The Aerodynamics of Gyroplanes

Safety Regulation Group
CAA Paper 2009/02
The Aerodynamics of Gyroplanes
www.caa.co.uk
CAA Paper 2009/02
August 2010
CAA Paper 2009/02 The Aerodynamics of Gyroplanes
Civil Aviation Authority 2010
All rights reserved. Copies of this publication may be reproduced for personal use, or for use within a
company or organisation, but may not otherwise be reproduced for publication.
To use or reference CAA publications for any other purpose, for example within training material for
students, please contact the CAA at the address below for formal agreement.
ISBN 978 0 11792 234 1
Published August 2010
Enquiries regarding the content of this publication should be addressed to:

Safety Investigation and Data Department, Safety Regulation Group, Civil Aviation Authority, Aviation
House, Gatwick Airport South, West Sussex, RH6 0YR.
The latest version of this document is available in electronic format at www.caa.co.uk/publications,

where you may also register for e-mail notification of amendments.
Published by TSO (The Stationery Office) on behalf of the UK Civil Aviation Authority.
Printed copy available from:

TSO, PO Box 29, Norwich NR3 1GN www.tso.co.uk/bookshop
Telephone orders/General enquiries: 0844 477 7300 E-mail: [email protected]
Fax orders: 0870 600 5533 Textphone: 0870 240 3701
List of Effective Pages
Part Page Date Part Page Date
iii August 2010 25 August 2010

iv August 2010 26 August 2010
v August 2010 27 August 2010
vi August 2010 28 August 2010
Contents 1 August 2010 29 August 2010
CAA Introduction 1 August 2010 32 August 2010
CAA Introduction 2 August 2010 33 August 2010
List of Figures and Tables 1 August 2010 34 August 2010
Executive Summary 1 August 2010 41 August 2010
Glossary 1 August 2010 42 August 2010
1 August 2010 47 August 2010
August 2010 Page iii


100 August 2010 148 August 2010
101 August 2010 149 August 2010
102 August 2010 150 August 2010
103 August 2010 151 August 2010
104 August 2010 152 August 2010
105 August 2010 153 August 2010
106 August 2010 154 August 2010
107 August 2010 155 August 2010
108 August 2010 156 August 2010
109 August 2010 157 August 2010
110 August 2010 158 August 2010
111 August 2010 159 August 2010
112 August 2010 160 August 2010
113 August 2010 161 August 2010
114 August 2010 162 August 2010
115 August 2010 163 August 2010
116 August 2010 164 August 2010
117 August 2010 165 August 2010
118 August 2010 166 August 2010
August 2010 Page iv

167 August 2010 Appendix 4 3 August 2010

194 August 2010 Addendum 1 1 August 2010
195 August 2010 Addendum 1 2 August 2010
Acknowledgements 1 August 2010 Addendum 1 3 August 2010
References 1 August 2010 Addendum 1 4 August 2010
Appendix 1 1 August 2010 Addendum 1 10 August 2010
August 2010 Page v

Addendum 3 1 August 2010 Addendum 3 49 August 2010

Addendum 3 11 August 2010
August 2010 Page vi

Contents
CAA Introduction
List of Figures and Tables
Executive Summary
Glossary
Abbreviations
Nomenclature
Report
Introduction
Background 1
Fundamental Research 1
Simulation and Model Validation 2
Application of Research 2
Literature Survey
Methodology 3
Results of Literature Review 4
Conclusions 6
Wind Tunnel Testing

Introduction 6
Wind Tunnel Facility 6
Wind Tunnel Model 6
Test Set-up 7
Test Matrix 8
Wind Tunnel Test Results 8
Configurational Differences Between the VPM M14 and
M16 Variants 15
Concluding Remarks 15
Development of a High Fidelity Gyroplane Simulation

Aims 43
Background 43
A Description of the RASCAL Model 44
Configuring the RASCAL Model to Represent a Gyroplane 46
Results from the Mathematical Model 48
August 2010 Contents Page 1

Flight Testing
Background 54
Flight Test Procedures and Techniques 55
Instrumentation 56
Conduct of Flight Trials 58
Processing of Recorded Data 60
Sample Flight Test Results 60
Model Validation
Methodology 73
Results 74
Discussion 77
Parametric Studies of Configurational Effects

Mass Variation (Figures 7.1 7.3) 93
Vertical c.g. (Figures 7.47.6) 94
Longitudinal c.g. Variation (Figures 7.77.9) 95
Rotor Mast Height (Figures 7.107.12) 95
Propeller Hub Height (Figures 7.137.15) 95
Pod on/Pod off (Figures 7.167.18) 95
Tailplane (Figure 7.197.21) 96
Pod and Tailplane On/Off Figures 7.227.24) 97
Blade Section (Figures 7.257.27) 97
Disc Loading (Figures 7.287.30) 97
Spindle Offset (Figures 7.317.33) 98
Environmental Parameters - Density Altitude (Figures 7.347.39) 98
Review of BCAR Section T

Inroduction 151
Background 151
Analysis 152
Comparison between VPM M16 and Montgomerie-Parsons 155
Analysis of BCAR Section T Chordal Balance Requirement 156
Discussion 156
Assessment of Rotor Blade Torsion Effects

Introduction 159
Background 159
Modelling 160
Results 160
Discussion 160

A Simulation Study of Air Command Autogyros

Introduction 167
Analysis of Two-Seat Air Command GBSSL 168
Analysis of Single-Seat Air Command GBRLB 171
Discussion 173
Investigation of Gyroplane Rotor Teeter Motion

Introduction 184
Teeter Angle Measurements 184
Aircraft Response in Recovery from Large Amplitude Stick
Displacement 184
Overall Conclusions
Acknowledgements
References
Appendix 1
Appendix 2
Appendix 3
Appendix 4
Appendix 5
Appendix 6
Appendix 7
Addendum 1
Addendum 2
Addendum 3

INTENTIONALLY LEFT BLANK
CAA Introduction to Glasgow University Report
1 Foreword
Five fatal accidents to Air Command Gyroplanes between April 1989 and March 1991
resulted in CAA requesting that AAIB undertake an Airworthiness review of the Air
Command. AAIB produced report EW/101/06 Airworthiness Review of Air Command
Gyroplanes dated 12 September 1991. Recommendation 4 of the report (see 2.0
below) was the driver that led to the initial research contract with Glasgow University,
requesting CAA to explore the possibility of assisting the gyroplane fraternity in a
research programme on aerodynamic and airworthiness characteristics of
gyroplanes.
2 AAIB Recommendation
AAIB Report EW/101/06 dated 12th Sept 1991, Airworthiness Review of Air
Command Gyroplanes made the following recommendations.
Recommendations:
1 It is recommended that the approved syllabus for prospective gyroplane pilots
includes training on a dual control gyroplane.
2 It is recommended that the Authority review its procedures for establishing that an
aircraft type has accumulated satisfactory service experience in its country of
origin when assessing it for a Permit to Fly
3 It is recommended that the CAA examine the possibility of some devolution of
Gyroplane modification approval. This will be facilitated by incorporation of Section
T into BCARs at the earliest opportunity.
4 It is recommended that the CAA explore the possibilities of assisting the gyroplane
movement to commission a programme of research into aerodynamic and
airworthiness characteristics.
3 Report Findings
This comprehensive report represents a significant step forward in the understanding
of the aerodynamics of gyroplanes. The conclusions presented are the scientific
results of the tests and studies carried out on a limited number of gyroplane designs.
The report does not cover all configurations of gyroplane designs and therefore the
conclusions may not be directly appropriate to gyroplanes not specifically addressed
by the report. It is important to appreciate that it represents the scientific findings of
the areas addressed and does not attempt to extrapolate beyond those boundaries as
this would be speculation and not appropriate for a scientific research report. Various
conclusions have been made in the report and are noted below:
1 CG / Thrustline offset. The recommendation that the vertical location of the centre
of mass is within 2 inches of the propeller thrust line is a result of the study and
therefore is reported as such. CAA accepts that closer alignment of the CG and the
thrustline is a sensible design aim to achieve pitch dynamic stability (phugoid
mode) but also has flight test experience of a gyroplane design that achieves good
stability but is well outside of the 2 inch criteria. CAA Flight Test Specialist
qualitative assessment implies that pitch dynamic stability may not be solely a
function of CG/Propeller Thrust alignment for all types of Gyroplane. It is
appreciated that in paragraph 8.3.1 (page 152) of the report it is stated that other
factors can affect the phugoid mode.
August 2010 CAA Introduction Page 1

MPD 2005-08 was issued 24 August 2005 mandating, in part, restrictions on pilot
experience, VNE and wind/gust speeds for single seat a/c. These could be removed
if acceptable evidence was presented to show that the CG/thrustline offset was
within 2 inches. However other restrictions noted in the MPD would still apply.
Advisory material to BCAR Section T.23 now includes 2" criteria.
2 Effect of tailplanes. The report also concludes that horizontal tailplanes are largely
ineffective in improving the long term response of pitch dynamic stability (phugoid
mode). This is the result of studies primarily on narrow tandem cockpit enclosures.
A CAA Flight Test Specialist qualitative evaluation of the effects of a horizontal
tailplane on a single side-by-side configuration gyroplane type indicated a degree
of improvement in the phugoid characteristics at higher speed. It is appreciated
that in paragraph 8.3.1 (page 152) of the report it is stated that other factors can
affect the phugoid mode.
3 Teeter Margins. Due to the possibility of excessive teeter angles under certain
phases of flight, that can lead to blade strike with the prop, pod or mast, flight tests
were undertaken. Flight instrumentation revealed small teeter angles in steady
flight conditions, but in one instance during a particularly extreme manoeuvre, the
angle reached 8 deg within an 11 deg safe envelope. Testing was limited, and it is
difficult to draw meaningful conclusions from the data, other than to recognise that
certain more extreme flight conditions will produce teeter angles that may go
outside of the safe envelope, and so could lead to a strike on some part of the
aircraft.
Advisory material to BCAR Section T.143 (a) is to be revised to require satisfactory
control margin and rotor clearance up to 1.1VNE.
4 Rotor aeroelasticity. One aspect of the fatal accidents over the past two decades
that has not been well understood is the sudden loss of rotor speed under certain
circumstances. To address this a study was undertaken to review rotor aeroelastic
instability. This is a study of the design parameters of the blade (e.g. bending
stiffness, torsional stiffness, cg position, lift characteristics, etc.) that affect the
behaviour of the blade in certain flight conditions. For gyroplanes the blade is not
mechanically driven, but generates lift in an autorotative process, relying on air
passing through the disc. Should the blade slow down, lift is reduced. The research
modelled the rotor and varied these design parameters to investigate their
sensitivity on aeroelastic instability (flutter and divergence). Their research
analysed a NACA aerofoil shape supplemented by physical data from a
McCutcheon blade.
Results noted that the key parameter to blade instability is the torsional dynamic
behaviour. To this end the research concluded that a safe rule of thumb would be
to ensure the blade chordwise cg was always forward of the 25% chord. This
would allow for the most pessimistic case of a blade with zero torsional stiffness,
and supports the current Section T 659 (b) requirement. CAA accepts this finding
yet notes that the stabilising affect of a number of other parameters was not
exhaustively investigated during this limited research, which would allow for CAA
acceptance of alternate means of compliance to this requirement that
demonstrates blade stability.
BCAR Section T 659 currently requires blades to have chordwise cg at or forward
of 25% chord.
August 2010 CAA Introduction Page 2

List of Figures and Tables
Figure 3.1: VPM M14 One Third Scale Wind Tunnel Model (dimensions in mm) 16
Figure 3.2: One Third Scale Wind Tunnel Model of a VPM M14 Gyroplane 17
Figure 3.3: The VPM M14 Model Without Cowling 17
Figure 3.4: VPM M14 Model in VZLU Wind Tunnel 18
Figure 3.5: Co-ordinate System for Aerodynamic Coefficients 19
Figure 3.6: Force Coefficient Variations ( = 0, Power On) 20
Figure 3.7: Force Coefficient Variations ( = 0, Power Off) 21
Figure 3.8: Moment Coefficient Variations ( = 0, Power On) 22
Figure 3.9: Moment Coefficient Variations ( = 0, Power Off) 23
Figure 3.10: Force Coefficient Variations ( = -15 deg, Power On) 24
Figure 3.11: Force Coefficient Variations ( = -15 deg, Power Off) 25
Figure 3.12: Moment Coefficient Variations ( = -15 deg, Power On) 26
Figure 3.13: Moment Coefficient Variations ( = -15 deg, Power Off) 27
Figure 3.14: Comparison of Measured and Estimated Cowling Off, Tail Off
Results for = 0, Power On 28
Figure 3.15: Comparison of Measured and Estimated Cowling Off, Tail Off
Results for = 15 deg Power On 29
Figure 3.16: Comparison of Power On and Power Off Force Coefficients,
= 0, Configuration 3 30
Figure 3.17: Comparison of Power On and Power Off Moment Coefficients,
= 0, Configuration 3 31
Figure 3.18: Comparison of Power On and Power Off Force Coefficients,
= -15 deg, Configuration 3 32
Figure 3.19: Comparison of Power On and Power Off Moment Coefficients,
= -15 deg, Configuration 3 33
Figure 3.20: Effect of Rudder Setting on Force Coefficients, = 0,
Configuration 3 34
Figure 3.21: Effect of Rudder Setting on Moment Coefficients, = 0,
Configuration 3 35
Figure 3.22: Effect of Sideslip Angle on Cowling On, Tail On Configuration
(Power Off) 36
Figure 3.23: Effect of Sideslip Angle on Cowling On, Tail Off Configuration
(Power Off) 37
Figure 3.24: Effect of Sideslip Angle on Cowling Off, Tail On Configuration
(Power Off) 38
August 2010 List of Figures and Tables Page 1

Figure 3.25: Effect of Sideslip Angle on Cowling Off, Tail Off Configuration (
Power Off) 39
Figure 3.26: Effect of Sideslip Angle on Cowling On, Tail On Configuration
(Power On) 40
Figure 3.27: Effect of Tailplane Endplates on CZ and CM, = 0, Power Off,
Cowling On 41
Figure 3.28: Effect of an Extended Tail Boom on the Cowling On, Tail On
Configuration (Power On) 42
Figure 4.1: Trim Results from RASCAL Mathematical Model for VPM M16 51
Figure 4.2: Response of VPM M16 to a 5s Pulse in Long. Stick (40mph, IAS) 52
Figure 4.3: Eigenvalues for VPM M16 53
Figure 4.4: Eigenvalues for VPM M16 Plotted Against Damping Ratio 53
Figure 5.1: VPM M16 G-BWGI in Unmodified Form 62
Figure 5.2: G-UNIV at FRA Bournemouth Airport for Initial Trials
(October 2000) 62
Figure 5.3: Standard Test Inputs 63
Figure 5.4: Flight Test Instrumentation for VPM M16 G-BWGI Flight Trials 64
Figure 5.5: G-UNIV Flight Test Instrumentation 66
Figure 5.6: Additional Modification of G-UNIV for Teeter Angle Measurement 68
Figure 5.7: Longitudinal Stick Frequency Sweep (VPM M16) 70
Figure 5.8: Longitudinal Stick Doublet (Montgomerie) 71
Figure 5.9: Trim with Teeter Measurement (Montgomerie) 72
Figure 6.1: VPM M16, Identified Model Verification, 30mph 79
Figure 6.2: VPM M16, Identified Model Verification, 70mph Short Term
Response 80
Figure 6.3: VPM M16, Identified Model Verification, 70mph Long Term
Response 81
Figure 6.4: Comparison of Flight and Model Data (VPM M16, Trim) 82
Figure 6.5: Comparison of Flight and Model Data (Montgomerie, Trim) 83
Figure 6.6: Comparison of Flight and Model Data, Derivatives, VPM M16 84
Figure 6.7: Comparison of Flight and Model Data, Derivatives, Montgomerie 86
Figure 6.8: Non-linear Model Comparison with Flight Test (VPM M16,
doublet input, 70mph) 88
Figure 6.9: Non-linear Model Comparison with Flight Test
(Montgomerie, Phugoid Input, 50mph) 89
(Montgomerie, Doublet Input, 50mph) 90
(Montgomerie, Doublet Input, 55mph) 91
Figure 7.1a: Trim Comparison for Mass Variation (Airframe Parameters) 99

Figure 7.1b: Trim Comparison for Mass Variation (Rotor Parameters) 100
Figure 7.2a: Comparison of Stability Modes (Mass Variation) 101
Figure 7.2b: The Phugoid Mode (Mass Variation) 101
Figure 7.3: Aircraft Response to 5s Pulse of Longitudinal Rotor Tilt -
Phugoid (Mass Variation) 102
Figure 7.4a: Trim Comparison for Vertical C.G. Variation (Airframe Parameters) 103
Figure 7.4b: Trim Comparison for Vertical C.G. Variation (Rotor Parameters) 104
Figure 7.5a: Comparison of Stability Modes (Vertical C.G. Variation) 105
Figure 7.5b: The Phugoid Mode (Vertical C.G. Variation) 105
Phugoid (Vertical C.G. Variation) 106
Figure 7.7a: Trim Comparison for Longitudinal C.G. Variation
(Airframe Parameters) 107
Figure 7.7b: Trim Comparison for Longitudinal C.G. Variation
(Rotor Parameters) 108
Figure 7.8a: Comparison of Stability Modes (Longitudinal C.G. Variation) 109
Figure 7.8b: The Phugoid Mode (Longitudinal C.G. Variation) 109
Phugoid (Longitudinal C.G. Variation) 110
Figure 7.10a: Trim Comparison for Mast Height Variation (Airframe Parameters) 111
Figure 7.10b: Trim Comparison for Mast Height Variation (Rotor Parameters) 112
Figure 7.11a: Comparison of Stability Modes (Mast Height Variation) 113
Figure 7.11b: The Phugoid Mode (Mast Height Variation) 113
Phugoid (Mast Height Variation) 114
Figure 7.13a: Trim Comparison for Propeller Thrust Line Variation
Figure 7.13b: Trim Comparison for Propeller Thrust Line Variation
Figure 7.14a: Comparison of Stability Modes (Propeller Thrust Line Variation) 117
Figure 7.14b: The Phugoid Mode (Propeller Thrust Line Variation) 117
Phugoid (Propeller Thrust Line Variation) 118
Figure 7.16a: Trim Comparison for Pod On/Pod Off (Airframe Parameters) 119
Figure 7.16b: Trim Comparison for Pod On/Pod Off (Rotor Parameters) 120
Figure 7.17a: Comparison of Stability Modes (Pod On/Pod Off) 121
Figure 7.17b: The Phugoid Mode (Pod On/Pod Off) 121
Phugoid (Pod On/Pod Off) 122

Figure 7.19a: Trim Comparison for Tailplane On/Tailplane Off

Figure 7.19b: Trim Comparison for Tailplane On/Tailplane Off
Figure 7.20a: Comparison of Stability Modes (Tailplane On/Tailplane Off) 125
Figure 7.20b: The Phugoid Mode (Tailplane On/Tailplane Off) 125
Phugoid (Tailplane On/Tailplane Off) 126
Figure 7.22a: Trim Comparison for Pod and Tailplane On/Pod and Tailplane Off
Figure 7.22b: Trim Comparison for Pod and Tailplane On/Pod and Tailplane
Off (Rotor Parameters) 128
Figure 7.23a: Comparison of Stability Modes (Pod and Tailplane On/Pod and
Tailplane Off) 129
Figure 7.23b: The Phugoid Mode (Pod and Tailplane On/Pod and Tailplane Off) 129
Phugoid (Pod and Tailplane On/Pod and Tailplane Off) 130
Figure 7.25a: Trim Comparison for Different Blade Sections
Figure 7.25b: Trim Comparison for Different Blade Sections
Figure 7.26a: Comparison of Stability Modes (Different Blade Sections) 133
Figure 7.26b: The Phugoid Mode (Different Blade Sections) 133
Phugoid (Different Blade Sections) 134
Figure 7.28a: Trim Comparison for Disc Loading Variation
Figure 7.28b: Trim Comparison for Disc Loading Variation
Figure 7.29a: Comparison of Stability Modes (Disc Loading Variation) 137
Figure 7.29b: The Phugoid Mode (Disc Loading Variation) 137
Phugoid (Disc Loading Variation) 138
Figure 7.31a: Trim Comparison for Spindle Offset Variation
Figure 7.31b: Trim Comparison for Spindle Offset Variation
Figure 7.32a: Comparison of Stability Modes (Spindle Offset Variation) 141
Figure 7.32b: The Phugoid Mode (Spindle Offset Variation) 141
Phugoid (Spindle Offset Variation) 142

Figure 7.34a: Trim Comparison for Temperature Variation

Figure 7.34b: Trim Comparison for Temperature Variation
Figure 7.35a: Comparison of Stability Modes (Temperature Variation) 145
Figure 7.35b: The Phugoid Mode (Temperature Variation) 145
Phugoid (Temperature Variation) 146
Figure 7.37a: Trim Comparison for Altitude Variation (Airframe Parameters) 147
Figure 7.37b: Trim Comparison for Altitude Variation (Rotor Parameters) 148
Figure 7.38a: Comparison of Stability Modes (Altitude Variation) 149
Figure 7.38b: The Phugoid Mode (Altitude Variation) 149
Phugoid (Altitude Variation) 150
Figure 8.1: Comparison of Flight Identified Modes. VPM and Montgomerie 158
Figure 9.1: McCutcheon Blade Torsion Mode Shape 161
Figure 9.2: McCutcheon Blade Tip Torsional Deflection with Load 161
Figure 9.3: Effect of Including Simple Blade Torsional Model 162
Figure 9.4: Correlation with Flight 166
Figure 10.1: Schematic of Main Rotor Thrust Line Relative to c.g. in
Undisturbed and Disturbed Flight - Propeller Thrust Line
Passing Through c.g. 174
Figure 10.2: Schematic of Main Rotor Thrust Line Relative to c.g. in
Undisturbed and Disturbed Flight - Propeller Thrust Line
Passing Below c.g. 174
Figure 10.3: Two seat Air Command 175
Figure 10.4: Stick Fully Aft 175
Figure 10.5: Stick Fully Forward 176
Figure 10.6: G-BOOJ Stick Position with Airspeed 176
Figure 10.7: Stick Gradient Comparisons 177
Figure 10.8: Two Seat Air Command GBSSL 177
Figure 10.9: Trim Comparisons, Simulation of GBSSL 178
Figure 10.10: Derivative Comparisons, Simulation of GBSSL 179
Figure 10.11: Comparison of Longitudinal Stability Roots 180
Figure 10.12: Single Seat Air Command G-BRLB 180
Figure 10.13: Trim Comparisons, Simulation of G-BRLB 181
Figure 10.14: Derivative Comparisons, Simulation of G-BRLB 182
Figure 10.15: Stability Comparisons, Simulation of G-BRLB 183
Figure 11.1: Blade Teeter Angles for all 3 Flights 187

Figure 11.2: Blade Teeter Angles Expressed in Multi-blade Co-ordinates

for Flight 2 188
for Flight 3 188
for Flight 4 189
Figure 11.5: Teeter Angle During Recovery from Large Amplitude Input 189
Figure 11.6: Time Histories for Recovery from Large Displacement
Longitudinal Stick Input 190
Figure 11.7: Teeter Angle vs. Azimuth for 948 - 952s of Flight 4 190
Figure 11.8: Stick Aft, Blade Teetered Through Propeller Disc 191
Figure 11.9: Stick Aft, 8 of Teeter 192
Figure 11.10: Teeter Measurement for a Propeller Strike 193
Table 4.1: RASCAL Mathematical Model Description 45

Table 5.1: Summary of VPM M16 Flight Trials 54
Table 5.2: Summary of Montgomerie-Parsons Two-Place flight trials
(model validation) 55
Table 5.3: Summary of Montgomerie-Parsons Two-Place Flight Trials
(Teeter Angle Measurement) 55
Table 5.4: Instrumentation Used in Flight Tests 57
Table 7.1: Summary of Simulation Model Parametric Variation 93
Table 8.1: Identified G-BWGI Short-period Mode Characteristics 154
Table 8.2: Identified G-BWGI long-period mode characteristics 155
Table 8.3: Identified G-BWGI long-period mode characteristics, Section T 155
Table 10.1: Calculated Mass and Inertia Properties, from Measurements 169
Table 10.2: Measured Geometric Properties 170
Table 10.3: Measured Mass and Inertia properties 171
Table 10.4: Measured Geometric Properties 172
Appendix 1
Table A1.1: Wind Tunnel Tests Conducted With Power On 1
Table A1.2: Wind Tunnel Tests Conducted With Power Off 3
Table A1.3: Additional Wind Tunnel Tests 4
Appendix 2
Figure A2.1: The VPM M16 Gyroplane 2
Figure A2.2: The Montgomerie-Parsons Two Persons Gyroplane 4

Appendix 3
Figure A3.1: Measurement of Longitudinal c.g. Position 1
Figure A3.2: Measurement of Vertical c.g. Position 2
Appendix 4
Figure A4.1: Conventions for Body Axes Set and State Variables 2
Figure A4.2: Characteristic Motions for Dynamic and Static Stability
(Reproduced and Modified from Figure 6.1: "Dynamics of Flight",
B. Etkin, 2nd Edition, Wiley, 1982) 9
Figure A4.3: Free Vibration of an Underdamped 2nd Order System
(x0 = 1.0, n = 5 rad/s, = 0.1) 10
Figure A4.4: Effect of Damping Ratio on System Response
(x0 = 1.0, n = 5 rad/s) 11
Figure A4.5: Stability Presented on Complex Plane 12
Table A4.1: State Variables for a Conventional 6 Degree of Freedom

Aircraft Model 2
Appendix 5
Figure A5.1: CAA Flight Test Card 7
Appendix 6
Figure A6.1: Attitude Angles During Phugoid Flight Test (70mph, VPM) 2
Figure A6.2: Summary of Rate Gyro Consistency Results (VPM) 3
Figure A6.3: Summary of Attitude Gyro Consistency Results (VPM) 4
Figure A6.4: Attitude Rates Recorded During a for/aft Doublet Input (VPM) 5
Appendix 7
Figure A7.1: Fit Quality of Flight and Identified Model Pitching Moment
Equation 3

Executive Summary
This report describes all work undertaken for the Civil Aviation Authority between November
1993 and June 2008 under the study entitled Aerodynamics of Gyroplanes and whose
genesis is to be found in the 1991 Air Accidents Investigation Branch review of the
airworthiness of Air Command series of gyroplanes. One recommendation of that review was
the commissioning of a programme of research into the aerodynamic and airworthiness
characteristics of light gyroplanes in general. Accordingly, the Authority tasked the University
of Glasgow with this work, and subsequent associated studies, by means of analytical, wind
tunnel and flight test activities.
The fundamental approach to such problems is to build a mathematical engineering model of
the aircraft so that design features, changes or hypothetical situations can be examined in
safety. However, such a model requires validation against actual flight test data in order that a
degree of confidence in its predictive capability can be established. This methodology is
consistent with contemporary industry practice, and was the one adopted here. Accordingly,
much of the work undertaken focussed on checking the fidelity of the model against flight test
data gathered during a number of campaigns using two fully instrumented light gyroplanes: a
VPM M16 Tandem Trainer; and a Montgomerie-Parsons Two-Place (Modified) aircraft. These
trials used specialised flight test techniques involving stylised and, to the lay pilot, unusual
control inputs. The resulting data were then analysed using sophisticated mathematical
analysis techniques, again typical of that used in industry and national research laboratories
worldwide.
The resulting validation confirmed that the model offered very good to excellent fidelity in most
aspects associated with the prediction of vehicle behaviour, certainly well up to the level
obtained by other agencies with helicopter modelling. As a result the model was deemed
suitable for use in a parametric study that posed a range of questions; e.g. what if the tailplane
is removed; what if the mass is increased; what if the rotor is raised on the airframe. It was
discovered that the static and dynamic stability, and hence compliance with BCAR Section T,
is sensitive to the vertical location of the centre of mass in relation to the assumed propeller
thrust line and this was found to be the case across the speed range. No other parameter had
any significant effect on the aircrafts likelihood of satisfying BCAR Section T. The simulation
study, together with an analysis of BCAR Section T made in the context of the VPM M16 flight
trials, resulted in the advisory recommendation that light gyroplanes certified against Section
T should ensure that the vertical location of the centre of mass is within 2 inches of the
propeller thrust line. A handling study of five single-seat gyroplanes was made independently
by the Civil Aviation Authority, and it was found that only one satisfied Section T dynamic
stability requirements - that was the machine that was specifically modified to comply with the
2 inch recommendation thereby providing independent verification of this recommendation.
These data are also included in this report.
The Report also presents results from a number of associated studies. First, a literature review
highlights a body of work that has increased during the 16 years this research has taken place,
indicating a growing interest in the gyroplane configuration although there remains little
indication that rigorous scientific or engineering investigation of airworthiness has occurred.
Second, a study of teeter margin is truly unique in terms of the instrumentation design and
installation, the flight trial itself and the data analysis. Coming as it did at the end of the work
is testament to growth in the confidence with which gyroplane problems can be tackled, and
indicates that adequate teeter margin exists in all but the most extreme dynamic response
situations. Finally an assessment of blade flexibility on rotor and aircraft stability, and a review
of BCAR Section T completes the Report.
August 2010 Executive Summary Page 1

Glossary
1 Abbreviations
AAN Airworthiness Approval Note

AIAA American Institute of Astronautics and Aeronautics
BCAR British Civil Airworthiness Requirements
CAA Civil Aviation Authority
DOF Degrees of Freedom
FAA Federal Aviation Administration
ISA International Standard Atmosphere
NACA National Advisory Council on Aeronautics
NASA National Aeronautics and Space Administration
NASA STAR NASA Scientific and Technical Aerospace Reports
RAE R and M Royal Aeronautical Establishment, Research and Memoranda
RASCAL Rotorcraft Aeromechanics Simulation for Control Analysis
VZLU Czech National Aeronautical Research and Test Institute
August 2010 Glossary Page 1

2 Nomenclature
A System matrix
ax, ay, az Absolute accelerations measured by accelerometers m/s2
B Control matrix
c Effective damping
ccrit Critical damping (i.e. = 1)
CL Aircraft rolling moment coefficient from wind tunnel data
CM Aircraft pitching moment coefficient from wind tunnel data
CN Aircraft yawing moment coefficient from wind tunnel data
CX Aircraft thrust coefficient from wind tunnel data
CY Aircraft side force coefficient from wind tunnel data
CZ Aircraft normal force coefficient from wind tunnel data
CD Drag coefficient
CL Lift coefficient
CLmax Maximum lift coefficient
Cm Pitching moment coefficient
Cm0, Cma Coefficients of pitching moment expression
Ct Propeller thrust coefficient
D Rotor diameter m
d Propeller diameter m
g Acceleration due to gravity m/s2
h Altitude m
I Identity matrix
Ixx Aircraft moment of inertia about x-axis kg m2
Iyy Aircraft moment of inertia about y-axis kg m2
Izz Aircraft moment of inertia about z-axis kg m2
Ixz Aircraft product of inertia kg m2
J Advance ratio
k Scale correction factor
l Distance from nose wheel to main undercarriage m
L, M, N External moments (roll, pitch, yaw) Nm
m Mass of aircraft kg
Ng, Nr Measured propeller and rotor speeds rev/min
P Roll rate (+ve roll to starboard) rad/s
p Perturbation in roll rate from reference trim state rad/s
Q Pitch rate (+ve nose up) rad/s

q Perturbation in pitch rate from reference trim state rad/s

Q Main rotor torque Nm
R Yaw rate (+ve nose to starboard) rad/s
r Perturbation in yaw rate from reference trim state rad/s
R Rotor Radius m
Re Reynolds number
T Rotor thrust N
T Period of an oscillatory mode s
thalf Time to half amplitude of stable mode s
tdouble Time to double amplitude of unstable mode s
U Fore and aft component of velocity (+ve forward) m/s
u Perturbation in fore and aft component of velocity m/s
u Control vector

uprobe, vprobe, Component air velocities measured by air data probe m/s
wprobe
Vf Indicated airspeed m/s
Vw Wind tunnel flow velocity m/s
V Lateral (sideslip) component of velocity (+ve to starboard) m/s
v Perturbation in lateral (sideslip) component of velocity m/s
v0, v1s, v1c Component of rotor inflow m/s
W Vertical component of velocity (+ve downwards) m/s
w Perturbation in vertical component of velocity m/s
W Weight of aircraft (c.g. calculation) N
Wnose Weight measured by load cell under nose wheel (c.g. calc) N
WL, WR Weights measured by load cells under left and right
wheels (c.g. calculation) N
X, Y, Z External forces N
x State vector

xcg, zcg Position of aircraft centre of gravity w.r.t. reference point m
xprobe, yprobe Positon of tip of air data probe with respect to fuselage m
xE, yE, zE Aircraft position as measured by GPS m
zprobe datum point m
Angle of attack ( +ve nose up ) rad

vane Angle of attack measured by air data probe vane rad
Sideslip angle ( +ve to port ) rad
1c, 1s Rotor disc orientation (long, lat) in multi-blade co-ords rad
vane Angle of sideslip measured by air data probe vane rad

t Teeter angle rad

bias or offset correction
r Rudder deflection ( +ve to port ) rad
e, a, t Control inputs for fixed wing aircraft (elevator, aileron,
throttle) rad
Error function
s, c Longitudinal and lateral stick displacement rad
p Pedal displacement
Throttle position
Bank angle (+ve roll to starboard) rad
Perturbation in bank angle from reference trim state rad
eigenvalue
Pitch angle (+ve nose up) rad
Perturbation in pitch angle from reference trim state rad
Keel inclination angle in c.g. calculation rad
Absolute viscosity of air kg/ms
Aircraft heading (+ve to starboard, measured from North) rad
Blade azimuth position rad
Density of air kg/m3
n Natural frequency rad/s
d Damped natural frequency rad/s
Rotorspeed rad/s
ss Rotorspeed in steady state (trimmed) flight rad/s
Damping ratio
Subscripts
b Body fixed axes set
cg Centre of gravity
e Equilibrium flight state (reference trim state)
E Earth fixed axes set
ph Phugoid Mode
probe Located at the pressure transducer in air data boom
sp Short period mode
vane Located at the vane on air data boom

Stability Derivatives
Stability (or aerodynamic) derivatives are expressed in the form:Ab, where:
A
Ab =
b e
that is the change in an external force or moment A with respect to a small

perturbation in a state or control b, evaluated at a reference trim state (denoted by the
subscript e). For example:
= perturbation in X - force due to perturbation in forward velocity, u
X
Xu =
u (known as speed damping)
M
Mq = = perturbation in pitching moment, M due to a perturbation in pitch
q rate q (known as pitch damping),
Q
Q = = perturbation in rotor torque, Q due to a perturbation in rotorspeed


1 Introduction
1.1 Background
The emergence of the gyroplane aircraft in the 1920's and '30's paved the way for the
development of the helicopter in the 1940's. Many of the technical problems
associated with rotary wing flight had been discovered and rectified by the early
gyroplane pioneers, most notably Juan de la Cierva's solution of installing flapping
hinges to accommodate non-symmetric lift from the rotor blades. The development
of the gyroplane receded as the helicopter became more popular and successful. In
recent years, however, there has been a resurgence of interest in this type of aircraft
both as a recreational aircraft and as a low cost alternative to the helicopter with
companies such as Groen and Cartercopter both seeking to market gyroplane
configurations to commercial and military operators.
In the UK the gyroplane is very much the domain of the private hobby flyer where the
aircraft is often purchased in kit form and built by the owner. Although the number of
gyroplanes registered in the UK is relatively small compared to, say, microlights or
light aircraft, gyroplane owners are often extremely enthusiastic and the UK gyroplane
community is a very active one. Despite (or in some part due to) the popular belief
that these aircraft are naturally stable, and that flying them was no more difficult than
piloting a light aircraft, a series of accidents between 1989 and 1991 occurred in the
UK. The statistics show 6 fatalities per 1000 hours of flying time during this period for
this type (compared to 0.015 per 1000 flying hours for general aviation). Given the
small number of aircraft of this type registered in the UK, this was clearly an
unacceptable situation.
Investigation of these accidents was hindered by a lack of contemporary published
research into this type of vehicle, particularly in its aerodynamic characteristics, its
flight dynamics and flying qualities. The Air Accidents Investigation Branch
recommended that the CAA undertake appropriate research in these areas to support
a major review of the British Civil Airworthiness Requirements for gyroplanes (BCAR
Section T). The aim was to improve the design standard of gyroplanes in the UK and
so improve their safety. A contract was awarded to the Department of Aerospace
Engineering, University of Glasgow in 1993. As the project developed, it became clear
that additional research was necessary, and over the period from 1993 to 2008 a
series of related research projects were commissioned by the CAA. The findings of
the various research projects have been fully reported in individual contract reports.
The purpose of this report is to set the various individual projects into context with
the main findings presented in an accessible manner, and to provide a single source
of reference for all of the research. This report has been structured to provide the
most logical route through the research, rather than a chronological list of project
descriptions.
1.2 Fundamental Research
Early efforts at investigating the gyroplane accidents mentioned above were
hampered by insufficient contemporary research material being available. In
particular, there were suggestions that some of the accidents might have been
caused by dynamic instability. Further, some of the accidents involved aircraft which
had been modified by the owner; for example, the fuselage pod removed to give an
open framed vehicle. One of the first tasks undertaken in 1993 was a literature
August 2010 Page 1

review. This was updated in 2007, and the results are presented in Section 2, with
the papers found listed in the references. This review showed an almost complete
lack of material on the flight dynamics of gyroplanes, and no detail whatsoever on the
aerodynamic properties. It was clear then that wind tunnel tests would be required to
establish, in the first instance, whether changes to the basic configurational design of
a gyroplane (tailplane, fin, pod etc.) might have any significant effect on its
aerodynamic characteristics. The results from an extensive series of wind tunnel
tests (see Appendix 1 for the test matrix) using a model of a typical gyroplane
configuration are presented in Section 3.
The approach used to study the dynamic stability characteristics of the gyroplane was
to modify an existing rotorcraft mathematical model to represent a gyroplane
configuration. This required creating a data set of parameters representing the
gyroplane to be simulated (mass, inertia, geometry etc.) see Appendix 2. An
important parameter is the location of the centre of gravity, and a practical method of
achieving this is detailed in Appendix 3. The aerodynamic coefficients measured in the
wind tunnel tests also form part of this data set. There were therefore two reasons
for performing the wind tunnel tests described in Section 3; to investigate whether
configurational changes can have any substantial influence on aerodynamic loading,
and to provide data for the simulation.
1.3 Simulation and Model Validation
The simulation used for this study, RASCAL, was developed in a modular and generic
form and so its conversion from primarily helicopter simulation to gyroplane
simulation was a relatively straightforward process, which is described in Section 4.
The resulting model is nonlinear and includes all body, rotor speed, teeter (flap) and
inflow degrees of freedom. Each blade on the rotor and propeller is represented as an
individual component with its own degrees of freedom. The model is used to
calculate trim states, response to controls and dynamic stability characteristics. It is
appreciated that many readers will be unfamiliar with the theory involved, and the
terminology used, and so Section 4 is supported by a tutorial in the theoretical
calculation of dynamic stability, given in Appendix 4. It is recommended that Appendix
4 be read before Section 4.
An essential element of any simulation development is its validation. This involves
comparing the results from the simulation with those measured from the real system.
In the case of the gyroplane, validation is achieved by comparing results from
RASCAL with data measured in flight. As described in Section 5, a total of 3 flight test
campaigns were carried out to collect sufficient data to successfully validate the
model. In the first trials, data from a VPM M16 aircraft was collected. Then, to provide
dissimilar data, a Montgomerie-Parsons aircraft was tested. In a third series of trials,
the focus was on the rotor teeter dynamics (see Section 11) and, again, the
Montgomerie-Parsons aircraft was used. The flight test instructions for this final trial
are given in Appendix 5.
The validation process itself is discussed in Section 6 of the report with comparisons
made between trim states, aircraft response to controls, and stability derivatives.
Significant post processing of flight data is necessary before validation can take place,
and the theory behind this is presented in Appendix 6 and Appendix 7.
1.4 Application of Research
Having discussed the development of the mathematical model and established its
validity, it is now possible to use the simulation to investigate specific issues relating
to the flight characteristics of gyroplanes. As mentioned above, during the period in
the early 1990s when the gyroplane accident rate was at its highest, there was a
August 2010 Page 2

certain level of evidence (often anecdotal and always unsubstantiated) that changes
to basic configuration had contributed to some of the accidents. Armed with the
results of the wind tunnel trials described in Section 3, it was possible to carry out
parametric studies varying many of the key design parameters which define the
gyroplane configuration. The results from the parametric studies are given in Section
7. This research has identified the vertical location of centre of gravity with respect to
the propeller thrust line as the parameter which is most influential in the dynamic
stability of gyroplanes. This result, and much of the other research carried out in this
study, has fed into the development of the airworthiness requirements for light
gyroplanes, BCAR Section T. A review of these requirements, and a description of the
contributions of this research to their formulation is given in Section 8. Two technical
papers supporting this work are included as Addenda to this report. Rotor blade
torsional effects are included in the mathematical model, and the influence this has
on the simulation results is discussed in Section 9. A third Addendum is provided
giving further insight into aeroelastic properties of gyroplane rotors. An investigation
of a specific gyroplane type, the Air Command, is presented in Section 10, whilst
unique results from tests to measure blade teeter angle in flight are given in Section
11. This report is concluded with a discussion of the main conclusions which have
emerged from the research in Section 12.
2 Literature Survey
It was clear from the very early stages of the research in 1993 that there was little
contemporary published research into gyroplane aerodynamics or flight dynamics.
The aim of the literature survey was to identify as much relevant information as
possible. Having obtained this information, the next task was to acquire copies of the
identified literature where possible. During the course of the research from 1993
2006, further relevant research has been published, perhaps stimulated by the
activities at Glasgow University, but certainly due to a resurgence in interest in the
use of the gyroplane in commercial and military roles. A second, internet-based
literature review was therefore undertaken in 2007, and these results are also
incorporated here.
2.1 Methodology
The initial survey was carried out in 1993 and was composed of three main sources
of information - a PC based CD ROM database available in Glasgow University Library,
a computer search of the main technical archives, and a search of the Department of
Aerospace archives. A more recent survey in July 2007 was conducted using internet
resources. Brief details of the methodology adopted are given below.
a) CD ROM Database
This database contains citations from all of the major science and engineering
journals (including all of the AIAA publications and the Journal of the American
Helicopter Society). Discs for the years 1987 - 1993 are available in Glasgow
University Library and these were searched using appropriate keywords
(gyroplane, gyrocopter, autogyro etc.) and various citations recorded.
b) Computer Search
A wider computer search was undertaken with the assistance of library staff. This
included accessing databases such as the NASA STAR archives and, as with the
CD ROM search, appropriate keywords were entered and a series of citations
recorded.
August 2010 Page 3

c) Departmental Archive
Until the mid 1970's the Department of Aerospace Engineering at Glasgow
University maintained a comprehensive collection of NACA/NASA and RandM/RAE
technical reports. The catalogues of these collections are still largely intact and
were used to obtain references to a series of gyroplane related documents.
d) Internet Search
The most popular search engines (Yahoo, Google Scholar) were used with
appropriate keywords to obtain lists of references. More specialised sites such as
AIAA and the NASA Technical Report Server were also used. This search returned
many of the papers previously found in the earlier reviews carried out in 1993,
however a number of additional papers were also discovered dating from before
1993 which had not been uncovered in the original search. Many papers published
after the original search were, of course, also found.
2.2 Results of Literature Review
2.2.1 General
The gyroplane helped to pave the way for the development of the helicopter,
introducing cyclic pitch control and blades attached to the rotor hub by means of a
hinge. Unfortunately the literature available before the current research was
undertaken did not specifically address vehicle stability and control. The literature on
gyroplanes is nonetheless considerable. However, in a contemporary context, much
of the early work is now primarily of historical significance. It provides the basis of the
understanding of gyroplane flight but does not address the issues of stability and
control. Examination of the literature shows a logical development of the study of
gyroplanes, from the elementary theory of gyroplane flight to an analysis of
aerodynamics and performance and ultimately rotor behaviour, but only for steady
flight. Interest then apparently waned and the next logical stage in the study of the
gyroplane, i.e. stability and control, was not examined.
2.2.2 Published work pre1940
The early work on the dynamics and aerodynamics of rotary wings was mainly in
support of the development of gyroplanes. Most notable amongst the early work is
that of Glauert [3, 5, 7] who developed methods for calculating rotor loadings and also
investigated the flapping motion of the blades. The rotor theory produced by Glauert
(which is still widely used today) was complemented by wind tunnel tests of
gyroplane rotors by Lock [4, 6, 7]. The work of Glauert and Lock in the 1920s was
advanced in the U.S.A. by Wheatly who, between 1932 and 1937, again concerned
himself primarily with the aerodynamics and dynamics of the rotor and also included
some wind tunnel testing [821]. The greater potential of the helicopter saw the
emphasis shift from unpowered to powered rotors from the late thirties onwards, and
NACAs interest in the gyroplane then seemed to wane, with only Bailey [22, 23]
continuing the work initiated by Wheatley. A couple of review papers written for
NACA also appeared during this era, Bregeut [24] and Schrenk [25]. The natural
progression from aerodynamics and dynamics through to stability and control seems
therefore not to have been made for the case of the gyroplane. Consequently there
appears to be no definitive study of the flight mechanics of a gyroplane. Likewise,
although there were many wind tunnel studies of gyroplane rotors there appear to
have been few of a gyroplane fuselage and empenage.
August 2010 Page 4

2.2.3 Published work 19411959

Towards the 1940s as the helicopter became a practical proposition, rotorcraft
research really focused on helicopter issues, and there is almost no mention in
published literature of the gyroplane aircraft again until the 1960s.
2.2.4 Published work 19601993
Perhaps due to increased use as a sport or recreational vehicle, technical papers
specifically dealing with gyroplane issues began to re-emerge during the 1960s. One
of the main exponents of the gyroplane was, of course, Wallis who published details
of his design in the Aeronautical Journal in 1963 [26]. Later in 1965, Schad [27]
published details of performance trials of Bensen and Avian gyroplanes, whilst
interest from academia appeared in the form of a Masters Thesis by Niemi [28]. Also
around this time Mouille [29] published a detailed analysis of the Turbogyre 330
gyroplane for Agard. Also in this era, the notion of using autorotation to decelerate
spacecraft on re-entry into the earths atmosphere was proposed, and papers by
Levin and Smith [30] and by Barzda [31] presented details of proposed systems.
Little appeared in the literature again until 1974 when Liss [32] published a theoretical
analysis of the deformation of a rotor blade in autorotative flight. The first indication
of handling problems associated with gyroplanes was reported by Przybylski et al [33]
at the 2nd European Rotorcraft Forum in 1976. They report on an investigation into
the lack of responsiveness of a light gyroplane to nose down inputs at certain
rotorspeeds. Johnson [34] used data gathered in flight tests using a gyroplane to
validate a wake model by comparing lateral flapping angles. The only other works of
interest are that of Nicholls et al [35] on the nature of gyroplane accidents which had
been occurring in Australia, and that of McKillip [36] who reports on the
instrumentation required to record data from a rotor of a towed gyroplane.
Due to the historical significance of the gyroplane there are citations in many
rotorcraft and aeronautical textbooks [37 - 40]. Mainly, these citations are concerned
with the evolution of rotorcraft rather than dealing specifically with the gyroplane.
2.2.5 Published work 1994present
During the course of this research the popularity of the gyroplane as a sport aircraft
has increased. It is its re-emergence as a potentially useful vehicle for commercial or
military operations, however, which has led to an upsurge in the volume of published
literature. Shippen [41, 42] published his findings from a flight test programme to
record control stick forces in flight. The aim was to determine the reason for a
previous in-flight failure. Interest in the gyroplane in Spain led to an analytical review
by Lopez-Diez et al [43] of aspects of the control of the Cierva C30 gyroplane, which
led to a more topical study of possible applications for the gyroplane by comparing its
performance to that of the helicopter in various roles [44]. Rapp et al [45] also reported
on an experimental measurement of blade loads in flight in 2000. Various other
authors [4651] have used contemporary techniques to analyse various aspects
associated with gyroplanes: rotor aerodynamics [46, 47], performance [48], and
aircraft design [4951].
As in other areas of aviation, the idea of an unmanned gyroplane has received some
attention in the past few years [5255]. Further, and as mentioned above, there is a
re-emergence of interest in the gyroplane concept for commercial and military use,
most notably from the Groen Brothers and Carter Aviation Technologies in the USA.
Both companies have produced a small number of publications of a technical nature
with most of their output naturally being of a promotional nature [5658]. There have
also been a couple of papers reflecting on historical aspects of the gyroplane [5960],
most notably the work by Leishman [60] which is rich in technical content as well as
historical background.
August 2010 Page 5

