6

0.5 0.4

0 0.2
-10 0 10 20 30

-0.5 0
Angle of Attack (degrees) -10 0 10 20 30
Angle of Attack (degrees)

(a) Lift coefficient. (b) Drag coefficient.


CY
1
Disc
Strut
0.5

0
-10 0 10 20 30

-0.5

-1
Angle of Attack (degrees)

(c) Side force coefficient.


CM CR
0.2 0.1
Disc Disc
Strut Strut
0.1
0.05

0
-10 0 10 20 30 0
-0.1 -10 0 10 20 30

-0.05
-0.2

-0.3 -0.1
Angle of Attack (degrees) Angle of Attack (degrees)

(d) Pitching moment coefficient. (e) Rolling moment coefficient.

Figure 4.8 Comparison of the disc test model and the support strut at zero spin rate,
AoA = −10o to 30o, V∞ = 20m/s, Re = 3.78×105.

- 128 -
Figure 4.9 L-shaped rig configuration, to measure the surface pressure distribution.
The capillary tubes were carefully wound around the support strut.

- 129 -
0.6
y/ c

0.5

0.4

0.3

0.2

0.1

a)
0
-0.3 -0.2 -0.1 0 0.1 0.2 0.3
x/ c

0.6 y/c

0.4

0.2

0
-0.6 -0.4 -0.2 0 0.2 0.4 0.6
x/c

-0.2

-0.4

b)
-0.6

Figure 4.10 (a) The pressure tappings in the surface of the disc-wing test models were
positioned on a spiral curve. (b) The spiral curve of points was stepped at 12°
increments, by yawing the model, to achieve full coverage of the disc-wing surface.
That is a polar grid array of 571 points on each surface, upper and cavity.

The non-dimensional position (normalised by the chord c) of the plotted points are
given relative to the distance along the surface profile cross-section from the centre.
- 130 -
- 131 -

Figure 4.11 Schematic diagram of (a) the smoke wire and oil supply system, (b) the oil feeder device.
CL CD
2 1

1.5 0.8

1 0.6

0.5 0.4

0 0.2
-10 0 10 20 30

-0.5 0
Angle of Attack (degrees) -10 0 10 20 30
Angle of Attack (degrees)

(a) Lift coefficient. (b) Drag coefficient.


CY
1

0.5

0
-10 0 10 20 30

-0.5

-1
Angle of Attack (degrees)

(c) Side force coefficient.


CM CR
0.15 0.1

0.1 0.05

0.05 0
-10 0 10 20 30

0 -0.05
-10 0 10 20 30

-0.05 -0.1
Angle of Attack (degrees) Angle of Attack (degrees)

(d) Pitching moment coefficient. (e) Rolling moment coefficient.

Figure 5.1 Load Characteristics at zero spin rate, AoA = −10o to 30o,
V∞ = 20m/s, Re = 3.78×105.

- 132 -
CL
2.5

1.5

0.5

0
-10 0 10 20 30 40 50
-0.5
Angle of Attack (degrees)

(a) Lift coefficient.


CD
1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
-10 0 10 20 30 40 50
Angle of Attack (degrees)

(b) Drag coefficient.


CM
0.25

0.2

0.15

0.1

0.05

0
-10 0 10 20 30 40 50
-0.05
Angle of Attack (degrees)

(c) Pitching moment coefficient.


Figure 5.2 Load Characteristics at zero spin rate, AoA = −10o to 50o,
V∞ = 20m/s, Re = 3.78×105.

- 133 -
CL
2.5

1.5

0.5

0
-120 -90 -60 -30 0 30 60 90 120
-0.5
Angle of Attack (degrees)
-1

(a) Lift coefficient.


CD
1.6

1.4

1.2

0.8

0.6

0.4

0.2

0
-120 -90 -60 -30 0 30 60 90 120
Angle of Attack (degrees)

(b) Drag coefficient.


CM
0.25

0.2

0.15

0.1

0.05

0
-120 -90 -60 -30 0 30 60 90 120
-0.05
Angle of Attack (degrees)
-0.1

-0.15

(c) Pitching moment coefficient.


Figure 5.3 Load Characteristics at zero spin rate, AoA = −100o to 100o,
V∞ = 20m/s, Re = 3.78×105.

- 134 -
CL
2
Re = 1.13*10^5
Re = 1.89*10^5
1.5 Re = 2.84*10^5
Re = 3.78*10^5
Re = 4.73*10^5

0.5

0
-10 0 10 20 30

-0.5
Angle of Attack (degrees)

(a) Lift coefficient.


CD
1
Re = 1.13*10^5
Re = 1.89*10^5
Re = 2.84*10^5
0.8 Re = 3.78*10^5
Re = 4.73*10^5

0.6

0.4

0.2

0
-10 0 10 20 30
Angle of Attack (degrees)

(b) Drag coefficient.


CM
0.15

Re = 1.89*10^5
Re = 2.84*10^5
Re = 3.78*10^5
0.1 Re = 4.73*10^5

0.05

0
-10 0 10 20 30

-0.05
Angle of Attack (degrees)

(c) Pitching moment coefficient.


Figure 5.4 Effect of Reynolds number at zero spin rate, AoA = −10o to 30o
tunnel speed varied to achieve different Re (V∞ = 6, 10, 15, 20, 25m/s).

- 135 -
CL Lift / Drag
2 3

1.5 2

1 1
Figure 5.5 Further load characteristics at zero spin rate,
AoA = −10o to 30o, V∞ = 20m/s, Re = 3.78×105.

0.5 0
-10 0 10 20 AoA ( o) 30

0 -1
CD
0 0.2 0.4 0.6 0.8 1 1.2

-0.5 -2

(a) CL plotted against CD. (b) Lift/Drag ratio plotted against AoA.
- 136 -

C CD
0.1 M,c/4 1

0
CL 0.8
-0.5 0 0.5 1 1.5 2
-0.1

0.6
-0.2

-0.3
0.4

-0.4
0.2
-0.5

-0.6 0
0 0.5 1 1.5 2 CL2 2.5 3

(c) CM (about the quarter chord) plotted against CL. (d) CD plotted against CL2.
CL
2

Zimmerman 1935
Potts & Crowther 2002
1.5 Finite Wing Theory

0.5

0
-10 0 10 20 AoA (o) 30

-0.5

(a) Lift coefficient. Finite wing theory governed by the equation,


CL = fn ( AR, e )(α − α o ) , [AR = 4/π, e = 1, αo = –2°].

CD
1

Zimmerman 1935
0.8 Potts & Crowther 2002
Finite Wing Theory

0.6

0.4

0.2

0 AoA (o)
-10 0 10 20 30

(b) Drag coefficient. Finite wing theory governed by the equation,


1 2
CD = CD 0 + C L , [CDo = 0.1, e = 1, AR = 4/π].
π e AR
Figure 5.6 Comparison of the present experimental data to circular
planform wings, both theoretical and experimental.

- 137 -
CL
2
Stilley & Carstens 1972
Ali 1998
Yasuda 1999
1.5 Hubbard & Hummel 2002
Potts & Crowther 2002

0.5

0
o
-10 0 10 20 AoA ( ) 30

-0.5

(a) Lift coefficient.


CD
1

Stilley & Carstens 1972


Ali 1998
0.8 Yasuda 1999
Hubbard & Hummel 2002
Pot ts & Crowther 2002

0.6

0.4

0.2

0
AoA (o)
-10 0 10 20 30

(b) Drag coefficient.


CM,c/2
0.2
Stilley & Carstens 1972
Ali 1998
0.15 Hubbard & Hummel 2002
Pot ts & Crowther 2002

0.1

0.05

0
-10 0 10 20 AoA (o) 30
-0.05

-0.1

(c) Pitching moment coefficient.


Figure 5.7 Comparison of the present experimental data
to Frisbee-like circular planform wings from the literature.

- 138 -
CL
2.5
Stilley & Carstens 1972
Pot
2 ts & Crowther 2002

1.5

0.5

0
-30 0 30 60 90
-0.5
Angle of Attack (degrees)

(a) Lift coefficient.


CD
1.6

1.4

1.2

0.8

0.6
Stilley & Carstens 1972
0.4 Pot ts & Crowt her 2002

0.2

0
-30 0 30 60 90
Angle of Attack (degrees)

(b) Drag coefficient.


