INTRODUCTION TO [yeTrtel MMA AelKe) Thomas W. StrganacINTRODUCTION TO FLIGHT TEST ENGINEERING Second Edition Revised Printing Donald T. Ward and Thomas W. Strganac Texas A&MCopyright © 1996 by Donald T. Ward ‘Copyright © 1998, 2001 by Donald T. Ward and Thomas W. Strganae. Revised Printing ISBN 0-7872-7500-X ‘Kendall Hunt Publishing Company has the exclusive rights to reproduce this work, to prepare derivative works from this work, to publicly distribute this work, to publicly perform this work and to publily display this work. ‘All sights reserved. No part ofthis publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any ‘means, electronic, mechanical, photocopying, recording, or otherwise, ‘without the prior written permission of Kendall/Hunt Publishing Company. Printed inthe United States of America 098765432PREFACE i ‘The purpose of this book is to consolidate the fundamental principles used in classical performance and flying qualities flight testing of manned aircraft. intended for use as ‘an introductory text for undergraduate students or as a self-teaching reference for an en- gineer newly engaged in flight tests. Worked examples are included in each chapter. The first half of the book covers performance measurements. Chapter 1 is devoted to an overview of why flight tests are conducted and what the most important constraints are. Chapter 2 reviews the standard atmosphere and applies basic aerodynamic equations to the basic measurement system found in virtually every airplane, the pitot-static system. ‘erated tum, and the level acceleration fight test methods. Chapter 4 briefly introduces propulsion systems, both for propelier-driven and jet-powered airplanes and summarizes Useful relationships for determining range and endurance for such airplanes. Speed power flight tests commonly used to collect cruise performance data are introduced to close out this chapter. Chapter 5 rounds out the performance section of the book by out- lining the equations used to estimate takeoff and landing performance and then describing measurements that must be taken to document appropriate measures of merit during these critical phases of fight. ‘The second half of the book deals with aircraft stability and control, concentrating on measurements that must be made to ascertain flying qualities. Chapter 6 covers the foun- dations of longitudinal static stability, concluding with a discussion of the fight test tech- niques often used to obtain such data. Chapter 7 does the same thing for maneuvering stability. Chapter 8 summarizes both the theoretical differences in the equations of mation ‘and the fight test methods used to measure static lateral-directional stability. Chapter 9 is ‘a concise summary of dynamic stability and control for both symmetric (longitudinal) mo- tions and asymmetric (lateral-directional) ones. The importance of qualitative pilot ratings in describing aircraft flying qualities and their usefulness as @ communications too! be- tween the test pilot and the test engineer are stressed. The chapter concludes with a dis- cussion of typical fight test techniques, including several practical ways of interpreting dy- namic response data. Chapter 10 introduces post-stall fight tests, intending to interest the beginning flight test student in more advanced topics. Gratitude is due to many people. First and foremost, for their unflagging support, | thank my family, especially my wife, Joyce. It is due to her patience with my early and late hours that the manuscript is finally finished. She also encouraged me throughout my twenty-seven years in the flight test profession, even though she undoubtedly often won- dered why | was so obsessed. | thank her and dedicate this effort to her. All the students who have been exposed to my attempts fo teach this subject at the United States Air Force Test Pilot School, at Texas A&M, at the University of Kansas, and at the United States Air Force Academy have contributed in no small way to this effort. Finally, thanks are due to all of the colleagues who reviewed these pages. Dr. Richard Howard of the Naval Postgraduate School deserves special mention for his contributions both as a PhD ‘student who understood the practicality of the subject and as a peer who made insightful suggestions. The faculty of the Aerospace Engineering Department at Texas A&M Unt versity were wholly supportive, especially Dr. Walter Haisler, Dr. John Junkins, and Dr. ‘Tom Pollock. | hope this subject will provide as much pleasure and challenge to the reader as the writing of this book has brought to me. Donald T. Ward March 1993,v PREFACE (Continued) Preface to the Second Edition The second edition of Introduction to Flight Test Engineering has been expanded to include a new chapter (Chapter 10) outlining the background and techniques used to Prepare for aeroelastic flight tests, a subject usually ignored in such an introductory vol- ume. This addition is the major change in this edition. Of course, the other material has been slightly rearranged to accommodate this insertion and known errors in these other chapters have been corrected. Dr. Thomas W. Strganac, joins me in providing this new material. This new addition to the text is based largely on the teaching experience of both authors, notably on their short course, taught since 1991, titled "Hazardous Flight Tests". This course, sponsored by the Continuing Education Division of the University of Kansas, has provided much of the new material. We are very appreciative of the assistance pro- vided by Mrs. Jan Roskam and her staff in this endeavor. Donald T. Ward July 1998 Of course, both of us are indebted to even more individuals than were mentioned in the original preface for their help in producing this text. | am indebted to those individuals who provided the many invaluable experiences afforded me during my 15 years as an engineer with NASA and 9 years on the faculty at Texas A&M University. The development and presentation of the "Hazardous Flight Test” short course with my co-author has been an extremely rewarding experience and collaboration. | dedicate my efforts as an engineer, researcher, and professor to my family. Sincere appreciation is given to my wife, Kathy, who has selflessly supported my pursuit of my professional interests. To my son, Christo- her, who has begun to build his educational foundation in his desired profession (Pale- ontology) at the University of Texas, and to my daughter, Kasey, who continues to follow her love of life and horses, | hope the two of you will pursue your dreams with devotion. Thomas W. Strganac July 1998TABLE OF CONTENTS v Chapter. Section CHAPTER 1 Planning, Discipline, and Safety 4.1. Why Flight Test?. 1.2 Types of Flight Test. 1.24 ‘Stage of Development 1.22 Developmental Versus Operational Testing. 1.3 Factors to be Considered in Test Planning 1.3.1 Safety. 1.3.2 Cost 1.3.3 Schedule 1.4 Summary CHAPTER 2 Pitot-Static System Calibration... 24 i i 22 2.2.1 Freestream Static Methods .... 2.2.1.1 ‘Tower Flyby Method. 2.2.1.2 Pacer Aircraft Technique 22.2 True Airspeed Methods... 22.3 Factors Affecting Position Error Measurements 23° Summary... BSRNBVRSaART AVA ooaanonaas CHAPTER 3 Ciimb, Descent, and Tum Performance Test 3.1 Theoretical Foundations 3.1.1 Goveming Equations. 3.1.1.1 Steady State Approximation .. 3.1.1.2 Accelerated Climb Equation 3.1.2 The Energy Approximation... 3.1.3 Forces in a Level, Unaccelerated Tum.. 3.1.3.1 Types of Tuming Performance .. 3.1.3.2. Limitations on Tuming Perfomance 3.1.3.2.1 Thrust-Limited (or Power-Limited) ‘Tuming Performance. 3.1.3.2.2 Lift-Limited Tuming PerformanceTABLE OF CONTENTS (continued) Chapter. Section Poge Number 3.2 3.3 CHAPTER 4 Cruise Performance Tests 44 42 43 3.4, 3 Load-Limited (Structural) Turing Per- formance 3.1.4 Experimental Versus Analytical Pe 3.1.4.1 Experimental Methods... 3.1.4.2 Analytical Methods... Climb Performance Test Methods... 3.2.1 Climb Schedule Determination 3.2.1.1 Sawtooth Climb Method. 3.2.1.2 Level Acceleration Method 3.2.2 Performance Climbs... 3.2.2.1 Performance Climb Procedures. 3.2.2.2 Performance Climb Data Reduction.. 3.2.2.3 Summary of Rate of Climb Corrections. 3.2.3 Tum Performance Flight Test Techniques... 3.2.3.1 Level, Unaccelerated Tum Technique. 3.2.3.2 Level Acceleration Technique. ‘Summary...... Theoretical Foundations Att Propelier-Driven Aircraft 4.1.1.1 Propeller Efficiency. 4.1.1.2 Power Available... 4.1.1.3 Corrections to BHP Available 4.1.2 Jet-Powered Aircraft. 4.1.2.1 Buckingham's Pi Theorem. 4.1.2.2. Combining Engine and Airframe Parameters. 413 Endurance... 4.1.3.1 Propeller-Driven Aircraft... 4.1.3.2 Jet-Powered Aircraft. 414 Range J 4.1.4.1 Propeller-Driven Aircraft. 4.1.4.2 Jet-Powered Aircraft. Cruise Performance Test Methods... 4.24 ‘Speed-Power Test Method for Propelier-Driven Airplanes. 422 ‘Speed-Power Test Method for Jet-Powered Airplanes Summary .... 73 75 75 80 81 81 83 83 ReesTABLE OF CONTENTS (continued) vi Chapter. Section Page Number CHAPTER 5 Takeoff and Landing Flight Tests. 52 53 CHAPTER 6 Longitudinal Stability Tests .... 641 Critical Field Length and Balanced Field Length. Critical Engine Failure Speed ... Refusel Speed and Refusal Distance Ground and Air Minimum Control Speeds. 1.1.9 Safe Single Engine Speeds . 5.1.2 Performance Equations... Ground Run Equations 5124 6.1.2.5 513 5.1.3.4 5.1.3.2 Basic Parametric Equation Flight Test Methods. 5.2.1 Requirements for Takeoff and Landing Flight Tests 5.2.1.1 Use of Mathematical Models and Simulation .. 5.2.1.2 Position and Velocity Measurement Techniques. Tica Part 23 Test Groups. ‘Theoretical Foundations 644 644.1 6.1.1.4 Angular Velocity ‘Transformations. 6.1.2 Straight Flight Paths... 6.1.2.1 Neutral Point Concepts......vil IBLE OF CONTENTS (continued) (Chapter Section, 6.1.2.2 Elevator Trim Angle. 6.1.2.3 Elevator-Free Considerations. 6.1.3 Other Concepts of Static Stability. 6.1.3.1 Speed Stability... 6.1.3.2 Force Gradient for Speed Stability. 6.1.3.3 Flight-Path Stability 6.2 Longitudinal Static Stability Test Methods. 8.2.1 Stability and Control Flight Test Measurements. 6.21.1 Trim. 6.2.1.2 Friction and Breakout Forces ... 6.2.2 Stabilized Method. 62.3 Slow Acceleration/Deceleration Method 6.2.4 — Power Acceleration/Deceleration Method 6.2.5 Flight-Path Stability Method. 63 Summary. CHAPTER 7 Longitudinal Maneuverabilty Tests ... 7.1 Theoretical Foundations ... TAA ‘Steady, Wings-Level Pull-up ... 7.4.1.1 Elevator Angle per G = 7.1.1.2 Stick-Fixed Maneuver Point. 143 7.1.1.3 Stick Force per G.... ‘Stick Free Maneuver Point. 74.2 Tums. 7.2 Longitudinal Maneuverability Test Methods... 7.2.1 Symmetrical Pull-up Method. 7.2.2 Steady Tum... 7.2.2.1 Stabilized Load Factor. 7.2.2.2 Slowly Varying Load Factor. 153 7.23 Windup Tum 7.2.4 — Concluding Remarks. 7.2.4.1 Instrumentation Required .. 154 7.24.2 Estimation of Stability Derivatives. 157 7.3 Summary... 159 CHAPTER & Static Lateral-Directional Stability Tests 81 Theoretical Foundations 8.1.1 Directional Stability and Control...Chapter. Section 8.1.2 8.1.3 8.2 Steady State Flight Test Methods..... 821 8.22 8.2.3 8.3 Summary... CHAPTER dynamic Stability Tests... 9.1. Theoretical Foundations .... 9.1.1 9.1.2 9.1.3 9.4.4 TABLE OF CONTENTS (continued) 8.1.1.1 Weathercock Stabilit 8.1.1.2 Directional Control. 8.1.1.3 Effect of a Freely Floating Rudder .. 8.1.1.4 Yawing Moment Due to Lateral Control Lateral Stability and Control. 8.1.2.1 Dihedral Effect... 8.1.2.2 Lateral Control Power. 8.1.2.3 Rolling Moment Due to Directional Control Lateral-Directional Matrix Equations Test Methods to Determine Dominant Stability Coefficients ....168 8.2.1.1 Steady, Straight Sideslips. 8.2.1.2 Slowly Varying Sideslips... 8.2.1.3 Steady Tums with Rudder Fixed 8.2.1.4 Steady Tums with Ailerons Fixed . 8.2.1.5 Use of Steady Sideslip Data... Test Methods to Determine Control Effectiveness 8.2.2.1 Steady, Straight Sideslips at Different C.G. Locations. 8.2.2.2 Yawing Moment Produced by a Wing Parachute. 8.2.2.3 Measurements with Weights at Known Spanwise Locations. Estimation of Stability Derivatives from Steady Lateral- Directional Tests . 170 Definition of Terms and Assumptions. Equations of Motion.. 9.1.2.1 Longitudinal Modes of Motion... 9.1.2.2 _Lateral-Directional Modes of Motion .. Handling Qualities ... 91.3.1 Clan Lop Respence versus Open Lop Response, 9.1.3.2 Pilot Rating Scales .. Longitudinal Dynamics ... 9.1.4.1 Dimensional Form of the Equation: 9.1.4.2 Short Period Approximation...x TABLE OF CONTENTS (continued) Chapter. Section. 9.1.4.3 Phugoid Approximation. 9.1.4.4 Sensitivity to Longitudinal Stability Derivatives 9.1.4.5 Summary of Linear Longitudinal Dynamics 9.5 _Lateral-Directional Dynamics.. Dimensional Form of the Equations Roll Mode Roll Mode Approximation. Dutch Roll Approximation... Sensitivity to Lateral-Directional Stability Derivatives . 9.1.5.9 Summary of Linear Lateral-Directional Dynamics 9.2 Dynamic Flight Test Method: 9.2.1 Types of Control Inputs ... 9.2.1.1 Step input. 9.2.1.2 Singlet... 9.2.1.3 Doublet... 9.2.2 Phugoid Test Methods......... 9.2.3 Short Period Test Methods.. 9.2.4 Dutch Roll Test Methods. 9.2.5 — Spiral Mode Test Method: 9.2.6 Roll Mode Test Methods.. 9.2.7 Basic Data Reduction Methods .. 1 Transient Peak Ratio Method 2 Time-Ratio Method 3 Maximum Slope Method. A Separated-Real-Roots Method 9.2.8 Calculation of Lateral-Directional Parameters from Response Data 9.3 Summary... CHAPTER 10 Principles of Aeroelasticity . 10.1. Analytical Foundations... 10.1.1 Aeroelastic Definitions.. 10.1.1.1 Aeroelastic Divergence. 10.1.1.2 Control Reversal. 10.1.1.3 Lift Effectiveness.Chapter. Section LE OF CONTENTS (continued) CHAPTER 10 (Continued) 10.2 10.1.2 10.1.3 10.1.4 10.1.5 10.1.6 10.1.7 10.1.7 10.1.8 Flight Test Methods . 10.2.1 10.1.1.4 Flutter Aeroelastic Equations of Motion Aeroelastic Similarity Parameters. Static Aeroelastic Phenomena. 10.1.4.1. Aeroelastic Divergence. 10.1.4.2 Control Reversal. Dynamic Aeroelastic Phenomena. 10.1.5.1. Bending-Torsion Flutter 10.1.5.2 Example: Quasi-Static Aeroelasticity. 10.1.4.1. Aeroelastic Divergence... 10.1.4.2 Control Reversal Matched Point Analysis... FAR Advisory Circulars: Parts 23 and 25 . Wind Tunnel Tests... Ground Vibration Tests Flutter Excitation in Flight.. 10.2.1.1. Natural Atmospheric Turbulence.xi Chapter. Section. CHAPTER 11 WA 11.4.5 11.6 WAT 1.4.8 11.2 Post-Stall/Spin Test Preparation and Flight Safety. 11.24 11.2.2 14.3, Summary IBLE OF CONTENTS (continued) Pege Number (Continued) 11.1.3.2. Developing Phase .. 11.1.3.3. Fully Developed Phase. Flight Path in a Spir 11.1.4.1 Assumptions.. 14.1.4.2. Balance of Forces .. ‘Aerodynamic Factors in a Spin. 11.1.5.1 Autorotative Couple of the Wing. 11.1.5.2 Effect of Forebody Shape. 11.1.5.3 Effect of Damping Derivatives .. Effect of Mass Distribution. 11.1.6.1. Principal Axes : 11.1.6.2. Relative Density . 273 11.1.6.3 Center and Radius of Gyration ........ 273 11.1.6.4 Relative Magnitude of Airplane Moments of inertia..273, ‘Simplification of the Post-Stall Equations of Motion... 214 11.1.7.1 Simplifying Assumptions... 274 11.1.7. Simplifed Large Amplitude Moment Equations... 275 ‘Aerodynamic Conditions for Dynamic Equilibrium. 277 11.1.8. Pitching Moment Equation... 27 11.1.8.2. Rolling and Yawing Moment Equation: 279 11.1.8.3 Estimation of Spin Equilibrium States. .280 .282 282 .266 267 268 268 .269 269 271 272 272 Pre-Test Planning and Preparation 11.2.1.1 Test Objectives . 283 11.2.1.2 Data Requirements. .290 11.2.1.3 Instrumentation for Post-Stall Flight Tests .290 11.2.1.4 Test Team Training 291 Safety Precautions and Special Equipment . 11.2.2.1 Emergency Recovery Devices... 14.2.2.2 Chase Aircraft. 11.223 Procedural Precautions 11.2.2.4 Real-Time Monitoring of Post-Stall Tests.UST OF FIGURES Pressure Forces Acting on a Unit Element of Air. Temperature Variations for Two Standard Atmospheres ‘Schematic Diagram of a Pitot-Static System ... Correction Factor (AV¢ Instrument Error Calibration Curve Locations for Static Pressure Orifices... ‘Supersonic Pitot-Static Tube Relationships. Ratio of Altimeter to Airspeed Indicator Position Error Correction Tower Flyby Geometry... Use of Ground Block Calibrated Altitudes. Dimensions of a Typical Tower Flyby Facility .. Velocity Position Error Corrections for M < 0.3 Altitude Position Error Corrections for M < 0.3. Pressure Coefficient Versus Mach Number...... Airspeed Position Error Correction for Transonic Flight Condition: Forces Acting on an Airplane in a Climb... Forces in a Level Turn = V-n Diagram Thrust Required at Various Bank Angles. Parameters That Affect Thrust-Limited Tuming Performance ‘Summary of Tuming Performance Limitations .. ‘Sawtooth Climb Procedure... ‘Weight Correction for Sawtooth Climb Data Effect of Vertical Wind Shear on Sensed Rate of Climb Uses of P, Contours. Determining P, Graphically Specific Excess Power versus Airspeed. Example Performance Climb Data Card, ‘Tapeline Altitude Correction for Rate of Climb. Apparent Acceleration Correction... Tuming Performance Measures of Merit Xixv UIST OF FIGURES (continued) CChapter-Figure Number Page Number CHAPTER 4 Fig. 4.1 Types of Drag Fig. 4.2 Generalized Power Curve Fig. 43 Geometric Variables for a Propeller. Fig. 4.4 Engine Power Chart for Normally Aspirated Engines. Fig. 4.5 Engine Power Chart for Supercharged Engines Fig. 4.6 Critical Altitudes for Reciprocating Engines... Effect of Mixture Setting Engine Operating Charts... Fig. 4. ‘9 Dimensional Engine Performance: Constant Altitude, Constant TAS... Fig. 4.10 Net Thrust versus Altitude and Engine RPM Fig. 4.11 Generalized Net Thrust. Fig. 4.12 Generalized Thrust and Mass Flow Rate. Fig. 4.13 Generalized Fuel Flow Rate and Specific Fuel Consumption. Fig. 4.14 Typical Performance Map for a Fixed Geometry Engine. Fig. 4.15 Envelope of Fuel Flow Curves... Fig. 4.16 Determining Altitude to Fly Fig. 4.17 Pitot-Static and Instrument Corrections for Example 4.3. Fig. 4.18 Speed-Power Flight Test Card Corrections for Example 4.3. Fig. 4.19 Specific Endurance and W/6 versus Mach Number. Fig. 4.20 Specific Endurance Crossplots. Fig. 4.21 Specific Range Crossplots.... CHAPTER S Fig. 5.1. Takeoff and Landing Phases. Fig. 5.2 Takeoff and Landing Air Phases Fig. 5.3 Balanced Field Length Components. Fig. 5.4 Velocity Profiles for RA = CFL. Fig. §.5 Forces during Ground Run Fig. 5.6 Coefficients of Friction during Landing Ground Run Fig. 5.7 _ Effect of Aerodynamic Braking and/or Drag Chute .. Fig. 5.8 Empirical Values of AC; used in Equation 5.9. Fig. 5.9 Transition Distance Geometry. Fig. 5.10 Distance to 35-Foot Obstacle Height for Large Jet Transports {All Engines Operating) .. Fig. 5.11 Distance to 35-Foot Obstacle Height for Large Jet Transports (One Engine inoperative) ... Fig. 5.12. Distance to 50-Foot Obstacle Height for Four-1 Engine Propeller Transports (One Engine inoperative) ...(Chapter. Figure Number CHAPTER 6 Fig. 6.1 Fig. 6.2 Fig. 6.3 Fig. 6.4 Fig. 6.5 Fig. 6.6 Fig. 6.7 Fig. 6.8 Fig. 6.9 Fig. 6.10 Fig. 6.11 Fig. 6.12 Fig. 6.13 Fig. 6.14 Fig. 6.15 Fig. 6.16 Fig. 6.17 Fig. 6.18 Fig. 6.19 UST OF FIGURES (continued) Ba ‘Sign Conventions. Definitions of a and B Definitions of Euler Angles Body Axis and Inertial Angular Velocity Components.. Family of Cry-a. Curves for Various 8g .. Dependence of Borin 0" og Position .. ‘Trim-Slope Criterion for Determining the Stick-Fixed Neutral Point..122 Control Surface Floating Angle. ‘Tab-Slope Criterion for Stick-Free Neutral Point. Stick-Foced Speed Stability. : Variation of Longitudinal Stability Coefficients with Mach Number... 1.27 Transonic Effects on Static Longitudinal Stability Flight-Path Stability Limits ..... Differences Between Level, Unaccelerated Flight and a Wings-Level Pull-up... Effective Increase in Horizontal Tail Angle of Attack Due to Pitch Rate.. : Stick-Fixed Neutral and Maneuver Points. Load Factor in a Tum. Variation of Control Forces in Different Test Maneuvers Examples of Good and Bad Wings-Level Pull-up Data.. ‘Stick Force per G from a Windup Tum. Longitudinal Maneuvering Flight Test Data.. Control Surface Gradients. Yawing Moment Produced by Sideslip.. 162 Rudder Defiection versus Sideslip. 162 Aileron Deflection versus Sides! Effect of og Shift on Crp Wing Parachute Installations.ai (Chapter. Figure Number. CHAPTER 9 Fig. 9.1 Fig. 9.2 Fig. 9.3 Fig. 9.4 Fig. 9.5 Fig. 96 Fig. 9.7 Fig. 9.8 Fig. 9.9 Fig. 9.10 Fig. 9.11 Fig. 9.12 Fig. 9.13 Fig. 9.14 Fig. Fig. 9.16 Fig. 9.17 Fig. 9.18 Fig. 9.19 Fig. 9.20 Fig. 9.21 Fig. 9.22 Fig. 9.23 Fig. 9.24 Fig. 9.25 Fig. 9.26 Fig. 9.27 Fig. 9.28 Fig. 9.29 Fig. 9.30 Fig. 9.31 Fig. 9.32 UST OF FIGURES (continued) Possible Responses of Linear Dynamic Systems . ‘Spring-Mass-Damper Analog Characteristic Roots in the Complex Plane.. Motion Parameters on the Complex Plane . Responses to a Step Input.. Use of Pilot Opinion Ratings. Closed and Open Loop Block Diagram. Cooper-Harper Pilot Rating Scale . Computed Time Histories After an Elevator input Lateral-Directional Dynamic Response Stable and Unstable Spiral Modes of Motion... Dutch Roll Transient Response (Full 6 DOF Response to Rudder) .201 Control Inputs for Dynamic Testing Phugoid Oscillation... Short Period Response to a Doublet fora Light Twin Dutch Roll Oscillation Excited by a Rudder Doublet. Roll Mode Tests: 45°-45° Bank Change for a Light Twin. Measurements for TPR and MTPR Method: Transient Peak Ratio Versus Damping Ratio. Measurements for TR Method Chart for Determining ¢ and o, by TR Method Measurements for the Maximum Slope Method... Chart for Determining ¢ and w, by the Maximum Slope Method.....215 Measurements for the Separated-Real-Roots Method... Measurements for Dolbin's Modified Separated-Real-Roots Method... Roll Rate Response to a Step Input Components of a Roll Rate Response to a Step Input. Extracting the Spiral-Dutch Roll Mode Envelopes... Combined Roll Mode-Dutch Roll Mode Components... Semilog Plot to Estimate Dutch Roll Mode Damping Ratio ... Estimation of the Roll Mode Component... Altemative Calculation of the Roll Mode Time Constant CHAPTER 10 Fig. 10.1 Fig. 10.2 Fig. 10.3 Fig. 10.4 Interdisciplinary Nature of Aeroelasticity Oscillatory Motion \Velocity-Damping Diagram... Frequency-Velocity DiagramUST OF FIGURES (continued) ii Chapter. Figure Number Page Number CHAPTER 10 (Continued) Fig. 10.5 Frequency-Velocity Diagram Indicating Divergence. Fig. 10.6 Decaying Motion in the Time Domain Fig. 10.7 Divergent Motion in the Time Domain . Fig. 10.8 Undamped Motion in the Time Domain. Fig. 10.9 Mass Ratio Geometry........ Fig, 10.10 Free-Body Diagram for Aeroelastic Divergence Example .. Fig. 10.11 Boundary for Static Aeroelastic Divergence .... Fig. 10.12 Boundary for Static Aeroelastic Divergence (Mach Number Effects). ee 287 Fig. 10.13 Control Surface Reversal. Fig. 10.14 Bending and Torsional Flutter Fig. 10.15 Coupling of Flutter Modes .. Fig. 10.16 Solution for the Matched Point . Fig. 10.17 V-g Diagram for Five-Mode Analysis Fig. 10.18 Velocity-Frequency Diagram for Five-Mode Analysis.. Fig. 10.19 Flutter Envelopes from V-g/Matched Point Solutions Fig. 10.20 The Flutter Boundary Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 10.24 Flutter Excitation Schemes. 10.25 Typical Frequency Reponse. 10.26 Results of Peak-Hold Method. 10.27 Location of Acoslerometers for the X-29. 10.28 Predicted X-29 Flutter Boundaries... 10.29 X-29 Midflaperon Flutter Boundary Prediction 10.30 X.29 Flutter Flight Test Matrix .. 10.31 Sweep Effects .... CHAPTER ff Fig. Fig. Fig. Fig. Fig. Fig. Fig. Fig. 11.1, Classical Aerodynamic Stall. 11.2. Spin Phases. 11.3. Helical Flight Path in a Fully Developed Spin 11.4 Changes in Lift and Drag Coefficients at High Angles of Attack 11.5 Aerodynamic Mechanisms for Autorotative Moments ... 11.6 Lift and Drag on Advancing and Retreating Wing Panels .. 11.7 Aerodynamic Contributions from the Forebody Shape... 11.8 Relationship Between Body and Principal Axes..il UST OF FIGURES (continued) Chapter.Figure Number CHAPTER If Fig. 11.9 Relative Magnitude of Aircraft Moments of Inertia Fig. 11.10 Pitching Moment Coefficient in a Steady Spi Fig. 11.11 Angular Velocity Components in a Typical Upright Spin .. Fig. 11.12 Stabilizing and Destabilizing Slopes for C.z and Cy. Fig. 11.13 Comparison of Aerodynamic and Inertial Pitching Moment Coefficients. Fig. 11.14 Emergency Parachute Recovery Subsystem Design Parameters. Fig. 11.15 Post-Stall Flight Test Team Communications Net .CHAPTER 1 None CHAPTER 2 Table 2.1 CHAPTER 3 Table 3.1 Table 3.2 CHAPTER 4 Table 4.1 Table 4.2 CHAPTER S Table 5.1 Table 5.2 Table 5.3 CHAPTER 6 None CHAPTER 7 None CHAPTER 8 None CHAPTER 9 Table 9.1 Table 9.2 Table 9.3 Table 9.4 Table 9.5 UST OF TABLES Constants for the Standard Atmosphere. FAR 23.65 Climb Requirements (All Engines Operating) .... FAR 23.67 Climb Requirements (Critical Engine inoperative, Propeller in Minimum Drag Position) Cruise Data for a Propeller-Driven Airplane .. Nondimensional Jet Engine Performance Parameters Values for Runway Surface Friction Coefficients. Takeoff Conditions for a Light Twin Certification Landing Tests for a Light Twin Certification .. Linearized, Small Perturbation Equations Using Dimensional Derivatives Linearized, Small Perturbation Equations Using Nondimensional Longitudinal Data for a Subsonic Fighter (M = 0.2, sea level). Stbiliy Derivatives Signifcanty Affecting Longitudinal Oscillations 2. Lateral Directional Data for an Attack Airplane (M = 0 f= sea level).x UST OF TABLES (continued) Chapter. Table Number Rage Number CHAPTER 9 (continued) Table 9.6 Effect of Altitude and Airspeed on the Roll Mode of an Attack Airplane... Table 9.7 Effect of Aitude and Airspeed on the Dutch Roll Mode of an Attack Airplane. Table 8.8 Effects of Stability Derivatives on Asymmetric Modes of Motion.....202 Table 9.9 Typical Spiral Stability Requirements: ime-to-Double-Amplitude (seconds) Table 9.10 a of Basic Graphical Techniques for Determining and. 2m CHAPTER 10 Table 10.1 Mode Type, Frequency, and Nature of instability CHAPTER If Table 11.1. Spin Mode Modifiers. Table 11.2 F-4E Spin Modes .. Table 11.3 F3H Demon Computed Spin Modes Versus Estimated Spin Modes... Table 11.4 Progressive Test Phases for Post Stall Demonstration Table 11.6 Typical Sensors Needed for Stability and Control Data ..Chapter 4 PLANNING, DISCIPLINE, AND SAFETY Flight testing piloted and unmanned aerospace vehicles is an interdisciplinary process fundamental to the development of new systems and to the advancement of aeronautical knowledge. Engineers from all branches of the engineering sciences are needed to successfully put a new system through a thorough and complete flight test program. ‘Asrospace engineers often lead such efforts because of their special familiarity with ‘aerodynamics and the effects of vehicle configuration upon performance and stability and control. Electrical engineers are indispensable in developing appropriate instrumentation, in evaluating feedback control loops, and in evaluating electronic subsystems. Mechanical ‘engineers provide special expertise in designing mechanical, hydraulic and pneumatic subsystems. Computer specialists are essential to @ reasonable integration of the on- board computers and their seemingly unlimited capacity to control and display information for the crew or the ground operators. Flight testing is a complex process. It requires many different technical skills, as well as good judgment in managing the process. “Therefore, the right place to start @ discussion of flight test methods is with a brief philos- ophy of cost effective flight testing. 4.1. WHY FLIGHT TEST? Flight test organizations have @ reputation for costing too much and taking too much time, the cardinal sins for a program manager. If the test engineer finds a flaw in the de- ‘sign, the usual reply to the recommended fix is: “We cannot afford such a drastic change!” or "Why did you not find the problem in the preliminary tests?" The fight test team is al the first to hear from the disgruntled user: "You gave me another piece of junk! Why did ‘you let them send it to the fieki?" The implications of these comments (and they are not altogether thetorical) is that the flight test engineer must do his job well and he must communicate effectively with those who control the purse strings if the product is to be useful. Exactly what is the purpose of flight testing? The introduction to the AGARD Flight Test Manual! lists three fundamental reasons for fight testing of piloted aircraft: © To determine the actual characteristics of the machine (as contrasted fo the computed or predicted characteristics) ‘© To provide developmental information © To obtain research information There is no reason to suggest that the testing of unmanned vehicles has purposes App; that is, the Ap of interest is error in sensed static pressure and, Correspondingly, dh —> - Ape, the altimeter's static pressure error correction. Then, eqn. (2.5) becomes: APp APp=PstaGoAlipe of Ah, (2.24) ome Pe” pata 90 AP 5 2 Then, Pstao = PoSstaIo = 0.07647490 54 (2.25) Psta9o 904 Pst _ Pata OF au ki )) * But ogi = P84 = Pst Pi _ Pi fy, Bith_p, + ky #0,0r sa "po Pi Po Po 7 ") -20(n-m) ") =e RO) ge 0 0 For /= 0, we substitute: --22_ = 5.25592, Ty= 518.67 °R, ky = -0.00356616 "RI koR 4.2550 Cott =( ee) (2.26) -22-(n-my) Fori= 1, CP TR 227 Po tropopause = 0.2979725¢7 (0000048063707 )(h-36089.239) Fortunately, these equations for ogiy can be, and usually are, plotted and/or programmed for table lookup. Approximate relationships between Ap, and airspeed indications can also be obtained in @ straightforward manner. Assuming no error in sensed total pressure, measured dynamic pressure is: oe = PT Ps-‘Chapter 2 Pitot-Static System Calibration: 19 ‘The dynamic pressure without static pressure error is just: %™ PT Po Static pressure error is, therefore, just the difference between q¢ and qo, APp= Ps Pa= I~ Tg Rearranging eqn. (2.10) ’ $e (1,05) 7 p BRT This pressure ratio can be applied to any set of temperature and velocity measurements so tong as the process is approximately isentropic. Specializing this ‘expression to calibrated airspeed, which implies substituting pg for p and To for T; ; peti ep ‘ feos) ‘and, for instrument corrected airspeed Vip, Jc, Feplaces qc 1 Gee (1-0=ee | 7 P 208 Now, we can obtain an approximate expression relating App and AVp¢ in the same way Herrington® does by using only the expression above and the definition q,,, = Pr - Ps- if pr is measured with negligible error, then dg,,, = - dps. Differentiating and replacing de,, with - dp gives eqn. (2.29): Vie (,, (-1B we ae (2.28) ‘Approximating the differentials with finite differences: App = dp and AVpc = - dVia Vi - Ve fee tall ME (2.30) AVpc 8G 89 We have developed this approximation based on eqn. (2.10), which is a valid ‘expression for an isentropic process; that is, no shock waves can exist. Equation (2.30) is valid only if Vi, < a9. However, a similar approach, starting with the Rayleigh supersonic20 Pitot Static System Calibration, Chapter 2 pitot formula, gives an approximation when the pitot tube does have an attached normal shock wave. w Ge. (s¥) | (ety? 4 P 202 v2 toys 2 (2.31) Substituting ¥ = 1.4 for air: 7( v2 de _ (4235 250 v > Oz Xe “) 72 If we now replace V with Vi. p with po, and @ with ao, we can solve for the instrument ‘corrected dynamic pressure read on the ASI by using the following equations: % Ve)’ ( Vie? _,)°* ee. = se0a015q{“e) (r¥e ) -1 Po 80, Ao, _ (2.32)Chapter 2 Pitot-Static System Calibration 2 Ap, where <7 is in inches of Hg/knot. (Changing the constant to 2214.786 in eqn. pe ao, (2.32) gives We in pstifps.) 