2.2.6 Publications Arising from this Study

A number of papers reporting work arising directly and indirectly from this study have
been published. Their content will be discussed in detail throughout this report
however, for convenience, they are listed together in the references [6178]
2.3 Conclusions
In the early days of this study there was very little useful information concerning the
dynamic stability of gyroplanes available in published works. Similarly, there was no
literature available pertaining to the aerodynamics of the gyroplane as a complete
vehicle. Through the course of this study the available literature has grown in no small
part through the efforts of the researchers, but also due to resurgence of interest in
the gyroplane configuration. Although the more recent material is of interest, none of
it has been of direct relevance to the current study which remains unique in its
breadth and depth of investigation into gyroplane aerodynamics and flight dynamics.
The policy adopted throughout by the researchers of publishing their findings in
learned, refereed journals has been a very important and significant outcome of the
work as it has greatly enhanced the quantity and quality of openly available technical
information on the gyroplane, particularly in relation to the location of the propeller
thrust line relative to the centre of gravity.
3 Wind Tunnel Testing
3.1 Introduction
A scale model of a VPM M14 gyroplane was tested in the 3m low speed wind tunnel
at the Aeronautical Research and Test Institute (VZLU) in Prague in the Czech
Republic. The results from this test series were analysed and then utilised in the
mathematical model used throughout this study.
The wind tunnel test programme described below had two distinct aims. The first
was to provide basic aerodynamic data on the effects of gyroplane configurational
characteristics with a view to assessing the degree to which specific design features,
such as cowlings and tailplanes, are beneficial to gyroplane performance. Secondly,
it was intended to provide aerodynamic data for input to mathematical models to
facilitate parametric studies of static and dynamic stability.
3.2 Wind Tunnel Facility
The tests were conducted in the VZLU 3m Low Speed Wind Tunnel. This institute has
provided high quality wind tunnel facilities for the Czech aircraft industry for many
years and is highly regarded throughout the world. The particular wind tunnel used in
this study is an atmospheric open-section, closed return, Gottigen style tunnel with a
maximum velocity of around 60m/s. Forces and moments were measured on a six
component fully-automatic overhead gravitational balance which is accurate to
between 0.01% and 0.05% full scale.
3.3 Wind Tunnel Model
The model used in this study was a powered, one-third scale model of a VPM-M14
gyroplane minus rotor. It is normal, in rotorcraft testing, to carry out wind tunnel tests
without the rotor since scaling considerations of a combined rotor-fuselage
configuration would require the use of a very large test facility and would be
prohibitively expensive. A VPM M16 model ideally would have been used, however
there were no sufficiently accurate engineering drawings of this vehicle available.
Accurate drawings of the M14 were available and the decision to build this model was
taken on the grounds that the basic configuration of the VPM M14 was typical of light
August 2010 Page 6

gyroplanes of this type, and the two vehicles used in this study (the VPM M16 and
Montgomerie-Parsons) were so similar that there would be no major differences in
their aerodynamic properties from those of the VPM M14.
The basic model frame was constructed from metal box-section on which was
mounted a water-cooled motor. This motor was connected by a toothed-belt to the
propeller drive system. The removable tailplane assembly was fabricated in
aluminium as were the representations of the aircraft wheels and wheel-covers. A
scaled representation of the VPM-M14 cowling was made from glass fibre. In addition
to the basic features of the aircraft, it was necessary to model the aerodynamic effect
of the pilot. This was achieved by creating a representation of the pilot's upper-body
which was mounted in an appropriate position on the model frame. The pilot's lower
body was adequately represented by the motor which was positioned just above the
main spar. A diagram of the model is shown in Figure 3.1, whilst Figure 3.2 shows a
photograph of the complete model, and Figure 3.3 shows the installation of the
motor.
3.4 Test Set-up
The model was mounted in inverted mode on the wind tunnel balance via a series of
connecting wires as shown in Figure 3.4. Services to and from the motor were
provided by a shrouded conduit which was located centrally above the model. This
conduit not only provided a power supply for the model but also housed the water
pipes necessary for the motor cooling system. The angle of attack and sideslip
settings were automatically adjusted using the balance control system. Rudder
deflection was achieved manually.
The model was tested in both power on and power off modes. In the latter case, the
propeller was removed from the model during testing as it would contribute
unrealistically to the measured drag of the model. For powered tests, the model
propeller operating conditions were appropriately matched to the cruise performance
of the full scale VPM-M14 propulsion unit. In particular, both the thrust coefficient and
advance ratio of the two propeller systems were matched by adjustment of the blade
pitch on the model propeller. The settings used were:
Advance Ratio
Vw
J= = 0.431
D
Propeller Thrust Coefficient

T
Ct = = 0.317
Vw2 d 2
All tests, including power off tests, were conducted at a test Reynolds number of:
Vw R
Re = = 2.5 10 6

which corresponds to a wind tunnel flow velocity of 31 m/s. The test Reynolds
number is approximately 40% of that of the full scale vehicle during cruise. This
discrepancy in the Reynolds number was a consequence of the model size and was
necessary to achieve propeller advance ratio matching between the model and the
full-scale vehicle. It is unlikely that the reduced Reynolds number of the tests would
August 2010 Page 7

produce any significant differences between the measured force and moment
coefficients and those experienced by the full-scale aircraft. This is primarily because
the basic gyroplane structure is non-streamlined and, consequently, insensitive to
Reynolds number changes. In addition, the surfaces which are streamlined, such as
the cowling and tail surfaces, all operate at test Reynolds numbers well in excess of
the generally accepted threshold (Re = 250 000) for low Reynolds number flows.
Hence, it unlikely that their performance would be adversely affected by the test
conditions.
All force and moment coefficients presented in the following sections were
measured with respect to the fuselage reference point and coordinate system shown
in Figure 3.5. The non-dimensional coefficients were calculated using the rotor radius
as the basic characteristic length. Thus, for example, the normal force and pitching
moment are given by:
1
Z= V 2 R 2CZ
2 w
and
1
M = V 2R 3CM
2 w
respectively.
3.5 Test Matrix
In total, one hundred and fourteen data polars were measured during the wind tunnel
test programme. These tests considered the performance of four different gyroplane
configurations over a range of incidence, sideslip and rudder angles for both power
on and power off settings. These four basic configurations were as follows:
Configuration 1: Cowling on, full tail on
Configuration 2: No Cowling, full tail on
Configuration 3: Cowling on, vertical fin on, no horizontal tail or end-plates
Configuration 4: No Cowling, vertical fin on, no horizontal tail or end-plates
In addition to these four basic configurations, individual tests were conducted on two
other configurations. These were
Configuration 5: Cowling on, full tail on, extended tail boom
Configuration 6: Cowling on, vertical fin on, horizontal tail on, no end-plates
The full test series, detailing sideslip and rudder angles, is outlined in Appendix 1.
3.6 Wind Tunnel Test Results
In the following sections, the main features of the results from the wind tunnel test
programme are presented and analysed. For clarity, only data which highlight the
salient features of the full dataset are presented in this report. The results are given
in graphical form and it should be noted that results labelled "tail off" relate to the
removal of the horizontal tail surfaces and associated end plates only.
August 2010 Page 8

3.6.1 Basic Configurational Aerodynamics

Of particular interest in this study is the extent to which the physical configuration of
a gyroplane can influence its aerodynamic performance. One of the primary
objectives of the wind tunnel test programme was, therefore, to obtain a quantitative
assessment of the effects which the addition of a cowling and/or tailplane would have
on the vehicle aerodynamics. This was achieved by measurement of the performance
of the four different gyroplane configurations described previously.
a) Zero sideslip
In Figure 3.6, the variations of the three force coefficients with incidence are
presented for the case of zero sideslip and power on. What is immediately
apparent from this figure is that, as may be expected, both CX and CY are relatively
insensitive to gyroplane configuration. It is noticeable, however, that there is a
greater difference between the four configurations at high negative angles of
attack.
In the case of CX, the most interesting effect is the higher drag at negative
incidence due to the horizontal tail surfaces in both power on and power off (Figure
3.7) modes. The insensitivity of the effect to power setting excludes the possibility
of increased tailplane performance due to the propeller wash and points more
towards an induced incidence effect resulting from forebody shielding. Forebody
shielding occurs when, at negative incidence, the tail moves into the wind shadow
of the gyroplane forebody. This does not happen at positive incidence where, as
the angle of attack is increased, the tail moves progressively further into clear air
and ultimately stalls. The effect of tail stall is particularly clear in the power off case
where, above 20 degrees incidence, the values of CX produced with no tail exceed
those of the tail-on configuration. In power off mode, the significant increase in
thrust produced by the cowling across the entire incidence range is due to an
additional thrust component from the cowling itself. This behaviour is consistent
with that which would be expected of any streamlined body but is masked in the
power-on case by the interference effect of the cowling on the propeller.
In the power off mode, the value of CY should ideally be zero over the entire
incidence range when the gyroplane is at zero sideslip. As may be observed in
Figure 3.7, the measured values of CY are very small for all configurations but not
exactly zero. This may be a result of a slight misalignment of the model on the
balance wires or a small flow angularity in the test section. It is also apparent that
the addition of the cowling at high negative incidence causes a marked increase in
the side force. One likely reason for this could be the effect of the flow which is
channelled into the open cowling at high negative incidence. Under these
conditions, the cowling acts like a scoop and, as may be observed in the CZ graph,
produces considerably more downforce. It is likely that this 'scoop' effect would
also contribute significantly to the sideforce if the cowling were not exactly aligned
with the fuselage.
In the power on case, CY also exhibits the cowling effect described above but the
general trend of the data comes from two specific effects. The first is due to the
basic propeller effect which is a combination of the sideforce produced at the
fuselage reference point by the propeller torque and the sideforce due to the
asymmetric loading on the propeller. On a fixed pitch propeller, a finite gross inflow
angle produces a blade angle of attack variation around the propeller disc and so
leads to a force imbalance. The second effect is the interaction of the vertical tail
surface with the swirl from the propeller. As is clearly visible in the figure, this
effect becomes more significant as the tail moves into the propwash at negative
angles of attack.
August 2010 Page 9

The variation in the normal force coefficient, CZ, with gyroplane configuration is
much more marked than the other two force coefficients. In both power on and
power off cases the effects of the cowling and tailplane are quite clear. The
increase in wetted area of the cowling increases the effective drag, in relation to
the onset flow, which is produced by the gyroplane forebody. Thus, at positive
angles of attack the upforce is increased and the opposite occurs at negative
angles. In addition, the 'scoop' effect of the cowling, at negative incidence,
increases the downforce further. The effect of the tail is, if anything, more
significant with the general effect being in the same sense as the cowling. At
positive incidence, there is clear evidence of tailplane stall at around 20 degrees.
The fact that stall is not obvious at negative incidence may be a consequence of
the forebody shielding effect discussed earlier.
Much of the information contained in the force coefficient plots translates directly
to the moment coefficient plots presented in Figures 3.8 and 3.9. In the power off
case, both CL and CN are almost zero over the entire angle of attack range. The
small moments which have been measured are probably due to slight model
misalignment or flow angularity as discussed earlier. In power on mode, both of
these coefficients are influenced by the propeller. The rolling moment, CL, is
almost constant and is produced by the reaction torque from the propeller and CN
varies according to the effect of the propwash on the vertical tail and the
asymmetric loading on the propeller disc as discussed in relation to CY above.
The pitching moment, CM, is the most sensitive of the three moments to the
gyroplane configuration. This is a direct consequence of the effects described in
relation to the normal force, CZ, above. In particular, it may be observed that the
general effect of the tailplane is to longitudinally stabilise the aircraft, whilst the
cowling has a destabilising effect. In this respect, the positive gradient of the CM
curve for the cowling on, tail off case indicates that this configuration is unstable.
It is also apparent that the onset of tail stall has significant influence on the stability
of the aircraft with the aircraft becoming marginally unstable above 20 degrees
incidence. The basic airframe plus pilot but without cowling or tail surfaces has
almost neutral stability.
b) -15 Degrees Sideslip
As may be observed from the tables in Appendix 1, power-off tests were only
conducted at negative sideslip angles. This is because, in all power-off
configurations, the gyroplane is symmetric about the XZ plane and so positive
sideslip angle results can be inferred from those measured at negative angles. In
the power-on mode, the influence of propeller rotation and swirl is significant and,
consequently, both positive and negative sideslip angle measurements were
made. In this section, the results for -15 degrees sideslip in power on mode are
compared to those measured with power off. The results not only highlight the
basic configurational effects, but also serve to illustrate the influence of the
propeller.
Figures 3.10 and 3.11 present the force coefficients measured in power on and
power off mode at -15 degrees sideslip. Examination of the results presented in
both of these figures indicates that the tailplane has almost no effect on the
developed thrust (CX) except at high negative incidence with the power on. This is
quite unlike the zero sideslip case where the shielding effect of the gyroplane
forebody resulted in increased thrust from the tail at negative incidence in both
power on and power off modes. It is therefore clear that, as may be expected, the
forebody shielding effect is reduced with increasing sideslip angle. Consequently,
the increased thrust from the tailplane in power on mode is due entirely to the
influence of the propeller wake on the tail as the aircraft incidence becomes more
negative.
August 2010 Page 10

The effect of the cowling on CX is unusual in that the influence which the cowling
has is different in the power on and power off modes. In Figure 3.11, it may be
observed that the effect of the cowling removal in power off mode is to reduce the
thrust. As discussed in the zero sideslip case, this is probably due to an additional
component of thrust developed by the streamlined shape of the cowling itself
when the aircraft is in yaw. In power on mode, however, there is little effect from
the cowling at positive incidence but a marked increase in thrust when the cowling
is removed at negative incidence. There was, in fact, some evidence of this at zero
sideslip, but in this case the effect is more pronounced. A probable cause of this
could be that the increased effective area of the cowling is reducing the mass flow
to the propeller and, so, reducing thrust.
It is immediately apparent from Figures 3.10 and 3.11 that the general variations in
sideforce are little affected by the power setting. It is also clear that a general
increase in sideforce is achieved in the tail-on case. Bearing in mind that "tail off"
refers only to the removal of the horizontal tail and end-plates, it is appears likely
that the inclusion of end-plates in this particular design was necessitated by lateral
stability considerations rather than any desire to enhance tail lift.
The cowling appears to have little influence on the sideforce at negative incidence
but is responsible for a substantial increase at positive angles of attack. This effect
on the sideforce would appear to be produced by the effective difference in shape
which the cowling presents to the onset flow in the two cases.
The normal force variations produced at -15 degrees sideslip are very similar in
trend to those of head-on flow. As before, the effect of the tail is to enhance the
absolute value of the normal force produced by the aircraft. In this case, however,
tail stall is not as well defined due to the strongly three-dimensional flow. The
effect of the cowling is, again, to increase the absolute normal force experienced
by the aircraft. This effect is more pronounced at negative angles of attack where,
as discussed previously, the open cowling acts like an air scoop.
Figures 3.12 and 3.13 present the moment coefficients measured at -15 degrees
sideslip for both power on and power off cases. Both figures exhibit the same
general trends for the rolling moment curves. In both cases, only small variations
are apparent across the incidence range with the most significant effect being the
change in the magnitude of the rolling moment between power on and power off
modes due to propeller reaction torque.
As in the head-on case, the pitching moment, CM, is very sensitive to gyroplane
configuration. Again, the tailplane and cowling have opposing effects with the
cowling acting to destabilise the aircraft and the tailplane providing an element of
stability. As before, the cowling on, tail off configuration is unstable with the
'scoop' effect of the cowling at negative incidence being particularly influential.
Apart from the change in magnitude due to the propwash on the vertical tail and
the propeller disc loading, the yawing moment curves for the power on and power
off modes are remarkably similar in form. In particular, the effects of the cowling
and tailplane are clearly in opposition in both cases. Given the position of these two
elements with respect to the fuselage reference point, and also in the light of the
preceding discussion on sideforce, this is not surprising. It is, however, important
to note that reduction in CN due to the removal of the tailplane and the addition of
the cowling may have an adverse effect on lateral stability.
August 2010 Page 11

3.6.2 Synthesis of Configuration Four Results

In flight mechanics modelling, it is common to supply input aerodynamic data in the
form of basic fuselage coefficients plus increments for any optional configurational
components. Thus, in the case of the gyroplane, the basic configuration could be
taken as cowling off, tail on and increments to each coefficient could be applied if, for
example, a cowling was added or the tail removed. This approach, however, relies
heavily on minimal aerodynamic interference between the constituent parts of each
aircraft configuration. In wind tunnel testing it is also possible to use this technique to
synthesize data if the subject of the test programme is suitable.
In the particular case of the gyroplane wind tunnel test programme, staff from
Glasgow University and VZLU were confident that the technique could be applied
successfully. For this reason, a full series of tests were conducted on three
configurations to establish cowling and tailplane increments and only a limited series
of tests were carried out for the tail off, cowling off case. This section compares the
measured tail off, cowling off results with those synthesized from the results
obtained for the other three configurations.
Figure 3.14 represents the most difficult case for the synthesis method; i.e. that of
zero sideslip where the cowling effectively shields the tail from the onset flow. In the
figure, the synthesized normal force, thrust and pitching moment coefficients are
compared to those measured. It is immediately apparent that, at positive angles of
attack, the level of agreement between the two datasets is exceptional. The same is
not true, however, at negative angles of attack when the tail moves into the shadow
of the cowling. Here, both CX and CZ show discrepancies whereas CM is well
modelled.
In Figure 3.15, a similar comparison is presented for the 15 degree sideslip case.
Again, the agreement at positive angles of attack is substantial. In this case, however,
the level of forebody shielding at negative incidence is much reduced and,
consequently, the agreement between the two datasets is considerably improved. A
similar quality of agreement is achieved for the remaining coefficients which are not
presented here.
Despite the anomalies at high negative incidence, the synthesis technique has been
shown to be suitable for the calculation of the tail off, cowling off coefficients within
the normal operating range of the gyroplane. The performance of the tail off, cowling
off configuration can, thus, be inferred from the discussion of configurational
aerodynamics given above.
3.6.3 Effect of Power Setting
In this section the results measured for the cowling on, tail off configuration are used
to illustrate the basic effects which the propeller has on the measured coefficients.
Figure 3.16 presents the measured force coefficients for the power on and power off
settings. The effect of the operating propeller on the thrust coefficient, CX, is to
simply shift the curve in the positive direction. Otherwise, the shapes of the two
curves are almost identical and there is little evidence of any change in the propeller
thrust with incidence.
The sideforce coefficient (CY) curves are again offset from each other but, this time,
the curves are not parallel. This is due to the previously discussed effect of the
propwash on the vertical tail and the forces generated as a result of the torque and
sideforce on the propeller itself.
August 2010 Page 12

Unlike the two previous coefficient variations, the normal force coefficient (CZ) curves
for the power on and power off cases do not exhibit large differences. It could be
expected that the power on case should produce greater downforce as a result of the
two degree offset in the thrust axis. This, however, is not borne out by the measured
data and, in fact, the power on case appears to produce less downforce over the
measured incidence range than the power off case. At positive incidence, this is a
consequence of the inflow angle to the propeller which determines the direction in
which thrust is developed on the propeller disc. At negative angles, however, the
opposite behaviour should be observed, but is not. Again, this is most likely a result
of the blocking effect of the aircraft forebody at negative angles of attack.
The three moment coefficient comparisons of the power on and power off cases are
presented in Figure 3.17. As may be expected, the rolling moment coefficient (CL)
curves are merely displaced from each other by a nearly constant amount when
power is applied. This effect is purely due to the reaction torque produced by the
propeller.
The pitching moment coefficient (CM) curves are offset from each other as a result
of the displacement of the thrust axis from the fuselage reference point. A divergence
is, however, apparent between the two curves as the gyroplane incidence is
increased. This behaviour is entirely consistent with that of the normal force
coefficient (CZ) described above. It has been established that the propeller produces
an upforce which increases as the angle of attack increases. Since the propeller is
located behind the fuselage reference point, a nose down pitching moment will be
produced. The increasing strength of this moment as the gyroplane incidence
becomes higher is responsible for the different gradients of the power on and power
off curves. It should be noted that the general effect of the propeller is a longitudinally
stabilising one.
The yawing moment (CN) effect of the propeller is very evident from the comparison
of the power on and power off cases. Again, it is clear that the influence of the
propwash on the vertical tail is substantial but reduces as the tail moves clear of the
forebody at positive angles of attack.
Figures 3.18 and 3.19 show the force and moment coefficient curves, in power on
and power off modes, for the same gyroplane configuration at -15 degrees sideslip.
Generally speaking, the observed trends are much the same as the zero degree
sideslip case although some differences are apparent. In particular, the sideforce
coefficient (CY) curves are almost parallel in the yawed case. In addition, the
behaviour of the power on and power off curves of CZ are more in line with that which
was originally anticipated. This would indicate that, as expected, the blocking effect
of the aircraft forebody is reduced as the sideslip angle increases. One consequence
of this result is a slightly more beneficial effect of the propeller on the pitching
moment characteristics.
3.6.4 Rudder Effectiveness
The influence of the rudder is illustrated for the cowling on, tail off configuration in
Figures 3.20 and 3.21. As may be observed in Figure 3.20, the effect of rudder
deflection on CX is to marginally reduce the overall thrust. At positive angles of attack,
this effect seems almost linear. At negative angles of attack, where the forebody
shields the tail and the propwash becomes more significant, the effect is most
noticeable between 0 and 10 degrees of rudder deflection.
The effects which the rudder have on the side force and yawing moments are, of
course, the most significant feature of these graphs. It is clear that the rudder effect
on side force is almost linear at a given angle of attack. It is also apparent that there
is a slight reduction in the effectiveness of the rudder as the absolute angle of attack
of the gyroplane increases and the tail is subject to strongly three-dimensional flow.
August 2010 Page 13

In light of the above, it is not surprising that the yawing moment also exhibits linear
behaviour with rudder deflection. In addition, the offset of the rudder centre of
pressure from the fuselage reference point results in a rolling moment when the
rudder is deflected. Again, the characteristics of this particular behaviour are benign.
The response of the aerodynamic characteristics of the gyroplane to rudder deflection
described above are indicative of those observed on all gyroplane configurations over
the range of sideslip measured in this study.
3.6.5 The Effect of Sideslip
Many of the effects of sideslip have already been addressed in previous sections.
Here, however, the effects which configurational changes have on lateral stability in
yaw will be examined.
In Figures 3.22 to 3.25, the sideforce and yawing moment coefficient variations for
the four basic gyroplane configurations, in power off mode, at three different sideslip
angles are presented. In Figure 3.25, the data for configuration four is synthesized
from the other three datasets. Examination of these four figures indicates that, in all
cases, the variation of sideforce (CY) with yaw angle shows no unusual
characteristics. The yawing moment coefficient (CN) variation with sideslip angle is,
however, particularly sensitive to gyroplane configuration. From a stability stand-
point, it is important that there exists a gradient in yawing moment with respect to
sideslip angle. In these figures the magnitude of this gradient can be inferred from the
spacing between each curve. It is clear that those configurations in which the cowling
is present appear to lack stability between 0 and -7.5 degrees of sideslip at positive
angles of attack. Indeed, the cowling on, tail off configuration is affected over a wider
range of sideslip angles. It is also particularly noticeable that the end plates on the
horizontal tail do have a beneficial effect on the lateral stability characteristics of the
aircraft.
In isolation, the above features would be particularly detrimental to the aircraft
handling characteristics. In practice, however, the stability of the aircraft is aided by
the relatively large propeller. This is illustrated in Figure 3.26. which presents the
corresponding power on measurements on the cowling on, tail on configuration. It is
clear from this figure that the effect of the propeller is to stabilise the configuration by
increasing the absolute magnitude of the yawing moment as the sideslip angle is
increased.
3.6.6 The Horizontal Tail
The configuration of the tail sections of the VPM M14 and M16 aircraft are quite
unusual in terms of the large endplates on the horizontal tail surfaces. It has already
been established that these endplates are beneficial for lateral stability but it was
unclear what effect, if any, they would have on the lifting effectiveness of the
tailplane. To investigate this, a single test was carried out with the cowling on, tail on
configuration but without the endplates. The measured normal force and pitching
moment coefficient curves are compared with those from the corresponding test
with the endplates on in Figure 3.27.
By preventing the flow of air around the tips of the horizontal tail, the end plates
restrict the flow over the horizontal tail to an almost two-dimensional state. This
should allow the tail to develop higher lift than a conventional tail at moderate
incidence but will result in a sharper stall. This is illustrated in Figure 3.27 where the
loss of normal force due to tail stall is apparent at around twenty degrees when the
endplates are on but there is a much more gradual loss when the endplates are
removed. There is no clear indication in the CZ curve of any increase in lift at moderate
incidence but the gradient of the CM curve is obviously greater between -20 and +20
degrees when the endplates are on. This is indicative of increased tailplane
effectiveness.
August 2010 Page 14

The effect of increasing the length of the tail boom by approximately 0.3m (on the full
scale vehicle) was also examined. The results of this study are presented in Figure
3.28 and were very much as expected with almost no change in the normal force
coefficient curve, but a substantial change in the gradient of the pitching moment
coefficient curve. This is simply due to the increased offset of the tail from the
fuselage reference point. It also appears, however, that the stall angle of the tail at
positive incidence is reduced with movement away from the gyroplane forebody. This
is undoubtedly a result of a change in the effective angle of attack experienced by the
tail due to reduced interference between the tail and the forebody.
Perhaps the most significant observation associated with the tailplane is the lift-curve
slope. This is a measure of efficiency and also an indicator of its likely impact on the
complete aircraft and it can be extracted from these data readily. For example Figure
3.7 compares Z-force coefficient with tail on and off; analysis shows that the lift-curve
slope is 2. A typical value for a blade is 5.7; for a helicopter horizontal tailplane 3.7; and
the theoretical maximum for a finite wing is 2. When combined with the short
moment arm of the typical pusher configuration it can be seen that tailplane
effectiveness is likely to be limited at anything other than high speed. Note that with
power on, where even the low-set tailplane might be expected to experience some
slipstream benefit, the lift-curve slope increases only by a relatively small amount to
a value of 2.7 - still well below even helicopter values.
3.7 Configurational Differences Between the VPM M14 and M16 Variants
The mathematical model used to generate all of the simulation results presented in
this report incorporates the aerodynamics data gathered for the VPM-M14
configuration. It is worthwhile noting that the aircraft flight tested was in fact a VPM-
M16 which differs slightly from the M14 variant (refer to Figures 3.1 and A2.1). The
use of M14 aerodynamic data in a simulating the M16 vehicle requires some
justification. However, there are only three significant differences between the two
variants:
a) the M16 is 738mm longer than the M14 - due to a longer cowling and tail boom;
b) the horizontal tailplane of the M16 is 20% larger than the M14 (the vertical tailplane
areas are almost identical);
c) the cowlings are different in shape, although the wetted areas are almost identical.
Given the geometric similarity of the two variants it is unlikely that their general
aerodynamic characteristics (coefficients) will differ substantially. It is therefore
reasonable to use the data for the M14 to simulate the M16 with suitable adjustments
made to account, for example, for the larger horizontal tail surface. Further, the data
collected for the VPM M14 should give a fair representation of other similar aircraft,
given the very minor differences in overall configuration between types: most light
gyroplanes consist of keel, pod, mast, horizontal and vertical tail surfaces and a
pusher propeller. Finally, the relatively low operating speed of these aircraft implies
that any minor differences in aerodynamics characteristics would manifest
themselves in very small differences in aerodynamics loads.
3.8 Concluding Remarks
The aerodynamic characteristics of the gyroplane configurations considered in this
study are generally benign. It is, however, pertinent to note that there are several
effects associated with the cowling that are detrimental to stability. Although the
cowling on the VPM-M14 is particularly large, it is likely that any 'open' cowling design
will be subject to similar effects in the longitudinal mode. Additionally, the length of
the VPM cowling is substantial; extending from well in front of the pilot up to the rotor
support column. The increased wetted area which this presents to the onset flow in
sideslip acts to oppose the stabilising effect of the tail. The tail itself benefits from the
additional sideforce produced by the endplates on the horizontal surfaces.
August 2010 Page 15

An important assumption also made here is that the VPM M14 and M16 variants are
sufficiently similar that their aerodynamic properties will be insignificantly different.
The implication is that it is reasonable to use M14 data in a flight dynamics model of
the M16 and other similar aircraft (such as the Montgomerie-Parsons) without greatly
affecting (if at all) the overall assessment of its dynamic stability characteristics.
It is worth noting that the modification proposed to solve the stability problems which
led to the grounding of the Air Command gyroplane type comprised principally, in
aerodynamic terms, of the addition of a pod and a tailplane. These tests suggest that
this modification would not be very effective and, indeed, this proved to be the case.
CAA Comment
The remarks concerning the addition of a tailplane to the Air Command are noted,
however CAA Flight Test Specialist qualitative evaluation of the effects of a
horizontal tailplane on the RAF 2000 indicated a degree of improvement in the
phugoid characteristics at higher speed.
Figure 3.1 VPM M14 One Third Scale Wind Tunnel Model (dimensions in mm)
August 2010 Page 16

Figure 3.2 One Third Scale Wind Tunnel Model of a VPM M14 Gyroplane
Figure 3.3 The VPM M14 Model Without Cowling
August 2010 Page 17

Figure 3.4 a) VPM M14 Model in VZLU Wind Tunnel (Front View)
Figure 3.4 b) VPM M14 Model in VZLU Wind Tunnel (Rear View)
August 2010 Page 18

Figure 3.5 Co-ordinate System for Aerodynamic Coefficients
August 2010 Page 19

Figure 3.6 Force Coefficient Variations ( = 0, Power On)
August 2010 Page 20

Figure 3.7 Force Coefficient Variations ( = 0, Power Off)
August 2010 Page 21

Figure 3.8 Moment Coefficient Variations ( = 0, Power On)
August 2010 Page 22

Figure 3.9 Moment Coefficient Variations ( = 0, Power Off)
August 2010 Page 23

Figure 3.10 Force Coefficient Variations ( = -15 deg, Power On)
August 2010 Page 24

Figure 3.11 Force Coefficient Variations ( = -15 deg, Power Off)
August 2010 Page 25

Figure 3.12 Moment Coefficient Variations ( = -15 deg, Power On)
August 2010 Page 26

Figure 3.13 Moment Coefficient Variations ( = -15 deg, Power Off)
August 2010 Page 27

Figure 3.14 Comparison of Measured and Estimated Cowling Off, Tail Off Results
for = 0, Power On
August 2010 Page 28

Figure 3.15 Comparison of Measured and Estimated Cowling Off, Tail Off Results
for = 15 deg Power On
August 2010 Page 29

Figure 3.16 Comparison of Power On and Power Off Force Coefficients,

= 0, Configuration 3
August 2010 Page 30

Figure 3.17 Comparison of Power On and Power Off Moment Coefficients,

August 2010 Page 31

Figure 3.18 Comparison of Power On and Power Off Force Coefficients,

= -15 deg, Configuration 3
August 2010 Page 32

Figure 3.19 Comparison of Power On and Power Off Moment Coefficients,

= -15 deg, Configuration 3
August 2010 Page 33

Figure 3.20 Effect of Rudder Setting on Force Coefficients,

August 2010 Page 34

Figure 3.21 Effect of Rudder Setting on Moment Coefficients,

August 2010 Page 35

Figure 3.22 Effect of Sideslip Angle on Cowling On, Tail On Configuration

(Power Off)
August 2010 Page 36

Figure 3.23 Effect of Sideslip Angle on Cowling On, Tail Off Configuration
(Power Off)
August 2010 Page 37

Figure 3.24 Effect of Sideslip Angle on Cowling Off, Tail On Configuration

(Power Off)
August 2010 Page 38

Figure 3.25 Effect of Sideslip Angle on Cowling Off, Tail Off Configuration
(Power Off)
August 2010 Page 39

Figure 3.26 Effect of Sideslip Angle on Cowling On, Tail On Configuration

(Power On)
August 2010 Page 40

Figure 3.27 Effect of Tailplane Endplates on CZ and CM, = 0, Power Off,

Cowling On
August 2010 Page 41

Figure 3.28 Effect of an Extended Tail Boom on the Cowling On,

Tail On Configuration (Power On)
August 2010 Page 42

4 Development of a High Fidelity Gyroplane Simulation
4.1 Aims
Before developing any mathematical model it is important to be clear on the purposes
to which it will be put. For the current study there are three distinct uses for the
model. Firstly, there is clearly a gap in the understanding of the fundamental dynamic
characteristics of the gyroplane configuration. The aim will be to use the model to put
the stability characteristics of this configuration into context alongside those of fixed
wing aircraft and other rotorcraft. Secondly, the model can be used to undertake
parametric studies of the configuration to identify any key configurational parameters
which might strongly influence the stability characteristics. Third, it is anticipated that
the model might be useful in predicting the characteristics of new or modified
vehicles, highlighting the possibility of characteristics which might be detrimental to
airworthiness. This last aim, of course, can only be considered once the validity of the
model has been established (see Section 6).
In the following section the RASCAL mathematical model, and its reconfiguration for
use in simulating gyroplanes, will be discussed in some detail. Readers unfamiliar
with the basic concepts of rotorcraft mathematical modelling, and the methods used
to determine the dynamic stability of aircraft are referred to Appendix 4 for
background information.
4.2 Background
Mathematical modelling of rotorcraft dynamic characteristics is considerably more
challenging than that of fixed wing aircraft. This is due mainly to the additional
degrees of freedom from the rotor but also due to the nonlinearities present and the
coupling between the longitudinal and lateral/directional dynamics. The equations of
motion are expressed in the form:
m(U& + QW VR) = X mg sin

m(V& + RU PW ) = Y + mg cos sin
m(W& + PV QU ) = Z + mg cos cos
(4.1)
I xx P& I xz R& + QR( I zz I yy ) PQI xz = L

I yy Q& + RP( I xx I zz ) + ( P 2 R 2 ) I xz = M
I zz R& I xz P& + PQ( I yy I xx ) + QRI xz = N
where U, V, and W are the components of translational velocity in the x, y, and z body
axis directions, and P, Q, and R are the angular velocities about the axes. The roll and
pitch attitudes are given by and respectively, m is the aircraft mass, and Ixx etc.
are its moments of inertia.
August 2010 Page 43

The real effort in modelling a rotorcraft (or any other rigid body) is in the development
of suitable methods of calculating the external forces and moments, X, Y, Z and L, M
and N. In general the procedure would be to break down the vehicle into a number of
subsystems or components so that in the case of a gyroplane, for example:
X = XRotor + XProp + XFuselage + XTailplane + XRudder
(4.2)
Components such as those from the fuselage and the tailplane are simply the
aerodynamic loads which, for the current study, are readily obtained directly from the
wind tunnel data described in section 3. The real complexity lies in establishing the
rotor components for a given flight condition.
The loads generated by a rotor are calculated using the well understood approach of
blade element theory. The blade is divided into a number of elements, the normal and
tangential components of velocity at these elements is derived from the blade
kinematics allowing the angle of attack of each element to be established. A look-up
table of aerodynamic properties of the blade section used (NACA 8H12, for example)
is then used to calculate the lift and drag of the element. The component forces and
moments acting on the blade are then obtained by summing the contributions from
all of the elements, and making appropriate transformations into body axes. The loads
on a given element vary with its position around the azimuth giving a periodic forcing
with time. Some mathematical models essentially average the load over one
complete revolution to give steady values for loading, however, higher fidelity
representations such as the model used here, retain the periodicity of the rotor's
motion. It is clear that in order for this scheme to be accurate, the kinematics of the
blade motion must be accurately represented. This involves considering each
individual blade and fully representing its motion as a function of azimuth and vehicle
state. The geometry of the rotor system - position of the teeter hinge for example, is
therefore important. The position of a particular blade element will be dependent on
the bending of the blade as well as its teetering, and a simple structural model which
assumes that the blade is attached to the hub by a stiff torsional spring may also be
included. It is also worthy of note that if the rotor model is expressed in a generic
manner, then the same model can be used to represent the propeller, albeit with
different representative data, and this is the approach adopted here.
The description above gives the reader a basic idea of the nature of the mathematical
model used in this study. Further background information on rotorcraft modelling can
be found in the many excellent text books available on the subject [39, 40, 826]. The
mathematical model used in this study, RASCAL, was developed in a manner which
allowed it to be configured for any rotorcraft type, although it had mainly been used
for helicopter simulation [8789]. Details of the model are summarised in the
following section, the reader being referred to the published literature for more
extensive information.
4.3 A Description of the RASCAL Model
The RASCAL model was developed primarily to study the flight dynamics of
helicopters but with sufficient generality to allow it to be modified to capture other
rotorcraft types. In this section, it is introduced in its most fundamental form, whilst
its reconfiguration to represent a generic gyroplane configuration is detailed in the
next section. The model has been influenced strongly by the literature in the field,
taking account of the significance of research by others into individual blade/blade
element modelling, and dynamic inflow modelling [90]. The key features of the model
used are summarised in Table 4.1.
August 2010 Page 44

Table 4.1 RASCAL Mathematical Model Description
Model item Characteristics
Rotor dynamics (main, tail, or propeller) up to 10 individually-modelled rigid blades

fully-coupled flap, lag and pitch motion
blade attachment by hinges and springs
lag damper
Rotor loads aerodynamic and inertial loads represented
by up to 10 elements per blade
Blade aerodynamics lookup tables for lift and drag as function of
angle-of-attack and Mach number
Wake model momentum-derived dynamic wake model
uniform and harmonic components of inflow
rudimentary interaction with tail surfaces
ground effect
Transmission coupled rotorspeed and engine dynamics
up to 3 engines
geared or independently-controlled rotor
torque
Airframe fuselage, tailplane and fin aerodynamics by
lookup tables or polynomial functions
Atmosphere International Standard Atmosphere
provision for variation of sea-level
temperature and pressure
The model takes the form
A x& = f ( x, u )
(4.3)
where the state vector x contains the airframe translational and angular velocity, blade
flap, lag and pitch angles and rates for each blade on each rotor (in the context of
gyroplanes this implies a main rotor and a propeller), the induced velocity states for
each rotor wake as well as the angular velocity of both rotors, and the engine torque.
Elements of the control vector are the four controls, which vary with aircraft type, e.g.
single main and tail rotor configurations will have three main rotor controls and one
tail rotor control. Depending on the number of blades on each rotor, there can be up
to 100 non-linear, periodic ordinary differential equations describing the coupled rotor/
airframe behaviour. The body dynamics are represented by equations (4.1). Note that
the matrix contains off-diagonal terms not simply associated with airframe products
of inertia, but which also include terms associated with the dynamic inflow and blade
equations of motion.
The model has been used previously for helicopter validation and simulation studies,
[8789], and simulation of gyroplanes [62, 66, 67, 91].
August 2010 Page 45

4.4 Configuring the RASCAL Model to Represent a Gyroplane

As described above, the RASCAL model was written to represent a generic rotorcraft,
and configuration as a gyroplane required synthesis of an appropriate aircraft data file.
The data files for the VPM M16 and the Montgomerie-Parsons are given in
Appendix 2.
4.4.1 Aircraft Measurements
Accurate measurements of the aircraft's principle dimensions were made. These
included blade chord and length, propeller location, and teeter bolt location. The
aircraft was also weighed with and without fuel. Lists of the measured data are given
in Appendix 2. Note that the body axes frame of reference to which all geometric
measurements are referred is the point at which the mast and keel lines intersect
(rear mast in the case of the Montgomerie-Parsons).
The centre of mass position was identified as a key parameter and care was taken in
every set of tests to ensure that its position was known accurately. In the first
campaign the centre of gravity location was established by hanging the aircraft
(GBWGI) from its teeter bolt (at a known location). An inclinometer was placed on
the aircraft keel and masses added at a known point until the keel became level. The
centre of mass location could then be established by simple geometry. For the trials
involving GUNIV an alternative method of locating centre of gravity involving the use
of a calibrated aircraft weighing system was employed (see Appendix 3).
The aircraft pitching and yawing moments of inertia, Iyy, and Izz were obtained by hang
tests. A single-point suspension method (from the teeter bolt) was used for the
former, and a bifilar method for the latter [39]. Periods of oscillation were taken and
converted into the moments of inertia. Note that Ixx could not be measured due to the
nature of the suspension point provided. The most likely lowest estimate was made,
based on rotor and pilot masses.
4.4.2 Aircraft Aerodynamic Data
The VPM M16
The aerodynamic data from the wind tunnel tests was incorporated into the RASCAL
model for the VPM M16 in the form of look-up tables. These look-up tables contained
body axis fore-and-aft, lateral and vertical forces, and roll pitch and yaw moments at
five angles of attack within the range 40 deg and five sideslip angles within the
range 40 deg. Linear interpolation is used to estimate coefficient values at general
intermediate points between measured data values. The data from the tests with
power off was used, as this was thought to give the worst case in terms of the fin /
rudder effectiveness. Without the benefit of the propeller flow field, the forces and
moments arising from these components are likely to be less than might otherwise
be the case.
It should be noted that the data obtained from the wind tunnel tests was for a VPM
M14 gyroplane despite our wish to simulate the M16 variant. In aerodynamic
coefficient terms, the differences between the two aircraft are largely superficial, the
general shape and dimensions of both aircraft are similar. The use of M14 data to
simulate the M16 should not significantly compromise the validity of results,
particularly as the gyroplane operates at low speed.
August 2010 Page 46

The Montgomerie-Parsons
As it was not possible to test a Montgomerie-Parsons model in the wind tunnel, the
strategy adopted was to adapt the data available for the VPM to suit. The premise was
that both aircraft types are largely similar in basic configuration and so it is likely that
their aerodynamic characteristics would be alike. Account of differences in size were
easily incorporated by scaling based on appropriate planar areas. Some additional
synthesis of the VPM data was required to obtain pod lift and drag coefficients, CL
and CD. The pitching moment was estimated from the simple expression:
Cm = Cm0 + Cm
where the coefficients C m 0 and Cm were estimated using the synthesised CL and
CD values and an assumed moment arm.
4.4.3 Blade Airfoil Section Data
It is a feature of an individual blade/blade element model that a comprehensive
description of the blade section aerodynamic properties can be included. For both
aircraft, the manufacturer describes the rotor blade section as a NACA 8H12 and data
for this profile was obtained [92]. It became apparent when comparing the 8H12
profile with that of the actual blade on the VPM that there were some differences.
There was the possibility that these differences may have a significant effect on the
predicted properties of the blades and hence may have an effect on the simulation
results. In particular, there is an obvious relationship between poor prediction of drag
properties and resulting poor prediction of rotorspeed. An aerodynamic assessment
of the actual blade was made by GKN Westland Helicopters Ltd., [93] and its main
recommendations were as follows:
a) The low drag characteristics of the 8H12 may be diminished as spanwise flow and
centrifugal effects will prevent laminar flow.
b) The actual aerofoil has a greater CLmax at lower Mach numbers than the 8H12 but
at high Mach numbers the performance reduces rapidly.
c) The NACA 8H12 exhibits positive pitching moment coefficient and a negative
gradient. The VPM M16 aerofoil has a similar pitching moment coefficient but its
gradient is much less marked.
The results from this study were used to modify the existing 8H12 data set in order
to make it more representative of the actual blade used on the VPM aircraft. The data
was also used for the Montgomerie-Parsons model as blades from the same
manufacturer are used in this vehicle. Data for the 8H12 section is available for a
variety of surface roughness conditions. Initially data for zero roughness was applied,
however the resulting rotorspeed and Q derivative comparisons suggested that
higher drag was present in the real aircraft. Section data for greater roughness (and
hence higher drag) was therefore applied which halved the rotorspeed error and
improved the correlation in Q. When a comprehensive validation exercise is
performed, as in this case, one of the main aims is to identify where it is possible to
adjust the model to improve the fidelity of the predictions. Any changes to the model
must of course be physically justifiable. In this case, the validation exercise (Section
6) has highlighted the need to increase the drag of the blade section, and the
independent information available on this blade section, which indicates a sensitivity
to surface roughness, certainly provides the necessary justification for this action.
August 2010 Page 47

4.5 Results from the Mathematical Model

The mathematical model can be used to provide 3 related sets of data. Firstly the
model can be configured to provide trim data. The body accelerations are made zero
so that there are no resultant unsteady forces on the aircraft (essentially x , on the left
hand side of equation 4.3 is set to zero), and the equations solved to provide the
control positions and resulting body attitude to achieve trimmed flight for a given flight
velocity, for example. Next it is possible to obtain the vehicle's response to a
particular pilot input by using equation 4.3. This time the control vector u , is perturbed

to simulate a pilot input (a step change, from the trim value, of one or more of the
controls, for example) and a numerical integration scheme (Runge-Kutta, for example)
used to calculate the response of the vehicle. The output is a set of time histories
showing how the various states of the aircraft change in response to the input. Finally,
it is possible to linearise the nonlinear equations of motion given by equations (4.3).
The linearization process is described by Padfield [89], however for the purposes of
this report it is sufficient to say that the nonlinear set of equations (4.3) when
linearised are expressed in state-space form as:
x& = A x + Bu
(4.4)
where the symbols and their meaning are as follows:
x The linearization process is defined from a known reference point - usually taken
to be a trim state. The vector x is known as the state vector and contains the n
states which represent the dynamic characteristics of interest expressed as
perturbations from the trim state.
u This is the control vector and contains the m control variables (again as
perturbations from trim).
A This is a square matrix of dimension n which has a series of constants known
as the stability derivatives which again are referred to a particular trim state, and
represent the dynamic characteristics of the vehicle.
B This is the control matrix which is of dimension n x m and contains the control
derivatives.
One structure of model which has been found to be useful for gyroplane flight
dynamics is an extension of the classical longitudinal set of equations familiar for fixed
wing flight, which includes the rotor speed degree of freedom. For this model we
have:
u Xu Xw Xq XT X: X Ks
w Z Zw Zq ZT Z : Z
u Ks
x q , u >Ks @, A M u Mw Mq MT M : , B M Ks

T 0 0 1 0 0 0
: Qu Qw Qq QT Q: QK
s
(4.5)
August 2010 Page 48

Note that the convention is that lowercase characters are used in the linearised
equations. For example, in the nonlinear equations (4.1), U is the translational velocity
in the direction of the x-axis whilst in the linearised equations u is defined as the
(small) perturbation from the reference trim state, Ue, that is U = Ue + u. The main
limitation of these equations is that they are only valid for small perturbations from
the trim state. As they are linear, however, they are extremely useful in determining
the aircraft's stability characteristics, as discussed in Appendix 4.
As an example of how the simulation may be used, consider the case of the VPM
aircraft in its nominal configuration described in Appendix 2. Figure 4.1 shows the
variation in the trim values of the controls, rotor speed, and pitch and roll attitude over
the speed range from 30 to 70mph (IAS). The decrease in longitudinal hub tilt with
increasing airspeed is equivalent to the disc tilting forward, and a forward migration
of the stick. As would be expected, this is accompanied by the pitch attitude of the
aircraft reducing from about 6 at low speed to 1 at higher speeds. At all flight speeds
the aircraft sits "left wing down", as indicated by the negative roll attitude. Given that
the full travel of lateral tilt is equivalent to 18 (i.e. 9), the lateral tilt for trim over the
speed range of 2 is significant. Finally, rotor speed is also shown, and from this plot
it is apparent that there is only a slight increase in the rotor speed for trim over the
speed range. Figure 4.1 represents a small subset of the data which is produced by
the trim calculation; all of the aircraft and rotor states are calculated along with
performance related parameters such as power required and propeller thrust.
Consider now the use of the model to calculate response to control inputs. As an
example, Figure 4.2 shows the response of the VPM M16 to a longitudinal stick input
(shown in the first plot). The aircraft is initially in trim at a velocity of 40mph then, after
3 seconds, a step input of longitudinal stick equivalent to an aft hub tilt of 1 input is
applied. This is maintained for 5 seconds after which the stick is returned to its original
trim position. This input is typical of one used in BCAR Section T compliance tests.
Initially the short period mode is excited, which can be characterised by a heavily
damped oscillation in pitch rate, which is visible in the first few seconds of the pitch
rate response. Whilst the aft stick is held the pitch rate is positive and the pitch
attitude is nose up, reaching around 20 before the forward stick step input is applied,
arresting the motion. The longer term response, the phugoid, becomes apparent with
the periodic variation in airspeed (and attitude). Note that the high frequency
oscillations on the pitch rate plot are due to the teetering motion of the rotor that is
captured by the model through its representation of the individual blade dynamics.
They are not visible on the pitch attitude plot, for example, as the high frequency
content of the pitch rate response will, effectively be filtered by the integration
process used to obtain pitch displacement.
Finally, the nonlinear model can be linearised as described above to produce the
simplified representation given by equation (4.4). Using the structure shown in
equation (4.5) it is possible to predict the stability characteristics of the aircraft over
its speed range by calculating the stability derivatives (Xu, Xw, , Q) at each trim
speed, to form a system matrix A at each speed. The eigenvalues (see Appendix 4)
can then be plotted on the complex plane to examine the aircraft's stability
characteristics. The results for this exercise are shown in Figure 4.3 where the
eigenvalues are plotted again from flight speeds in the range 3070mph, as indicated.
As in fixed wing aircraft, the characteristic modes are a short period mode which is a
heavily damped oscillation in pitch, and the phugoid which is a longer period mode
characterised by oscillations in airspeed and attitude. The third mode is a rotor speed
mode which is aperiodic and varies very little over the speed range. The short period
mode damping varies from around 0.85 at low speed to 0.6 at higher speeds. The
phugoid is unstable at low speeds (i.e. the eigenvalues are to the right of the
August 2010 Page 49

frequency axis, see Figure A4.3, Appendix 4), but stabilises as speed is increased (the
eigenvalues move over to the left of the axis). The point of neutral stability occurs at
about 40mph, and the frequency here is approximately 0.04cycles/sec (equivalent to
a period of 25 seconds). Figure 4.4 is equivalent to Figure 4.3 however the
eigenvalues are plotted in an alternative form as frequency vs. damping ratio (see
Appendix 4), which is a more convenient format when investigating compliance with
BCAR Section T. As explained in Appendix 4, one consequence of plotting
eigenvalues in this way is that the stable region of the plot is now to the right of the
vertical axis. The short period mode is now on the right hand side of the plot as
positive damping ratio indicates a decaying amplitude of oscillation (i.e. stability). Note
that the rotor speed mode, cannot be displayed on this plot (damping ratio cannot be
defined for aperiodic modes), which is another consequence of plotting eigenvalues
in this way.
Figure 4.2 shows the response of the aircraft calculated using the nonlinear version
of RASCAL for a flight case initiated at a trim speed of 40mph. Examining the pitch
attitude plot it is apparent that the period of the phugoid oscillation is indeed around
25s. It should be appreciated that in this case the linear and nonlinear models are in
agreement as the perturbations from trim experienced are small. The linear model
(developed from the nonlinear model) would not be suitable for simulating larger scale
manoeuvres.
Although the results from the RASCAL simulation presented here can all be
discussed qualitatively and logically, the issue of just how representative these values
are with respect to those of the actual aircraft can only be established by a full model
validation exercise. Such an exercise is discussed fully in Section 6.
August 2010 Page 50