CM,c/2
0.25

0.2

0.15

0.1

0.05

0
-30 0 30 60 90
-0.05
Stilley & Carstens 1972
-0.1 Pott s & Crowther 2002

-0.15
Angle of Attack (degrees)

(c) Pitching moment coefficient.


Figure 5.8 Comparison of the present experimental data
to a similar Frisbee-like circular planform wing.

- 139 -
CL CD C
2 1 0.25 M,c/2
Zimmerman 1935
Greif, Kelly & Tolhurst 1960 Zimmerman 1935
Greif, Kelly & Tolhurst 1960 Zimmerman 1935
Demele & Brownson 1961
Demele & Brownson 1961 0.2 & Hromas 1958
Kentzer
Greif & Tolhurst 1963 0.8 Greif, Kelly & Tolhurst 1960
1.5 Tutjowitsch 1976
Greif & Tolhurst 1963
Tutjowitsch 1976 Demele & Brownson 1961
Potts & Crowther 2002 Greif
Pott s & Crowther 2002 0.15& Tolhurst 1963
Potts & Crowther 2002

1 0.6
0.1

0.4 0.05
0.5

0
AoA ( o)
0 0.2 -10 0 10 20 30
- 140 -

o
-10 0 10 20 AoA ( ) 30 -0.05

0
-0.5 AoA ( o) -0.1
-10 0 10 20 30

(a) Lift coefficient. (b) Drag coefficient. (c) Pitching moment coefficient.
Figure 5.9 Comparison of the present experimental data to a variety of circular planform wings,
with chordwise cross-sections as seen below.
CL
2
Zimmerman 1935
Greif, Kelly & Tolhurst 1960
Demele & Brow nson 1961
Greif & Tolhurst 1963
Stilley & Carstens 1972
1.5 Tutjow itsch 1976
Ali 1998
Yasuda 1999
Hubbard & Hummel 2002
Potts & Crow ther 2002
Finite Wing Theory

1
- 141 -

0.5

0
AoA ( o)
-10 -5 0 5 10 15 20 25 30

-0.5

Figure 5.10 Lift coefficient: Comparison of the present experimental data to that found in the literature for similar circular planform wings,
with chordwise cross-sections as seen to the right of the figure.
CD
1

Zimmerman 1935
Greif, Kelly & Tolhurst 1960
Demele & Brow nson 1961
Greif & Tolhurst 1963
Stilley & Carstens 1972
0.8 Tutjow itsch 1976
Ali 1998
Yasuda 1999
Hubbard & Hummel 2002
Potts & Crow ther 2002
Finite Wing Theory

0.6
- 142 -

0.4

0.2

0
-10 -5 0 5 10 15 20 AoA ( o) 25 30

Figure 5.11 Drag coefficient: Comparison of the present experimental data to that found in the literature for similar circular planform wings,
with chordwise cross-sections as seen to the right of the figure.
CM,c/2
0.25

Zimmerman 1935
Kentzer & Hromas 1958
Greif, Kelly & Tolhurst 1960
0.2 Demele & Brow nson 1961
Greif & Tolhurst 1963
Stilley & Carstens 1972
Ali 1998
Hubbard & Hummel 2002
0.15 Potts & Crow ther 2002

0.1
- 143 -

0.05

0
-10 -5 0 5 10 15 20 AoA ( o) 25 30

-0.05

-0.1

Figure 5.12 Pitching moment coefficient: Comparison of the present experimental data to that found in the literature for similar circular
planform wings, with chordwise cross-sections as seen to the right of the figure.
CL CD
2 1
AdvR 0.00 AdvR 0.00
AdvR 0.17 AdvR 0.17
1.5 AdvR 0.35 0.8 AdvR 0.35
AdvR 0.69 AdvR 0.69
AdvR 1.04 AdvR 1.04
1 0.6

0.5 0.4

0 0.2
-10 0 10 20 30

-0.5 0
Angle of Attack (degrees) -10 0 10 20 30
Angle of Attack (degrees)

(a) Lift coefficient. (b) Drag coefficient.


CY
0.5

0.3

0.1

-10 -0.1 0 10 20 30

AdvR 0.00
AdvR 0.17
-0.3 AdvR 0.35
AdvR 0.69
AdvR 1.04
-0.5
Angle of Attack (degrees)

(c) Side force coefficient.


C CR
0.15 M 0.1
AdvR 0.00 AdvR 0.00
AdvR 0.17 AdvR 0.17
AdvR 0.35 AdvR 0.35
0.1 AdvR 0.69 0.05 AdvR 0.69
AdvR 1.04 AdvR 1.04

0.05 0
-10 0 10 20 30

0 -0.05
-10 0 10 20 30

-0.05 -0.1
Angle of Attack (degrees) Angle of Attack (degrees)

(d) Pitching moment coefficient. (e) Rolling moment coefficient.

Figure 5.13 Load characteristics at various advance ratios,


AdvR = 0 to 1, AoA = −10o to 30o, V∞ = 20m/s, Re = 3.78×105.

- 144 -
CY
0.2

0.1

0
-10 0 10 20 30

AdvR 0.00
-0.1 AdvR 0.17
AdvR 0.35
AdvR 0.69
AdvR 1.04
-0.2
Angle of Attack (degrees)

(a) Side force coefficient.


CM
0.01

0
-5 0 5 10 15

-0.01 AdvR 0.00


AdvR 0.17
AdvR 0.35
AdvR 0.69
AdvR 1.04
-0.02
Angle of Attack (degrees)

(b) Pitching moment coefficient


CR
0.02
AdvR 0.00
AdvR 0.17
AdvR 0.35
0.01 AdvR 0.69
AdvR 1.04

0
-10 0 10 20 30

-0.01

-0.02
Angle of Attack (degrees)

(c) Rolling moment coefficient.

Figure 5.14 Load characteristics (detail) at various advance ratios,


AdvR = 0 to 1, AoA = −10o to 30o, V∞ = 20m/s, Re = 3.78×105.

- 145 -
Figure 5.15 Effect of spin on the Lift Coefficient: Comparison set of spin rates (0, 4, 8, CL CL
2.5 2.5
16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15, 20m/s), AoA = -10o to 30o.

AdvR 0.00
AdvR 0.58 AdvR 0.00
2 AdvR 1.15 2 AdvR 0.38
AdvR 2.30 AdvR 0.69
AdvR 3.46 AdvR 1.38
1.5 1.5 AdvR 2.07

1 1

0.5 0.5

0 0
-10 0 10 20 30 -10 0 10 20 30
-0.5 -0.5
Angle of Attack (degrees) Angle of Attack (degrees)

(a) V∞ = 6m/s (b) V∞ = 10m/s


- 146 -

CL CL
2.5 2.5
AdvR 0.00 AdvR 0.00
2 AdvR 0.23 2 AdvR 0.17
AdvR 0.46 AdvR 0.35
AdvR 0.92 AdvR 0.69
AdvR 1.38 AdvR 1.04
1.5 1.5

1 1

0.5 0.5

0 0
-10 0 10 20 30 -10 0 10 20 30
-0.5 -0.5
Angle of Attack (degrees) Angle of Attack (degrees)

(c) V∞ = 15m/s (d) V∞ = 20m/s


Figure 5.16 Effect of spin on the Drag Coefficient: Comparison set of spin rates (0, 4, 8, 1.4
CD
1.4
CD
16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15, 20m/s), AoA = -10o to 30o.