2o, ‘The approximate expressions in eqns. (2.30) and (2.32) for es are valid only for loc fairly small errors in pressure measurement; Herrington® says that AV,, must be less than 10 knots, which is usually quite adequate for well-designed pitot-static systems. Since both Apc and AVpe are functions of the error in measuring static pressure, they must be related. Having determined one, we should be able to calculate the other. Dividing eqns. (2.30) or (2.32) by eqn. (2.25), for a <1or Me > 4, respectively, and 0 then using the appropriate form of asig from either eqn. (2.26) or (2.27), Ah 2 Fe. «ons Ye |r024) Vie for Me ct (2.33) Ssta#0 25 8 Ane 48, AV p¢ (2.34) Equations (2.33) and (2.34) are plotted in Fig. 2.8 and are valid approximations if ‘Ape < 1,000 feet and AVp¢ < 10 knots, For larger position error corrections, Herrington hes developed and plotted more exact equations®, good for AVpc < 50 knots. If increased range is needed for this correction, the interested reader should consult Herrington's handbook. Example 2.3: An alplene is fying at 203 KIAS at an indicated pressure altitude of 29,750 feet. The out- side aic temperature ia -25°F for theee test conditions. The instrument corrections include AV;, =~ 3 knots, lige = 260 fet, and Ah, = 75 toot. Caladate Pe aYpe Vy, Voy 14 Me pc First, we will use our definitions to relate indicated and true sirepeed Vig = Vig + AVgg + AVe, Dut Vig # Vj + AVig = 903-3 = 300 knots Also, using ag = \(TANTTIESE(STEGT) = 1116.4 feetteec = 661.489 knots, from eqn. (2:33),2 Pitot-Static System Calibration, Chapter 2 (508001900) 02 3002 ia AV pe \ 0 99 661489 661.4892, To obtain ogy wo Need p from the pressure altitude, which is: N= + Ade + Age = 29,760 + 75 + 250 = 30,075 toot Equation (2.62) gives P= 0.29596 and p = 626.29 psf. Then, Po oa 89 __ 9.35914, 50 Mee ea19 (1716.55)(434.67)(0.0023769) ‘BVp0 it _250ft = AV pq = 80notS Vee = 35.169 0 tad From Fig. 2.4, AV, = - 15K18 = Vq-Voap 80 Vg = Vou) ~ 15kt9 = 300-15 Vo = 285knots Vo 285 Vi. = 47850 knots= 80898 fps ne nm My = te = 00 98 __ Ma = 079 VA1716.55)(494.67) 2.2 POSITION ERROR CALIBRATION METHODS From the preceding discussion of pitot-static system errors it should be obvious that any pitot-static system calibration must be primarily concemed with determining the static Position error. Gracey'9 classifies position error calibration methods under four headings: ‘© Freestream static pressure methods in which Ap, is obtained from mea- ‘surement of ps and Pao. © True airspeed methods in which App is derived from values of Vey calculated from ground speed measurements or from anemometer readings. ‘ A temperature method in which Ap, is determined from measured tempera- ture and a pressure-temperature survey. @ Mach number methods in which App is obtained from Mach number. Of these four types of calibration methods the first two are the ones most often used. They are especially well-suited for low speed and low altitude, although the first category of methods includes several techniques useful at high altitudes and airspeeds. Instru- mentation needs can be kept to a minimum with these two approaches. For brevity, we will describe in detail only three calibration procedures, two that are freestream static Pressure methods and one that is a true airspeed method. We will also restrict our de- tailed descriptions to subsonic methods. For the reader whose needs go beyond this in- troductory material, Gracey lists fourteen methods of calibration and summarizes relative accuracies (when data are available) and limitations. After describing the three selected techniques, the discussion will focus on the sensi- tivity of the position error correction to three major effects: gross weight changes, altitude variations, and Mach number deviations.Pitot-Static System Calibration Chapter 2, uoeu0g 40113 UORIsod so}eOIPU| peedsily O} soeWY JO CHE gz “Bld siouy - |24__________ Pitot Static Syston Calibration__ Chapter 2 221 Freestream Static Pressure Methods ‘These methods depend on accurate experimental determination of the difference be- tween measured static pressure and the actual freestream static pressure at each of the flight conditions. There are at least four ways to determine this differential pressure. First, ‘App can be obtained from a reference pressure source moving with the aircraft but located outside the pressure field of the vehicle. The test aircraft may camry this reference pressure sensor along (trailing cone or trailing bomb) or the calibrated reference sensor may be carried in another airplane (pacer aircraft) flying formation with the test vehicle. ‘Second, the reference value of pi. may be obtained by interpolating the pressure gradient optically (tower flyby) for low altitude calibrations or with tracking radar for medium and high altitudes. Third, p,, may be calculated by measuring p and T at the ground and as- suming a temperature gradient up to the altitude where temperature is measured. Fourth, Px at @ given altitude can be obtained from measuring indicated pressure and temperature and noting the changes in height from an altitude where Ap, is known. Ordinarily, for these freestream static methods where pressure differences are measured by aircraft instruments, the altimeter position error, Afipc, is the measured correction bCaUS® Mg iS relatively easy to determine and the altimeter provides better resolution than the ASI. For example, at sea level and 150 knots, @ AVpc of 2 knots corresponds approximately to a Alipg Of about 38 feet and to @ Ahipg of approximately 116 feet at 500 knots. Let us now single out the tower flyby technique and the pacer aircraft technique and discuss in detail the data collection and data reduction procedures for each of them. We will also suggest some of the advantages and limitations of each approach. Fig. 2.9 Tower Flyby Geometry 2.2.1.1 Tower Elyby Method. In this technique, the calibrated altitude of the vehicle is determined by triangulation relative to a surveyed point on the ground (Fig. 2.9). The airplane is flown down the tower fiyby fine at an altitude approximately level with the eyepiece in the tower. The altitude and the airspeed both must be completely stabilized as the airplane passes directly abeam the tower. The height of the airplane in the sighting grid (y) and the pressure altitude in the tower should be recorded for each pass. Then the calibrated altitude of the airplane's altimeter can be calculated by adding the pressure attude of the tower to Ah, where Ah=z X.Chapter 2 Pitot-Static System Calibration 2 Figure 2.10 illustrates how calibrated altitudes from the two ground blocks are used to obtain he at each test point. The data reduction for tower flyby information proceeds along the following lines. Determine the calibrated altitude of the test airplane's altimeter at each test point during the flight. The so-called "ground block" method, which assumes a constant change of calibrated altitude with time of day, is the most commonly used means. of obtaining he for each test point. The flight test crew must obtain a preflight and postfight reading of the aircraft altimeter (set at 29.92 in Hg, of course) at a point of known elevation on the ramp. Then, Meaany = Mame + Mic Fig. 2.10 Use of Ground Block Calibrated Altitudes Eyepiece or Film Plane I Optical Grid i! Elevation of Zero Grid Line ‘315.55 feet MSL Fig. 2.11. Dimensions of a Typical Tower Flyby Facility Using this interpolated value for the calibrated altitude on the ramp, calculate the calibrated altitude at the eyepiece in the flyby tower. Figure 2.11 is a sketch illustrating @ typical tower flyby course. The eyepiece in the tower is 315.55 feet above mean sea level, the optical grid is 139.5 inches from the eyepiece and grid lines are 1 inch apart. The bot- tom grid line is at the same level as the eyepiece. The flyby line is 977.75 feet in front of the tower eyepiece. In using this method you must also take into account the height above the ground of the altimeter for the perticular airplane you are testing. ealerapiece = Healanp * AAcaleepiece~ AM calatnaar28 Pitot-Static System Calibration, ‘Chapter 2 ‘The calibrated altitude of the test aircraft is then obtained from - I, Neal ealgepece * Tah (2.35) where gz is the correction for nonstandard temperature conditions. Next, we must find the test day pressure-related parameters with instrument correc- tions. For this introductory course, we will consider only ho, Vi, and do,» leaving Mj, to more detailed descriptions. ig, = hy+ Ate and Vig. = Vj + AVje. AS we saw in Section 2.1.3.3, 2)35 Uo -{r-02!@) -1 3 Finally, we standardize Alp, and AVic to a reference altitude approximately equal to standard day elevation at the test site. The tower eyepiece elevation in Fig. 2.11 is ap- proximately 315 feet; therefore, we will correct each test day measurement to standard day at this elevation. No lag correction is necessary for steady state test data, so: Alipo, = Peat ~ icy To correct to standard day conditions at 315 feet, considering only smail altitude and angle of attack changes: = 0315 Ahpesis = Mpc 9 where @ = 1 - 0.00000687559h ca) 0.9978343 $0, Ahpeass™ Mpc Gy (2.38) To correct Vic, 10 Vicgy, assume that each test point was flown at 315 feet and the test day Mach number. Then, we observe that Freus Tig _ Weg “Peas Py POBDIE For subsonic airspeeds, we solve this expression for Vies,¢ 0.2857142 ( eats ; eat. P0315 AVpeays Can then be obtained from eqn. 2.33 or from Fig. 2.8, knowing Apa, This data reduction scheme has been simplified considerably by assuming small dif- ferences in altitude measurements and test day calibrated altitudes, small correction terms, and subsonic fight conditions. For other conditions (Mj. > 1, for example), different ‘equations must be used. Do not apply these data reduction equations to data for which Vicars = 20 (2.37)Chapter 2, Pitot-Static System Calibration. a Ahpe > 1000 feet Or AVpo > 10 knots. Consult Herrington® and/or other fight test handbooks for complete details that are beyond the scope of this book. ‘The tower flyby method requires very lttle instrumentation and is quite a straightfor- ward, though tedious, procedure. Photographic or theodolite equipment are often used to record the aircraft elevation at each measurement point, but it is not essential and hand- recorded optical data are also usable. Gracey"? concluded from tests at NASA's Langley Research Center that App can be measured within + 1% of qo, even for very low speeds (M= 0.15 and 90 knots). ‘The method is least accurate for such low speed points because angle of attack effects dominate. For higher speeds (M = 0.3 and 190 knots), accuracy within + 0.2% of qc was found. Obviously, altitude effects cannot be investigated using the tower flyby method and the speed range is somewhat limited because the technique is a low altitude procedure. The lowest test speeds must offer a safe margin above stall ‘speed and the highest test speed is limited to the maximum level flight speed of the vehicle at low attitude. 2.2.1.2 Pacer Aircraft Technique. There are at least two variations of the pacer method, one in which the reference source is in an airplane flying formation with the test aircraft and one in which the test aircraft fles at various speeds past the reference air- plane. The reference carrier simply maintains a constant speed and altitude. The latter method, though less demanding for the reference airplane, is not as accurate as the for- mation method because of the practical difficulties in determining small differences in the height of the two aircraft and in obtaining simultaneous readings. Lag errors are also more likely to affect data from the latter technique. When the calibration tests are flown in formation, the test aircraft may be either the lead or the wing man. The important points are that the two aircraft are flown at the same level, that their speeds are perfectly matched, and that they are far enough apart so that neither of the pitot-static systems are affected by the pressure field of the other. The last requirement means that at least helf a wing span in spacing is required between the two aircraft. However, the precise formation flying required to guarantee matched altitudes and airspeeds for each point suggests that the lateral spacing should be little more than this half @ wing span. Level, in the case of dissimilar aircraft, means that the two altime- ters being used should be at the same height. Data reduction for the pacer aircraft (formation) method is relatively simple. First, find hieai and Vea for the pacers reference altimeter using its indicated readings, the laboratory corrections for its instruments, and the known static position error corrections. eapacar™ Mpacer * Mepser + AM, Vealpecer= Vipscor * AVicpecer + AV pe, Itis assumed, of course, that the pacer aircraft system has a full and accurate set of cali- brations. Next, obtain Mig, and Vig for the test aircraft just es was described for the tower flyby technique in Section 2.2.1.1. Then, calculate Ape and AVpc for the test aircraft under the ‘assumption that both aircraft were at the exactly the same test day calibrated altitude and airspeed at each test point. hoc, = Mealpecer~ Meee” SPictect28 Pitot Static System Calibration, Chapter 2 Correct Aig, to the nominal altitude for the test point just as we did the tower flyby data with eqn. (2.36). AV pce = Veiner” View” Vices This approach gives AVp, independent of the approximate conversion equations de- veloped previously (eqns. 2.33 and 2.34) and offers a crude method of checking the total pressure error if any exists. (Caution: The measurement errors in the pacer method may be large enough to imply erroneously that the total pressure error is significant. Use this ‘comparison only as an indication that further tests may be required. Remember you are obtaining the correction by differencing two large numbers.) The pacer aircraft method, particularly when the two aircraft are closely matched and ‘can be flown in formation, is a very accurate calibration method. Gracey0 reports an ac- curacy in Be of 20.7% at M = 0.5 and +0.2% at M = 1.0 for this method. Also, altitude effects can be fully investigated within the envelope of the two vehicles and ‘supersonic calibrations are feasible. Low speeds and, for some aircraft, transonic speeds may cause ‘some difficulty in stabilizing in formation because of the handling qualities of the ‘specific aircraft. Finally, the pacer aircraft method lends itself to collecting static position error cor- fections rapidly, though the cost of operating a pacer aircraft may not be warranted, 2.2.2 True Airspeed Methods Static position errors can also be calculated by measuring true airspeed and deter- mining Voqy from the measurements. The speed-course technique is that uses this ap- proach. Obviously, the wind must be recorded at the time and altitude of the test; wind effects can be minimized by flying reciprocal headings if the wind is constant. Tests should be conducted when the wind speed is as near zero as possible to reduce such uncertainties. Calibrations are often carried out near sunrise or sunset for this reason. Fig. 2.12 Speed Course Geometry Figure 2.12 is a sketch depicting the ground track of a speed course calibration test. Notice that the aircraft is aligned with a heading parallel to the desired course; no drift cor- rection is used. The data needed from each pass are hj, Vj, T;, and time between the two Points. If no other instrumentation is available, observers can note the time the airplane asses over each of the known points. However, radar or optical trackers will reduce the likelihood of human error in recording time accurately. The test should be planned so the timing interval is large compared to the error in marking the start and stop times. TheChapter 2, Pitot-Static System Calibrations. 29 most important parameter to be held constant is indicated airspeed and, if turbulence or poor pilot technique causes a variation of over +1 knot, the data point is suspect. Accu: rate ambient temperature at the test altitude is also necessary. This temperature becomes particularly important whenever medium altitude calibrations are attempted. To reduce speed course data we must first find the average true airspeed. +) fy te. where d is the course length; ty and f2 are the times required to traverse the course in reciprocal directions. Next, find the freestream temperature at the test altitude. If precise weather data col- lection equipment is available, one might obtain this information from the meteorologist. If the indicated temperature from the sirplane is used, it must be corrected for the gage labo- ratory correction and the temperature recovery factor. Since the temperature gage reads the total temperature except for the kinetic term whitch requires a recovery factor because the flow is not perfectly adiabatic, Tr V2 Te ro DyRT ” Te, where K is the temperature recovery factor. Usually K is essentially constant throughout the flight envelope and for a good flight test measurement system it will be in the range of 0.95 to 1.0. However, with sloppy de- sign or careless installation of the temperature probe, K may range as low as 0.7 and may be a function of airspeed. To summarize, ambient temperature is calculated from: 1+K(y-1): v2 459.67) - K(y- 1) 52 Too (b+ ATi # 459.87) -Ker- 1) 555 where tis in °F. Next, find the test day instrument corrected altitude and airspeed. Notice that these values are based on the average indicated gage readings during the run. If large varia- tions in either of these readings occur, the data will be unusable. hey Pigg * Ac 894 Vie, = Ving + Vic ‘Then, obtain Veg) from the average true airspeed and the ambient temperature. Cal- culate the density ratio, using the perfect gas law and the measured temperature from the second step above, and substitute in the expression for equivalent airspeed. Ve= V0 = Vogt AVe Figure 2.4 offers a convenient means of graphically iterating on this expression to obtain Voas Then, calculate AVpc) 8Nd Ahipe,ff0M AVpcy = Voay ~ Vice aNd Fig. 2.8. Finally, stan- dardize Aho, and AVpq, 10 nominal altitude conditions exactly as was done for the tower flyby test data.30. Pitot-Static System Calibrations______Chapter 2 The speed course method is one of the least accurate techniques for obtaining the static position error correction. Gracey, taking a different approach to the reduction of the data than shown above, says that it is valid for a very limited airspeed range (speeds safely above the stall speed up to approximately 130 knots at sea level or about M = 0.2) and offers no estimate of the accuracy or precision’®, On the other hand, the technique requires almost no instrumentation, apart from observers, a stopwatch or two, and an ac- curately known distance between two points on the ground. This simplicity does offer a simple way to calibrate the pitot-static system but questions about accuracy will linger if the flight test team does not take meticulous care in flying and recording the data. 2.2.3 Factors Affecting Position Error Measurements it.can be shown by dimensional analysis that the static pressure anywhere in the pres- sure field surrounding a body moving through the air is a function of at least five variables, ‘two dealing with the flow direction and three non-dimensional parameters: angle of attack (2), sidesiip angle (B), Mach number (M), Reynolds number (Re), and Prandt! number (Pn). if we neglect heat transfer effects, Pn is of no consequence; and if the static source is outside the boundary layer, Re effects will be negligible. Finally, we will assume that ssideslip can be minimized for those maneuvers for which we insist that pitot-static system measurements be reliable. Then, Ps -(M,0, Po (MO where f(x, y) is an unspecified function of two variables, with x = M and y= a. For vehicles operating in the linear angle of attack range, C, varies directly with a, ‘Therefore, with a slightly altered functional relationship, we may write a =f(M,C,). cp =-2L_. 20W____2nw____ aw “pv2s "(PZ \y2s (Srpov? © 8M21p 9S" ° (Phrlv2s (S100 fer) 0 ‘So, the functional expression becomes: (m2) (2.38) This simple expression clearly shows that there are three major variables affecting the sensed static pressure, p,, from the pitot-static system. The three variables ere Macht number, the weight supported by the lifting surfaces, and the altitude. To retum to our standard usage, the pressure difference is the prime variable of interest, so the functional relationship of eqn. (2.38) can also be written as: Mp _ {met} (2.39) If Mic is less than about 0.3, the flow field is essentially unaffected by compressibility ef- fects. [Some references even suggest using a much higher limit for Mach number'®.]Chapter 2, Pitot Static System Calibrations 34 Then the pressure coefficient of eqn. (2.39) is a function of lift coefficient only. It is also true that for M < 0.3, altitude effects are generally negligible. Then eqn. (2.39) becomes: oa) YT} wn WE (2.40) ‘At constant Vig, this equation suggests that Ap, is a function of angle of attack only (which must change to accommodate changes in either load factor or weight while main- taining level fight). For aircraft where the weight change as fuel is used or payload is changed is a large fraction ofthe total weight, there is a family of curves for AV. Figure 2.13e illustrates such a family of curves, On the other hand, if angle of attack changes are ‘small for the fight envelope, these curves collapse to a single curve as shown in Fig. 2.436. AVpc Voc Constant weight, Level flight 2 Vie Vic Fig. 2.13 Velocity Position Error Corrections for M < 0.3 ‘nW increasing However, Aligc is a function of altitude even if there are no angle of attack effects ‘The pressure difference is a function of altitude because the geometric altitude takes on a different value for every pressure level in the atmosphere. Figure 2.14 depicts the changes in Ahipg 88 a function of Vj, and altitude. In the transonic flight regime, especially near M,, = 1, it is very difficult to accurately measure static position errors. Certainly the complete functional, ean. (2.39), is neces- sary. In fact, at high speeds the importance of changes in angle of attack diminishes be- cause angle of attack changes litle is small in any case. Mach number then becomes the dominant term. Figure 2.15 illustrates a complete variation of pressure coefficient with both low speed effects and the discontinuities in slope that occur in the transonic region. ‘These discontinuities are the result of shock waves forming ahead of the static orifice. Figure 2.15 emphasizes the point that in this region the pressure coefficient is primarily & function of Mach number, Mach number is different for each Vig, 80 AVpe is function of altitude in this region. Figure 2.16 illustrates this variation. Plotted versus Vic, Apc 'e- sults in a family of curves displaced both vertically and horizontally for different altitudes, though if plotted versus Mj, the horizontal displacement is removed.2 sé itot-Static System Calibrations Captor 2 Abie h increasing Vie Pat) < Pat), 8), 03 04 OF 06 O07 08 O09 10 1.4 Mic Fig. 2.15 Pressure Coefficient Versus Mach Number Voc h increasing Vie Fig. 2.16 Airspeed Position Error Correction for Transonic Flight Conditions.Chapter 2, Pitot-Static System Calibrations 33, 2.3 SUMMARY In this chapter the subject of pitot-static system calibrations has been introduced. Starting with the mathematical model that produced the standard atmosphere, the theo- retical basis for these calibrations was laid. Experience has shown that only the static po- sition error correction is important for many flight test purposes, so three different methods of obtaining this information were discussed in detail. ‘Since such tests are rather tedious both to conduct and to reduce the data, no attempt is made in this introductory course to discuss ail the available methods of calibration. The reader who needs or desires detailed information on other methods of calculation will find ‘a wealth of material in the list of references. While the engineer who is responsible for ‘such calibrations is often accused of “measuring with a micrometer what is later to be cut with a hacksaw," careful pitot static calibrations are the foundation for accurate perform- ance and stability and control flight test data. The calibration process cannot be shorted without paying the price later.34. Pitot Static System Calibrations, Chapter 2 REFERENCES ‘Diehl, W. S., "Standard Atmosphere — Tables and Data," NACA Technical Report 218, 1925. 2 “Standard Atmosphere —Tables and Data for Altitudes to 65,800 Feet," NACA Report 1235, 1955. 3 U.S. Standard Atmosphere, 1962," U. S. Government Printing Office, Washington, 1962. 4 Minzner, R.A. Champion, K. S. W., and Pond, H. L., "The ARDC Model Atmos- phere," AF CRC-TR-59-267, 1959. 5 "U.S. Standard Atmosphere Supplements, 1966," U. S. Government Printing Office, Washington, 1966. ® Hamlin, B., Flight Testing Conventional and Jet-Propelled Airplanes, The Mac- millan Company, New York, 1946. 7 Anderson, J. D., Jr., Introduction to Flight, McGraw-Hill Book Company, New York, 1978. ‘Schoolfield, W. C., "A Simple Method of Applying the Compressibility Correction in the Determination of True Airspeed,” Journal of the Aeronautical Sciences, Vol. 9, Oct. 1942, pp. 457-464. ® Herrington, R. M., Shoemaker, P. E., Bartlett, E. P., and Dunlap, E. W., “Flight Test Engineering Handbook," AFFTC TR-6273 (AD 636392), Air Force Flight Test Center, Edwards AFB, California, May 1951 (Revised June 1964 and January 1966). 10 Gracey, W., "Measurement of Aircraft Speed and Altitude," NACA Reference Publica- tion 1046, May 1980. Huston, W. B., "Accuracy of Airspeed Measurements and Flight Calibration Proce- dures," NACA Report 919, 1948. Wildhack, W. A., "Pressure Drop in Tubing in Aircraft Instrument Installations,” NACA TN 593, 1937. "Performance," FTC-TIH-70-1001, Air Force Flight Test Center, Edwards AFB, Cali- fomia, 1970. Kline, R. E., Crews, A. H., and Knight, W. J., "F-5A Category II Performance Test,” FTC-TR-65-15, Air Force Flight Test Center, Edwards AFB, Califomia, July 1965.Chapter 3 CLIMB, DESCENT, AND TURN PERFORMANCE TESTS Every airplane on every flight; takes off, climbs, tums, descends, and lands. Thus, immediately after the pitot-static system is calibrated the test team can begin collecting performance data for these phases of flight. Since takeoff and landing performance is @ more difficult measurement task, climb, descent, and tum performance will be tackled first. ‘The climb performance of a vehicle is of utmost importance to the operator and is di- rectly related to lift, drag, and thrust capabilities of the aircraft. Generally, measurements ‘are made to determine either a speed or Mach number profile that will optimize a single performance parameter like minimum time to altitude, minimum fuel to altitude, minimum time to a total energy level, or such like. The actual altitude and velocity measurements: can be manipulated to describe the maneuver capability of the airplane or to evaluate the tactical capability of the vehicle relative to an adversary. ‘Turning performance is an important measure of merit for some classes of aircraft (like fighters or acrobatic craft) and is hardly ever used for others (like general aviation or commercial jets). It is more important for airplanes with an intended aggressive maneu- vering combat role than for any other group of airplanes. The measurements can be made in a number of ways and with relatively simple sensors (like a stopwatch and the normal cockpit instruments). Therefore, it is useful, at least in a teaming environment, to measure tum rates and radii even on airplanes for which tum performance is not used for certification or to qualify contractual guarantees. ‘This chapter is concemed then with performance measurements that are relatively straightforward: climbs, descents, and tums. All of them are strongly related to the ex- cess thrust or power available from the engine-sirframe combination. However, these categories are extremely important in assessing the utility of a design. 3.1 THEORETICAL FOUNDATIONS Table 3.1 FAR 23.65 Climb Requirements (All Engines Operating) Power Set- Configuration ‘Speed Minimum Steady Minimum Angle Climb Rate ‘Max Con- Max Wt< 6000 Ibs Most > 300 fpm at sea tinuous Flaps TO, Gear UP, vorable Cow fiaps COOL Takeoff Max Wt<6000Ibs MostFa- —_>300 fpm or No Require- Flaps TO, Gear UP, —vorable > 11.5V,, ment Cow flaps COOL Nota: W Vqy <1 Knots, no cb requrement is apectiod. Va, < 61 knots, use the requirement fora max rmumtweight. > 6000 ibs. Vg, Is the power off sia spesd, landing configuation. The power of stall speed, loan (cruise) I Vy ‘The climb performance required of a given aircraft may be spelled out in one of sev- eral documents. Tables 3.1 and 3.2 summarize the climb requirements stated in FAR 23.65 for small multiengine airplanes. For military airplanes, these requirements most of- ten come from the contractual specifications provided by the procurement agency. For ‘commercial aircraft, the buyer (an airline, for example) may have specific climb perform-36 Climb, Descent, and Tum Performance Tests, Chapter 3 ance requirements, but such unique requirements are not likely to be less stringent than those of the applicable Federal Aviation Regulations (FAR). Table 3.2 FAR 23.67 Climb Requirements (Critical Engine inoperative, Propeller in Minimum Drag Position) Weight/Performance Minimum Rate of Climb in Power Set- fpm at 5000 feet ting/Configuration Max Wt > 6000 Ibs RIC = 0.027V,,2 Maximum Continuous (Ve, in knots) (or less) Max Wt < 6000 Ibs RIC = 0.027V,,2 Gear UP, Flaps Vs, > 61 knots (Vz, in knots) OPTIMUM, Cow! Flaps as per FAR 23.1041-23,1047 Max Wt <6000 Ibs No minimum specified, ac- Vs, < 61 knots tual rate must be deter- mined Military requirements are similar to those listed in these tables although they are often more specific and more detailed. In any case, these examples suggest that the fight test team must verify or determine the optimum climb speed schedule as one of its first tasks for measuring the climb performance of the vehicle. Then, and only then, can either the minimum rate of climb or the minimum angle of climb be determined for a required con- figuration. However the requirement is stated, determination of climb (or descent) and tuming performance is based on the familiar lift-weight and thrust-drag force balances, which we will now review briefly. Fig. 3.1. Forces Acting on an Airplane in a ClimbChapter3__Glimb, Descent, and Tum Performance Tests a7 3.1.1 Governing Equations ‘A free-body diagram of the basic forces acting on an airplane in a climb is sketched in Fig, 3.1. ‘The forces perpendicular and parallel, respectively, to the fight path are: v2 L+ Tpsin(a. + @7)-Woosy = @1) g Re WV. dt a rans on notre The instantaneous velocity (or true airspeed) of the airplane's center of gravity is re- lated to the time rate of change of the flight path angle ¥, which is obtained by: sal) -Seta At At dt ° Veo Or, we can write: 7= pe e Thus, eqn. 3.1 can be rewritten as Ty0os (2 +7) -D- Weiny =a (3.2) L-Woosy +7,sin (a+ @7)=1 Vi (3.3) If fight path curvature is slight (Re->) and a + @r= 0, eqns. 3.1 and 3.3 yield: L=WoosY @4) Eqn. 3.2 becomes: Ty - -0- Was =W sin ‘The angle of climb is then ees from siny= nee -i% (3.5) Recaling thet rate of climb is S = Vin, we substitute to obtain: (6) 3.1.1.1 Steady State Approximation. Equations 3.4, 3.5, and 3.6 are the governing ‘equations for climb (or descent) performance. To further simplify the equation, it is often assumed that there is no acceleration along the flight path. Then, Me Oand @.7) (3.8)38__ iim, Descent, and Tum Performance Tests Chapter 3 Equations 3.7 and 3.8 are very common approximations for low performance aircraft. If indicated airspeed is held constant and climb rate is low, the rate of change of true air- speed with altitude (and time) is small and this approximation is adequate. Or, if true air- speed is changing rapidly during the climb, rate of climb can be rewritten as dh _(Tn~DNe Vo We dh dt Wg dh at (Sh) (3) Seca eo Vio Ve where AF = acceleration factor = =. <= 3.1.1.2 Accelerated Climb Equations. More generally, the performance equations can be constructed by assuming only that the angle a + ®; is small and allowing both y and V,, to be functions of time. Then the climb performance equations can be expressed concisely as a set of first-order nonlinear differential equations. Tyla 14 Iolo+or) cosy j Ted s-(i- 9 7 -siny A Ven sin'y ‘These equations can be extended to include other time-dependent variables such as R=V,,cosy or W= wy where R= horizontal distance covered or ground range and w= fuel flow rate This set of first order differential equations suggests the use of state variables, which ‘opens the door for use of more powerful mathematical tools (like the calculus of variations) to optimize performance parameters. But our purposes are served in this chapter by sim- ply using this formulation to lead into a discussion of Rutowski's energy approximation’. 3.1.2 The Eneray Approximation If the airplane comprises a dynamical system having a total energy made up of the ‘sum of its kinetic energy and its potential energy, a slightly different, yet equivalent, set of climb performance equations can be written. Then, total energy is given by e=Lmv2 +mgh To more easily compare aircraft of different masses, it is customary divide by mg and re- write total energy in terms of specific energy = with units of distance or altitude:Ctimb, Descent, and Turn Performance Tests 39 (3.10) specific energy or energy height The time rate of change of specific energy, or specific excess power Pa, is Pane =f +VnY® (3.11) Recalling the V.. equation from the state variable set in Section 3.1.1.2, Mutipying by Ve, wo get: VaNln =Vao (=> -siny) But, h=Vey siny , 80 = D)Vey , Vo Co Plo = +g XE =P (3.12) Equation 3.12 merely confirms our earlier assertion that the energy approach to obtain specific excess power is equivalent to the force balance state equation relating Vi. to 7, D, W, and 7. Several books?.3 elaborate on the utility of the energy approach, but it suf- fices for our purposes to simply point out that the energy definitions allow computation of climb performance from thrust and drag measurements under accelerated flight conditions, not just constant true airspeed maneuvers. This simple result merely scratches the sur- face in demonstrating the uses of the energy approximation. Rutowski's approach leads naturally to performance optimization of not only climb and descent performance, but iso ‘of maneuvering fight. As an example of such maneuvers, consider a level tum. 3.1.3 Forces in a Level, Unaccelerated Tum Fig. 3.2 Forces ina Level Turn Consider first the idealized force and moment diagram (Fig. 3.2). It can be inferred from the diagram that the forces are in equilibrium (as long as you accept the nation of40 Climb, Descent, and Tum Performance Tests, Chapter 3 "centrifugal" forces being produced by @ constant acceleration). This idealization really describes a constant airspeed, constant altitude, and constant bank angle tum - often simply called a stabilized tum or a level, unaccelerated tum. As we will see from the sim- ple force equilibrium, the load factor in such a tum is constant also. ‘The force equations are almost trivial, but they do show the important variables we need to measure. Loos¢-W=0 or Leos}=W, so 1 W ~ cosé (3.13) In this expression we have used the definition of load factor n to show why a constant bank angle in a steady, level tum implies that n is a constant. Quite obviously, bank angle is a parameter we are likely to want to measure and that can be done by recording the load factor if the test airplane is equipped with an accelerometer (g-meter) or with an accu- rate vertical gyro. We will also see later that other parameters can be obtained from such ‘simple relationships and that even simpler instrumentation will suffice. Another equation springs from the horizontal force "balance" -- equating the compo- nent of lift acting in a horizontal direction to the “centrifugal force”: 2 ting = af =) (3.14) where ris the radius of the tum (as suggested in Fig. 3.2) Vis the true airspeed (constant) Using eqn. 3.13 with lft written in coefficient form and solving for turn radius: W, ) re af 4s (3.15) © eC, sing Equation 3.15 shows clearly that the tum radius: (1) varies directly with wing loading (WIS), (2) inversely with altitude (that is, increasing altitude increases r), and (3) inversely with lift coefficient. Minimum tum radius will occur, therefore, at low gross weights, low altitude, and at maximum lift coefficient. Other expressions that are useful in fight test are related to the previously developed expressions. For example, we can take eqn. 3.14 again and rearrange it to express load factor in terms of the other parameters. Knowing that me and using basic trigono- 2 metric definitions, we notice that sing = gives: - Making this substitution and solving for r (3.18)Chepter3__Climb, Descent, and Tum Performance Tests a Similarly, starting with the same expressions and recognizing that the tum rate is sim- ply the angular velocity associated with the level tum and is therefore defined by Vir, we can easily repeat the algebra and obtain an expression for the tum rate: V_gvn?