Figure 4.1 Trim Results from RASCAL Mathematical Model for VPM M16
August 2010 Page 51

Figure 4.2 Response of VPM M16 to a 5s Pulse in Longitudinal Stick (40mph, IAS)
August 2010 Page 52

Figure 4.3 Eigenvalues for VPM M16
Figure 4.4 Eigenvalues for VPM M16 Plotted Against Damping Ratio
August 2010 Page 53

5 Flight Testing
5.1 Background
Flight testing, as with any experimental activity, is a vital component of aeronautical
engineering studies requiring understanding of real behaviour. It is important however
to recognise that there are various categories of flight test as described by Bailey [79].
Since the design of the experiment and the resulting data is tuned exactly to the
nature of the specific application, the activity may appear abstract and obscure to the
lay person or pilot. The application for which data is required here, is model validation.
The flight testing is therefore best described as 'experimental' or 'research' flying and
by Bailey's definitions means that the nature of the tests are unusual.
Three test campaigns were undertaken. The first, using VPM M16 G-BWGI, was
conducted in 10 hours of flying based at Flight Refuelling Ltd's Bournemouth base in
October 1996, Figure 5.1. The second, using Montgomerie-Parsons Two-Place
G-UNIV, took place at Bournemouth (October 2000) and Carlisle (February 2001),
Figure 5.2. The third and final trial was conducted at Perth in July 2005, again using
G-UNIV.
The first two trials were designed specifically to gather data that could be used to
validate the mathematical model used for simulation studies. It was felt necessary to
undertake two sets of validation trials to gather data on different aircraft of the same
class to confirm the general applicability of the model. The third trial was designed to
gather rotor teeter data in steady and manoeuvring flight, with a view to making an
assessment of the behaviour of an actual rotor. It was hoped that these data could
also be used to further support mathematical model validation, but this was not the
primary aim. The first two trials were conducted under Flight Refuelling Ltd's 'B'
conditions approval; the third campaign was conducted under a Permit to Fly for Test
Purposes, with close CAA supervision and direction. Tables 5.1 5.3 summarise the
content of these three campaigns.
Table 5.1 Summary of VPM M16 Flight Trials
Flight number Airspeed (mph) Nature Location
1-4 - Pre-trials test flights Bournemouth
5 25 - 75 Steady, level flight. Zero sideslip Bournemouth
6 25 - 75 Steady, level flight. Sideslip Bournemouth

angles 5, 10
7 30, 50, 70 Longitudinal stick doublets; Bournemouth

phugoid (step) inputs
8 30, 50, 70 Longitudinal stick doublets; Bournemouth

phugoid (step) inputs
9 30, 50, 70 Lateral stick and rudder Bournemouth

doublets, steps, frequency
sweeps (70mph only)
10 70 Longitudinal, lateral and rudder Bournemouth

frequency sweeps
August 2010 Page 54

Table 5.2 Summary of Montgomerie-Parsons Two-Place Flight Trials

(Model Validation)
1,2 - Pre-trials test flights Bournemouth
3 60 Steady, level flight.Phugoid Bournemouth

(step) and doublets, all controls
4 60 Steady, level flight.Phugoid Carlisle

(step) and doublets, all controls
5 60 Frequency sweeps, all controls Carlisle
6 40 Phugoid (step) and doublets, all Carlisle

controls
7 40 Longitudinal frequency sweeps Carlisle
8 50 Phugoid (step) and doublets, all Carlisle

controls
9 40, 50, 60 Phugoid (step), doublets and Carlisle

frequency sweeps (longitudinal
stick only)
Table 5.3 Summary of Montgomerie-Parsons Two-Place Flight Trials

(Teeter Angle Measurement)
1 - Permit test flight Perth
2 30-70 Steady, level flight. Steady, Perth

turning flight (up to
30bank)Climb/descent
various power settings
3 40, 50, 60 Longitudinal stick; phugoid Perth

(step) inputs
4 40, 60 Steady, level flight. Sideslip Perth

angles 5, 10
5.2 Flight Test Procedures and Techniques

It is important to understand the terminology and purpose of these individual tests. A
general, steady flight condition can be defined in terms of airspeed; turn rate; rate of
climb or descent; and sideslip angle. Steady, level flight is therefore an equilibrium
condition at a constant airspeed and altitude. Nominally conducted with zero sideslip
(i.e. balanced flight), some tests were flown in an unbalanced condition with up to 10
deg of sideslip. The simulation model can then be set up in exactly the same flight
condition and a direct comparison made of the control positions, roll and pitch angles,
rotorspeed etc. required to hold that equilibrium condition. The control input cases
(step, doublet, frequency sweep and phugoid) perturb the aircraft from this
August 2010 Page 55

equilibrium condition, producing non-steady (i.e. time-varying) changes in the

measured parameters. These resulting 'time histories' can be used to characterise
the aircraft behaviour, or be compared directly with the simulation model disturbed by
exactly the same control input, or be compared mathematically in exactly the same
form as the model giving a direct comparison of the mathematical framework of the
simulation.
The fundamental objective when flight testing to provide data that characterises an
aircraft's response is to excite it in frequency and amplitude. Amplitude is
straightforward - the test pilot simply applies larger inputs with the appropriate flying
control. Frequency testing is less easy, or straightforward. It may be required to excite
a particular frequency or a range of frequencies. Airworthiness and flight safety
considerations are paramount. Large inputs may cause departure from controlled
flight or over-stress; certain frequencies may excite structural modes rather like flutter
in fixed-wing aircraft. These inputs must be applied from a steady flight condition to
have any meaning. Samples of what they should look like are shown in Figure 5.3.
These are not examples from flight but illustrate a mathematically perfect input that
the pilot needs to replicate as closely as possible. The first is called a step; the second
is a doublet; and finally a frequency sweep is shown.
The frequency sweep clearly excites over a range of frequencies. It is normal to start
at low frequency and build up to the faster cycles. The sweep shown here starts with
a 24 sec period, followed by 20 then 16, 12, 8, 4 and 2. The input then moves on to
higher frequencies up to a limit defined by flight safety considerations. The doublet
looks like a square sine wave and, indeed, in terms of frequency of excitation this is
how it impacts on the aircraft - doublets excite at a particular frequency. In this case
it is about 0.5 Hz (the frequency is not exactly the same as the sine wave analogue).
Finally, the step excites the aircraft at a range of low frequencies, although it is
obviously difficult to see any 'wave' content in the data (sophisticated mathematical
and numerical analysis tools [80], are required to see the frequency content of any
waveform, and will be referred to later in the data analysis).
5.3 Instrumentation
A range of parameters were recorded during each flight trial, and the instrumentation
used is summarised in Table 5.4. The fundamental flight mechanics data consists of
airspeed, angle of attack and sideslip; roll, pitch and yaw rates; roll, pitch and yaw
(heading) attitudes; longitudinal, lateral and normal acceleration (these with reference
to a set of axes fixed in the aircraft); rotor and propeller rotational speeds; and the
positions of the flying controls (control column pitch and roll, rudder and throttle
positions). These parameters were common to all three flight test campaigns. A
selection of the sensors used across the first two campaigns is shown in Figures 5.4
and 5.5.
The aircraft used for the first campaign of test flying was a privately owned VPM M16
gyroplane (G-BWGI). It is shown in unmodified form in Figure 5.1. The aircraft was
modified to carry appropriate measurement instrumentation by FR Aviation at their
Bournemouth Airport site. Additional specialised sensors were fitted for this trial in
the form of 5 strain gauges applied to various elements of the flying control system.
These had no part to play in this study but were fitted at the request of the CAA to be
analysed in-house as part of a follow-up investigation to the incident involving G-
YRAT, [81].
August 2010 Page 56

Table 5.4 Instrumentation Used in Flight Tests
Parameter Description Units Measurement Device
ax Longitudinal Acceleration m/s2 Accelerometer
ay Lateral Acceleration m/s2 Accelerometer
az Normal Acceleration m/s2 Accelerometer
P Roll Rate rad/s Rate Gyro
Q Pitch Rate rad/s Rate Gyro
R Yaw Rate rad/s Rate Gyro
Pitch Attitude rad Attitude Gyro*

Angle Indicator**
Roll Attitude rad Attitude Gyro*

Angle Indicator**
Heading rad Pick-up from existing sensor*

Angle Indicator**
Vf Indicated Airspeed knots Air Data Probe
h Altitude m Air Data Probe*
Sideslip Angle rad Air Data Probe
Angle of Attack rad Air Data Probe
s Longitudinal Stick Position % Displacement Transducer
c Lateral Stick Position % Displacement Transducer
p Pedal Position % Displacement Transducer
t Throttle Position % Displacement Transducer**
Ng Propeller Speed rpm Induction coil*

Optical sensor**
Nr Rotor Speed rpm Pick-up from existing sensor*

Optical sensor**
xE, yE, zE Aircraft position m GPS**
t Teeter Angle rad Rotary Transducer**
Blade Azimuth rad Rotary Transducer**
* VPM M16 (G-BWGI) ** Montgomerie-Parsons (G-UNIV)

For the second set of trials the University of Glasgow-owned Montgomerie-Parsons
gyroplane, G-UNIV was used, Figure 5.2. This two seat aircraft was modified by the
manufacturer to accommodate the necessary instrumentation, essentially by
removing the second seat. The sensor and recording system was designed and built
by University of Glasgow technical staff using "off the shelf" components.
August 2010 Page 57

The third campaign (the second involving G-UNIV) involved measurement of teeter
angle and this required a newly-manufactured rotor head assembly. Considerable
investigation was undertaken to establish the best means of doing so with regard to
robustness of signal with minimum impact on the existing rotor head design as
shown in Figure 5.6 a) The teeter angle was measured by means of an angular
displacement transducer driven by a gear secured to the teeter bolt, Figure 5.6 b). As
the transducer drive gear is made of plastic, any malfunction of the sensor would
result in the gear teeth being stripped thereby having no effect on the controllability
of the aircraft. This philosophy was applied consistently throughout the design of the
new instrumentation to ensure that the airworthiness and flight safety of the vehicle
were not compromised. The rotary transducer is mounted on a modified teeter bolt
support plate, which was extended to carry the transducer mounting bracket and a
plate holding circuitry for signal conditioning and amplification. During flight this
circuitry is protected by a plastic cover, Figure 5.6 c). Note that the modification to the
support plate extended above the teeter bolt and so did not weaken the overall head
assembly. It should also be stressed that all modifications to the rotor head were
conducted after consultation with the aircraft manufacturer, Jim Montgomerie to
ensure the appropriate material specifications and manufacturing standards were
maintained.
It was decided to use slip rings to transfer data from the rotating to non-rotating frame
of reference rather than using any form of telemetry. The slip ring arrangement is
shown in Figure 5.6 d). Four slip rings were bonded to the underside of the pre-rotator
ring gear, one of these carried the signal from the teeter sensor which was then
picked up by the non rotating contacts before being recorded by the on-board
computer. Again great care was taken to ensure that the instrumentation, wiring
contacts etc. did not interfere with the normal working of the aircraft and its systems.
The contacts were mounted on a modified hub bar. The modification involved
extending the hub bar beyond its original pivot point. There is therefore no change to
the load path or the loads carried by the bar. The final modification to the aircraft was
a display of vane sideslip for the pilot to assist accurate setting of test points.
A key choice to be made with a digital recording system is the sample rate used. High
sample rates can be good for frequency resolution with long record lengths (such as
frequency sweeps) but demanding of storage capacity. Low sample rates can
adversely affect frequency resolution and introduce aliasing in the recorded data, i.e.
introduction of 'ghost' or 'spoof' frequencies. As usual, a compromise is required. The
sample rate for the VPM M16 trial was set at 10 samples per second; for the two
Montogmerie campaigns it was variable but for flight mechanics parameters such as
air data and inertial measurements a sample rate of 50 per second was used. A much
faster sample rate, 200kHz, was required for the teeter angle campaign.
5.4 Conduct of Flight Trials
In each campaign, once the instrumentation had been installed and ground tested, the
aircraft was weighed and the new centre of mass position determined. All appropriate
documentation was completed, and there was close consultation with the CAA at all
times to ensure that the tests were carried out in a completely safe manner. The tests
were carried out in an incremental manner with replay facilities to examine the
recorded data available immediately after landing. There were some specific issues
which arose with each trial, and these are documented briefly below.
August 2010 Page 58

5.4.1 VPM M16 (G-BGWI) Trial

For the VPM M16 trial, mass and centre of mass position were determined for the
aircraft in standard condition, as well as modified. It was necessary to calculate the
latter for airworthiness and flight safety reasons as the aircraft would be flown from
the rear seat, normally prohibited solo. It was decided that, provided the weight and
balance of the aircraft in modified form lay within the (undefined) envelope for normal
operation, then the likelihood of the aircraft handling being changed would be
minimised. This normal operational envelope was calculated from hang-test
measurements taking into account the mass of fuel (from minimum to maximum) and
the pilot.
During the trials, engine management problems, particularly at low power settings,
resulted in frequent stoppages and poor running, and resulted in the decision not to
apply frequency sweep or multi-step inputs to the throttle. It was felt by the test pilot
that this might compromise engine integrity and hence flight safety, given his growing
concerns with the serviceability of the engine.
5.4.2 Montgomerie-Parsons (G-UNIV) Model Validation Trials
a) Carlisle, July 2000, Pre-Trials Verification of Aircraft and Instrumentation
Before the main series of trials took place it was considered prudent to undertake
a series of "shakedown" flights primarily to test the functionality of the
instrumentation suite, but also to practice procedures for downloading data from
the onboard computer. These trials were funded by the Department of Aerospace
Engineering, University of Glasgow and took place during the month of July 2000.
The tests were restricted to steady trimmed flight and simple manoeuvres to
gather performance data (steady climbs etc.). A problem with an incorrectly
installed roll rate sensor became apparent and was immediately rectified.
Thereafter the aircraft and instrumentation proved to be reliable, and a procedure
for recovery and initial processing of data was developed and perfected. Inspection
of the recorded results provided confidence in the ability of the aircraft system to
generate the data required to fulfil the objectives of the main test programme.
b) Bournemouth, 16th - 27th October 2000
A number of test flights were flown to ensure that the aircraft was ready before it
was then handed over to FRA and placed under B conditions. However, the
following day the aircraft developed a fuel tank leak. The aircraft was grounded and
the instrumentation system removed to allow repair of the fuel tank. Following the
repair, one flight was then performed to record data in trim, and responses to step
and doublet inputs at 60 mph. Inspection of the data from this flight also
highlighted a problem involving mis-wired pitch and yaw rate gyros (pitch rate data
being recorded to the yaw rate channel on the PC and vice versa). This problem,
which had not been detected in the earlier shake-down trials as these involved
relatively benign steady flight manoeuvres, was easily rectified.
Bad weather caused a short delay in flying, and when the weather improved a
further leak in the fuel tank was discovered. The aircraft was grounded and the
flight trials were postponed until a replacement tank could be obtained. It was
agreed, with the approval of the CAA, to complete the test flying at Carlisle airfield
as soon as a replacement fuel tank could be sourced and at a date agreeable to all
concerned. Operations at Carlisle were to remain under FRA's B conditions.
August 2010 Page 59

c) Carlisle, 19th - 23rd February 2001

The remainder of the data required was to be generated during trials from Carlisle
Airport during the week 19th - 23rd February, 2001. Carlisle was seen as a sensible
choice because, due to the relatively low volume of air traffic in the area, it was
possible to perform the tests close to the airfield minimising transit time and
thereby increasing the number of flights possible in a day. This was particularly
significant when considering the very low temperatures encountered that week.
By the end of the period all scheduled tests had been flown, and the test
programme had been completed.
5.4.3 Montgomerie-Parsons (G-UNIV) Teeter Trials
The flight test programme was carried out at Perth Airport on the 20th and 21st of
July 2005. Before proceeding with the experimental flights, a Flight Trials Instruction
was drawn up by the CAA Flight Test Engineer, agreed by the University of Glasgow
team and the pilot - see Appendix 5, before finally being authorised by the CAA.
Details of personnel involved, and the various operational aspects are also given in the
Appendix. Particular care was taken in this trial in reviewing the recorded data after
each flight as this trial represented the first ever measurement of gyroplane rotor
teetering motion and there was, naturally, a degree of uncertainty as to what would
be encountered.
5.5 Processing of Recorded Data
During each set of trials, data was recorded for the complete duration of each flight.
The first stage of processing was to download the data from the onboard recording
system to a PC, converting it to engineering units, and formatting it into a convenient
form (Matlab or Mircosoft Excel). The second stage of processing was to extract only
the events of interest. Both aircraft were equipped with event markers to indicate
specific test points on the data records to facilitate identification.
Data quality can be assessed in a rudimentary fashion by inspection and, as happened
during the trials, minor errors such as incorrectly labelled data streams were easily
rectified. Drop-outs or spikes can be identified for smoothing, typically using an
averaging process based on neighbouring points. Magnitudes and signs can be
checked using engineering judgement as well as knowledge of the calibrations
applied, however, a formalised method of checking the quality of rate and attitude
sensors, and air data and inertial sensors, can be employed. This is called kinematic
consistency, and details of its nature and use are given in Appendix 6.
5.6 Sample Flight Test Results
Typical results are shown in Figures 5.7 - 5.9. Figure 5.7 shows a longitudinal stick
frequency sweep conducted during the VPM M16 trial. Figure 5.8 shows a
longitudinal stick doublet input applied to the Montgomerie-Parsons. Figure 5.9
shows teeter angle measurement in steady, level flight, taken from the
Montgomerie-Parsons trial during the third flight test campaign. These data are really
designed for analysis using mathematical techniques to 'tease out' the dynamic
features of interest. Nonetheless, it is the case that best practice demands that the
data be 'eyeballed' and engineering judgement applied as much can be learned about
the status of the instrumentation system, and the nature of the aircraft itself, which
can shape consideration of airworthiness issues in relation to the experiments
underway. Figure 5.7 is typical in this regard. Notice the presence of 'dropouts', or
spikes. This is an electrical phenomenon, and clearly is not the result of aircraft
behaviour. Dropouts are a feature of older, analogue systems, such as the one used
in the VPM M16. The Montgomerie-Parsons data does not suffer from dropouts
which are in any case easily corrected numerically. Note that the frequency sweep is
August 2010 Page 60

similar to the ideal shown in Figure 5.3, although it is difficult for the pilot, as
frequency increases, to keep the amplitude constant. This can have airworthiness
implications as imposing large-amplitude inputs at high frequency can accelerate
fatigue damage despite the fact that at these frequencies the airframe response
diminishes greatly. The frequency sweep is therefore an input that needs application
with care. However, the benefits of the frequency sweep input can be seen, even
without mathematical analysis, if the records between 100 and 200 s are examined.
The rotorspeed variation is at its greatest, 40 rpm, when the magnitude of the
control input is at its smallest. This is indicative of a resonance, or mode, that might
not be obvious in normal operation.
The Montgomerie-Parsons sample shown in Figure 5.8 is of a doublet input. This
aircraft proved a particular challenge to flight test using these specialised flight test
control inputs. Note that the shape of the doublet lacks the sharp edges of the
idealised sample shown in Figure 5.3. The fundamental problem with this aircraft is
the undamped pitch oscillation, most clearly seen in the pitch rate time history. The
case shown here, at 40mph, is benign in this regard but at higher speeds this
oscillation is intrusive. The aircraft has a very lightly damped mode that rendered
accurate inputs difficult to make. Note also that the airspeed time history is
meaningless, being unrealistically low. The air data system proved prone to blockage
by moisture and airspeed time histories had to be reconstructed from inertial data
using the kinematic consistency approach. The rotorspeed time history also suffers
from lack of resolution, making the dynamic behaviour difficult to identify. However,
the sensor was changed for the teeter angle trial, Figure 5.9 where it can be seen to
produce a much 'cleaner' signal.
The teeter trial data shown in Figure 5.9 illustrates the quality of data taken from the
rotor head using the slip ring assembly. Two sensors were used, the second being
provided for redundancy. Only the master is shown. The rotor azimuth position is vital
to interpret teeter data. For example, the maximum teeter angle of approximately 2.5
occurs when the blades are approximately 40 from the fore-and-aft position. To aid
the understanding of the teetering motion an individual to multi-blade transformation
can be made. This has the effect of replacing individual blade motion by a disc
oriented at particular longitudinal and lateral tilt angle. This transformation can be
expressed mathematically as:
1 = t cos
c
and 1 = t sin
s
The teeter angle t, represents the orientation of the hub bar relative to the shaft
which is then transformed to disc co-ordinates 1c, being the longitudinal tilt and 1s
being the lateral tilt of the disc. The blade azimuth position, , is taken to be zero
when the blade lies on the centre-line of the aircraft ahead of the pilot, and the teeter
angle t, is taken to be positive when the hub bar is tilted forward. This gives aft
longitudinal disc tilt as being negative, and for a rotor rotating in an anti-clock wise
direction (when viewed from above), a positive lateral disc tilt indicates a tilt to the
right. Referring again to Figure 5.9, a maximum teeter angle of 2.5 means that the
tip-path plane is therefore tilted aft by approximately 1.9 (= 2.5 cos 40), and to the
left by about 1.6 (= 2.5 sin 40). Data from the teeter trial is of extremely high quality
and forms a rich and unique resource.
August 2010 Page 61

Figure 5.1 VPM M16 G-BWGI in Unmodified Form
Figure 5.2 G-UNIV at FRA Bournemouth Airport for Initial Trials (October 2000)
August 2010 Page 62

Figure 5.3 a) Standard Test Inputs (A Step Input)
Figure 5.3 b) Standard Test Inputs (A Doublet Input)
Figure 5.3 c) Standard Test Inputs (A Frequency Sweep)
August 2010 Page 63

Figure 5.4 a) Flight Test Instrumentation for VPM M16 G-BWGI Flight Trials
(G-BWGI in Modified Form for Flight Trials)
Figure 5.4 b) Flight Test Instrumentation for VPM M16 G-BWGI Flight Trials
(Air Data Probe)
August 2010 Page 64

Figure 5.4 c) Flight Test Instrumentation for VPM M16 G-BWGI Flight Trials
(Rear Cockpit)
Figure 5.4 d) Flight Test Instrumentation for VPM M16 G-BWGI Flight Trials
(Inertial Unit)
August 2010 Page 65

Figure 5.5 a) G-UNIV Flight Test Instrumentation

(General Layout of Instrumentation)
Figure 5.5 b) G-UNIV Flight Test Instrumentation

(Main Instrument Pallet)
August 2010 Page 66

Figure 5.5 c) G-UNIV Flight Test Instrumentation

(Sensors On-board Main Instrumentation Pallet)
Figure 5.5 d) G-UNIV Flight Test Instrumentation

(Displacement Transducers for Measurement of Stick Position)
August 2010 Page 67

Figure 5.6 a) Additional Modification of G-UNIV for Teeter Angle Measurement

(Rotor Head Prior to Installation of Teeter Angle Instrumentation)
Figure 5.6 b) Additional Modification of G-UNIV for Teeter Angle Measurement

(Additional Modification of G-UNIV for Teeter Angle Measurement)
August 2010 Page 68

Figure 5.6 c) Additional Modification of G-UNIV for Teeter Angle Measurement

(Rotor Head Assembly After Modification)
Figure 5.6 d) Additional Modification of G-UNIV for Teeter Angle Measurement

(Slip Ring and Teeter Bar Assembly)
August 2010 Page 69

Figure 5.7 Longitudinal Stick Frequency Sweep (VPM M16)
August 2010 Page 70

Figure 5.8 Longitudinal Stick Doublet (Montgomerie)
August 2010 Page 71

Figure 5.9 Trim with Teeter Measurement (Montgomerie)
August 2010 Page 72

6 Model Validation
An engineering model is a powerful tool for supporting studies into aircraft stability,
control and handling qualities but confidence in the results can only be quantified if
the model has a known level of validity. An important precursor to validation is
verification. The verification process involves determining whether the simulation is
functioning correctly, that is, are the numerical processes within the computer code
working, is the computer code "bug free", and most importantly, can the results be
interpreted in a sensible manner when considered in the context of the known
physics of the actual system. This process is often performed by comparison of
results with other simulations, or in the case of a simulation undergoing development,
with results from a previous version. A successful verification gives sufficient
confidence in the functionality of the simulation to allow validation (that is comparison
of simulated data with real life data) to proceed. Verification is an important process
as it eliminates the possibility of a deficient (computational) implementation being the
cause of any mismatch between simulated and test data at the validation stage.
Validity, at its simplest level, describes whether or not a model is 'good enough' for
predicting behaviour. Hence validation must involve comparison, in some way, with
flight. A valid criticism of the process is that if the model compares well with a specific
test case, then it is only valid for those flight conditions or control input. In an attempt
to generalize any comparison to give confidence in wider applicability, model fidelity
is generally described in terms of amplitude and frequency, and here this is addressed
by comparing the model's trim, linearised small perturbation derivatives and the
response to large amplitude control inputs, with corresponding results from flight test
data. Note that what constitutes 'good enough' depends very much on the
engineering judgment of the user - formal approaches to the general problem of
validation do not exist.
Notwithstanding this, the overall objective remains the definition of the predictive
capability across a wide range of operating conditions from steady flight, to the
transient response to large control inputs. Note that the mathematical model is
regarded as generic, and is configured as a specific gyroplane type by means of a data
file that defines the model as that type. In principle then, the model can be used for
any gyroplane and if validated against a specific type, should exhibit the same level of
fidelity when used to simulate other aircraft. It is accepted that this may not be the
case due to type-specific attributes of particular aircraft and, to this end, the model
was validated against both the VPM M16 and Montgomerie-Parsons gyroplanes.
Although gyroplanes have been simulated previously [93, 94], the models and the
results obtained were severely limited in their applicability. For example, simple disc
representations of the main rotor were used, and analytical expressions for linearised
stability and control derivatives were formulated. Reference 94 was further limited by
using data for a generic aircraft type, i.e. a specific aircraft was not modelled. No
validation was attempted in either case.
6.1 Methodology
Comparison of model with flight for trim cases is very straightforward -
measurements of parameters such as stick position or pitch attitude made in flight
are compared with the calculations made by the model. Even in 'trim', the flight
measurements tend to vary with time, so an average over a period such as 10
seconds is taken to define 'trim'. Generally, however, the quality of the trim flight
results are excellent, and even visual inspection confirms that the aircraft is in steady
flight. Nevertheless an average over 10 seconds is calculated and the average is taken
to represent trim.
August 2010 Page 73

Validation in the dynamic or manoeuvre cases is much less clear-cut. The lay reader
might expect that comparison of model and flight time histories is all that is required.
This is an attractive prospect as time histories are intuitive. Time history comparison
does indeed have a role to play, particularly when inputs of varying amplitude are
applied. However, any particular input and resulting time history has a very specific
frequency characteristic, as the input itself has specific frequency content. The
objective is therefore to find a means of validating the model that is less input-
specific. The derivatives (see Appendix 4) form the basis of this analysis as, in
combination, they define the response characteristics across a range of frequencies.
Appendix 7 describes the method by which these derivatives are extracted from flight
test data. Since they are going to be used as validation metrics against the model, it
is important that they are verified (note, not validated) by assessing how they predict
the behaviour of the real aircraft to a control input. Note that in this process it is
important to use an input that was not used to identify the derivatives, in order that a
degree of independence in the verification is preserved.
The derivatives identified from a flight test input are used to create a linearised model
of the aircraft (see Appendix 4). The fundamental linearised model structure is:
x& = A x + Bu
with the derivatives filling the rows and columns of A and B. Typical results are shown
in Figures 6.1 to 6.3, all for the VPM M16. Figure 6.1 is for 30mph, and the responses
are represented as perturbations from trim (linearised models can only represent
perturbations). The M16 displays no real separation of the traditional short-period and
phugoid modes at lower speeds, and hence the combined response is presented.
However, Figures 6.2 and 6.3 show short-period and phugoid-type responses at 70
mph where these modes do apparently separate into their traditional forms. In all
three cases the close similarity between the flight response and that predicted by a
model, itself derived from another flight response, give confidence that the
derivatives accurately represent the behaviour of the real aircraft and hence can be
used to assist model validation. Note that where discrepancies do exist (most notably
in the VPM M16 rotorspeed short-period response at 70mph), these tend to be where
the response diverges significantly from trim and, since the identified model is linear,
it is therefore only really suitable for use with small perturbations.
6.2 Results
The RASCAL simulation model was configured with data appropriate to the flight
condition for each test point. Mass and inertia information was available from ground-
based measurements. The aircraft mass for each test point was influenced only by
fuel state which could not be measured accurately. Accordingly, simulation results
were obtained for the two limiting conditions of full fuel (maximum weight) and zero
fuel (minimum weight).
6.2.1 Equilibrium Flight
Comparison of the model with measurements taken in steady, level flight for the VPM
M16 and Montgomerie-Parsons is shown in Figures 6.4 and 6.5, respectively.
Generally, there is very good or excellent agreement between model and flight. The
longitudinal parameters (stick position, pitch attitude, rotorspeed) show excellent
correlation between flight and theory. The lateral variables (lateral stick position and
roll attitude) are less good. However, correlation of these parameters is sensitive to
sideslip and blade torsional effects (see Section 9). The model (both its physical
August 2010 Page 74

formulation and the engineering data used to configure it as a specific type) can
therefore be assessed to offer a realistic facsimile of the actual aircraft in trim flight.
Often in rotorcraft model validation it can be difficult to achieve this level of fidelity.
Note that with regard to the longitudinal parameters in particular, this level of fit easily
satisfies the requirements stipulated by the FAA for Level D (i.e. zero flight-time)
piloted flight simulators [95].
6.2.2 Small-Perturbation Flight
The RASCAL simulation model can be linearised using either the process used to
extract derivatives from the flight test data, or by a simple but quite effective
numerical process called forwards and backwards differencing [88]. The latter is used
in this case. Figure 6.6 shows comparisons of key longitudinal stability and control
derivatives for the VPM M16. The method of extracting the derivatives from flight test
produces bounds on the confidence level with which the derivative has been
estimated. Here, the flight-derived values are presented as 95% confidence bounds,
while the model results are those for minimum and maximum mass. Note that if the
95% bounds are far apart the method has estimated that particular derivative with low
confidence, and vice-versa.
Some simulation and flight derivatives show a consistent trend with speed.
Simulation values can lie within, or overlap, the uncertainty bounds of the flight
results. It is argued that in such cases, model and flight can be said to be in agreement
within the limitations of derivative identifiability and modelling uncertainty. Some
comparisons are worthy of consideration. For example, the drag damping term Xu,
which is of paramount importance in phugoid mode damping, is predicted by the
model to lie just on or outside the lower boundary of the flight result. However the
flight boundaries are fairly wide, indicating that Xu is estimated with low confidence,
except at high speed. Heave damping Zw plays an important role in short-period mode
damping, but its magnitude is substantially over-estimated by the simulation model.
This is a difficult result to reconcile, and it is not a result specific to the RASCAL model
- even simple theory [83] gives a result very similar to RASCAL. If the flight result is
accurate and simple theory is correct, this suggests that the blade lift-curve slope of
the real aircraft is unrealistically small. Of less concern, however, is the mismatch in
Mw. The model results appear simply shifted relative to the flight values; significantly,
the model captures the unusual result for unaugmented rotorcraft, that Mw <0. A
physical reason for this result is given by Houston [66]. Coupling of rotorspeed with
the pitch degree of freedom, manifested in terms of M, is well-represented by the
simulation model.
The simple forms of the derivatives given by Bramwell, [83] can give some indication
as to the engineering source of any modelling deficiencies and this is where
derivatives have a pivotal role to play in insight. For example, Mw and M are
influenced by the vertical location of the c.g. [66] so a good match in both indicates
that this parameter is correctly modelled. However, Mw is also influenced by pod and
tailplane aerodynamics (and given the apparent weakness of the tailplane lift-curve
slope suggested by the wind tunnel model, it is the pod that dominates), hence the
pod aerodynamic data is possibly responsible for the apparent offset between flight
and model. Note that the correlation in Mw and Mq could be improved if the pitch
moment of inertia was some 50% greater, but this would then worsen correlation in
Mu and M. These deliberations are typical of any model validation exercise and
serve, if not to highlight errors in the model, then certainly the detailed role that
derivatives and only derivatives can play.
August 2010 Page 75

The rotor torque derivatives (the Q derivatives) are unique to rotors in autorotation,
and are therefore of major importance to simulation of gyroplanes. It is believed that
this work is the first time that rotorcraft torque derivatives have been identified in
autorotation so there is no body of work against which either the estimation or
calculation can be compared. In addition, given the sensitivity of rotorspeed to a
multiplicity of factors, good modelling of these terms gives confidence that the
physical modelling is accurate. Some of the key derivatives are presented in Figure
6.6, which shows that the simulation model results exhibit excellent agreement with
flight. In particular Q is proportional to blade drag coefficient (in simple modelling,
[62]), so the match here is a good indication that the detailed blade element drag
modelling is correct in terms of overall effect.
Figure 6.7 presents comparable results for the Montgomerie-Parsons aircraft. The
exercise was repeated with this aircraft in an attempt to provide corroboration of
modelling assessments made with the VPM M16. It should be remembered that, at
the time, the VPM data were unique; there was no body of literature that provided
alternative evidence as to how well rotorcraft models performed (in terms of their
validity) as gyroplanes, unlike the helicopter literature which is well-populated with
case studies. The Montgomerie-Parsons comparisons were therefore undertaken as
a check on the VPM study, in an attempt to enhance confidence in the use of the
RASCAL code. However the Montgomerie-Parsons flight trials and modelling differ
from the VPM case. First, the rotorspeed sensor fitted to the Montgomerie-Parsons
lacked resolution (as can be seen by inspection of Figures 5.8 and 5.9, for example),
and this adversely influenced the confidence with which rotor torque (Q) estimates
could be extracted from the flight data. Second, the airframe aerodynamics used in
the RASCAL code were gross simplifications relative to the high quality tunnel-
derived data used for the VPM modelling. This was because no tunnel testing of the
Montgomerie-Parsons configuration was carried out. The consequences of these
aspects can be seen in Figure 6.7 - there is generally poorer correlation between flight
and theory in terms of trend and matching of values. However, there is some
consistency with the VPM results, in that the Montgomerie-Parsons flight results
show low values for Xu and Zw.
6.2.3 Large-Amplitude Response
The derivatives summarise the small-amplitude or linearised response of the
mathematical model. For all their utility and power they represent a view of the model
that is limited in amplitude and frequency. Time response of the full, non-linear
individual blade/blade element model from which the derivatives can be extracted, is
itself limited in frequency and amplitude to the characteristics of the control input
applied, yet is viewed usually by the lay reader as the ultimate test of a mathematical
model - if it can predict the response to a control input, it's an accurate representation
of flight. It is re-iterated however that because we can only ever test one or two
cases, response to control inputs is part of the validation process, not a validation
process in its own right, and the following cases illustrate this.
Figure 6.8 shows comparisons of response to a longitudinal stick input for the VPM
M16. The longer-term rotorspeed response, i.e. greater than 10 seconds, is more
heavily damped in simulation model than in flight. This is entirely consistent with
linearised model validation which revealed that the real aircraft has little drag
damping, Xu, and consequently little phugoid or long-term damping. The opposite is
true with the linearised simulation model. The pitch rate response is simulated fairly
well, the only anomalous period in time being around 10 sec, consistent with the
mismatch in rotorspeed. However, the off-axis response in roll rate displays behaviour
familiar to helicopter flight dynamicists, where the amplitude is reasonably predicted
in magnitude, but not in phase.
August 2010 Page 76

Figure 6.9, 6.10 and 6.11 show the response of the RASCAL code configured as the
Montgomerie-Parsons aircraft, when compared with flight test data. Three separate
control input responses are shown: the first is a BCAR Section T compliance
demonstration case, where it is desired to excite the phugoid or low-frequency mode;
the second and third are doublet inputs designed to excite only the short-period or
higher-frequency mode. In Figure 6.9 (phugoid response) the model is adequate only
for about 5 sec after the input has been made, capturing pretty well the angle of
attack, pitch rate and rotorspeed response during this time. From 15 sec onwards, the
aperiodic instability that is predicted by the model takes over and it can be seen that
all three model responses diverge rapidly, whereas the flight data does not. Figure
6.10 shows the response to a doublet input, where it is to be expected that the
longer-term response is not excited, and this is indeed seen to be the case for both
model and flight. Short-period pitch rate response is captured well, the angle of attack
less so. Rotorspeed judgment is less easy to make due to the limitations of the
sensor alluded to earlier, but the model and flight correlate at least in that the
magnitude of the rotorspeed response is small. Figure 6.11 shows a very similar input
however, in this case, the longer-term response is quite dissimilar to that shown in
the previous figure despite the input being very similar (the nominal airspeed is only
5 mph greater and has no bearing on the dynamic response). In this case both the
short-period pitch rate and angle of attack are captured accurately. However there is
a longer-term, neutrally-damped oscillation not present in the previous case despite
the similarity of pilot control input but which is nonetheless captured accurately by the
model, at least in terms of damping and frequency if not phase. While this is of credit
to the model, the difference in the long-term response between Figure 6.10 and 6.11
is perplexing until one inspects the respective control inputs - note that in the case of
Figure 6.11 the pilot control input varies sinusoidally, driving the oscillation seen in the
response, whereas in Figure 6.10 the control is held rigidly fixed after the doublet
input is made. This emphasizes that comparing only the response to control inputs
does not constitute model validation in the widest sense, but only narrowly verifies
that the model can capture a specific control input response.
6.3 Discussion
The criteria used to assess whether or not the rotorcraft model can adequately
simulate the gyroplane are necessarily subjective, relying on engineering judgement
since there are no formal criteria for gyroplane simulation. However, comparisons
with FAA Level D simulator requirements for helicopters, [95], can be instructive.
They are expressed in terms of trim and time response comparisons, and the model
generally satisfies the requirements for control position prediction to be within 5%,
and attitude to be within 1.5 deg. Primary axis time responses would fall within
acceptable envelopes, although the cross-coupling would not. The criteria for
rotorspeed simulation in autorotation is not pertinent to the gyroplane.
Although there are no criteria for derivative comparisons, Tischler, [96], is developing
criteria for frequency response validation. It is accepted that the linearised 6 DOF
model structure is of limited utility in the context of frequency response methods,
being applicable to a much narrower bandwidth. However, the validation envelope
defined by the approach taken in this report is consistent with pilot control strategy
for gyroplanes, which is essentially low-frequency. In addition, the derivative allows a
high degree of insight into the causes of specific modelling deficiencies, such as the
heavily damped phugoid (Xu).
Notwithstanding the lack of formal criteria, it is argued that validation discrepancies
are typical of that obtained for helicopter models, [97]. For example, the poorly-
predicted directional characteristics are determined largely by airframe aerodynamics
which are obtained from wind tunnel tests. However, discrepancies in the drag and
August 2010 Page 77

heave damping derivatives Xu and Zw could be directly attributed to modelling of

physical phenomenon unique to the gyroplane, specifically the induced velocity in
autorotation. However, the functionality of the Peters wake dynamic inflow model for
the gyroplane appears acceptable and, given that no previous gyroplane model
validation has taken place, a definitive statement cannot be made until the predictive
ability of the model for simulating other types reveals whether or not errors in Xu and
Zw in this case are indicative of a generic problem or are type-specific. Note that if the
model can simulate gyroplane flight, it should in principle be capable of simulating
helicopters in this flight regime as well. It is therefore important that it is capable of
being used without major modification to the underlying physical modelling.
Assessment as to whether or not the model is suitable for studies of gyroplane flight
is ultimately a matter of engineering judgment, relying upon assessment as to
whether or not the model is 'good' enough. Given its use for parametric studies, i.e.
to establish trends or tendencies with respect to changes in configuration
parameters, it is judged that the modelling is at least as good as if not better than
contemporary helicopter mathematical modelling practice.
August 2010 Page 78

Figure 6.1 VPM M16, Identified Model Verification, 30mph
August 2010 Page 79

Figure 6.2 VPM M16, Identified Model Verification, 70mph Short Term Response
August 2010 Page 80

Figure 6.3 VPM M16, Identified Model Verification, 70mph Long Term Response
August 2010 Page 81

Figure 6.4 Comparison of Flight and Model Data (VPM M16, Trim)
August 2010 Page 82

Figure 6.5 Comparison of Flight and Model Data (Montgomerie, Trim)
August 2010 Page 83

Figure 6.6 Comparison of Flight and Model Data, Derivatives, VPM M16
August 2010 Page 84

Figure 6.6 Comparison of Flight and Model Data, Derivatives, VPM M16
(continued)
August 2010 Page 85

Figure 6.7 Comparison of Flight and Model Data, Derivatives, Montgomerie
August 2010 Page 86

Figure 6.7 Comparison of Flight and Model Data, Derivatives, Montgomerie

(continued)
August 2010 Page 87

Figure 6.8 Non-linear Model Comparison with Flight Test (VPM M16,
doublet input, 70mph)
August 2010 Page 88

Figure 6.9 Non-linear Model Comparison with Flight Test (Montgomerie,

Phugoid Input, 50mph)
August 2010 Page 89


Doublet Input, 50mph)
August 2010 Page 90


Doublet Input, 55mph)
August 2010 Page 91

7 Parametric Studies of Configurational Effects
The RASCAL mathematical model was used to examine the sensitivity of the
behaviour of a typical aircraft to a range of design and operational variations. The
design variations in particular are those that are either impractical or potentially unsafe
to examine in actual flight test. A baseline, or default, configuration was selected and
variations applied by modifying the aircraft datafile that defines that configuration.
Simulated differences between the modified aircraft and the default then allow an
assessment to be made of the importance of the design or operational difference.
The objective is to catalogue these changes with a view to determining which ones
are likely to have the greatest impact on the airworthiness of the aircraft.
The model was validated against VPM M16 and Montgomerie-Parsons Two-Place
flight test data (see Section 6), and demonstrated good predictive ability in both cases
so either can be chosen as the basis of the parametric study. However, the default
configuration chosen was the VPM M16 modified for flight test, Figure 5.4 a), since
more robust and detailed airframe aerodynamic data is available from wind tunnel
testing. It also has a much larger tailplane than the Montgomerie-Parsons and so it is
somewhat easier to interpret the scale of this parametric variation.
Table 7.1 shows the range of configurations simulated. The 13 configurations or flight
conditions examined can be broadly classified as airframe geometry; airframe
aerodynamic; rotor system; and operational variations. Engineering analysis then
typically takes the form of comparison of steady (equilibrium, or trim) flight, dynamic
stability and control response. Much can be learned from such comparisons but the
specific form that they take can seem esoteric to the lay reader, hence a particular
presentation of the results is made here. First, the trim comparisons are made against
airspeed for airframe (roll and pitch attitudes, and flying control positions) and rotor
(flapping angles and rotorspeed) parameters. Second, the dynamic stability
comparisons are normally expressed in terms of stability and control derivatives, and
aircraft modes of motion. Derivatives in particular can have little relevance to a pilot
although great insight can be gained from their consideration. The modes of motion
can also appear mathematically abstract. However, they can be converted to the form
required of BCAR Section T dynamic stability compliance demonstration, and hence
compared directly with those boundaries. Finally, the time response to controls, while
adding little to an engineering analysis, can serve to illustrate how the aircraft actually
behaves in a very specific case. Here, a control input is simulated that a pilot might
typically use in demonstrating compliance with BCAR Section T
August 2010 Page 92

Table 7.1 Summary of Simulation Model Parametric Variation
Variation type Parameter Variation Figure No.
Airframe geometry Mass 426, 459 kg 7.17.3
Airframe geometry Vertical c.g. +6 in 7.47.6
Airframe geometry Longitudinal c.g 3 in 7.77.9
Airframe geometry Mast height 3 in 7.107.12
Airframe geometry Propeller thrust line 3 in 7.137.15
Airframe aerodynamics Pod Pod on and off 7.167.18
Airframe aerodynamics Tailplane Tailplane on and off 7.197.21
Airframe aerodynamics Combination Pod and tailplane on 7.227.24

and off
Rotor system Blade section NACA 8H12 and 0012 7.257.27
Rotor system Blade radius 1 foot 7.287.30
Rotor system Spindle offset Nominal, zero 7.317.33
Operational Temperature ISA and ISA+20 7.347.36
Operational Altitude Nominal, 13000 feet 7.377.39
7.1 Mass Variation (Figures 7.1 7.3)

It is instructive to consider one example in detail before examining all other
comparisons. Figures 7.1 7.3 show the trim, stability and time response
comparisons of the default configuration with the high-mass case across the speed
range. Speeds are given in mph, and are indicated or equivalent airspeeds. The default
case is the aircraft as illustrated in Figure 5.4, with pilot but zero fuel giving an all-up
mass of 426 kg. The 'high mass' case is obtained by simulating full fuel (33 kg) giving
an all-up mass of 459 kg. Figure 7.1 suggests that there is little difference between
the two configurations; the heavier aircraft requires a more aft tilt of the main rotor
and a greater rotorspeed as might be expected, but the differences are small. Note
the very small flapping angles required, as might be expected from a rotor system
that has no cyclic pitch applied. The change in stick position is only about 5% of the
available travel, or just over 0.5 in. At high speed there is little difference in the pitch
attitude between the two cases.
The dynamic stability results are shown in Figure 7.2 presented as eigenvalues, (see
Appendix 4 for further explanation). The three modes are present, the short period
mode (cluster of points to the left) the phugoid mode (the cluster of points around the
vertical axis), and the rotorspeed mode (the series of points on the real, or damping
axis). The stable region of the plot is to the left of the vertical axis, and hence we see
that the short period is stable over the whole speed range, whilst the phugoid is
unstable at lower speeds. The BCAR Section T boundaries consistent with T181
period and cycles to damp are also plotted. Any point to the right of, or above, this
boundary is non-compliant. The cluster of points farthest to the left is analogous to
what aeronautical engineers would recognise as the 'short-period' mode. It is
important to recognise that this mode is the one that determines whether or not a
pilot finds aircraft handling qualities to be acceptable or not and, if acceptable, just
August 2010 Page 93

how good they are. This mode will tend to be compliant with Section T requirements,
as evidenced by the fact that it is so far from the boundary. Yet the handling qualities
may be poor if this mode has inappropriate damping and/or frequency. As can be
seen, Section T is designed not to prescribe good short-period handling qualities, but
rather to protect the pilot from unacceptable 'phugoid' mode characteristics, and in
this case the phugoid mode is the one that is on the borderline of Section T
compliance. Points to the right of the vertical axis represent unstable aircraft, but
Section T allows a degree of instability. In this case, the model is predicting that
stability of this phugoid mode improves with increasing speed (the points move right
to left). Both configurations are non-compliant at 30 mph but as speed is increased
the eigenvalues cross the boundary into the compliant region. Note that relative to a
fixed-wing aircraft, this mode is faster (i.e. further up the vertical axis), and this is the
attribute that makes the gyroplane airworthiness more problematic if unstable - hence
the Section T focus on it. Although the short-period mode plays no part in Section T
compliance, it is worthwhile to note that it moves up the plot with increasing speed,
so the mode in effect becomes faster and more responsive.
In general through all of the results presented in this section, the 70 mph cases show
the greatest difference between default and modified configurations (this is indeed
the case for default and heavy cases here), and so are chosen for time history
comparisons presented in Figure 7.3. The exception is the comparison of default with
ISA+20 where the lowest speed case is used, as it is borderline compliance with
Section T and therefore of use in illustrating what the response of such a
configuration might look like. The control input is a 5 sec pulse of the longitudinal rotor
tilt, similar to that which should be used in Section T compliance demonstration.
Again, the default (light) and heavy aircraft show very similar characteristics. These
plots help the reader to visualise the response of the configurations shown in Figure
7.2. As described in Appendix 4, the damping, natural frequency, period and time to
half (or double) amplitude of a particular mode can be calculated from the eigenvalue.
The eigenvalues plotted in Figure 7.2 for 70 mph (m = 426 kg) give the following:
Phugoid = -0.0318 0.0313i Damping Ratio = 0.1597, Period = 31.9s
Short Period = -1.2201 0.188i Damping Ratio = 0.7185, Period = 5.32s
Rotorspeed = -0.5126
The overall response is influenced by all 3 modes. However, as the phugoid is the
longer period mode, its influence is much more obvious.
7.2 Vertical c.g. (Figures 7.47.6)
The dominant effect of this parameter is on the angle of attack stability derivative Mw.
While the derivatives are not being compared in the parametric study, the reader
should appreciate that this term will have powerful effects on both the short-period
and phugoid modes. This can be seen in Figure 7.5 - lowering the c.g. will reduce the
frequency of the short-period mode and, although this mode is deep within the
compliance region, the aircraft will tend to feel more sluggish in manoeuvres which
can adversely affect handling qualities. The phugoid mode frequency, however, is
increased to the point where a number of the speed cases move out of compliance.
Whilst this is only just outside in this case, it is intuitive that any marginally-stable
mode that becomes faster (and this is what increased frequency means) will require
greater pilot attention, compensation and workload.
The time responses, Figure 7.6, particularly the pitch attitude, highlight this clearly.
The case shown is for 70 mph where both cases are Section T compliant, but the case
with the low c.g. has a phugoid mode frequency considerably greater that of the
default case where the c.g. is close to the propeller thrust line (see Figure 7.5 b). The
longer-term (phugoid) response is visually much faster than that of the default case.
August 2010 Page 94