AdvR 0.00 AdvR 0.00


1.2 AdvR 0.58 1.2 AdvR 0.38
AdvR 1.15 AdvR 0.69
AdvR 2.30 AdvR 1.38
1 AdvR 3.46 1 AdvR 2.07

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
-10 0 10 20 30 -10 0 10 20 30
Angle of Attack (degrees) Angle of Attack (degrees)

(a) V∞ = 6m/s (b) V∞ = 10m/s


- 147 -

CD CD
1.4 1.4
AdvR 0.00 AdvR 0.00
1.2 AdvR 0.23 1.2 AdvR 0.17
AdvR 0.46 AdvR 0.35
AdvR 0.92 AdvR 0.69
1 AdvR 1.38 1 AdvR 1.04

0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
-10 0 10 20 30 -10 0 10 20 30
Angle of Attack (degrees) Angle of Attack (degrees)

(c) V∞ = 15m/s (d) V∞ = 20m/s


Figure 5.17 Effect of spin on the Side Force Coefficient: Comparison set of spin rates (0,
4, 8, 16, 24Hz) tested at various flow speeds (V∞ = 6, 10, 15, 20m/s), AoA = -10o to 30o.
CY CY
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
-10 0 10 20 30 -10 0 10 20 30
-0.2 AdvR 0.00 -0.2 AdvR 0.00
AdvR 0.58 AdvR 0.38
AdvR 1.15 AdvR 0.69
-0.4 -0.4 AdvR 1.38
AdvR 2.30
AdvR 3.46 AdvR 2.07
-0.6 -0.6
Angle of Attack (degrees) Angle of Attack (degrees)

(a) V∞ = 6m/s (b) V∞ = 10m/s


- 148 -

CY CY
0.8 0.8

0.6 0.6

0.4 0.4

0.2 0.2

0 0
-10 0 10 20 30 -10 0 10 20 30
-0.2 AdvR 0.00 -0.2 AdvR 0.00
AdvR 0.23 AdvR 0.17
AdvR 0.46 AdvR 0.35
-0.4 AdvR 0.92 -0.4 AdvR 0.69
AdvR 1.38 AdvR 1.04
-0.6 -0.6
Angle of Attack (degrees) Angle of Attack (degrees)

(c) V∞ = 15m/s (d) V∞ = 20m/s


C C
0.15 M 0.15 M

0.1 0.1
Comparison set of spin rates (0, 4, 8, 16, 24Hz) tested at various
Figure 5.18 Effect of spin on the Pitching Moment Coefficient:

0.05 0.05
flow speeds (V∞ = 6, 10, 15, 20m/s), AoA = -10o to 30o.

0 0
-10 0 10 20 30 -10 0 10 20 30
-0.05 -0.05 AdvR 0.00
AdvR 0.00 AdvR 0.38
AdvR 0.58 AdvR 0.69
-0.1 AdvR 1.15 -0.1 AdvR 1.38
AdvR 2.30 AdvR 2.07
AdvR 3.46
-0.15 -0.15
Angle of Attack (degrees) Angle of Attack (degrees)

(a) V∞ = 6m/s (b) V∞ = 10m/s


- 149 -

C C
0.15 M 0.15 M

0.1 0.1

0.05 0.05

0 0
-10 0 10 20 30 -10 0 10 20 30
-0.05 AdvR 0.00 -0.05 AdvR 0.00
AdvR 0.23 AdvR 0.17
AdvR 0.46 AdvR 0.35
-0.1 AdvR 0.92 -0.1 AdvR 0.69
AdvR 1.38 AdvR 1.04

-0.15 -0.15
Angle of Attack (degrees) Angle of Attack (degrees)

(c) V∞ = 15m/s (d) V∞ = 20m/s


CR CR
0.1 0.1
AdvR 0.00 AdvR 0.00
AdvR 0.58 AdvR 0.38
AdvR 1.15 AdvR 0.69
0.05 AdvR 2.30 0.05 AdvR 1.38
AdvR 2.07
Comparison set of spin rates (0, 4, 8, 16, 24Hz) tested at various

AdvR 3.46
Figure 5.19 Effect of spin on the Rolling Moment Coefficient:
flow speeds (V∞ = 6, 10, 15, 20m/s), AoA = -10o to 30o.

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

-0.05 -0.05

-0.1 -0.1
Angle of Attack (degrees) Angle of Attack (degrees)

(a) V∞ = 6m/s (b) V∞ = 10m/s


- 150 -

CR CR
0.1 0.1
AdvR 0.00 AdvR 0.00
AdvR 0.23 AdvR 0.17
AdvR 0.46 AdvR 0.35
0.05 AdvR 0.92 0.05 AdvR 0.69
AdvR 1.38 AdvR 1.04

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

-0.05 -0.05

-0.1 -0.1
Angle of Attack (degrees) Angle of Attack (degrees)

(c) V∞ = 15m/s (d) V∞ = 20m/s


CL CD
0.6 0.6

Re = 1.13*10^5 Re = 1.13*10^5
0.5 Re = 1.89*10^5 0.5
Re = 1.89*10^5
Re = 2.84*10^5 Re = 2.84*10^5
0.4 Re = 3.78*10^5 0.4 Re = 3.78*10^5

0.3 0.3

0.2 0.2

0.1 0.1

0 0
0 1 2 3 4 0 1 2 3 4
Advance Ratio Advance Ratio

(a) Lift coefficient. (b) Drag coefficient.


CY
0.5

Re = 1.13*10^5
0.4
Re = 1.89*10^5
Re = 2.84*10^5
0.3 Re = 3.78*10^5

0.2

0.1

0
0 1 2 3 4
-0.1
Advance Ratio

(c) Side force coefficient.


Advance Ratio CR
0 0.02
0 1 2 3 4

-0.01
AdvR
0
0 1 2 3 4
-0.02

-0.02
-0.03

Re = 1.13*10^5 -0.04 Re = 1.13*10^5


-0.04 Re = 1.89*10^5 Re = 1.89*10^5
Re = 2.84*10^5 Re = 2.84*10^5
Re = 3.78*10^5 Re = 3.87*10^5
-0.05 -0.06
CM

(d) Pitching moment coefficient. (e) Rolling moment coefficient.

Figure 5.20 Effect of Advance Ratio at 0o AoA for various Reynolds numbers,
tunnel speed varied to achieve different Re (V∞ = 6, 10, 15, 20m/s),
AdvR = 0 up to 3.5, each curve utilises a consistent range of spin rates (0 to 31Hz).

- 151 -
CL CD CM
2 1 0.2
t/c = 0.01 t/c = 0.01
t/c = 0.06 t/c = 0.06
t/c = 0.14 t/c = 0.14 0.15
1.5 0.8
t/c = 0.16 t/c = 0.16

0.1
1 0.6
0.05
0.5 0.4
0
-10 0 10 20 30
0 0.2 t/c = 0.01
-0.05 t/c = 0.06
-10 0 10 20 30
t/c = 0.14
t/c = 0.16
-0.5 0 -0.1
- 152 -

Angle of Attack (degrees) Angle of Attack (degrees)


-10 0 10 20 30
Angle of Attack (degrees)

(a) Lift coefficient. (b) Drag coefficient. (c) Pitching moment coefficient.
Figure 5.21 Comparison of disc-wings with various thicknesses achieved by modifying the rim height to increase the cavity depth,
with chordwise cross-sections as seen below, th / c = 0.01 to 0.16.
CL
2

Hollow Cavity
1.5 Solid - No Cavity

0.5

0
AoA ( o)
-10 0 10 20 30

-0.5

(a) Lift coefficient.


CD
1

Hollow Cavity
0.8
Solid - No cavity

0.6

0.4

0.2

0
AoA ( o)
-10 0 10 20 30

(b) Drag coefficient.


CM,c/2
0.2

Hollow Cavity
0.15
Solid - No Cavity

0.1

0.05

0
-10 0 10 20 AoA ( o) 30

-0.05

-0.1

(c) Pitching moment coefficient.


Figure 5.22 Comparison of the baseline (hollow) Frisbee configuration
to a cavity filled (solid) disc-wing geometry.