-1 jo=r=— y_] 8.17) As suggested earlier, eqns. 3.16 and 3.17 demonstrate that calculating the important tum performance parameters like radius of tum and rate of tum, requires only a stop watch to time a specified heading change (usually 360°) and the pitot-static measurements that give true airspeed. An accelerometer could also be used, but it is not absolutely nec- essary. Having now developed some simple (but quite effective) ‘equations to calculate tuming performance, let us now tum our attention to the kinds of tuning performance that are of interest in our introduction to flight tests. ‘Structural Limit Load Factor -n Dynamic Pressure Limit ‘Structural Limit Fig. 3.3. V-n Diagram 3.1.3.1 Types of Tuming Performance. The equations in the preceding section need to be related to operational considerations and modified to provide better indications of the actual capability of a given aircraft. The unaccelerated level tum is closely related to sustained tuming performance, but instantaneous tuming is often of even greater inter- ‘est for combat maneuvering and is more closely related to specific excess power (espe- cially when the aircraft is limited by thrust). The V-n diagram (Fig. 3.3) contains many of the elements important to both sustained and instantaneous tum performance. However, as we show later, there are more informative ways to present turing performance results. Sustained turn performance is usually described in terms of energy state; it is a measure of the excess power of the airframe-engine combination, especially at load fac- tors greater than one, The V-n diagram fimits clearly affect and constrain the sustained tum performance since they describe the aerodynamic and the structural limits for the air- frame. The stall characteristics of the airfoil, the planform, and other configurational fea- tures of the airplane prescribe lift limits. The maneuvering loads that the airframe can withstand typically set the upper and lower structural limits. In today's modem fighters,a2 Climb, Descent, and Tum Performance Tests, Chapter 3 these limits are usually expressed in g's and for these load limits are usually +9g and -3g, accelerations which are now approaching the physiological limits of the human pilot. Said another way, the tuming performance of modem fighter airplanes is often not an airframe limit; instead, the pilot is may be unable to tolerate the achievable accelerations. G-Loss- of-Consciousness (G-LOC) is an example of such a physiological limit. On the other hand, instantaneous turning performance ignores any change in air- speed that may occur during the maneuver. It is not necessarily a steady state maneuver, It is important in combat when the aircraft has weapons that can be employed with the nose pointing (perhaps only momentarily) at the adversary. It is particularly important in either getting off a first shot or in defeating an enemy's tracking solution. Instantaneous tuming performance is important any time it is advisable to sacrifice energy to gain or re- gain tactical advantage. Again, the limits shown on the V-n diagram are useful in showing important characteristics of the tum performance, but there are more meaningful plots. The point where maximum lift coefficient and maximum load factor coincide is the maneu- ver point . The true airspeed associated with that point is called the comer velocity; at this speed the airplane can tum with the smallest radius and the highest tun rate, 3.1.3.2 Limitations on Turning Performance. Though we have alluded to tuming performance limitations already, we have not clearly spelied out the types of limitations that are associated with the airframe and its powerplant. They can be classified and dis- cussed under three general headings for any airplane. 3.13.21 Thrust-Limited (or Power-Limited) Turing Performance. To tum at constant altitude the airplane's lifting surfaces must produce lift over and above that re- quired to maintain level fight with the wings level (Fig. 3.2). However, if additional lift must be generated, that lift inevitably produces additional induced drag. This additional drag must be balanced with additional thrust to maintain constant airspeed as assumed at the outset of this discussion. Of course, in many aircraft this additional drag grows to a point that available thrust from the powerplant cannot overcome the additional drag and the air- craft will decelerate in the tum. The airframe and powerplant are thrust-limited in the tum. Thrust Required Fig. 3.4 Thrust Required at Various Bank Angles Naturally, as bank angle increases, the thrust required increases rapidly (Fig. 3.4). Of course, there are many fighter aircraft today with very high thrust-to-weight ratios. Rarely do they encounter thrust limitations during tums, which suggests we must consider theChapter 3, Cilmb, Descent, and Turn Performance Tests “3 limits on the left side of Fig. 3.4 that are based on the maximum lift the wings of the aircraft can develop. Hf you are conducting tuning performance tests in a propeller-driven airplane, it will be necessary to consider how changes in induced drag change the power required curves rather than thrust required curves as shown in Fig. 3.4. However, power required is sim- Thrust Required / Available ‘Airspeed or Mach Number Fig. 3.6 Parameters That Affect Thrust- The thrust-limited tum is usually the primary focus of tum performance. This type of performance is closely tied to the energy concepts discussed earlier. In fact the same test technique (the fevel acceleration) provides data for turing performance and is used to determine optimum climb performance. This fact has led us to discuss tuming perform- ance in the same chapter as climb tests. Since P, is relevant to both types of perform- ance, we can readily see that thrust-limited tuming performance is affected by exactly the same factors that affect climb performance (Fig. 3.5); anything that affects excess thrust changes tuming performance. Corrections to tuming performance are based on similar considerations: (1) Temperature effects on thrust (3) Weight effects on induced drag (2) Pressure attitude effects on thrust (4) Pressure altitude effects on drag 3.1.3.2.2 Lift-Limited Tuming Performance. The dashed line in Fig. 3.4 indicates the maximum lift coefficient the wing can generate; that is, the lift limit is indicated by this line, simply because the wing will stall if it is exceeded. Stall speed in a tum is directly proportional to the load factor and the stall speed in level flight; that is, Vs, =nV,, where Vey is the stall speed in an unaccelerated turn at @ constant bank angle. This result sug- imited Tuming Performance gests that the dashed line in Fig. 3.4 increases as oF44 Climb, Descent, and Turn Performance Tests, Chapter 3 This lit boundary for turing performance is a part of the envelope that is of particular interest in fighter and trainer aircraft. Sometimes specially designed tests are laid out to explore the lift boundary. However, the interest then is primarily one of controllability. For example, what levels of buffet are to be expected? Will they adversely affect the ability of the pilot to operationally maneuver and track an adversary for weapons delivery? Is there a strong likelihood that the airplane will go out of control with litte or no waming? Though the lift limit does bound the performance as suggested in Figs. 3.3 and 3.4, these consid- erations are properly discussed under handling qualities testing. For that reason we will not dwell on them in this section. 3.1.3.2.3 Load-Limited (Structural) Turing Performance. The loads that the air- frame is designed to carry are statically tested on the ground and monitored during early envelope clearance fights. These load limits are normally expressed in load factor (both positive and negative) as suggested in Fig. 3.3. These limits are typically specified in the contract and/or by the certification agency. Federal Aviation Regulations in the United States spell out limits of +4.4/-2.2 g for utility category aircraft certified under FAR Part 23. These load limits are seldom achieved in fight and there is always a margin of safety ap- plied by the designer to allow for unexpected turbulence in the atmosphere or inadvertent overstressing of the aircraft. LEVEL, UNACCELERATED TURNING PERFORMANCE TURN RADIUS TRUE AIRSPEED Fig. 3.8 Summary of Tuming Performance Limitations Figure 3.6 is yet another way to show and summarize possible limitations on turing . At speeds less than the maneuver speed shown in this sketch, even aircraft that are not thrust-Jimited cannot reach the limit load factor. They are aerodynamically (lift)Chapter 3 45 limited, At higher speeds the vehicle may be limited by either the strength of the structure or by the available excess thrust. For example, most general aviation airplanes are thrust- limited throughout much of their envelope. Climb, Descent, and Tum Performance Tests 3.1.4 Experimental Versus Analytical Performance Methods Performance measurements have little meaning unless the data are reduced to some ‘common basis of comparison. This common basis is the standard atmosphere. Hence, performance reduction refers to the data reduction schemes by which test data taken un- der nonstandard day conditions are converted to standard day results. As Lush puts it, reduction methods are necessary to rationalize the test results so that they may be used for valid prediction and comparison.” ‘As the title of this section suggests, there are two broad categories of performance data reduction. Experimental methods are so named because they require no advance knowledge of the components of the dynamical system. For an airplane, for example, no knowledge of the power plant characteristics would be required under this method. Ane- tytical methods do require a priori sources to complete the analysis and are somewhat more widely used. Analytical methods are further broken down into differential methods ‘and performance analyses. Differential methods are based on the notion, that for small ‘corrections, linearization is appropriate. These methods depend on a generalized char- acterization of airframe drag and engine behavior. The differential technique is not useful, for example, when compressibility effects are significant. Performance analyses rely on ‘advance information of engine behavior and are generally used when experimental meth- ‘ods are inconvenient or simply impractical. ‘One of the most perplexing problems in performance reduction is the number and type of variables to be considered, some of which are controllable while others are not. Engine parameters are, for example, usually quite controllable over the range of interest but out- ‘side air temperature is not. Tests are ordinarily planned to cover a suitable range of the controllable variables, while those variables that cannot be controlled are “standardized”. Choice of a standard weight, like 95% of the takeoff weight, is typical of the latter case. 3.1.4.1 Experimental Methods. Dimensional analysis is a tool that allows us to re- duce the number of variables by grouping them. Furthermore, if uncontrolled variables ‘can be associated with controlled variables, experimental methods may be used to deduce standard day performance from nonstandard day test data. One of the simplest illustra- tions of this technique is in reducing performance data for turbojet airplanes. We will briefly introduce the subject for now and illustrate it in greater detail in Chapter 4. The success of dimensional analysis rests upon the assumption that all pertinent variables can be conceptually identified, though the mathematical form of the relationships may remain unknown. We could assume, for instance, that the climb rate of a turbojet-powered sir- plane depends explicitly on true airspeed, engine compressor speed, aircraft weight, out- side sir temperature, and pressure, if such a function exists, it can be represented by: fy(Vex N, W. Tay, Pao) = 0. (Notice that .. has been used with T. to differentiate it from 7 for thrust, as used previously. Wee will stick to that convention for the rest of the book.) If Buckingham's Pi Theorem Is applied (we will retum to this theorem later), the variables8 Climb, Descent, and Turn Performance Tests, Chapter 3 Va ND _W. I - can be reduced to three: {ee}. in the first expression there are five in. dependent variables and two of them, W and T.,, are not easy to control or adjust during flight tests. Dimensional analysis reduces the number of independent variables by two. This reduction of the number of independent variables by two is generally the level of sim- plication realized by application of dimensional analysis. Pragmatically, reducing the number of independent variables by only two often makes it difficult to use experimental methods. Usually, there are still too many variables af- fecting the performance reduction. However, if one is able to assume that only a few of the variables affect performance strongly, experimental methods can be used. A good ex- ample of such a simplification is the level fight performance of a jet aircraft (discussed in detail in Chapter 4). If the constant test altitude is reasonably close to standard condi- tions, it is usually acceptable to omit viscosity, for example, as an independent variable in determining the level speed of the airplane. 3.1.4.2 Analytical Methods. Of the two “analytical” methods, the differential method is the easiest to use when the background data are available. This type of performance data reduction is with conventional reciprocating engine powered aircraft. The database for reciprocating engines is rather extensive and most of this class of airplanes do not ap- proach any compressibility effects, making the differential method both convenient and ac- curate to use. The a priori knowledge necessary to use this method does not have to be extremely precise because the differential corrections for nonstandard conditions are usu- ally relatively small. Turboprop-powered aircraft are generally analyzed using this ap- proach, though the jet thrust introduces more variables and reduces the precision of the corrections. The second analytical method, performance analysis, is based on determi- nation of drag and/or power curves (both required and available) over the altitude and air- speed envelope of the airplane. Again, one needs a priori information on the output of the propulsion system and propeller efficiencies, along with the effects of changes in temp- erature and forward speed on these parameters. Choosing the appropriate performance reduction method is a difficult task. The most important considerations involved are the amount of data required (number and character of variables to be measured), the size of the flight envelope (altitude, Mach number, air- speed, load factors) to be covered, and the availability and nature of generalized data on the airplane and its power plant. In virtually every instance, the choice is a compromise {and it is often a combination of the various methods. Now, let us tum our attention from these general considerations to the specifics of climb performance reduction. 3.2 CLIMB PERFORMANCE TEST METHODS Climb performance tests fall into two broad categories: those designed to determine ‘an optimum speed or Mach number schedule or those designed to measure the actual performance of the vehicle. For the first category of tests, the optimization may be carried out with any of the following purposes: ‘© minimum time to altitude ‘@ minimum fuel to energy level © minimum time to energy level ‘@ maximum climb angle ‘© minimum fuel to altitudeChapter 3. Cilmb, Descent, and Tum Performance Tests. a7 ‘The second category of climb tests are sometimes called "check" climbs or "performance" climbs since they are meant to verify that a chosen profile meets the specified optimization objectives. This validation should also ensure that cooling considerations, lack of forward visibility, poor handling qualities, or some other operational consideration does not dictate deviation from the optimized climb schedule. Both types of these tests will be introduced in this section, along with some of the techniques used. The reduction of climb data from such tests to standard day conditions will also be described briefly. 3.2.1 Climb Schedule Determination Optimization of a climb schedule is based on the airplane's lif, drag, and thrust rela- tionships (reviewed in Section 3.1) over an appropriate speed range. The measurement may be either direct or, more often, indirect. The two most common techniques, the sew- tooth oF partial climb test and the level acceleration test, will be discussed in this intro- ductory text. Of these two test methods, the sawtooth climb is easier to understand and requires only the most rudimentary instrumentation. However, it is only useful for low climb rates. The level acceleration test provides more data for the same investment of flying time (data that are useful for more than just determining a climb schedule, too) and is applicable to high performance sircraft with high climb rates. It does require some means of automatically recording altitude and airspeed as a continuous function of time. We consider sawtooth climbs first. Nominal. test. altitude ui Start times. ooh note IAS ceaencend ‘Set power; recheck configuration, trim at IAS; fly 90° to wind; stabilize 200300 feet BelOW hentry: start recorder, if available Fig. 3.7 Sawtooth Climb Procedure 3.2.1.1 Sawtooth Climb Method. The sawtooth climb, so named because of the ‘shape of the altitude trace it produces, is used to determine the airspeed for maximum rate of climb at a nominal pressure altitude under test day conditions. A series of timed climbs is made over an altitude band bracketing the nominal pressure altitude chosen. Figure 3.7 ilustrates the data collection. Increments of altitude used for each airspeed point should be chosen so that timing can be carried out precisely; if a stopwatch is manually started ‘and stopped es suggested in Fig. 3.7, the altitude increments should be chosen so that the elapsed time is at least 1 minute. Indicated airspeed or Mach number must be main-ry Cilmb, Descent, and Turn Performance Tests, Chapter 3 tained very precisely (+1 knot is desirable) to obtain valid results without excessive scatter. The landing gear, flaps, cow! flaps (in other words, the configuration) as well as the de- sired power setting must be set before commencing the data run. Trim should only be adjusted between runs. The pilot should be briefed to make data runs so that the heading is approximately 90° to the wind for the test altitude and he should attempt to conduct each climb in approximately the same air mass. Enough different constant airspeeds ‘should be flown to define the shape of the rate of climb curve. Typical time to climb data are shown in Fig. 3.8 (open circles). For each nominal altitude, a minimum and a maxi- mum speed point (using the same configuration and power setting) should be flown first to anchor the ends of the curve. A running plot of the observed time to climb through the alti- tude band, as shown by the open circles in Fig. 3.8, often gives clues as to which of the data points should be repeated to better define the maximum rate of climb speed. ‘© Measured rate of climb =t- Rate of climb corrected for weight bh 7667 12 3 "Test poin Fig. 3.8 Weight Correction of Sawtooth Climb Data Dommasch® suggests that no corrections are necessary for nonstandard weight or for Power changes due to nonstandard atmospheric conditions if sawtooth climb data are Used solely to determine the maximum rate of climb speed schedule. It is current practice to correct for both these conditions®.5 and those due to wind gradients. Both these ref- erences provide detailed discussion of the data reduction techniques, but for our intro- ductory purposes we will describe only the simplest two of the possible ways to correct for changes in gross weight. Ways to minimize the effects of wind shear will also be briefly discussed. The simplest way to correct sawtooth climb data for differences in gross weight during a set of climbs is illustrated in Fig. 3.8. it amounts to prorating the change in rate of climb over the time interval and requires that a second climb be repeated at a selected airspeed. This repeated point should be flown near the end of the series to provide a comparison with @ point flown at @ heavier weight earlier in the series. Points 3 and 9 in Fig. 3.8 illus- trate this repetition. It is assumed that rate of climb varies linearly with time due to the change in gross weight. The solid triangles in Fig. 3.8 illustrate how the measured times to climb (or rates of climb) might change with this form of correction. Alternatively, if the ‘weight of the airplane is known for each data point, the rate of climb may be corrected by a simple ratio of weights. e ts Wy h * hy 'omected * 7,Chapter 3, Ciimb, Descent, and Tum Performance Tests “9 Both of these approximations produce roughly the same resuits and both of them ignore fan induced drag correction term. However, they are easy to use and provide resutts that are little different from more elaborate schemes. Nonstandard temperature conditions may also require corrections to the power output of the propulsion system. If 80, the effects of atmospheric variations on engine perfor- mance must be known in advance and outside air temperature measurements become necessary. The Empire Test Pilots' School (ETPS) Handbook’ gives a detailed account of British practice in making such corrections for reciprocating engine aircraft. Since these types of corrections depend entirely on the engine data available and are totally unique to each installation, we will discuss them no further. wa Vow Fig. 3.9 Effect of Vertical Wind Shear on Sensed Rate of Climb Finally, sawtooth climb data may be affected by both changes in wind speed and di- rection. Shifts in wind direction are usually of little consequence since the altitude band traversed is usually smell and directional variations affect only the measured pressures. where AVs = Viv, g 47 Vw, Ifthe vertical wind gradient “Y=. is positive, rate of climb increases and, if the vertical wind gradient is negative, rate of climb decreases. But this knowledge is not easy to use. Determining wind gradients is very difficult and since they constantly change, correcting for them mathematically is impossible. The best way to handle wind effects on sawtooth climb data is by flying data runs close to perpendicular to the wind direction. One other possible technique is to repeat each data point (each Vj.) on reciprocal headings. Unfor- tunately, this approach also doubles test time required to complete determination of the climb schedule. In short, it is difficult to correct sawtooth climb data for vertical wind shears; shears problem should be avoided with good flight planning if possible. Example 3.1: The following data were collected in flight on a twin-engine STOL transport during a series of sawtooth climb tests. The nominal (midpoint) test pressure altitude was 8,750 feet. Assume that the first ‘ix points were flown at constant weight and that the last six points were flown at a different (lighter) con- ‘tant weight. Betwoen the two sets of sawtooth points, a maximum speed of 120 KIAS was obtained at the limb power setting used. Plot the two curves thus obtained and find the average curve between the two50 Climb, Descent, and Tum Performance Tests, Chapter 3 ‘ets of faired data, Based on this average curve, what is the best rate of climb speed (KIAS) for this plane at 8750 feet pressure altitude? 20; Vj At(eec) ah Rate of Climb KAS (rect) __(fsec 71870: 900—«15.79 Bi 73 55.0 900 16.36 if e595) 0015.43 6 65.5 900 13.74 Er 109 139.0 900 6.47 3 10s 1000 0 90 §e 5 630 0014.23 cy 51.0 900 17.65 7% 485 «0018.56 te) arey ‘6 indtsted Arspoed —"erats 1”? ‘The plot on the right above shows the resulting sawtooth curves with the peak rate of climb occurring at ‘spproximately 77 KIAS. Thus, these sawtooth tests suggest that, for maximum rate of climb, an indicated airspeed of 77 KIAS should be maintained as the airplane passes through 8,750 feet. 3.2.1.2 Level Acceleration Method. Equation 3.6 shows that by maintaining a con- ‘stant altitude and recording the change in true airspeed with time one can measure indi- rectly the excess power available from any airframe-powerplant combination. Further- more, this measurement is easy enough to make even in high performance airplanes with a suitable means of recording airspeed variation with time. Motion picture fim, video, magnetic tape, or telemetered signals can provide such time histories readily. Further- more, acceleration data can also be used to define the maneuvering capability of the de- sign, especially if the purchaser is interested in the energy maneuverability. Finally, the acceleration performance itself is sometimes used as a measure of merit in the contractual specifications. ‘The level acceleration technique is a simple concept. The airplane is merely accel- erated, with a fixed power setting and configuration, from near its minimum level fight ‘speed to its maximum level flight speed. The piloting task is simply to maintain constant pressure altitude. Of course, there are complications. First, the pilot does not usually have a pressure altitude indicator and altitude measurements must be corrected for pitot- static errors. Since the pilot must keep these corrections as small as possible, knowledge of static pressure error as a function of indicated airspeed is used to plan data runs. Be- cause the altimeter has hysteresis error, reversals in altimeter readings should be avoided. Therefore, an acceleration or deceleration run should be planned to hold a slight climb or descent (100-200 fpm maximum), whichever minimizes the altitude error after the static pressure error correction is applied. The flight test team must take care to obtain ground block readings with the altimeter set at 29.92 (remember ail test data taken at @ pressure altitude must be taken with this altimeter setting). Typically, the airplane is trimmed at ‘some midpoint speed for the range of speeds to be covered and the trim setting is not changed. The pilot must exercise care in passing through this point to transition from back pressure to forward pressure smoothly and not allow an altimeter reversal. If the airplane is capable of a wide speed range, it is possible to piece together several different ac- celeration runs. If such a procedure is used, the test team may have to account for smallChapter 3, Climb, Descent, and Tum Performance Tests 51 changes in drag due changes in trim. A particularly sensitive part of the acceleration run ‘occurs at speeds near Mach 1 when the pitot-static system is affected by shock waves. Ordinarily it is best to simply maintain the same rate of change of attitude (by using either visual references or guidance system commands derived from the airplane's inertial refer- ence system) until the pitot-static system settles down in supersonic flight. Maximum rate of climb \ Maximum rate of change of h, Fig. 3.10 Uses of Ps Contours Level acceleration data can be used to optimize climb schedules in any one of the several ways suggested at the beginning of this section (Fig. 3.10 Illustrates). For the purposes of these introductory notes, we will only discuss the reduction of level ac- celeration data for either the maximum rate of climb schedule or the minimum time to an energy level. For either of these purposes, specific excess power Pa is the preferred mea- sure of merit. As Fig. 3.10 shows, the maximum rate of climb can be estimated from con- stant altitude (horizontal) tangent lines to the various P, contours and the maximum change in energy level can be estimated from contours of constant total energy level (the parabolic tangents in Fig. 3.10). Obviously, the two climb schedules are different, even for ‘a subsonic airplane as is depicted in the sketch. This abbreviated discussion merely hints at the usefulness of specific excess power contour maps in terms of altitude and velocity parameters. ‘With regard to standard day corrections, P, data from level accelerations can be treated as rates of climb. To be complete, the data reduction process should include cor- rections for nonstandard temperature effect on the thrust output of the propulsion system, both inertia and induced drag weight corrections, and wind gradient corrections. In most cases not all of these corrections will be significant and the test team must choose the ‘ones to be made. In this section we will discuss the weight correction only, leaving the nonstandard temperature correction discussion to the check climb section. The wind gra- cient correction ie identical that of the sawtooth climb technique. The only means of taking such gradients into account is through careful fight planning and execution.Climb, Descent, and Tum Performance Tests, Chapter 3 In considering the weight correction for sawtooth climbs, no mathematically rigorous analysis was used. Another approach is to express test day thrust as a function of the other variables and then see how weight affects them. As a part of this analysis, it will be convenient to refer all calculations to a standard weight for ease of comparison. The gen- eral expression for P, on a given test day is (Ta =2e) Mor y WM Rearranging to express test day thrust as: Ww, Tuy Pe +0; The effects of weight on the acceleration due to thrust are given by differentiating with re- ‘spect to weight. Ty We PPoy | Poy GWM, ~ Ving OM * Vay we aw, Typically, thrust is affected little by changes in weight at constant power setting (usually military or maximum) as the airplane accelerates in level fight. Angle of attack also has little effect on thrust available for acceleration, 80 (3.18) Defining the weight difference at any instant in the maneuver as AW = Werg - Wi and expressing all differentials in eqn. 3.18 as differences PgAW , AD Mey 10a.) ar, »-(Sa tm (3.19) Equation 3.19 assumes that the only change in drag is because the weight is not the nominal weight, which leads to changes in induced drag. This simplified differential cor- ection method yields two primary correction terms, the first due to inertia of the vehicle and the second due to induced drag changes between test weight and standard weight. The induced drag term can be further expanded by writing D; as: cos? y eM Then AD; becomes wr of Be (COP ))- Geeta) aad where kis the ratio of specific heats (typically, 1.4) and 7 is the fight path angle. =O(nw? where C; = 2 kxPoChapter 3, Cilmb, Descent, and Tum Performance Tests 53 The P, correction for nonstandard weights then follows: Vo = nggW3q Ot} 2b rst te le (3.20) Since Neg = m= 1 and cos ¥ = 1 for the level accelerations, eqn. 3.20 can be written as AW 5 AW | —S74—|| w? -w, #) (3.2 aa (Grane OM 5 com With regard to standard day corrections, P, data from level accelerations can be treated as rates of climb. To be complete, the data reduction process should include cor- rections for nonstandard temperature effect on the thrust output of the propulsion system, both inertia and induced drag weight corrections, and wind gradient corrections. Usually, not all these corrections will be significant and the test team must choose the ones to be mede. In this section we will discuss the weight correction only, leaving the nonstandard temperature correction to the check climb section. The wind gradient correction is identi- cal as discussed under the sawtooth climb technique and the only satisfactory means of taking such gradients into account is careful fight planning and execution. Typical level acceleration data reductions include the following: 1. Correct the indicated gage readings for instrument corrections for each indi- vidual gage and each data point to be reduced; that is, for every V;, fi, and Trop calculate the corresponding Vie hs and Tix, Then, apply the appropri- ate position error corrections to obtain calibrated values of each parameter. Fig. 3.11. Determining P, Graphically 2. Calculate the test day true airspeed from calibrated airspeeds for each data point. Plot these points versus time as shown in Fig. 3.11 or arrange them ina time history array for step 5 in this process. 