Whilst the VPM M16 simulations illustrate the general trend effect of lowering c.g. on
angle of attack stability and hence these modes of motion, it is the case that of all the
simulations conducted during 15 years of study this aircraft requires a fairly large c.g.
offset to drive it non-compliant. As will be seen in later in Section 10 dealing with the
Air Command series of gyroplanes this is not generally the case, and for the Air
Command a 4 in offset produces a gross change in dynamic stability of short-period
and phugoid modes.
7.3 Longitudinal c.g. Variation (Figures 7.77.9)
Longitudinal c.g. position is an important consideration in the stability and
controllability of fixed-wing aircraft, and the range of c.g. position allowed is, in
relation to the overall size of the aeroplane, quite small indeed. The stability and
controllability of rotorcraft on the other hand is much less sensitive to longitudinal c.g.
position. Rather, it is the longitudinal control required to trim that is the consideration,
where c.g. must not compromise control margins. This is clearly illustrated in Figure
7.8, where an excessive (at least for an aircraft of the VPM M16's size, geometry and
payload) longitudinal c.g. variation of 6 in is seen to have a very small impact on
Section T compliance modes of motion, the short-period mode and the response to
the compliance control input. Trim on the other hand is considerably different (Figure
7.7) with the 6in change in position changing the pitch attitude to trim by 4-5 deg, and
the corresponding rotor tilt by 5-6 deg. A comparison of time responses is shown in
Figure 7.9.
7.4 Rotor Mast Height (Figures 7.107.12)
The height of the mast (to the pivot where the pitch and roll control inputs are
transferred to the hub) was varied by 3 in. Note that in reality this will raise or lower
the c.g. but this effect is not included, hence it is purely the aerodynamic impact of
moving the rotor vertical location that is being examined.
Trim, stability and control response plots show that this parameter has an insignificant
impact on the flight mechanics of the aircraft.
7.5 Propeller Hub Height (Figures 7.137.15)
It should be expected that moving the propeller height on the airframe (assuming that
the c.g. is in a fixed position and that there is no airframe/propeller interference) would
have a broadly similar outcome to that obtained by raising or lowering the c.g. The
trim curves (Figure 7.13) show that this is indeed the case, although the pitch attitude
and rotor tilt angle differences between these two configurations are about 2 deg
whereas the comparisons for vertical c.g. movement showed only a 1 deg difference.
The dynamic stability results (Figure 7.14) are however consistent with lowering the
c.g. - raising the propeller hub and therefore thrustline tends to make the short-period
mode more sluggish and the phugoid to be of higher frequency. This is shown in the
time history (Figure 7.15) where the configuration with a raised propeller thrust line
has faster and more oscillatory response.
7.6 Pod on/Pod off (Figures 7.167.18)
Airframe pods are generally regarded as destabilising, adding as they do a lifting
surface ahead of the c.g., and given that the force is a function of dynamic pressure
it is expected that pod effects will increase with increasing airspeed.
The trim plots (Figure 7.16) show a clear airspeed effect on both the pitch attitude and
the longitudinal tilt of the rotor required to trim. Since the aircraft flies nose-up, i.e.
with a positive angle of attack, the pod is lifting across the speed range and hence
producing a nose-up moment. This is why without the pod the nose is lower and a
more aft rotor tilt is required. The lift force is not, however, significant enough to
unload the rotor by an amount sufficient to reduce the rotorspeed.
August 2010 Page 95

The dynamic stability results, Figure 7.17, show a short-period mode little affected by
having a pod fitted, but a phugoid that perversely is more stable with pod fitted. This
counter-intuitive result is clarified when it is considered that while the pod in isolation
does indeed confer a degree of instability, the nose-up aerodynamic moment is
sufficient to require having the rotor thrust vector placed sufficiently farther aft (i.e.
more stabilising) relative to the c.g. to more than overcome the destabilising
tendency.
The time histories, Figure 7.18, quantify the degree of difference that might be seen
at 70mph and this is consistent with the stability diagram.
7.7 Tailplane (Figure 7.197.21)
The issue of horizontal tailplanes, or stabilisers, is foremost in the mind of the
gyroplane community. While nothing of archival note has come to light, internet
forums and association magazines and newsletters have focussed on whether or not
to have a tailplane (although the majority seem to agree that to include one in a design
is a good thing), where it should be located, the impact of propeller slipstream on
effectiveness, etc. It is certainly the case that the Glasgow University work has, and
continues to, come in for a degree of criticism for promulgating the view in the early
days of the study that tailplanes, while having a role to play, really were insignificant
in terms of providing a safe gyroplane. It is hoped that the results presented here, and
this brief debate, will clarify and justify the position taken by the University team many
years ago.
The results compare a VPM M16 with and without a horizontal tailplane. This is
exactly the role for which an engineering model is ideally suited as to conduct actual
flight trials would be expensive and have airworthiness implications that, in terms of
approval, might be difficult to resolve. The M16 has perhaps one of the largest
tailplanes fitted to a light gyroplane, and the reader concerned with fitment of
anything smaller should consider any technical judgement in that context. Trim is
insignificantly affected by the tailplane - at most there is a 1 deg difference in pitch
attitude and rotor longitudinal tilt at 70 mph; less at lower airspeeds (Figure 7.19).
Dynamic stability shows (Figure 7.20) that the absence of the tailplane has an
enormous impact on those modes that are of short-period, or 'fast'. However, these
modes easily comply with Section T and come nowhere near the boundaries. It is
certainly the case that their character changes and, perhaps because of this,
gyroplane pilots will indeed feel a difference with and without a stabiliser. However,
the phugoid mode is negligibly affected by the presence of a tailplane at low to
medium speeds. Even at 70 mph where one might expect the greatest impact, the
time to half amplitude of the mode is 7 sec without the tailplane, and 14 sec with. The
latter is still a well-damped mode in the context of aircraft phugoid motions; both with
and without the tailplane configurations easily satisfy Section T compliance; and at
low speed there is no benefit at all, the phugoids for both configurations being nearly
identical. The counter-intuitive result that the phugoid is less damped with the
tailplane, is specific to the VPM and is a consequence of the pod generating such a
significant lift force at high speed due to the nose-up pitch attitude without the
tailplane, that the main rotor thrust vector is placed in a more favourable (stabilising)
location to trim this increased moment. If the aircraft flew nose-down at high speed,
the pod would generate a downforce and the result is reversed. The conclusion is
reached that in terms of Section T compliance and the dynamic stability of the mode
that is likely to be marginally stable (the phugoid), a tailplane offers no overall benefit.
Remember, that is not to say that a pilot will not feel that his aircraft has better
handling qualities with a tailplane fitted, as the short-period mode comparisons
testify.
August 2010 Page 96

Other aircraft with smaller stabilisers are likely to experience less benefit. The
fundamental problem, as illuminated by the wind tunnel results, is that the lift-curve
slope is weak at around 2 per radian. Typical helicopter values are 3-3.5, an aeroplane
overall (including fuselage etc.) may be around 5.5, and finite wing theory limits this
parameter to per radian. Combine this with the very short moment arm of the pusher
configuration and it can be seen that very large tailplanes will be required and, as the
results above show, even then it will only be the short-period mode that is affected.
There is some debate in the gyroplane community over the optimum location for a
stabiliser in relation to the propeller wake. The wind tunnel chapter discusses this
point and indicates that, for the VPM M16 configuration, there is little benefit to be
obtained from the propeller wake. If the tailplane can benefit from the full value of the
slipstream velocity by being appropriately placed, simple momentum theory allows an
assessment to be made of its potential effectiveness. At 30 mph, the tailplane would
experience an airspeed of 43 mph, while at 70 mph it would be 73 mph. This assumes
a uniform distribution of induced velocity.
The time histories, Figure 7.21 confirm that there is little difference in the response
of the two configurations to the control input at 70 mph, where the difference in the
phugoid mode is greatest. (Note that this input fails to excite the short-period mode,
indeed it is designed to avoid exciting it, and hence the gross changes seen in the
compliance diagram are hidden in the time response plots).
7.8 Pod and Tailplane On/Off Figures 7.227.24)
Simulation of pod and tailplane removed shows little difference in the trim variables
relative to the standard aircraft (Figure 7.22) or in the phugoid mode, and hence
compliance with Section T is not influenced by airframe aerodynamics. However, the
short-period mode is significantly different (Figure 7.23) much more so than in any of
the parametric variations seen previously. Further, the small control input in this case
produces large excursions from trim where the linearised analysis that underpins the
stability diagram is of dubious value, as seen from the time response, (Figure 7.24).
7.9 Blade Section (Figures 7.257.27)
The VPM M16 blade section is cambered, while the NACA 0012 is symmetrical and
was for a number of years the section of choice for helicopter rotor blades. The
gyroplane community fly with a number of different blade sections but this
comparison was chosen because the 0012 is dissimilar to the majority of sections in
use; it is intended therefore as an 'extreme' comparison. The trim results (Figure 7.25)
show fairly small differences in rotor tilt and airframe attitude (no more than about 1
deg). The rotor with the 0012 section, however, turns some 40 rpm faster.
The dynamic stability results (Figure 7.26) show for the 0012 a short-period mode
similar at low speeds but more lightly damped at high speed. The phugoid mode is of
similar frequency to the standard aircraft, but is slightly less well damped. This is clear
from the time histories (Figure 7.27).
7.10 Disc Loading (Figures 7.287.30)
Disc loading was varied by reducing blade radius by 0.3 m; and increasing it by the
same amount relative to the standard aircraft. This results in very considerably
different rotor and blade loadings. The small rotor has a blade loading in these
simulations of 248.4 kg/m2, and a disc loading of 9.1 kg/m2. By contrast the large
rotor blade and disc loadings are 214.6 kg/m2 and 6.8 kg/m2. The scale of the
following differences should therefore be seen in this context.
The trim results (Figure 7.28) show, as might be expected, a large difference in the
rotorspeed of 60 rpm, and a 2 deg difference in pitch attitude; the more lightly loaded
rotor operates at a lower angle of attack.
August 2010 Page 97

The dynamic stability results (Figure 7.29) show that the aircraft with the smaller
loadings has better short-period mode damping, but poorer phugoid damping to the
extent that it only becomes Section T compliant above 40 mph. The better short-
period mode damping is due to the greater pitch damping which comes from a lower
rotorspeed (the disc will precess more in response to pitch rate disturbances). The
time histories (Figure 7.30) quantify the magnitude of the difference in phugoid
damping, showing that the response is more oscillatory with the larger and therefore
more lightly-loaded rotor.
7.11 Spindle Offset (Figures 7.317.33)
All light gyroplanes fly with what are basically slight variations in the design of the
Bensen rotor head. A feature of the design is that the teeter mechanism is not
mounted directly above the rotor hub pivot point, but is displaced aft by a few
centimetres. This endows the aircraft with a degree of stick-free stability when
disturbed by, say, atmospheric changes. All the analysis conducted here assumes
stick-fixed, and it is the case that stick-free the aircraft is more stable. Trim
differences (Figure 7.31) are limited to pitch attitude and rotor tilt, while Section T
compliance (Figure 7.32) and time responses (Figure 7.33) both confirm that this
parameter has a negligible impact on the dynamic stability and response of the
aircraft.
7.12 Environmental Parameters - Density Altitude (Figures 7.347.39)
The density altitude is affected by temperature and the height at which the aircraft is
flown. The results for ISA+20 deg C (Figures 7.347.39) show that the density
changes resulting from high temperature at low altitude have an insignificant impact
on trim, stability or control response.
Flight at 13,000 feet, however, producesx more variability. Overall the results hint at

proximity to retreating blade stall. The trim curves (Figure 7.37) differ by only a degree
or so, but the dynamic stability (Figure 7.38) shows a short-period mode highly
sensitive to flight at altitude. This can be attributed directly and mathematically to the
influence of true air speed (rather than indicated air speed) at altitude. The phugoid
mode is slightly less stable across the speed range. The time responses (Figure 7.39)
confirm that there is little difference in the long-term response, but odd behaviour that
appears to be non-linear in nature (most likely the model capturing dynamic stall
effects) afflicts the short-term response to the control input.
Consideration of these results, particularly in the context of design advice or
airworthiness considerations in general, requires careful interpretation and insight.
For example at face value the design or operational parameter that has the greatest
relative impact is a change in disc loading (rotor blade radius). A relatively small
change (0.3 m) drives a considerable change in the control angles and rotorspeed
required to trim the aircraft, as well as the damping of the phugoid mode (and hence
Section T compliance). However, damping alone is only one of the parameters
considered when assessing dynamic stability; frequency is also important and while
pilots can cope with a degree of phugoid mode instability, that ability is compromised
if the mode becomes faster (i.e. the period becomes shorter). The parametric study
shows that the relative location of the propeller thrust line in relation to the centre of
gravity is the only parameter investigated that produces a significant variation in the
period of the phugoid mode for a relatively small change in distance - Figure 7.14 b)
for example shows that the frequency of the mode can be doubled, while Figure 7.15
clearly shows a livelier pitch attitude response with a thrust line only 3 in higher than
the datum aircraft. Of particular note is that this change occurs across the speed
range. Addenda 1 and 2 provide independent experimental validation of the
importance to dynamic stability of the relationship between propeller thrust line and
c.g. location.
August 2010 Page 98

The tailplane effect on Section T compliance, by comparison, is limited. It fails to

change the frequency of the phugoid across the speed range; improves the damping
only with speed; but has a gross effect on the short-period mode (Figure 7.20 a)). The
Section T compliance mode is therefore negligibly affected by a horizontal tailplane at
low speeds (consistent with the poor aerodynamic efficiency identified in tunnel test),
whereas propeller thrust line location has an impact across the speed range. Heresay
evidence from the gyroplane community however is that tailplanes are significant
components for the improvement in gyroplane stability. This perception is probably
associated with their impact on the short-period mode seen in Figure 7.20 a) the
short-period mode is not an issue for compliance demonstration, but it is the mode
that pilots excite to manoeuvre their aircraft and hence the mode that determines
how handling qualities are perceived.
CAA Comment
CAA accepts that closer alignment of the CG and the thrustline is a sensible design
aim to achieve pitch dynamic stability (phugoid mode) and whilst conclusive Flight
Test evidence has supported this and directly led to the issue of MPD 2005-08 for
single seat gyroplanes, other flight tests provide evidence of a gyroplane design
that achieves good stability whilst being outside of the 2 inch criteria. CAA Flight
Test Specialist qualitative assessment implies that pitch dynamic stability may not
be solely a function of CG/Propeller Thrust alignment for all types of Gyroplane.
The report also concludes that horizontal tailplanes are largely ineffective in
improving the long term response of pitch dynamic stability (phugoid mode). This
is the result of studies primarily on narrow tandem cockpit enclosures. CAA Flight
Test Specialist qualitative evaluation of the effects of a horizontal tailplane on
a single side-by-side configuration gyroplane type indicated a degree of
improvement in the phugoid characteristics at higher speed.
Figure 7.1a Trim Comparison for Mass Variation (Airframe Parameters)
August 2010 Page 99

Figure 7.1b Trim Comparison for Mass Variation (Rotor Parameters)
August 2010 Page 100

Figure 7.2a Comparison of Stability Modes (Mass Variation)
Figure 7.2b The Phugoid Mode (Mass Variation)

Figure 7.3 Aircraft Response to 5s Pulse of Longitudinal Rotor Tilt - Phugoid

(Mass Variation)

Figure 7.4a Trim Comparison for Vertical c.g. Variation (Airframe Parameters)

Figure 7.4b Trim Comparison for Vertical c.g. Variation (Rotor Parameters)

Figure 7.5a Comparison of Stability Modes (Vertical c.g. Variation)
Figure 7.5b The Phugoid Mode (Vertical c.g. Variation)


(Vertical c.g. Variation)

Figure 7.7a Trim Comparison for Longitudinal c.g. Variation

(Airframe Parameters)

Figure 7.7b Trim Comparison for Longitudinal c.g. Variation

(Rotor Parameters)

Figure 7.8a Comparison of Stability Modes (Longitudinal c.g. Variation)
Figure 7.8b The Phugoid Mode (Longitudinal c.g. Variation)


(Longitudinal c.g. Variation)

Figure 7.10a Trim Comparison for Mast Height Variation (Airframe Parameters)

Figure 7.10b Trim Comparison for Mast Height Variation (Rotor Parameters)

Figure 7.11a Comparison of Stability Modes (Mast Height Variation)
Figure 7.11b The Phugoid Mode (Mast Height Variation)


(Mast Height Variation)

Figure 7.13a Trim Comparison for Propeller Thrust Line Variation


Figure 7.13b Trim Comparison for Propeller Thrust Line Variation

(Rotor Parameters)

Figure 7.14a Comparison of Stability Modes (Propeller Thrust Line Variation)
Figure 7.14b The Phugoid Mode (Propeller Thrust Line Variation)


(Propeller Thrust Line Variation)

Figure 7.16a Trim Comparison for Pod On/Pod Off (Airframe Parameters)

Figure 7.16b Trim Comparison for Pod On/Pod Off (Rotor Parameters)

Figure 7.17a Comparison of Stability Modes (Pod On/Pod Off)
Figure 7.17b The Phugoid Mode (Pod On/Pod Off)


(Pod On/Pod Off)

Figure 7.19a Trim Comparison for Tailplane On/Tailplane Off


Figure 7.19b Trim Comparison for Tailplane On/Tailplane Off (Rotor Parameters)

Figure 7.20a Comparison of Stability Modes (Tailplane On/Tailplane Off)
Figure 7.20b The Phugoid Mode (Tailplane On/Tailplane Off)

Figure 7.21 Aircraft Response to 5s Pulse of Longitudinal Rotor Tilt Phugoid

(Tailplane On/Tailplane Off)

Figure 7.22a Trim Comparison for Pod and Tailplane On/Pod and Tailplane Off

Figure 7.22b Trim Comparison for Pod and Tailplane On/Pod and Tailplane Off
(Rotor Parameters)

Figure 7.23a Comparison of Stability Modes

(Pod and Tailplane On/Pod and Tailplane Off)
Figure 7.23b The Phugoid Mode (Pod and Tailplane On/Pod and Tailplane Off)


(Pod and Tailplane On/Pod and Tailplane Off)

Figure 7.25a Trim Comparison for Different Blade Sections


Figure 7.25b Trim Comparison for Different Blade Sections (Rotor Parameters)

Figure 7.26a Comparison of Stability Modes (Different Blade Sections)
Figure 7.26b The Phugoid Mode (Different Blade Sections)


(Different Blade Sections)

Figure 7.28a Trim Comparison for Disc Loading Variation (Airframe Parameters)

Figure 7.28b Trim Comparison for Disc Loading Variation (Rotor Parameters)

Figure 7.29a Comparison of Stability Modes (Disc Loading Variation)
Figure 7.29b The Phugoid Mode (Disc Loading Variation)


(Disc Loading Variation)

Figure 7.31a Trim Comparison for Spindle Offset Variation (Airframe Parameters)

Figure 7.31b Trim Comparison for Spindle Offset Variation (Rotor Parameters)

Figure 7.32a Comparison of Stability Modes (Spindle Offset Variation)
Figure 7.32b The Phugoid Mode (Spindle Offset Variation)


(Spindle Offset Variation)

Figure 7.34a Trim Comparison for Temperature Variation (Airframe Parameters)

Figure 7.34b Trim Comparison for Temperature Variation (Rotor Parameters)

Figure 7.35a Comparison of Stability Modes (Temperature Variation)
Figure 7.35b The Phugoid Mode (Temperature Variation)


(Temperature Variation)

Figure 7.37a Trim Comparison for Altitude Variation (Airframe Parameters)

x

x

Figure 7.37b Trim Comparison for Altitude Variation (Rotor Parameters)

Figure 7.38a Comparison of Stability Modes (Altitude Variation)
Figure 7.38b The Phugoid Mode (Altitude Variation)


(Altitude Variation)

8 Review of BCAR Section T

8.1 Inroduction
Although there is now a broad literature on the subject of gyroplanes, there is little
substantive, validated data available to use for making informed judgement to guide
the analysis or development of BCAR Section T. There is an extensive body of
experience in the UK, but this resides largely in the knowledge of few individuals who
have formed their opinions as pilots or homebuilders, and without the benefit of a
rigorous and objective analysis of gyroplane aerodynamics.
The work described in this report has produced an extensive database of results from
mathematical modelling, wind tunnel and simulation experiments. In addition, flight
tests with two fully-instrumented gyroplanes have been used to generate insight into
actual aircraft behaviour by performing highly specialised control inputs and then
processing the resulting data using contemporary analysis techniques.
These data, whether derived from simulation or flight tests, are unique. As such, they
provide a benchmark for testing BCAR Section T. Perhaps the most important aspect,
however, is that these data are objective and quantitative. Opinion, experience and
knowledge gained by pilots and homebuilders is essentially subjective and qualitative.
It is argued then, that these data can be viewed with a greater degree of confidence
than the existing body of experience, and are therefore ideally suited for use in
assessing Section T.
8.2 Background
8.2.1 A review of Section T [99] was conducted in 1997, with specific reference to the
material affected by the study of the Aerodynamics of Gyroplanes, [98]. Accordingly
it focused on the following:
a) Subpart B Paragraph T23 "Load distribution limits";
b) Subpart B Paragraph T173 "Static Longitudinal Stability";
c) Subpart B Paragraph T181 "Dynamic stability".
That review was simply intended as a survey of Section T in the context of new
information generated subsequent to the Code's initial development from Section S,
for which only sparse levels of handling and stability guidance was available. The only
substantive, quantitative information that had then been generated was from the
modelling and testing described earlier in this report. (No further literature review was
conducted to determine if other pertinent studies had generated complementary
results. The current, up-to-date review in Section 2 confirms that this remains the
case).
8.2.2 Item c) above is directly relevant to handling qualities and, as a result, other material
was used in support of the review, [100 - 102]. The 1997 review made the following
recommendations:
a) It is recommended that Subpart B Paragraph T23 "Load distribution limits" be
modified to specify that weight and balance envelopes be determined that
encompass fore-and-aft and vertical centre-of-mass positions for all possible
loading configurations. For clarity this recommendation should be expressed as a
requirement for the propeller thrust line to pass through the c.g. envelope defined
by limiting loading conditions; or that the normal distance between propeller thrust
line and c.g. be no more than 2 inches.
b) It is recommended that Subpart B Paragraph T173 "Static Longitudinal Stability" be
amended to require stick position as well as force gradients be demonstrated.

c) It is recommended that Subpart B Paragraph T181 "Dynamic stability" be amended

to state that short-period oscillations be "damped" and not "heavily damped", unless
"heavily" can be further quantified.
8.2.3 These recommendations were adopted and are incorporated in BCAR Section T,
[103], as follows:
a) T23 (a) and AMC T23;
b) T173 (a);
c) AMC T181.
The analysis that led to these recommendations is developed in the following section.
8.3 Analysis
Note that paragraph references in this section refer to the original CAA BCAR Section
T paper, [98].
8.3.1 Subpart B Paragraph T23 "Load distribution limits"
Paragraph T23 interprets load distribution limits purely in terms of the conventional
(for gyroplanes) hang check. This is done, effectively, to determine that known control
limits are not compromised. There is no requirement for a conventional weight and
balance schedule to be determined. Aeroplanes require this to be done primarily to
ensure that the longitudinal position of the centre-of-mass (known to have a direct
impact on static and dynamic stability) is kept within set limits. Simulation has shown
that longitudinal centre-of-mass position does not have the same impact on dynamic
stability for gyroplanes as it does for aeroplanes so, at first inspection, it may seem
reasonable not to require a conventional weight and balance schedule.
However, a theoretical study of gyroplane longitudinal stability, [62], indicates that the
vertical location of the centre-of-mass of the aircraft, in relation to the propeller thrust
line, may have a significant impact on dynamic stability. Analysis of the flight test data,
[104], produced evidence to support this theory. Simulation of the impact of varying
this parameter indicates that the phugoid mode can be destabilised, resulting in
unstable rotorspeed response to control inputs, controls fixed. Results in Section 7
show significant movement (towards non-compliance) of the phugoid mode with
changes to propeller thrust line and vertical c.g. location. The mechanism for this
effect is described in Refs. 62 and 104. Strictly speaking, if the propeller thrust line is
canted at some angle to the longitudinal axis of the aircraft, it is the normal distance
between thrust line and centre-of-mass, i.e. the propeller thrust moment arm, that is
the significant parameter.
There is further uncorroborated evidence that this parameter may be an issue in
gyroplane stability. The Air Command accidents, [105], are largely unexplained but are
thought to have their root in handling problems. The final fatal Air Command accident
in the UK to G-BOVP joins this category, although the community seems to think that
structural issues and poor post-accident repair had a role to play, despite the AAIB
report to the contrary. In 1997 weight and balance information were available for two
Air Commands. The owner of one, unregistered in the UK, wrote to the CAA in 1992
stating that the centre-of-mass for his aircraft was 11.5 in below the propeller thrust
line. In [62] an early version of the VPM M16 simulation became unstable when the
centre-of-mass was only 3.9 in below the propeller thrust line. Data for Modac 503 G-
BMZA, dated May 1993, shows the propeller thrust line passing through a calculated
centre-of-mass envelope, that was very small. Heresay evidence suggests that this
aircraft represents a significant improvement in terms of handling qualities over other
Air Commands.

These data were the evidence used in 1997 to propose that the relationship between
propeller thrust line and centre-of-mass position can have an impact on longitudinal
stability and handling qualities. This parameter can tend to de-stabilise the phugoid
mode: whether or not this mode actually becomes unstable depends on other factors
such as the drag damping (rotor as well as fuselage), pod and tailplane effectiveness
and airspeed. If Paragraph T23 is modified to require determination of the fore-and-aft
and vertical centre-of-mass positions, then the difficulty arises as to what quantitative
information to supply as interpretative material.
The normal distance between the propeller thrust line and the centre-of-mass, i.e. the
moment arm about which the propeller thrust acts, was calculated to be 2.44 in at
minimum weight (zero fuel), and 3.36 in at maximum weight (full fuel) for the VPM
M16 used for flight trials, when modified with instrumentation and with a pilot in the
rear seat. These distances result in the propeller thrust line being above the centre-
of-mass. The VPM M16 tested had a marginally stable phugoid oscillation, (see Figure
6.3) so this seems consistent with the study in [62] where the configuration became
unstable when the centre-of-mass was 3.9 in below the propeller thrust line.
It would therefore seem reasonable to modify Paragraph T23 to require determination
of a centre-of mass envelope for all possible loading configurations. It was suggested
that material could be included to specify that the propeller thrust line ought to pass
through this envelope, and a conservative figure (in an effort to capture "safe" values
for all types addressed by Section T) is that it should not be a normal distance of more
than 2 in from the centre-of-mass at any loading condition. This is indeed the course
of action that was adopted by the CAA.
It is fair to say that this recommendation came in for a very considerable degree of
criticism from the UK gyroplane community. However, two studies conducted since
then serve, in retrospect, to confirm the validity of that original analysis. First, a weight
and balance study of two Air Commands (single and twin-seat) was conducted for the
CAA as part of a wider simulation study of the type [106], which confirmed that the
type had a low centre-of-mass position. Second, and arguably more important, a
weight and balance survey of five single-seat machines [107] was carried out in
support of the CAA flight tests, [108]. These data validate the earlier
recommendation, in that only one of the five aircraft tested satisfied the Section T
handling criteria, and it was the aircraft modified to meet the proposed 2in propeller
thrust line/c.g. offset limit. For clarity, Reference 107 and 108 are included in this
document as Addendum 1 and Addendum 2 respectively. Note that the tests
reported in Reference 108 were conducted separately and independently (and after)
the flight tests reported in Reference 107. They were conducted to validate the flight
test conclusions.
8.3.2 Subpart B Paragraph T173 "Static Longitudinal Stability"
This paragraph expressed static stability in terms of stick force. For completeness, it
was recommended that stick position should be included as well. Stick position may
be easier to measure accurately and repeatably given the rudimentary nature of light
gyroplane trim systems.
8.3.3 Subpart B Paragraph T181 "Dynamic stability"
This paragraph is perhaps the most difficult to deal with, for a variety of reasons. It
offers a particular challenge in seeking to find easy to apply criteria that are rigorous,
complete and consistent. The emphasis on oscillation characteristics may indicate
that the intention of the criteria is to deal with the potential for pilot-induced-oscillation
(PIO). PIO is difficult to predict, particularly from elementary dynamic response criteria
of the type given in T181. Other issues, such as pilot control strategy, task demands

and aircraft frequency response characteristics all have a role to play. Notwithstanding
this, T181 can be examined in the context of results from flight test experiments
conducted on G-BWGI and the existing literature on handling qualities prediction from
dynamic response criteria.
8.3.3.1 Short-period oscillations
T181 (a) states that any short-period oscillations should be "heavily damped". The
interpretative material places this in context, that an oscillation with a period of less
than 5 sec should damp to half amplitude in not more than one cycle. Consider the
limiting case: a 5 sec oscillation, damping to one half amplitude in one cycle, i.e. 5 sec.
This gives an oscillation with a damping ratio of 0.11. Critically-damped systems have
a damping ratio of 1, and "well-damped" systems, 0.35. The interpretative material is
therefore inconsistent with the requirement "heavily damped". Even an oscillation
with a 1 sec period, damping to half-amplitude within 1 sec, will have a damping ratio
of 0.11, still not "heavily" damped (generally, a damping ratio greater than about 0.6
would indicate "heavy" damping). If the interpretative material were to be modified to
read "...should damp to one half amplitude in not more than half a cycle", then the
damping ratio increases to 0.22 in these limiting cases, still not even "well" damped.
Any further modification to the criteria will then result in compliant aircraft having
dynamic characteristics that could call into question the test pilot's ability to identify
any oscillation of this sort quantitatively.
Additional information may help in deciding what, if anything, should be done with
these requirements. First, the VPM M16 G-BWGI, as identified, has the short-period
mode characteristics described in Table 8.1. The aircraft easily satisfies T181
interpretative material at both speeds but the damping, particularly at 70 mph, cannot
be described as "heavily damped". Note that MIL-F-8785C, a now obsolete fixed-wing,
essentially fighter, handling criteria states that damping ratios of 0.35 and above are
required for "satisfactory" handling qualities, 0.25 being the lower limit for Level 2
("acceptable") handling qualities, [101]. The relevance of MIL-F-8785C to gyroplanes is
not clear, however.
However, the undamped natural frequency is also important and, combined with the
damping requirement MIL-F-8785C criteria, predicts VPM M16 handling qualities to
be borderline poor/unacceptable. Yet it satisfies the criteria in Section T. Further, in
previous performance and handling assessments, the aircraft is well-regarded, [109].
There seems little point in changing the interpretative data without any further
handling qualities research specific to the gyroplane.
The review therefore recommended that the word 'heavily' be removed from the
description of the short-period damping requirement (T181), and this was accepted.
Table 8.1 Identified G-BWGI Short-period Mode Characteristics
undamped natural frequency

speed (mph) damping ratio
(rad/s)
30 0.75 0.59
70 0.38 1.40
8.3.3.2 Long-period oscillations

The interpretative material for longer-period oscillations was contained in App. T 181
(b) and (c), the former dealing with stable and the latter with potentially unstable,
oscillations. Identified results for G-BWGI are given in Table 8.2. Note that this
oscillation is very lightly damped even in a machine that is well regarded in the
community. These data are converted in terms of Section T in Table 9.

Table 8.2 Identified G-BWGI Long-period Mode Characteristics
undamped natural frequency

speed (mph) damping ratio
(rad/s)
30 0.042 0.58
70 0.042 0.40
App. T181 (c) is then the relevant section, and the aircraft passes, as the requirement
is only for the oscillation to be damped. However, at 30 mph, the aircraft only just
misses App. T181 (b) being the relevant section, where it would fail, as the number
of cycles to half amplitude is greater than 2. Again, previous flight test work does not
indicate poor handling qualities as a result of long-period oscillations. (It is interesting
to note that in this previous work, [109], the test pilot used his judgement and skill to
measure phugoid characteristics. Interpretation of this previous data gives a damped
natural frequency of 0.5 rad/s, and a damping ratio of about 0.1-0.2, quite different to
that identified using parameter estimation methods it is argued that this is an
indication of how challenging the compliance demonstration against Section T can
be).
Table 8.3 Identified G-BWGI Long-period Mode Characteristics, Section T
speed (mph) No. of Cycles to Half Amplitude Period (s)
30 2.6 10.9
70 2.6 15.6
Some previous, now dated work, [100] suggests that phugoid damping ratios of
0.042, as identified for G-BWGI, will give marginally acceptable, tending to poor, pilot
opinion. However, these data were for a period of 50 sec, i.e. quite slow, whereas the
VPM M16 has a much faster phugoid oscillation than this. Intuitively one might expect
an even higher level of damping is required for satisfactory or acceptable handling
qualities.
8.4 Comparison between VPM M16 and Montgomerie-Parsons
The foregoing analysis was developed at a time when only VPM M16 flight test data
were available with which to evaluate the compliance demonstration criteria
proposed for BCAR Section T. The aircraft complies with the dynamic stability
requirements and, since the heresay opinion of the aircraft is good, it can be taken
that the boundaries are in the right place. However, corroboration of this assessment
requires comparison of other types, and only the Montgomerie-Parsons Two-Place
G-UNIV has been subjected to a rigorous flight test programme that allows dynamic
characteristics to be identified mathematically from the flight test data.
The Section T compliance boundaries given in T181 are best rendered graphically as
shown earlier, rather than as a narrative as given in BCAR Section T. The diagram is
by now familiar, and is shown in Figure 8.1. Note that as the x-axis is "damping ratio"
and not "damping" as in previous plots presented in this report, the stable region is
now to the right of the vertical axis (see Appendix 4). The aircraft have three
longitudinal modes: short-period, long-period (more colloquially known as the
phugoid), and rotorspeed. The data are for two speeds, 30 and 70 mph IAS in the case
of the VPM, 40 and 70 mph IAS in the case of the Montgomerie-Parsons. Arrows
against each mode show the progression from low to high speed.

As described above the VPM M16 meets the compliance demonstration criteria. Only
the long-period (phugoid) mode is close to the boundary, but is nonetheless
compliant. The Montgomerie-Parsons on the other hand fails compliance by virtue of
the fact that the long-period mode is placed too far to the left at low speed. This result
is broadly consistent with pilot opinion of the aircraft, which was a challenge to fly
when conducting specialized inputs such as steps, doublets and especially frequency
sweeps where the inadequacy of the handling required an alternative approach to be
adopted by the pilot.
A formalized approach correlating these data with Cooper-Harper rating of each
aircraft in applied flying tasks would be required to confirm this assessment. Such
flying never featured as part of any flight test campaign. Nonetheless, the results tend
to confirm the validity of the criteria for predicting handling qualities in the broadest
sense.
8.5 Analysis of BCAR Section T Chordal Balance Requirement

A study of the aeroelastic properties of gyroplane rotors has been undertaken in
parallel with the work reported here [78], and a full report on this is provided as
Addendum 3 to this report. From this work it is clear that aeroelastic instabilities do
exist. Occurrence of a type of flutter that is unique for autorotating rotors was
predicted by the model developed, both during axial descent in autorotation and
autorotative forward flight. This aeroelastic instability is driven by blade pitch / bending
/ rotor speed coupling and differs from both flutter of a helicopter rotor and flutter of
a fixed wing. The instability results in catastrophic decrease of the rotor speed and
significant increase of speed of descent. Given this result, it is considered prudent to
verify (or otherwise) that the requirements for chordal balance currently in BCAR
Section T are appropriate. Specifically, BCAR Section T (light gyroplanes) paragraph
T 659 states:
Mass balance
b) The chordwise balance of the blades must be at, or forward of, the 25% chord.
The chordwise balance of each blade in a pair must be the same, or within a
tolerance to be agreed with the CAA.
The research looked in detail at this requirement in the light of the identified
aeroelastic instability issues. Addendum 3 concludes that BCAR T 659 provides a safe
rotor, in terms of pitch flap flutter and divergence, for conventionally designed light
gyroplanes with teetering rotors. In addition, the offset gyroplane hub was shown to
be slightly stabilizing in forward flight. The predicted flutter frequency for low stiffness
control systems is around 1/rev. This low frequency, and the fact that the dominant
(low stiffness) degrees of freedom are longitudinal and lateral disc tilt, means that any
flutter phenomenon arising from an aft c.g. will manifest itself as a 1/rev varying disc
tilt.
8.6 Discussion
The essential fact is that the technical ability of the gyroplane community and the test
and analysis facilities available to them, may severely compromise or inhibit the
collection of meaningful, or even correct data with which to demonstrate compliance.
Any recommendations that might be made in respect of Section T, therefore, may be
only of academic consequence, and do little to enhance Section T and therefore
airworthiness.

For example, the previous, largely subjective analysis conducted by the CAA's
approved gyroplane test pilot, [109], estimated short-period frequencies of the order
of 2.5 Hz, or 15 rad/s. The parameter estimation exercise conducted on G-BWGI
identifed short-period frequencies, depending on airspeed, of between 0.5 and 1.5
rad/s, i.e. a full order of magnitude less. The phugoid frequency estimated previously
was 0.5 rad/s at 70 mph. The exercise conducted on G-BWGI identified 0.4 rad/s so,
although test pilot judgement was better in the latter case, it was still in error by 20-
25%. With regard to phugoid mode damping, the identification exercise indicates that
the damping is less than half of that subjectively measured in the previous exercise.
Thus, locating the c.g. appropriately relative to the propeller thrust line offers the best
prospect for acceptable handling qualities across the speed range.
Relative to appropriate handling qualities criteria in the literature, the VPM M16 fares
badly, with Level 2/3 (acceptable/poor) predictions of short-period and phugoid mode
characteristics, [100, 101]. This is at odds with subjective and heresay evidence; the
aircraft handling qualities are regarded as an enhancing feature of the type, particularly
for the training role. Despite this inconsistency, the aircraft possesses longitudinal
dynamics that, for other aircraft performing other tasks, would be unacceptable.
However, there is some consistency with Section T in regard to long-period (phugoid)
oscillations, where the aircraft is only marginally acceptable at 30 mph. Given that the
VPM M16 may "set the standard" (it certainly, by virtue of the flight test programme
conducted on it, benchmarks the understanding of gyroplane stability and control), it
is difficult to see how other types will then satisfy Section T's stability requirements.
In essence, based on experience with the VPM M16, other gyroplanes of similar
configuration and with two-bladed teetering main rotors, will find demonstrating
compliance with the stability criteria of Section T a challenge.

Arrows indicate increasing airspeed
Figure 8.1 Comparison of Flight Identified Modes. VPM and Montgomerie

9 Assessment of Rotor Blade Torsion Effects
9.1 Introduction
The mathematical modelling of rotor blade behaviour and the resulting forces and
moments transmitted to the aircraft employed for the studies reported here has
embodied a rigid, rather than flexible, representation of the blade. The blades are of
course free to flap or teeter, but outboard of the hinge or teeter bolt are regarded as
rigid. This is consistent with contemporary approaches to flight modelling of
rotorcraft. It is rare indeed to find a rotorcraft flight mechanics model that includes
blade flexibility.
The phenomena that arguably have the greatest impact on aircraft behaviour are
those that generate changes in angle of attack at a blade element. These include
aircraft motions, blade flap velocity, inflow and control inputs. However, if a blade is
flexible then the bending of the blade can further change its angle of attack and hence
the loads generated. Bending can certainly produce such effects but tends to have a
small impact by comparison with elastic twisting, or torsion, of a blade. This produces
a direct geometric change in a blade element angle of attack as opposed to that from
bending which results from small fast vertical movement of the blade. The total blade
motion can then be seen to be a complex and complicated mix of rigid blade motions
with elastic deformations in torsion and bending superimposed.
It is possible to couple a simplified representation of torsion with the rigid blade
model, but it must be appreciated that it fails to couple the dynamics of torsion such
as the time-varying response of the blade's twist with applied load. Instead the blade
is assumed to twist instantaneously to an appropriate angle given by the applied load,
resulting in a quasi-steady rather than truly dynamic response. However, the
relationship between load and twist is predicated on the results from the quite
separate and sophisticated model of torsion that, for its part, does not capture the
flight behaviour of the whole aircraft. There is nothing in the technical literature that
indicates this has been attempted previously but, in principle, it can be expected that
this approach offers an appreciation of at least the gross effects of blade torsion on
flight mechanics.
9.2 Background
Interest in the effects of blade flexibility as it affects gyroplane rotors is driven by two
principle questions: first, can bending be sufficiently large during benign manoeuvres
as to cause airframe/rotor strikes or propeller/rotor strikes; secondly, can torsion as a
consequence of loading cause a rotor to speed up or slow down sufficiently to cause
a flight safety problem. Relatively early studies on the first of these concerns has
taken place, [110, 111]. Interest in the torsion problem is fairly recent and was driven
by the CAA interest in Wallis gyroplane rotors which anecdotal and circumstantial
evidence suggest are torsionally 'soft' and therefore likely to strongly couple
rotorspeed and elastic twist effects, perhaps beneficially, [112]. However, the very
limited application of blade flexibility models in helicopter flight mechanics modelling
does not suggest that blade elasticity has a seminal impact in the fidelity of the
modelling, [113]. In this latter study an attempt to resolve poor modelling of cross-
coupling by detailed finite-element modelling of blade flexibility, coupled with a more
complex wake model, produced at best inconclusive results with regards the
modelling improvements obtained. This modelling, and the comparison with flight
test data, benefited from a comprehensive flight test database obtained using a fully-
instrumented UH60 helicopter.

9.3 Modelling
The implementation of blade elastic deformation is enabled by modifying the static
twist distribution (normally zero for a gyroplane blade which is typically untwisted).
This elastic twist is derived from a blade torsion mode 'shape', and this is shown in
Figure 9.1. This shape is the twist adopted by the blade under any load. To derive the
actual twist at any position, this shape is correlated with the deflection at the tip, itself
a function of the blade load. Calculation of a blade load is an integral part of any flight
mechanics code, rigid blade or otherwise, and is therefore readily available in the flight
model.
Figure 9.2 shows the relationship between the tip torsional deflection and the total
load acting on the blade. So, the torsional deflection (twist) from elastic deformation
under load is obtained by multiplying Figure 9.1 by 9.2, scaled to the actual load
calculated by the flight model.
The model of course requires data to configure the blade as a specific type. In this
case a McCutcheon blade was tested to derive the appropriate characteristics, [112].
The McCutcheon blade is used on the Montgomerie-Parsons gyroplane, so the
simulation results are specific to that aircraft.
9.4 Results
Results are presented in the same form as that previously used for the parametric
study and model validation. Figure 9.3 a) shows the airframe parameters in trim flight
across the level flight speed range, Figure 9.3 b) the corresponding rotor parameters.
Figure 9.3 c) shows comparison of the dynamic stability results in the context of the
BCAR Section T boundaries. Figure 9.3 d) compares the response to the aft tilt control
input used to demonstrate Section T compliance for the 40mph case. Above 40 mph
both cases fail Section T compliance and become aperiodically unstable.
The lateral tilt of the rotor (directly proportional to the pilot's lateral stick input) is
greater with the elastic torsion effect included, which improves correlation with flight,
Figure 9.4. Correlation with flight-measured teeter angle, converted into lateral
flapping, is also improved. In all other respects inclusion of this simple torsion model
has a negligible impact on trim, stability and response. Note that the elastic distortion
at the tip of a blade is relatively small, 0.5 deg.
9.5 Discussion
These results tend to suggest that inclusion of a simple model of elastic twist will
have a limited impact on prediction of trim, stability and control response. Whether
considering dynamic stability (Figure 9.3 c)), which is pertinent to linearised or small
perturbations from trim), or the large-amplitude non-linear response to a control input,
Figure 9.3 d), the results suggest a negligible difference. In particular the small
amount of torsion is insufficient to alter the rotorspeed response by a sufficient
margin to present flight safety issue. Lateral control required for trim and the lateral
flapping of the disc are, however, affected by a small amount that nonetheless
improves correlation with flight experiment. The difference in lateral control is
approximately 5%, at most, of the available lateral travel of the pilot's control column.
A comparison of the stability and control derivatives has not been presented, but the
only noticeable difference in evidence is in pitch/roll and roll/pitch cross-coupling.
These results are, however, specific to one blade type that is known to be torsionally
stiff. A more definitive statement is not possible without similar comparisons of a
dissimilar blade type which has not been possible. It must also be remembered that
inclusion of this simple model can, at best, emulate only quasi-steady elastic
behaviour, in this case of only one torsion mode. Coupling with a full aeroelastic
model is a major undertaking but is required if the full time-varying nature of the
dynamic behaviour is to be captured, to include other modes and coupling with
bending. This has been one of the aims of a related study sponsored by the CAA, and
a summary of results is presented in Addendum 3.

Figure 9.1 McCutcheon Blade Torsion Mode Shape
Figure 9.2 McCutcheon Blade Tip Torsional Deflection with Load

a) Aircraft Parameters
Figure 9.3 Effect of Including Simple Blade Torsional Model

b) Rotor Parameters
Figure 9.3 Effect of Including Simple Blade Torsional Model (continued)

c) Dynamic Stability Longitudinal Modes

d) Response to BCAR Section T Compliance Demonstration Control Input

Figure 9.4 Correlation with Flight

10 A Simulation Study of Air Command Autogyros
10.1 Introduction
From the preceding chapters a general picture of gyroplane flight dynamics has been
drawn, albeit with the focus on the test vehicles, a VPM M16, and a Montgomerie-
Parsons. The successful validation of the RASCAL mathematical model against data
from these aircraft allows its further use to predict the characteristics of other
gyroplanes of the same class. Investigation of the Air Command series of gyroplane
aircraft brings the work full circle, as it was a spate of fatal accidents to this type that
initiated the study of gyroplane characteristics [1].
The essential result of the study is that, for the conventional light gyroplane operating
at speeds up to 70 or 80 mph, only the vertical position of the centre-of-mass (c.g.) in
relation to the propeller thrust line has a significant impact on the longitudinal
characteristics of the aircraft across its speed range. Anecdotal evidence suggested
that the effects of a low c.g. was consistent with experience in the pilot community.
For example, one criticism of the Air Command was the low c.g. resulted in
undesirable characteristics such as aft migration of the control column with increasing
speed, reduced control margins at high speed and a substantial nose-down pitch
change during low "g", high power manoeuvres.
Air Command analyses are, in a contemporary context, now only of academic interest
as the type is no longer approved in the UK. However, the research reported in this
section has an important objective, viz. to confirm that the correct relationship
between propeller thrust line and c.g. location is the single most powerful design
feature for endowing a gyroplane with good handling qualities that will satisfy BCAR
Section T, across the speed range. Specific aims were first; to gather engineering
data to allow RASCAL to be configured as single and two seat aircraft; second, to
simulate both configurations and assess the results in the context of handling
difficulties; finally, to assess the efficacy of the AAN modifications required to allow
issue of a new Permit to Fly.
Full understanding of the impact of design changes on aircraft handling qualities can
only be obtained if the reader is willing to grasp the concept of the stability derivative.
Until this point in the report the stability derivative and its meaning have been
consigned to appendices for the more curious reader to savour. However, in bringing
the study of the aerodynamics of gyroplanes full circle it is necessary for the reader
to grasp the physical importance of these mathematically abstract parameters.
Classical fixed-wing modes of longitudinal motion are the phugoid and short-period.
Helicopters can also exhibit similar behaviour, and Padfield [114] has presented and
used appropriate approximations to these modes. These approximations emphasise
the importance of the pitching moment stability derivatives Mu, Mw and Mq. The
additional complicating factor with the gyroplane, to an extent that can only be
quantified by the engineering model, is the rotorspeed degree of freedom. The
rotorspeed, unlike the helicopter, is not governed but is free to vary with flight
condition. It is therefore reasonable to assume that the derivative M might also have
an impact on gyroplane longitudinal stability. Stabilising moments will be those for
which:
Mu > 0
Mw < 0
Mq < 0
M < 0

The latter derivative may not be familiar, even to helicopter flight dynamicists. The
rationale for it to be stabilising if negative is as follows. Given an increase in
rotorspeed, the rotor thrust will increase if the aircraft then pitches nose-down, the
reduced axial flow through the disc will in turn tend to reduce the rotorspeed.
The dominant forces and moments acting about the aircraft centre-of-mass are likely
to be those of the main rotor and the propeller. In principle, rotor behaviour for a
gyroplane will be the same as that for a helicopter, so the extensive theory of blade
flapping behaviour developed for helicopters will also be applicable to the gyroplane.
For example, with a main rotor thrust line passing through the centre-of-mass in
equilibrium flight, Mu will tend to be stabilising, and Mw destabilising. Simple rotor
theory, e.g. Bramwell [83], explains that this is because a perturbation in u will result
in the rotor disc flapping back and the thrust increasing, Figure 10.1. The tendency is
therefore to pitch the aircraft up and hence to reduce the perturbation in u. Likewise,
a perturbation in w will result in the rotor disc flapping back and the thrust increasing.
The tendency is therefore to pitch the aircraft up and hence to increase the
perturbation in w. A similar reasoning indicates that M>0 would be destabilising.
However, with the gyroplane configuration the propeller thrust can be used to modify
the location of the main rotor thrust line in relation to the centre-of-mass. For
example, if the propeller thrust line is placed below the centre-of-mass, the
consequent nose-up moment in equilibrium flight is balanced by placing the main
rotor thrust line behind the centre-of-mass, Figure 10.2. The possibility then exists of
producing a situation where Mw and M are now stabilising, with only Mu
destabilising.
10.2 Analysis of Two-Seat Air Command GBSSL
10.2.1 Flight Testing
The gradient of longitudinal stick position with speed is an important handling qualities
parameter and, although derived from steady flight, is a measure of the dynamic
stability derivative Mu. This derivative has been shown to have a significant impact on
phugoid mode characteristics, particularly its frequency [114]. Hence one simple test
can reveal much about static as well as dynamic stability. The two-seat Air Command,
G-BOOJ similar to that shown in Figure 10.3, was used. The two-seater was chosen
for reasons of availability and the additional safety afforded by having two crew.
Measurements were averaged over two runs. Note that a different two seat Air
Command, G-BSSL, was used for the static tests.
The geometry of the flight control system on this aircraft is such that fore and aft tilt
of the rotor head is accomplished by, in effect, vertical movement of the stick, Figures
10.4 and 10.5. The quantitative measurements were taken simply with a 12 inch. ruler
by the non-handling crew member. Figure 10.6 shows the stick position curves
against speed, for an arbitrary datum (relative, rather than absolute data is all that is
required for establishing gradient, and the datum of zero does not reflect a control
limit, but rather the stick position in equilibrium flight at 40 mph). Conditions for the
test were reported as flat calm.
A simple second order polynomial fit was made through these data, and
differentiation of the polynomial with respect to airspeed then gives a linear equation
representing stick gradient as a function of airspeed. This result is shown in Figure
10.7, together with results taken from Ref. 115. The VPM M16 results were obtained
during the CAA-funded work using G-BWGI in 1996, and were recorded using the
instrumentation system installed to support these tests. The RAF 2000 data was
obtained in a similar manner to that for this Air Command exercise using G-BXDD. G-
BOOJ data are converted to in/knot for consistency with Ref. 115. A stick/rotor tilt
calibration was not available for the RAF 2000, hence VPM and Air Command flight
results have been converted for consistency.