- 153 -
Figure 5.23 2D flat surface colour weighted contour plot of pressure coefficients Cp
derived from measurements of the 3D upper surface pressure distribution of a
non-spinning disc-wing, taken over the entire surface,
not over half the disc and mirrored to the other side,
flow direction from top to bottom,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

- 154 -
Figure 5.24 2D flat surface colour weighted contour plot of pressure coefficients Cp
derived from measurements of the 3D cavity surface pressure distribution of a
non-spinning disc-wing, taken over the entire surface,
not over half the disc and mirrored to the other side,
flow direction from top to bottom,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

- 155 -
- 156 -

Figure 5.25 2D flat surface colour weighted contour plot of pressure coefficients Cp derived from measurements of the 3D upper (a) & cavity (b)
surface pressure distribution of a non-spinning disc-wing, taken over the entire surface, not over half the disc and mirrored to the other side,
flow direction from top to bottom, AoA = 5º, V∞ = 15m/s, Re = 2.84×105.
Figure 5.26 Central cross-sectional pressure profile at the half span station. Pressure
coefficients Cp derived from measurements of the 3D surface pressure distribution of a
non-spinning disc-wing. Non-dimensional disc geometry shown below pressure plot,
leading edge on the left of the figure. Unbroken line - upper surface,
dashed line - lower surface, AoA = 5°, V∞ = 15m/s, Re = 2.84×105.
- 157 -
Figure 5.27 Central cross-sectional pressure profile at the half chord station, pressure
coefficients Cp derived from measurements of the 3D surface pressure distribution of a
non-spinning disc-wing. Non-dimensional disc geometry shown below pressure plot, port
wing tip on the left of the figure. Unbroken line - upper surface,
dashed line - lower surface, AoA = 5°, V∞ = 15m/s, Re = 2.84×105.
- 158 -
- 159 -

Figure 5.28 2D flat surface colour weighted contour plots of pressure coefficients Cp for a range of AoA, derived from measurements of the 3D upper
surface pressure distribution of a non-spinning disc-wing, taken over the entire surface, not over half the disc and mirrored to the other side,
flow direction from top to bottom, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.
- 160 -

Figure 5.29 2D flat surface colour weighted contour plots of pressure coefficients Cp for a range of AoA, derived from measurements of the 3D cavity
surface pressure distribution of a non-spinning disc-wing, taken over the entire surface, not over half the disc and mirrored to the other side,
flow direction from top to bottom, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.
Figure 5.30 Central cross-sectional pressure profiles at the half span station, for a range
of AoA. Pressure coefficients Cp were derived from measurements of the 3D surface
pressure distribution of a non-spinning disc-wing. Non-dimensional disc geometry shown
below pressure plot, leading edge on the left of the figure. Unbroken line - upper surface,
dashed line - lower surface, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.
- 161 -
- 162 -

Figure 5.31 Central cross-sectional pressure profiles at the half span station, for a range of AoA. Pressure coefficients Cp were derived from
measurements of the 3D surface pressure distribution of a non-spinning disc-wing. Non-dimensional disc geometry shown below pressure plot, leading
edge on the left of the figure. Unbroken line - upper surface, dashed line - lower surface, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.
Figure 5.32 Cross-sectional pressure profiles at various span stations (a) 1/2 b, (b) 3/8 b,
(c) 1/4 b (d) 1/8 b. Pressure coefficients Cp were derived from measurements of the 3D
surface pressure distribution of a non-spinning disc-wing. Non-dimensional disc cross-
section shown below pressure plot, leading edge on the left of the figure. Unbroken line -
upper surface, dashed line - lower surface, AoA = 5º, V∞ = 15m/s, Re = 2.84×105.
- 163 -
- 164 -

Figure 5.33 Cross-sectional pressure profiles at various span/chord stations (i) 1/2 b, (ii) 3/8 b, (iii) 1/4 b, (iv) 1/8 b, (v) 1/2 c and a range of AoA.
Pressure coefficients Cp were derived from measurements of the 3D surface pressure distribution of a non-spinning disc-wing. Non-dimensional disc
cross-section shown below pressure plot (e), leading edge (i, ii, iii, iv) and port wing-tip (v) on the left of each plot. Unbroken line - upper surface,
dashed line - lower surface, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.
Figure 5.34 2D flat surface colour weighted contour plots of pressure coefficients Cp for
a range of Re, derived from measurements of the 3D upper surface pressure distribution
of a non-spinning disc-wing, taken over the entire surface, not over half the disc and
mirrored to the other side, flow direction from top to bottom,
AoA = 5º, V∞ = 6 to 20m/s, Re = 1.13 to 3.78 ×105.

- 165 -
Figure 5.35 2D flat surface colour weighted contour plots of pressure coefficients Cp for
a range of Re, derived from measurements of the 3D cavity surface pressure distribution
of a non-spinning disc-wing, taken over the entire surface, not over half the disc and
mirrored to the other side, flow direction from top to bottom,
AoA = 5º, V∞ = 6 to 20m/s, Re = 1.13 to 3.78 ×105.

- 166 -
Figure 5.36 Central cross-sectional pressure profiles at the half span station, for a range
of Re. Pressure coefficients Cp were derived from measurements of the 3D surface
pressure distribution of a non-spinning disc-wing. Non-dimensional disc geometry shown
below pressure plot, leading edge on the left of the figure. Unbroken line - upper surface,
dashed line - lower surface, AoA = 5º, V∞ = 6 to 20m/s, Re = 1.13 to 3.78 ×105.
- 167 -
CL
2
Load Data
Integrated Pressure Data
1.5

0.5

0
-10 0 10 20 30

-0.5
Angle of Attack (degrees)

(a) Lift Coefficient.


CD
1
Load Data
Integrated Pressure Data
0.8

0.6

0.4

0.2

0
-10 0 10 20 30
Angle of Attack (degrees)

(b) Drag Coefficient.


CM
0.15

Load Data
0.1 Integrated Pressure Data

0.05

0
-10 0 10 20 30

-0.05
Angle of Attack (degrees)

(c) Pitching Moment Coefficient.


Figure 5.37 Pressure load characteristics, integrated from the pressure data over the 3D
disc-wing geometry, and a comparison to the load measurements, taken from the wind
tunnel balance, AoA = -10º to 30º, V∞ = 20m/s, Re = 3.78×105.

- 168 -
Figure 5.38 Upper surface paint patterns for a non-spinning disc-wing,
flow direction from top to bottom,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

- 169 -
Figure 5.39 Lower (cavity) surface paint patterns for a non-spinning disc-wing,
flow direction from top to bottom,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

- 170 -
- 171 -

Figure 5.40 Upper (a) and Lower (cavity) (b) surface paint patterns for a non-spinning disc-wing,
including superimposed labels to aid the explanation of surface flow features, flow direction from top to bottom,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.
- 172 -

Figure 5.41 Upper surface paint patterns for a non-spinning disc-wing, over a range of AoA,
flow direction from top to bottom,
AoA = 0º to 30º, V∞ = 15m/s, Re = 2.84×105.
- 173 -

Figure 5.42 Lower (cavity) surface paint patterns for a non-spinning disc-wing, over a range of AoA,
flow direction from top to bottom,
AoA = 0º to 30º, V∞ = 15m/s, Re = 2.84×105.
Figure 5.43 Upper surface paint patterns (detail) for a non-spinning disc-wing,
depicting boundary layer reattachment features, flow direction from top to bottom,
(a) Straight line reattachment (half disc) at 20° AoA,
(b) Nodal point reattachment at 30° AoA. V∞ = 15m/s, Re = 2.84×105.

- 174 -
Figure 5.44 Half surface comparison plot: 3D pressure surface colour weighted contour
plots of pressure coefficients Cp derived from measurements of the upper (a) & lower
(cavity) (b) surface pressure distribution over a non-spinning disc-wing,
superimposed onto the half surface flow visualisation images,
flow direction from top to bottom, AoA = 5°, V∞ = 15m/s, Re = 2.84×105.
- 175 -
Figure 5.45 Comparison plot: Central cross-sectional, upper surface pressure
profile at the half span station, superimposed onto a half surface paint flow
visualisation image. Leading edge on the left of the figure. Pressure coefficients
Cp derived from measurements of the upper surface pressure distribution over a
non-spinning disc-wing. Separation (S1, S2) and reattachment (R1) lines are
marked on for comparison, flow direction from left to right,
AoA = 5º, V∞ = 15m/s, Re = 2.84×105.

- 176 -
- 177 -

Figure 5.46 Half surface comparison plot: 3D pressure surface colour weighted contour plots of pressure coefficients Cp derived from measurements of
the upper surface pressure distribution over a non-spinning disc-wing, superimposed onto the half surface flow visualisation images,
over a range of AoA, flow direction from top to bottom, AoA = 0º to 30º, V∞ = 15m/s (flow vis) & 20m/s (pressure), Re = 2.84×105 & 3.78×105.
- 178 -

Figure 5.47 Half surface comparison plot: 3D pressure surface colour weighted contour plots of pressure coefficients Cp derived from measurements of
the lower surface pressure distribution over a non-spinning disc-wing, superimposed onto the half surface flow visualisation images, over a range of
AoA, flow direction from top to bottom, AoA = 0º to 30º, V∞ = 15m/s (flow vis) & 20m/s (pressure), Re = 2.84×105 & 3.78×105.
Figure 5.48 Central cross-section of the flow field over a non-spinning disc-wing,
for a range of AoA, flow direction from right to left.
An electrically heated wire vaporised oil to create smoke,
visualising the flow field when illuminated.
AoA = 0° to 50°, V∞ = 3m/s, Re = 5.67×104.