3, Calculate test day P, from Ps = fg= hotel and plot these points on the same graph with the true airspeed time history as suggested in Fig. 3.11. ‘Alternatively, arrange the he points in a time history array similar to the true ‘airspeed array so that energy height can be numerically differentiated.4 Climb, Descent, and Turn Performance Tests, Chapter 3 4. Estimate P, for test day conditions by numerically differentiating the energy height time history to obtain an approximate rate of change of he for each instant of time and the corresponding true airspeed. Again, this process is easiest to visualize as a graphical process, though the actual work can be carried out by manipulating the appropriate data arrays in a computer rela- tively easily. 5. Correct P,, to standard day and standard weight conditions as suggested previously for rate of climb data. Vou OF Mux Fig. 3.12 Specific Excess Power versus Airspeed 6. Plot P, versus V., (or M,.) for several altitudes. Notice that the resulting curves, as sketched in Fig. 3.12, are very similar in appearance to sawtooth climb curves for low performance airplanes. However, the shape of P, curves for supersonic airplanes is quite different. 7. Crossplot P, as a function of pressure altitude and velocity by taking lines of constant P, across the constant altitude curves of Fig. 3.12. The resulting crossplot of constant P, curves as a function of altitude and true airspeed, sketched previously in Fig. 3.5, is perhaps the most common way to depict, level acceleration data. Recall that this format allows the determination of both the maximum rate of climb schedule and the minimum time to energy height climb schedule. This procedure is by no means the only way to get P, contours from level acceleration data; for example, you may choose to plot both test day true altitude and test day true air- ‘speed versus time and numerically differentiate both quantities. Using these estimates for a and we in eqn. 3.11 gives P, for the test day also. Whichever of the measured quantities is numerically differentiated, the test engineer should be careful to smooth the data and remove fluctuations with a haif-cycle less than 20 to 30 seconds. Potential to ki- netic energy exchanges simply do not occur more rapidly than this rate and higher fre- quency disturbances are usually noise in the data. Level accelerations retum a considerable amount of useful data in a relatively short ‘amount of test time. It is a very efficient flight test technique. Optimized climb schedules for both subsonic and supersonic airplanes can be extracted using this approach. The ac- celeration characteristics of a given airframe-powerplant combination are direct mea- surements. Finally, the maneuvering capability of the machine can be derived from level accelerations conducted at different load factors. This introductory text will not go into this latter technique; interested students may refer to references 1,3, and 5 for details of thisChapter 3, Climb, Descent, and Tum Performance Tests 55 extension of the level acceleration method. Data collected from level accelerations in high performance airplanes are usually more consistent and repeatable than data collected with the sawtooth climbs, both because of the use of the acceleration term and because verti- cal wind shear has little effect on level fight data. However, data handling is slightly more complex than itis for sawtooth climbs. Data must be recorded automatically and synchro- ized with time; so slightly more sophisticated instrumentation is required. Data reduction itself is rather more tedious and begs for automated processing. Neither of these objec tions is particularly troublesome with modem instrumentation and computer capabilities. Example 3.2: Lovel acceleration data for a fictitious smal jet trainer are tabuleted below. These data ‘were collecied at a pressure atitide of 20,000 feet and they have been partially reduced (that is, stan- dard day true airspeed and pressure alitute have eireedy been calculated from the test messure- tmerts). Calculate and plot the P, curve for this eirplane at 20,000 feet. Based on this curve, what is the true airspeed for best rate of climb? é bh av Fi Band V from wand a ist, wo estimate v ape hand Fe 199707 109700 « 6p ont V = V. For example, the first two lines in the table give: f= Time TAS Ate fh V, 7 (86.1 + 86.8(3.5) Oe ee ae a0 868 120707 a! 30 080 882 199730 7:00 350 16587 Pe =12.004pe| 420 801 199780 2000 260 26982 » 160 915 190850 1500 450 27.735 200 923 199910 1500 450 27.006 240 9568 199980 1000 650 20182 S40 280 982 200000 1000 650 28.708 320 1009 200037 1350 5.50 30654 360 1039 200070 10.00 7.00 32453, 20) 400 1078 200096 450 1000 37.632 4M 4148 200117 550 1050 41578 go 4980 1156 200149 450 11.00 43.648 $20 1204 200167 3.00 1250 49.201 6 cd ao) 140 560 1264 200129 -1600 1600 47.062 True Aispeed ~ featisecond 600 1325 2007.1 -13.00 1450 46.061 We could have found the slopes of the €40 1376 200028 -11.00 12.50 41.974 curves with elaborate numerical schemes, 6.80 1421 199980 -10.00 10.00 33.855 but this straightJine average is adequate with 7.20 1447 1967.0 5.00 450 25.175 closely spaced data points. The plot above, 760 1461 190081 050 4.50 20.871 _with a spline fit, shows an approximate maxi- 001474 19909.2 2.00 3.50 __ 13.987 _ mum maximum climb of 123 fps. Vint ng 2123 1 3.2.2 Performance Climbs ‘Once a climb schedule for a given purpose is determined, it must be validated by ac- tually flying the schedule and measuring performance. Time to climb, climb angle, fuel used during the climb, and distance traveled during the climb are measured along the op- timized path to prove that contractual specifications or certification requirements have been met by the design. These easily measured quantities are not all that the test team must evaluate in performance climbs, however. Operational factors, even though they may be qualitative rather than quantitative, are also important. Pilot visibility and general6 Climb, Descent, and Tum Performance Tests, Chapter 3 handling qualities fall in this category. Subsystems operation may also be affected by the climb schedule. Engine cooling, critical altitudes for supercharging, and other powerplant characteristics are typical of the subsystem operational characteristics that must be ex- amined during the performance climb. All in all, performance climbs are the validity check of not just the climb speed schedule established by the flight test team, but a verification that the entire vehicle and alll its subsystems perform as expected in a climb. a ny Mi a ze | is 6000, 175 Fig. 3.13 Example Performance Climb Data Card 3.2.2.1 Performance Climb Procedures. Figure 3.13 illustrates a typical data card for a performance climb and emphasizes some of the data collection requirements. Gene- rally, the piloting task during a performance climb demands considerable concentration on the part of the pilot. Therefore, it is best to have an automatic data recording device (video camera, photopanel, magnetic tape, or telemetry) if the airplane is occupied by only the pilot. With a second crewmember and a relatively low performance configuration, data can be hand recorded. Ordinarily, it is best to treat hand-recorded data as backup to be used only in case of instrumentation failure. One of the keys to a successful performance climb is the entry to the maneuver. The desired power, configuration, and altitude (or rate of change of altitude) should be estab- lished and stabilized before data collection starts. The climb schedule itself must be relig- iously followed; if, for example, indicated airspeed is used as the primary cue for the pilot, it should be maintained as close to the desired as possible. A sensitive instrument (air- ‘speed or Mach meter) must be used to see these small changes in the controlling variable quickly. (Even a +0.5 knot variation calls for control actions on the part of the pilot). As with sawtooth climb tests and level accelerations, plan ahead to minimize the ef- fects of wind gradients. Since a significant change in altitude is involved, you should fly 90° to the average wind. Each airplane and each configuration will dictate different control techniques, but the pilot must be positive and make smooth changes at all times. Where there are abrupt ‘changes in the profile, either due to mechanical constraints in the propulsion system, limits ‘on the airframe, or an unusual climb profile, the pilot must anticipate the required changesChapter 3 Climb, Descent, and Tum Performance Tests s7 and the flight test engineer must expect to correct the data for variations from the opti- mum. At high altitude and/or low airspeed (low dynamic pressure), deviations from the Profile must also be expected. 3.2.2.2 Performance Climb Data Reduction. The data reduction procedures for sawtooth climbs and level accelerations are subsets of the corrections required for the complete performance climb. To round out the discussion of climb corrections, it is con- venient to restate the kinds of corrections that may be required. Three types of correc- tions are usually of interest: @ Correction for nonstandard atmospheric conditions © Acceleration errors © Weight corrections Of these three classes of corrections, nonstandard weight corrections and induced drag corrections have already been discussed. Therefore, only the first two types of corrections need explanation to complete performance reduction methods for climb performance. Fig. 3.14. Tapeline Altitude Correction for Rate of Climb ‘The correction for nonstandard atmospheric conditions takes into account the effects that test day temperatures at the measured pressure (geopotential) altitudes have on alti- tude measurements, airspeed measurements, and available thrust from the powerplant. Figure 3.14 shows that as test day altitude increments (as measured by an altimeter set at 20.92 in. Hg.) are not the same as actual or tapeline altitude changes if temperature does not follow the assumed standard temperature profile. This temperature deviation may be due to either a nonstandard lapse rate or to a warmer (or Cooler) surface temperature with a standard rate of decay of temperature. The latter condition is sketched in Fig. 3.14 and is the first correction normally made to rate of climb data. Indicated ambient pressure change Ap., reflects a potential energy change of Ah; not the actual potential energy change Ahg. To correct for this shortcoming in the measurement system, the hydrostatic ‘equation (eqn. 2.5) can be written for each altitude increment. =—APe Pata 90 Then, dividing Ah; by Aha APw hy and = Ahg=——= . P9058 Climb, Descent, and Tum Performance Tests, Chapter 3 Finally, this so-called tapeline altitude correction (good for small deviations only, of course) can be expressed as the limit of the altitude increment divided by the corresponding time increment. By convention this correction is labeled with the subscript 2 to indicate that it is the tapeline altitude correction. #) dh Teas G fg dt Try (322) Example 3.3: During @ check climb on a new jet trainer an indicated rate of climb of 1100 fpm was re- corded at 9000 feet pressure altitude. The outside air temperature was 32° F at that test altitude. Cal- culate the tapeline altitude correction for this measured rate of climb under these conditions. To make the correction, we must have absolute temperatures at the test altitude on the test day and on a standard day. Te = 459.69 + 32 = 491.69°R and T= 486.61°R from standard atmosphere tables, Then, eqn. 3.22 gives: 2) th Tog _ 4, 48651 1008635 fpm F Jy at To, 1 497.69 ” The next correction made is for changes in thrust output of the powerplant due to non- standard temperature variation. Thrust horsepower available can be written as THP A= TyV. and THPR= DV, (Note: Net thrust is denoted by T, in the following development. Do not confuse this sym- bol with any temperature. in the literature conflict is sometimes avoided by denoting thrust with F,,) If sig and ¢ continue to denote standard day or test day conditions, respectively, the ratios between standard day power and test day power are: THP pete _ TeVews Towa [Testa THPy, Voie Ties THP Reig DstMorsis _ [Torsig TH, DiVay Toy ‘The square root of the temperature ratio is obviously an important correction parame- ter for thrust output on a nonstandard day. The key to this simple relationship is that Mach ‘number for test day conditions is the same as for standard day conditions. Defining a new corrected rate of climb as: Since M;= Msig implies that Dgtg = DyChapter 3 Climb, Descent, and Tum Performance Tests 58 2) THP agg, ~ THE = (3.23) (3), i . Th Aly Lot the increment of net thrust be: ATy=Theuy —Try OF git Int Int ‘Substituting into eqn. 3.23: dh" Theta 1 [Teo _ ATh 1 [Te (2), -(2 THPa, ~THPR, ag VT, THP a, THER, +7, THPA WT Hf we make the tapeiine altitude correction first and assume that it fully reflects test day power available and power required, THP,, - THI (2) _ HPAL THPR: hich directly substitutes into the correction above, 2 at Ww z a) Tro, ( Tien) a) =e] +e (3.24 te 19 V Tom 2” Wy, (324) The climb correction expressed in eqn. 3.24 now includes both the tapeline altitude correction and an allowance for the difference in engine thrust available and required on the test day rather than on a standard day. Notice, however, that to make this differential power correction we need to measure the difference between test day net thrust and stan- dard day net thrust. It is very difficult to make such a measurement for most airplanes; usually, we must rely on engine manufacturer's data rather than direct measurement. ‘Since this thrust differential is so hard to measure directly, (*) is often one of the most 3 controversial rate of climb corrections. Example 3.4: For the climb of Example 3.3, 100% of rated rpm of its nonafterburning engine was used. ‘The calibrated test engine was instrumented with pressure taps and net thrust was calculated as 4627 Ibs while climbing at a true airspeed of 375 fps. The engine manufacturer has tested the engine and ‘shown thet on @ standard day at 9000 feet the engine should develop 4800 Ibs of thrust. The initial ‘weight of the airplane was 12780 Ibs and 2100 ibs of fuel have been bumed. Calculate the rate of climb ‘correction for variation in thrust horeepower due to nonstandard temperature for this test point. ‘atin, Tay, 801 Ta, 8r0 needed. Using the values previously calculated, eqn. 3.24 gives: (2), = EE (cmerson + Mt ARMIES IO), ec et consistont between the two additive torms inside the () (2) = 1445.8 fprn| A The third and last correction in this series of rate of climb corrections takes into ac- count the effect of nonstandard temperature lapse rates on measured true airspeeds. Figure 3.15 (following page) illustrates the rationale for this correction, remembering that, even for constant indicated airspeeds, true airspeeds for each indicated pressure altitude60 Climb, Descent, and Tun Performance Tests, Chapter 3 will be affected by nonstandard temperature. This apparent acceleration will occur whether the climb is made at constant airspeed or at constant Mach number. Standard Standard hy day hy day 1 A \ --4 Hot day with = ha 1 same lapse rate as ‘standard day Hot day with \ lower lapse rato ~ythan standard day Fig. 3.15 Apparent Acceleration Correction ‘The apparent acceleration correction due to nonstandard temperature lapse rate can be made using the acceleration factor introduced earlier. Applying eqn. 3.9 to obtain the accelerated rate of climb for both standard day and test day conditions (*) _ aig = Pats Vong a (2) Mg = Dt)Meoe DE gg (1 ea ens) at), wr Yor Mer) 9 dhstg g oh After dividing to obtain the ratio between these two rates of climb, apply the binomial theorem to each of the terms containing the acceleration factor and retain only first order terms since only small changes (differential corrections) will be considered. The result is dhstg /dt ( Veo, Sow Veo sat) ( Veo AVeor Veogiy AVenstg ) =(1-—Patt eats 4g Moe Oe) fy, Mee tee Means HOT, dhy/at 9 Ah NOG ah) tg hg aha Ifitis also assumed that Visgig® Vaoy ANd that Agi ~ Ah Bag” (a (tS ) ¢ stg SO, gah (avon ~A¥ar) Reverting fo serial numbering of the rate corrections & F Ql te (¥en - Ave) 025) Note that in eqn. 3.25 Ah and AV,, are smaill increments in altitude and true airspeed ‘over which the measured climb rate is measured. These differences must be small for our approximations to be valid; so climb rate must be measured over a small altitude change and the corresponding airspeeds must also be small. USChapter 3, Climb, Descent, and Tur Performance Tests, 61 Example 3.8: Continuing our performance climb corrections as introduced in Examples 3.3 and 3.4, the ‘airplane's rete of climb is measured between 8500 feet and 9500 feet to obtain the measured rate of climb. At each of these attitudes, the outside air temperature was 32.7°F and 31.3°F, respectively, and ‘constant Mach number was maintained for this segment of the climb. For the same pressure altitude band, calculate the epperent acceleration correction to the rate of climb previously corrected with tape- line altiucde corrections and for power available due to nonstandard test day conditions. Firat, calculate AV,, for the Ah in question. The temperatures at each altitude are: T.,, = 459.69 + 32.7 = 492.30°R and Ta, = 469.09 + 91.9 = 490.90°R. Alto, the constant M., (identical at both test and ‘standard day conditions) comes from standard day conditions: 375 Me Taayarie sonaseay 0 "°* For the test day lapse rate, the AV between the two measurement altitudes is: Ve, =MeyR (Ta, - fo, )=-8:9879 fo For a standard lapse rate: AV, = MoV7R (\Feay - (fom, )=-1:878 fos ‘Substituting into eqn. 3.25, (2) = mae a 13,7578 + 5.3679) 3), = 1887.0 fpm (62.17405)(1000); Example 3.5 clearly illustrates how a nonstandard test day temperature lapse rate af- fects the correction of test day climb corrections to standard day conditions. The need for dh’ rate of climb corrections is obvious from the size the cumulative correction, (? ‘i 4 3.2.2.3 Summary of Rate of Climb Corrections. Collecting all rate of climb correc- tions (including those developed earlier), the complete correction is: #) = [Fests [ (om) ATny e ec” Y Toy ($), +a art nn a), aa (Cea aah )\ Tuy -V, Vey Viv) [Tropa ( (ah AT, St - ew | j Sant) | a (= (8Veoaes Ne) ( aah ) 7 (2) + THEA ae (3.28) 3.2.3 Tum Performance Flight Test Techniques ‘Tuming performance is usually determined in one of two ways: (1) directly from level tuming flight at constant power settings or (2) from level acceleration tests. The instru- mentation required is typically quite minimal for the first kind of tests and the results are presented in several different ways. Cockpit instruments are often used to simply set the bank angle, the load factor, and the airspeed. Timing may be done with a stopwatch and the results are adequate. Thrust-limited turning performance can often be determined in this way quite satisfactorily.62___________Chimb, Descent, and Turn Performance Tests, Chapter 3 3.2.3.1 Level, Unaccelerated Turn Technique. The simplest approach to measur- ing thrust-limited tuming performance is to simply fly tuns at constant altitude, constant Power setting, and selected constant airspeeds. Typically, the powerplant is operating at a Power setting at or near military or maximum power (since we are discussing thrust-imited tums). Obviously, in such a stabilized level tum the bank angle and the load factor are also constant. The instrumentation required is simply a stopwatch to time the total head- ing change. Frequently, a stabilized tum of 360° is used for convenience in this timing, though the tum could be through any heading change desired. To avoid significant timing errors due to human reaction times, it is usually best to turn for at least a minute, prefera- bly two minutes. Tums should be made in both directions to check for any secondary ef- fects on the tum performance. Care must be exercised by the pilot to guarantee that alti- tude and speed variations are kept small, usually less than +50 feet and £1 to 2 knots throughout the maneuver. It is far more important to maintain a constant airspeed throughout the timing period than it is to make the tun at exactly the aim airspeed. De- Pending on the resolution of the cockpit sensors available, the pilot may use either the at- titude indicator (or some more precise bank angle indicator) or the accelerometer to maintain constant load factor. At high load factors with the bank angle increasing beyond 60° for level fight, greater care must be exercised to stay within acceptable tolerances. If the tum continues beyond 360° and the pilot has maintained the altitude precisely, he may encounter the jet wash of the test airplane. Such an event can upset the stabilized tum, but usually is nothing more than a nuisance. The reliability of cockpit heading indications should be checked; if they are reliable in tums at all oad factors, the heading indicator can provide starting and stopping points that are useful. Hand recorded data are usually quite adequate, but if other instrumentation (rate gyros, stable platforms, event markers, and the like) is available, it should be used and backed up with hand-recorded data. As with all tests, care must be taken to observe all aircraft structural and serodynamic limits. At speeds near the predicted maneuvering speed (comer velocity) the pilot is responsible for insuring that the maximum allowable load factor is not exceeded. Another caution in this part of the performance envelope: remember to relax the load factor before rolling out of the maneuver; typically, the load factor limit in a rolling maneuver is considerably lower than in a pure pitching maneuver with zero rol rate. 3.2.3.2 Level Acceleration Technique. This technique for level acceleration data is identical to that described in section 3.2.1.2. The only change is in how the data are used to describe the performance of the airplane. It is important to carry the level acceleration ‘out to maximum airspeed so that the point for zero excess thrust is established. The usefulness of the data to characterize turing performance depends on the thrust limita- tions. Said another way, Ps (again as discussed in section 3.2.1.2) is just as useful as a Measure of merit for thrust-imited tuming performance as it is for climb performance. Fig- ure 3.16 (next page) illustrates one of the plots commonly produced from level accelera- tion data to depict the tum performance capability of a fighter aircraft. These plots, some- times called “doghouse” plots because of their characteristic shape, often include both the tum performance measures like rate and radius of tum and P, contours.Chapter 3 Climb, Descent, and Tum Performance Tests 6 Mach Number Fig. 3.16 Tuming Performance Measures of Merit By overlaying the P, contours, whether the turn rates and radii are for sustained or in- stantaneous tums can be determined. If the tum rate and Mach number are such that P, is negative, the indicated tuming performance is an instantaneous tuming capability. The ‘aerodynamic and structural limits also are clearly indicated in the “doghouse” plot. 3.3 SUMMARY This chapter has introduced the most common flight test methods used to optimize climb schedules, to validate climb schedules, and to measure tum performance. The methods are straightforward, though ‘somewhat time-consuming, with the level accelera- tion technique holding a definite advantage over sawtooth climbs and level, unaccelerated tums in efficiency of data collection. Like most flight tests, climb tests are relatively easy to describe in theory, but they demand careful attention to detail in data reduction to achieve the accuracies needed for certification or specification compliance purposes. Tum performance fiight tests are related to cimb and descent performance largely be- cause they are usually limited by available excess thrust. This fact allows test data to be collected in level accelerations for use in characterizing both optimized climb paths and instantaneous and sustained tuming performance. Stabilized level tums with little or no instrumentation other than what is available in the cockpit also provide considerable in- sight into the capability of the airplane and certainly provide opportunity for a simple intro- duction to the measurement of this important aspect of aircraft performance.64 Climb, Descent, and Turn Performance Tests, Chapter 3 REFERENCES * Rutowski, E. S., "Energy Approach to the General Aircraft Performance Problem," Journal of the Aeronautical Sciences, Vol. 21, Mar. 1954, pp. 187-195. Nicolai, L. M., Fundamentals of Aircraft Design, METS, Inc., San Jose, California, 1975. “Performance,” Volume 1, FTC-TIH-70-1001 (Revised December 1976), USAF Test Pilot School, Edwards AFB, Califomia, 1976. Lush, K. J., "A Survey of Performance Reduction Methods,” Chapter 3, Vol. |, AGARD Elight Test Manual, Pergamon Press, New York, 1959. “Fixed Wing Performance, Theory and Flight Test Techniques,” USNTPS-FTM_No. 104, Naval Air Test Center, Patuxent River, Maryland, Jul. 1977, 5 Dommasch, D. O., "Data Reduction and Performance Test Methods for Reciprocating Engine Aircraft," Chapter 6, Vol. 1, AGARD Flight Test Manual, Pergamon Press, New York, 1959. "Performance, Book B-2," Empire Test Pilots’ School, Royal Aircraft Establishment, Famborough , England, 1966. Bryson, A. E., Jr., Desai, M. N., and Hoffman, W. C., "Energy-State Approximation in Performance Optimization of Supersonic Aircraft,” Journal of Aircraft, Vol. 6, Nov. 1969, pp. 481-488, E. and Roskam, J., Airplane Aerodynamics and Performance, Roskam Aviation and Engineering Corporation, Ottawa, Kansas, 1980, McNamar, L. F. and Gordon, H. C., "T-38A Category I! Performance Test," TDR No. 63-27, Air Force Flight Test Center, Edwards AFB, California, Nov. 1963, 11 Payne, J. H., "Flight Test Handbook," JP Aviation, USAF Academy, Colorado, 1990. 12 “Performance Flight Test Techniques," Volume Ill, FTC-TIH-70-1001 (Revised August 1975), USAF Test Pilot School, Edwards AFB, Califomia, 1975. 13 Roberts, S. C., "Light Aircraft Performance for Test Pilots and Flight Engineers," Flight Research, Inc., Mojave, Califomia, 1980.Chapter 4 CRUISE PERFORMANCE TESTS “The cruise performance of any vehicle is one of the most important matters to be de- termined by fight test. Estimates of cruise performance are made from the very beginning of the conceptual design. Range and endurance under specified mission profiles are con- tractual requirements that must be validated for any aerospace vehicle. Hence, the topic is one of prime concem to the fight test team. ‘The elemental aerodynamic variable in determining cruise performance is the drag produced by the airframe-engine combination for the particular configuration or configu- rations and the specified mission profile. Unfortunately, there is no direct way to measure drag in fight. Test engineers therefore usually rely on approximate mathematical models for drag to start cruise performance analyses. Figure 4.1 is a block diagram representing types of drag and suggesting possible ways to simplify the mathematical model. TOTAL DRAG [PaRasiTE DRAG INDUCED DRAG WAVE DRAG — INTERFERENCE DRAG| | PROFILE DRAG |SKIN FRICTION DRAG| | PRESSURE DRAG Fig. 4.1 Types of Drag ‘One of the simplest mathematical models for low speed airplanes is the subsonic drag polar given in eqn. 4.1. co a _ Cp =Cp, + (4.1) where Cp is the total drag coefficient and Cp, is the effective parasite drag coefficient Equation 4.1 is valid only for airplanes in which compressibility drag rise is not a factor, in this book the parabolic drag polar is assumed to be a good approximation for airplanes powered by reciprocating engines. 4.1 THEORETICAL FOUNDATIONS Cruise performance is determined by both the available thrust (or power) from the powerplant and the thrust (or power) required by the airframe. Thus, it is necessary to ‘correct the outputs from different propulsion systems to standard day conditions, much like the corrections required in climb performance data reduction. However, because of the fundamental differences between propeller-driven and jet-propelled airplanes, it will be convenient to discuss the performance reductions under separate sections. Different performance reduction methods are used for the two types of powerplants. Propeller-6. Cruise Performance Tests Chapter 4 driven aircraft will be considered first since they tend to be lower performers and the data reduction procedures are simpler (though perhaps more tedious). Typically, propelier- driven airplanes are analyzed using an analytical performance reduction method that draws upon the extensive data bank supplied by reciprocating engine manufacturers. Tur- boprop-powered aircraft usually are also analyzed with an analytical method, although a mixture of the analytical approach with differential techniques from the experimental per- formance method is often necessary to account for missing information on the powerplant. Turbojet and turbofan powered airplanes typically are analyzed with an experimental per- formance reduction method. 4.1.1 Propeliler-Driven Aircraft ‘The analytical performance reduction method by which propeller-driven aircraft per- formance is usually analyzed relies heavily upon data collected by the engine manufactur- er, usually brake horsepower data taken at sea level. For any airplane, THP, = =-5-, with V,, in fps and D in pounds. With the usual definitions of dynamic pressure q, aspect ratio AR, Oswald's efficiency factor e, and assuming a parabolic drag polar with L = W. PVECOpSVo | PaVICPSVn _ PaVCopS w? 1100 1100neAR 1100 275peSVoneAR THP, = Inflight measurement of thrust, especially from a reciprocating engine-propeller com- bination, is both uncertain and requires expensive instrumentation. Consequently, it is ‘common practice to fall back on measured brake horsepower data from the engine manufacturer. But the propeller efficiency must be included in the horsepower relation- ship. npBHP, = THP, PaVeCo eS w2 So, BHPr= F700, * 276p.SVaneARng Cop’ 1 Setting A= 22 and B= ——" ,—, and multiplying by Vas, 4d 700m 275neb* np eeed BHP Vay = PaAVen + 4 BW? (4.2) P This relationship, with weight and density identified as dependent variables, clearly shows the dependence of power required upon airplane weight and altitude, the two vari- ables that must be standardized if performance data are to be comparable between diffe- rent tests. However, the power requirements can be further generalized to allow all test data for a given configuration to fall on a single power required curve. In other words, this further generalization of the power equation will allow direct comparison of test data and easy interpolation of the generalized curve to points within the envelope where test data ‘were not collected to verify the mathematical model. Such an approach allows minimum, flight test data collection to validate the mathematical model. Choose a standard weightChapter 4, Cruise Performance Tests, 87 Warg for the configuration and use pq to define a generalized velocity parameter VIW and a generalized power parameter PIV. [Pe pv __1_ [2Wade5 vw = |= and PIW=—2 =o, (4.3) Pos 550 ~ 550 poC7S oa since L= Wag =4pav28G, and = Yaifo 2 Co nd THP, = = pwc a ae *= B50 pots {f each term in eqns. 4.3 is multiplied by “one in the appropriate form, Ata given altitude and C,: Ve 2Wets vw =v, Yes _y, [oWata. (44) 2w (4.5) 2 Using eqn. 4.2, THP, can be written as THP, = noloooavd + aol ) W. 3 Multiplying both sides by Wast) : 3 2 3 = Watt) - 3, BW! jo( Mae PIW = THP,. o(: Tat) = o|PomAVe +o WC w ) Since pp and Weg are constants and np is approximately constant, we readily obtain PIWIW= AyVIW4 + By (4.6) PoC, S wa, where Ay=——-2— and By = Sid > A= 4900 21 = eponeb™ ‘A generalized power curve (Fig. 4.2) is a reasonable approximation for a parabolic drag curve and constant propeller efficiency except at either the high or the low speed ends. At low speed (high a) nonlinearities due to separated flow appear. At low Cz, com- pressibility effects are important and the curve often splits into muttiple branches. So, eqn. 46 is a simplified, but useful, mathematical model of the cruise performance for a low68, Cruise Performance Tests, Chapter 4 ‘speed, propeller-driven aircraft. The slope and intercept of this generalized curve readily give estimates of Oswald's efficiency factor and of the minimum drag coefficient. Wa e= 3 (4.7) 275Bypoxb* (4.8) viw4 Fig. 4.2 Generalized Power Curve ‘Table 4.1 Cruise Data for a Propelier-Driven Airplane TAS TAS BAP, WW PW vi? Pravin) (ts) bs) (109) 55 828) 6512 (6433486 0.0423 2700257 60 101.267 $430 88.63 20565 0.0617 26203.83 65 © 108.706 5376 96.50 260.06 0.0867 © 25004.84, 7o 118.144 5322 104.45 296.45 0.1190 24606.38, 75 126.583 5288 112.27 221.34 0.1588 24848.28, 90 135.022 5236 12034 20204 0.2097 2431.51 90 151.900 5198 135.88 177.12 0.3409 24086.47 100 168.778 5165 151.46 16223 0.5262 24571.56, 110 185656 221 183.430 5111 167.48 161.88 0.7868 2712.15, 120 202533 225 186750 5079 183.28 166.37 1.1285 © 30492.86 130 219.411 235(195.050 5021 199.70 176.78 1.5004 35303.81 140 236.289 «252 200.160 4948. «21664 19378 2.2028 41981.70 160 270.044 © 302,-«-280660 4875 «240.44 237.47 3.8712 5923366, 180 303.800 375 «311.250 4805 28265 301.34 63829 8517416 200 337.556 __458__380.140__4722__316.81_377.78 10.0740 _119683.50 Example 4,1: Table 4.1 shows data for a general aviation airplane with a propelier efficiency of 0.83. Data were collected at 6000 feet pressure altitude with an outside air temperature of 40°F. The airplane's “stan-Chapter 4, Cruise Performance Tests Cy dard" weight is 5000 Ibs. Wing area is 175 f2 and aspect ratio is 6.5. Calculate Cp, and e by plotting = (generalized power required curve end measuring the slope and the intercept ofthat curve. First, density is calculated from p= -P-., with T= 40 + 459.69 = 499,69°R, that is, or RT 1606 __ o.0010773 | P= TIrTRBRHABEER) ~ 20019773 stugsmM®. Then, ©; 0.0973769 0.83188. With V,, = 55 knots = 3 92.828 fps, VIW = 98828) [(083188)(5000) . g0.628 fps. Also, PIW = THP,, oa) . 