Airspeed calibration errors were unknown for both the Air Command and RAF 2000
aircraft and hence observations are difficult to compare directly. In general, however,
the VPM demonstrates better speed stability than either of the other two aircraft at
low to medium speeds. The Air Command and RAF 2000 speed stability is insensitive
to airspeed changes, with the Air Command having better speed stability than the
RAF 2000. However a previous study (Ref. 116) documents a single seat Air
Command stick gradient of -0.045 inches/knot at 40 mph, but does not quantify the
two seater result due to difficult test conditions, although this suggests that it is
weaker that the single seat machine, i.e. less than -0.045 inches/knot. This is
significantly dissimilar to the result presented in Figure 10.7 and serves to highlight
the difficulty in reconciling different pilot opinion and assessment without on-board
instrumentation.
10.2.2 Model Configuration
The aircraft is shown in Figure 10.8. Mass, c.g. and inertias were calculated from
geometric measurements, wheel reactions and oscillation periods. Single point
suspension was used for roll and pitch inertias, and a bifilar method [117] was used
for yaw inertia calculation. The aircraft had zero fuel, and measurements were made
with an 84 kg pilot on board. (Two sets of measurements were taken: with and
without pilot. This allowed determination of pilot c.g., and hence calculation of weight
and balance with the normal complement of two crew). The three methods used for
determining single seat vertical c.g. could not be used with this aircraft, as with a pilot
on board, it would not sit on its tailwheel. However, sufficient confidence has now
been achieved with the incline method, that the result from a single test can be
regarded as accurate. This method is now used at Glasgow University in preference
to the single- or multi-point suspension for vertical c.g. calculation. The reaction at
each of the three wheels is measured on a level surface allowing the longitudinal
position of the c.g. to be found. The same measurements are then taken on an
inclined surface which, by simple geometry, allows the vertical location also to be
determined, see Appendix 3.
Table 10.1 presents the mass-related properties of the aircraft, together with the
location of the propeller. The reference point for all measurements was one of the
bolts through the mast plates approximately half-way up from the keel. The engine is
inclined 3 deg downwards relative to the aircraft datum, resulting in the normal
distance from the propeller axis to the c.g. being 4.5 in. Note that this configuration is
an extrapolation of measurements taken with the 84kg pilot, but now assuming two
crew, which gives the worst case in terms of vertical c.g. position.
Table 10.1 Calculated Mass and Inertia Properties, from Measurements
Quantity Value
mass 310.2 kg
c.g. location (0.2502,0,0.2590) m
propeller location (-0.625,0,0.100) m
Ixx 73.0 kgm2
Iyy 160.0 kgm2
Izz 90.0 kgm2

Other measurements were taken to allow creation of a datafile for the RASCAL
simulation program, which are summarised in Table 10.2. The lift curve slope for the
various aerodynamic surfaces is based on previous experience with the VPM wind
tunnel model. This may be of limited applicability for the pod, which unlike that of the
single seat aircraft, is dissimilar to that of the VPM in terms of planform, shape and
outline. The rotor blades are of NACA 8H12 section, data for which is published in Ref.
92. Note that rotor radius is increased relative to the single-seat aircraft.
Table 10.2 Measured Geometric Properties
Quantity Value
rotor radius 3.8105 m

rotor location (0,0,-0.877) m
pod frontal area 0.94 m2
pod plan area 1.57 m2
pod side area 1.57 m2
fin area 0.69 m2

tailplane area 0.47 m2
pod drag coefficient 1.0
pod lift-curve slope 1.0 /rad
tailplane lift-curve slope 2.0 /rad
fin lift-curve slope 3.5 /rad
10.2.3 Results
Simulations were conducted in three configurations: the aircraft with both the pod
and horizontal stabiliser fitted (compliant with the 1994 AAN produced to return the
aircraft to flight following grounding); without these modifications (i.e. the
configuration at the time of the fatal accidents); and finally without the AAN
modifications but with the engine thrust line tilted down by 10 deg rather than 3 deg,
the additional 7 deg. required to place the thrust line and c.g coincident. This may then
be considered as an alternative modification to the AAN whose principle aerodynamic
effect, inter alia, was to introduce a pod and tailplane. The results therefore quantify
the effect of the principal changes arising from the modified AAN against a simple
thrust line modification. Note that the simulations were configured for zero fuel and
two crew, which as noted above gives the worst case in terms of c.g. position.
Figures 10.9, 10.10 and 10.11 show comparisons of the trim, principal longitudinal
stability derivatives and stability roots, between 40 and 75 mph. The 1994 AAN
modifications have little impact on the trim, stability and hence BCAR Section T
compliance characteristics. However, simply tilting the engine thrust line down by 10
deg as opposed to 3 deg renders the aircraft stable in the phugoid (that is the real part
of the eigenvalues are now negative), and Section T compliant across the speed
range, that is the eigenvalues for the phugoid move from the right, non-compliant side
of the boundary to the left compliant side.

The other main effect of the thrust line modification is on the trim of the aircraft. The
model predicts that the hub tilt would be about 2 deg further forward, and pitch
attitude about 2 deg more nose-up across the speed range. Again, Mu is reduced in
magnitude, as predicted by elementary consideration of the rotor and propeller forces
presented earlier in Section 10.1. Reduction in Mu will tend to reduce the frequency
of the phugoid oscillation, i.e. increase the period. Note that Mw is rendered negative
(stabilising) with the thrust line modification which will tend to improve the damping,
i.e. reduce the instability. The pitching moment due to rotorspeed changes (M) is
also more stabilising with the thrust line modification. The phugoid mode is now
stable throughout the speed range, i.e. the eigenvalues remain to the right of the
vertical axis within the stable region (as frequency is plotted against damping ratio -
see Appendix 4) and with positive damping causing the amplitude of any oscillatory
motion to decay. The modified aircraft would easily satisfy Section T.
10.3 Analysis of Single-Seat Air Command GBRLB
10.3.1 Model Configuration
This aircraft is shown in Figure 10.12. Once again, mass, c.g. and inertias were
calculated from geometric measurements, wheel reactions and oscillation periods.
Single point suspension was used for roll and pitch inertias, and a bifilar method was
used for yaw inertia calculation [117]. The aircraft had zero fuel, and all measurements
were made with a 78 kg pilot on board. Three methods of measuring vertical c.g.
were used: suspension; wheel reaction when sitting on an incline, nose-up; and
wheel reaction when sitting on an incline, nose-down. The three methods produced
a range in vertical c.g. position of 1.2 in, with the mean value being used in
simulation. Note that three tests were performed to provide confidence in the final
result however, subsequently it has been accepted that a single carefully conducted
test using the inclination method should provide results with sufficient accuracy.
Table 10.3 presents the mass-related properties of the aircraft, together with the
location of the propeller. The reference point for all measurements can be seen on
Figure 10.4, as the top aft bolt on the engine suspension plate immediately behind
the crew seat. This places the propeller 0.625 m behind and 0.1 m below the
reference point, corresponding to a point mid-distance between the front and aft
faces of the propeller hub. The engine is inclined 3 deg downwards relative to the
aircraft datum, resulting in the normal distance from the propeller axis to the c.g.
being 3-5 in, depending on measurement method.
Table 10.3 Measured Mass and Inertia properties
Quantity Value
mass 240.5 kg
c.g. location (0.1574,0,0.2376) m
propeller location (-0.625,0,0.100) m
Ixx 70.8 kgm2
Iyy 130.1 kgm2
Izz 60.0 kgm2

Table 10.4 Measured Geometric Properties
Quantity Value
rotor radius 3.5075 m

rotor location (-0.0160,0,-0.877) m
pod frontal area 0.33 m2
pod plan area 0.85 m2
pod side area 0.65 m2
fin area 0.47 m2

tailplane area 0.38 m2
pod drag coefficient 1.0
pod lift-curve slope 1.0 /rad
tailplane lift-curve slope 2.0 /rad
fin lift-curve slope 3.5 /rad
Other measurements were taken to allow the creation of a datafile for the RASCAL
simulation program. These are summarised in Table 10.4. Note that the lift curve
slope for the various aerodynamic surfaces is based on previous experience with the
VPM wind tunnel model. The rotor blades are of NACA 8H12 section, data for which
is published in [49].
Simulations were, again, conducted in three configurations: compliant with the 1994
AAN; the basic aircraft; and with the thrust line modification applied to the basic
aircraft.
The single and two seat aircraft are almost identical geometrically apart from the pod,
minor dimensional differences in the rotor head and, of course, the rotor diameter.
10.3.2 Results
Figure 10.13 shows the trim comparisons between 35 and 75 mph, the upper speed
being some 10 mph faster than Vne. It is clear that the pod and tailplane have an
insignificant impact on trim. Note that the model predicts the aircraft to have positive
speed stability, since the longitudinal tilt of the rotor spindle is increasingly forward
with increasing speed.
Figure 10.14 shows the principal longitudinal stability derivatives across the same
speed range. The positive speed stability is confirmed by Mu>0, but the angle of
attack stability is negative, since Mw>0. The stability roots associated with the model
put these derivatives in the context of dynamic stability and Section T compliance,
Figure 10.15. In terms of dynamic stability, there are insignificant differences
between the AAN-compliant version and unmodified aircraft other than at high speed
where the modified aircraft is less unstable. However, both configurations fail to
comply with Section T and, again, it is only the thrust line modified version that
satisfies Section T.

10.4 Discussion
If the Air Command aircraft did have a handling qualities problem, it was likely to lie
in poor phugoid mode characteristics. Simulation tends to suggest a very unstable
mode, whether fitted with pod and tailplane as required by the current AAN or not.
However, the instability is not too dissimilar to some unaugmented helicopters, e.g.
the Lynx, and any pilot difficulty with the Air Command was likely to be due to either
lack of currency, recency or experience, or the fact that the mode has a high natural
frequency (almost 1 cycle every 10 sec) relative to other aircraft. Indeed, existing
criteria for phugoid mode characteristics [100] were previously applied to the VPM
M16 flight data, but these criteria do not extend to the relatively high frequency that
is suggested by the Air Command simulation, so it is impossible to grade the aircraft
in these terms.
However, the RASCAL mathematical model has known limitations in respect of
simulation of the VPM M16, specifically with regard to the phugoid. This was isolated
to the drag damping Xu which was significantly overestimated by simulation. (Phugoid
damping is determined primarily by drag damping Xu although an increasingly positive
Mw will tend to destabilise this mode). It is by no means certain that these limitations
map directly across to validity in respect of other types. The model may be more or
less valid for simulation of the Air Command types but, without a full validation
exercise that is type-specific, this cannot be assessed. The simulation model is
known to be optimistic in respect of the prediction of VPM M16 drag damping Xu, and
this directly affects the damping of the mode of most concern to the Air Command
problem. If this optimism reads across to the Air Commands, the degree of instability
highlighted earlier will be even worse.
If there is concern regarding the applicability of the model for analysing the Air
Command, the actual physical measurements of speed stability (weight, balance,
inertia etc.), constitute valuable entities in their own right that communicate much
about the real aircraft. For example, a low c.g. position will result in an arrangement
of forces and resulting moments that will tend to be destabilising. Hence the low c.g.
of both single and two-seat Air Commands points to a potential problem with pitch
axis handling qualities, but will not, in isolation be sufficient to quantify this problem.
The measurements from the very limited flight testing that was conducted are the
most that can be attempted without instrumentation, but the results can be related
directly to the prediction of speed stability and the derivative Mu. Speed stability is
comparable to the VPM M16, and better than the RAF 2000.
The low c.g. position has a direct influence on the angle of attack stability Mw, as
highlighted in Section 10.1 earlier. The simulation model has been demonstrated to
be excellent for predicting VPM M16 Mw across the speed range. If this is also true
for the Air Command aircraft, the impact on this derivative of increasing engine
inclination from 3 to 10 deg has been accurately modelled. The stabilising influence
of the modification will therefore also be accurately modelled.
A final point relates to observations of differences in control stick geometry. The two-
seat variant was fitted with a 'pump-action' stick; perhaps they should have been
fitted with a more conventional control stick that displaced in a fore-and-aft sense,
rather than vertically. The already good speed stability might therefore have been
translated into a conventional, and therefore readily interpreted form, for the pilot. It
seems odd that pilots trained on the two-seater with one stick geometry, but had to
progress onto the single-seater with fundamentally dissimilar control geometry.

Figure 10.1 Schematic of Main Rotor Thrust Line Relative to c.g. in Undisturbed
and Disturbed Flight - Propeller Thrust Line Passing Through c.g.
Figure 10.2 Schematic of Main Rotor Thrust Line Relative to c.g. in Undisturbed
and Disturbed Flight - Propeller Thrust Line Passing Below c.g.

Figure 10.3 Two seat Air Command
Figure 10.4 Stick Fully Aft

Figure 10.5 Stick Fully Forward
Figure 10.6 G-BOOJ Stick Position with Airspeed

Figure 10.7 Stick Gradient Comparisons
Figure 10.8 Two Seat Air Command G-BSSL

Figure 10.9 Trim Comparisons, Simulation of G-BSSL

Figure 10.10 Derivative Comparisons, Simulation of G-BSSL

Figure 10.11 Comparison of Longitudinal Stability Roots
Figure 10.12 Single Seat Air Command G-BRLB

Figure 10.13 Trim Comparisons, Simulation of G-BRLB

Figure 10.14 Derivative Comparisons, Simulation of G-BRLB

Figure 10.15 Stability Comparisons, Simulation of G-BRLB

11 Investigation of Gyroplane Rotor Teeter Motion
11.1 Introduction
Sufficient rotor teeter margin is a compliance demonstration requirement of BCAR
Section T. This requires flight beyond the flight envelope boundary (VNE) where
nothing is known about teeter behaviour. Indeed teeter angle measurements have
not previously been measured in flight. There is sufficient evidence (factual and
anecdotal) to suggest that teeter angles can be excessive in certain circumstances,
and can lead to blade strike with the propeller, pod or mast. There is therefore a need
to take such measurements to establish just how significant teeter motion is, and
whether it can have implications on aircraft safety. This will have secondary value
both for use as further data for validation of the simulation model, and in the
continued development of BCAR Section T.
11.2 Teeter Angle Measurements
One of the unique aspects of this series of flight tests was that this is the first time
that a gyroplane rotor's teetering motion has been recorded. Detail of the
experimental procedure is given in Section 5 of this report. Figure 11.1 shows the
complete recording of teeter angle for flights 2 - 4 sampled at a rate of 200Hz. Note
that the total recording time begins before the aircraft is started up and continues to
record all data until the total record time has elapsed. For example, for flight no 2 the
total recording time was 50mins (3000s), the first 500s (approx) represents the
aircraft on the ground prior to engine start, and the final 500s (approx) is the aircraft
back on the ground engine off. The very large teeter angles measured at the start and
end of each recording are due to the rotor teetering due to wind or ground handling.
Referring to the data for the remainder of the flight it is clear that the teeter angle
remains small for most of the flight (less than 5 degrees). The blade time histories
shown in Figure 11.1 are referred purely to a time base and it is difficult to establish
at which points on the azimuth that the peak values of teeter occur.
Figure 11.2 shows the disc tilt angle for flight 2. This shows the disc tilted aft (-ve) by
about 1.5 - 2 degrees and by a similar amount to the right (+ve) over most of the flight.
Figure 11.3 shows the disc tilt angles for flight 3. This particular flight was focused on
dynamic testing with a series of pulse and step inputs applied. Despite this, the disc
tilt angles are still small with the disc tilting slightly aft and to the right relative to the
hub. Finally, the disc tilt angles for flight 4 are given in Figure 11.4, and, as before,
small angles are observed for most of the flight.
11.3 Aircraft Response in Recovery from Large Amplitude Stick Displacement
Figures 11.111.4 show relatively small teeter angles during most of the flights. The
exception to this observation is an event just before 1000s during Flight No. 4 (see
Figure 11.1 and Figure 11.4). For a short period of time teeter angles of 8 degrees
were recorded. Figure 11.5 shows the teeter angle time history for the few seconds
around this event. The very rapid increase in teetering is evident. This is a significant
occurrence as the allowable range as quoted by the manufacturer is 9 degrees. It
should be noted that on the ground the physical restraint (i.e. the teeter stop) permits
teetering motion beyond the manufacturer's quoted limits. This can be observed in
Figures 11.111.4 where teeter angles of 12 degrees (the design range of the
sensor) are recorded in ground handling.
This occurrence corresponds to a recovery from a 1 inch step input of forward stick
at a flight speed of 60mph. Although a 1 inch step was requested, evidence from the
stick displacement transducer indicates a displacement somewhat larger than this.
Given the rudimentary design of the aircraft cockpit it is extremely difficult to apply

control inputs of a particular amplitude with great accuracy. This event is presented
in more detail by examining the 20s from 940s to 960s on the recording, Figure 11.6.
The forward stick input is shown on the first plot which is, in effect, of shaft tilt angle,
a forward stick input is applied which gives 4 degrees forward shaft tilt. The
immediate effect is for the nose to pitch down to -20 degrees, airspeed to increase
from a nominal 60mph to around 75 mph, and most surprisingly, rotorspeed increases
dramatically from around 360rpm to 450rpm. The aircraft clearly has entered an
uncomfortable flight state and the pilot has had to take rapid action to recover. The
recovery involves a sudden chopping of power and an aft stick displacement
equivalent to around 6 degrees of shaft tilt at roughly 950 seconds. The aircraft then
begins to recover its attitude and airspeed, and rotor speed decays to its original
values.
It is clear that the aircraft has experienced a severe manoeuvre, and that large teeter
angles have been observed. The question arises as to where exactly the maximum
teeter occurs in the azimuth. Figure 11.7 shows the teeter angle plotted against
azimuth for the several revolutions of the blade during the period 948 - 952 seconds.
It is clear that the peak value of teeter is occurring at around 220 of azimuth. The
sensor is calibrated such that 180 is the rearmost position, and noting that the rotor
is rotating in an anti-clockwise direction, the blade is reaching its maximum teeter in
a location just over the pilot's right shoulder.
It should be noted that the teeter angle reaches a maximum of 7 in the 180 position
raising the question of whether a propeller strike in flight is possible.
11.3.1 Ground Checks on Propeller Clearance
The results described above prompted a series of ground tests on the aircraft to
determine propeller clearances in conditions where large teeter angles occurred. The
first thing to note is that the strike plate which limits teetering motion is ineffective -
with the stick fully aft the blade can be teetered through the plane of the propeller
until its tip touches the ground, Figure 11.8. In this position there is still around 2cm
of clearance between strike plate and under surface of the blade. To replicate the
flight state observed during Event 1 of Flight 4, the stick was held in an aft position,
and a teeter angle of 8 applied to the rotor, Figure 11.9. The clearance between the
tip of the propeller and the under surface of the blade was 5cm. In comparing this with
the actual event, it should be noted that in the ground test the stick was fully aft, and
in flight there would be some coning of the blades providing additional clearance.
Finally, the lowest position of the blade (i.e. maximum teeter occurs at around 220
of azimuth, Figure 11.7) which is to the side of the propeller disc. These three factors
combined suggest that in flight the clearance is likely to have been a little greater than
the 5cm measured on the ground.
The final test performed was to estimate the minimum teeter angle required to cause
a blade strike on the propeller. The stick was held in its fully aft position to provide the
worst possible situation. The blade was then teetered until it touched the tip of the
propeller, Figure 11.10. The measured teeter angle was 11.1. Again it should be
stressed that this is the worst case - with the stick in a more central position, taking
into account coning and the off-centre position for maximum teeter, a teeter angle in
excess of 11.1 would likely be required to cause a strike in flight.


The following three conclusions are based on the results presented in this paper,
which were recorded during flight trials in the period 20-21st July 2005.
1 The rotor teeter angle is very small in all steady flight states.
2 In typical transient manoeuvres teeter angles equivalent to one third to one half
of the allowable range are observed.
3 A single event indicates that gross manoeuvres can excite teeter motion such
that the teeter angle can approach its limits. This has implications in recovery
procedures from unusual flight states.
The data gathered has addressed a lack of knowledge of rotor teetering behaviour and
has confirmed the applicability of a proposed change to BCAR Section T 143 (a)
(Interpretive Material). The current issue of BCAR Section T (Issue 3) has interpretive
material in support of T 143 (a) that states:
"The gyroplane's VNE will normally be dictated by the need to have a positive teeter
margin up to VNE+15%".
Concern has been expressed that it is improper to relate an acceptance criteria to an
airspeed 15% greater than VNE and, hence, 4% greater than VDF. Furthermore,
available technology does not provide for a ready means of measuring rotor teeter
angle and until the flight test programme described above, teeter angle had not been
measured in flight. Following this work, alternative text requiring demonstration of
satisfactory flight characteristics at speeds up to 1.11 VNE has been drafted by the
CAA.
Consequently, the CAA has proposed that the next issue of Section T should contain
the following revised text for AMC T 143 (a):
".. it will be expected that satisfactory flight characteristics have been demonstrated
at speeds up to 1.11 VNE. This demonstration must include satisfactory control margin
and rotor clearance."

Figure 11.1 Blade Teeter Angles for all 3 Flights

Figure 11.2 Blade Teeter Angles Expressed in Multi-blade Co-ordinates for Flight 2

10
0
946 947 948 949 950 951 952
-2
-4
-6
-8
Time (s)
Figure 11.5 Teeter Angle During Recovery from Large Amplitude Input

Figure 11.6 Time Histories for Recovery from Large Displacement Longitudinal
Stick Input
Figure 11.7 Teeter Angle vs. Azimuth for 948 - 952s of Flight 4

Figure 11.8 Stick Aft, Blade Teetered Through Propeller Disc

Figure 11.9 Stick Aft, 8 of Teeter

Figure 11.10 Teeter Measurement for a Propeller Strike

12 Overall Conclusions
Due to extremely poor accident statistics in the early 1990s, the AAIB recommended
that the CAA undertake a research study into the aerodynamics of gyroplanes to
address the lack of contemporary knowledge of this type of aircraft, and to provide
the technical information required to update the airworthiness requirements for this
class of aircraft, BCAR Section T. Research at the University of Glasgow has focused
on supporting the UK Civil Aviation Authority to achieve these goals. The depth and
breadth of knowledge on the flight characteristics of gyroplanes has without doubt
been extended by this project. The following observations and conclusions can be
drawn from the research.
a) At the outset of the project there was only a limited volume of archival
documentation on the fundamental aerodynamic and dynamic properties of
gyroplanes. Dissemination of the work reported in this document through journal
articles, conference presentations and public lectures has gone some way to
address this. Due to the interest generated by this project (both nationally and
internationally), much more technical information is now available in the open
literature.
b) Wind tunnel tests have shown that the aerodynamic properties of gyroplanes are
relatively insensitive to configurational changes. Even at the high speed end of the
range, the aerodynamic properties of the vehicle pod and tailplane have little
influence.
c) The RASCAL mathematical model, both in its linear and nonlinear formulation, was
validated using flight test data. Analysis of these results show that RASCAL
provides a valid predictive tool for analysing gyroplane stability characteristics.
Consistent results were obtained allowing the model's applicability and limitations
to be defined. Certainly, in parametric studies where it is a trend in relation to
changes in configurational parameters which is to be identified, RASCAL is as good
as contemporary helicopter mathematical modelling practice.
d) In general gyroplanes exhibit a mix of stability characteristics typical of those from
fixed wing aircraft and helicopters. Notably, they possess a lightly damped phugoid
mode. The lateral/directional dynamics of the gyroplane are broadly similar to those
of a conventional aircraft and are usually benign.
e) The single most significant factor in determining the longitudinal stability of
gyroplanes is the vertical location of the centre of gravity relative to the propeller
thrust line. It has been shown that a centre of gravity location below the propeller
thrustline can destabilise the phugoid mode. Flight tests to demonstrate
compliance with BCAR Section T can be difficult to perform, particularly without
sophisticated measurement and data acquisition systems to provide objective
documentation of aircraft performance. Locating the centre of gravity appropriately
in location to the propeller thrust line then offers the best prospect for acceptable
handling qualities across the speed range. Gyroplane stability is insensitive to
changes in most other configurational parameters with the exception of the
horizontal tailplane where gross changes in area (complete removal, for example)
can influence the short period mode and hence the handling qualities of the
aircraft.
f) Although the teeter motion of the rotor is generally of low amplitude, large scale
inputs or severe manoeuvring can rapidly lead to large amplitude teeter oscillations
with serious consequences for flight safety.

The work described in this Paper represents a substantial advance in the

understanding of gyroplane flight dynamics. The development of the mathematical
models now provide a significant predictive capability, and the data gathered in the
flight tests, as well as being essential for validation purposes, has also provided
insight into the flight characteristics of this vehicle type. The primary aim of the
research was to improve gyroplane safety. This project has undoubtedly increased
awareness of the issues associated with gyroplane configuration. Along with the
redrafting of BCAR Section T, this project has contributed to a lower accident rate and
so has fulfilled its main aim.
CAA Comment
Effect of tailplanes. The report also concludes that horizontal tailplanes are largely
ineffective in improving the long term response of pitch dynamic stability (phugoid
mode). This is the result of studies primarily on narrow tandem cockpit
enclosures. CAA Flight Test Specialist qualitative evaluation of the effects of a
horizontal tailplane on a single side-by-side configuration gyroplane type indicated
a degree of improvement in the phugoid characteristics at higher speed. It is
appreciated that in paragraph 8.3.1 (page 152) of the report it is stated that other
factors can affect the phugoid mode.
Teeter Margins. Due to the possibility of excessive teeter angles under certain
phases of flight, that can lead to blade strike with the prop, pod or mast, flight
tests were undertaken. Flight instrumentation revealed small teeter angles in
steady flight conditions, but in one instance during a particularly extreme
manoeuvre, the angle reached 8 deg within an 11 deg safe envelope. Testing was
limited, and it is difficult to draw meaningful conclusions from the data, other than
to recognise that certain more extreme flight conditions will produce teeter angles
that may go outside of the safe envelope, and so could lead to a strike on some
part of the aircraft.
Advisory material to BCAR Section T.143 (a) is to be revised to require satisfactory
control margin and rotor clearance up to 1.1Vne.

Acknowledgements
In over 15 years of research for this project the authors have sought advice from experts in all
areas of gyroplane flight from owners, operators, manufacturers and those few individual
researchers scattered over the globe who have at some point in their careers dared to
investigate gyroplane flight. Their assistance was wide ranging and invaluable. There are too
many to mention here in full, however the assistance of the following people was particularly
valuable: Roger Savage, Jim Montgomerie, and the late Phil Barlow.
A major feature of the work was the flight testing of the gyroplane - a world first to the
knowledge of the authors. This could not have been achieved without the invaluable
assistance of Colin Handley and Vic Lockwood of FR Aviation. Also, the contribution of Colin
Massey of GKN Westland Helicopters Ltd. in auditing the work on behalf of the CAA should
not be overlooked. The authors would also like to acknowledge the professionalism and
piloting skills of those who carried out the test flying: Chris Chadwick, Roger Savage and B. de
Saar.
The authors would also wish to acknowledge the assistance provided by their colleagues, Prof.
Frank Coton on the aerodynamic testing in the early stages of the project and, latterly, Dr Eric
Gillies who has contributed significant expertise in the area of aeroelastic modelling. Particular
credit must go also to Robert Gilmour, the Department's Research Technologist for all of his
efforts in building, maintaining and operating the instrumentation on board G-UNIV. Tony
Smedley and his technical staff should also be acknowledged for their contributions which
were often above and beyond their job descriptions!
Finally, the authors wish to thank the Civil Aviation Authority for the support and funding for
this research. It has given the University of Glasgow, and the authors particularly, the
opportunity to be involved in an important and unique project. It is therefore appropriate here
to list all of those who were involved: Robin Ablett, Dave Chapman, Julie Denning, Keith
Dodson, Alistair Maxwell, Paul Spooner, Kevin Russell, Carl Thomas, Joji Waites, Jonathan
Howes, Andrew Goudie, and Ray White. Particular acknowledgement of the efforts of Dave
Howson who has borne the responsibility of managing the project should also be made.
August 2010 Acknowledgements Page 1

References
[1] Anon, "Airworthiness Review of Air Command Gyroplanes", Air Accidents

Investigation Branch Report, Sept. 1991.
[2] "British Civil Airworthiness Requirements, Section T, Light Gyroplane Design
Requirements, " Civil Aviation Authority CAP 643, Issue 1, March 1995.
[3] Glauert, H., "A General Theory of the Autogyro", ARC R and M No. 1111, Nov. 1926.
[4] Lock, C.N.H., Further Development of Autogyro Theory Parts 1 and 2", ARC and R
and M No. 1127, March 1927.
[5] Glauert, H., "Lift and Torque of an Autogiro on the Ground", ARC R and M No. 1131,
July 1927.
[6] Lock, C.N.H., Townend, H.C., "Wind Tunnel Experiments on a Model Autogyro at
Small Angles of Incidence", ARC R and M No. 1154, March 1928.
[7] Glauert, H., Lock, C.N.H., "A Summary of the Experimental and Theoretical
Investigation of the Characteristics of an Autogyro", ARC R and M No. 1162, April
1928.
[8] Wheatly, J.B., Lift and Drag Characteristics and Gliding Performance of an Autogiro
as Determined in Flight", NACA TR 434, 1932.
[9] Wheatly, J.B., Simplified Aerodynamic Analysis of the Cyclogiro Rotating Wing
System", NACA TN 467, 1933.
[10] Wheatly, J.B., Wing Pressure Distribution and Rotor-blade Motion of an Autogiro as
Determined in Flight", NACA TR 475, 1933.
[11] Wheatly, J.B., An Aerodynamic Analysis of the Autogiro Rotor with a Comparison
Between Calculated and Experimental Results", NACA TR 487, 1934.
[12] Wheatly, J.B., The Aerodynamic Analysis of the Gyroplane Rotating Wing System"
NACA TN 492, 1934.
[13] Wheatly, J.B., Hood, M.J., "Full-Scale Wind Tunnel Tests of a PCA-2 Autogiro Rotor",
NACA TR 515, 1935.
[14] Wheatly, J.B., Influence of Wing Setting on the Wing Load and Rotorspeed of a PCA-
2 Autogiro as Determined in Flight", NACA TR 523, 1935.
[15] Wheatly, J.B., Windler, R., "Wind Tunnel Tests of a Cyclogiro Rotor" NACA TN 528,
1935.
[16] Wheatly, J.B., Bioletti, C., "Wind Tunnel Tests of a 10-foot-Diameter Gyroplane Rotor",
NACA TR 536, 1935
[17] Wheatly, J.B., Bioletti, C., "Wind Tunnel Tests of a 10-foot-Diameter Autogiro Rotor",
NACA TR 552, 1935
[18] Wheatly, J.B., A Study of Autogiro Rotor Blade Oscillations in the Plane of the Rotor
Disk", NACA TN 581, 1936
[19] Wheatly, J.B., Bioletti, C., "Analysis and Model Tests of Autogiro Jump Take-off", NACA
TN 582, 1936
[20] Wheatly, J.B., An Analytical and Experimental Study of the Effect of Period Blade
Twist on the Thrust, Torque, and Flapping Motion of an Autogyo Rotor", NACA TR 591,
1937.
August 2010 References Page 1

[21] Wheatly, J.B., "An Analysis of the Factors that Determine the Periodic Twist of an
Autogiro Rotor Blade, with a Comparison of Predicted and measured Results", NACA
TR 600, 1937.
[22] Bailey, F.J. Jr., Gustafon, F.B., Observations in flight of the region of stalled flow over
the blades of an autogiro rotor, NACA-TN-741, 1939.
[23] Bailey, F.J. Jr., A Study of the Torque Equilibrium of an Autogiro Rotor, NACA-TR-623.
[24] Breguet, L., The Gyroplane Its Principles and its Possibilities, Technical
Memorandums, NACA, No. 816, 1937
[25] Schrenk, M., Static Longitudinal Stability and Longitudinal Control of Autogiro
Rotors, Technical Memorandums, NACA, No. 879, 1938
[26] Wallis, K.H., Design and Construction of a lowcost Autogyro Aircraft and Detailed
Prospects for Future Application, The Aeronautical Journal, Vol. 67, pp. 111 118, Feb.
1963.
[27] Schad, J.L., "Small Autogyro Performance", Journal of the American Helicopter
Society, Vol. 10, 1965.
[28] Niemi, E.E. "Effect of Various Cabin Designs on the Performance of a Small
Unstreamlined Autogyro", Master of Science Thesis Paper, Worcester Polytechnic
Institute, January 1964.
[29] Mouille, R., "The Turbogyre 330", Agard Report, Helicopter Development, 1966 (In
French).
[30] Smith, R.C., Levin, A.D., The Unpowered Rotor: A Lifting Decelerator foor Spacecraft
Recovery, Proceedings of AIAA Aerodynamic Deceleration Systems Conference,
September 1968, AIAA-1968-969
[31] Barzda, J.J., Rotors for Recovery, AIAA Journal of Spacecraft and Rockets, Vol. 3,
No. 1 pp 104 109, 1966.
[32] Liss, A.I.U., Consideration of Control Elasticity in Calculating the Deformation of
Gyroplane Rotor Blades, Aviatsonnaia Tekhnika, Vol. 17, No. 1, pp 64 71, 1974.
[33] Przybylski, J., Toms, R.L., Cheeseman, I.C., The Investigation of Some Unusual
Handling Characteristics of a Light Autogyro, Proceedings of the 2nd European
Rotorcraft Forum, September 1976
[34] Johnson, W., Comparison of Calculated and Measured Helicopter Rotor Lateral
Flapping Angles, Journal of the American Helicopter Society, Vol. 26, pp 46 50, April
1981.
[35] Nicholls, J.M., Dymock, R.B., James, R.A., "Ultralight Aircraft and Gyrocopter
Fatalities in South Australia", Medical Journal of Australia, Vol 149, Iss 2, p111, 1988.
[36] McKillip, R.M., Chih, M.H., "Instrumented Blade Experiments Using a Light Autogiro",
Proceedings of the 16th European Rotorcraft Forum, Glasgow, Sept. 1990.
[37] Barnard, R.H., Philpot, D.R., "Aircraft Flight", p35-36, Longman Scientific and Technical,
1993, ISBN 0-582-00338-5
[38] Kermode, A.C., "Mechanics of Flight", p191-196, Longman Scientific and Technical,
1991, 9th Edition, ISBN 0-582-42254-X
[39] Prouty, R., "Helicopter Performance, Stability, and Control", p142, 154, 662, Krieger
Publishing, 1990, ISBN 0-89464-457-2
[40] Johnson, W., "Helicopter Theory", p10-11, 15-16, 301-302, Princeton University Press,
1980, ISBN 0-07917-691-4

[41] Shippen, J., Data Acquisition and Analysis of Service Loads in a Light Weight
Autogyro, Proceedings of 52 Annual Forum of the American Helicopter Society, June
1996.
[42] Shippen, J., Measurement of Autogyro Control Forces, Proceedings of the I.Mech.
E Part G. Journal of Aerospace Engineering, Vol. 212, No. 1, pp 57 60, 1998.
[43] Lopez-Diez, J., Cuerno-Rejado, C., Lopez-Ruiz, J.L., Directional Control of a Non-
rudder Autogiro-Landing Manoeuvre of the C-30, Proceedings of the 24th European
Rotorcraft Forum, pp. FM11.1 FM11.9, September 1998
[44] Lopez-Diez, J., Cuerno-Rejado, C., Lopez-Ruiz, J.L., Study of Competitive Missions
for Autogyros, Proceedings of the 25th European Rotorcraft Forum, September 1999.
[45] Rapp, H., Wedemeyer, P., Teuber, C., Measurement of in-Flight Rotor Blade Loads of
an Autogyro. Proceedings of the 26th European Rotorcraft Forum, pp 101.1 101.8,
September 2000.
[46] Wang, H-J., Gao, Z., Aerodynamic Virtue and Steady Rotary Speed of Autorotating
Rotor, Acta Aeronautica et Astronautica Sinica, Vol. 22, No. 4, pp. 337 339, July
2001.
[47] Kalmykov, A.A., Realizable Combinations of Gyroplane Main Rotor Overload and
Starting, Aviatsionnaya Tekhnika, No. 2, pp 6 10, April June 2004.
[48] McCormick, B.W., A Numerical Analysis of Autogyro Performance, Proceedings of
the Biennial International Powered Lift Conference and Exhibit, AIAA Paper No. AIAA-
2002-5950, November 2002.
[49] Traum, M.J., Carter, R.G., Pitch Control Benefits of Elevators for Autogyros in Low-
Speed Forward Flight, 43rd AIAA Aerospace Sciences Meeting and Exhibit, AIAA
Paper No. AIAA 2005 26, January 2005.
[50] Li, Y., DeLaurentis, D, Mavris, D., Advanced Rotorcraft Concept Development and
Selection Using a Probabilistic Methodology, Proceedings of the 3rd Annual AIAA
Aviation Technology, Integration, and Operations Forum, Paper No. AIAA 2003-6759,
November 2003.
[51] Ahn, B-H, DeLaurentis, D., Mavris, D.N., Advanced Personal Air Vehicle Concept
Development Using Powered Rotor and Autogyro Configurations, Proceedings of the
2nd Annual AIAA Aviation Technology, Integration, and Operations Forum, October
2002.
[52] Jensen, D.I., Un-manned Autogyro for Cinematography and Reconnaissance,
Proceedings of Aircraft, Technology Integration and Operations Forum, AIAA Paper
No. AIAA-2001-5228, October 2001.
[53] Lopez, C.A., Wells, V.L., Dynamics and Stability of an Autorotating Rotor/Wing
Unmanned Aircraft AIAA Journal of Guidance, Control, and Dynamics, Vol. 27, Part 2,
pp 258 270, 2004.
[54] Somov, Y.I., Polyntsev, O.Y., Nonlinear Dynamics and Robust Control of a Wind-
milling Gyroplane Rotor, Proceedings of the International Conference on Physics and
Control, August 2003.
[55] Somov, Y.I., Polyntsev, O.Y., Nonlinear Dynamics and Robust Control of a Gyroplane
Rotor, Proceedings of the IFAC Congress, 2005.
[56] Carter, J. Jr., CarterCopter A High Technology Gyroplane, Proceedings of the
Vertical Lift Aircraft Design Conference, Jan. 2000.
[57] Whitney, M.J., Shah, A.H., Dynamic Analysis of the Hawk 4 Rotor Blade,
Proceedings of the 43rd AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics
and Materials Conference, Paper No. AIAA-2002-1605, April 2002.

[58] Groen, D., Gyroplane Technology, Proceedings of the International Air and Space
Symposium and Exposition: The Next 100 Years, Paper No. AIAA-2003-2519, July
2003.
[59] Gibbings, D., The Fairey Rotodyne: Technology Before Its Time?, The Aeronautical
journal, Vol. 108, No. 1089, pp. 565 574, 2004.
[60] Leishman, J.G., Development of the Autogiro: A Technical Perspective, AIAA Journal
of Aircraft, Vol. 41, No. 4, July-August 2004.
[61] Houston, S.S., Thomson, D.G., "A Study of Gyroplane Flight Dynamics", Paper No. VII-
6, 21st European Rotorcraft Forum, St Petersburg, Russia, August 1995
[62] Houston, S. S., "Longitudinal Stability of Gyroplanes", The Aeronautical Journal, Vol.
100 No. 991, 1996, pp. 1-6.
[63] Houston, S.S., Thomson, D.G., Flight Investigation of Gyroplane Longitudinal Flight
Dynamics, Paper 108, Proceedings of the 23rd European Rotorcraft Forum,
Dreseden, Germany, September 1997.
[64] Houston, S.S., Thomson, D.G., Identification of Gyroplane Stability and Control
Characteristics, Proceedings of NATO Research and Technology Organisation
Symposium System Identification for Integrated Aircraft Development and Flight
Testing, Madrid, Spain, May 1998
[65] Spathopoulis, V.M., Thomson, D.G., Houston, S.S., Flight Dynamics Issues Relating
to Autogyro Airworthiness and Flight Safety, Proceedings of the 54th American
Helicopter Society Annual Forum, Washington, May 1998
[66] Houston, S. S., "Identification of Autogyro Longitudinal Stability and Control
Characteristics", Journal of Guidance, Control and Dynamics, Vol. 21, No. 3, 1998, pp.
391-399.
[67] Houston, S. S., "Identification of Gyroplane Lateral/Directional Stability and Control
Characteristics from Flight Test", Proc. Inst. Mech. Engrs, Part G, Vol. 212, 1998, pp.
271-285.
[68] Coton, F., Smrcek, L., Patek, Z., "Aerodynamic Characteristics of a Gyroplane
Configuration", AIAA Journal of Aircraft, Vol. 35, No. 2, 1998, pp.274-279.
[69] Houston, S.S, Thomson, D.G., Spathopoulos, V.M., "Experiments in Autogyro
Airworthiness for Improved Handling Qualities", Proceedings of 57th Annual Forum of
the American Helicopter Society, May 2001.
[70] Houston, S.S., Analysis of Rotorcraft Flight Dynamics in Autorotation, AIAA Journal
of Guidance, Control and Dynamics, Vol. 25, Vol. 1., pp 33 39, 2002.
[71] Bagiev, M., Thomson, D.G., Houston, S.S., Autogyro Inverse Simulation for Handling
Qualities Assessment, Proceedings of the 29th ERF, Sept. 2003
[72] Houston, S.S., Modelling and Analysis of Helicopter Flight Mechanics in
Autorotation, AIAA Journal of Aircraft, Vol. 40, No. 4, pp 675 682, 2003
[73] Houston, S.S., Brown, R., Rotor Wake Modelling for Simulation of Helicopter Flight
Mechanics in Autorotation, AIAA Journal of Aircraft, Vol. 40, No. 5, 2003
[74] Bagiev, M., Thomson, D.G., Houston, S.S., Autogyro Handling Qualities Assessment,
Proc. of American Helicopter Society 60th Annual Forum, June 2004.
[75] Thomson, D.G., Houston, S.S., Experimental and Theoretical Studies of Autogyro
Flight Dynamics, Proceedings of the 24th ICAS Congress, Sept. 2004
[76] Thomson, D.G., Houston, S.S., Application of Parameter Estimation to Improved
Autogiro Simulation Model Fidelity, Special Issue on Parameter Estimation, AIAA
Journal of Aircraft, Vol. 42, No. 1, Jan/Feb 2005.

[77] Thomson, D.G., Houston, S.S, Spathopoulos, V.M., "Experiments in Autogyro

Airworthiness for Improved Handling Qualities", Journal of American Helicopter
Society, October 2005
[78] Trchalik, J., Gillies, E., Thomson, D.G., Aeroelastic Modelling of a Rotor in
Autorotative, Axial Flight, Proceedings of the 32nd European Rotorcraft Forum,
Maastricht, Netherlands, September 2006
[79] Bailey, Roger, The Role of the Test Pilot, Aerogram, Vol. 8, No. 3. Cranfield University
(October 1996)
[80] Tischler, M. B., "Identification Techniques, Frequency Domain Methods", AGARD
LS178, pp. 6-1 to 6-4, October 1991.
[81] Anon, AAIB Bulletin No: 4/95 Ref: EW/C95/1/5, April 1995
[82] Leishman, J.G., Principles of Helicopter Aerodynamics, Cambridge University Press,
ISBN 0-521-85860-7, 2006
[83] Done, G., Balmford, D., Bramwells Helicopter Dynamics, Butterworth Heinmann,
ISBN 0-7506-5075-3, 2001.
[84] Padfield, G., Helicopter Flight Dynamics, Blackwell Science, ISBN 0-632-05607-X,
1996
[85] Dreier, M., Introduction to Helicopter and Tiltrotor Simulation, AIAA Education
Series, ISBN 1-5637-873-0, 2007
[86] Newman, S., The Foundations of Helicopter Flight, Edward Arnold, ISBN 0-340-
58702-4, 1994
[87] Houston, S., "Rotorcraft Aeromechanics Simulation for Control Analysis -
Mathematical Model Definition", University of Glasgow Dept. of Aerospace
Engineering Report No. 9123, 1991.
[88] Houston, S. S., "Validation of a Non-linear Individual Blade Rotorcraft Flight Dynamics
Model Using a Perturbation Method," The Aeronautical Journal, Vol. 98 No. 977, 1994,
pp. 260-266.
[89] Houston, S. S., "Validation of a Blade-Element Helicopter Model for Large-Amplitude
Manoeuvres", The Aeronautical Journal, Vol. 101 No. 1001, 1997.
[90] Peters, D. A., HaQuang, N., "Dynamic Inflow for Practical Applications", Journal of the
American Helicopter Society, Vol. 33, No. 4, 1988.
[91] Houston, S., "Flight Mechanics of Gyroplanes", University of Glasgow, Dept. of
Aerospace Engineering Report No. 9323, 1993.
[92] Schaefer, R.F., Smith, H.A., "Aerodynamic Characteristics of the NACA 8-H-12 Airfoil
Section at Six Reynolds Numbers". NACA Tech. Note 1998. December 1949.
[93] Brotherhood, P., "Stability and Control of the Wallis W117 Autogyro", Royal Aircraft
Establishment Technical Memorandum TM Aero. 1461, 1972.
[94] Arnold, U. T. P., "Untersuchungen zur Flugmechanik eines Tragschraubers", Technische
Universitat Braunschweig Diplomarbeit 88-5D, March, 1988.
[95] Anon, "Helicopter Simulator Qualification", US Dept. of Transport Federal Aviation
Administration Advisory Circular AC 120-63, Nov. 1994.
[96] Tischler, M. B. (ed.), "Advances in Aircraft Flight Control", Taylor and Francis, 1996.
[97] Tischler, M.B., System Identification Methods for Handling Qualities Evaluation,
Rotorcraft System Identification, AGARD LS 178, October 1991.
[98] "British Civil Airworthiness Requirements, Section T, Light Gyroplane Design
Requirements" Civil Aviation Authority Paper No. T 860 Issue 2, Jul. 1993.