- 179 -
- 180 -

Figure 5.49 Central cross-section of the leading edge separation bubble over a disc-wing, for a range of AoA, flow direction from right to left.
An electrically heated wire vaporised oil to create smoke, visualising the separation bubble when illuminated by a laser light sheet.
AoA = 0° to 30°, AdvR = 0.7, V∞ = 3m/s, Re = 5.67×104.
- 181 -

Figure 5.50 Central cross-section of the separated shear layer and cavity flow over a non-spinning disc-wing,
for a range of AoA, flow direction from left to right.
An electrically heated wire vaporised oil to create smoke, visualising the cavity flow when illuminated by a laser light sheet.
AoA = 0° to 30°, V∞ = 3m/s, Re = 5.67×104.
Figure 5.51 Visualisation of the wake, downwash and trailing vortex structures
downstream of a non-spinning disc-wing, for a range of AoA. An electrically
heated wire vaporised oil to create smoke, visualising the flow field when
illuminated by ambient lighting (i) or a laser light sheet (ii). The camera was
placed downstream of the disc model, flow direction is out of the page.
The wake cross-sections (ii) were taken at a distance of one diameter from the
trailing edge, AoA = 0° to 10°, V∞ = 3m/s, Re = 5.67×104.

- 182 -
Figure 5.52 Visualisation of the wake, downwash and trailing vortex structures
extending to just over five diameters downstream form the trailing edge
of a non-spinning disc-wing, for a range of AoA. An electrically heated
wire vaporised oil to create smoke, visualising the flow field when
illuminated, flow direction from right to left,
AoA = 0°, 5° & 10°, V∞ = 3m/s, Re = 5.67×104.

- 183 -
Figure 5.53 Cross-sectional slices through the wake, downwash and trailing vortex
structures. Each slice was located aft of a non-spinning disc-wing at a range of
positions, up to two chord lengths downstream from the trailing edge. An electrically
heated wire vaporised oil to create smoke, visualising the flow field when illuminated
by a laser light sheet, flow direction is out of the page.
x/c = 0 to 2, AoA = 10°, V∞ = 3m/s, Re = 5.67×104.

- 184 -
Figure 5.54 Visualisation of flow structures over the upper surface planform including
the separation bubble and wake over a disc-wing, for various AoA and advance ratios.
An electrically heated wire vaporised oil to create smoke, visualising the flow field
when illuminated, flow direction from top to bottom.
i) AdvR = 0 & 0.9, AoA = 0°, V∞ = 1.5m/s, Re = 2.84×104,
ii) AdvR = 0 & 0.7, AoA = 5°, V∞ = 3m/s, Re = 5.67×104.
b) Spinning Disc: Advancing side on the left of the figure, retreating side on the right.

- 185 -
- 186 -

Figure 5.55 Cross-sectional slices through the wake, downwash and trailing vortex structures at various spin rates. Each slice was located aft of a disc-
wing model at a couple of positions, up to two chord lengths downstream of the trailing edge. An electrically heated wire vaporised oil to create smoke,
visualising the flow field when illuminated by a laser light sheet, flow direction is out of the page.
Advancing side on the left of the figure, retreating side on the right. x/c = 1 & 2, AdvR = 0, 0.7 & 1.6, AoA = 10°, V∞ = 3m/s, Re = 5.67×104.
Figure 5.56 Cross-sectional slices through the wake, downwash and
trailing vortex structures at various advance ratios including opposite spin
directions. Each slice was located aft of a disc-wing model at a range of
positions, up to two chord lengths downstream from the trailing edge. An
electrically heated wire vaporised oil to create smoke, visualising the flow
field using laser light sheet illumination, flow direction is out of the page.
x/c = 0 to 2, AdvR = −0.7, 0 & 0.7, AoA = 5°, V∞ = 3m/s, Re = 5.67×104.
i) Advancing side on the left of the figure, retreating side on the right.
iii) Advancing side on the right of the figure, retreating side on the left.

- 187 -
Figure 6.1 Two-dimensional central cross-sectional flow topology for a non-rotating disc-
wing at 10° AoA. The proposed topology is based upon the flow visualisation, depicting
many of the flow structures, namely the separation bubble, cavity flow and turbulent
wake. Flow direction from right to left.
- 188 -
- 189 -

Figure 6.2 Three-dimensional flow topology for a non-rotating disc-wing at 10° AoA.
The proposed topology is based upon the flow visualisation, depicting many of the flow structures,
namely the separation bubble and trailing vortices. Flow direction from right to left.
Figure 6.3 Two-dimensional central cross-sectional flow topology for a non-rotating disc-
wing for a range of AoA. Each proposed topology is based upon the flow visualisation,
depicting many of the flow structures, namely the separation bubble, cavity flow and
turbulent wake. Flow direction from right to left, AoA = 0º to 30º.
- 190 -
Figure 6.4 Two-dimensional central cross-sectional flow topology for a non-rotating
disc-wing at 45º AoA, pre- (a) and post-stall (b). Each proposed topology is based upon
the flow visualisation, depicting many of the flow structures, namely the separation
bubble and turbulent wake. Flow direction from right to left, AoA = 0º to 30º.

- 191 -
ft
Dr

Li
ag
u2 u3
x2 α3 x3
β2
- 192 -

y3 z4
v2 u3 w3=w2
v3=0
y2 x3 z3 x4

(a) Zero sideslip body axes (xyz)3, obtained (b) Relative wind axes (xyz)4, obtained
by rotation of body axes (xyz)2 through -β2 by rotation of zero sideslip body axes
(xyz)3 through -α3

Figure 7.1 Additional axes systems used for the disc-wing simulation.
Figure 7.2 Comparison of simulated results (no dots) with experimental
time histories (dots) from flight tests (flight f2302; Hummel, 2003),
for an identical set of initial conditions, given below.
Xeo Yeo Zeo uo vo wo φo θo ψo po qo ro
m m m m/s m/s m/s rad rad rad rad/s rad/s rad/s
-0.90 -0.63 -0.91 4.48 12.52 1.84 -0.07 0.21 5.03 -26.3 -5.19 52.85

- 193 -
g)

h)
f)
e)

b)

d)
a)

c)
Figure 7.3 Trajectory plots and parameter time histories for a simulated Frisbee flight
path. Solid lines correspond to results from using the full non-linear aerodynamic model.
Grey lines correspond to results from using a linear approximation.
- 194 -
CLift CDrag CM,c/2
2.5 1.2 0.15
Wind Tunnel Data Wind Tunnel Data
2 1 Wind Tunnel Data
Linear Approximation Linear Approximation
0.1 Linear Approximation
1.5 0.8

1 0.6 0.05

0.5 0.4
0
0 0.2 -0.2 0 0.2 0.4 0.6
-0.2 0 0.2 0.4 0.6
- 195 -

-0.5 0 -0.05
Angle of Attack (rads) Angle of Attack (rads)
-0.2 0 0.2 0.4 0.6
Angle of Attack (rads)

(a) Lift coefficient. (b) Drag coefficient. (c) Pitching moment coefficient.
C Lift = C Lift 0 + C Lift α α C Drag = C Drag 0 + C Drag α ( α − α 0 ) 2 C M = C M 0 + C Mα α

[CLift 0 = 0.13, CLift α = 3.09] [CDrag 0 = 0.085, CDrag α = 3.30, α 0 = −0.052] [CM 0 = −0.01, CM α = 0.057]

Figure 7.4 Comparison of the Frisbee-like disc-wing wind tunnel data to the linear approximations for use in the simulation.
The alpha dependent aerodynamic coefficients are governed by the equations beneath each respective figure,
linear derivatives given also. Angle of attack range given in radians, AoA = −0.2 to 0.6 rads.
Z(m)
-10

0 10 20 30 40 50
X(m)
0

10

20 Launch Advance Ratio


0 0.1
0.4 1
8
30
a)

Y(m)
-10

0 10 20 30 40 50
X(m)
0

10

20 Launch Advance Ratio


0 0.1
0.4 1
8
30
b)

Figure 7.5 Effect of launch conditions (advance ratio) on disc-wing XZ and XY


trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o , θ L = 0o , ψ L = 0o , α L = 0o , VL = 19 m/s, AdvR = 0 to 8.