1 h $ 3 npBHP,. oe) oF PW =(0.89)(512) o.seo( 202) =335 hp. Finally, carying out the mu- tiplications for (PIW)(VIW) and VIW* gives the last two columns in Table 4.1. The plot below shows the re- sulting generalized power curve, A linear curve was fitted to the last 8 data points in Table 4.1. The result- ing straight line had a slope of 0.0000100502 end a PIW-VIW intercept of 19573.9. Utilizing eqns. 4.7 and 4.8, we estimate the required perameters: 50007 (e=0648) © = (376)(062.6(0.0023769)(19873.9) * (0.0000100502)(1100) = Co, =" (6,0023769)(175) Co, = 2.0266) 0 21108) 4109) 61109) 4109) viw* 4.1.1.1 Propeller Efficiency. Propeller efficiency is not easily measured in flight, al- though Bull and Bridges! have used specially designed instruments and a maximum like- linood mathematical modeling technique to measure it. Their approach shows consider- ‘able promise, but the method is not yet widely used. More commonly, propeller efficiency is estimated by either an analytical technique” or a semi-empirical method. Each of these methods is relatively simple, provided the appropriate inputs are known. To use them, one must know altitude, true airspeed, brake or shaft horsepower, and number of propeller blades, all readily obtainable configuration constants. In addition, variables like advance ratio, activity factor, solidity ratio, and integrated design lift coefficient that are peculiar to propeller performance must be derived. Advance ratio J is defined as the forward speed of the propeller divided by the rotational speed:7______Cnuise Performance Tests Chapter 4 J= % , Where n is propeller rotational rate in revolutions/second Activity factor AF is a measure of the power absorbed by the propeller and is defined as 19 = 100000 'f ¢ V3, (4.9) AF =e J pew (4.9) where x=, D=2R, and c= ox) (Fig. 4.3.) The non-dimensional hub radius, usually taken to be about 0.15, is x. The solidity ratio ‘SRis simply the ratio of total blade area to the disk area swept by the rotating blades. SR= sr= where Bis the total number of blades. Fig. 4.3 Geometric Variables for a Propeller The integrated design lift coefficient is defined as 10 Cy, =4 fog, xPax (4.10) xy where cy, is the individual blade section design lift coefficient and is a function of x. Optimizing the propeller for cruise performance generally means that C,, will be small (on the order of 0.35), while optimizing the propeller for high thrust at lower forward speeds (climb speeds, for example) will result in an integrated design lift coefficient of about 0.60. Itis convenient (for the same reasons that we use nondimensional coefficients like Cy, Cp, and Crm) for engineers to describe propeller performance using nondimensional coeffi- cients. Three terms typically characterize an aircraft powerplant’s output: power, thrust, and torque. The three important propulsive coefficients include: © Power coefficient Cp = 550 BHP Poo” D © Thrust coefficient cp-—, Poh*D' © Torque coefficient Cg-—2 PeotD*Chapter 4 Cruise Performance Tests m Having indicated how propeller efficiency can be estimated, consider next how we es- timate the power available from the engine. 4.1.1.2 Power Available. Earier, it was pointed out that for propeler-driven aircraft it is common to use engine manufacturer's data to estimate the power output available from the engine. Certainly this approach is a widely accepted practice for reciprocating en- gines. However, itis possible to directly measure the output power to the propeller shaft. ‘A torque meter can be installed on many engines for this purpose. Production torque meters, which generally do not provide the accuracies needed for flight test, are routinely installed on airplanes powered by turboprop powerplants. To find BHP, you only need the torque meter reading, the rotational rate of the propeller shaft, and a calibration constant for the torque meter (usually found during ground tests using a dynamometer). 2pnQ BHP = AF OF BHP = Kn, Qic Flight test quality torque meters are hard to install, require frequent calibration, and are expensive. Smaller engines cannot accommodate commercially available torque me- ters and many project budgets cannot afford their cost. So, engine characteristic curves, ‘obtained from ground test stands, provide estimates of available engine power. [SEA LEVEL PERFORMANCE “ALTITUDE PERFORMANCE | (Part Throttle) 6 (Full Throttle) ig 20 22 24 26 28 0 ‘5000 70000 75000 ~~ “20000 MANIFOLD PRESSURE PRESSURE ALTITUDE (in Hg abs) td Fig. 4.4 Engine Power Chart for Normally Aspirated Engines Since the manufacturer's data sheets (similar to Figs. 4.4 and 4.5) are usually based ‘on ground tests, the flight test analyst must estimate engine output data for a given test condition from charts like those shown. Remember that measurements are usually taken ‘at of near sea level for such charts. The altitude estimates are usually calculated by the powerplant engineers. For reciprocating engines these data are usually quite accurate72, Cruise Performance Tests Chapter 4 and their ready availability makes them attractive for use. Also, they are usually based on ‘standard day conditions; corrections must be made for nonstandard conditions. The flight test engineer must carefully study the engine data sheets for his powerplant to be sure that all nonstandard conditions have been considered. Once again, a few terms must be defined to make this process easier. SEA LEVEL PERFORMANCE| ALTITUDE. PERFORMANCE Part Throtie) ‘ull Throtto) 20 24 28 32 36 40 44 MANIFOLD PRESSURE PRESSURE ALTITUDE fin Hg abs} tt Fig. 4.5 Engine Power Chart for Supercharged Engines 2 et g 2 2 Full Throttle = * id >. High Blow) al Full Throttle Gritical. = BMEP (psi) Fig. 4.6 Critical Altitudes for Reciprocating Engines For a reciprocating engine the critical altitude is the altitude at which the throttle must be fully opened to develop rated power at rated RPM (Fig. 4.6). The British give this alti- tude a more descriptive name, full-throttle height, because critical altitude is the altitude at which the throttle must be fully open to produce the specified horsepower. Critical alt- itude is the altitude at which the power begins to decrease regardiess of the position of the throttle. Whenever an engine reaches a structural limit because of cylinder pressures or if the manufacturer seeks longer life, or for any other practical operational reason, ratedChapter 4, Cruise Performance Tests. 73 power may be produced at part throttle up to the critical altitude. Such “de-rating” is, in fact, more common than having not having any operating limits. Figure 4.6 illustrates this idea for the case when the power output of the engine is limited by the pressure allowed in the cylinders. Turbosuperchargers are the most common form of supercharging in modem reciprocating engines and engine limits may be set by BMEP, BHP, turbine RPM, or tur- bine temperatures. Of course, multi-stage supercharging results in more than one critical altitude (again, see Fig. 4.6). ‘A reciprocating engine that takes in ambient air and does not compress it prior to the combustion process is normally aspirated. If a geared compressor is attached to the crankshaft accessory drive, the pressure of the air used in the combustion process may be raised to produce more power. Such engines are called ground-boosted engines. Both normally aspirated and ground-boosted engines may have sea level as a critical altitude. Fuelair mixture setting also strongly affects available brake horsepower. Figure 4.7 ilustrates how deviation from the desired fuel-air ratio affects reciprocating engine per- formance. Obviously, the mixture must be set carefully during cruise performance tests. i i i ' Backing, i "Backfirng, Roughness! 'Stoppage Overheating; i ten MIXTURE SETTINGS ae Fig. 4.7 Effect of Mixture Setting 4.1.1.3 Corrections to BHP Available. For part throttle operation with no mechani- cal limitations, corrections to power available (primarily those dealing with temperature) from a reciprocating engine are usually given on engine charts (Fig. 4.8). To estimate actual test day horsepower from manifold pressure, engine rpm, altitude, ‘and inlet temperature on such typical charts, the following procedure is recommended. ‘The outlined procedure helps explain how and why fight tests are necessary to verify in- stalled engine characteristics from manufacturer's data. Keep in mind that these charts are usually based on test data taken near sea level; altitude performance is calculated from that information. A typical estimate of test day power available is obtained by: 1. Locate point A on the full throttle altitude curve for a given dry manifold pressure-full throttle rpm curve. 2. Find the corresponding point B on the sea level performance curve for the manifold pressure and rpm used in step 1. 3. Project the horsepower for point B to the ordinate on the altitude perform- ance scale (point C) and connect points A and C with a straight line.74, Cruise Performance Teste Chapter 4 4. At the test day pressure altitude and standard day temperature (nomogram at the bottom of the altitude performance chart in Fig. 4.8) project upward to the straight line found in step 3 to locate point D. 5. Modify the horsepower at point D using the following expression to estimate actual (nonstandard) power output available under test conditions. BHP aq = BHP, | 780. or MAPyg = MAP, | std. ‘std t Th ‘std t. Th (Manifold pressure limit for continuous operation 18 20 22 24 26 28 "5000 10000-15000 20000 Manifold Pressure Pressure Altitude fin Hg abs) (feet) Fig. 4.8 Engine Operating Charts Full throttle operations often involve other limits or constraints that require corrections including: ‘© Nonstandard intake temperature (carburetor or supercharger inlet) @ Nonstandard exhaust back pressure @ Nonstandard BHP @ Nonstandard turbine RPM | For a given engine and installation, consult the manufacturer's data base for such lim- its that apply to the specific configuration under test. Any limit that may differ because test atmospheric conditions are not standard day conditions is a subject for careful study. Having introduced the performance reduction methods most commonly used for pro- peller-driven aircraft, attention must now be given to jet-powered airplanes for which ex- perimental methods are more appropriate.Chapher 4, 4.1.2 Jat-Powered Aircraft Experimental performance reduction methods are quite commonly applied to jet- ‘powered eireraft because their ift-drag relationship is complicated by their dependence on ‘Mach number. Wave drag, shown in dashed lines in Fig. 4.1, must be included for these higher performing airplanes. The simple parabolic drag polar is usually not an adequate model, except for a few specific flight conditions. Performance measurements for jet air- , therefore, are somewhat more complicated by the number of variables that must considered. Dimensional analysis, based on Buckingbam's Pi theorem. is frequently used to reduce the number of parameters that must be covered in the test program. Even purposes. The objective of this introductory text is to introduce the fundamentals with a typical set of variables used for cruise performance and give the student a foundation for ‘when he is called upon to apply the principles to a particular problem. 4.1.2.1 Buckingham's Pi Theorem. Buckingham's Pi Theorem can be stated as: If @ physical problem is characterized by n variables having p fundamental units, the functional relationship between the variables can be expressed dimensionless numbers. For the typical flight mechanics problem (like the turbojet performance problem), the fundamental units are mass, length, and time, abbreviated m, L, and t. Then, assuming that there only n variable that significantly affect performance, Buckingham’s Pi Theorem asserts that the total number of parameters that must be measured to characterize the performance is reduced from n to n - 3, if one measures performance in appropriate di- mensionless variables rather than the n physical variables. For flight test purposes, this reduction in the total number of variables is extremely important to make the testing more cost-efficient and less time-consuming. The steps in this process include: ‘ Select p variables from the physical variables. These variables should in- clude all the fundamental units so the following algebra can be completed. © Form n- p dimensionless equations by combining each of the variables from step 1 in tum with each of the others. This procedure will be itustrated with an example in the next paragraph. But, it must be emphasized that the process is totally dependent on the original assumptions made; that is, if one or more of the significant physical variables is omitted from the set p in step 1, its effect will be left out of the answers. Consider now the output of a jet .. The important measurements might be taken 18s net thrust Ty, weight flow rate of air wa, weight flow rate of fuel wr, and characteristic temperatures 7). (7; might be turbine inlet temperature, exhaust gas temperature, or a host of other temperatures depending on the specific engine under consideration.) These measurements may be functions of true airspeed V,, ambient temperature 7. ambient static pressure p... a physical dimension of the engine like the diameter of the compressor rotor D, a characteristic rpm N of the rotating parts like the compressor rotor, air viscosity 1s, and various engine component efficiencies like inlet efficiency nj, compressor efficiency76. Cruise Performance Tests Chapter 4 Ng, bumer efficiency np, turbine efficiency nj, and nozzle efficiency np. The list of vari- ables might be simplified by assuming that each of the component efficiencies is also a function of V,,. Tz. Pa, N, and D. Further, viscosity can be eliminated as an independent variable if it can be described adequately as a function of p, and T,. Notice that this as- sumption is tantamount to saying that Reynolds number effects are not important. If Rey- olds number effects do significantly affect the engine performance, the model will not ac- ‘count for them. Example 4.2: Applying Buckingham's Pi Theorem to the functional relationships described above, that is, Tn = T(r Tes Pax N, D). Select the variables T,, N, and D and operate on the dimensions of V.;, Ti, and p. t0 group T,, N, and D as dimensioniess constants xy, na. x3 Firat, we select x, = T,!NPD°V., end apply Buckingham's Theorem: eee ITE) ‘Solving for equal exponents on both sides of this equation & 0 = atcet aso m OFa & + tb 0 = 2e-b-4 ena ‘Therefore, 4 = (4.1) ND ‘Simitarty, the other two nondimensional groupings can be found by solving the dimensional equations ‘or Ty, and D-with temperature first end then with pressure, Consider nest, x = T;*NPDST., which ives! owe BIE (4.12) (4.13) Continuing to specialize the example to jet engine performance, any of the functional relationships Suggested by eqne, 4.14, 4.42, oF 4.13 can be written in reciprocal form. For example, a= fay, 2) imple no (5 ee (S 2)Chapter 4, Cruise Performance Tests, 7 Noting thet ND has units of length over time, and that ./T,, 8 proportional to sonic velocity and has the ‘same units, the nondimensional groups can be altered to the following form without affecting the results of the dimensional analysis. Mele) el) If the engine has fixed geometry so that the characteristic length D is not a variable, the functional re- lation can be further simplified. While reordering the constants for engine geometry in the functional rela- tionship, itis also convenient to reference the net thrust and temperature to standard day sea level condi- tions, rather then to ambient conditions alone. Making these adjustments in the constants results in Tan *) (4.14) 5715 Mao Te, ) 120 3 ), FUEL FLOW (Ibs/hr) phages EXHAUST GAS TEMPERATURE (°C) STATIC THRUST § Br zs (ibs fuel” thrust perhr) AIR FLOW (Ibs/hr) §8s8sses 90 100 50 60 70 80 ENGINE RPM (%) Fig. 4.9 Dimensional Engine Performance: Constant Altitude, Constant TAS You may still question the usefulness of nondimensional parameters. Figures 4.9 and 4.10 depict representative sets of variables that describe jet engine performance. Figure 4,9 illustrates engine performance for a single altitude and a single true airspeed. (Static thrust, rather than net thrust, is shown; but that fact is not important to understanding the tity of nondimensionalization.) Figure 4.10 shows what happens when just one of these parameters is charted for just two altitudes and three throttle settings. ‘Six curves result for net thrust alone. Obviously, merely presenting the data becomes @ cumbersome task. Collecting flight test data at each of the needed altitudes and at a complete set of engine rpm settings is simply not feasible. And, even if such a data collection effort were com- pleted, the resulting data base would be quite unwieldy.78. Cruise Performance Tests Chapter 4 wae * 3 3 1000 "Te 40,000 ft - 80% rpm = — — oO a2 08 1.0 0.4 06 MACH NUMBER Fig. 4.10 Net Thrust Versus Altitude and Engine RPM Nondimensional parameters are used to generalize data from several different alti- tudes, throttle settings, and atmospheric conditions into a single curve ideally. The prin- ciple is exactly the same as that used in generalizing the power curve for propeller-driven airplanes in section 4.1.1. Of course, no generalization is perfect; single curves rarely re- sult. Instead, fairy closely related families of curves usually result. Figure 4.11 illustrates how Mach number effects (and other parameters that were ignored in the orginal a8 tions) generate a family of generalized net thrust curves for -2- versus 7. ye Tn 20 i ys os a 8 N ve Fig. 4.11 Generalized Net Thrust Part of the reason that these plots do not generalize perfectly lies in the assumptions used to develop the nondimensional relationships. Reynolds number effects were ignored by ignoring viscosity. But, if the component efficiencies depend strongly on Reynolds number, the generalization we seek does not occur. A family of curves results, rather than a single curve, just as Fig. 4.12 illustrates. Figures 4.12 and 4.13 suggest that mass flowChapter 4, Cruise Performance Tests, 79 rates and fuel flow rates do not generalize as well as thrust. Other variables (and Rey- nolds number is clearly one of them) could be included in the nondimensionalization pro- cedure. However, this simplified model serves our purpose of introducing the Buckingham Pi Theorem and its usefulness in flight test planning. ma & & Ce Te 3. Le N we Fig. 4.12 Generalized Thrust and Mass Flow Rate we SFC FA , Se fe |e Sa g & N N N Fig. 4.13 Generalized Fuel Flow Rate and Specific Fuel Consumption Developments similar to the one illustrated in Example 4.2 result in nondimensional groups as shown in Table 4.2. Nondimensional parameters corresponding to each of the20. Cruise Performance Tests Chapter 4 Performance measures of merit can be obtained with the Buckingham Pi approach; they fully describe engine performance. And, equally important to the flight test engineer, they allow test day performance data to be corrected to standard conditions easily. Table 4.2 Nondimensional Jet Engine Performance Parameters Dimensional Parameter Dimensions _ Nondimensional Groups Net Thrust mL Th of NL Ma) ? 3 We" * Fuel Flow me Me nf Ny, ) 8 we We. Air Flow a t Temperature Le rd SFC none Dommaschs lists and derives a more complete set of nondimensional parameters. It is quite clear that grouping sets of physical variables systematically, as the Buckingham Pi theorem allows us to do, is extremely helpful in planning flight tests for cruise perform- ance. These techniques are most useful for jet airplanes, though many of the concepts can also be applied to propeller-driven aircraft. 4.1.2.2 Combining Engine and Airframe Parameters. To utilize the power of di- ‘mensional analysis and to reduce the size of the test matrix for an experimental perform- ance reduction, we now need to express airframe parameters as well as engine parame- ters in nondimensional terms. This nondimensionalization is straightforward if you simply arrange the aerodynamic forces in functional form. Lift can be written ca 2WRTa au WS o, =Fi(Me) PoVoS PoVsS YPoMZS PoMZS L (4.1) Similarly, Cp = Fx( Mot) (4.16) In steady, unaccelerated flight net thrust is equal to drag and eqns. 4.16 and 4.14 can be equated: fy (mot) = Fie( Mor). Though the form of the functional relationship is unknown, this expression can be used to "solve" for any one of the three variables in terms of the other two. Choosing Mach number as the independent variable, (4.17)Chapter 4 Cruise Performance Tests 81 the task has been simplified to controlling two variables (. and *) and holding “ ‘constant by adjusting altitude. Thus, dimensional analysis makes experimental perfor- mance data reduction tractable for jet-powered airplanes. We have also avoided measur- ing thrust directly. Instead, performance is obtained indirectly from more easily measured variables like engine RPM, Mach number, weight, atmospheric pressure, and atmospheric temperature. These measured variables can be readily evaluated. One difficulty in this indirect approach is that engine RPM may not be @ constant measure of engine thrust over the entire life of the engine. Mechanical wear will take its toll. There is also variation from ‘engine to engine in the relationship between RPM and thrust. If these differences are relatively small, the performance of a fixed geometry turbojet engine can be validated ex- perimentally using the nondimensional parameters shown in Fig. 4.14. We sve Ww 5 al Fig. 4.14 Typical Performance Map for a Fixed Geometry Engine For variable geometry engines the functional relationship of eqn. 4.17 must include at least one more parameter. Therefore, it is usually best to map variable geometry engine performance in terms of one of the nondimensional thrust parameters. The same sug- {gestion holds for dual rotor engines because the additional performance variable compli- cates the performance relationships in the same way. In any case, the use of nondimen- sional variables simplifies the task of experimentally measuring the performance of the jet- powered airplane. Without this simplification, validating the performance would be an ‘even more tedious process. 4.1.3 Endurance ‘The total time that an airplane can loiter on a specified amount of fuel is simply a func- tion of how much fuel is carried and how fast is it consumed. The most obvious difference in the endurance equation between propelier-driven airplanes and jet-powered airplanes is in how fuel consumption is expressed. 4.1.3.1 Propeliler-Driven Aircraft. Fuel flow rate is the basic parameter of interest for endurance. Typically, during cruise tests, the entire weight change is due to fuel con- sumed. Consequently, wy 2%. For a propeller-driven airplane, specific fuel con-2, Cruise Performance Tests Chapter 4 sumption is expressed as pounds of fuel bumed per horsepower produced at the propeller shaft: c= me = SFC. Rearranging this relationship and integrating E={ where Wo aircraft weight at start of cruise segment We = aircraft weight at end of cruise segment — = endurance for a given fuel load, We- Wo (Note: Eis often defined simply as the maximum time the airplane can fly with a full fuel load. In this book, that time is denoted by Emax) AV But BHP, = "p for level fight. Then (4.18) Constant Weight Constant Altitude Fig. 4.15 Envelope of Fuel Flow Curves ‘The quantities to be measured include primarily fuel flow rate and weight of the air- cts plane (eqn. 4.18). The aerodynamic term 7 guarantees that maximum endurance will D occur at @ specific airspeed. Speed for maximum endurance can be obtained by plotting a family of constant altitude and constant weight fuel flow curves x increasing propeller RPM (see Fig. 4.15). For small changes in weight, changes in "2. are often negligible. if ‘so and if the drag curve is parabolic, the true airspeed for maximum ‘endurance can be es-Chapter 4 Cruise Performance Tests 83 fimated from the commonly used Breguet approximation: Ven = ViC,"51Co)mm Recall that, under these simplifying assumptions, 3Cp, = Cp’. Propelier-driven aircraft endurance is readily measured by recording fuel flow rates accurately. Altitude has a small effect, but care must be taken to ensure that the mixture settings are consistent between comparable data runs. Weight flow rate of the fuel should be measured, since the specific gravity of aviation gasoline varies significantly with temp- erature, an uncontrollable variable. Configuration, including power settings, must be set Up painstakingly to avoid excessive scatter in cruise data. 4.1.3.2 Jet-Powered Aircraft. Measuring endurance for a jet airplane is very similar to measuring it for propeller-driven airplanes. Endurance is still defined from measure- ment of weight fuel flow rates; hence, the integral equation for endurance is unchanged. However, specific fuel consumption for the jet engine is expressed in different units and is called thrust specific fuel consumption to highlight that difference. TSFC = = 4 in ‘where c; is in pounds of fuel per unit time per pound of thrust ‘The endurance equation can then be written % Wo % c= fv. Tw 1taw_ 1c, aw aaa we “We tT yet DW Wet Cp W Notice that airframe aerodynamics again appear through the L/D ratio. For simplifying assumptions similar those used earlier (constant c;, parabolic drag polar), maximum en- durance (Emax) Occurs at the velocity for maximum LID where Cp, = Cp, 4.1.4 Range Analogous to the definition of endurance, range is the distance an aircraft can fly on a given amount of fuel. Range, too, is affected by how the specific fuel consumption is ex- pressed for the type of powerplant. The distance term in the definition for range perform- ance can be either air distance or the distance covered over the ground. In the first case, the range is called air range and no wind effects are considered. 4.1.4.1 Propelier-Driven Aircraft. The integral expression for range in calm air (which is valid for general drag polars and for variations in np and c) comes from the defin- ition of true airspeed for steady, straight cruising flight. Using this definition and the rela- tionships described in section 4, i thrust horsepower, and aerodynamic forcss, % We = = f vee awe five Haw fA a4: R= fu nae fs aew™ [v Ww i w«C«*20) where R= “ange fora given fue lod, We-Wo8, Cruise Performance Tests Chapter 4 To maximize range, the integrand or at least me St. must be maximized. If np is con- 0 stant and the drag curve fits the usual polar form, Rinax Occurs with Cp, = Cp. 4.1.4.2, Jet-Powered Aircraft. The form of the range expression for jet-powered air- planes is affected in much the same way as endurance is by the definition of specific fuel ‘consumption. [ute Ww t R= [Vadt =<, but V, “ \pe5C, 4.21) Maximizing the integrand of eqn. 4.21 must include consideration of both an altitude effect and an aerodynamic term, even with c constant. The density term in the denomi- nator indicates why jet-powered vehicles fly at high altitude for their best range perform- ‘ance. An upper limit on the altitude for best range is not apparent because it depends on how component efficiencies of the engine are affected by altitude. When these efficien- cies become small, ¢; will dominate the integrand. If the usual simplifying assumptions are made, the aerodynamic term is maximized when Cp, = 3Cp, 4.2 CRUISE PERFORMANCE TEST METHODS Cruise performance tests determine the related performance parameters, endurance ‘and range. Endurance is time the airplane can spend in the air for a given fuel load and range is the total distance (usually measured in air miles) that can be flown on a given fuel oad. For both propeller-driven and jet-powered airplanes, both fuel consumption and drag Vs, where V, is the stall speed for the takeoff configuration. 5.1.1.9 Safe Single Engine Speeds. During takeoffs or landings with one engine inoperative, a multiengine airplane must be operated with some margin of safety. Some literature calls this speed the takeoff safety speed, Vos, but the most common terminol- ‘ogy in the United States is safe single engine speed, Vsse-Chapter 5 ‘Takeoff and Landing Flight Tests [Safe single engine speed (Vssz) is the airspeed below which a multiengine air [craft must not be operated after leaving ground effect with one engine inoperative. Vege must exceed both V; and Vm, by @ safe margin. While Vsse is most important during takeoff, prudence dictates that the airspeed not be reduced below Vgse during single engine landing until the landing is assured. With this terminology firmly established, we now turn our attention to basic physics. 5.1.2 Performance Equations Because of the large number of factors that affect TO&L performance, it is practically impossible to model it perfectly. Major influences include: Gross weight Thrust available Ambient temperature Pressure altitude Wind direction and velocity ‘Slope of the runway Coefficient of friction ooooooe ‘To make any rational analysis of performance measurements, we will have to make sev- eral simplifying assumptions. pn Thrust (Jet Engine)| Thrust (Propeller) [Net Accelerating Force (Jel Forces W at Liftoff Le Drag + Rolling Friction + Slope Term ——$ Drag. SG Fig. 5.5 Forces During Ground Run 5.1.2.1 Ground Run Equations. First, let us consider the distance while the airplane is strictly rolling on the ground during either a takeoff or a landing. The takeoff or landing98 ‘Takeoff and Landing Flight Tests. Chapter 5 flares are not a part of this analysis; the distance covered during rotation during takeoff will be approximated separately. Hence, we will call this distance sg, for the takeoff, differentiating it from sg (the total ground run distance to liftoff). The forces acting on an airplane during this first portion of the ground run of a takeoff or the last portion of a landing are sketched in Fig. 5.5. The contributing aerodynamic forces are defined in the usual sense: L and D, as shown, are conventional, as are T and W. The frictional force Fy is rather difficult to ascertain, but it is usually defined in terms of a friction coefficient x and the net normal force. That is, F,= L- W. Lan and Roskam2 give typical values for 1 as shown in Table 5.1. Of course, the slope of the runway also produces a small component of W(W sin © ~ WW) acting parallel to the takeoff surface, along with a small reduction in normal force. This latter component is usually neglected, since © is typically a very small angle (usually less than 3°). ‘The runway gradient, or slope, is positive if the takeoff is Uphill. So the net force accelerating the airplane parallel to the surface of the runway is F=ma=T-D-wW(W-L)-Wo (4) TABLE 5.1 Values for Runway Surface Friction Coefficients ‘Type of Surface Range for Concrete 0:02-0.030 Hard turf or short grass 0.05 Long grass 0.10 Soft ground 0.10-0.30 We also want to consider wind effects: let +V, = wind speed. Here, "+" implies a tail wind and "-" a head wind. So, at any point in the takeoff ground roll the ground speed is. d = =ViVy (6.2) 1 which we rearrange to at, and Of course, the acceleration is a=, dsg ave) S (5.3) Integrating eqn. 5.3 yields an exact expression for takeoff ground run distance, pro- vided we know a, V, and V, throughout the takeoff roll and we can accurately determine the true airspeed at liftoff. so- ft 64 Of course, we can not generally express each of these parameters analytically, but they can be measured by several means. If we solve the basic force equation (eqn. 5.1) for the acceleration and substitute it into eqn. 5.4, WV tVy)dV 80= \F7=p-ww-1)-we] 65)Chapter 5 ‘Takeoff and Landing Flight Tests 99 Equation 5.5 is the basic equation for ground roll distance for either takeoffs or tand- ings; it will be used in modified form for both purposes. We will use a second set of sub- scripts (",") to denote landing conditions. Notice that in eqn. 5.5 we have adhered the our sign convention that © > 0 for uphill takeoffs. Remember that the acceleration during the ground roll after landing is negative and that the thrust term can be either positive or negative, depending on whether or not the test vehicle has thrust reversing or not. ‘Though we have not included it, another similar term must be added if the airplane uses a deceleration device, like @ drag parachute or a speed brake, to reduce the ground roll. We choose to lump this added drag with the aerodynamic drag. Even more important, the braking friction coefficient (11g) is quite different than it was for takeoff because the brakes are applied at some point during this segment of the landing. Typically, rolling friction coefficients vary considerably. Notice that yp is a strong function of runway surface conditions (wet or icy surfaces) and whether or not the tires are skidding or not (Fig. 5.6). Modem aircraft often have antiskid brake units to avoid tire slippage during heavy braking. itis usually not feasible to test for these stopping distances under actual runway surface conditions like those shown in Fig. 5.6. It becomes the fight test engineer's job to identify and plan for simulating such conditions. Often this task is one of the more difficult ones faced by the fight test engineer. The approach that is now most common is to develop a mathematical mode! and then carry out a limited number of tests to validate this model. Naturally, one of the more difficutt facets of this kind of testing is measuring actual braking coefficients for other than dry runway surfaces. Sometimes, aerodynamic braking is 80 important that during the early stages of the rollout the recommended procedure is to maintain a high pitch attitude, applying brakes only after the nose is lowered. 