[99] Houston, S. S., Review of BCAR Section T (issue 2), The Aerodynamics of
Gyroplanes CAA Contract No. 7D/S/1125, July 1997.
[100] O'Hara, F., "Handling Criteria", Journal of the Royal Aeronautical Society, Vol. 71 No.
676, April 1967.
[101] Anon, "Military Specification, Flying Qualities of Piloted Airplanes", MIL-F-8785C, Nov.
1980
[102] Anon, "Handling Qualities Requirements for Military Rotorcraft", US Army
Aeronautical Design Standard ADS-33D, July 1994.
[103] "British Civil Airworthiness Requirements, Section T, Light Gyroplanes" Civil Aviation
Authority CAP 643 Issue 3, Aug. 2005.
[104] Houston, S. S., "Identification of Autogyro Longitudinal Stability and Control
Characteristics From Flight Test", University of Glasgow Dept. of Aerospace
Engineering Report No. 9701, 1997.
[105] Anon, "Airworthiness Review of Air Command Gyroplanes", Air Accidents
Investigation Branch Report, Sept. 1991.
[106] Houston, S.S, Thomson, D.G., Airworthiness Investigation of the Stability and
Control of the Air Command Series of Gyroplane Aircraft, CAA Contract No. 7D/S/
1125/1, Final Report, April 1999.
[107] Houston, S. S., A Survey of Gyroplane Centre-of-mass and Propeller Thrust Line
Locations, Issue 2, Sept. 2005.
[108] Talbot, N., White, W.F.R., Montgomerie Bensen Autogyro Stability Investigation,
CAA Internal Report Ref. 9/50/5/C66/1, Issue 3, Nov. 2005.
[109] Chadwick, C. J., "Flight Test Report No. 2/93 on the VPM M16 Gyroplane G-BUPM
Against the Requirements of BCAR Section T". Arrow Engines UK Flight Test Report
No. 2/93. June 1993.
[110] Przybylski, J., Toms, R.L., Cheeseman, I.C., The Investigation of Some Unusual
Handling Characteristics of a Light Autogyro, Proc. of the 2nd ERF, 1976.
[111] Anon, Gyrocopter Rotor System Behaviour, Preliminary Aeronautical Report,
Commonwealth of Australia, Department of Civil Aviation, Nov. 1972.
[112] Trchalik, J., Gillies, E.A., Thomson, D.G., Development of an Aeroelastic Stability
Boundary for a Rotor in Autorotation, Proceedings of the AHS Specialists Conference
on Aeromechanics, San Francisco, Jan. 2008.
[113] Turnour, S. R., Celi, R., 'Modeling of Flexible Blades for Helicopter Flight Dynamics
Applications,' Journal of the American Helicopter Soc., Vol. 41, No. 1, 1996, pp.52-61
[114] Padfield, G. D., "On the Use of Approximate Models in Helicopter Flight Mechanics",
Vertica, Vol. 5, No. 3, 1981
[115] Houston, S. S., "Investigation of the Longitudinal Stability Characteristics of RAF 2000
GTX-SE Gyroplane G-BXDD", University of Glasgow, Dept. of Aerospace Engineering
Report No. 9718, Oct.1997.
[116] Chadwick, C. J., "Flight Test Report No. CJC 01/94 on Modified Air Command
Gyroplanes and the Modac 503 Gyroplane Against the Requirements of BCAR
Section T". 12 March 1994.
[117] Anon, "FRA Post Flight Report", No. M16-99-01

Appendix 1 Wind Tunnel Test Series
Table A1.1 Wind Tunnel Tests Conducted With Power On
MEASUREMENT
NUMBER
r COWLING TAIL
1 0 0 YES YES
2 -30 0 YES YES
3 -15 0 YES YES
4 -7.5 0 YES YES
5 7.5 0 YES YES
6 30 0 YES YES
7 15 0 YES YES
8 15 -20 YES YES
9 7.5 -20 YES YES
10 0 -20 YES YES
11 -7.5 -20 YES YES
12 -15 -20 YES YES
13 -15 20 YES YES
14 -7.5 20 YES YES
15 0 20 YES YES
16 7.5 20 YES YES
17 15 20 YES YES
18 15 10 YES YES
19 0 10 YES YES
20 -15 10 YES YES
21 -15 -10 YES YES
22 15 -10 YES YES
23 0 -10 YES YES
24 0 -15 YES YES
25 0 -5 YES YES
26 0 5 YES YES
27 0 15 YES YES
28 0 15 NO YES
29 0 10 NO YES
30 0 5 NO YES
August 2010 Appendix 1 Page 1

Table A1.1 Wind Tunnel Tests Conducted With Power On (Continued)
MEASUREMENT
NUMBER
r COWLING TAIL
31 0 -5 NO YES
32 0 -10 NO YES
33 0 -15 NO YES
34 0 -20 NO YES
35 -15 -20 NO YES
36 -7.5 -20 NO YES
37 7.5 -20 NO YES
38 15 -20 NO YES
39 15 20 NO YES
40 7.5 20 NO YES
41 0 20 NO YES
42 -7.5 20 NO YES
43 -15 20 NO YES
44 -15 0 NO YES
45 -7.5 0 NO YES
46 0 0 NO YES
47 7.5 0 NO YES
48 15 0 NO YES
49 15 0 YES NO
50 7.5 0 YES NO
51 0 0 YES NO
52 -7.5 0 YES NO
53 -15 0 YES NO
54 -15 20 YES NO
55 -7.5 20 YES NO
56 0 20 YES NO
57 7.5 20 YES NO
58 15 20 YES NO
59 15 -20 YES NO
60 7.5 -20 YES NO
61 -7.5 -20 YES NO
62 -15 -20 YES NO

Table A1.1 Wind Tunnel Tests Conducted With Power On (Continued)
MEASUREMENT
NUMBER
r COWLING TAIL
63 0 -20 YES NO
64 0 15 YES NO
65 0 10 YES NO
66 0 5 YES NO
67 0 -5 YES NO
68 0 -10 YES NO
69 0 -15 YES NO
70 0 0 NO NO
71 15 0 NO NO
72 30 0 NO NO
Table A1.2 Wind Tunnel Tests Conducted With Power Off
MEASUREMENT
NUMBER
r COWLING TAIL
73 0 -15 YES YES
74 0 -10 YES YES
75 0 -5 YES YES
76 0 -20 YES YES
77 -7.5 -20 YES YES
78 -15 -20 YES YES
79 15 20 YES YES
80 -7.5 20 YES YES
81 0 20 YES YES
82 0 0 YES YES
83 -7.5 0 YES YES
84 -15 0 YES YES
85 -15 0 NO YES
86 -7.5 0 NO YES
87 0 0 NO YES
88 0 20 NO YES
89 -7.5 20 NO YES

Table A1.2 Wind Tunnel Tests Conducted With Power Off (Continued)
MEASUREMENT
NUMBER
r COWLING TAIL
90 -15 20 NO YES
91 -15 -20 NO YES
92 -7.5 -20 NO YES
93 0 -20 NO YES
94 0 -15 NO YES
95 0 -10 NO YES
96 0 -5 NO YES
97 0 -5 YES YES
98 0 -15 YES YES
99 0 -10 YES YES
100 -15 -10 YES YES
101 -15 10 YES YES
102 -15 20 YES YES
103 -7.5 20 YES YES
104 0 20 YES YES
105 0 -20 YES YES
106 -7.5 -20 YES YES
107 -15 -20 YES YES
108 -15 0 YES YES
109 -30 0 YES YES
110 -7.5 0 YES YES
111 0 0 YES YES
Table A1.3 Additional Wind Tunnel Tests
MEASUREMENT
NUMBER
r COWLING TAIL POWER DESCRIPTION
112 0 0 YES YES NO Standing Propeller
113 0 0 YES YES YES Tail Boom Extended
114 0 0 YES YES NO No End Plates on Tail

Appendix 2 Configurational Data
VPM M16 Gyroplane

Airframe reference point is the intersection of the projection of the mast centreline and the
keel centreline with the x-body axis aligned with the keel. The rotor rotates in an anti-clockwise
direction when viewed from above. The following represents the default data set used in the
simulations presented in this report.
Gross Mass 450 kg CY (Side Area) -2.0

Moments of Inertia CZ (Plan Area) 2.0
IXX 195 kg m2
IYY 637 kg m2 Tailplane Data:
IZZ 4425 kg m2 Area 0.9 m2
IXZ -46 kg m2 Lift curve Slope 3.5 /rad
Setting Angle 0
Co-ordinates (in metres) for:
Nominal Centre of Mass (0.319, 0, -0.823) Fin Data:
Hub Plate Pivot Point (0.0969, 0, -2.105) Area 0.19 m2
Propellor Hub (-0.68, 0, -0.885) Lift curve Slope 3.5 /rad
Fuselage C.P. (1.2, 0, -0.3) Setting Angle 0
Tailplane C.P. (-1.52, 0, 0)
Fin C.P. (-1.4, 0, -0.3) Endplate Data:
End Plate C.P. (-1.91, 0.85, -0.13) Area 0.18 m2
Rudder C.P. (-1.74, 0, -0.39) Lift curve Slope 3.5 /rad
Setting Angle 0
Rotor Blade Parameters:
Radius 4.267m Rudder Data:
Chord 0.22m Area 0.34 m2
Mass 8.5 kg Lift curve Slope 3.5 /rad
Flapping Inertia 51.6 kg m2
Lift Curve Slope 5.7 /rad Propellor Data:
Profile Drag Coeff. 0.015 Blade Radius 0.86 m
Shaft Length 0.15m Blade Chord 0.1 m
Blade Twist 0
Fuselage Data: Blade Mass 0.9 kg
Side Area 1.4 m 2 Orientation of thrust 2
line
Plan Area 1.6 m2 c.g. normal distance 0.027m
from thrust line (below)
Frontal Area 0.59 m2
CD (Frontal Area) 1.0

Figure A2.1 The VPM M16 Gyroplane

Montgomerie-Parsons Gyroplane G-UNIV

Airframe reference point is taken as the intersection of the projection of the mast centreline
and the keel centreline with the x-body axis aligned with the keel. The rotor rotates in an anti-
clockwise direction when viewed from above. The following represents the default data set
used in the simulations presented in this report.
Gross Mass 355 kg Tailplane Data:

Moments of Inertia Area 0.356 m2
IXX 72.96 kg m2 Lift curve Slope 3.5 /rad
IYY 297.21 kg m2 Setting Angle 0
IZZ 224.25 kg m2
IXZ 0 kg m2 Fin Data:
Area 0.281 m2
Co-ordinates (in metres) for: Lift curve Slope 3.5 /rad
Nominal Centre of Mass (0.199, 0, -0.757) Setting Angle 0
Hub Plate Pivot Point (-0.038, 0, -1.968)
Propellor Hub (-0.91, 0, -0.795) Endplate Data:
Fuselage C.P. (1.626, 0, -0.48) Area 0.107 m2
Tailplane C.P. (-1.02, 0, -0.057) Lift curve Slope 3.5 /rad
Fin C.P. (-1.0, 0, -0.268) Setting Angle 0
End Plate C.P. (-1.09, 0.45, -0.063)
Rudder C.P. (-1.633, 0, -0.392) Rudder Data:
Area 0.368 m2
Rotor Blade Parameters: Lift curve Slope 3.5 /rad
Radius 3.81m
Chord 0.197m Propellor Data:
Mass 17.255 kg Blade Radius 0.787 m
Flapping Inertia 83.492 kg m2 Blade Chord 0.09 m
Lift Curve Slope 5.75 /rad Blade Twist 0
Profile Drag Coeff. 0.011 Blade Mass 0.9 kg
Shaft Length 0.137m Orientation of thrust 1.0
line
Shaft Offset 0.025m c.g. normal distance 0.018m
from thrust line (below)
Fuselage Data:
Side Area 0.798 m2
Plan Area 0.916 m2
Frontal Area 0.448 m2

Figure A2.2 The Montgomerie-Parsons Two Persons Gyroplane

Appendix 3 Experimental Measurement of Centre of

Gravity Position
Since the vertical location of the centre-of-gravity has been highlighted as an important design
parameter for light gyroplanes, it is appropriate to consider a suitable method for determining
this quantity. The method described here is easy to perform and provides robust results most
importantly error bounds on the calculated values are easily determined. Note that
determination of c.g. position is not to replace the hang check the hang check is really
performed to determine if suitable control margins will be available, and is not really a weight
and balance exercise in itself. A weight and balance exercise is therefore an additional item to
be performed.
Equipment required includes three weight scales that each main wheel, plus nose (or tail)
wheel is to rest upon. A tape measure is required to determine wheelbase, and a clinometer
is necessary for measuring the pitch angle of the aircraft. The method is performed in two
separate stages. First, the aircraft is placed on the scales as shown in Figure A3-1, on a level
surface. The position of the vertical reference line is arbitrary, and is a matter of choice. For a
symmetric aircraft, the left and right wheel reactions should be the same if they are not,
advice should be sought. The wheelbase should be measured, and then the longitudinal
position of the c.g., with respect to the vertical reference line, is given by:
xcg = Wnose l
W
Figure A3.1 Measurement of Longitudinal c.g. Position

Note that W is the weight of the aircraft i.e. WL+WR+Wnose.
The nose of the aircraft should then be raised so that the keel is inclined no less than 5 deg to
the level surface, and preferably 10 deg. Note that the scales should not be inclined, and are
to remain level. From geometry shown in Figure A3.2, the vertical location of the c.g., relative
to the original level surface, is given by:
zcg = l(Wnose W nose)

W tan

where is the angle by which the keel has been raised. Wnose is the nose wheel reaction from
stage 1, and Wnose is the nose wheel reaction from the inclined test. It can be seen that only
a few simple measurements need to be taken. However, the above equation emphasises the
need for accuracy, especially in the measurement of the wheelbase. This is because of the
tanq term. For inclined angles of around 5 deg, a 1 mm error in wheelbase will produce a 1 cm
error in zcg. For incline angles of 10 deg, this error is reduced by half, consequently angles
greater than 10 deg. are desirable, however in practical terms aircraft handling can become
difficult above this angle.
Figure A3.2 Measurement of Vertical c.g. Position

Appendix 4 Aircraft Dynamic Stability
The aim of this Appendix is to give the reader who is unfamiliar with theoretical flight dynamics
some background and insight into the subject. For a more complete treatment, the reader is
referred to the various textbooks on the subject which are widely available.
1 Aircraft Nonlinear Equations of Motion
A conventional fixed wing aircraft has 6 degrees of freedom (3 translational and 3

angular), and therefore, as described in Section 4 of this report, the aircraft nonlinear
body equations of motion are given in the form of the Euler equations:
m(U QW VR) X mg sin 4

m(V RU PW ) Y mg cos 4 sin )
m(W PV QU ) Z mg cos 4 cos )
I xx P I xz R QR( I zz I yy ) PQI xz L
I yy Q RP( I xx I zz ) ( P 2 R 2 ) I xz M
I zz R I xz P PQ( I yy I xx ) QRI xz N
(A4.1)
In addition, the rotational kinematics are given by the expressions:
& = P + Q sin tan + R cos tan

& = Q cos R sin

& = Q sin sec + R cos sec

(A4.2)
The various symbols used, and the sign convention for the state variables are given
in convenient form in Table A4.1 and Figure A4.1.
For rotorcraft applications other degrees of freedom must be added: rotor speed,
rotor inflow and most significantly, the blade degrees of freedom (flap, teeter, lag and
pitch).

Table A4.1 State Variables for a Conventional 6 Degree of Freedom Aircraft Model
Aircraft TRANSLATIONAL ANGULAR

Axis
Motion Velocity force Motion Displacement Rate Moment
xb Fore/Aft U X Roll P L
yb Sidewards V Y Pitch Q M
zb Heave W Z Yaw R N
V
y
b
Z
U L
xb P
N
zb
Figure A4.1 Conventions for Body Axes Set and State Variables
The mathematical model can then consist of a large set of nonlinear equations, often
expressed in general form as:
x f ( x, u )

(A4.3)

where x is the state vector which, for a conventional aircraft, will consist of the

body states:
x

>UVW PQR)4@T
but supplemented by the various rotor states for a rotorcraft, for example:
x

>UVW PQR)4:v v 0 1s 1s @
v E1
T
The control vector, u , is given for an aircraft with conventional controls (throttle,

elevator, aileron, rudder) as:
u

>G t G eG aG r @T
For a gyroplane with controls consisting of throttle, longitudinal and lateral stick, and
rudder, we write the control vector as:
u

>G tKsKcG r @T
The real modelling effort is in deriving analytical or empirical expressions for the
external forces and moments (X, Y, Z, L, M, N). It is in the way these are calculated
(and of course in the nature of the parametric data used) that different configurations
may be simulated using basically the same equations. Equations (A4.3) are most
often used to calculate the response of the aircraft, x , due to a single or series of

perturbations of the control vector, u . Due to the complexity of the equations, a

numerical algorithm such as the Runge-Kutta method is used for this calculation.
2 The Linearised Equations of Motion
The usual approach to determining the dynamic characteristics of a vehicle is to

consider the linearised representation of the equations of motion. In these equations
each state variable is considered to be composed of an equilibrium (trim) component
and a small perturbation component, i.e. U = Ue + u etc. The linearised equations
become functions of the perturbations rather than the full values of the states. The
major limitation of the linearised equations is that they are only valid for small
perturbations from trim. For a fixed wing aircraft it is possible to linearise analytically,
and for most aircraft types it is possible to decouple longitudinal and lateral directional
dynamics. Written in state-space form:
x A x Bu
(A4.4)

the two sets of equations are firstly, for longitudinal flight:
u Xu Xw 0 g cos 4 e u 0
w Z Zw Ue g sin 4 e w Z G e
u >G e @
q M u M w M q M T q M G e

T 0 0 1 0 T 0
and then for the lateral/directional motion:
v Yv 0 Ue g cos 4 e 0 v YG r 0
p L L
L
0 0 p L G r L G a
v p r
G r
r N v N
N
0 0 r NG r N G a
Ga
p r

I 0 1 tan 4 e 0 0 I 0 0
\ 0 0 sec 4 e 0 0 \ 0 0
The terms in the equations such as Xu, Xw etc. are known as the "stability derivatives",
and are defined as:
X X
Xu = ,X w = etc.
u w
These are calculated (analytically or numerically) for a particular trim state and
mathematically they are the first order terms in a Taylor series expansion representing
an external force or moment (for example, X = Xe + Xu u + Xw w + .). Physically
they also have significance; for example Xu is the change in X-force experienced due
to a small change in velocity, u. It is known as the "speed damping" derivative and
plays an important role in the damping of the phugoid. An aircraft with a particularly
small value for Xu will have insufficient phugoid damping, or even an unstable
phugoid. Many of the other derivatives have important roles in stability analysis, and
the reader is again referred to the standard textbooks for further information.
For the rotorcraft problem, the complexity of the forces and moments calculations,
and the very strong coupling between longitudinal and lateral directional motion
makes analytical derivation impossible, and a numerical method is used instead. To
represent the longitudinal dynamics of the light gyroplane it is found that the rotor
speed degree of freedom has to be included. The equations of motion can be written
as:
u Xu Xw Xq XT X : u X Ks
w Z
u Zw Zq ZT Z : w ZK s

q M u Mw Mq MT > @
M : q M K s K1s

T 0 0 1 0 0 T 0
: Qu Qw Qq QT Q: : QK s

(A4.5)

3 Prediction of Dynamic Stability
Consider the state space equation, equation (A4.4) applied for the stick fixed situation
(i.e. u = 0). We then have:

x Ax
(A4.6)
This equation represents a system of n first order, linear ordinary differential
equations where n is the number of states and therefore degrees of freedom. This
will have a general solution:
xi x0i e Ot
(A4.7)
for i = 1, n, or in vector form:
x x0 e Ot
differentiating this and substituting in (A4.6):
O x0 e Ot A x0 e Ot
Noting that this is a vector-matrix equation we can write:
OI A x0 0
where I is the (nxn) identity matrix. In this expression OI A is an (nxn) matrix, while
x0 is a vector of dimension (n). This equation has a trivial solution: x0 = 0, or the more
useful solution that the determinant of OI A should be zero:
OI A 0
(A4.8)
Equation (A4.8) is the characteristic equation which will be a polynomial of order n,
and on factorisation its solution will yield n roots, or "eigenvalues", from which the
stability of the system can be determined. Consider the general case of an eigenvalue
being a complex number:
= Re() r i Im()

Substituting this into the general solution given by equation (A4.7):
x x0 e Ot x0 e (Re( O ) i Im(O ))t x0 e Re( O ) t e i Im(O ) t
or
x x0 e Re( O ) t cosIm(O )t r i sin Im(O )t

(A4.9)
The response will therefore be oscillatory in nature with the imaginary part of the
eigenvalue giving the frequency of the oscillation. If the real part of the eigenvalue is
positive then the amplitude of the oscillations will increase exponentially, an unstable
situation. Should however, the real part of the eignevalue be negative, then the
amplitude will decay a stable situation. Should the eigenvalue be real, then the
solution is simply:
x x0 e Ot
and the characteristic response will be exponential in form. If O is negative then after
a disturbance the aircraft's response will decay tending to return it to a steady state
condition (a stable situation). Should O be positive then the disturbance will grow and
we have an unstable situation.
It is clear then that once the linearised equations are written in state space form,
(A4.4), then it is possible to obtain the characteristic equation from (A4.8), the solution
of which will give the aircraft's eigenvalues which can then be interpreted in order to
predict its stability. The nature of the response can be quantified by calculating the
damping and natural frequency of the mode from its eigenvalue. Consider a general
eigenvalue:
O V r iZ
(A4.10)
This will have been derived from a quadratic factor of the characteristic equation
(A4.8):
O2 2VO V 2 Z 2 0
(A4.11)
that is:
O V iZ O V iZ O2 2VO V 2 Z 2

Equation (A4.11) can be compared to the characteristic equation of a general 2nd order
damped system. Consider first the general equation of motion for a damped 2nd order
system:
x 2]Z n x Z n2 x 0
(A4.12)
where is the damping ratio and n is the undamped natural frequency. The
undamped natural frequency is the frequency at which the system would oscillate
without the presence of damping (essentially the frequency in simple harmonic
motion). Making the substitution x x0 e Ot again, we get the characteristic equation:
O2 2]Z n O Z n2 0
This gives: Z 2n V 2 Z 2 and ]Z V , or:
V
Zn V 2 Z 2 and]
V 2 Z2
(A4.13)
It is therefore apparent that the damping and natural frequency associated with a
dynamic mode can be derived directly from the real and imaginary parts of its
eigenvalue. As there is damping present we require the damped natural frequency,
d, which in general is given by:
Zd Zn 1 ] 2
or
Z d2 Z n2 1 ] 2 Z n2 Z n2] 2
Noting that ] 2 Z 2n V 2 , we can write:
( )
2d = 2 + 2 2 = 2 or d =

Hence, when a complex conjugate pair of eigenvalues O V r iZ is present, the

frequency of the resulting oscillation is given by the imaginary part of the eigenvalue,
. This is also apparent on inspection of the general solution given by eqn (A4.9). The
period of the oscillation is therefore given by:
2S 2S
T
Z Re(O )
The way in which eigenvalues can be interpreted with respect to aircraft stability can
be summarised, with reference to Figure A4.2, as follows:
Real and Negative : This gives an aperiodic mode with an exponential decay
or a "Convergence". This situation is indicative of static
stability.
Real and Positive : This gives an aperiodic mode with exponential growth
or a "Divergence". This situation is indicative of static
instability.
Complex -ve Real Part : This gives an oscillatory motion with a decreasing
amplitude. This situation is indicative of dynamic
stability.
Complex +ve Real Part : This gives an oscillatory motion with an increasing
amplitude. This situation is indicative of dynamic
instability.

4 Damping Ratio,
Any force (or moment) acting on a system which is proportional to the velocity of the
system is referred to as a "damping". The damping ratio relates the effective damping
of the system, c, to its critical value:
c
=
ccrit
Figure A4.2 Characteristic Motions for Dynamic and Static Stability (Reproduced
and Modified from Figure 6.1: "Dynamics of Flight", B. Etkin, 2nd
Edition, Wiley, 1982)
The critical value of damping is best explained by considering the solution of (A4.12).
The solution of this differential equation is dependent on the value of the damping
ratio .
i) < 1 (Underdamped Damped System)
This is the most common situation, and the solution of equation (A4.12) for an initial
displacement x0 is:
]
x(t ) x0 e ]Z nt cos Z d t sin Z d t
1] 2

This gives an oscillation with damped natural frequency, Zd, given by:
Zd Zn 1 ] 2
but with decaying amplitude, Figure A4.3
Figure A4.3 Free Vibration of an Underdamped 2nd Order System

(x0 = 1.0, n = 5 rad/s, = 0.1)
ii) ] > 1 (Overdamped System)

Here the solution is:
]Z n t Z n ] 2 1 t Z n ] 2 1 t
x(t ) x0 e Ae

Be

where A and B are constants related to the damping ratio, . This gives an exponential
response a indicated in Figure A4.3.
iii) = 1 (Critically Damped System)
Critical damping is defined as = 1 or c = ccrit. For this particular case the response
is again exponential and is given by:
x(t ) x0 (1 Z nt )e Z nt

x ( t ) = x0 (1 + nt )e n t
Figure A4.4 Effect of Damping Ratio on System Response

(x0 = 1.0, n = 5 rad/s)
The effect of damping on a second order system is presented in Figure A4.4. For all
values of < 1 the response will be oscillatory. The critical value is = 1 after which
the response is non-oscillatory and has exponential form. For stable, lightly damped
systems (say 0 < < 0.2) any disturbance will result in an oscillation which can take
several cycles to settle, whilst moderately damped systems (say, 0.3 < < 0.6) may
only take a few cycles before settling on a new steady state. For heavily damped
systems the nature of the response is determined by whether the system is under-
damped or over-damped. For an under-damped system, in response to a disturbance,
a heavily damping system (say, 0.7 < < 1) would have a response which had perhaps
only one or two cycles before settling. As damping is increased beyond the critical
value (i.e. ] > 1), the response type is exponential, and as damping is increased, the
speed of the response drops so that systems with >1 take a significant time to
settle.
5 Representing Stability Characteristics on the Root Locus Plot
An aircraft's stability properties can vary with flight condition or with a configurational
parameter (c.g. position or a control system gain, for example). Mathematically, the
values of the stability derivatives change with flight condition and configurational
parameter, and consequently so do the eigenvalues. As the eigenvalues are often
complex numbers, it is convenient to plot them on a complex plane, Figure A4.3. The
convention is to label the axes as the Real (part of the eigenvalue) versus the
Imaginary (part of the eigenvalue). Referring to equation (A4.10) essentially, once the
eigenvalue is calculated the real part is plotted against the imaginary part . The
plot of a particular eigenvalue as a flight state or configurational parameter is varied is
known as its "root locus". Any eigenvalue with a negative real part is of course stable,

and so any eigenvalue to the left of the imaginary axis indicates stability, any
eigenvalue to the right (positive real part) instability. Any real eigenvalue will lie on the
real axis, and again if this is to the left of the imaginary axis the mode is stable. Any
complex eigenvalue appears as a conjugate pair and therefore the root locus diagram
is symmetrical about the real axis, however conventionally only the top half of the
plane is drawn. Examples of a root loci plots are given throughout this report, for
example, Figure 4.3.
As demonstrated above, equation (A4.13), the damping and natural frequency of the
mode may be derived from the real and imaginary parts, and so it is also possible to
plot natural frequency vs damping ratio. It should be appreciated however that only
complex eigenvalues (oscillatory modes) can be plotted, unlike the conventional root
locus which includes real eigenvalues. Further, as a positive damping ratio indicates
stability (that is decaying amplitude of oscillation), when eigenvalues are plotted as
frequency vs damping ratio, values lying to the right hand side of the frequency axis
indicate stable modes.
Figure A4.5 Stability Presented on Complex Plane

6 Quantitative Example
Referring to Section 4 of this report, the use of the RASCAL model is described, and
calculations for the VPM M16 are presented. Trim calculations and time response
results are shown. Here, for clarity, the way in which the eigenvalues for Figure 4.3
and 4.4 are calculated is presented. The model structure used is as in equation (A4.3),
and expressing this in the form of equation (A4.4),that is for stick fixed flight, we have:
u Xu Xw Xq XT X : u
w Z
u Zw Zq ZT Z : w

q M u Mw Mq MT M : q

T 0 0 1 0 0 T
: Qu Qw Qq QT Q: :

Substituting the derivative values from the linearised RASCAL model for the 30mph
flight condition we get:
u 0.141 0.160 0.191 9.090 0.050 u

w 0.164 0.903 13.324 0.903 0.287 w

q 0.018 0.063 1.597 0.306 0.009 q

T 0 0 1 0 0 T
: 0.137 0.631 1.464 0.001 0.155 :

Applying equation (A4.8) we get the 5th order polynomial, characteristic equation:
O5 2.796O4 2.8825O3 0.8643O2 0.1961O 0.07387 0

When factorised this becomes:
O2

2.4408O 1.9625 O2 0.078O 0.0869 O 0.4331 0
giving the eigenvalues:
= 1.2204 0.6876i
= 0.039 0.2922i
= 0.4331
The real part will have units of s-1, whilst the imaginary part (i.e. the frequency) has
units of rad/s. In referring to BCAR Section T where criteria are expressed in cycles/s
the frequency is converted by dividing by 2, and hence the final form of the
eigenvalues is:

= 1.2204 0.1094i
= 0.039 0.0465i
= 0.4331
which correspond to the points on Figure 4.3 marked for the 30mph case.
The first of these eigenvalues is stable (real part is negative) and corresponds to the
short period mode. This mode is characterised by a heavily damped pitching
oscillation. In this case, the natural frequency, damping, and damped natural
frequency are calculated as follows:
Z n2 1.9625,? Z n 1.4rad/s ( 0.1094cycles/s)

2]Z n 2.4408,?] 0.8717
Zd Z n 1 ] 2 ,? Z d 0.686rad/s (or Z d Re(O) )
The second mode is unstable (real part is positive) and corresponds to the phugoid
which is a longer period oscillation in airspeed and altitude. The characteristic
parameters are:
Z n2 0.0869, ?Z n 0.2948rad/s ( 0.0465cycles/s)
2]Z n 0.078, ?] 0.1323
Zd Zn 1 ] 2 , ?Zd 0.2922rad/s (again, Z d Re(O) )

The third mode is a rotor speed mode which is stable (negative real part). Aperiodic
modes are characterised by the time to half amplitude (for a stable mode) or time to
double amplitude (for an unstable mode). Here we have:
0.69
t1 ,? t 1 1.593s
2 Re(O ) 2
The above calculation was repeated at a series of flight speeds to give the set of
eigenvalues plotted on Figure 4.3 which shows the root locus of the aircraft over its
normal flight range. Both the rotor speed and short period modes remain stable
across the speed range, however the phugoid is unstable at low speed but stable at
higher speeds (the root locus showing that the eigenvalue migrates from the right to
the left hand half plane as airspeed increases).
The alternative presentation of the eigenvalues where frequency is plotted against
damping ratio is presented in Figure 4.4. At 30 mph the phugoid mode, from above,
gives the point (],Z n ) as (-0.1323, 0.0465) which is unstable (negative damping
indicating increasing amplitude) whilst the short period mode at 30 mph gives the
point (0.8717, 0.1094) which lies on the right hand side of the axis, which is stable for
this format of plot.

Appendix 5 Flight Test Instruction for Teeter Trials
Civil Aviation Authority

Flight Test Instruction
Investigation of Gyroplane Teeter Margin Limits
Montgomerie-Parsons Gyroplane G-UNIV
Prepared ------------------------------- Date ----------------

W F R White (Flight Test Engineer)
Checked ------------------------------- Date ----------------

Dr S Houston (Trials Sponsor)
Checked ------------------------------- Date ----------------

B de Saar (Trials Pilot)
Authorised ------------------------------- Date ----------------

N Talbot (Deputy Chief Test Pilot)
Distribution
Glasgow University
Dr S Houston
CAA
Mr N Talbot
Mr G Greene
Mr A Maxwell
Mr K Dodson
Mr B de Saar
FR Aviation
Mr C Handley
1 Introduction
An investigation into gyroplane stability by Glasgow University has been conducted

using the RASCAL simulation model, validated by flight trials of 2 instrumentated
gyroplanes, VPM M16, G-BWGI and Montgomerie-Parsons, G-UNIV. The
investigation has concluded that degraded longitudinal stability characteristics are to
be expected, should the thrust line be more than 2" higher than the vertical CG for
contemporary light gyroplanes.

The model validation is, however, deficient in that it was not possible to instrument
the rotor teeter (flapping) angle on either aircraft.
Review of BCAR Section T has identified that the definition of VNE in terms of control
margins is less than ideal, and compliance demonstration is very difficult, if not
impossible without specialised instrumentation.
AMC T143: The gyroplane's VNE will normally be dictated by the need to have a
positive teeter margin up to VNE + 15%.
Accordingly, a further flight trial has been proposed to investigate the teeter behaviour
of the McCutcheon rotor of Montgomerie-Parsons, G-UNIV, the aim being:
a) To validate the model teeter behaviour with actual flight test data.
b) Review the results of the research investigation with the enhanced model.
c) To use the model with confidence to predict the stability characteristics of other
light gyroplanes.
d) To support a technical review of the BCAR Section T definition of VNE and
associated control margins.
2 Aircraft Details
The trials aircraft will be the Glasgow University owned Montgomerie-Parsons

2-place, modified autogyro, G-UNIV.
The aircraft is equipped with a standard "Merlin" type cockpit nacelle for the forward
seat. AAN27261 defines the installation of an instrumentation package. The rear seat/
fuel tank has been replaced by an alternative tank with a faired mounting for a laptop
computer which is used for flight test data acquisition.
A 25 ft McCutcheon rotor is installed. Engine is a Rotax 618, 75 hp, and propellor is
an Ivoprop 3 blade 60".
3 Instrumentation
The following data are recorded on the on-board laptop PC:

- Control positions, longitudinal, lateral and yaw
- Pitch, Roll and Yaw attitudes
- 3 axis body rates and accelerations
- GPS position and x, y velocities
- Airspeed
- Altitude
- OAT
- Angle of attack
- Sideslip angle
- Rotor rpm
- Engine rpm
- Event number
- Voice

For this trial, a rotor teeter angle potentiometer is installed with the data transmitted
over slip rings to the recording system. 360 rotor azimuth is also recorded.
The teeter angle instrumentation has been developed by Glasgow University with the
support of the gyroplane manufacturer, JM Montgomerie, who has also been
responsible for the reassembly of the modified rotor head.
4 Personnel
CAA WFR White (Flight Test Engineer and co-sponsor)

N Talbot (Deputy Chief Test Pilot)
CAA (under contract) B de Saar (Trials Pilot)
Glasgow University Dr S Houston (Trials Sponsor)
Dr D Thomson (Head of Dept.)
Mr R Gilmour (Responsible for Instrumentation Design and
installation)
5 Operational Aspects
5.1 Trials Location

The trials will be conducted from Perth Aerodrome.
5.2 Trials Area
Test flights will be conducted away from built up areas, but where observation from
the ground, by means of binoculars, as required to monitor aircraft behaviour and
assist with pilot evaluation, is possible.
5.3 Communications
To facilitate the conduct of the trial, a dedicated trials frequency will be required.
5.4 Weather
Tests will be conducted in VMC by day only in as calm conditions as possible. No
flights will be conducted in precipitation. Minimum cloudbase 1000 ft, minimum
visibility 5 km, with good visual horizon.
5.5 Trials Management
Each test flight will be briefed thoroughly beforehand. A flight card (Figure A5-1) will
be prepared during each briefing.
5.6 Trials Pilot
The trials pilot, Mr de Saar, is an experienced gyroplane pilot and instructor who has
been given basic training in flight test techniques and who has already conducted a
number of supervised handling and stability investigations on Bensen and Air
Command gyroplanes.
Basic serviceability checks of the aircraft and instrumentation system prior to
commencement of actual data gathering will ensure a suitable level of familiarisation
with the type.

5.7 Aircraft Servicing

Normal aircraft serviceability checks will be performed by Mr de Saar as approved by
the CAA AMSD. In the event that major maintenance action is required, additional
approved engineering personnel will be engaged as required.
5.8 Insurance
No hull insurance is available. Third party liability is currently limited to 250,000.
6 Limitations
6.1 Tests will be conducted within the normal approved flight envelope of the aircraft,
except that flight to VDF (1.11VNE) may be investigated.
6.2 The programme will be conducted incrementally. Basic analysis of data from each
flight will be conducted before continuing the programme.
6.3 Sideslip tests will initially be limited to 10. Following preliminary analysis of flight
test data, consideration will be given to progressively increasing the sideslip angle,
particularly at lower airspeeds.
7 Safety of Flight Test
7.1 All test flying will be carried out in an incremental manner.

7.2 All manoeuvres involving a possible reduction of normal 'g' must be carried out in a
very progressive manner. Account must be taken that nose up control inputs could
lead to reductions of normal 'g', e.g. nose up input to establish phugoid behaviour is
likely subsequently to result in nose down pitch rates greater than that generated by
the initial nose up input.
7.3 Manoeuvres must be stopped if a noticeable reduction in normal 'g' or a rapid
reduction in rotor rpm is perceived by the pilot.
7.4 Manoeuvres must be stopped if high or rapidly increasing pitch rates occur,
particularly nose down.
7.5 Extreme attitudes are to be avoided. For guidance, pitch attitudes beyond 30
degrees with low pitch rates should be considered excessive. Lower pitch attitudes
could be excessive with high pitch rates. Roll attitudes of greater than 60 degrees
should be considered excessive
8 Test Programme
8.1 Datum Flying

A brief handling qualities evaluation will be conducted to assess the effect of the
modified rotor head. This phase of testing will also be utilised to optimise the
instrumentation system.
8.2 Data Gathering
The following flight conditions will be investigated:
8.2.1 Airspeed Calibration
The aircraft is equipped with an airspeed probe mounted on a boom. This will be used
to calibrate the standard airspeed indicator during level flight constant speed runs.

8.2.2 Steady State Conditions

a) Zero sideslip, trimmed, Vmin to VNE, in 5 mph increments.
b) Min power speed, zero sideslip.
c) Entry to, steady state and recovery from 30 turns at Vmin power and 65 mph.
d) Entry to, steady state and recovery from 10 turns at VNE.
e) Max power climb, 50 mph.
f) Cruise climb, 65 mph.
g) Cruise descent, 65 mph.
h) Idle descent, 65 mph.
i) Idle descent, 50 mph.
j) Min. control speed, idle power.
8.2.3 Throttle slam
From 65 mph and Vmin power, with 1 sec pilot intervention time, and progressively
increasing rates of power application.
8.2.4 Throttle Chop
From 65 mph, 40 mph and Vmin with 1 sec pilot intervention time and progressively
increasing rates of power reduction.
8.2.5 Phugoid
At Vmin power, 60 and 65 mph, IAS. The phugoid will be investigated by means of an
approximately 5 pitch disturbance, returning the cyclic to the trim position and
recording the open loop response of the aircraft, stick fixed. Nose up inputs will be
made first at each trim speed, and nose down inputs will be made with caution, and
consideration of the response to the nose up input. In each case, the flight safety
guidance detailed in paragraph 7 of this document will be observed.
8.2.6 Control response
Control response tests will be conducted at Vmin power, 60 and 65 mph, IAS. An aft
cyclic step input will be introduced and maintained, all other controls remaining fixed,
and the aircraft response observed and recorded. Initially, the input will be
approximately 1 cm (0.4"), sufficient to generate a pitch rate of approx 5/sec. The
amplitude of the cyclic input will be increased incrementally to generate an adequate
pitch rate. In each case, the flight safety guidance detailed in paragraph 7 of this
document will be observed particularly in determining the criteria for implementation
of recovery action.
The tests will be repeated for left and right cyclic step inputs in turn.
Following analysis of aircraft response to the nose up inputs, consideration will be
given to a cautious, progressive examination of the response to forward cyclic inputs.
8.2.7 Steady Heading Sideslips
Steady heading sideslips will be examined at Vmin power, 60 and 65 mph IAS. Initially,
5 sideslip angles will be investigated and analysed before progressing to 10.
Following preliminary analysis of results, further testing may be conducted at greater
sideslip angles. (see paragraph 7.4)

8.3 Additional Tests

Following analysis of the data gathered in paragraph 8.2, consideration will be given
to investigating flight to VDF (1.11 VNE), and also turns on one control, pedal only.
8.3.1 VDF
Flight to VDF will only be undertaken following satisfactory exploration of the normal
operating envelope to VNE. In particular, satisfactory handling characteristics and
adequate teeter margin will have been demonstrated.
For the actual test, the aircraft will be accelerated in level flight to the lesser of VNE or
VH (maximum continuous power). Airspeed will then be progressively increased,
initially by further throttle opening if available, then in a shallow dive to stabilise at VDF.
An instrumentation record will be taken and then a gentle deceleration to less than
VNE will be conducted by raising the nose as required and gently reducing power.
8.3.2 Turns on 1 Control
Turns on pedal only may be carried out to examine the rolling moment due to sideslip
and the transient effect on lateral teeter margin. This test will only be conducted if
examination of steady heading sideslips shows adequate lateral steady state teeter
margin.
The aircraft will be stabilised at 60 mph level flight with zero sideslip. A pedal input
sufficient to change heading by approx 5 - 10 will be introduced and maintained.
Cyclic will be maintained in the trim position. The condition will be maintained until
either a stable condition is achieved, or until flight safety considerations such as
excessive roll or pitch attitude require a recovery. An instrumentation record will be
taken throughout the manoeuvre. The test will be repeated for the opposite pedal
input, and again at 65 mph and Vmin power.

CAA FLIGHT TEST CARD
A/C TYPE: Montgomerie Parsons Reg: G-UNIV
DATE FLIGHT
OBJECTIVE ...........................................................................................................................................................................
ZFW: TO FUEL:
AUM CG:
CONFIGURATION.................................................................................................................................................................
No. TEST DESCRIPTION/REMARKS
FLIGHT TEST ENGINEER TEST PILOT
TRAILS SPONSOR
Figure A5.1 CAA Flight Test Card

Appendix 6 Kinematic Consistency
This Appendix is not intended as a tutorial on methods for analysing flight test data. The reader
is referred to the excellent AGARD LS178 "Rotorcraft System Identification", [80] for detail on
rotorcraft-specific applications. It is sufficient here to illustrate the approach taken by means
of a trivial example.
If only pitching motion occurs, Euler's equations for kinematic angular motion reduce
intuitively to:
T q
i.e. the rate of change of pitch angle is equal to the pitch rate. Given pitch rate from the rate
gyroscope, integration of this signal should yield the pitch signal measured by the attitude
gyroscope. If it does not, the data is kinematically inconsistent, and corrections need to be
applied to the measured signals to render them consistent. This can be formalised as a least-
squares optimisation problem, if the kinematic equation above is re-written as:
T kq G
where k is a scale correction factor, and G is a bias or offset correction. For a given k and G ,
integration of this equation yields Tint, the pitch attitude calculated from the rate information.
Now if an error function is constructed as:
= int m
where Tm is the attitude gyroscope pitch information duly corrected itself for scale factor and
bias, then a simple least-squares algorithm can be used to calculate the bias and scale factors
required to minimise (to zero hopefully) the error HT.
A similar process can be constructed for the air data and accelerometers - assuming still air
conditions, integration of accelerometer information will give velocity components in aircraft
body axes. The airspeed, sideslip and angle of attack can be used to calculate explicitly the
corresponding velocity components, and exactly the same process as that described above
can be applied to calculate scale factor and bias error corrections for these sensors.
Figure A6.1 shows attitude gyro roll and pitch outputs as measured during a phugoid test
(70mph) using the VPM aircraft, and once corrected by the process described above. Of note
is the fact that the scale factor in each case is close to 1, the principal corrections to render
consistency with the rate information being to apply bias corrections. In roll, this gives an
attitude more "left wheel down", and although difficult to see from the noisy pitch information,
a more nose-down attitude. (The noisy pitch attitude was attributed to the gyroscope
mechanism requiring maintenance).
All data used for analysis and model validation was subjected to kinematic consistency
checking. For the purposes of demonstration only the technique for consistency checking
angular parameters is presented here for clarity. The velocities and accelerations were treated
in a similar manner. Results for all events in one particular flight are presented in Figures A6.2
and A6.3. In Figure A6.2, scale factor and bias errors are shown for the three rate gyroscopes.
Note that most scale factors are clustered around 1, although most of the roll rate scale factors
lie around 0.8. Bias errors are small, except in yaw rate, particularly towards the end of the
flight. Attitude scale factors and biases are shown in Figure A6.3. Given the earlier statement
regarding the noisy pitch attitude signal and the maintenance state of the attitude gyroscope

mechanism it is perhaps not surprising that the pitch attitude scale factors are not clustered
around 1, although they are consistently about 0.6. Bias errors are consistently small. Figure
A6.4 shows measured and corrected rates. Roll and pitch axes appear identical. The relatively
large yaw rate bias error manifests itself in this case as a small shift in the yaw rate signal.
AGARD LS178 makes interesting reading regarding kinematic consistency. Results (in terms
of biases and scale factor variations, event-to-event) can, in general, be variable and
inconsistent. The problem is basically over-parameterised, i.e. so much flexibility is given to
the matching process, least-squares or otherwise, that anything can be fitted to anything else.
Techniques much more sophisticated than simple least squares can be employed, such as
maximum likelihood methods, but experience has shown that this does little to improve
confidence in the results. Indeed, one contributor to LS178 who employed a sophisticated
matching algorithm, suggests that it is probably appropriate to filter data, and not apply
kinematic consistency checking at all. Another reports that it is necessary to assume some
scale factors are one, and biases zero, to get good, confident estimates of the remainder.
The view the authors take here is that if the basic data quality is good, and sound engineering
judgement can be applied then meaningful results can be achieved from kinematic
consistency checking. This despite its limitations and pitfalls and even using a simple algorithm
as was the case here. It is judged that the results suggest that basic data quality is indeed
good, but scale factors and biases as calculated need to be applied to the angular information.
Figure A6.1 Attitude Angles During Phugoid Flight Test (70mph, VPM)

Figure A6.2 Summary of Rate Gyro Consistency Results (VPM)

Figure A6.3 Summary of Attitude Gyro Consistency Results (VPM)

Figure A6.4 Attitude Rates Recorded During a fore/aft Doublet Input (VPM)

Appendix 7 Data Analysis and Model Synthesis
The model structure for which coefficients are to be identified, is of conventional state-space
form, i.e.
x A x Bu
(A7.1)
where
Xu Xw Xq XT X: X Ks
Z Zw Zq ZT Z : Z
u Ks
A M u Mw Mq MT M : , B M Ks

0 0 1 0 0 0
Qu Qw Qq QT Q: QK
s
(A7.2)
and
x = [u w q T ]T , u = [Ks]
(A7.3)
This constitutes the longitudinal subset of the conventional 6 degree-of-freedom rigid-body
flight mechanics model, with the important (and unique) addition of the rotorspeed degree of
freedom. The rigid body states are taken to be with respect to a mutually orthogonal, right-
handed frame of reference whose origin is at the centre of mass. The longitudinal and vertical
axes are respectively parallel and normal to the keel of the aircraft.
The angular quantities in the state vector, and the control position, are all measured directly.
The translational velocities and are obtained from airspeed, sideslip and angle of attack data
measured at the nose-mounted boom, as follows:
u = uprobe q(zvane zcg) + r(yvane ycg)

w = wprobe p(yvane ycg) + q(xvane xcg)
(A7.4)
and
V f cos vane
uprobe = ; wprobe = u probe tan probe
1+ tan2 vane
(A7.5)

The time histories of each variable were then converted into frequency domain information
using a Discrete Fourier Transform, [80], given by:
N 1
X(kf ) = t x ne i2 ( kn)/ N ; k = 0,1,2,..., N 1
n =0
(A7.6)
which gives real and imaginary parts of X:
N 1 N 1
Re[X (kf )] = t x n cos(2 (kn) / N); Im[ X(kf )] = t x n sin(2 (kn) / N)
n =0 n= 0
(A7.7)
The quality of these frequency domain data can be enhanced by standard processing
techniques such as applying overlapped and tapered windows to the data, as recommended
by Tischler, [80].
Each degree of freedom can then be treated separately, and formulation as a linear regression
problem allows estimation of the coefficients. The state-space description is converted to the
frequency domain, i.e.
i x( ) = Ax( ) + Bu( )
(A7.8)
Note that this assumes that any process noise is zero. The unknown coefficients of the A and
B matrices are determined by solutions of the frequency domain equations:
Im[ x ( )] = A (Re[ x ( )]) + B(Re[u( )])

Re[ x ( )] = A(Im[ x ( )]) + B(Im[u( )])
(A7.9)
This solution applies equal weighting to real and imaginary part errors, which is consistent with
the standard weighting for system identification on a Bode plot. The pitching moment equation
for example, is then expressed as the two equations:
Im[q( )] = Mu Re[u( )] + Mw Re[w( )] + Mq Re[q( )]+ M Re[ ( )]

+ M Re[( )] + M s Re[ s ( )]
Re[q( )] = Mu Im[u( )] + Mw Im[w( )] + Mq Im[q( )] + M Im[ ( )]
+ M Im[( )] + M s Im[ s ( )]
(A7.10)

The other degrees of freedom are in a similar form. Figure A7.1 shows the real and imaginary
parts of the pitching moment equation in the above form, and the corresponding fit using the
derivatives identified using the above process. This result is typical of that achieved in general
for both VPM M16 and Montgomerie.
Figure A7.1 Fit Quality of Flight and Identified Model Pitching Moment Equation

Addendum 1 Technical Report
A Survey of Gyroplane Centre-of-mass and Propeller Thrust Line Locations
Dr Stewart Houston
Department of Aerospace Engineering University of Glasgow
August 2010 Addendum 1 Page 1

"A Survey of Gyroplane Centre-of-mass and Propeller Thrust Line Locations"
Report
(issue 2)
September 2005
Dr Stewart Houston
Dept. of Aerospace Engineering

James Watt Building, The University of Glasgow, Glasgow G12 8QQ
Tel. (0141)-330 6865; Fax. (0141)-330 5560
This Proposal is supplied under the conditions that it is not copied, disclosed or used
other than for the purposes for which it is supplied without the express written
consent of The University of Glasgow
Copyright The University of Glasgow 2005

1 Introduction
The amendments and additions to BCAR Section T proposed by Glasgow University

are based on criteria associated with the vertical position of the centre-of-mass.
These criteria were developed from parametric studies using the RASCAL simulation
model, and validated by flight test using the VPM M16 G-BWGI, and Montgomerie-
Parsons G-UNIV. Limited supporting data from hearsay evidence associated with
other types has helped to increase confidence in the proposed changes.
Five Bensen-type gyroplanes have been the subject of the CAA flight trials conducted
to establish compliance demonstration against the dynamic stability requirements of
BCAR Section T. This report documents the results of a weight and balance survey of
these five aircraft. The objective is to provide a database of results that can be used
to correlate flight compliance demonstration with c.g. location, and hence serve to
validate the proposed limits between c.g. and propeller thrust line location.
2 Background
The role of the vertical location of the c.g. in relation to the propeller thrust line is a
significant issue of which gyroplane pilots have always been keenly aware, but only
in relation to the power-pushover phenomenon during flight with high power and at
low rotor loading. However, its role in relation to static and particularly dynamic
longitudinal stability has only recently been explored. Simulation studies highlighted
the sensitivity of dynamic stability, particularly the phugoid-type mode, to vertical c.g.
location in 1994 [1], and in 1997 a recommendation was made that the normal
distance between the propeller thrust line and c.g. should not exceed 2 in, [2]. In
1998 analysis of flight test data obtained using a VPM M16, which had a thrust line
close to the 2 in limit, validated the simulation studies and postulated a simple
mechanism that explained how raising the c.g. location would tend to improve
stability [3].
The CAA recently undertook a series of flight tests using five Bensen-type
gyroplanes, including compliance demonstration against the dynamic stability
requirements of BCAR Section T. No quantitative weight and balance information was
available for these aircraft, although one incorporated modifications claimed to place
the propeller thrust in line with the c.g. This report documents results obtained from
measurements taken using the same method with all five aircraft.
3 Analysis and experimental method
The analysis method chosen to calculate c.g. position is described in Appendix 3 of

the main body of this report. It requires measurements of wheel reactions,
wheelbase and airframe pitch attitude. It is accepted that it is not the only method of
determining c.g. location and it is further accepted that the vertical position of the c.g.
can be calculated with large error bounds. However it is practical and simple, and the
errors are rigorously quantifiable. An alternative method, by suspension from multiple
points, is illustrated in Figure 1 the aircraft is G-BUJK, one of the subjects of this
report, although the test shown in the figure was conducted over 10 years ago. It
requires suspending the aircraft and hence the use of a winch or crane and it is argued
that the photographs show this method is also prone to errors subject to
interpretation and are therefore qualitative only. The preferred method here is widely
accepted, and is perhaps the only means of dealing with large aircraft. The largest
aircraft that this author has seen examined using multi-point suspension is Lynx

ZD559 (around 4 tonnes). In summary, it is argued that the practicalities of the chosen
method outweigh disadvantages of accuracy, which are in any case controllable and
quantifiable. Note that the results are calculated in Earth axes (Figure A3-1, main
report), but can easily be transformed to airframe axes if required.
The experimental method was to roll the aircraft onto the load cells using the ramps
provided. Depending on configuration and type, the aircraft would sit either on its
tailwheel, or nosewheel and the load cell placed accordingly. A digital clinometer was
used to measure the pitch attitude of the aircraft. Wheelbase was obtained by
chalkmarking a rectangular box around each wheel and then measuring distances
between box centroids.
It is important that no lateral force is placed on the load cells as this has been found
to introduce bias errors, hence the importance of rolling the aircraft on and off.
Lowering or lifting the aircraft onto the load cells, or even allowing the pilot to climb
in once placed on the cells, has the potential to splay the main wheels or scrub the
nosewheel laterally and hence introduce sideloads. Multiple tests of the same
configuration were made to ensure repeatability and consistency. The aircraft was
inclined by placing the cells on top of wooden pallets.
4 Results
Five aircraft were tested as detailed in Table 1. Each aircraft was examined in four
configurations for which two sets of measurements were required to compute c.g.
position (i.e. aircraft level, then inclined). This gave a total of 40 cases, most of which
were repeated at least twice to address consistency.
Table 1 Aircraft surveyed
Aircraft Date Location Appendix No.
Bensen B8MR 25 May 2005 Barton A2

G-YJET
Bensen B8MR 5 Jan 2005 North Coates A3
G-BIHX
Montgomerie-Bensen 6 Jan 2005 Shipdham A4
B8MR G-BZJR
Bensen B8MR 7 Jan 2005 Poole A5
G-BOZW
Montgomerie-Bensen 25 May 2005 RAF Benson A6
B8MR G-BUJK
The intention was to survey aircraft with zero fuel, and full fuel, although this was not
possible and only G-BZJR was tested with full fuel. At each fuel state the aircraft
would be assessed with and without pilot. A summary of configurations is given in
Table 2. Note that G-BUJK was tested in the same configurations with a fuel load but
G-YJET was only tested empty.
Centre of gravity positions were calculated for each aircraft, and the results are
summarized in Table 3, as well as new data from the additional measurements taken
using G-YJET and G-BUJK.