- 196 -
Z(m)
-20
Launch Pitch Angle ( o)
0 10
20 30
45 60
90
-10

-10 0 10 20 30 40
0
X(m)

a)
AdvR = 1.1
10

Y(m)
-20
Launch Pitch Angle ( o)
0 10
20 30
45 60
90
-10

-10 0 10 20 30 40
X(m)
0

b)
AdvR = 1.1
10

Figure 7.6 Effect of launch conditions (pitch angle) on disc-wing XZ and XY


trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o , θ L = 0o to 90°, ψ L = 0o , α L = 0o , VL = 19 m/s, AdvR = 1.1.

- 197 -
X(m)
40
Range (m)

30

20

10

0
0 10 20 30 40 50 60 70 80 90
o
Launch Pitch Angle ( )

Figure 7.7 Range as a function of launch pitch angle.


Maximum range occurs at θ L = 10o .

- 198 -
t (sec)
5
Duration (sec)

0
0 10 20 30 40 50 60 70 80 90
o
Launch Pitch Angle ( )

Figure 7.8 Flight duration as a function of launch pitch angle.


Maximum duration occurs at θ L = 20o .

- 199 -
Z(m)
-10

0 10 20 30 40
0
X(m)

10 Launch Angle of Attack ( o)


0 10
20 30
45 60
90
a) AdvR = 0.5
20

Y(m)
-20
Launch Angle of Attack ( o)
0 10
20 30
45 60
-10 90

0 10 20 30 40
0
X(m)

b) AdvR = 0.5
10

Figure 7.9 Effect of launch conditions (angle of attack) on disc-wing XZ and XY


trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o , θ L = 0o , ψ L = 0o , α L = 0o to 90°, VL = 19 m/s, AdvR = 0.5.

- 200 -
Z(m)
-10

0 10 20 30 40
X(m)
0

10
Launch Roll Angle ( o)
-20 -10
-5 0
5 10
20
AdvR = 0.5
a) 20

Y(m)
-10

0 10 20 30 40
X(m)
0

10 Launch Roll Angle (o)


-20 -10
-5 0
5 10
20 AdvR = 0.5
b) 20

Figure 7.10 Effect of launch conditions (roll angle) on disc-wing XZ and XY


trajectories. The time interval between each data point is 0.2 seconds.
φL = 0o to 90°, θ L = 0o , ψ L = 0o , α L = 0o , VL = 19 m/s, AdvR = 0.5.

- 201 -
Y(m)
-20

Launch Roll Angle = -30o

-20o

-10

-10o

0 10 20 30 X(m) 40
0
-5o
-2o

0o
30o 2o
10

20o

10o 5o
Landing Position
20

Figure 7.11 Disc landing position (XY location at which Z=0) as a function of launch
roll angle. Re-plotted from data in Fig. 57.8a.

- 202 -
g)

h)
f)
e)

b)

d)
a)

c)
Figure 7.12 Trajectory plots and parameter time histories for a simulated spiral turn
manoeuvre obtained using a constant rolling moment control input.
φ L = −45o, θL = −30o, ψL = 0o, αL = 5o, VL = 13.5 m/s, CLcontrol = 0.1, AdvR = 0.5.
- 203 -
g)

h)
f)
e)

b)

d)
a)

c)
Figure 7.13 Trajectory plots and parameter time histories for a simulated spiral roll
manoeuvre obtained using a constant pitching moment control input.
φ L = −0o, θL = −20o, ψL = 0o, αL = -3o, VL = 19 m/s, CMcontrol = 0.2, AdvR = 0.34.
- 204 -
CLu2
4
Uncorrected Data
3 Best Fit Line

0
0 0.2 0.4 0.6 0.8 CDu 1

-1
CLu2 = 3.448 CDu - 0.4015

(a) Uncorrected lift data taken in the wind tunnel CLu2 plotted against uncorrected
drag data CDu, equation of the best fit line given.
CDu
1
CD
CDo
0.8
CDo + CDi

0.6

0.4

0.2

0
2
0 0.5 1 1.5 2 2.5 CLu 3

(b) Disc-wing drag analysis for a lifting wing, after Maskell (Fig. 8, 1965).

CLu2
0.5
Uncorrected Data
0.4 Best Fit Line

0.3

0.2

0.1

0
0 0.2 0.4 0.6 CDu 0.8
CLu2 = 0.892 CDu - 0.0822

(c) Uncorrected wind tunnel data CLu2 plotted against CDu, for the negative AoA range,
equation of the best fit line given.
Figure C.1
- 205 -
CD
1
CDu
0.8 CDc

0.6

0.4

0.2

0
-10 0 10 20 30
Incidence Angle ( o)

(a) AoA –10o to 30o


CD
1.6

CDu
CDc
1.2

0.8

0.4

0
-10 0 10 20 30 40 50
Incidence Angle ( o)

(b) AoA –10o to 50o


CD
1.8

CDu 1.6
CDc 1.4

1.2

0.8

0.6

0.4

0.2

0
-120 -90 -60 -30 0 30 60 90 120
Incidence Angle ( o)

(c) AoA –100o to 100o.


Figure C.2 Corrected and uncorrected drag results for various AoA ranges.
- 206 -
CD
1.6

1.4
1.4
1.29
1.24
1.17 1.19
1.2

1
0.85
0.8

0.6

0.4
0.4

0.2

3D Hemispherical Cup, Rim facing downstream (White 1999)


Disc-wing, Rim facing downstream (Potts & Crowther 2002)
Flat Plate Disc (White 1999)
Flat Plate Disc (Potts & Crowther 2002)
Disc-wing, Rim facing upstream (Stilley & Carstens 1971)
Disc-wing, Rim facing upstream (Potts & Crowther 2002)
3D Hemisperical Cup, Rim facing upstream (White 1999)

Figure C.3 Experimentally measured open face bluff body drag coefficients for various
disc planform configurations (Potts & Crowther, 2002), compared to similar shapes
detailed in White (1999) and Stilley & Carstens (1971), AoA = +90o or –90o.

- 207 -
Appendices

- 208 -
Appendix A History of the Frisbee

A.1 Introduction
‘As the sun gleams through the trees, the lingering memory of the stifling oven heat
fades away, deeply engrossed in the newly discovered leisure activity. Tracking a flying
iron plate as it crosses the grass this way and that, falling over themselves to pluck the
low-flyer out of the sky and re-launch onto the next human target.’

This was the scenario outside the Olds Baking Company in Connecticut, USA around
the middle of the 20th century, workers would toss their deep rimmed backing tins to
and fro during lunch breaks. Through some modification to the basic pie tin shape,
coupled with the birth of plastics in the 1950s, this small town anomaly was exported to
the world as the endearingly familiar form of the Frisbee disc.

A.2 Wham-O’s FrisbeeTM disc


William Russell Frisbie moved to Bransford, Connecticut, USA where he began to
manage a branch of the Olds Baking Company in 1871. His success led to buying the
bakery to form the Frisbie Pie Company, increasing production to supply a number of
East Coast shops. It is this humble pie making family from which the Frisbee derives
not only its name but the basic shape also. There is some dispute as to whether Cookie-
tin lids or Pie-tins were thrown as the first prototype but the commercial value was
realised thereafter with the birth of plastic in the 1940’s and growing interest in flying
saucers from outer space.

Walter Frederick Morrison of West Coast USA bought an injection mould in the late
1940’s and began production of his first flying disc design. His first design was the
somewhat brittle Morrison’s Flyin’ Saucer disc which had a tendency to shatter on
impact with a hard surface. His superior design, the Sputnik, flew much more
successfully and became the inspiration for all subsequent Frisbee designs.

Rich Knerr and A. K. Melin started a toy company on leaving the University of
Southern California in the early 1950’s. They saw Morisson’s Flying Saucers thrown on
Californian beaches and targeted Morrison in 1955 making him a proposition to
increase production. In 1957 the Wham-O Pluto Platter was sold across the US and

- 209 -
succeeding models became collectively labelled Frisbee after the misspelling of the
original Frisbie pie family.

A.3 Professional Sport Disc Development


Many flying disc related sports were developed largely during the 1960’s with Wham-
O’s vice president Ed Headrick driving the development of the Professional Frisbee
Model in 1964 and founding the International Frisbee Association. The Guts Frisbee
Championship among other events were initiated during this period. Ultimate, probably
the most widely played flying disc sport today, was devised in 1969. From small
beginnings the flying disc matured from its toy status to become a serious sport
implement.