08 Dry Concrete| ne Si Concrete Ho Ol Cah Rain} Heavy Fair) | O2 an [Smooth Clear Ice| 0.0 L oO 20 40 60 80 100 Percent Slippage Between Tire and Runway Fig. 8.6 Coefficients of Friction during Landing Ground Run Figure 5.7 illustrates this procedure and also shows how useful a drag chute is for some designs. While the landing ground roll distance equation is of the same form as the takeoff equation, the values of the terms may be significantly different; there are terms present that are not in the takeoff equation. These differences are indicated by the changed symbols and the apparent reversed order of integration (which accounts for the deceleration) in the following expression:100 ‘Takeoff and Landing Flight Tests. Chapter § ° wivtv)av Se, = f L(V #Vw)al (5.6) Ti, - Dy - np(W, - 4.) - Wo] Vj The T,, Hy, and the D, in this equation are the landing roll expressions for the correspond- ing quantities in eqn. 5.5; the subscripts are merely used to remind you that these pa- rameters are considerably different from those in the takeoff roll expression. g With Drag Chute 1600) ‘Aerodynamic Braking “t 1200 8 Deceleration Force in Pounds a 8 a 740100 60 20 Landing Roll True Airspeed in Knots Fig. 5.7 Effect of Aerodynamic Braking and/or Drag Chute Example 6.1: An airplane with a wing area of 230 square feet weighs 9900 pounds and the speed brakes. ‘are opened for landing. In this configuration it has a drag equation: Cp = 0.144 + 0.055C,?. The speed brakes remain extended down to approximately 75 ktas. Idle thrust ofthe engine at sea level, standard day conditions is constant at 700 pounds with the airspeed at or below final approach speed. Touchdown on the main wheels occurs at 100 knots. if the average rolling coefficient of friction on dry concrete is 0.03, ‘and the average braking coefficient of friction is 0.5, estimate the total landing distance from touchdown to ‘complete stop. Assume no wind. The nose wheel is lowered (instantaneously) to the runway at 75 knots, but the test pilot maintains a constant angle of attack after touchdown until he lowers the nose, giving C, = 1.25 during this phase of the landing. As soon as the nose wheel is on the runway, the wheel brakes are ‘applied and braking begins. With the airplane in this three-point attitude, Cy » 0.1. Viktas) Vifps)_——LiR(C,~1.25) AD —sintegrand iE. EAs, (feet) 100 198.889, 9745.88 1792.75 47.36 — = 95 (160.444 8795.65 16179 51.91 49.36 419.4 90 152.000 7894.16 1452.13 67.58 10437 = 881.42 85 143.556 7041.40 120526 64.86 «= 16559 © 1308.4 80 135.111 6237.36 1147.36 7461 -238.33 19872 75 126.667 5482.06 1008.42 88.39 316.83 ___—2675.4 ‘This problem breaks down into two parts: (1) from main wheel touchdown until the nose whwel is on the runway and braking begins and (2) from nose wheel touchdown until the airplane comes to rest. The first phase is dominated by aerodynamic braking. Wheel brakes provide most of the deceleration during the second part of the rollout. Equation 5.5 applies to the first phase and eqn. 5.6 to the second. At main wheel touchdown L = W. As the airplane brakes aerodynamically, eqn. 5.5 shows that the friction term be- ‘comes @ more important part of the retarding force. But the principal deceleration still comes from aerody- ramic drag. The aerodynamic drag decreases as V decreases, even though C, is held constant. Tabulat-Chapter 5, Takeoff and Landing Flight Tests. 101 ing these tif and drag changes at discrete speeds during the landing rol, in the table above we numerically ‘approximate the integration of eqn. 6.5, The table summarizes the results of this numerical integration and ‘estimates the ground roll prior to commencement of braking at 2675 feet. ‘The second phase of the ground roll is calculated using the same procedure, except that the terms in ‘eqn. 5.6 are changed to the appropriate values for Cand to pip (0.1 and 0.5, respectively) for this phase of the rollout. This calculation is illustrated below and shows that the ground roll distance after the nose wheel is lowered is only 584 feet. Deceleration distances (after the brakes re applied) are tabulated on pege 101. V(etas) Vhs) AL __intogrand EAs; (feet) 75 128.667 r 33.04 70 118.222 . 552.24 8.12 636 65 (109778 476.16 1877 1932 60 101.333 405.72 22 1936 55 92.889 30, “2987 249.7 50 84.444 281.75 35:70 301.4 45 78.000 228.22 41.28 348.8 40 (67.556 90.22 46.32 301.2 35 (S841 138.06 50.80 4200 30 80.867 101.43 $470 461.9 2 © 33.778 45.08 6074 5129 to = 16.880 7.80 14.27 6531 551.5 o 0.000 2.00 0.00___0.00___—-65.62 554.1 So, our approximation forthe total landing rol isthe sum of the two increments: (so, = 3230 feet 5.1.22 Rotation Distance. As was pointed out eartier, most takeoff distance pre- dictions for modem high performance aircraft should include an allowance for the distance covered during the takeoff fire or the rotation down from touchdown attitude to the attitude where all wheels are on the ground and brakes are applied. We will call this dis- tance either sg, (for takeoff) or 8, (for landing). Typically, the estimate is made by simply ‘assuming a reasonable average time to complete the pitch rotation. The time chosen will be different for each airplane considered, but 3 to 4 seconds is a reasonable ap- proximation for many types. The distance is assumed to be traversed at a constant ‘speed, either VioF oF Vip (again, for liftoff or landing, respectively). S6,='VLor Of 812=4VrD (6.7) where ¢, is the approximate average time used to cany out the takeoff of landing flare, respectively 5.1.2.3 Transition Distance. Part of the air distance estimate is based on rather arbitrary assumptions. After liftoff the aircraft must transition from an essentially level ‘acceleration along the runway surface to either a constant airspeed climb or a constant angle climb. For certification and for design purposes, the latter is usually taken to attain clearance height above any obstacles in the flight path near the ground. Later, another transition to a constant airspeed or a constant Mach number climb path is may when well clear of any ground-based obstacles. During the landing approach flare maneuver is used to transition from the ideal constant glide path angle descent to touchdown. We will ‘concem ourselves in this paragraph with how far the airplane travels in @ horizontal direction while making these transitions. To simplify the estimate (and this distance is highly approximate), we will assume that this transition path is a circular arc. To maintain102 ‘Takeoff and Landing Flight Tests. Chapter 5 2 a circular path, the velocity will be constant and the acceleration is ¥, where Re is the a radius of the circular transition path. Clearly, to fly through such a circular path requires an acceleration toward the center of the circular arc. The force producing this acceleration comes from additional lift generated by the lifting surfaces and this added lift can be measured as an increment in load factor given by: 2 2. Anu Y— = AGeV"S Re Ww The velocity in eqn. 5.8 (and in eqn 5.9 below) can be either Vio¢ or Vrp depending on whether we are dealing with a takeoff rotation or a landing flare. In either case we can rearrange eqn. 5.8 to calculate R,. v2 2w -- (5.9) gan gACipv?S 69) ‘Another variation uses the Re found from the total lift coefficient during transition (C;,.,,)- (68) wv2c, 2 t-te w+ WH where Vg is the stall speed for the appropriate con- V2 Chnax OR . figuration and C,,,,, is the corresponding maximum lift coefficient (Fig. 5.8). Then (6.10) 1.2 13 14 Vavg Vs Fig. 6.8 Empirical Values of AC; used in Equation 5.9 Of course, the estimate improves if the velocity in the calculation of R, is improved. One approach is to use an average speed during the takeoff transition or the landing flare‘Chapter 5. ‘Takeoff and Landing Flight Tests 103 in the equations above. This mean speed is essentially the average of Vio and the speed during the initial climb for a takeoff transition or of Vrp and the final approach speed for 8 landing flare. Lan and Roskam? point out that Williams used this approach, including the effects of AC;’s in the maneuvers. Figure 5.6 is an adaptation of their data. Knowing the radius of the circular arc, the transition distance (Fig. 5.9) is a simple matter of trigonometry. The horizontal distance covered during either transition is: Stran = Re|sin®cz| (6.11) ‘The absolute value accounts for negative flight path angles during a landing approach. Obstacle Clearance Stren So. Fig. 5.9 Transition Distance Geometry 5.1.2.4 Climbout Distance. The last part of the air distance phase of the takeoff or landing performance estimation is the distance covered during the initial climb before the standard obstacle distance is cleared or during the final approach after the obstacle is cleared but before the landing flare begins. Naturally, in some cases this part of the air distance may be zero. If the altitude at the end of the takeoff transition is greater than the obstacle clearance height or if the landing flare begins above the obstacle clearance height, then this distance has no meaning and is not included in the air distance. There- fore, we must also find the height at the end of the takeoff transition (or the height at the beginning of the fanding flare), Atran (Fig. 6.9). The geometry is again quite simple: Stran_ hen = Re(t-|c08 8¢z|) = —"="-7(1 -|eos @cx| (6.12) ron = Fal esde)= 28 (- rosea) If Fran > obstacle clearance height, then we are through with estimating the required distances.” If ay < obstacle clearance height, we again refer to the geometry sketch (Fig. 5.9) and observe that for an obstacle clearance height of 50 feet: pie (6.13) ScL= tan’ 7 Example 6.2: Consider the airplane of Example 5.1. It has a stall speed of 92 knots and it ifts off at 1.2V,. ‘Estimate the transition distance and the climbout distance to the climb path. Assume that the transition104, Takeoff and Landing Flight Tests Chapter 5 path is a circular are and Siem. = 018 curing tension tothe ci path ‘Also, assume that V = Viog and ae that this speed is constant during this part of the takeof. Of course, Cg, Cou at Ve. The thrust during this transition phase is constant at 7000 pounds. Then, Cig ey = 210000) =15. With this value of Cj nq, and also Usi ' Goma SSvF ~ o002aTER2 x1 6668807230) ma ee V= Vior = 1.2¥g during the transition, we obtain, either from eqn. 5.9 oF from Fig. 58, AC, = 0.2767, Calculating R from the other form of eqn 5:9: Ro= a 2 9000) °~ SaG,Sp ~ 010002376H(0.2767)(230)52. 17405) ‘As the note on Fig. .8 emphasizes, this value of the fight path radius is besed on an assumption that the piloting technique is aggressive; the rotation is called a “maximum effort rotation. Equation 5:8, on the ‘other hand does not require this assumption; f we calculate R without a maximum effort rotation, 2 ee 12192 1.608888); = 7109 feet. {1.20 = 0.12+ 0.0407 (0.8) -1]- 3217405 Notice that there is a wide range in the results ( @ change of about 75%) depending on whether or not the “maximum effort” assumption is used. This oversimplified calculation underscores just how sensitive take- ‘off and landing performance estimates are to piloting technique. rs Climb angle is set by the excess thrust available during the transition period; that is, sinéqy = ad with each of the variables evaluated during the transition phase. In the takeoff configuration the speed ‘brakes are retracted and the flaps are at their takeoff setting. So, Cp = 0.12 + 0.04CP.. Since the maximum usable C, = 1.50.8, Cp = 0.12 + 0.04(1.44)? = 0.1776. Thus, the average drag during transition is about = 7000 = 1688 = 4079 feet. 1686 pounds. With these intermediate estimates available, sin@c. =——So—5-—. This expression gives, Oc = 34.5". Now, we can finally estimate the transition distance, using a “maximum effort fare to minimize Re; that i, tran = Re = sin Gc. IStran = 2310feet| iran = Rel - 008 8c) = (4079 f(t - cos 34.5%) = 717 feet. Since hyan > 60 foot, sey = 0 feet] Arplanes with high thrust-o-weight ratios often reach or exceed obstacle clearance height before complet- ing the takeoff flare. Lower performance airplanes will require calculation of soy using eqn. 5.13, 5.1.2.5 Summary of Equations. Having developed the component expressions making up total takeoff or landing distance, we simply sum the component parts to com- plete our estimate and our discussion of the takeoff and landing performance analysis. S70 = 8g + Sg, + Stran * SL (6.14) SLND ™ SL + 86+ Sian SFA (6.15) 5.1.3 Parametric Analysis of Takeoff Performance ‘ShevellS advocates a different approach to takeoff performance that is simple and al- lows straightforward comparison of competing designs. With judicious assumptions, he generalizes takeoff performance for most commercial airliners so that takeoff and field lengths can be estimated from a few charts. The underlying premise of this parametric approach is that "...takeoff performance is basically an acceleration to the required speed Plus a climb segment to a 36-ft height (civil turbine-powered transports) or a 50-ft height (piston-powered, general aviation, or military aircraft).” Required runway length is definedChapter 5 Takeoff and Landing Flight Tests 105 a the distance from the start of the takeoff point to the point where these obstacle heights are reached. 5.1.3.1 Key Parametric Assumptions. In laying out this parametric approach, Shevell makes the following basic assumptions: @ Even though acceleration is not constant (as we have seen in preceding sections, it depends on thrust and drag and both are functions of speed), the effecive average excens thrust (7-0) occurs at 7 = “ABE, © Drag is small compared to thrust during the takeoff run. 5.1.3.2 Basic Parametric Equation. Considering only the first assumption above ‘and denoting the acceleration at V as 2, we can approximate ground run distance simply: via Sgn oe (6.16) FAR Part 25.1074 mandates that the minimum takeoff speed (V2,,,) cannot be less than 1.2V, for two-engine and three-engine turboprop and reciprocating engine-powered airplanes and for ail large turbojet transports that do not have provisions for significantly reducing the one-engine inoperative power-on stall speed. So, we choose this nominal value and let Viog =1.2V5 = 1.2, |_2W then eqn. 5.16 becomes: PSC Linas w2 PSCiipg, Tn - OV ‘The constant in eqn 5.17 depends solely on Moe. Under certain constraints, this /s value can be reduced to 1.15. Then, the constant in eqn §.17 becomes 1.3225 in stead of 1.44, In general, the distance for accelerating a body is directly proportional to the square of the speed and inversely proportional to the average acceleration. But V itself depends directly on wing loading W/S, inversely on C,,,., , and inversely on p. Acceleration, un- der the restriction of the second assumption on the previous page, is directly proportional to TW. Lumping all these terms together and expressing sg in functional form to allow for different constants and to correct for nonstandard density conditions: Sq =144 (6.17) 2 so" wx] 6.18) Equation §.18 is a generalized takeoff performance approximation similar to the gen- eralized performance curves introduced for turbojet-powered aircraft in Chapter 4. Shevell asserts, moreover, *...that if we plot takeoff distance for each airplane against the pa- w2 OSC Lpax V. rameter sg: ) pits wi form sine curve.” The scatter is small for108, ‘Takeoff and Landing Flight Tests. Chapter 5 similar airplanes and a single fairing gives a good approximation to the takeoff distance. Since the ground run is usually greater than the air distance plus the transition distance (by as much as 4 to 1), the total distance to the obstacle clearance height can also be ap- proximated by an expression like eqn. 5.18. Figure 5.10 (adapted from Shevell's data) approximates required runway lengths under FAR Part 25. It was obtained by plotting a large number of takeoffs for different turbojet and turbofan-powered commercial transports and fairing a single curve through the points. This curve applies only to takeoff field Jengths with all engines operating. FAR Part 25.113 requires a field length of 115% of the measured obstacle clearance distance; the faired curve includes this safety margin. e e i 7 ; Ss 10}--- includes 15% Increasel Z...| x (Par Part 25.113) #8, ag ze a ae, 3 : fat 0.707Viog z > ! 2 50 100 150 2 200 250 300 w Sh Fig. 5.10 Distance to 35-Foot Obstacle Height for Large Jet Transports {All Engines Operating) Similarly, Fig. 5.11 is a chart constructed from measured takeoff field lengths, again based on FAR Part 25 requirements, for the same aircraft as Fig. 5.10 but with one engine inoperative. Again, the required 115% margin for uncertainty required by the certification regulation is applied in this chart. For large commercial transports powered by reciprocating engines the takeoff field lengths required can be approximated from similar parametric charts, However, the ac- cceleration is directly proportional to average installed horsepower, rather than average in- stalled thrust (notice that both terms are installed values, not thrust or power available from engine manufacturer's charts). Then, eqn. 5.18 is modified to: (5.19) Also, the obstacle height in the regulations is 50 feet for propelier-driven airplanes, rather than the 35 feet used for commercial jet transports. The required field length, underChapter 5: ‘Takeoff and Landing Flight Tests 107 the assumptions listed at the beginning of this section, collapses to a single curve. Figure §.12 illustrates such a curve for a DC-6. 3 Required Takeoff Field Length x 1000 ft (Based on FAR 25) = o x 50 300 osc, Tv Fig. 5.11. Distance to 35-Foot Obstacle Height for Large Jet Transports (One Engine inoperative) 10 Includes 15% Increase] 2 © - Required Takeoff Field Length/1000 ft (Based on FAR 25) Fig. 5.12 Distance to 50-Foot Obstacle Height for Four-Engine Propeller Transports (One Engine inoperative) ‘Turboprop takeoff performance may be estimated from charts like Fig. 5.12 if total in- stalled power is known or from curves like Figs. 5.10 and 5.11 if the total installed thrust is108, Takeoff and Landing Flight Tests, Chapter 5 known. In every case the form of the expressions is the same; only the experimental measurements (which change the slope and curvature of the generalized curve) differ. ‘Typically, this approximation is more difficult to apply to such hybrid powerplants. Finally, Shevell strongly underscores the value of such a parametric estimation proc- ess. He says, "...the parametric method can be used to check the reasonableness of de- tailed calculations of takeoff (performance).” This basic awareness is essential to a fight test engineer engaged in planning a series of takeoff and landing tests, From such rea- sonable approximations come those elusive qualities called "intuition" and “engineering Judgment’. They are acquired skills, not innate intelligence! 5.2 FLIGHT TEST METHODS The nature of takeoff and landing measurements leaves much to be judgment of the individual fight test team; there is no well-defined “standard” for making these measure- ments as there is for pitot-static calibrations, climb performance, or cruise performance. This statement does not imply that the requirements documents for military full scale de- velopments or for certification under FARs are imprecise or vague. Rather, these docu- ments are often so specific that the test team must tailor a takeoff and landing test to the individual test program. Consequently, this section is more general and can offer only guidelines. Fortunately, the theory and "analytical" equations presented in section 5.1 are quite pragmatic; so, tailoring to the specific requirements of a given fiight test program is, usually straightforward. 5.2.1 Requirements for Takeoff and Landing Flight Tests ‘Takeoff and landing tests are made to provide data for the operator during these criti ccal phases of fight. Apart from special tests conducted with a specific research goal in mind (for example, changes in stability derivatives in ground effect), both military and civil Tequirements are heavily slanted toward collecting information that goes into the Airplane Flight Manual (AFM). These fiight tests are a combination of performance and handling qualities assessments. As such, they involve measurement of performance parameters of interest. But they also usually involve some degree of subjective evaluation as to how the airplane's stability and controllability affect the pilot's ability to precisely maneuver the sir- plane during these terminal phases of flight. Partly because of the dual nature of such evaluations, certification requirements and military specifications spell out a very compli- cated matrix of test conditions. This matrix includes not only normal operating conditions, but also a large number of abnormal or emergency procedures in order to "demonstrate" that each standard is met. Ordinarily, this phrasing means that a flight test demonstration is required. Specifically, Part 23.21 states: (a) Each requirement of this subpart must be met at each eppropriate combi- nation of weight and center of gravity within the range of loading conditions for which certification is requested. This must be shown — (1) By tests upon an airplane of the type for which certification is re- quested, or by calculations based on, and equal in accuracy to, the results of testing; and (2) By systematic investigation of each probable combination of weight and center of gravity inferred from combinations investigated.Chapter 5 “Takeoff and Landing Flight Tests 109 () The following general tolerances are allowed during flight testing. However, greater tolerances may be allowed in particular tests: ee ____ tem. Tolerance. Weight +5%, -10% Critical items affected by weight +5%, -1% Center of gravity +7% of total travel While each of the subparts of FAR Part 23 listed below is applicable for takeoff and landing demonstrations and should be carefully studied by the test team, those subparts marked with an asterisk pertain directly to these flight tests. Most of the other compliance requirements can be extracted from tests that these subsections dictate. Since this book introduces the subject and does not attempt to comprehensively cover all aspects, the dis- cussion is limited to these most important subparts. Part 23.33 Propeller speed and pitch limits Part 23.153. Control during landings” Part 23.49 Stalling Speed Part 23.185 Elevator control force in Tekeot™ Part 23.51 maneuvers: Pat23.75 Landing Part 23.157 Rate of roll Part 23.143 Controllability and maneuverability: Part 23.161 Trim general Part 23.171 Stability: general Part 23.145 Longitudinal control Part 23.173 Static longitudinal stability Pert 23.147 Directional and lateral control —-—Part 23.175 Demonstration of longitudinal Part 23.149 Minimum control speed stabllty Part 23.177 Directional and lateral stability Part 23.235 Ground and water handling Part 23.179 _ instrumented stick force maneuvers. ‘characteristics: Taxiing condi- Part 23.181 Dynamic longitudinal stability tions Part 23.201 Wings level stall Part 23,239 Ground and water handling Part 23.207 Stall warning characteristics: Spray condi- tions. Pert 23.231 Ground and water handling characteristics: Part 23.251 Vibration and buffeting Part 23.233 Ground and water' 5.2.1.1 Use of Mathematical Models snd Simulation. One of the more significant recent trends in evaluating takeoff and landing performance is a shift from overall meas- urements (which are rather imprecise mainly due to variations in piloting technique) to meesurement of segments of these maneuvers and use of modeling or mathematical simulation to tie segments together. The primary objective of most TO&L measurements is to determine and demonstrate for the user of the airplane how to achieve reasonable takeoff and landing distances and safe speeds for everyday operations. Flight tests are typically used to verify individual portions of the mathematical model rather than attempting to measure the performance on complete maneuvers. These “patched” maneuvers (or maneuver segments) can be compared and standardized more easily than can measure- ments from complete takeoffs or landings. Both military and civilian flight test engineers ‘are permitted to use either approach. Of course, complete takeoffs and landings, with careful attention to consistent piloting techniques, are still demonstrated. But the modeling approach allows thoughtful choice of valid maneuvers and better control of test variables.110, ‘Takeoff and Landing Flight Tests, Chapter 5 In practice the test team should plan to collect data on almost every takeoff and/or landing until a rather complete data base is established for normal procedures. Even though repeatability in technique is emphasized, there will be wide variation in results. But, individual segments where nearly identical rotation rates and usable airspeeds are matched can be collected from several different runs and statistically evaluated for con- sistency before the complete maneuvers are flown for validation of the modeled takeoff or landing. Such mathematical simulations are particularly useful in establishing takeoff ac- celeration profiles and in predicting landing rollout under slippery runway conditions. The segmented approach allows the test engineer to use portions a of a maneuver to establish performance, even though the entire maneuver may have flaws in it that make the overall distances and velocities invalid. From this perspective, segmented testing is more effi- ient; it allows almost every maneuver flown to contribute to the data base and thereby adds statistical relevance. However, complete maneuvers must also be done as final validation of the takeoff or landing model. As one FAA document® puts it: ",,,actual takeofis using the AFM (recom- mended technique) should be conducted to verify that the actual distance to the 50-foot height does not exceed the calculated takeoff distance to the 60-foot height.” Complete takeoffs and landings, with attention to consistent piloting techniques, must still be demon- strated: but not many of them. 5.2.1.2 Position and Velocity Measurement Techniques. The instrumentation used to measure distances and velocities varies widely in precision, in complexity, and in versatility. The most useful measurement tools allow simultaneous capture of both hori- zontal and vertical position and velocity during either the takeoff or the landing. Probably the most widely used equipment for measuring ground roll distances is some form of tracking device — laser, radar, optical, or video. In each case the flight test engi- neer is responsible for choosing the type of tracker and for guaranteeing (1) that its preci- sion is adequate to satisfy the certifying agency, (2) that any special transponders or illu- minating devices operate reliably, and (3) that the data reduction effort required to convert the spatial measurements into meaningful engineering units is compatible with program budget and schedules. With regard to (2), the test engineer should be alert for any “unde- sirable side effects" result from the transponder or illuminator. A laser tracker often needs a reflective surface painted on the airplane and the laser beam is intended to point at this surface only. The flight test engineer will have to satisfy safety review authorities that a misdirected laser beam cannot harm (even temporarily) the pilot and/or the airplane and its ‘systems. A radar tracker may also need a radar reflector or a transponder to retum accu- rate spatial information; the test engineer must be sure that such an electronic device does not interfere with other aircraft or test instrumentation subsystems. Optical devices need clearly discernible levels of contrast so that data reduction does not become overly tedious. The same statement is also true of video imaging systems, which do lend them- selves to at least partially automated data reduction”. At least for the ground roll portions of takeoff and landing trajectories, distances can also be measured by integrating accelerations and then velocities from inertial navigation ‘systems (INS) if the airplane is so equipped. Cheney and Pham® describe how such an approach [along with instrument landing system (ILS) and radio altimeter measurements} was used to validate and certify Category Ill autoland performance for the MD-80 and the MD-87 commercial airliners. This indirect form of distance measurement is accurate pro-Chapter 5 ‘Takeoff and Landing Flight Tests 114 vided (1) the INS is not strongly affected by acceleration errors during takeoff and (2) the ‘events in question (liftoff, touchdown, and passage through obstacle clearance height can be precisely defined on the data records. This latter point means that the test planner must thoroughly evaluate techniques to define these events. Sometimes the flight test engineer may have to design additional equipment or improvise procedures to define the events critical to his measurement data. Landing gear microswitches and photoelectric beams are two such specialized devices that have been used for this purpose. Often the preparation for a full-blown set of takeoff and landing tests must include qualification and proof testing of these kinds of specialized instrumentation before useful data can be col- lected. The wise fight test planner includes time in his schedule for such proof tests. ‘These secondary measurements are crucial since the fight test engineer must correct his data for nonstandard conditions. The growing use of segmented, mathematical simu- lations to reduce test time and cost underscores the necessity for these data in making appropriate corrections to measurements. All of these atmospheric variables affect pro- pulsive thrust and/or directly affect ground speed as measured by tracking equipment. ‘Takeoff and landing measures of merit are drastically affected if this atmospheric informa- tion is missing; however, it is often necessary to use expensive instrumented runways (available at all major test facilities). As always, the cost of having such instrumentation is ‘often prohibitive to small or underfunded development efforts. Of course, spatial measurements (tracking data, whether radar-derived, optical, or otherwise), made repeatedly and rapidly against an accurate time base, are a satisfactory way to estimate velocities only if atmospheric conditions are recorded at the same time. Consequently, most major test ranges have runways that are heavily instrumented with atmospheric sensors. As a minimum, TO8L flight test data used for certification or proof of performance, require the following atmospheric instruments®: (1) an altimeter or @ sensitive barometer set to 29.92 inches of mercury to measure altitude; (2) one or more anemometers that measure wind velocity and direction at the takeoff or landing surface (It is also useful is some means of measuring wind velocity and direction up through the obstacle height is available.); (9) an accurate thermometer to provide free air temperature, again at the take- off or landing surface; and (4) when appropriate (as when tests are conducted in high temperature, high ‘humidity conditions), relative humidity at the takeoff or landing surface. 5.2.2 Typical Part 23 Test Groups 5.2.2.1 Takeoff Tests. In laying out a series of takeoff tests, most of the data is col- lected during other tests; after all, every successful flight must begin with a takeoff and end with a landing. However, every test plan should include a matrix of test conditions that covers specific TO&L test maneuvers. The test engineer can then simply cross off } the appropriate data as it is collected, analyzed, and validated. For a typical light twin to be certified under FAR Part 23, the takeoff test matrix groups that must be expanded into detailed test cards for individual flights might look like those in Table 5.2 on the following page.112 Takeoff and Landing Flight Tests. Chapter 5 Table 5.2 Takeoff Conditions for a Light Twin Certification ‘Test Description Conditions Primary Purpose __ References Determination of V, Normal takeoff Selecttechniques Part 23.51, AC-233% Demonstrate takeoff per- At nominal weights®, og Proof of compliance, Part 23.51, formance?, including locations’, flap settings, data for AFM AC-23-0 ground roll,flare distance, and atmospheric vari- distance to clear obstacle, _ables® ‘speed at obstacie height Demonstrate minimum con- Specified weights®, cg Proof of compliance, Part 23.51, trol speeds? locations’, sidestip data for AFM AC-23-1x conditions, and bank angles Demonstrate required roll At nominal weights, cg Proof of compliance, Part 23.51, responsiveness in the focations*, and atmos- data for AFM AC-23-1% takeoff configuration? heric variables® Part 23.157 Notes: 1. Stall speeds must be determined before this matrix is begun. 2. Emphasis is on achieving a speed so that all maneuvers can be performed safely in the ‘event of a power failure in reasonably turbulent conditions. 