Table 2 Aircraft configurations
Wheel
Wheel
base Wheel
Aircraft Pilot Fuel base
(tail) track
(nose)
Bensen B8MR 96.0 kg 0.0 kg (01) 1.192 m 1.214 m 1.627 m

G-YJET
Bensen B8MR 81.5 kg 15.0 kg (21 1) 1.268 m 1.152 m 1.567 m
G-BIHX
Montgomerie-Bensen 74.0 kg 32.0 kg (45 1) 1.300 m n/a 1.605 m
B8MR G-BZJR
Bensen B8MR 92.5 kg 15.0 kg (21 1) 1.140 m 1.143 m 1.574 m
G-BOZW
Montgomerie-Bensen 78.0 kg 14.5 kg (20 1)* 1.245 m 1.190 m 1.620 m
B8MR G-BUJK
84.4 kg (9.5) 15.3 kg 11.3) 1.229 m 1.175 m 1.599 m
Mean (SD)
(0.06) (0.03) (0.03)
* Estimated - no zero fuel configuration tested.

Table 3 Calculated c.g. locations, ( x cg , y cg , z cg ) - m
August 2010
Aircraft Zero fuel, no pilot Zero fuel, with pilot Fuel, no pilot Fuel, with pilot
CAA Paper 2009/02
Bensen B8MR -0.15,-0.03,-0.85 0.15,-0.03,-0.79 n/a n/a

G-YJET
Bensen B8MR -0.14,-0.01,-0.81 0.16,-0.02,-0.70 0.10,-0.02,-0.77 0.17,-0.01,-0.73

G-BIHX
Montgomerie-Bensen n/a 0.22,0.01,-1.02 0.07,0.01,-1.00 0.23,0.02,-0.99

B8MR G-BZJR
Bensen B8MR -0.19,0.01,-0.95 0.18,0.01,-0.84 -0.12,-0.02,-0.93 0.20,-0.01,-0.80

G-BOZW
Montgomerie-Bensen n/a* n/a* -0.11,0.01,-0.914 0.13,0.01,-0.78

B8MR G-BUJK
-0.16,-0.01,-0.87 -0.18,-0.01,-0.84 -0.065,-0.01,-0.90 -0.18,-0.00,-0.83

Mean (SD)
(0.026,0.020,0.072) (0.040,0.021,0.145) (0.090,0.017,0.097) (0.043,0.015,0.114)
* No zero fuel configuration tested.

Addendum 1 Page 6
The normal distance between the propeller thrust line and the c.g. can be calculated
from:
d ( z p zcg ) cos(J T ) ( x p xcg ) sin(J T )
where (xp, yp, zp) is the position of the propeller hub in body axes, (xcg, ycg, zcg) is the
c.g. position, J is the inclination of the propeller thrust line (positive down) and T is
the pitch attitude. The results are shown in Table 4. Positive values indicate c.g. below
the thrust line.
Table 4 Distance of c.g. from propeller thrust line (m)
Zero fuel, Zero fuel,

Aircraft Fuel, no pilot Fuel, with pilot
no pilot with pilot
Bensen B8MR 0.089 0.104 n/a n/a

G-YJET
Bensen B8MR 0.030 0.072 0.081 0.047

G-BIHX
Montgomerie-Bensen n/a -0.100 -0.083 -0.075

B8MR G-BZJR
Bensen B8MR 0.024 0.042 0.052 0.077

G-BOZW
Montgomerie-Bensen n/a* n/a* 0.030 0.130

B8MR G-BUJK
0.048 0.030 0.062 (0.025) 0.045 (0.087)

Mean (SD)
(0.036) (0.090)
Note: 2 inches = 0.0508 m * No zero fuel configuration tested
The two podded machines appear to have the largest thrust line offset, of between
4-6 in above the c.g. The two open-frame aircraft have slightly smaller thrust line
offsets of between 1.7-3 in. Only G-BZJR has a thrust line below the c.g., of between
3-4 in. This result appears consistent with Montgomerie's result, Figure 2, if the black
line drawn on the pod to represent the propeller thrust line is discarded and a more
appropriate line drawn that takes into account the engine tilt (which the line as drawn
arguably does not).
5 Discussion
Four of the five aircraft surveyed displayed c.g. locations below the propeller thrust
line, the average normal distance lying between 0.051 m (without pilot) and 0.078 m
(with pilot). This is in excess of the 0.05 m (2 in) limit proposed for Section T, Ref. 2.
However, one aircraft (G-BZJR) was designed to comply with Section T but does so
only by having its c.g. 3-4 in above the thrust line. A large part of this characteristic is
due not only to having a c.g. located considerably higher on the airframe relative to
the other aircraft, but in also having the engine inclined downwards by 3.8 deg relative
to the airframe, combined with a significantly further forward c.g. location which also
makes a noticeable contribution to placing the thrust line below the c.g.

The two anomalous results discovered in earlier work have now gone as a direct
consequence of the new data indicated earlier. G-BUJK does however remain worthy
of discussion, as it has been the subject of a third weighing exercise to directly
compare the method used here with a double-suspension point method proposed by
the BRA and conducted at RAF Benson on 5 August 2005. On this occasion both
methods gave identical results, with the propeller thrust line being a normal distance
of 0.08 m above the c.g. Note that the result here gives 0.130 m. The differenceTcan
be attributed to the pilot sitting on foam cushion of substantial depth on the later
exercise. Montgomerie's result for this aircraft with pilot does suggest a c.g. that is
below the propeller thrust line, but not by 0.2 m, Figure 1.
Experience has shown that great care is required if accurate results are to be
achieved. Primarily, no sideloads should be placed on the load cells, and consistent
total loads should be obtained with the aircraft level and inclined. Inclinations of 5 deg
seem to produce error bounds of 1-2 in, so consideration should be given to angles
of 10 deg.
6 References
[1] Houston, S., "The Aerodynamics of Gyroplanes CAA Contract No. 7D/S/1125 Phase 2
Interim Report". 5 May 1994.
[2] Houston, S., "The Aerodynamics of Gyroplanes CAA Contract No. 7D/S/1125 Review
of BCAR Section T (issue 2). July 1997.
[3] Houston, S., "Identification of Autogyro Longitudinal Stability and Control
Characteristics". AIAA Journal of Guidance, Control and Dynamics Vol. 21 No. 3, pp.
391-399 (1998)

Figure 1

Figure 2

Addendum 2 Internal Report
Civil Aviation Authority

Montgomerie-Bensen Autogyro Stability Investigation
W.F.R. White
N. Talbot

1 Montgomerie Bensen Autogyro Stability Investigation
1.1 Introduction
Permit to Fly autogyros have the worst safety record of all types of flying machine.
Based on the available evidence and AAIB reports, the record is up to 13 times worse
than that for Microlight fixed wing aircraft (AAIB Bulletin No: 9/2004). Some other
countries in which autogyros are allowed to fly would appear to have a similar
situation.
A number of the accidents have been put down to the flight characteristics of this
type of aircraft, combined with the ability of pilots to fly them satisfactorily which has
lead to the question of whether the high accident rate is due to excessively
demanding flight characteristics or to inadequate pilot training and ability.
In order to attempt to resolve this issue, and in response to AAIB Recommendations
following fatal accidents to RAF 2000 G-CBAG and Ponsford Bensen G-BIGU, the
CAA has carried out a number of evaluations of autogyro handling characteristics.
Accident data showed that after the Air Command Elite 532, the Bensen B8MR series
of autogyros has a significantly worse record than other types, so it was decided to
investigate these first. A number of evaluations were required because the B8MR has
been produced in a number of significantly varying configurations.
The evaluations were carried out by an experienced autogyro pilot and instructor who
was also an experienced professional helicopter pilot, working under contract to the
CAA. He was given basic instruction in flight test techniques. This was considered a
good solution to avoid a heavy handed approach to recreational machines yet provide
appropriate evaluation and help create a good flight evaluation capability within the
autogyro industry for future use within a self regulatory context.
The aircraft evaluated so far, described in the next section, were all provided freely by
the owners. Their generosity, support and interest in the evaluations was invaluable
and is much appreciated by the CAA.
Some additional points from a Permit to Fly issue evaluation that was carried out on
a Speich Air Command autogyro are also included.
1.2 Aircraft Tested
Five aircraft have been tested. The details of each are as follows:
Aircraft Type Bensen B8MR

Aircraft No. G-01-2003
Airframe Hrs. 95:45
Engine Type Rotax 503
Engine No. 4154285
Rotor 22 ft Dragon Wings
Propeller 52 inch, 3 blade
GSC
Built from plans
Location North Cotes
Figure 1 Aircraft Tested - G-BIHX


Airframe Hrs. approx 200
Engine No. 3722444
Rotor 22 ft RotorHawk
(original type)
Propeller 52 inch, 3 blade
GSC
Built from plans
Location Newton Peverill
Figure 2 Aircraft Tested - G-BOZW

Airframe Hrs. 164
Engine No. 3798747
Rotor 22 ft RotorHawk
Propeller 60 inch, GSC TEC 3
Built and supplied by Montgomerie
Location Abingdon
Figure 3 Aircraft Tested - G-BUJK

Airframe Hrs. 180 approx
Engine No.
Propeller 60 inch, Warp Drive
Location Barton
Figure 4 Aircraft Tested - G-YJET

Airframe Hrs.
Engine No.
Propeller 62 inch, Ivoprop
Location Shipdham
Figure 5 Aircraft Tested - G-BZJR

Each aircraft was in generally good condition and had been inspected by PFA
immediately prior to the trial. G-YJET had not been flown for some time prior to the
CAA visit and had a number of unserviceabilities, principally a failure of the prerotater
engage cable and a cracked bracket for the radiator coolant header tank.
The final aircraft G-BZJR has been modified by Montgomerie Gyroplanes to raise the
vertical c.g. position and lower the engine thrustline, but is otherwise virtually
identical to G-BUJK.
1.3 Test Programme
Flight tests were conducted from the home airfield of each aircraft. The evaluation
pilot was Mr B de Saar, and each exercise was managed/supervised by N Talbot and
WFR White.
Table 1 Flight details
Aircraft Date Airfield Total flight time

G-BIHX 14 June 2004 North Coates 1:10
G-BOZW 15 July 2004 Newton Peverill 1:30
G-BUJK 11 Aug 2004 Abingdon 2:20
G-YJET 18 Oct 2004 Barton 1:55
G-BZJR 10 Nov 2004 Shipdham 1:55
Weather conditions were acceptable for each exercise, with winds of 5-10kt.
The following tests were completed:
- Cockpit Assessment
- Ground Handling
- Take Off and Landing
- Airspeed Position Errors
- Longitudinal Static Stability
- Longitudinal Dynamic Stability
- Lateral and Directional Stability
- Power/attitude coupling.
1.4 Initial Findings
a) Cockpit Assessment
The cockpits of this class of aircraft fall into two main categories, being open or
faired.
i) Open Frame Cockpits
The open frame cockpits are indeed completely open with no structure
surrounding the pilot's seat (see photographs in previous section). Instruments
were mounted in a binnacle in front of the pilot, generally low down and the
predominant effect was of sitting in open space. There was clearly no
protection from the elements, but more importantly, little or no structure to
provide pitch reference.
It was also noticeable that the rotor tip path plane was very high in the normal
flight condition, being 'flapped back' to substantially above the horizon, further
removing cues to attitude that would, for example, be present in a helicopter.

Little effective instrumentation was generally provided. Most consisted of

airspeed, altitude, a whiskey compass, with basic powerplant instruments, e.g.
engine RPM and coolant temperature. Readability of airspeed and altimeters
was generally poor with, sometimes, small scales covering large ranges. The
accuracy of engine RPM gauges was questionable. There was typically no
vertical speed indicator (VSI), or attitude indicator. Nr indications, where
provided, were typically digital readouts of questionable accuracy especially at
lower rpm and in general little use was made or importance placed on rotor
speed.
The combined effect of the lack of pitch reference from structure, no attitude
indicator, no VSI with, sometimes, very crude altimeter indications resulted in
great difficulty in defining precise flight conditions for test purposes and
undoubtedly contributed to the difficulty in maintaining stabilised flight. These
factors may be influential in explaining the initial opinions on pitch dynamic
stability between open and faired aircraft discussed later.
ii) Faired Cockpits
The 'faired' cockpits typically had a 'nose cone' shaped fairing with windscreen.
The cockpit structure enclosed the pilot and seat up to just below shoulder
height with the pilot sitting in the normal position (See photographs in previous
section) but with very little space to spare. The sitting position in the faired
cockpit aircraft was much more recumbent than the open framed machines
owing to the rudder pedals being much higher.
The instrumentation standard was similar to that of the open frame machines.
The primary difference from the open frame machines was that the cues to
pitch attitude were somewhat improved by the greater amount of structure
ahead of the pilot, although even this was of limited width.
iii) Cockpit Ergonomics
It was noted that some machines had poorly placed throttles that were difficult
to use. Others had poor mechanical characteristics, i.e. high gearing or a
tendency to create high gearing due to lack of arm support causing the throttle
to be gripped close to it's fulcrum. These generally poor design features
impeded smooth control of engine power in some cases particularly when
combined with the non-linear power delivery characteristics of the Rotax 532
engine.
Furthermore, there was a lack of standardisation in the location of prerotater-
engage and wheel brake controls and in some cases, the wheel brake lever was
particularly poorly placed. Wheel brake, throttle and prerotater must be used
simultaneously during initial rotor engagement and this poor cockpit layout
made rotor engagement difficult.
b) Ground Taxy, Take-Off and Landing
i) Ground Taxy
Ground taxy did not present any particular problems, all the machines having
wide track main wheels and powerful rudder effect from propeller slipstream.
The characteristics of the nose wheel steering systems were variable. Braking
was to the nosewheel only, either by means of a pad applied directly to the tyre,
or by a conventional single leading shoe drum brake, which was somewhat
more effective. Taxying over rough ground caused a significant amount of
pitching due to the short-coupled semi-rigid nature of the undercarriage. The
final aircraft examined, G-BZJR, had main gear suspension, but all operations
were from paved surfaces, so no observation on rough ground handling was
possible.

ii) Take-Off
The three faired aircraft had rotor pre-rotator mechanisms, allowing Nr to be
increased before commencing the take-off run. The pre-rotator mechanisms
were very prone to vibration and resonance during operation, and although this
was only for a short period of time, and on the ground, it was plain from witness
marks that damage was being caused to the main control rods on several of the
aircraft due to contact with the airframe during pre-rotator operation.
Aircraft without a pre-rotator relied on natural wind or forward motion to spin up
the rotor, and unless the natural wind was brisk, long take-off runs could result.
Without the general provision of Nr gauges, there was little definitive feedback
to the pilot that Nr had risen adequately to achieve a take-off. Adequate Nr was
judged by various methods, such as general 'feel', inability to count the passing
blades, general experience, etc. Lack of adequate Nr during take-off results
typically in a roll over accident, and these occur, at least anecdotally, quite
frequently. Reliable useable Nr indications with proper operating handbook
information could be influential in reducing this type of accident.
In other respects, take-offs were straightforward on smooth surfaces, but less
so on rough surfaces. In this case, the short-coupled semi-rigid nature of the
undercarriage resulted in quite severe pitching and associated difficulty in
achieving smooth acceleration and Nr build up. Although manageable by
experienced pilots, rough ground take-offs should be treated with caution and
avoided where possible by less experienced pilots.
iii) Landing
Landings presented no particular problems with low speed, flared touchdowns
possible, particularly with some wind.
c) Airspeed Calibrations
Detailed results are included in the individual flight test reports. There was a
general trend for the airspeed indicator (ASI) to over read significantly at higher
speeds, by up to 25% (see Figure 1 below). This appeared to be the case for both
venturi and pitot type systems. The venturi systems had no obvious static source,
which caused confusion as to how they functioned at all. The pitot systems
typically had a static vent on the back of the ASI case, which clearly exposed the
system to large potential errors depending on the location of the instrument. The
large airspeed errors are perhaps one reason why some machines are felt to be
more benign in their characteristics than others, in that these aircraft are never
flown at higher actual speeds due to the indications.
Although it could be argued that gross over reading protects from exposure to poor
high speed characteristics and is therefore a safety benefit, it would be more
appropriate to have consistently accurate indications and correct maximum speed
limitations. Consideration should be given to including airspeed calibration with
appropriate accuracy requirements in Section T.

75
Zero Error
55 G-BUJK
EAS (mph)
G-YJET
35 G-BZJR
Position Error G-BUJK
15 Position Error G-YJET
Position Error G-BZJR
-5
-25
20 40 60 80
IAS (mph)
Figure 6 Bensen B8MR Airspeed PE Calibration
d) Forward Flight Longitudinal Characteristics

Forward flight longitudinal characteristics were examined formally at two speeds,
typically slow and fast cruise conditions in level flight. Details are given in the
relevant flight test reports. A qualitative evaluation was carried out at Vne.
i) Controllability
Stick margins were adequate in all cases with relatively small amounts of the
available range being required to maintain control. Power on yaw controllability
was good at all speeds with propeller slipstream working on the large rudders,
although large rudder offsets were required for many flight conditions to
counter propeller slipstream asymmetric effects such as torque and thrust
variation across the propellor disk due to varying blade angle of attack in very
steep descents. A similar characteristic is apparent in high power steep ascents
and, in particular, in nose-high throttle slams. Power off yaw controllability
steadily degraded with reducing airspeed to a point at which it was no longer
possible to maintain yaw control, typically 25mph (indicated). It follows that
engine failures at or below this speed could be problematic and this is discussed
later.
ii) Longitudinal Static Stability
Longitudinal static stability was typically positive, with good stick gradients as
shown below in Figure 2 below. The results for G-BIHX are suspect owing to
airflow effects on the spring tape used to measure stick position. Aircraft
configuration, open frame or faired, with or without horizontal stabiliser did not
make a significant difference.

70
65 G-BOZW
G-BOZW
60
G-BUJK
55 G-BUJK
% fwd stick
G-YJET
50
G-YJET
45 G-BIHX
G-BIHX
40
G-BZJR
35 G-BZJR
30
20 30 40 50 60 70 80 90
CAS (mph)
Figure 7 Bensen B8MR Longitudinal Static Stability

iii) Longitudinal Dynamic Stability
Attempts were made to examine both stick fixed and stick free dynamic
characteristics in the longitudinal axis. It was difficult to carry out stick free tests
as most of the aircraft could not be trimmed due to a lack of an in-flight
adjustable trim system, other than longitudinally at one speed and usually not
laterally, i.e. it was not possible for the pilot to let go of the stick. It was equally
difficult to carry out stick fixed tests due to the difficulty in returning the cyclic
precisely to the original position once the disturbance had been made. The tests
completed were therefore predominantly stick fixed but with a degree of
variability. The dynamic stability tests were very dependant on pilot technique,
and this developed and improved as the evaluations continued. Nevertheless, a
good general impression of the longitudinal dynamic stability characteristics
was obtained.
To summarise the characteristics, all the aircraft demonstrated dynamic
instability at higher speeds, and at VNE were reported to be noticeably 'twitchy,'
based on general workload assessment. The open framed aircraft were less
unstable than the faired cockpit aircraft, which could generate high nose up or
nose down pitch rates on the first or second overshoot following an initial nose
up disturbance, to the extent that the pilot was reluctant to make an initial nose
down disturbance. Some aircraft had small, short coupled horizontal stabilisers,
but as might be expected, these seemed to have no effect.
A potential anomaly was that initial impressions of stability during general flight
were more favourable with the faired cockpit aircraft, whilst the formal
examination of stability showed them to be worse. This is possibly explained by
the additional references available from the cockpit structure allowing the pilot
to detect and deal with the attitude changes caused by instability more quickly
than with the open framed machines. (See also paragraph 4.a.(i) of this
Addendum)
On more than one occasion, faired cockpit machines were seen to rotate rapidly
nose down, giving the appearance of rotating about the centre of gravity. Such
a manoeuvre can result in a considerable reduction in normal G and with less
experienced pilots might induce large aft cyclic inputs with an increased
probability of high flapping angles and rotor/airframe contact. Of equal concern

were the situations when the dynamic instability resulted in a rapid nose up
departure with the potential for less experienced gyroplane pilots to respond
with a sharp nose down cyclic input leading to G reduction, mast bumping and/
or rotor/airframe contact.
iv) Ability to Trim
As mentioned above, the ability to trim was generally poor, with no in-flight
capability to reduce stick forces to zero. This resulted in permanent lateral
forces and only one speed at which the aircraft would be trimmed longitudinally.
No rudder trim was provided for and control force was invariably needed to
maintain balanced flight. Such characteristics would not conventionally be
acceptable and some form of in-flight trimming would reduce tendencies to
flight path deviations and unstabilised flight. These characteristics would be
positively dangerous in conditions of poor external visual cues, so it is fortunate
that this class of machine is rarely flown in such conditions.
e) Forward Flight Lateral and Directional Characteristics
i) Lateral Static Stability
Lateral static stability, as might be expected, was neutral to occasionally
negative with the faired cockpit aircraft exhibiting poorer characteristics in this
respect. The lateral static stability characteristics observed are generally
regarded as acceptable for VMC flight, but not for IMC or other cases of poor
external visual references.
ii) Directional Stability
Steady Heading Sideslips and turns on cyclic were carried out power on and off,
at various airspeeds. Directional stability was found to be weak but positive,
becoming neutral at lower power off speeds. Combined with the high power-
on yaw controllability, the weak yaw stability resulted in the ability to generate
large yaw angles easily with a corresponding increase in pilot workload for the
yaw control task. The directional characteristics could usefully be improved.
iii) Spiral Stability
Spiral characteristics could not be examined with any certainty owing to the trim
characteristics, but there was no reported tendency to dutch roll, which
confirms the results obtained from the lateral and directional tests.
f) Inter Axis Couplings
There was a certain amount of pitch to roll coupling, little yaw to roll coupling and
some roll to yaw coupling, which reflect the basic stabilities of the aircraft. The
predominant couplings were from power to yaw and power to pitch, with some
power to roll effects.
i) Power to Yaw
Any changes in power resulted in sideslip due to the slipstream effects on the
fin/rudder being in very close proximity to the propeller, with large sudden
power changes having a correspondingly greater effect. Although there was
adequate yaw control power to deal with this, the coupling effect and the poor
directional stability resulted in significant attitude changes and high pilot
workload during the attempts to maintain stabilised flight during power
changes, e.g. manoeuvring flight or a go-around. The characteristics of the
aircraft could be much improved by reducing the power to yaw coupling, e.g. by
lengthening the fin/rudder moment arm and thereby increasing the separation
of the fin/rudder from the propeller.

ii) Power to Pitch

The power to pitch coupling was found to be dependent on the relationship
between propeller thrust line and vertical centre of gravity. For example, those
aircraft with a high thrust line suffered significant nose down pitching with rapid
power increase. This type of characteristic, when combined with longitudinal
dynamic instability and poor pitch attitude references could be the precursor to
an inability to maintain stabilised flight in the pitch axis.
Even without the pilot making power changes, some variation in thrust will
occur in forward flight in turbulent conditions or whilst manoeuvring and such
variations coupling into the pitch axis will tend to destabilise the aircraft.
It could be that aircraft with otherwise good pitch stability characteristics, could
accommodate such power/pitch coupling, but for most of the aircraft evaluated,
significant improvements could be made by either reducing the inherent
disturbance and/or by improving the pitch stability characteristics.
The final aircraft evaluated in the series had been modified to bring the propeller
thrust line/vertical c.g. offset within the 2" recommended by the Glasgow
University Autogyro Stability Research programme and this resulted in a
significant reduction in power/pitch coupling and gave more docile handling
characteristics generally. A similar result was also noted during the previous
Permit to Fly Section T evaluation of the Speich Air Command Autogyro that had
been similarly modified.
iii) Power to Roll
The torque effects from power variations caused roll disturbances that reflected
the amplitude and rate of power change. Roll control power was adequate to
deal with the attitude changes, but workload was increased.
g) Engine Handling, and Failures
i) Engine Handling
Engine handling was generally acceptable but with some tendency for power
settings to vary for a fixed throttle position, particularly the Rotax 532.
Furthermore, the Rotax 532 engine had rather non-linear power delivery
characteristics which when combined with poor throttle characteristics could
cause inexperienced pilots difficulty in setting and maintaining accurate and
stabilised power. As mentioned previously, the physical and mechanical
characteristics of some throttles' ergonomics did not enhance the ease of use,
and improved design would be beneficial.
ii) Engine Failures
Formal evaluation of limiting Height/Velocity engine failure cases was not
carried out, although the concept was established during the permit to fly
testing of the Spieich Air Command. Introduction of appropriate methodologies
for this into Section T would be beneficial to safety.
Engine failures were carried out at the minimum power on speed established
during controllability testing. These resulted in significant height loss (circa
300ft) to re-establish forward flight, the height loss being very dependant on
aircraft behaviour at the point of failure, and subsequent piloting techniques. For
the high thrust line aircraft, there were significant pitch, roll and yaw attitude
disturbances caused by the couplings mentioned above. In particular, a pitch up
tendency caused speed to reduce which led to yaw control problems and
increased height loss during recovery. The low thrust line aircraft tested last, did
not suffer from the couplings to the same extent or the speed loss and the
recovery was much more straightforward.

1.5 Summary and Significant Findings

Successful evaluations were carried out of a range of B8MR aircraft configurations,
including faired/open cockpit, high and low thrust line machines.
The aircraft became progressively more difficult to fly as speed increases.
The pitch dynamic characteristics are poor at higher speeds, leading to the
possibility of unstabilised flight, large flapping angles and rotor/ airframe contact.
Significant inter-axis coupling with the high thrust line aircraft exacerbated the poor
pitch dynamic characteristics, significantly increasing workload and the ability to
deal with adverse conditions.
Tests on a low thrust line aircraft demonstrated much improved characteristics.
Airspeed indicating systems were grossly inaccurate, and more precise definition
of flight envelopes based on accurate airspeed could be beneficial in protecting
against poor high-speed characteristics.
Engine failure characteristics were significantly improved with a low thrust line.
It was not generally possible to trim in flight.
The lack of physical cockpit structure with the open framed aircraft degraded the
ability of the pilot to control the aircraft precisely.
1.6 Recommendations
a) Section T should be modified taking into account the contents of this summary
report and the detailed flight test results, to protect the situation for future
autogyro designs.
b) Careful consideration needs to be made as to whether modifications should be
required to the B8MR series of aircraft, considering that some owners fly
successfully with no incentive to make changes. Possible options are:
Require modification to achieve a low thrust line;
Require open frame aircraft to have some attitude reference structure, e.g.
T bar, rod antenna;
Improve airspeed indications and re-define VNEs;
Apply minimum experience and greater training requirements for pilots
wishing to fly unmodified aircraft.
1.7 Future Work
a) The relationship between thrust line and vertical c.g. position should be
established for each of the aircraft tested during this investigation.
b) A number of Montgomerie Merlin aircraft, based on the B8MR type have been
modified to use a 4 stroke Rotax 912 engine of increased power output.
Consideration should be given to extending the test programme to investigate the
stability characteristics of this modified variant in the light of the findings of this
report.
N Talbot WFR White

Deputy Chief Test Pilot Flight Test Engineer

Addendum 3 ARB Scholarship Final Report
Aeroelastic Modelling of Gyroplane Rotors
Josef Trchalik
Dr Eric Gillies
Dr Douglas Thomson
August 2008

Air Registration Board Fellowship and Research Grant Scheme

CAA Contract No.905
Aeroelastic Modelling of Gyroplane Rotors
FINAL REPORT
Josef Trchalik
Dr Eric Gillies
Dr Douglas Thomson
Department of Aerospace Engineering

University of Glasgow
Glasgow
G12 8QQ
August 2008

1 Abstract
A mathematical Aeroelastic Model of a Rotor in Autorotation (AMRA) was created to

simulate the aeroelastic behaviour of a rotor during autorotation. The model captures
transverse bending and teeter, torsional twist and lag-wise motion of the rotor blade
and hence it can be used to investigate couplings between blade flapping, torsion and
rotor speed. Lagrange's method was used for the modelling of blade flapping and
chord-wise bending. Torsional twist of the rotor blade was modelled with the aid of
finite element method (FEM), and blade transverse bending could also be modelled
in FEM. The model can switch between using a full FEM model for bending and
torsion, or an FEM model for torsion and simple blade teeter, depending on the
complexity that the user requires.
The AMRA model was verified against experimental data obtained during a CAA
sponsored flight test programme and published results of modal analysis of helicopter
rotor blades and other data published in open literature were used to validate the FEM
model of the rotor blade. As a further verification of the modelling method,
Arospatiale Puma helicopter rotor blade data were compared on a Southwell plot
showing comparison between experimental results and AMRA estimation.
The aeromechanical behaviour of the rotor during both axial flight and forward flight
in autorotation was investigated. A significant part of the research was focused on
investigation of the effect of different values of torsional and flexural stiffness, and
the relative positions of blade shear centre/elastic axis and centre of mass of the
blade on stability during the autorotation. The results obtained with the aid of the
model demonstrate the interesting, and unique, characteristics of the autorotative
regime with instabilities possible in bending and torsion, but also in rotorspeed.
To conclude, the model is used to demonstrate that the requirements currently
embedded in BCAR Section T are adequate and should provide a rotor free of the
types of instabilities identified elsewhere in the report.

2 Nomenclature
a Lift-curve slope [1/rad]

aL Local lift-curve slope [1/rad]
A Rotor disc area [m2]

b Blade length [m]
c Blade chord [m]
cD Drag coefficient [1]
cL Lift coefficient [1]
cM Pitching moment coefficient [1]
cR Resultant force coefficient [1]
D Drag force [N]
1/f Thrust coefficient based on descending velocity [1]
1/F Thrust coefficient based on resultant air velocity [1]
G Gravitational force [N]
H In-plane force [N]
k Torsional Stiffness [Nm2/rad)
L Lift force [N]
m Weight of the blade [kg]
M Weight of the vehicle [kg]
Mc Blade tip mass [kg]
M Blade forcing moment in flap [N.m]
M Blade forcing moment in torsion [N.m]
M Blade forcing moment in flap [N.m]
N Number of blades [1]
Q Blade torque [N.m]
R Rotor radius [m]
S Rotor blade area [m2]
t time [s]
T Blade thrust [N]
T Kinetic energy [J]
U Inflow velocity [m/s]
Up Vertical component of inflow velocity [m/s]
Ut Horizontal component of inflow velocity [m/s]
V Free-stream velocity [m/s]
V Potential energy [J]

Vd Speed of descent [m/s]

vi Induced velocity [m/s]
yc Chord-wise offset of aerodynamic centre from the elastic axis [m]
yg Chord-wise offset of centre of gravity from the elastic axis [m]
D Blade angle of attack [rad]

DD Rotor disc angle of attack [rad]
Blade flapping angle [rad]
Inflow angle [rad]
J Angle of descent/climb of the vehicle [rad]
Blade fixed angle of incidence [rad]
, p Vertical inflow ratio [1]
Blade angle of induced twist [rad]
Air density [kg.m-3]
Blade geometric twist [rad]
Time constant matrix

Dynamic inflow static gain matrix
Rotor speed [rad/s]
Blade azimuth [rad]

3 Introduction
The gyroplane represents the first successful rotorcraft design and it paved the way
for the development of the helicopter during the 1940s. Further development of the
gyroplane was ceased during the following decades as the helicopter became more
successful. Interest in this type of aircraft as a recreational vehicle was resurrected in
recent years thanks to the simplicity of its design and low operational costs.
Most gyroplanes use two-bladed teetering rotors for generation of lift. Unlike in
helicopters, the rotor is not powered by an engine but rotor torque is generated by
the aerodynamic forces and the rotor has to be pre-rotated before take-off.
Gyroplanes do not need a tail rotor as there is no torque acting on their fuselage.
Longitudinal and lateral tilt of the rotor disk is used for longitudinal and lateral control
of the vehicle. Most gyroplanes use the combination of two-stroke or four-stroke
engine and a propeller in pusher configuration, for thrust.
Unfortunately, autogyros, or gyroplanes, have been involved in a number of fatal
accidents during the last two decades. Sudden loss of rotor speed or mechanical
failures of the rotor blades as delamination were involved in many of the accidents.
Very little data on gyroplane flight mechanics and handling qualities are available in the
literature. Accordingly the UK Civil Aviation Authority (CAA) began investigation of
these problems by contracting the Department of Aerospace Engineering, University
of Glasgow to research the aerodynamics and flight mechanics of a gyroplane [1 - 4].
Much more is now understood on the flight dynamics and flying qualities of
gyroplanes, however, rotor aeroelastic instability has not yet been investigated as a
possible issue with regard to flight safety. The aim of the present work is to
investigate this possibility.
The aim of the investigation is to identify flight conditions or configurations of the
rotor that might have catastrophic consequences, and work out basic design criteria
for gyroplane blades. The resulting aeromechanical model of gyroplane rotor blades
can also be used for prediction of stability of new or modified gyroplane rotor
configurations. The first stage of the investigation is to develop a suitable
mathematical model of the rotor system.

4 Development of an Aeroelastic Model of a Gyroplane Rotor
4.1 Background Autorotative Flight

There are substantial differences between dynamics of a helicopter rotor and
dynamics of a rotor in autorotation. During autorotation, both torque and thrust are
generated exclusively by flow through the rotor disc. Thus, in comparison with
dynamics of a helicopter rotor, the system has one extra degree of freedom (i.e. rotor
speed). Thrust and torque are functions of rotor speed and distribution of local angles
of attack along the blade span. Further, angles of attack are dependent upon blade
twist, rotor speed, speed of descent and induced velocity. It can be easily shown that
both speed of descent and rotor angular velocity are strongly dependent upon rotor
torque and rotor thrust. Therefore, simulation of the aeroelastic characteristics of a
rotor in autorotation is iterative process that involves large number of algebraic loops.
This makes modelling of autorotation significantly more challenging than powered
flight.
During steady autorotation, overall torque generated by flow through the rotor disc is
zero and rotor thrust is equal to the weight of the vehicle [5, 6]. There are several
design parameters of the rotor that determine whether steady autorotation is
possible. Perhaps the most important are blade incidence angle (i.e. angle of attack
of the blade relative to the rotor disc plane) and blade torsional stiffness. Torque
equilibrium can not be achieved for high incidence angles due to high value of blade
drag. If torsional rigidity is too low, extensive blade twist has the same effect. The
extra degree of freedom in rotor speed has significant implication for gyroplane rotor
stability. Unlike helicopter rotors, gyroplane rotors can experience significant
variations in rotor speed during changing torque generated at the rotor. Decrement of
the rotor speed decreases centrifugal stiffness of the rotor and the resulting higher
deflections in flap and twist generate more drag and cause further drop in rotor speed.
Conditions in which the rotor enters autorotative regime are also of great importance.
If rotor speed during vertical descent is too low or even zero, the rotor does not
autorotate, requiring pre-rotation to be included in the simulation. A constant value of
torque is applied to the rotor in order to reach conditions that make autorotation
possible.
4.2 An Overview of the Model
The most flexible and practical approach to developing the aeromechanical model of
gyroplane rotor was to programme with the aid of the MATLAB - SIMULINK
computer package which offers powerful tools for modelling of complex mechanical
systems. The model was named AMRA, which stands for 'Aeroelastic Model of a
Rotor in Autorotation'.
A blade element method combined with quasi-steady aerodynamics (Theodorsen's
Theory) is used for calculation of aerodynamic forces and moments generated by the
rotor blade. The general arrangement of the blade model is shown in Figure 1. Two
rotor blades and arbitrary number of blade span-wise elements can be used. The
aerodynamic characteristics of the aerofoil for the full range of angles of attack are
approximated with the aid of wind tunnel data [7]. A NACA 0012 aerofoil was chosen
for the first version of the AMRA model since aerodynamic characteristics for the full
range of angles of attack of the aerofoil are available [8]. A semi-empirical method of
induced velocity calculation was used in the first versions of the AMRA model. The
original calculation [9] was improved in order to capture blade stall and compressibility
of the airflow. A simplified version of Peters HaQuang inflow model replaced semi-
empirical approach in the later versions of the AMRA model in order to improve
fidelity of forward flight simulations.

Lagrangian equations of motion were used to describe dynamics of the rotor blade.
Chord-wise locations of elastic axis (EA), centre of gravity (c.g.) and aerodynamic
centre (AC) can be set in each span-wise station. Values of flexural and torsional
rigidity of the blade can be set to investigate the behaviour of the rotor for different
physical properties of the blades. The AMRA model also allows placement of single
concentrated mass at any span-wise station of the blade.
4.3 Modelling the Aerodynamics of a Blade Element
The blade element method represents a widely used tool for description of the flow
through a rotor disc. The theory has to be modified in order to capture the
aerodynamics of a rotor in autorotation. During autorotation, the flow through the
rotor has the opposite direction to that in the case of powered flight of a helicopter.
Hence the blade aerodynamic angle of attack, D, can be expressed as [9 - 12]
D I T
(1)
where T is the angle of induced twist and I is the inflow angle, Figure 2, given by:
U p
I tan 1
UT
(2)
where UP is the component of flow normal to the blade element and UT is the
tangential component. The method of calculating normal and tangential and radial (UR)
components of velocity at a blade element is developed in Appendix 1. Local values
of vertical and horizontal components of inflow velocity have to be calculated in order
to determine aerodynamic angle of attack of any blade section. Inflow velocity is a
function of angle of attack of the rotor disc, DD, that is given by sum of incidence angle
of the rotor disc L (i.e. angle between rotor disc plane and the horizontal) and pitch
angle of the vehicle here given the symbol J. Note that during axial flight, rotor disc
angle of attack is 90. Using the symbols Vd and Vh to signify the vehicle body velocity
components in the x and z directions, the disc angle of attack is given by:
DD L J
(3)
where
V
J tan 1 d
Vh
(4)
From Appendix 1, the result for axial flight is:
dE dT
UP :R O x T y
d\ d\
(5)

dT
UT :R x T y
d\
(6)
UR :ROE
(7)
where: is the rotational speed of the rotor,
R is the rotor blade radius,
Vd vi
= non-dimensional inflow velocity ratio =
:R
vi = inflow velocity,
= blade flapping angle,
\ = blade azimuth,
r
x = normalised radial position of element =
R
y = non-dimensional chordwise position of aerodynamic centre with
respect to elastic axis = 1 3
c y EA
R4
The equations above allow the angle of attack and velocity at each blade element to
be determined and hence, given appropriate data for the particular blade section,
aerodynamic loads can be found. An assumption of a linear lift curve and parabolic
drag curve is often used in order to simplify the model and speed up computations.
However, this approach does not allow the effects of blade stall, drag divergence and
most importantly air compressibility to be captured. Since the aerodynamic
characteristics of rotor blade elements depend upon local values of angle of attack
and Mach number, it is convenient to express aerodynamic characteristics of the
blade aerofoil as functions of these two variables. This can be achieved by expressing
aerodynamic characteristics of the blade in terms of polynomial functions of angle of
attack and Mach number [5].
It was shown by Prouty [7] that it is possible to obtain full-range angle of attack
aerodynamic data of an aerofoil with the aid of polynomial fit. Prouty uses the
example of the NACA 0012 aerofoil in his book [7]. Prouty's empirical equations were
derived from data published in Carpenter [13]. Since full-range AOA aerodynamic data
for the same aerofoil are available from numerous sources [8, 14], Prouty's approach
was amended and incorporated into the AMRA model. Figures 3 and 4 show the
variation of lift and drag coefficient of a NACA 0012 section obtained using the
polynomial representations suggested by Prouty.
When the values of aerodynamic coefficients at all span-wise stations are obtained,
the forces generated by the blade elements can be calculated:
1 1 1
dL UU 2C L cdx dD UU 2C D cdx dM c / 4 UU 2CM c 2 dx
2 2 2

It can be seen from equations (5) (7) that inflow velocity does not depend upon
azimuth in axial flight. This symmetry makes modelling of axial flight much easier
since model of single blade can be created and resulting aerodynamic forces can be
obtained by integrating the elemental loads across the blade span and multiplying by
the number of blades (NB). In forward flight, inflow angle of the blade is a function of
azimuth. Therefore, assumption of uniform rotor disc loading cannot be made.
4.4 Dynamic Inflow Calculation
Many models of helicopter aerodynamics utilise momentum theory for computation
of induced velocity. However, for small negative values of speed of climb,
momentum theory fails to estimate induced velocity correctly (see Figure 5).
Therefore, classical momentum theory cannot be used for calculation of induced
velocity of autorotating rotor. Instead, a modified version of Peters - HaQuang inflow
model that was introduced by Houston [15] was used. In forward flight induced
velocity is assumed to vary radially and around the blade azimuth and can be
expressed as:
vi
vi0 x vis sin \ vic cos\
(8)
These components of the induced velocity are calculated from the following system
of differential equations [16]:
vi0 vi0 T

>W @vis vis [/] L
vi vi M
c c
(9)
Matrix W is the time constant matrix and is defined as [17]:
4R R tan F 2
0
3SVT C0 12U m
>W @ 0
64 R
0
45SU m 1 cos F
5R tan F 2 64 R cos F
0
8VT 45SU m 1 cos F
(10)
whilst matrix / is defined as the dynamic inflow static gain matrix and is given by:
R 15S tan F 2
0
2VT 64U m
>/@ 0
4
0
U m 1 cos F
15S tan F 2
4 cos F
0
64VT U m 1 cos F
(11)

In the equation (9) T, L and M are rotor thrust, rolling moment and pitching moment.
If Vx, Vy and Vz are component free-stream velocities, the remaining variables form
equation (9) are given as follows:
V x2 + V y2 + (2u mom V z )(u mom V z )

Um =
VT
VT = V x2 + V y2 + (V z u mom )
2
T
u mom =
2A
V 2 +V 2
= tan 1 x y
u mom V z

8
C0 =
3
(12)
From the system of equations (9), only the first equation was used in the simulation
and the remaining two components of induced velocity were assumed to be
negligible. This modification decreases computing time and reduces complexity of
the AMRA model significantly. The equation below shows the solution for the rate of
change of vertical component of induced velocity.
T T
3C0 2SUR 2 vi0 Vx2 V y2 Vz2 2Vz T
USR 2
2 USR 2

vi 0
8 UR 3
(13)
Hence, the value of induced velocity during forward autorotative flight can be
obtained by integration of the above equation according to time.
vi | vi0 v i0 dt
(14)
4.5 Modelling Blade Dynamics
The AMRA model can use Lagrange's equations of motion or the finite element
method (FEM) for modelling of rotor blade structural dynamics in bending, teeter and
torsion. Alternatively, combinations of both approaches can be used. This allows
comparison of differences between results obtained with the aid of models with
different levels of complexity. In Lagrange's method, the blades are assumed to be
perfectly rigid and blade stiffness is modelled with springs located at the root of each
blade. Bending and torsional deflections are assumed to be constant along the blade
span. This requires less computational time but it does not provide a full picture of the
dynamics of the system since it is significantly simplified. Finite element analysis
(FEA) of coupled bending-torsion of the blade on the other hand is highly complex and
requires significantly higher computational time.