Ed Headrick appears as the inventor of the ‘Flying Saucer’ on a patent application


granted in 1976 (Headrick E, 1976), the first of many flying disc designs protected
under law. Disc design also matured driven largely by the increasing popularity of Disc
Golf, which is much the same idea as ‘ball’ Golf. Just as the professional golfer has a
variety of clubs, the pro-disc golfer carries a number of discs for various situations. For
example, a well designed driver disc has minimum drag for maximum range and little
tendency to divert from a straight-line trajectory. The distance record currently stands at
250m held by Christian Sandstrom thrown on 26th April 2002 with an InnovaTM DX
Valkyrie disc.

A more detailed history of the Frisbee disc is given by Johnson (1975).

- 210 -
Appendix B Disc-wing Aircraft R&D

B.1 Introduction
Disc-wing based flight vehicles fall into two distinct categories:
1. Non-rotating, non-axisymmetric body.
2. Spin-stabilised, axisymmetric body.
The first type will typically have a conventional airfoil cross-section when viewed from
the side with a rounded leading edge and sharp trailing edge and thus defined flight
orientation. This type of vehicle has the characteristics of a flying wing aircraft with
low aspect ratio and as such is relatively conventional. The second type of flight vehicle
by definition has an airfoil section with fore and aft symmetry and a centre of gravity at
the centroid of the disc. This configuration will typically be unstable in pitch and for
practical purposes must be inertially stabilised by spinning. Such a disc has no
predefined flight orientation.

B.2 Literature Review: In Brief


Research and development into circular planform air vehicles has a diverse history. The
concept of a circular planform flying wing (XF-5U-1) was developed during the 1930s
& 40s based on a family of three dimensional circular wings with cross-sectional profile
based on the Clark Y airfoil (Zimmerman, 1935).

In the 1950s a circular VTOL (vertical take off and landing) air-vehicle known as the
Avrocar (VZ-9AV) was developed with a central ‘turborotor’ which generated lift and
control forces through a combination of annular nozzles and peripheral jets (Frost &
Earl, 1959; Greif & Tolhurst, 1963; Murray, 1990).

In the early 1960s, NASA researchers investigated the suitability of disk shapes for
reentry vehicles (Demele & Brownson, 1961; Ware, 1962). The proposed lenticular (bi-
convex lens-like) shape was part of a preliminary investigation to develop a lifting-body
(wingless) reentry vehicle with conventional landing capabilities.

In 1972 the U.S. Navy commissioned a project to further the development of a spin-
stabilised self-suspended flare (Stilley & Carstens, 1972; Stilley, 1972), which was
essentially a spin-stabilised axi-symmetric cambered wing with circular planform.

- 211 -
The Moller M200X, a disc planform aircraft with distributed engines driving eight
individual fans, was the culmination of over thirty years of research and development.
Flown in and near hover (1989) the pilot and chief designer, Dr. Paul Moller, furthered
this work to develop a personal air vehicle known as the M400 Skycar (Moller, 1998).

In 1996 Imber & Rogers (1996) looked at a low aspect ratio circulation control wing
with circular planform, similar to the geometry tested by Demele & Brownson (1961),
at zero angle of attack. Annular slots were cut into the rim to allow active tangential
blowing over a Coanda surface to modify the near surface flow structures and the global
circulatory flow. The tangential blowing, particularly at the trailing edge, favourably
modified the lift. The pitching and rolling moments were modified by symmetric and
asymmetric blowing, respectively.

With the recent development of such a wide variety of UAV body shapes it is not
surprising to find the re-emergence of the disc-wing. The ‘Cypher’ VTOL UAV
(Sikorsky Aircraft Corp.) with circular planform encompasses a central rotor and ducted
airstream similar to the Avrocar. The Cypher II Marine variant, ‘Dragon Warrior’, adds
a conventional wing and another smaller ducted tail fan for longer endurance and range.

The SiMiCon Rotor Craft (SRC) UAV (Glaskin, 2002) has a circular planform fuselage
with airfoil cross-section for cruise and retractable rotor blades for VTOL capability.

Slightly smaller was the AeroVironment (DARPA funded, Phase I) Micro Air Vehicle
(MAV), a six-inch diameter circular planform wing, remotely piloted with propeller and
elevons for control, developed in the mid-1990s. Phase II, a rectangular planform MAV
now, due to redefined design criteria, is known as the Black Widow MAV (Grasmeyer
& Keenon, 2001).

Smaller still is a silicon wafer integrated MAV under development by Washabaugh et al


(2000) at the University of Michigan. It comprises an array of Helmholtz resonators
distributed over a thin disc-plate, harnessing synthetic jet technology to provide both lift
and control.

Higuchi et al (2000) investigated the flow over a similar disc-wing using smoke wire
flow visualisation and PIV (particle image velocimetry) measurements aiming to

- 212 -
provide information for UAV applications.

B.3 History of the Circular Planform


The history of the circular planform for aerospace applications is drawn from sporadic
military research and development for various purposes such as signal flares, ground
effect research, supersonic cruisers and re-entry vehicles.

B.3.1 Chance-Vought XF-5U-1


Charles H. Zimmerman conceived the idea for a circular planform aircraft concept in
the 1930’s, based on a family of three dimensional wings with cross-sectional profile
based on the Clark Y airfoil. He won a National Advisory Committee for Aeronautics
(NACA) design competition in 1933 with a disc planform concept aircraft capable of
flying at high speeds and hovering. NACA chose not to fund the project further but
undeterred Zimmerman moved to Chance-Vought in 1937 and began work on an
aircraft based upon this disc-wing design. The preliminary stage of the programme to
develop a high-speed fighter-bomber for the US Navy, which could operate off small
carriers, went ahead. The experimental prototype V-173 was tested in the wind tunnel
and test flown from 23 November 1942 proving Zimmerman’s circular wing concept
practical. It was capable of Short Take Off and Landing (STOL), ideal for carrier
missions, but could not quite manage hover with the existing engines. It did however
demonstrate remarkable low-speed performance, the circular wing planform design
providing the ability to fly at extremely high angles of attack without stalling. The Navy
was impressed and ordered the construction of a successor. The XF-5U-1 aircraft was
completed but never flown, the programme was cancelled in 1947 as jet engines began
to offer greater high-speed advantages.

B.3.2 Avro’s Special Projects Group


In 1947 John Frost, a British aeronautical engineer, joined Avro-Canada to oversee their
CF-100 fighter programme. He became chief of Avro’s Special Projects Group (SPG)
in 1952, formed to develop advanced aircraft for the Canadian Defence Research Board.
The board funded start-up studies to develop a Vertical Take Off and Landing (VTOL)
fighter aircraft capable of supersonic speeds under the codename ‘Project Y’. Frost and
his team began design and feasibility studies to meet these criteria and started to build a
prototype. Development problems and investment limitations brought the involvement
of the US Air Force (USAF) in 1953, who saw great promise in the flying disc concept.

- 213 -
The USAF assigned US scientists to the programme and made US research facilities
available to the SPG. The Air Force renamed the programme ‘Project Silver Bug’
continuing research to develop a supersonic disc-shaped aircraft. Frost and his team
developed various prototypes over the next five years including supersonic fighters and
spy-planes with circular planform. Escalating costs finally halted work on the latest
supersonic version (Weapon System 606A) in 1958 and work began on the joint Army-
Air Force Avrocar (VZ-9AV) programme. Utilising WS 606A research, Frost built a
subsonic Ground Effect Take Off and Landing (GETOL) disc-shaped aircraft, the
‘flying jeep’ concept. The GETOL principle harnessed a circular jet curtain to form an
air cushion, similar to a hovercraft. The Avrocar had a central ‘turborotor’ which
generated lift and control forces through a combination of annular nozzles and
peripheral jets. This enabled the vectoring of air jets to provide lateral thrust but once
the vehicle moved forward it became unstable, increasingly so as it lifted out of ground
effect (~2m). Modifications to solve the stability problems caused the programme to
overrun and in 1960 the Avrocar concept was discontinued.