3. Takeoff distance tests are conducted at maximum weight and at a range of og locations. 4. Takeoff distance demonstrations are performed at the most critical og location (usually the most forward ca. ‘5. Humidity is measured and accounted for (Part 23.45) and wind velocity and direction is ‘measured adjacent to the takeoff surface. Wind velocities should be as low as possible, not to exceed 1.2Vp, oF 10 knots 6. The airplane is presumed to have the desired maneuvering capability at the obstacle clearance height if the speed attained at this point is the higher of 1.1Vinc, 1.3Vs,, OF ‘some speed > 1.3Ve, down to Vx + 4 knots (for multiengine airplanes). Since Ve has ‘no meaning for single airplanes, the frst speed requirement at the obstacle height does ‘not apply. 7. Minimum control speed tests, obviously apply only to multiengine airplanes and are not takeoff tests in the strictest sense. The test team must determine the “critical” engine ‘and must investigate all flap settings recommended for use, Dynamic or static Vine, Whichever is greater, wll be included in the AFM. 8. Demonstrations should be accomplished by rolling the airplane in both directions. 5.2.2.2 Landing Tests. Landing tests are also outlined to help organize the test matrix during the test planning phase. The primary difference between them and takeoff tests lies in the demonstration of adequate longitudinal control authority for the landing flare maneuver. Table 5.3 summarizes a typical grouping of such tests, again for a small twin-engine airplane. Again, the detailed test planning involves laying out detailed test cards that cover all items in this matrix. The test planner must also anticipate consider- able variation in the landing data, since the performance during this maneuver is quite un- certain due to a number of factors including: (1) atmospheric turbulence is common for landing, (2) pilot technique varies widely for making a "good landing, and (3) ground effect changes the handling qualities for each airplane and pilot combination. For ail these rea- sons, it is imperative to measure a statistically significant number of landings carried out by a representative sample of pilots to obtain quality landing data.Chapter 5, Table 5.3 Landing Tests for a Light Twin Certification ‘Test Description Conditions Primary Purpose References Determination of Approach Landing at all proposed Select techniques Part 23.75, ‘and landing speeds? flap settings, og loca- AC-23-0« tions’, and power Demonstrate landing per- _ Atnominal weights, cg Proof of compl Part 23.75, formance?, including ground locations, fiap set- ance, data for AC-23.« rol, flare distance, distance —_tings, and atmos- AFM to dear obstacle, speed at —_pheric variables obstacle height Demonstrate longitudinal ‘Specified weights?,cg Proof of compl: Part 23.153, ‘control during landing® locations* ance, data for AC-23-xx AFM Demonstrate required roll re At nominal weights®, og Proof of compi- Part 23.187, sponsiveness in the landing locations, and at ance, data for AC-23-x« configuration”: mospheric variabiesS = AFM Notes: 1 Stall speeds must be determined before this matrix is begun. 2 Emphasis is on achieving a steady speed at the obstacle height with the power set so the ‘irplane passes through this reference point in stabilized conditions. A smooth flare to touchdown should be made. Normal pilot reaction times should be used for power re- duction, brake application, and actuating other deceleration devices. At least six landings on the same wheels, tres, and brakes should be demonstrated to ensure serviceability. Landing distance tests are conducted at maximum allowable landing weight and at a range of lesser weights expected in normal service. Landing tests ere performed at the “critica” (usually the most forward) cg location and the ‘most aft og location certified, as well es a range of other og locations expected in normal service. Humidity is measured and accounted for (Part 23.45) and wind velocity and direction is measured adjacent to the takeoff surface. Wind velocities should be as low as possible, ot to exceed 1.2Vq, oF 10 knots. ‘The primary purpose of this demonstration is to ensure that airplanes over 6000 pounds ‘9r08s weight have sufficient flare capability to overcome any excessive sink rate thet may Gevelop at a speed 5 knots lower than recommended normal approach speed and to en- ‘ture that control forces are not excessive. If the airplane is to be certified at approach ‘speeds < 1.3Ve,, compliance with Part 23.153 must be demonstrated at the selected ap- ‘proach speed. Demonstrations should be accomplished by roling in both directions. Takeoff and Landing Flight Tests 13114, Takeoff and Landing Flight Tests Chapter 5 5.3 SUMMARY This chapter introduces methods of estimating takeoff and landing distances and ‘speeds critical to attainment of this performance. We review both piecewise estimation techniques and parametric computations. Test methods for these flight tests are neither as clearly defined nor as precise as for some performance measurements; pilot technique plays an important role. Although it is usually easy to gather large amounts of takeoff and landing data, it is difficult to obtain repeatable data for complete maneuvers. It is common practice to select consistent segments of each of several similar maneuvers and use them to verify mathematical models of takeoff and landing for a given configuration. This ap- proach is a good example of how simulation augments experimental measurements and makes flight testing more efficient. Finally, general test groups for a simple airplane are ‘suggested that outline detailed test matrices for certification compliance demonstrations. REFERENCES 1 Dekker, F. E. D. and Lean, D., "Takeoff and Landing Performance: Chapter 8, Volume 1," AGARD Flight Test Manual, Pergamon Press, New York, 1959. 2 Lan, C. E. and Roskam, J., Airplane Aerodynamics and Performance, Roskam Aviation and Engineering Corporation; Ottawa, Kansas, 1980. 3 Shevell, R. S., Fundamentals of Flight, Prentice Hall, Englewood Cliffs, New Jersey, 1983, Federal Aviation Regulation, FAR 25, Federal Aviation Administration, Govemment Printing Office, Washington, Mar. 30, 1967. Federal Aviation Regulation, FAR 23, Federal Aviation Administration, Govemment Printing Office, Washington, Mar. 30, 1967. Flight Test Guide for Certification of Small Airplanes, Advisory Circular 23-xx, Federal Aviation Administration, Goverment Printing Office, Washington, Nov. 14, 1983. Dorsett, K., Robson, R., Pollard, S., and Albright, R., and Ward, D. T., "Takeoff and Landing Measurements with a Video Imaging Technique,” Unpublished report, Texas A&M University, 1988. Cheney, H. K. and Pham, C. T., "A New Method to Confirm Category II Autoland Per- formance,” AIAA Paper 88-2126, Aug. 1988.15 Chapter 6 LONGITUDINAL STABILITY TESTS Longitudinal stability is usually the first subject introduced when one begins a study of ‘the stability and controllability of airplanes. For relatively mild maneuvers and perturba- tions, this subject is a convenient introductory topic since only one moment equation is involved and the complications of coupled motions can be avoided. The development of flight test methods has generally followed the same pattem. It is also convenient to start the discussion with the fundamental definitions and assumptions for static stability before going on to the dynamics of the problem. To evaluate the relative merit of any airplane's flying qualities, it is essential that all parties accept a yardstick. For years the flight test community struggled with standards (and on occasion still does!), but in time both the civil regulatory bodies and the military produced requirements documents that spell out what is expected in any class of aircraft. ‘Van Pelt! gives some of this history to go with his examples? of such requirements. The student should recognize that these requirements change frequently and may be modified by specific contractual documents. Each flight test team must carefully examine the specifications that are applicable to the design being evaluated before they design the test matrix needed to adequately explore its flying qualities. 6.1 THEORETICAL FOUNDATIONS To lay the groundwork for longitudinal stability test methods, we first briefly review the meaning of static and dynamic stability and the basic equations that gover such motions. To evaluate airplane stability rationally, the test team must be keenly aware that flying qualities are important primarily as they affect the pilot. Therefore, in this book how the pilot reacts fo an engineering modification will be a most important consideration. 6.1.1 Definitions To understand longitudinal stability, the concept of equilibrium must be clearly under- stood. The most basic definition of equilibrium for any body simply requires constant linear momentum and constant angular momentum. Thus, a rigid body of constant mass must either be at rest or in unaccelerated motion to be in an equilibrium state. For an airplane, oth the forces and the moments about the center of gravity must be balanced. An airplane is in equilibrium or immed when the sum of the extemal forces is zero and the sum of the moments about the center of is zero. ‘That is, the equilibrium state for an airplane is one of unaccelerated linear transiation. Aircraft static longitudinal stability is typically defined in terms of its initial tendency to retum to equilibrium after a disturbance. Dynamic longitudinal stability, on the other hand, considers behavior about the pitch axis ss a function of time after a perturbation. Disturbances can take many forms, but the one most often of interest is an angle of attack change. (Many older texts use C; rather than o. as the independent variable; however, as Etkin’ points out, there are cogent reasons for using Crp, 8s the measure of static116 Longitudinal Stability Tests, Chapter 6 longitudinal stability.) Perturbations in true airspeed often are also important to flying qual- ities investigations but they are studied less often than are disturbances in a. ‘An airplane is said to be statically stable in pitch if a disturbance in angle of attack produces an initial pitching moment that tends to restore the airplane to its trim angle of attack. This definition is primarily of use to an engineer; the pilot senses static longitudinal stability in an entirely different way. He senses static longitudinal stability through the forces he must apply to move the elevator and the displacement of the control stick or column available to him in the cockpit. He perceives positive static stability as an aft stick movement or @ pull force on the stick as airspeed is reduced from the trim airspeed. Similarly, a push force or a forward movement of the stick is expected as speed increases (ais reduced) from the trim condition. Both the control movement and the contro! forces indicate to the pilot whether or not the airplane is statically stable. The fact that there are two ways to sense static longitudinal stability indicates that there are two different types of stability of concem to the flight test team. Each is important for different reasons. First, stick-fixed stability is a measure of the free response of the airplane being tested. The stick or control wheel is held stationary by the pilot so that the control sur- faces will not move with changes in the aerodynamic forces on the surfaces. ‘Stick-fixed static longitudinal stability is positive when the pilot's longitudinal controller must be deflected aft for a greater than oy The converse, of course, must be true for a < cm For conventional tail-aft airplanes, this definition means that the elevator must move trailing edge up (TEU) when the speed is reduced below the trim speed and trailing edge down (TED) when the speed is in- creased above the trim speed. Stick-fixed static longitudinal stability is fundamentally im- portant for all other kinds of stability. It applies to airplanes with both reversible and irre- versible flight control systems. When the elevator is allowed to float freely (zero stick force applied by the pilot), a dif- ferent form, as a different level, of static longitudinal stability can be evaluated. The pilot's force that must be applied to trim the airplane is @ key issue and becomes a primary ‘measure of this form of stability. Stick-free static longitudinal stability is positive when the force applied to the pilot's longitudinal controller is an aft (pull) force for a greater than aj With these basic concepts reviewed, we now tum our attention to quantifying them But to do so, we must first establish sign conventions for our analysis. 6.1.1.1 Sign Conventions. Figure 6.1 shows the sign conventions commonly used in the literature and in this book. The positive coordinate directions x, y, z and the right- hand rule define positive force, angular velocity, and moment vectors as illustrated. The most common (though by no means universal) sign conventions for control surface de- flections are shown. While most stability and control textbooks use the “standard” convention (Fig. 6a) for contro! surface deflections, we will use the flight convention shown in Fig. 6b. In former convention the observer is assumed to be sighting along the positiveee Chapter6____ Lon gituctinal Stability Tests 17 z axis and using the right hand rule to define positive angular deflections of the rudder, that is, left rudder deflection is positive. Positive elavator deflection is obtained by sighting along the positive y direction and again applying the right hand rule. This approach gives {8 positive deflection of the elevator surface when it is moved trailing edge down (TED). Aileron movement is defined to positive when the right aileron is moved trailing edge up (TEU). Similarly, if @ "rolling tail” is used for lateral control, TEU for the right half of the stabilizer is considered positive. On the other hand, the fight test community typically uses a less positive moment about the respective axis. ZWZ (a) “Standard” Convention (b) Flight Test Convention Fig. 6.1 Sign Conventions Fig. 6.2 Definitions of a and B 6.1.1.2 Angle of Attack and Sidesiip. To analyze the longitudinal handling qualities of an airplane with respect to disturbances in angle of attack (a), a must be precisely de- fined. Figure 6.2 defines a. and 8 (sideslip angle) which will be used to evaluate lateral-di- rectional handling qualities. Looking at longitudinal stability first, we note that a is defined by vector components of true airspeed lying entirely in the vehicle's plane of symmetry. 7 W] U lo = tan’ 6.1)118 Longitudinal Stability Tests, Chapter 6 Similarly, even though this definition will not be used in this chapter, asin? ¥ p= sin (6.2) Of course, Veo = VU? + V2 + W? 6.1.1.3 Inertial Orientation. Euler attitude angles are often used to specify the orien- tation of the body-fixed stability axes relative to an inertial coordinate system located at the center of the earth. These inertial angles must be measured in a specified order to fit the usual aircraft conventions: 1) a yaw rotation (‘¥) about the inertial Z axis; 2) a pitch rotation (@) about the position of the y body axis after the first rotation; and 3) a roll rota- tion (©) about the position of the x body axis after the first two rotations. These inertial at- titude angles are shown in Fig. 6.3. Earth Plane Horizon z y Fig. 6.3 Definitions of Euler Angles Itis important to realize that the Euler attitude angles, ¥, ©, and @ are not vectors, nor can they be thought of as vector components since the order of rotation spells out the final position. However, the angular velocity of the aircraft is a vector and can be expressed in either inertial components [> © ¥]7 or body axis components [p qr)’. 6.1.1.4 Angular Velocity Transformations. Figure 6.4 shows the projections of ‘7 on each of the body axis coordinate directions. First, is a rotation about the positive z-axis. if there is no pitch or roll rotation, r; = Y. If the aircraft is then pitched nose up, as shown in Fig. 6.4b, r2 = ‘¥.cos @. If the airplane is banked through an angle ©, r= ¥.c0s @cos ©. Similarly, after three rotations, p = -‘¥ sin © and q = Ycos sin ©. Next, resolving 6. in the same way, q= .cos ® and r= [email protected] ®. Resolving the roll rate (@) into body axis components, p = . Collecting the individual vector components, and waiting the resulting linear equations in matrix form: fume) -sino \(o =|0 cos cos@sin® || 6 (6.3) 0 -si ¥Chapter 6 Longitudinal Stabitity Tests, 119 inverting the matrix to solve for the Euler angular velocities: ©) (1 sin@tand tan®cosd)(p’ 0 cos = -sin® iq (64) r c. © Rotation Fig. 6.4 Body Axis and Inertial Angular Velocity Components 6.1.2 Stralaht Flight Paths Logically, the next terms to be defined are those used to specify static longitudinal stability quantitatively. From a fight test point of view, the discussion always starts with an ‘engineering concept and proceeds to the pragmatism of how a pilot recognizes and de- termines the degree of static longitudinal stability for the airplane under consideration. 6.1.2.1 Neutral Point Concepts. The stick-fixed and stick-free neutral points are terms used to relate center of gravity (cg) location to the degree of stability of the configu- ration. Quantitative equations for both neutral points will be developed later, but to fix the concept, consider the following definition. “The stick-fixed neutral point is that og position for which the pitching momer of «with the longitudinal control surfaces held stationary. ‘Again, much of the flight test literature’.5.6 defines neutral point using C,_ as the indepen- dent variable. As suggested earlier, neutral points defined in this way and associated with zero values of 222. are not always directly related to siabity with respect to angle of L attack disturbances. Both Cj and C; are functions of Mach number, Reynolds number, thrust coefficient, and dynamic pressure. Even with the longitudinal control surface and120 Longitudinal Stability Tests, Chapter 6 the og position held fixed, both the derivatives of interest depend on the parameters listed in addition to a. Consequently, setting 22m. oa = Oto obtain the neutral point is only valid if the other variations are unimportant ae disturbances in angle of attack. For low speed airplanes, however, this assumption is valid. To sum up, the neutral point is that og position for which Crm, = 0. The floating characteristics of the longitudinal control surfaces also affect the stick-free Neutral point definition. (Of course, "stick-free” conditions may be ignored for irreversible control systems since the control surfaces do not float with aerodynamic loads.) independent of a with the longitudinal control surfaces unrestrained. Comparing these definitions to the aerodynamic center, the neutral point is conceptually identical to the aerodynamic center of the entire airplane. Recall that the total pitching moment coefficient for the airplane can be written as: Cm = Cm +Cm,o where Cm, pitching moment coefficient about the og when lit is 0. Cm, can also be written as: Cig = Cima, +VHCL, (it +£0) where Cm4, pitching moment coefficient for the wing-body combination (n0 horizontal tail) at zero wing-body lift Va horizontal tail volume coefficient, Set Ciyy ‘tli curve slope SEL. evaluated at tim conations it tall-setting or — £9 downwash angle when cee littiso Etkin has shown that Keg _ ae, Cla (y_ 28 : Cm ee ine (2 Ma ae +Cz, (it +€0) where Xo 9 position ce mean aerodynamic chord Xecyp __erodynamic center or neutral point of the wing-body alone Differentiating with respect to a (while also assuming 0: = cys) Xog | Xac, Gy, a ones (FF }weet-8) Setting Cm, = 0 and solving for Xoq gives the stick-fixed neutral point::‘Chapter 6 Longitudinal Stability Tests. 121 0p. *eeu0. 4 ee(-2) 65) te MON ee : here Vi, = Value of Vi forthe 09 located at Xpp, The diference ~2°. £2. is called the stick-fixed static margin and it is a measure of the static stability with respect to angie of attack disturbances. ; a 8.(7F- = C1, (-StaticM argin) (6.8) 6.1.2.2 Elevator Trim Angle The point was made earlier that pilots sense static longitudinal stability of an airplane through the control movements and forces, mainly the forces. The pilot's feel for the neutral point also comes from elevator movement per unit change in angle of attack. Said another way, the change in elevator angle required to trim the airplane is the pilot's index to stick-fixed static margin. Consider how longitudinal con- trol surface deflections affect the pitching moment curves, as sketched in Fig. 6.5. For each elevator setting, equilibrium conditions can be maintained at only one angle of attack. For example, with 8g = 0, the trim angle of attack occurs at point A. Increments of lift and pitching moment are usually assumed to vary linearly with small changes in a; the Cpa curves depicted are not strictly straight lines, which is common behavior. The pertinent include: AC, = C,,5e, with C.., > 0 forthe sign convention specified in Fig. 61 and AC = Cma8o: ‘with Ge. < 0, uding the same sign conventions. Then, we assume linearty in both C, = Cat C1480 and C= Co + Cm, + CngS—- Setting Gmn= 0 in this last expression gives the eiSvator angle for trim, (6.7) Fig. 6.5 Family of Ca Curves for Various 8 ‘The trim lift coefficient is: Crim = Cetin + CLSerim122 Longitudinal Stability Tests, Chapter 6. Eliminating apm from eqn. 6.7: = F0bg + Sng Chim (68) ‘eri Cina Pts ~ CL Omsy In equation 6.8 we have assumed that 5, is measured with respect to a trail positi that is, 59 = O when ot = ori. ‘The above expressions are valid approximations for con- ventional tail-aft airplanes. However, for tailless aircraft or other unconventional configura- tions, the equations must be developed assuming that Cm, also varies with 5. Fig. 6.6 Dependence of S94, 0n og Position Equation 6.8 ideally assumes a straight line variation of 89,,,, with angle of attack (Fig. 6.6). This relationship between 3eniq, and cg location (or static margin) suggests a way to determine the stick-fixed neutral point from flight test data. Since the slopes of these curves are clearly functions of cg location, we differentiate equation 6.8 with respect (69) Fig. 6.7. Trim-Slope Criterion for Determining the Stick-Fixed Neutral Point Figure 6.7 illustrates the use of eqn. 6.9 and depicts the so-called “trim-siope" criterion. Notice that the derivative was taken with respect t0 Ciyiq, NOt a. This change of independent variable means that this equation is strictly true only if Mach number effects and speed effects are negligible. Etkin4 emphasizes this fact and lays out the rationale quite clearly. Nonetheless, this criterion is a common approach for estimating the stick- fixed neutral point for low speed airplanes and serves well to introduce flight test meas- urement of the neutral point.Chapter 6 Longtitudinal Stability Tests, 123 6.1.2.3 Elevator-Free Considerations. In section 6.1.2.1 the concept of a stick-tree neutral point was introduced. We must now detail the description of static longitudinal stability theory. Even with a perfectly rigid body assumed for the airframe, a manual con- trol system will result in some freedom of movement for the control surfaces. Cable ‘stretch and pilot inputs alone dictate at least some movement. An imeversible contro! system closely approximates the stick-fixed assumption. But a manual control system de- mands that the flight test engineer understand the effects of aerodynamic hinge moments on the longitudinal stability of the sirplane. Generally, configurations are less stable when the control surfaces are free to rotate under the influence of these hinge moments. The hinge moments are the result of the force distribution on the control surface. Since friction ie alvays present in a control system, the actual stability perceived by the pilot lies some- place between the two extremes of the stick-fixed and the stick-free analyses. Fig. 6.8 Control Surface Floating Angie Consider the force distribution illustrated in Fig. 6.8. The surface hinge moment can be written as a linearized function of the local angle of attack at the surface (a,), the con- trol surface deflection angle (6), the tab setting (8jgp), and a term describing the mass bal- ance of the surface. (The subscript j is generic; it can indicate any control surface at- tached to the trailing edge of a ling panel - elevator, aileron, or rudder.) Wyx; Ch, =Cny +Cr,, 25 +Cn,,8) + Ch, Btab ase (6.10) where Wy weight of the jth surface (can stand for », a OF » depending (on which control surface is being considered) x cg position of the jth surface relative to the hinge line (If the surface is perfectly mass balanced, x)= 0. gq freestream dynamic pressure (to distinguish it from pitch rate) Cho hinge moment coefficient at a= 5 = 0 Chay rate of change of hinge moment coefficient with local angle of stack, vated ai coins (22 FH trim Che, rate of change of hinge moment coefficient with surface de- flection, evaluated at trim conditions (2 4 Jim Coby rate of change of hinge moment coefficient with trim tab de- flection, evaluated at trim conditions (2) tab Jim124 Longitudinal Stability Tests, Chapter 6 The leading term in equation 6.10, Chg, depends upon several geometric factors like airfoil shape, nose bluntness, and gap seal effectiveness. Oversimplifying for the sake of clarity, we will assume a symmetrical airfoil, a surface with perfect mass balance, and a fixed trim tab. Then, with the surface floating freely so C;, = 0, equation 6.10 gives: Chafty + CngBj= OOF, 8, =-— (6.11) ime Gp, | Both Ch, and ©; usually negative, although Cp, may be positive when a large hom balance’is desigted into the surface or when the hinge line is set far back from the surface's leading edge. Taking the usual case with both Cp, and Cp, negative, a positive local angle of attack at the surface produces: G1, =C aj + Ch, 8) =O, aj) 1-——L— tree 4 Cr, CL, acy, Ss where ay (S. and C1, = 0) If we now focus on the longitudinal axis and define the free elevator factor as: G ac, : Shag 0 (S and C,,, =%9 St on, O., ori OS Then the coefficient of lift developed by the horizontal tail is y= FC. (6.12) If F< 4, a freely floating surface is less effective than a fixed one. Of course, if F> 1 (as would be the case if Cp, were positive), then the surface produces more lift when free than when fixed. Denoting’stability coefficients with primes when the elevator is free to float under the influence of the hinge moment, the aircraft lift coefficient is: Fet- , where a G'= Oa" o where 961, (FSH 3) ind a= ayy —F—Hleg +i) The aircraft pitching moment coefficient with the elevator free is: Gm’ = Cg! Cg 2” a ipand Cl, =c;,,(%2.-*00 where Cio’ = Cmag + FC Vig (60 * i) and Geil" 2] Solving eqn. 6.5 for the stick-free neutral point,Chapter 6 Longtitudinal Stability Tests_ 125 ete oe) + CL, S(,_ 66) - Xap _ Xe, » CL, i oe ee ea (613) ee MG Sl ma)" ee Cu where Vij,’ = tail volume coefficient when the og is at the stick-free neutral point. Vit = tail volume coefficient when the og is at the wing-body aero- dynamic center. Ifthe variation of tail volume coefficient wth og position can be neglected, } G Xap _*00 (4 F Min oS -2) (6.14) Chae s aa, Etkin4 gives exact expressions for the difference between the two types of neutral points, but this approximate expression will suffice for our purposes. ‘Analogous to the trim-slope criterion outlined in section 6.1.2.2, the trim tab angle re- quired to produce various free floating angles gives an estimate of stick-free neutral point. If 8jap i$ not zero, equation 6.11 becomes: Pe Shag Bt + Chg Std Cn, Solving for the tab angle necessary to trim and using eqn. 6.7, Ch, Ch, Btabvm “ae{em +Cm,a-Cmy, an (6.15) "tab My (6.16) C ay -f1-2)}-o wif er Using this expression to eliminate oy from eqn. 6.15, Ch (( Cm, Sh, ae Cm, Stab, =| | ee (1-2) +a +60 +i tabeim ~ Gp, (é Sm, ea) PCy, oi Differentiating with respect to a: ‘Substituting expressions for Cn,,, Crs, and F and rearranging: Ptadon Sno, | Xan _ X00 (17 oa. Chey c Equation 6.17 is exact only when Vy = Vy". Notice that the negative of the paren- thetical term is exactly like the stick-fixed static’ margin defined earlier, except that now it126, Longitudinal Stability Tests, Chapter 6 should be called the stick-free static margin. Obviously, eqn. 6.17 indicates that the slope of the elevator trim tab required to trim longitudinally with changing angle of attack is @ measure of stick-free static longitudinal stability. Recalling the difference between Viy and Vj, and using the approximate difference between the stick-fixed and the stick-free neutral Points (eqn. 6.14), eqn. 6.17 quantifies a “tab-slope” criterion for obtaining the lo- cation of the stick-free neutral point from measured variations of trim tab versus angle of attack. Figure 6.9 illustrates this approach. 28 tad een aa Xeg = Xip Xeq Fig. 6.9 Tab-Slope Criterion for Stick-Free Neutral Point 6.1.3 Other Concepts of Static Stability Currently, military specifications? do not directly use neutral points in spelling out re- quirements for piloted airplanes. The authors of such standards apparently fee! that these concepts are no longer appropriate for high performance aircraft that often operate with re- laxed static stability and very complex automatic control systems. Whether one agrees with this approach or not, it is important that the flight test engineer be prepared to work in this environment by understanding the terminology. 6.1.3.1 Speed Stability. The basic stability criterion set down by the military specifi- cation is that of speed stability. Paragraph 3.2.1 of the primary document? states: there shall be no tendency for airspeed to diverge aperiodically when the air- Plane is disturbed from trim with the cockpit controls fixed and with them free. This requirement will be considered satisfied if the variations of pitch control force and pitch control positions are smooth and the local gradients stable... he bhp Sonim Fig. 6.10 Stick-Fixed Speed Stability The specification goes on to say that the speed stability criterion must be met in level flight at constant altitude with no change in throttle setting or trim movements over a range of #50 knots or 415% about the trimmed equivalent airspeed. Figure 6.10 illustrates aChapter 6 Longtitudinal Stability Tests, 127 stable airplane with respect to stick-fixed speed stability; the elevator trim position to maintain equilibrium meets the requirement. Etkin4 has succinctly related speed stability to the neutral point for those cases where compressibility, aeroelasticity, and propulsive effects are negligible. Figure 6.10 also summarizes this concept. SinC® Benim 18 @ unique function of C, with these restrictions and since C,,,, = , the trim elevator angle is @ unique function of Veas also. As poveS speed increases, 5g,,, increases monotonically for a stable airplane. Hf Xog = Xpp the re- quired trim setting is the same for all equivalent airspeeds. Hence, the eirplane is neutrally stable, both in the speed sense and under the neutral point concept. However, the speed stability concept is more general in that the trim setting can also be considered a function of propulsive effects, Mach number, and dynamic pressure, as well as angle of attack. Equation 6.8 can be rearranged to give Since each of these stability coefficients varies with Mach number, for example, then also varies with Mach number. A sketch of typical variations with Mach number is shown in Fig. 6.11. Cty) Cte Cm, my Cing 1 2.0 0.4; 6.0 — 0.8 | 1.6 C + 06 41.2 C., 0.2} 4.0 0.4 408 |e Sma 0.2 40.4 0.0! 2.0 0.0 | 0.0 0.0 1.0 2.0 Mach Number. Fig. 6.11 Variation of Longitudinal Stability Coefficients with Mach Number In transonic flight Mach number dependence demands modification of the concept of longitudinal speed stability as a measure of static longitudinal stability. MIL-F-8785C re- laxes the requirement for an airplane that has a trim curve like that shown in Fig. 6.12. This airplane is not speed stable in the transonic region. Physically, the airplane may start ‘ut in equilibrium at point A. If the speed is perturbed to point B by a gust or some other B disturbance, ae will be negative even though speed is increasing. Thus, the ordinary Cm, Criterion for static longitudinal stability has no meaning in this case.128 Longitudinal Stability Tests Chapter Fig. 6.12 Transonic Effects on Static Longitudinal Stability Hy > 0 Fig. 6.13 Control System Schematic 6.1.3.2 Force Gradient for Speed Stability. Even though most of the airplanes covered by MIL-F-8785C have irreversible flight control systems, the document is more concemed with force the pilot must apply than in the elevator travel itself. At least one proposed revision” recommends omitting the constraint on a stable surface displacement. ‘Once again, the pilot's perception of stability is the most important issue. Consider the control system schematic sketched in Fig. 6.13, noting that a positive stick force is a pull force (in the aft direction). A positive stick force results in a negative elevator defiection, though it produces a positive pitching moment. The gearing ratio (or gain) is a constant relating stick force and hinge moment: Fs= GHe. Applying the principle of virtual work, FeAS + HeASg= 0Longtitudinal Stability Tests. 