4.5.1 Blade Dynamics Using Lagrange's Method

Lagrange's method is an elegant way of obtaining the equations of motion and it is
useful for modelling of complex dynamics of rotorcraft rotor blades. In order to use
span-wise distributions of blade physical properties and also allow coupling of
Lagrange's equations of motion with FEA of blade dynamics, each rotor blade is
discretized into a number of lumped masses. Lagrange's equations of motion are
generated for each lumped mass. Linearized and simplified versions of equations of
motion for coupled torsion-flap of a rotor blade are published in the open literature
[18]. Further details of the development of these equations is given in Appendix 2.
4.5.2 Galerkin's Method Development of Finite Element Approach to Blade
Dynamics
Galerkin's method of solution is used in the AMRA model. The model can perform
dynamic FEA of blade torsion or blade bending for teetering rotors. Alternatively, FEA
of coupled bending-torsion dynamics with inclusion of teeter can be used. A 1D FEA
was used in order to reduce complexity of the model. Each blade element has two
nodes and number of degrees of freedom differs from one (torsion only) to three
(torsion with flap-wise bending). Differential equation of blade torsion can be written
in following form [19]
ww f wT
GJ w f ixT w f AT w f q 0
wr wr
(15)
The term AT represents sum of terms that represent the effect of coupling of blade
torsional dynamics with degrees of freedom in flap and rotation.
Although solution of the differential equation of blade torsion with the aid of FEM
does not require shape functions of higher order and linear shape functions can be
used, cubic shape function was chosen for modelling of blade torsion in AMRA. Cubic
shape function is defined as [19]
H1 = 3(xi+1 -x)2 -2(xi+1 -x)3
H2 = 1 -H1 = 3(x -xi)2 -2(x -xi)3
(16)
Corresponding mass and stiffness matrices and forcing vector are as follows [19]
6 6
5l 5li
[ K i ] GJ i
6 6
5li 5li
(17)

13li 9li

[ M i ] ix 35 70
9l 13li
i
70 35
(18)
li

^ fi ` fi 2
l
i
2
(19)
Alternatively, diagonal (or diagonally lumped) mass matrix can be used. It speeds up
the computations as inversion of this mass matrix is much easier than in case of
consistent mass matrix [19]
li
0
[ M i ] ix 2
li
0
2
(20)
The finite element model of blade bending has two nodes per element but in contrary
to the FEM model of blade torsion it requires two degrees of freedom per node -
vertical displacement and flap-wise rotation. Differential equation of blade bending is
shown below [19].
w2wf w2w
EI 2 w f P b w
w f AB w f t 0
wr 2 wr
(21)
Again, the term AB in the above equation represents a sum of all coupling terms.
Unlike FEA of blade torsion, modelling of blade bending with the aid of FEM requires
shape functions of higher order. Hence, two different shape functions have to be
used in order to describe distribution of both vertical displacement of blade nodes and
slope of blade elements over a blade element. These shape functions are called
Hamiltonian shape functions and they are based on the cubic shape function
described in equation (16) [19]. Application of Hamiltonian shape functions results in
the following forms of stiffness matrix, consistent mass matrix and forcing vector
[19].

12 6li 12 6li
6l
EI i 4li2 6li 2li2
[Ki ]
l 3i 12 6li 12 6li

6li 2li2 6li 4li2
(22)
156 22li 54 13li

l
22l 4li2 13li 3li2
[M i ] Pi i i
420 54 13li 156 22li

13li 3li2 22li 4li2
(23)
6li
2
f i li
^ fi `
12 6li
li2
(24)
A diagonal form of mass matrix is more convenient for dynamic analysis. Parameter
has to be a positive number smaller than 50. Kwon et al [19] recommends DM = 1/78,
which is used in AMRA.
1
2 0 0 0
0 D l2 0 0
[ M i ] P i li M i
0 1
0 0
2
0 0 0 2
D M li

(25)

5 Experimental Measurement of Blade Properties
Since the majority of gyroplane rotor blades are manufactured by small private
companies, it is relatively difficult to get any information of structural properties of
these blades. A pair of blades from the Montgomerie-Parsons gyroplane were
subjected to a series of experiments in order to assess its physical properties and
mass distribution. Data gathered during the experiments were used as input values
of the simulations.
The first blade was cut up into 20 sections and each was measured and weighed so
as to ascertain span-wise mass distribution of the blade. Chord-wise position of
centre of gravity was also estimated for each blade element from the arrangement of
internal structure of each blade section (i.e. position and size of the spar, thickness of
the skin and distribution of potential filling material). Figure 6 shows the internal
structure of the blade at blade root and at the tip. It can be seen that both mass
distribution and chord-wise positions of c.g. are mainly given by span-wise
distribution of the spar. Span-wise distributions of blade mass and c.g. locations that
were obtained from the experiments are depicted in Figure 7 and Figure 8.
Experimental measurements accomplished with use of second Montgomerie-
Parsons gyroplane rotor blade were focused on structural properties of the blades.
Torsional stiffness and chord-wise positions of elastic axis of the blade were
measured at three span-wise stations. Span-wise positions of these the stations were
x = 0.25 (quarter-span), x = 0.5 (half-span) and x = 0.75. The rotor blade was firmly
fixed at the root and clamped into outboard clamp at the appropriate span-wise
station. The outboard clamp was then used for loading of the blade with a torsional
moment. Constant weight was used and loading moment was changed by shifting of
the weight along the clamp arm. Consequent measurements of blade angular
deflections allowed calculation of appropriate stiffness coefficients.
MT
kT
T
(26)
Angular deflections of the blade in pitch were determined with the aid of calibrated
angle measuring instruments that were fixed to the upper surface of the clamp. The
arrangement of the equipment during the experiment is shown in Figure 9.
Measurements were carried out for different values of torque at each span-wise
station to increase higher accuracy of stiffness estimation. Graphical interpretations
of the results are given in Figure 10. Torsional stiffness was determined for each
span-wise station of the blade (see Table 1 below).
Table 1 Locations of elastic axis and torsional stiffness as obtained during the
experiment
Span-wise station [1] 0.25 0.5 0.75
Location of EA [%c] 35.5 25.3 27.24
GJ [N.m2/rad] 1534 1443 1409

Measurement of the first flexural natural frequency of the blade was used to estimate
flexural stiffness of the blade. Determination of blade flexural stiffness that was used
during the experiment is described below.
1
Z 2 R 3 mx 4 dx
1 N cycles
f Z 2Sf EI | 0
T t 4
(27)
Data gathered during the experiment are shown in Table 2 below.
Table 2 Characteristics of oscillations in flap of Montgomerie-Parsons rotor blade
Number of oscillations Time [s] T [s] f (Hz) [rad/s]

60 47.63 0.793833 1.25971 7.914993
60 47.62 0.793667 1.259975 7.916655
60 47.67 0.7945 1.258653 7.908352
The resulting estimated value of flexural stiffness is EI = 1166.2 N.m2.

6 Model Verification and Validation
Although the AMRA model is fully functional standalone model of rotor

aeromechanics in autorotation, its main purpose was to test model of blade dynamics
in autorotation so that it can be incorporated into a rotorcraft flight mechanics model.
Therefore, the main objective of the validation phase of the project was to make sure
that blade dynamics are computed correctly by the AMRA model. FEM model of
blade torsion represents the key block of AMRA and extra care was taken during its
validation. Values of teetering angle predicted by the AMRA model were verified
against G-UNIV gyroplane flight data. Estimations of flight performance during flight
in autorotation were also successfully compared with experimental data published in
open literature [6, 18].
6.1 Verification in Axial Flight
A series of simulations of aeromechanical behaviour of a gyroplane rotor in axial
autorotative flight was performed as a simple method of model verification. The input
parameters of the simulations can be found in Table 3. Simulations of a gyroplane
rotor in steady autorotative descent have revealed that the AMRA model captures all
key features of the system. The rotor speed has to be increased by application of
external torque during pre-rotation. Once the rotor speed reaches sufficient value, the
external torque is removed and the system enters autorotative regime. Both
acceleration of the rotor from lower rotor speed and transition from helicopter regime
(i.e. deceleration from higher rotor speed, see Figure 11) can be demonstrated by the
simulation (see Figure 12). Note that rotor speed always stabilises at the same value
once steady autorotation is established since its configuration did not change.
It should be noted that rotor speed in autorotation is much lower than rotor speed of
helicopter rotor during flight. It can be seen from Figure 11 that the rotor speed is
stabilized and the system reaches torque equilibrium within few seconds. At this
point, the total thrust of the rotor is in balance with the weight of the vehicle and the
value of speed of descent is approximately 11.5m/s (see Figure 13). The value of
speed of descent agrees with the results of experimental flight measurements that
were carried out by NACA and several other research bodies [5, 6]. Equation (28)
below shows empirical relationship of disc loading of a gyroplane and speed of
descent that was derived from the experimental results [5, 6].
Table 3 Input parameters of the simulations
PARAMETER VALUE
Blade length (R) 3.63 m
Blade chord (c) 0.2 m
Chord-wise position of EA (yEA) 0.08 m = 40%c
Chord-wise position of c.g. (ycg) 0.066 m =33%c
Chord-wise position of AC (yAC) 0.05 m = 25%c
Offset of c.g. from EA (yg) 0.014 m
Offset of AC from EA (yc) 0.03 m
Blade weight (m) 13 kg
Number of blades (NB) 2

Table 3 Input parameters of the simulations (Continued)
PARAMETER VALUE
Blade fixed incidence angle 0 rad
Span-wise distribution of blade geometric twist = 0 rad

Weight of the gyroplane (M) 400 kg
Blade torsional stiffness coefficient (kT) 1600N.m2/rad
Blade flexural stiffness coefficient (k ) 1200N.m2
Blade section geometry NACA 0012
Torque used for pre-rotation (QPR) 2000N.m
Vd | 1.212 T A
(28)
The weight of the vehicle is M = 400kg and rotor radius is R = 3.63m, hence rotor disc
loading is T/A = 96N.m-2 and equation (28) gives speed of descent Vd = 11.8m/s.
A characteristic span-wise distribution of blade torque for a rotor in the autorotative
regime is observed. The inboard part of the blade generates positive torque and the
outboard part of the blade generates negative torque. In steady autorotation, the total
value of torque generated by the blade is zero. Figure 14 shows a comparison of span-
wise distribution of torque obtained from the simulation and torque distribution as
described in open literature [5, 6]
The so-called coefficient of resultant force is another important characteristic of
autorotative regime [5, 6]. It is defined by
2R
cR =
V 2 A
R = L2 + D 2
V = Vh2 + Vd2
(29)
Previous research involving experimental flight measurements [5, 6] found that CR on
typical rotor of a gyroplane during steady autorotative flight at large rotor disc angles
of attack (DD > 30deg) is about 1.25. It is important to realize that the majority of
gyroplane rotors have very small or zero fixed blade angle of incidence (in effect a
collective pitch setting, in helicopter jargon). Figure 15 shows a comparison of
experimental values of cR [5, 6] and the outcome of the simulation. The AMRA model
predicts value of CR to be 1.19.

6.2 Verification in Forward Flight

Since gyroplanes operate mostly in forward flight regime, modelling of forward
autorotative flight represents the key task in investigation of aeroelastic behaviour of
a gyroplane rotor blade. In comparison to the simulation of axial autorotative flight,
simulation of forward flight in autorotation induces some complication. Both direction
and value of the inflow velocity are functions of azimuth if horizontal speed is not zero
(see equations (A1.2) and (A1.3)). This means that there is no torque equilibrium
during steady forward flight and the value of torque oscillates around the zero value
(see Figure 16). The amount of vibrations induced by the rotor blade during steady
forward flight is therefore significantly higher than in axial descent. In addition, free-
stream velocity at the advancing side of the rotor disc is higher, and thus the values
of the forcing moments are higher too. It can be expected that gyroplane rotor blade
in the forward flight regime is more prone to undergo aeroelastic instability than the
same blade during axial autorotative flight.
6.3 Validation of the AMRA Model
a) Teeter Motion
Data gathered during the CAA sponsored series of flight test of the University of
Glasgow Montgomerie-Parsons gyroplane (G-UNIV) were used for validation of
model of blade teeter that is included in the AMRA model. AMRA uses a NACA
0012 airfoil that has different aerodynamic characteristics in comparison with
NACA 8-H-12 that are used in McCutcheon rotor blades. The NACA 8-H-12 is
normally chosen due to its speed stability properties however in this case a
complete and reliable set of aerodynamic data was required, and hence the 0012
profile was chosen. In order to reach similar flight conditions during simulations
(i.e. rotor speed and speed of descent), rotor speed was set to mean value of rotor
speed measured during the flight tests. Two different regimes of steady level flight
were chosen for the validation. Predictions of the model were found to be in a
good agreement with teeter angles measured during flight trials as it can be seen
from Figures 17 and 18. Table 4 summarizes the results of validation of AMRA
model of blade teeter.
Table 4 Comparison of predictions of rotor blade teeter and G-UNIV

experimental data
CASE VH exp AMRA
[m/s] [rad/s] [rad] [rad]
A 14 38 0.031 0.026
B 27 41 0.058 0.056
b) Validation of FEM Model of Blade Torsion

Deflections of beams of several lengths and with different torsional stiffness that
were loaded statically by a torsional moment at the tip were computed by the
AMRA FEM model of blade torsion. The results were then compared with
analytical estimations of beam tip torsional deflections according to the St. Venant
theory and were found to be in very good agreement. Mean relative deviation was
less than 2%.
The shape of the first torsional mode predicted by the FEM model was also
compared to corresponding torsional mode shapes that were published in open
literature [18]. Predictions of the model are in good agreement with published data.

Comparison of the first torsional mode shape computed by the FEM model of
blade torsion with data published in open literature, Bielawa [18], is depicted in
Figure 19.
Since the span-wise distribution of blade torsion has a strong influence on blade
aerodynamics, it is absolutely crucial that it is modelled correctly and the
aerodynamic forcing of the blade is estimated realistically. Figures 20 and 21 depict
a qualitative comparison of distribution of torque generated by rotor blade over the
rotor disk as predicted by the AMRA model and qualitative sketch of torque
distribution reproduced in open literature [6].
Comparisons of the first natural frequencies in torsion and bending of two different
rotor blades with results of experimental measurements and predictions of other
models of blade dynamics represent another phase of AMRA validation. Data
obtained during experimental measurements of physical properties of the
McCutcheon blade along with comprehensive data on physical properties of
Arospatiale SA330 Puma helicopter rotor blade were used [21]. A Southwell plot
of the McCutcheon rotor blade is shown in Figure 22. Although first natural
frequency in torsion is slightly under-predicted by the AMRA model, general
agreement with experimental data is good. Simple dynamic model of blade
bending using spring stiffness and rigid blades was used during testing of the FEM
model of blade torsion. Predictions of first natural frequency in bending of
McCutcheon rotor blade is consistent with both theory and published shake tests
of similar rotor blades [22, 23].
As can be seen from Figure 23 predictions of first natural frequency in bending of
Puma rotor blade is in reasonable agreement with results of METAR/R85.
CAMRAD and RAE/WHL models predict lower values of the first natural torsional
frequency than both METAR/R85 and AMRA. Bousman et al [21], however,
describes estimations of modal frequencies of METAR/R85, CAMRAD and RAE/
WHL models as consistent.
c) Validation of FEM Model of Blade Bending
A coupled FEM model of blade torsion and bending was validated in similar manner
as the FEM model of blade torsion. Predictions of both static and dynamic
loading of the blade were compared with analytical predictions, experimental
measurements and results of other structural analysis codes.
Predictions of static bending as obtained from the AMRA model are in good
agreement with analytical predictions. Values of blade vertical displacement
obtained with the model are roughly by 5% lower than analytical predictions and
relative deviation of blade gradients is roughly 6% and these values do not change
with loading, blade flexural stiffness or blade length.
Estimations of blade bending behaviour were validated against the same set of
data that was used for validation of the FEM model of blade torsion [21]. The
conclusion can be made that both static and dynamic behaviour of AMRA
structural dynamics block was validated and that the model is likely to give realistic
estimations of both rotor blade torsion and bending. Figures 24 and 25 depict
distribution of torsional deflections and flexural vertical displacements over the
rotor disc during forward flight.

7 Aeroelastic Stability of a Rotor in Autorotation
Unlike helicopter rotors, rotors in autorotation can experience significant variations in

rotor speed during manoeuvres. Decrement of the rotor speed decreases centrifugal
stiffness of the rotor and the resulting higher deflections in flap and twist generate
more drag and may cause further drop in rotor speed. It is clear that thrust and torque
of the rotor are functions of rotor speed and distribution of local angles of attack along
the blade span. Further, angles of attack are dependent upon blade twist, rotor speed,
descent rate and induced velocity. It can be easily shown that both descent rate and
rotor angular velocity are strongly dependent upon rotor torque and rotor thrust. The
AMRA model has shown that the extra degree of freedom in rotor speed has
significant effect on the aeromechanics and aeroelastic stability of an autorotating
rotor.
A series of parametric studies that was carried out with the aid of the model [24]
shows that blade induced twist (torsion), fixed angle of incidence of the blade and
blade geometric twist have by far the strongest influence on the aeromechanical
behaviour of a rotor in autorotation.
AMRA simulations were performed for three different levels of model complexity.
The simplest configuration of the model used perfectly rigid rotor blades in both
bending and torsion and blade flexibility was modelled with the aid of spring stiffness
located at blade root. The second configuration of AMRA employed the FEM model
of blade torsion while only spring stiffness was used for modelling of blade bending.
The most complex variant of the model utilized coupled FEM models of both torsion
and bending. Comparison of results obtained for these three model configurations
allowed assessment of the effect of complexity of the structural dynamics model on
performance and fidelity of AMRA simulations. In order to investigate rotor stability
boundary in torsion, AMRA simulations for various values of torsional stiffness and
chord-wise positions of centre of gravity (c.g.) were carried out.
The results of the simulations have revealed that low torsional stiffness of the blade
leads to an aeroelastic instability (flutter) that manifests as coupled rotor speed / pitch
/ flap oscillations. These oscillations result in catastrophic decrease of rotor speed.
This is a demonstration of strong rotor speed / pitch / flap coupling that exists only
during autorotation.
Decrement of the rotor speed decreases centrifugal stiffness of the rotor and the
resulting higher deflections in flap and twist generate more drag and cause further
drop in rotor speed. This type of flutter is unique for rotors in autorotation since it
differs from both helicopter rotor flutter and flutter of a fixed wing.
7.1 Aeroelastic Stability of a Rotor in Axial Autorotative Flight
Since blade aerodynamic forcing during steady axial autorotative flight is not
dependent upon blade azimuth, it becomes constant after the rotor reaches
equilibrium state. During steady vertical autorotation, overall torque generated by
flow through the rotor disc is zero and rotor thrust is equal to the weight of the
vehicle. The rotorspeed converges towards its steady value during torsional
equilibrium. A characteristic span-wise distribution of blade torque for a rotor in the
autorotative regime is observed. The inboard part of the blade generates positive
torque and the outboard part of the blade generates negative torque [5, 6]. The AMRA
model describes all major features of aerodynamics of a rotor in autorotative axial
descent very well [24].

Results of the AMRA simulations obtained for different levels of complexity of the
model show that torsion is the most important parameter to compute accurately;
torsion therefore requires a FEM. The aeroelastic behaviour of the rotor for two
different levels of model complexity is shown in Figures 26 and 27. In Figure 26, the
simplified model of blade dynamics using equivalent spring stiffness for modelling
both torsion and bending is usued, whilst the full FEM model was used to calculate
the results shown in Figure 27. As it can be seen in the figures, reduction of rotor
speed from a steady value to zero takes only few seconds. Speed of descent
increases to unacceptable value during this time due to dramatic decrease of rotor
thrust. Parametric studies carried out with an earlier generation of the model [24] had
shown that chord-wise position of c.g. seems to have much stronger influence on the
stability of autorotation than chord-wise position of EA. Note that the c.g. location in
Figures 26, 27, 28, and 29 is destabilising, as it is 8% after the elastic axis.
7.2 Aeroelastic Stability of a Rotor in Forward Autorotative Flight
Since gyroplanes operate mostly in the forward flight regime, modelling of forward
autorotative flight represents the key task in investigation of aeroelastic behaviour of
a gyroplane rotor blade. Since both direction and value of the inflow velocity are
functions of azimuth if horizontal speed is not zero there is no torque equilibrium
during steady forward flight and the value of torque oscillates around the zero value.
The amount of vibration induced by the rotor blade during steady forward flight is
therefore significantly higher than in axial descent. In addition, free-stream velocity at
the advancing side of the rotor disc is higher than at the retreating side, and thus the
values of the forcing moments are higher and also asymmetrical.
Since aerodynamic forcing of the blade during steady forward flight has harmonic
character, blade motion in both bending and torsion has harmonic components too.
As in case of axial flight, simplification of the model of blade torsion seems to
significantly degrade predictions of the model. Only small changes in behaviour are
seen when bending is modelled by an equivalent spring stiffness rather than a FEM.
Similarly as in case of axial flight in autorotation, simulations for various torsional
stiffness and chord-wise positions of centre of gravity were performed.
Computations carried out with the aid of the AMRA model have shown that the rotor
suffers of aeroelastic instability if c.g. lies aft EA. Aeroelastic behaviour of the rotor
for all three different levels of model complexity has very similar character to the
aeroelastic instability predicted in autorotative vertical descent and it is shown in
Figures 28 and 29. The resulting aeroelastic stability boundary can be found in Figure
30 which is a plot of the blade torsional stiffness at which instability occurs, kTcrit , for
specific chord-wise location of c.g., ycg. It can be seen from the figure that position of
c.g. aft EA is destabilizing, similar to fixed wing classical flutter and helicopter pitch
flap flutter.

8 Investigation of Aeroelastic Stability Boundaries of Gyroplane Rotors

Impact on BCAR Section T
8.1 Background
The poor safety record of the gyroplane in the UK in the early 1990s led to a review
of the airworthiness requirements of the type, BCAR Section T [25]. From the mid-
1990s the Civil Aviation Authority has funded a series of research programmes in the
University of Glasgow in support of the development of BCAR Section T including
simulation, wind tunnel testing and flight testing. The results of this work have fed
directly into certain key areas of BCAR Section T, notably the requirements for
dynamic stability [2, 3]. The current work reported here has been to study the
aeroelastic properties of gyroplane rotors, and from the previous section it is clear that
instabilities do exist. The aim of the work reported in this section is to further the
aeroelastic stability boundary, carrying out parametric studies to identify if particular
design features might lead to an unstable rotor system. The aim is to verify (or
otherwise) that the requirements for chordal balance currently in BCAR Section T are
appropriate.
8.2 Analysis of BCAR Section T Chordal Balance Requirement
BCAR Section T (light gyroplanes) paragraph T659 states:
Mass balance
a) ......
b) The chordwise balance of the blades must be at, or forward, of the 25% chord. The
chordwise balance of each blade in a pair must be the same, or within a tolerance
to be agreed with the CAA.
This paragraph is similar to the requirement which achieves pitch-flap flutter stability
for helicopter rotors, where it may be shown (see for example Bramwell's Helicopter
Dynamics, by Done and Balmford, 2nd Ed, Chapter 9, section 9.3 [26]) that flutter and
divergence stability for a helicopter blade may be assured, regardless of the control
system stiffness, if the blade chordwise c.g. is ahead of the collinear pitch axis and
quarter chord point. A linear stability analysis shows that the flutter boundary for a
rigid blade, pitching, about the feather hinge at the quarter chord point, and flapping
about a flap hinge, is a hyperbola, as shown in Figure 31. The divergence boundary is
linear, and is usually below the flutter boundary. Note that Figures 30 and 31 are
similar, however the detail is different as Figure 31 shows the boundary for a
helicopter blade pitching around the quarter chord position and is generated by a
linear model, whilst Figure 30 is generated by a nonlinear model of a gyroplane blade
twisting about the elastic axis.
The dashed line in Figure 31 is the maximal aft position of c.g. that assures flutter
stability regardless of control system stiffness, and this always lies at, or slightly aft
of the quarter chord. This is only strictly true for a blade pitch axis coincident with the
aerodynamic centre. The linearized, non-dimensional model used to elucidate this
stability boundary is
1 0 E 1 0 E

I *
x IT* T I x*
IT ZT 1 T
* 2
(30)

for the mass and stiffness matrices, which are added to the following aerodynamic
terms (developed using thin aerofoil, and Theodorsen's theory):
J
8 C (k ) 0 E 0 J C (k ) E

J 1 T 8
T
0 2
0 0
32 AR
(31)
Rotorspeed does not enter into these linearized, non-dimensional equations of
motion.
To elucidate the stability properties of a gyroplane, a similar linearized model is
developed here and is compared to time marched results of a non-linear Finite
Element Model (FEM) of an existing gyroplane rotor. Modifications to the helicopter
model are required to account for the different hub design of a gyroplane to that of a
helicopter, and, in the FEM case, the inclusion of rotorspeed as a degree of freedom.
The results are only pertinent to conventional two bladed teetering light-gyroplane
hubs with no blade feather hinge.
A gyroplane hub has a slightly different construction to a helicopter rotor hub: the gyro
blades are usually mounted to the hub bar at the quarter chord point, but do not have
a separate pitch degree of freedom, and also have a rotorspeed degree of freedom.
The blades pitch about the quarter chord when they are fore and aft (0/180 azimuth),
but pitch about an offset axis, due to the offset hinge on the gyro hub (an introduction
of Igor Bensen in 1948 to provide stick free pitch stability). This offset hinge
arrangement is shown in Figures 32(a) and 32(b).
A rigid blade model (similar to the above linearized equations of motion for a
helicopter blade) of a gyroplane hub therefore has flap and pitch blade degrees of
freedom, with the pitch axis moving cyclically about the quarter chord. Alternatively,
the blade degrees of freedom can be expressed as a lateral disc tilt about the centre
of rotation and a longitudinal disc tilt slightly ahead of the centre of rotation. An
oscillation in flutter would therefore manifest itself as an oscillatory disc tilt.
As a result of this pitch about an axis unequal to the quarter chord, extra terms come
into the aerodynamic matrices for the gyro flutter boundary. The aerodynamic terms
for a single flapping gyro blade become:
J J
C (k ) 0 E 0 C (k ) E
8 8
J 1 1 J 1 1 J 1 1
T T
x C (k ) x 0 x C ( k )
12 AR 2 32 AR 2 2 12 AR 2
(32)
For the teetering gyro rotor as a whole, the mass and stiffness matrices become:
1 0 E 1 0 E
2 * 2
I x IT* T I x*
IT ZT 1 T
* 2
(33)

and the aerodynamic terms for both blades sum together to become:
J
8 C (k ) 0 E 0 J C (k ) E
2
J 1 T 8
T
0 2
0 0
32 AR
(34)
i.e., the extra aerodynamic terms for the offset pitch axis average out to zero (one
blade has a rear pitch axis, and the other a forward pitch axis- the net effect is zero).
Again it should be noted that rotorspeed does not enter in to these linearized
equations of motion. The flutter stability boundary for the gyro blade is therefore
equivalent to the boundary for a helicopter rotor blade and so, for axial flight, and
based on a linear stability analysis, BCAR Section T should produce a rotor free from
'pitch-flap' flutter and divergence.
The stability boundary for the gyro rotor is shown in Figure 33. For forward flight,
however, the offset pitch axis terms in the aerodynamic matrices do not cancel out
because of the added asymmetry in local velocity on the rotor disc at a particular
advance ratio. The effect of non-zero advance ratio, for a typical hub pitch offset is
shown in Figure 34 to be stabilizing when the blades are not at the 0/180 azimuth.
This means a c.g. further aft is permissible when the blades are not at the 0/180
position. The critical blade mass balance condition is the same as for axial flight,
however, which we have shown to be equivalent to that of a helicopter blade (despite
the different hub design). Linear theory therefore suggests that compliance with
BCAR T 659 will assure flutter stability in forward or axial flight.
The critical case is in axial flight, which is never encountered in operation. In forward
flight, the hub offset is slightly stabilizing. The flutter, or divergence, boundary for the
rotor is always furthest left (at its lowest xcg/c) at low (but usually non-zero) values of
control system stiffness. However, it should be remembered that the results of this
stability analysis assume a control system with zero friction. A small amount of
friction is undoubtedly present in actual gyroplane control linkages. When friction is
added to the blade pitch terms in the system of equations it is stabilizing, meaning
that a farther aft c.g. is tolerable. It is, however, difficult to quantify the amount of
friction present in an actual gyroplane- to give a conservative flutter boundary, it is
usual in aeroelastic analyses to assume zero friction, although, in the 1940s,
Theodorsen found that small amounts of friction added to flutter equations often
produced more realistic results than for systems without friction [27].
The rigid blade equations, used for this linear stability analysis were time integrated,
to confirm that a small amount of friction in the control system was stabilizing. The
effect of changing CM was also investigated using this model. For gyroplanes,
0
positive CM0 is often added to the aerofoil section in order to prevent rotor overspeed.
For a single blade, positive CM0 is slightly destabilizing in the model and negative CM0
slightly stabilizing. For a rotor as a whole, however, the pitching moment arising from
CM0 on one blade is counteracted by that of the other blade and so a change to CM0
has no effect on the linear stability boundary for the gyroplane rotor.
The predicted flutter frequency for low stiffness control systems is of the order of 1/
rev.

8.3 Parametric Analysis

A finite element analysis was undertaken for a gyroplane rotor model to confirm the
conclusions of the linear stability analysis. 10-20 finite elements were used to model
blade torsion and bending (on top of teeter), rotorspeed was included in the model as
a degree of freedom, and blades could either pitch about the control system, or about
their own elastic axes. A thin aerofoil, Theodorsen, strip theory aerodynamic model
was used in the analysis. Moreover, non-zero CM0 was added to the model. Various
blade c.g. positions were input to the model, together with varying torsional stiffness,
and the solution time-marched in order to see if the rotor model was stable. The
results are summarized below. As time marching is computationally intensive, points
on the flutter boundary were calculated for c.g.s at 0.28, 0.36, 0.40 and 0.46 chord
and polynomials fit, in a least squares sense, to the data. Linear polynomials provided
the best fit. The flutter boundaries are similar in character to the linear stability
analyses.
The results of varying CM0 are shown in Figure 35. These results are for blades
twisting about their elastic axes (i.e. each blade has its own twist degree of freedom).
Positive CM0 is seen to be slightly de-stabilizing (it shifts the boundary to the left) and
negative CM0 is predicted to be slightly stabilizing. Unlike the case where blades are
pitching about the control system axis, the blade elastic torsion degree of freedom is
associated with significant stiffness (in fact, gyro blades are often much stiffer than
helicopter blades). Again, BCAR T 659 should provide a rotor safe from flutter, given
that actual measured blade stiffness' are high.
Bending stiffness did not affect the stability of the rotor in any of our simulations, but
only small deflections were considered in the model, and no off-axis bending was
modelled. It is noted, however, that off-axis bending is a mechanism by which an
apparently low torsional stiffness can be achieved.
Figure 36 shows the effect of modelling the blade pitch to be about the control
system (the lowest stiffness DOF) or about the blade elastic axes. This figure shows
two boundaries, one for a blade pitching about the quarter chord (i.e. the pitch control
axis) and one for a blade twisting at a nominal value of the elastic axis (at 32% chord
in the figure). At zero stiffness, the destabilizing effect of pitching about the elastic
axis is more pronounced due to the dominant effect of the extra aerodynamic
coupling introduced by the elastic axis/quarter chord offset. This effect is only of
academic interest, however, as the blade elastic axis is always associated with a high
stiffness.
In conclusion, BCAR T 659 provides a safe rotor, in terms of pitch flap flutter and
divergence, for conventionally designed light gyroplanes with teetering rotors. In
addition, the offset gyroplane hub is seen to be slightly stabilizing in forward flight.
The predicted flutter frequency for low stiffness control systems is around 1/rev. This
low frequency, and the fact that the dominant (low stiffness) degrees of freedom are
longitudinal and lateral disc tilt, means that any flutter phenomenon arising from an
aft c.g. will manifest itself as a 1/rev varying disc tilt.

9 Conclusions
An aeromechanical model of a gyroplane rotor AMRA was developed in the MATLAB

programming language and used in predicting the aeroelastic behaviour of a rotor.
Two regimes were investigated autorotative axial flight (vertical descent) and
forward flight in autorotation. Simulations have shown that autorotation is a complex
aeromechanical process with auto-stabilizing characteristics. In order to obtain input
parameters for the structural model of the blade, a series of experimental
measurements were carried out on a typical gyroplane blade. Blade mass distribution,
position of elastic axis, span-wise distribution of c.g. and torsional and flexural
stiffness was determined during the experiments.
Results from the AMRA model were verified and found to be in reasonable
agreement with experimental measurements. Several parametric studies were
performed so as to gain more knowledge on the effect of blade geometry and
structural properties on performance of the rotor during autorotation.
It was found that blade twist / bending / rotor speed coupling has major effect on the
stability of autorotation when the rotor is in a stable configuration. Computations were
performed for three different levels of complexity of the model of blade structural
dynamics. Results of the AMRA simulations have shown that detailed modelling of
rotor blade torsion (i.e. blade induced twist) has major influence on aeromechanics of
a rotor in autorotation. It has been demonstrated that it is sufficient to model only first
bending mode, i.e. to assume constant span-wise distribution of flap angle. The
conclusion can be made that any aeroelastic simulation of a rotor in autorotation
should contain model that gives realistic predictions of blade dynamics in torsion.
Occurrence of a type of flutter that is unique for autorotating rotors was predicted by
the model both during axial descent in autorotation and autorotative forward flight.
This aeroelastic instability is driven by blade pitch / bending / rotor speed coupling and
differs from both flutter of a helicopter rotor and flutter of a fixed wing. The instability
results in catastrophic decrease of the rotor speed and significant increase of speed
of descent.
Finally, the research undertaken in this project was applied to a review of elements
of BCAR Section T, and it was found that BCAR T 659 provides a safe rotor, in terms
of pitch flap flutter and divergence, for conventionally designed light gyroplanes with
teetering rotors.
CAA Comment
CAA accepts this finding yet notes that the stabilising affect of a number of other
parameters was not exhaustively investigated during this limited research. As
such the CAA would consider accepting blades that do not meet this single
parametric requirement, as long as blade stability could be demonstrated by other
means. This would allow for CAA acceptance of alternate means of compliance
that demonstrate an equivalent level of safety to this requirement for blade
stability.

10 References
1 Thomson, D.G., Houston, S.S., Spathopoulos, V. M., "Experiments in Autogiro

Airworthiness for Improved Handling Qualities", Journal of American Helicopter
Society, Pg. 295, No. 4, Vol. 50, October 2005.
2 Houston, S. S., "Longitudinal Stability of Gyroplanes", The Aeronautical Journal,
Vol. 100 (991), 1996, pp. 1-6.
3 Houston, S. S.: "Identification of Autogiro Longitudinal Stability and Control
Characteristics", Journal of Guidance, Control and Dynamics, Vol. 21, No. 3,
1998, pp. 391-399.
4 Coton, F., Smrcek, L., Patek, Z., "Aerodynamic Characteristics of a Gyroplane
Configuration", Journal of Aircraft, Vol. 35, No. 2, 1998, p. 274 - 279.
5 Leishman, J.G., The Development of the Autogiro: A Technical Perspective,
Journal of Aircraft, Vol. 41, (2), 2004, pp. 765-781.
6 Leishman, J.G., Principles of Helicopter Aerodynamics, Cambridge University
Press, 2nd Edition, 2006, ISBN 0-521-85860-7.
7 Prouty, R. W., Helicopter Performance, Stability and Control, Robert E. Krieger
Publishing Co., Malabar, FA, USA, 1990.
8 Sheldahl, R. E., Klimas, P. C., Aerodynamic Characteristics of Seven Aerofoil
Sections Through 180 Degrees Angle of Attack for Use in Aerodynamic Analysis
of Vertical Axis Wind Turbines, SAND80-2114, Sandia National Laboratories,
Albuquerque, New Mexico, USA, 1981.
9 Nikolsky, A.A., Seckel, E., An Analytical Study of the Steady Vertical Descent in
Autorotation of Single-Rotor Helicopters, NACA TN 1906, Washington, 1949.
10 Wheatley, B., Bioletti, C., Wind-Tunnel Tests of a 10-foot-diameter autogyro
Rotor, NACA TR 536.
11 Wheatley, J.B., The Aerodynamic Analysis of the autogyro Rotating-Wing
System, NACA TN 492, Langley Memorial Aeronautical Laboratory, 1934.
12 Wheatley, J.B., An Aerodynamic Analysis of the Autogiro Rotor with a
Comparison between Calculated and Experimental Results, NACA TR 487,
1934.
13 Carpenter, P.J., Lift and Profile-drag Characteristics of an NACA 0012 Airfoil
Section as Derived from Measured Helicopter-rotor Hovering Performance,
NACA TN 4357, Langley Memorial Aeronautical Laboratory, Langley Field, VA,
USA, 1958.
14 www.cyberiad.net/foildata.htm, visited in the fall 2005
15 Houston, S. S., Modelling and Analysis of Helicopter Flight Mechanics in
Autorotation, Journal of Aircraft, Vol. 40, No. 4, 2003.
16 Chen, R. T. N., A Survey of Nonuniform Inflow Models for Rotorcraft Flight
Dynamics and Control Applications, NASA TM 102219, Ames Research Centre,
California, USA, 1989.
17 Houston, S. S., Brown, R. E., Rotor Wake Modelling for Simulation of
Helicopters Flight Mechanics in Autorotation, Journal of Aircraft, Vol. 40, No. 5,
2003.

18 Bielawa, R. L., Rotary Wing Structural Dynamics and Aeroelasticity, Second

edition, AIAA Education series, 2006. ISBN 1-56347-698-3
19 Kwon, Y. W., Bang, H., The Finite Element Method Using MATLAB, Second
Edition, CRC Press, 2000, ISBN 0-8493-0096-7.
20 Sissingh, G., "Contribution to the Aerodynamics of the Rotating Wing Aircraft",
NACA 921.
21 Bousman, W.G., Young, C., Toulmany, F., Gilbert, N.E., Strawn, R.C., Miller, J.V.,
Maier, T.H., Costes, M., A Comparison of Lifting-Line and CFD Methods with
Flight Test Data from a Research Puma Helicopter, NASA TM 110421, Ames
Research Center, Moffett Field, CA, USA, 1996
22 Friedman, P. P., Rotary-Wing Aeroelasticity: Current Status and Future Trends,
AIAA Journal, Vol. 42, No. 10, October 2004.
23 Wilkie, W.K., Mirick P.H., Langston, Ch.W., Rotating Shake Test and Modal
Analysis of a Model Helicopter Rotor Blade, NASA TM 4760, ARL TR 1389,
Langley Research Center, Hampton, VA, USA, 1997.
24 Trchalk, J., Gillies, E.A., Thomson, D.G., Aeroelastic Behaviour of a gyroplane
Rotor in Axial Descent and Forward Flight, 32nd European Rotorcraft Forum,
Maastricht, Netherlands, October 2006.
25 Anon, "British Civil Airworthiness Requirements, Section T, Light Gyroplanes"
Civil Aviation Authority CAP 643 Issue 3, Aug. 2003.
26 Bramwell, A.R.S., Done, G., Balmford, D., 'Bramwell's Helicopter Dynamics',
9th Edition, Butterworth Heineman, Oxford, 2001
27 Theodorsen, T. and Garrick, I., "Flutter calculations in three degrees of freedom'
NACA TN-741, 1941

11 Acknowledgments
The authors would like to acknowledge the support for gyroplane research provided
by the UK Civil Aviation Authority. This work is funded through a CAA ARB Fellowship.
The support and advice from Steve Griffin, Jonathan Howes, Alistair Maxwell,
Andrew Goudie and Joji Waites is highly appreciated. Many thanks go also to Dr
Richard Green and the departmental team of technicians from Acre Road laboratories
for help with the experimental measurements.

Figure 1 General arrangement of the AMRA gyroplane rotor blade model
Figure 2 Relation of blade twist, flap and inflow angle of a rotor in autorotation

Figure 3 Trends of NACA 0012 lift coefficient obtained for different values of Mach
number. Obtained with the aid of approach described in Ref.5
Figure 4 Trends of NACA 0012 drag coefficient obtained for different values of
Mach number. Obtained with the aid of approach described in Ref. 5

Figure 5 Induced velocity relations in vertical flight (speed of climb on the x-axis,
induced velocity on the y-axis). Reproduced from Ref. 23
Figure 6 Internal structure of the rotor blade from Montgomerie-Parsons research

Gyroplane. Left section comes from the root of the blade while the right
one was located close to the tip of the rotor blade

Figure 7 Span-wise mass distribution of the blade. The total weight of one blade
was 11.95kg
Figure 8 Span-wise distribution of blade c.g.

Figure 9 Arrangement of the experimental measurements of EA position and

torsional stiffness of the blade
Figure 10 Dependence of blade twist upon torsional loading at three span-wise

stations

Figure 11 Stabilization of rotor speed and corresponding torque equilibrium during

steady axial autorotative flight

Figure 12 Time behaviour of rotor speed for different lengths of pre-rotation
Figure 13 Trends of thrust (for single blade) and speed of descent during steady
autorotative descent

Figure 14 Span-wise distribution of torque during steady autorotation. The right-

hand side figure was reproduced from Ref. 14 and Ref. 20

Figure 15 Comparison of values of resultant force coefficient as a function of rotor

disc angle of attack obtained by experimental flight measurements (top)
and value of cR predicted by the simulation for axial flight (i.e. DD = 90deg)
and zero blade incidence angle. The figure on the top was reproduced
from Ref. 14 and Ref. 20

Figure 16 Rotor speed and blade flap angle in stable forward flight

Figure 17 Comparison of predictions of rotor blade teeter and G-UNIV experimental

data
Figure 18 Comparison of predictions of rotor blade teeter and G-UNIV experimental

data

Figure 19 Comparison of the first torsional mode shape computed by the AMRA and
published data
Figure 20 Distribution of blade torque over rotor disk during steady forward flight in
autorotation estimated by AMRA

Figure 21 Distribution of blade torque over rotor disk during steady forward flight in
autorotation published by Leishman [Ref. 6, pg 247]
Figure 22 Southwell plot of McCutcheon rotor blade showing correct qualitative

behaviour

Figure 23 Southwell plot Arospatiale SA330 Puma helicopter rotor blade showing
correct qualitative behaviour. Data in the plot are non-dimensionalised
with : = 28rad/s
Figure 24 Distribution of blade torsional deflection obtained with the aid of AMRA

Figure 25 Distribution of blade vertical displacement in bending obtained with the

aid of AMRA
Figure 26 Aeroelastic instability during axial flight in autorotation as predicted by

AMRA with simplified model. Data used: blade torsional stiffness of
100N.m2/rad, elastic axis at 32%, and blade centre of gravity at 40% of
blade chord

Figure 27 Aeroelastic instability during axial flight in autorotation as predicted by

AMRA. Coupled FEM model of blade torsion and bending used. Data
used: blade torsional stiffness of 400N.m2/rad, elastic axis at 32%, and
blade centre of gravity at 40% of blade chord
Figure 28 Aeroelastic instability during forward flight in autorotation as predicted by

AMRA. Simplified model of blade dynamics (equivalent spring stiffness
for modelling both torsion and bending). Data used: blade torsional
stiffness of 300N.m2/rad, elastic axis at 32%, and blade centre of gravity
at 40% of blade chord

Figure 29 Aeroelastic instability during forward flight in autorotation as predicted by

the AMRA. Coupled FEM model of blade torsion and bending used. Data
used: blade torsional stiffness of 600N.m2/rad, elastic axis at 32%, and
blade centre of gravity at 40% of blade chord
Figure 30 Stability boundary for autorotative flight

Figure 31 Flutter Boundary for a Helicopter Rotor Blade
Figure 32a) Offset Hinge

Figure 32b) Hub offset sketch
Figure 33 Stability boundary for a gyroplane rotor in axial flight

Figure 34 Gyroplane rotor stability boundary at advance ratio 0.2
Figure 35 FEM stability boundary

Figure 36 Effect of pitch axis

Appendix 1 - Calculation of Blade Element Velocities
Referring to Figure2, if i is the incidence angle of the rotor disc, the angle between
the blade longitudinal axis and the horizontal plane can be expressed as:
iB E i cos\
(A1.1)
therefore, the vertical component of the inflow velocity can be expressed as
3c
UP Vd cosE i cos\ Vh sin E i cos\ cos\ vi cos E Er T y EA cosT
4
(A1.2)
The horizontal component of inflow velocity is given by the following equation.
UT Vh cos i Vd sin i sin\ :r cos E T 3c y EA sin T

4
(A1.3)
Component of inflow velocity that is tangential to the rotor disc and parallel with blade
axis is usually neglected, especially if axial flight is considered. It is given below.
Ur Vh cosE i cos\ cos\ Vd sin E i cos\ vi sin E
(A1.4)
The above equations describe inflow velocity components for general flight
conditions and they were used in AMRA model of a gyroplane rotor. However, the
equations can be modified and simplifying assumptions can be made when
describing axial flight or high speed forward flight. In axial flight, horizontal speed is
S rad. Thus,
negligible and D J i D D J 0 and:
D
2
3c
UP Vd cos E vi cos E Er T y EA cosT
4
3c
UT :r cos E T y EA sin T
4
Ur Vd sin E vi sin E
(A1.5)

If both flapping angle and pitch angle of the blade are small, equations (A1.5) can be
rewritten.
3c
UP vi Vd Er T y EA
4
3c
UT :r TT y EA
4
Ur Vd E vi E
(A1.6)
Components of inflow velocity can be expressed in different if following substitutions
are made13
r Vd vi 1 3
x O y c y EA
R :R R4
V
cos D D P P:R V cos D D
:R
(A1.7)
Hence, equations (A1.7) can be written in different form
E T
UP :R O x y
: :
T
UT :R x y T
:
Ur :ROE
(A1.8)
The above equations can be further modified with the aid of the equation below
1 d dt d d
: dt d\ dt d\
(A1.9)

Using the above transformation, we get
dE dT
UP :R O x T y
d\ d\
dT
UT :R x y T
d\
Ur :ROE
(A1.10)
In case of high speed forward flight, rotor disc incidence (i) is very small and hence
the assumption can be made that DD|Y. Therefore, equations, (A1.2 - 1.4) can be
rewritten in the following manner:
3c
UP Vd cos E Vh sin E cos\ vi cos E Er T y EA cosT
4
3c
UT Vh sin \ :r cos E T y EA sin T
4
Ur Vd E Vh cos E cos\ vi sin E

(A1.11)
Since rotor disc incidence is negligible, horizontal and vertical components of free-
stream velocity are Vh=V cos DD and Vd=V sin DD. Further simplifications in the above
equations can be made with the aid of the assumption that flapping angle (E) and
blade twist (T) are small.
3c
UP Vd vi Er T y EA cosT VE cos D D cos\
4
3c
UT :r T y EA T V cos D D sin \
4
Ur Vd E Vh cos\ vi E
(A1.12)

Ignoring negligible terms, it follows from (A1.7), that:
x y
UP :R O E T PE cos\
: :
UT :r x P sin \
Ur :ROE P cos\
(A1.13)
Finally, transformation shown in (A1.9) yields in the equations below.
dE dT
UP :R O x y PE cos\
d\ d\
UT :r x P sin \
Ur :ROE P cos\
(A1.14)

Appendix 2 - Blade Equations of Motion Using a Lagrangian Approach

The rotor blades are modelled as perfectly rigid beam, which means that both flap and
twist of the blade are constant along the blade span. The blade has three degrees of
freedom - in flap, torsion and rotor speed. Lagrangian equations of motion were used
for the dynamic model of the blade. The system of equations consists of three
differential equations that describe dynamic behaviour of the blade in pitch (twist),
flap and rotation. General arrangement of blade equation motion is shown below.
d wT wT wV
ME
dt wE wE wE
d wT wT wV
M\
dt w: w\ w\
d wT wT wV
MT
dt wT wT wT
(A2.1)
In the above equations, T is blade kinetic energy and V is potential energy of the blade
and ME, M\ and MT are forcing moments. The potential energy of the blade consists
of a component due to flexibility of the blade (VF) and a component due to mass of
the blade (VM). Span-wise mass distribution of the blade was assumed to be
homogenous during derivation of the equations of motion. Hence, T and V of a
gyroplane rotor having weight m, length b, flexural stiffness kE, torsional stiffness kT
and offset of elastic axis from centre of gravity yg are as follows:
b
m
2b 0
T r. r dr
V VF VH
1 1
R R
mg mg
y g 1 sin T dr r 1 sin E dr
R 0
VF k E E 2 kTT 2VH
2 2 R 0
(A2.2)
After integration, equations describing kinetic and potential energy of the blade give
1 2 2 1 2 2 1 1
T mb E my gT mby g ET cosT my g2 E 2 sin 2 T my g2 :E sin T cosT cos E
6 2 2 2
1 1
my g2 :T sin E mby g E: sin E cosT my g2 : 2 cos 2 T sin 2 E sin 2 T
2 2
1 1 1
mby g :T sin T cos E mb 2 : 2 cos 2 E mby g : 2 sin T sin E cos E
2 6 2
(A2.3)

1
V
2

R
k E E 2 kTT 2 mg 1 sin E y g 1 sin T
2
(A2.4)
Both blade kinetic energy and the final form of Lagrange's equations of motion were
verified with the aid of MAPLE and MATLAB software packages. Forcing
moments of the blades are computed from the output of the aerodynamic model of
the blade. Each blade is divided into ten span-wise elements and aerodynamic forces
and physical properties of the blade are defined in the middle of each element (see
Fig.A2.1). Thrust, in-plane force, torque and pitching moment of the j-th span-wise
blade element can be calculated using equations below. Boundaries of the element
of the blade are defined by dimensionless span-wise coordinates xj-1 and xj. Variable
yc is the offset of aerodynamic centre of the blade element from its centre of gravity
and R is blade length (i.e. radius of the rotor).
dT j dL j cos I dD j sin I
dH j dL j sin I dD j cos I
x j x j 1
dQ j R dH j
2
yc j yc j1
dM j dL j cos D dD j sin D dM c / 4, j
2

(A2.5)
Using the variables defined in the previous equation, forcing moments in system of
equations (A2.1) are
n
x j x j 1
ME R dT j
j 1 2
n
M\ dQ
j 1
j
n
MT dM
j 1
j
(A2.6)

Figure A2.1 Blade Element Arrangement

The Aerodynamics of Gyroplanes

Uploaded by

Copyright:

Available Formats

The Aerodynamics of Gyroplanes

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

The Aerodynamics of Gyroplanes

Uploaded by

Copyright:

Available Formats

What is the document about?

What is the document about?

What are Lagrangian equations of motion used for?

What are Lagrangian equations of motion used for?

Safety Regulation Group

CAA Paper 2009/02