B.3.3 Lenticular Re-entry Vehicle


In the early 1960s NASA became interested in various geometries conceived for the
development of an unpowered, reusable, wingless, lifting body aircraft capable of
conventional landing after atmospheric reentry. Three prototypes were investigated as
part of a preliminary study (M2-L1, M1-L, L-2C) including the ‘Lenticular Re-entry
Vehicle’. Scientists at the NASA research facilities developed various designs, such as
the Ames M2-F1 ‘flying bathtub’, the M1-L ‘half-cone’ and the Langley L-2C
‘lenticular body’ which were all variations on the wingless lifting body concept. The
lifting body reentry concept had many advantages over non-lifting reentry capsules such
as lower peak heating rates and decelerations during atmospheric reentry, conventional
horizontal landing, manoeuvrability to a pre-selected landing site and a deeper entry
corridor into the atmosphere. The Lenticular body is so called due to the same generic
shape as an optical lens i.e. a convex curve on both the upper and lower surfaces with
circular planform. Such a vehicle would achieve reentry at high angles of attack (~90°)
to utilise the maximum drag exhibited by a disc planform configuration. After reentry a
decrease in AoA would allow the generation of lift and aerodynamic control for
manoeuvrability to achieve stable glide and horizontal landing. After wind tunnel tests
to investigate the feasibility as a practical reentry vehicle, the lenticular lifting body was
considered ‘too exotic’ for the programme. The favoured M2-F1 design was constructed

- 214 -
by NASA Dryden in conjunction with a local glider company that incorporating elevons
and rudders for control. It was first flown by Milton Thompson in 1963 on-tow from a
ground vehicle, enabling a short flight-time of 20 seconds. The prototype was shipped
to NASA Ames where it was wind tunnel tested in the 40’ by 80’ wind tunnel. The data
provided important information for aero-tow flight tests, the aircraft released from
12,000 feet descended at 3600 feet/second into smooth landing at 150 mph. This
demonstrated the lifting body concept and paved the way for subsequent designs, during
the 1960’s the HL-10 and X-24A, revisited in the 1990’s with the X-33 ‘VentureStar’
and X-38 ‘Crew Return Vehicle’.

B.3.4 Self-suspended Flare for Special Ordnance


In the late 1960’s the US Navy began development of a self-suspended flare based on
the Frisbee flight principle. The spin-stabilised axi-symmetric flare was essentially a
cambered wing with circular planform, which would ignite and air-launched to retard
the vertical descent of the flare, avoiding the parasitic weight and volume of
conventional parachute flares. A self combusting pyrotechnic payload contained within
the cavity was ignited providing illumination whilst the flare was held in a horizontal
attitude utilising gyroscopic stabilisation. Wind tunnel tests of analogous spin-stabilised
projectiles such as the Frisbee, clay pigeon, right circular cylinder (low thickness to
chord ratio) were tested to establish the optimum flare configuration and coupled with a
computer simulation of flight trajectories the optimum delivery conditions were
established also.

- 215 -
Appendix C Blockage Corrections & Analysis

C.1 Maskell’s Blockage Correction


Consider Maskell’s well known dynamic pressure equation,

1 S pf 5 S pf
qc
= 1+ CDo + (CDu − CDi − CDo ) (C.1)
qu 2 S wt 2 S wt

where the third term on the RHS accounts for the contribution of separated flow. CDo is
the profile drag, CDi is the induced drag, CDu is the uncorrected drag measured in the
wind tunnel, Spf is the projected frontal area of the model i.e. the area that the flow
‘sees’, Swt is the wind tunnel test section area, qc the corrected dynamic pressure and qu
the uncorrected dynamic pressure. The blockage factor 5/2 defined as adequate for most
purposes (Maskell, 1965), is adequate in this case.

The ratio of the projected disc area to wind tunnel cross-sectional area is an indicator of
the significance of the blockage on the results. At 0o AoA, Spf/Swt ~ 0.01 whereas at 90o
AoA, Spf/Swt ~ 0.07. This suggests that the blockage at typical flight angles of attack the
blockage is negligible but at higher AoA and certainly above the stall AoA the blockage
will likely cause around 10% error in the wind tunnel results.

Now, the corrected drag coefficient is given by,

qc
C Dc = C Du . (C.2)
qu

So to calculate the dynamic pressure correction (C.1) the profile drag CDo = 0.085 is
needed, as is the uncorrected drag CDu which is dependent on the incidence angle and
the induced drag which is given by,

2 dC Du
C Di = C Lu 2
+ ∆C Du − CDo (C.3)
dC Lu

(Barlow, Rae & Pope, 2001).

- 216 -
Plotting a graph of dCLu2 against dCDu (Fig. C.1a) gives an approximately straight line
with gradient dCLu2/ dCDu = 3.4 and CDu axis intercept ∆CDu = 0.1. The linear nature of
the curve suggests that the drag component due to separated flow is minimal and the
induced drag dominate over the pre-stall incidence range. This is more evident when
plotting the disc-wing case, see Fig. C.1b.

Although the boundary layer separates even at 0° AoA the drag increment due to the
separation bubble on the leading edge does not contribute significantly to the total drag,
pre-stall the induced drag dominates. Post-stall, separated flow predominates and CDo,
CDi are considered much smaller than CDu. Approximating equation (C.1) by the form,

qc 5 S pf
= 1+ CDu (C.4)
qu 2 S wt

gives the fully separated flow correction valid for bluff bodies with leading edge
separation only (Cooper et al, 1996).

The negative AoA range can be treated similarly, correcting the drag measurements
using the above analysis, equation (C.1) down to –35° AoA and equation (C.4)
thereafter. However, when calculating the induced drag component equation (C.3),
dCLu2/ dCDu = 0.9 and ∆CDu = 0.1 based on the drag results for negative AoA, see Fig.
C.1c.

Comparing corrected and uncorrected drag results for the various ranges of AoA. The
blockage has little effect on the pre-stall drag (Fig. C.2a) with the corrected drag curve
largely unchanged. However post-stall the fully separated wake creates a large closed
fluidic recirculating separation bubble obstruction, which causes a large error in
uncorrected drag (Fig C.2b). A similar situation occurs for the negative AoA range,
minimal error in pre-stall (0° to –35°) drag, whereas post-stall (-40° to -100°) the
blockage is much greater requiring a significant blockage correction, although the error
is not nearly as much as for the high positive AoA range (Fig. C.2c).

- 217 -
C.2 Analysis
The defining factor for the difference in drag at +/–90° is the rim, which points
upstream at 90° AoA and downstream at –90°. The recirculating bubble aft of the post-
stall disc has maximum diameter equal to or larger than the disc diameter itself, for a
flat plate disc the separation bubble diameter is 1.3 diameters (Fail et al, 1957). For the
disc-wing model at 90° there will be more spillage due to the cavity and rim which
enlarges the separation bubble diameter and increases the blockage in the working
section, which provides a large error in the drag. The validity of the blockage correction
is reinforced by values for the drag coefficients of similar disc shapes found in the
literature with planform normal to the airstream i.e. at +/–90° AoA. Fig. C.3 shows
experimentally measured and corrected drag coefficients for a disc-wing and flat plate
disc compared to those quoted in White (1999) for a 3D hemispherical cup and a flat
plate disc.

The wind tunnel measured and corrected drag results (Potts & Crowther 2002) for the
disc-wing are a CD of 1.29 with the rim pointing upstream and 0.85 when the rim points
downstream. The maximum measured CD, 1.29, for a disc-wing at 90° compares well
with that of Stilley & Carstens (1971, 1972), 1.24, for a similarly shaped disc-wing
(Fig. C.3). The measured disc-wing results lie within the comparable upper and lower
values quoted in White (1999) for a 3D hemispherical cup or shell (Fig. C.3). The cup
shape has a CD of 1.4 with the rim pointing upstream and 0.4 when the rim points
downstream (White, 1999).

Now the disc-wing shape is somewhere in between the hemispherical shell and the flat
plate disc and the measured drag results agree accordingly (Fig. C.3). For a definitive
comparison, the drag coefficient for a flat plate disc (with identical diameter) was
measured in the wind tunnel alongside the disc-wing. The measured and corrected CD
for the flat plate disc was 1.20 which compares well with the quoted values in the
literature namely, 1.17 (White, 1999), 1.18 (Soong, 1976), 1.12 (Fail et al, 1957). So
using Maskells blockage correction for separated flows, the measured CD for the flat
plate disc agrees very well with that quoted recently in the literature (White, 1999). This
provides a good degree of confidence in the blockage correction applied to the disc-
wing also, which was the same method as for the flat plate disc.

- 218 -

View publication stats

You might also like