129 Writing the stick force in terms of hinge moment coefficient, Fy =GCp,SoeoF ‘Assuming that Vj, = constant and that Ch =0; Chg = Chat + Crs Setim * Choy Stad- | we can also expect C.,, to be small, eqn. 6.8 reduces to: equation 6.16, Fg =A+BG (6.18) Cru, (Xo ¥h where A=-GSece a so Xe Ms, Chg Stab C, C, a B=GS.06Cn,Cry, (Aa ge-, ori (-e2(-2)] Equation 6.18 leads to the following conclusions: Stick force is directly proportional to the product of the elevator area and the elevator chord, that is, the size of the elevator. + Stick force is directly proportional to the gearing ratio. ¢ Retrimming (altering 8.,, ) changes the shape of the parabola as indicated by the dashed lines in Fig. 6.14. Positive 5, reduces the force to be held by the pilot at a given speed. ¢ The weight atlects stick force through the wing oading term “2 end an in- crease in wing loading has the same effect as a forward shift in og. ¢ Altitude affects only the second term through 7. For a given true airspeed, in- creasing altitude reduces F,. Naturally, the second term in the stick force ex- pression varies directly with the square of the true airspeed. ‘¢ Shifting the og location only affects the constant term (A) of equation 6.18. Moving the og forward translates the curve upward and moving it aft translates the curve downward. Figure 6.14 also illustrates this translation.Longitudinal Stability Tests Chapter 6 Fs eG a. Effect of Tab Deflection on Stick Force Gradient Fe b. Effect of cg Location on Stick Force Gradient Fig. 6.14 Stick Force Variation with Speed The slope of the stick force curve is a very important handling qualities parameter. The pilot judges loss of longitudinal stability by the rate of change of stick force with air- ‘speed change. Hoh’ clearly spells out the recommended maximum allowable size of any unstable gradients in the transonic regime. Equation 6.18 leads to an expression for the stick force gradient. Taking the derivative with respect to Vig, ee BpVz, oF a Cisgy Stab Cm, fg Cher St(y_ oe SS = pVipGSeCeCn, Ch, | 2 —- — - Cp, (eg + ip} 1-H 1 - SS Ve een Cn Sm, Ci, oa. At the trim point (where F, = 0), the stick force gradient is: Fe. «26505 te W( Xc0_* |. ve35405 SM ( SMyge) On) Vn "Tm, Slee eo, S aoe) J 6.1.3.3 Flight-Path Stability. Paragraph 3.2.1.3 of MIL-F-8785C describes yet another type of longitudinal stability related to "backside-of-the-power-curve” operations:Longitudinal Stability Tests, 131 ‘Chapter 6 Flight-path stability is defined in terms of the change in flight-path angle when the airspeed iltered by use of 7) Moin” 5 Va, (kts) y | NEGATIVE ® | stores FOR LEVEL 1 gen 5 Fre < 0.08 %t de) /de) (2), (Sz) scm Fig. 6.18 Local Slope Definitions for Flight-Path Stability ‘This type of stability is especially important for precise control of airspeed, as is es- ‘sential for carrier approaches or other short-field landings. If the elevator alone is used to control airepeed, it can be shown? that a closed (pilot-in-the-loop) loop instability can occur at speeds below the minimum drag speed. The requirement is quantified in terms of the local siopes for each level of handling qualities. For level 1, the local slopes must be ei- ther negative or less positive than 0.06 “/knot. Figure 6.15 illustrates the allowable local slopes. Compensation for poor fiight-path stability can be made if a secondary fight path controller, like power or direct lft control, is incorporated into the design. Short takeoff and landing (STOL) aircraft often utilize this means of compensating for poor flight-path stabil- ity since their landing approaches are frequently made at speeds well below the minimum drag speed. Naturally, the secondary flight path controller itself must have satisfactory re- sponse characteristics if tis to have acceptable closed loop handling qualities. 6.2 LONGITUDINAL STATIC STABILITY TEST METHODS. ‘The commonly used flight test techniques to obtain longitudinal stability data are rele- tively simple and straightforward, although very careful trimming and fiying must be done, To obtain reliable quantified data, it practically imperative that the instrumentation in- clude some form of automatic recording and the sensors used are almost always some kind of electronic device. The net effect of both these factors is that stability and control data are harder to obtain than are performance data. The so-called stabilized method and the slow acceleration-deceleration method are the techniques most often used to gather information about the speed stability or neutral points. These methods will be discussed first, followed by a brief discussion of the way that fight path stability tests are conducted. However, before the data procedures are detailed, it is necessary to examine two factors that govem the quality of the data collected whatever the method used. 6.2.1 Stability and Control Flight Test Measurements Stability and control testing is an exacting process. As suggested previously, virtually all interestina oarameters must be measured indirectiv and the control surfaces and con-132 Longitudinal Stability Tests, Chapter 6 trol tabs are the most common measurements used to back out the desired information. Forces are often measured directly and they are variables of primary interest, but even 80, the data must be further manipulated to obtain engineering terms of interest like neutral points and force gradients. For these reasons, data must be taken with as much precision as possible and it is particularly important that the fight test engineer understand the un- certainties introduced by the mechanical parts of the control system. 6.2.1.1 Trim. The first of these uncertainties is introduced because most longitudinal static stability measurements are taken about an equilibrium condition, The requirement quoted in section 6.1.3.1 measures static stability from such a trim condition. Trimming to an absolute equilibrium condition is not trivial, but flight test engineers must insist that the flight crew establish an accurate trim condition. A good rule of thumb is to require the pilot to release all controls for a minimum of ten seconds after he has established the trim air- speed and altitude. If the airplane deviates at all within that period of time, trim should be reset and the check repeated. This process should be repeated until no deviation from the desired conditions can be detected during the specified time interval. To adequately cover the flight envelope, it will be necessary to set up separate trim conditions at convenient airspeed intervals as illustrated in Fig. 6.16. The exact incre- ments of airspeed to be used will depend upon the data to be collected and the regime of flight. Smaller increments between trim conditions should always be used where surface deflection gradients and/or stick force gradients are expected to change rapidly. The tran- sonic region is certainly such a region for those airplanes designed to operate there. h INSTABILITY ALLOWED IN TRANSONIC REGION, ve Fig. 6.16 Trim Conditions for a Specified Altitude 6.2.1.2 Friction and Breakout Forces. The second uncertainty affecting the preci- sion of stability and control measurements is the mechanical friction of the mechanisms through which the pilot controls the machine. For the pilot to have a feel for the neutral Positions of the controls, there is usually an initial force to be overcome in moving the contro! out of the neutral position. While this kind of force serves a useful purpose in de- fining the neutral position for the pilot, the control stick will also require some force to ‘overcome the initial resistance to motion at any point in the control throw. This type of force is lumped under the term "breakout force."Chapter 6 Breakout force is defined as that stick force which must be applied to the control system before the control surface begins to move. Itis customary to call the remaining frictional forces simply friction. Longitudinal Stability Tests, 133 Friction is that stick force required to overcome sliding, rolling, or any other form of dynamic frictional forces in the control linkages. SPEED |PUSH <2 Fig. 6.17 Control System Friction Band Figure 6.17 illustrates how friction and breakout forces appear if measured carefully at various airspeeds near an equilibrium condition. ‘Though friction and breakout have been defined in terms of forces, they both also affect the movement of the pilot's controller. Be- ‘cause of friction and breakout there is no guarantee that resetting the pilot's controller to exactly the same position will set the control surface to the same setting. If either the stick force of the control surface gradients are very small, friction and breakout can introduce large percentage uncertainties into longitudinal stability data. The fight test team must plan and execute the tests in a manner that minimizes these uncertainties to the maximum ‘extent possible. The separation of the two curves resembles the hysteresis loop common in almost any physical system when precise measurements are made. The circles indi- cate data taken when the pilot was moving the controller aft and the squares represent ‘stick movement in the opposite direction. If care had not been taken to keep the controller moving in the same direction at all times, the data points would be scattered between the ‘two curves. 6.2.2 Stabilized Method ‘One flight test method for obtaining longitudinal static stabilty is to fly a series of stabi- lized airspeeds above and below the trim airspeed and measure the appropriate control parameters. As in all the techniques, the trim conditions must be established carefully and care must exercised to insure that the stick force, contro! surface, and control tab meas- urements are made with the surfaces on one side or the other of the friction band for any points to be joined es a continuous curve. The stabilized method is based on use of the134 Longitudinal Stability Tests, Chapter 6 longitudinal control to vary airspeed at constant power setting and most of all at constant trim setting. Once the trim conditions are established and recorded, the longitudinal con- trol alone is used to slow down or speed up to predetermined airspeed points. Altitude is ‘obviously not constant, but should be controlled within a specified band, usually no more than 41000 feet from the trim condition. For military requirements, this variation of air- speed about the trim condition should be set to cover a range of +15% or +50 KEAS, whichever is less. A good rule of thumb is to plan for at least three points (preferably more) at speeds equally spaced above and below the trim airspeed. As the speed is sta- bilized at each of these airspeeds, care must be exercised to have the stick on one or the other sides of the hysteresis loop sketched in Fig. 6.17. Generally, itis easiest for the pilot to accomplish this rather delicate task by attempting to keep the control stick moving for- ward as speed is increased and aft as the speed is decreased. Example 6.1: A light twin general aviation airplane was tested for static longitudinal stability at two og lo- cations. The stabilized technique was used, resutting in the data give in the table below. ‘Stabilized Static Longitudinal Stability Data S=11.0 feet Vy = 0.42 Find the stickfixed static margin when the og is 1.98 feet aft ofthe datum and the elevator angle required for vim wth C, = 0.63, Calculate C., forthe horizontal Cy = 0.04. What i the poston of the stick- fixed neutral point? Plotting the data given in the table above: CL A least squares curve it of these two lines gives slopes of -5.14889 end -0.29423. Notice that the in- tercepts of these two lines will never be identical, simply because no measurements are perfect. A simple ‘average is usually a reasonable approximation. Clearly, it would be preferable to have measurements at ‘more than two og locations. The desirability of having multiple measurements is illustrated in the plot below of these two straight lines. This plot, using the slopes of the two lines versus Xeq (utilizing our assumed linear relationship), illustrates how the neutral point is obtained. Notice that these fictitious data indicate the airplane was flown very near @ neutrally stable condition, which is very unlikely! Instead, the fight test ‘engineer usually must extrapolate more than is shown here to find the neutral point. The need for this ex- ‘rapolation makes it imperative that data be collected for several cq locations, rather than just two.Chapter 6 Longitudinal Stability Tests. 135 {A straight line is obtained by connecting these two slopes on a plot of elevator angle with change in C, {could also be elevator angle change with angle of attack) as shown above. Solving this expression when Be _ 9 gi . 2¢, 70 sives us the nevtral point locaton: AB 0.2625 or xnp = 2.777 feet aff of datum 100 00 os to” 15 20 Xoq_~foet aft of datum 25 30 For the third location, wo ealauate 5°. = 8.60100 from our straight line average of the intercopis for L the tabulated data on the previous page. The average intercept ofthe two plotted lines is 0.80067, giving 5, = - 8.60109C, + 0,80067 for the given Xa With C, = 0.63: [50 =-7.80" and SM. = 0.0725 ing 200. 00 “The stick-xed static margin ie calculated using —P - —2- € 1.24679 - 0.180. Recognizing that the in- G, tercept in the above calculation should be a reasonable approximation for -7—*- , we get my Cm, 0-208 [m,, = -0.0499) ™ ”""0.80067 Me 6.2.3 Slow Acceleration/Deceleration Method This test method is perhaps the one most commonly used to collect static longitudinal stability data because it is very efficient. A large amount of data can be collected in a short time and it is particularly useful for large airspeed envelopes. It is not quite as accu- tate 2s the stabilized method, since absolute equilibrium is never attained. Generally though, if the acceleration/deceleration rates are limited to no more than 2 knots/second, the data is entirely adequate for most purposes. ‘An automatic recording scheme is abso- lutely necessary, preferably one from which the data can be acquired and formatted di- rectly into data files for further reduction on a computer. ‘Asis the case for all stability and control fight tests, the first step in performing a slow ‘acceleration/deceleration is to trim carefully. The importance of this step cannot be over- emphasized; the airplane must be trimmed precisely at this equilibrium condition because this point is the only one where a true equilibrium is attained in this method. The pilot or196 Longitudinal Stability Tests Chapter 6 test engineer should activate the data recording system to automatically record the trim conditions, Figure 6.18 shows how data from a slow acceleration/deceleration should look after reduction and plotting. Leave the data recorder running and mark the point where a ‘smooth acceleration or deceleration is begun. Apply smooth control inputs throughout the acceleration/deceleration. Power (or thrust) for level flight should be noted during the trim shot. Use as small a change as possible from this trim power setting to accomplish the acceleration or deceleration. About 5% change in power usually avoids large changes in pitching moment with power. The longitudinal trim should not be changed throughout the maneuver. Continue the acceleration or deceleration to the appropriate end of the air- speed range and then reverse the sense of the rate of change of the longitudinal control force. It is important that this reversal occur only at the extremes of the airspeed range; otherwise, the hysteresis effects pointed out in paragraph 6.2.1.2 will inject considerable uncertainty in the data. Continue the acceleration or deceleration to the other end of the airspeed range and again reverse the force rate of change. The maneuver should be completed by coming back to the original trim condition and again recording the trim shot. The advantages of using an automated data formatting and reduction scheme should be ‘obvious from the number of data points shown for this one maneuver. Fe A Les PULL LBS “8T pusH Fig. 6.18 Slow Acceleration/ Deceleration Data 6.2.4 Power Acceleration/Deceleration Method This technique is identical to the slow acceleration/deceleration method except that larger power changes (up to maximum power or back to idle) are used to accelerate and decelerate and attitude is maintained approximately constant. This procedure can only be used with configurations for which the change in pitching moment with power setting is negligible. An automatic recording device is again essential. Like the previous method, equilibrium is attained only at the initial trim condition and the rate of acceleration or decel- eration is constantly changing. As usual, start with a careful trim shot. Smoothly set the throttle, allowing the recorder to run continuously. Mark the minimum speed and the top of the speed range. Reset the throttle and reverse the direction of the force change and continue the maneuver to theChapter 6 Longitudinal Stability Tests. 137 ‘opposite end of the speed range. It may be desirable to shut off the recording device while making these changes outside the data range and then restarting the recorder be- fore the data band. The same care must be taken to avoid reversing contro! forces, as mentioned before. Upon reaching the other end of the speed range, reset the power and reverse the rate of change of stick force until reaching the equil point. Constant pressure altitude should be maintained as nearly as possible by fying a slight climb or de- ‘scent appropriate for the position error correction of the test airplane. 10 5 Vy +5 +10 +16 (knots) rin Fig. 6.19 Flight-Path Stability Limits 6.2.5 Flight-Path Stability Method The technique to measure flight-path stability (applicable to power approach configura- tions) is a specialization of the stabilized method. Figure 6.19 is a sketch of allowable lim- its. After setting gear and flaps, the airplane is trimmed in a normal glide path for a power- (on approach st Von, the normal final approach speed. The airspeeds for this condition over a range of weights are calculated by the flight test engineer during preflight planning along with anticipated rates of descent for a 3° glide path. Trimming must be precise, ‘even though the configuration and airspeed combination make trimming difficult. The trim condition should be set up at a predetermined nominal altitude. Once the configuration and the rate of descent are selected for the test weight, the pilot climbs approximately 2000 feet above the nominal altitude and starts the maneuver by setting the power to give the desired rate of descent with a tolerance of no more than +100 fpm. Altitude, airspeed, and rate of descent are recorded as soon as they stabilize. The nose of the aircraft is then raised to smoothly reduce the airspeed about 5 knots. Altitude, airspeed, and rate of de- ‘scent are again recorded with airspeed is stabilized 5 knots below Von, Airspeed control within 1/2 knot is essential for good data. This procedure is repeated fOr Von, ~ 10: Vomin +5, and for Von, again. At least four points are required to define the slope at Voj4, aNd Vomin~ 5 Knots for MIL-F-8785C. These data must be obtained as quickly as possible to minimize altitude loss; more than 4000 feet of change in altitude leads to inaccurate thrust corrections. Repeating Vonj, 88 the last point provides a reference for correcting even hand-recorded data. Example 6.2: The data in the table below were collected on a standard day to measure the flight path sta- Dilty of a jet trainer. Does the airplane meet the level 1 criteria of MIL-F-8785C? Careful plot the curve ‘and show the slopes used to make this determination. Vo, = 148 knots (TAS) forthe test weight.138 Longitudinal Stablilty Tests. Chapter 6 Flight Path Stability Example Data Calibrated Airspeed (knots) Pressure Altitude (feet) 130 41,000 128 10,700 124 10,400 120 10,200 136 9,550 130, 9,100 Flight Path Stability Calculated Performance Calibrated Air- [ Pressure Al- | Density | True Air- | Corrected Rate | Flight Path An- speed (knots) | titude (feet) | Ratio _| speed (knots) | of Descent (fpm) | gle (degrees) 130 11,000 | o7tss | 153.68 1000.0 3.6816 128 10,700 | 0.7224 | 148.25 1001.2 3.9610 124 10,400 | 07292} 145.21 963.2 3.7531 120 10,200 | 07338 | 140.08 884.2 3.5713 136 9550 | 0.7490 | 157.15 902.6 3.2404 130 9100 | o7se6 | 149.16 1000.0 3.7934 ‘The first step is to find the density ratio for each of the pressure altitudes at which data were hand re- ‘corded. using eqn. 2.72, we can calculate density ratios for each of the pressure altitudes listed above in the deta table. Then, the equivalent airspeed could be calculeted using the expression thet generated Fig. 2.4 (see page 13). In fact Fig. 2.4 shows clearly that Vp ~ Vow (OF, at least, AVoy < 1.0) for the attitudes where the data were taken. Using this approximation, the true repeed becomes Vn = Ee» “a. Hav. le” Jo ing this true airspeed estate allows 400 calculation of the fightpath ange from the indicated vertical velocity siny = Xe, where both volostion must bein the same units of course, Moreo- | ver, the vertical velocities recorded from the vorical veloaty indicator must be. & corrected for changes in engine thrust ‘over the 1900 feet of atte spanned in the descents, Tat is, \, (1000? a ‘Ten, he definion of fight path angle (given above) is used to calculate the 3004+ r 1 = flight path angle for each of the test 0 Ms 180 155 180 points, Notice that each of these angles Tiue Aispeed — knots {is negative, which is consistent with the descent that is typical for both @ normal landing approach and the ‘way the data were taken. The results of the calculations described above are summarized in the table on the previous page. Finally ploting the calculated fight path angles versus rue eirspeed end fiting @ second-order curve to the data, we obtain the plot above. ‘As was discussed in section 6.1.2.2, the slope at point 1 must be less than 0.06 “Amot and the differ- ‘ence between the slopes at points 1 and 2 must be less than or equal to 0.05 “Aknot to satisfy Level 1 han- dling qualities for fight path stability. Taking the second-order polynomial least squares curve fit shown above as a reasonable farina for the verv few measurements taken, the equation for vis: = Wing EOChapter 6 Longitudinal Stability Tests, 139 -y=-0.00562206V? +1,65626V -118.127 “Tking the derivative wih respect to V to get the required slopes, -21--~0.01124572V +1.65626, he slope at the recommended irue airspeed for final approach (146 knots) is: 0.014385 “knot, which meets the frat pat ofthe requirement The slope at Vo,,, 6 knots (141 knots) is: 0.076135 Mknot. Subtracting we find the necessary diflerence in slopes is 0.05623 “/knot, which does not meet the requirement. 6.3 SUMMARY This chapter introduced stability and control flight test techniques used to evaluate the static longitudinal stability of an airplane. The theoretical concepts of stick-fixed and stick- free neutral points were reviewed and this background will be built upon in succeeding chapters. The importance of the pilot's perception of engineering measurements was em- phasized throughout. How elevator control surface deflections and trim tab deflections can be utilized as a measure of stick-fixed and stick-free static longitudinal stability was explained in some detail. Speed stability, defined in MIL-F-8785C, was related to the neu- tral point concept for airplanes not affected by compressibility or Mach number effects. Fiight-path stability was introduced as a specialized, but important type of stability, to be measured. The importance of trim techniques, friction and breakout forces, and piloting precision for stability and control measurements was emphasized. Finally, the stabilized, the slow acceleration/deceleration, and the power acceleration/deceleration methods were introduced. With this basic introduction to stabilty flight testing, the student should be ready to proceed to more difficult tasks.140 Longitudinal Stability Tests Chapter 6 REFERENCES 1 Perkins, C. D., “Introduction,” Chapter 1, Volume Il, AGARD Flight Test Manual, Per- gamon Press, New York, 1959. 2 “Military Specification, Flying Qualities of Piloted Airplanes", MIL-F-8785C, November 1980. 3 “Federal Air Regulations, Volume Ill, Part 23, Airworthiness Standards: Normal, Uti- ity, and Acrobatic Category Airplanes, Federal Aviation Administration, Washington, 1969. 4 Etkin, B., Dynamics of Fliaht-Stability and Control, (2nd Edition), John Wiley & Sons, New York, 1982. 5 “Stability and Control Flight Test Theory,” Vol. 1, Chapter 3, AFFTC-TIH-77-1, USAF Test Pilot School, Edwards AFB, California, Revised February 197. 8 “Fixed Wing Stability and Control: Theory and Flight Test Techniques," USNTPS- FTM-No. 103, Naval Air Test Center, Patuxent River, Maryland, January 1975 (Re- vised November 1981). 7 Hoh, R. H., et al, "Proposed MIL Standard and Handbook — Flying Qualities of Air Ve- hicles," Vol. I: Proposed MIL Handbook, AFWAL-TR-82-3081, Air Force Wright Aero- nautical Laboratories, Wright-Patterson AFB, Ohio, November 1982.141 Chapter 7 LONGITUDINAL MANEUVERABILITY TESTS In Chapter 6 we discussed longitudinal static stability strictly in terms of unaccelerated, ‘equilibrium conditions. Now we need to consider how to test an airplane under conditions in which the flight path is curved by either banking the wings or performing a wings-level pull-up or pushover. In either of these cases the vehicle's center of gravity will be acceler- ated by the force imbalance. Obviously, static longitudinal maneuvering tests are related to tuming performance tests, but the purposes are entirely different and the techniques are quite different. The concepts of stick-fixed and stick-free maneuver points, which are very much kin to stick-fixed and stick-free neutral points, are reviewed and fiight test tech- niques for determining maneuvering characteristics are discussed. 7.1 THEORETICAL FOUNDATIONS Longitudinal maneuvering stability is concemed with disturbances in angle of attack ‘and load factor, but not speed. In other words, the subject matter in this chapter will con- centrate on load factor as the parameter of primary interest and ignore speed changes in contrast to Chapter 6 where load factor was always one. Parallel to the development of stick-fixed neutral point concepts with respect to elevator deflection, stick-fixed maneuver- ing stability concepts will be based on elevator angle deflections. Similarly, the stick-free maneuver point will be studied through the changes in stick force experienced by the pilot. These ideas lead to the following figures of merit: @ Elevator angle per g (stick-fixed maneuvering) © Stick force per g (stick free maneuvering). 9 To begin our study of accelerated flight, let us consider one of the simplest cases, a wings-level pull-up. 8Cm = Cmg Aa +Cmg, S50 + Sn Aq (7.1) “tam w (@) Straight and Level Trim Condition (©) Wings _Level Pullup sw) (C= nw) Fig. 7.1 Differences Between Level, Unaccelerated Flight and a ‘Wings-Level Pull-up142 Longitudinal Maneuverability Tests, Chapter 7 7.1.1 Steady, Wings-Level Pull-up Consider an airplane that has been trimmed for level flight at a given true airspeed. ‘Assume that it is climbed and then dived without altering the power setting (changing both the altitude and airspeed) and then a pull-up is initiated using the elevator alone. Instan- taneously, the airplane will pass through the level flight trimmed (equilibrium) condition during the maneuver and if care is taken this condition can be attained at the original alti- tude. As suggested in Fig. 7.1, both angle of attack and load factor will be greater in the accelerated maneuver than in level flight. Lift coefficient will increase and there will be a positive pitch rate generated. These changes in angle of attack and in pitch rate cause a change in pitching moment coefficient compared to the original level flight equilibrium Hinge moments for the control surface are also affected; that is, both C, and Cp, are functions of a, 8, and q. Notice that q is pitch rate, not dynamic pressure. To avoid'con- fusion, dynamic pressure is denoted by 7 in this book. Considering pitch rate an instantaneous source of centrifugal acceleration (that is, at any instant of time, the fight path can be approximated by a circular arc) and using New- ton's second law relates pitch rate to normal load factor and true airspeed as follows: quate (7-2) Von This pitch rate contributes to the total pitching moment experienced by the airplane by ef- fectively increasing the angle of attack at the horizontal tail or the canard. (For simplicity, ‘we will consider only conventional tail-aft configurations in this introductory course.) Fig- Ure 7.2 illustrates how the angle of attack at the horizontal tail is affected by the airplane's pitch rate. a TL stosconta tei Vv P ir Velocity component Airplane c.g. due to pitch rate q Fig. 7.2 Effective Increase in Horizontal Tail Angle of Attack Due to Pitch Rate 7.1.1.1 Elevator Angle per G. Following Etkin’s approach’ with minor modifications, we can derive the first of our measures of merit. The pitching moment increment due to the pitch rate in curvilinear fight depends on changes in angle of attack, elevator position, ‘and pitch rate. If the total pitching moment coefficient can be considered a linear combina- tion of contributions due to each of these variables suggested in eqn. 7.1. The increment Of Crm due to the pitch rate during @ wings-level pull-up that has negligible angular ac- celeration (this restriction implies a nearty circular path) can be written as: ACm ~C, +7239 4G, bg With our assumption of a near circular path, the pitching moment increment due to rotation is 0, just as itis in straight and level fight equilibrium condition. Then,Longitudinal Maneuverability Tests 143 (7.3) The increment in angle of attack due to pitch rate can be related to the increment in lift coefficient necessary to sustain the maneuver. AL _(n-1W _ ac, == LM n-10sim where CLiim = the lift coefficient in straight and level flight at the given speed and altitude. The change in angle of attack is then given by: aw BOL=OLs, 80 Cla ‘Substituting this expression for Aa and eqn. 7.1 for g, Cma(8C. ~Cig, 80) , 00m g Cin, 0-14 ~ Cbs, 880], OC (0-1) Cu aq CL aq Ve Cm, Cm, which simplifies to: (4) In-4 CiCm, ~Cm,Cly, 7.1.1.2 Stlek-Fixed Maneuver Point. Paralleling our previous definition of stick-fixed neutral point, we now define the stick-fixed maneuver point as the airplane center of144 Longitudinal Maneuverability Tests Chapter 7 gravity position for which elevator angle per g vanishes. Mathematically, we set eqn. 7.4 to O.and define Xmp as the stick-fixed maneuver point: : Xmp_Xnp_\ , Gm O= CG ne _ Xa s a which is solved for “10 give: o|_—_t__ a) ¢ 2 (CL. Chee Of course, Be can be rewritten with this expression for stick-fixed maneuver point: ~ (78) The quantity in parentheses suggests another margin analogous to the stability margin in Chapter 6 (page 121). The stick-fixed maneuver margin is a measure of aircraft ma- neuverabilty and is defined as: (7.7) Fig. 7.3. Stick-Fixed Neutral and Maneuver Points Typically, the stick-fixed maneuver point is about 0.062 behind the stick-fixed neutral Point for a conventional tail-aft airplane’. Considering the definition of the relative mass parameter, am _ 2%) oa PSS — PoE itis clear that as either altitude or wing loading increases, 1: increases. Equations 7.5 and 7.8 imply that either of these changes causes the maneuver point to move closer to the neutral point. Taken as a whole, a large airplane with a large tail and long tail moment arm, having a low wing loading, and flying at sea level will lead to the greatest difference between the stick-fixed neutral point and the stick-fixed maneuver point. Said anotherChapter 7 Longitudinal Maneuverabllity Tests 145 ‘way, large transport airplanes tend to be more sluggish than fighter airplanes and they are hardest to maneuver at sea level. Figure 7.23 illustrates the relative positions of the two stick-fixed points, Notice also that the elevator angle required to maintain a constant load factor increment in the pull-up does not reverse sign until the cg moves aft of the maneu- ver point, that is, the airplane can be maneuvered even though it is statically unstable. Of course, this statement does not take into account the dynamics involved and should not be construed to suggest that one should always design for this condition. It does, how- ever, suggest one of the reasons that relaxed static stability has become an attractive al- temative for high performance airplanes. 7.1.1.3 Stick Force per G. In Chapter 6 we leamed that stick force is proportional to the elevator hinge moment. It is natural to infer that the increment of stick force necessary to maintain an increment in load factor in a wings-level pull-up is also proportional to the increment in hinge moment. AFs = C9S.E.Cr, ‘Still considering conventional tail-aft tail configurations only, the contribution of pitch rate to local angle of attack at the horizontal tail is illustrated in Fig. 7.2. The angle © is defined by the ratio of the velocity component due to rotation to the velocity component due to translation. tone = $6280" or, for small changes in angle of attack o~ Hein radians (79) ‘Then the total increment in angle of attack at the horizontal tail is due to both the incre- ment in wing angle of attack required to provide the necessary additional lift and the in- crement in angle of attack due to pitch rotation at the tail. a Aay =Aay(1-E}+0 Ac, -1) =) But bay = Ac, == ond q= ‘ACL (,_ 2), 9% Then =(n-1) 4 1-— |+— say =(0 fe ) By exsuming that Bay =5CL. , we neglect any contribution to C;,due to elevator de- Le flection. This assumption is equivalent to taking Cy = 0 for this maneuver. Recalling ‘eqn. 7.4 with this assumption incorporated, and the incremental change in elevator hinge moment is:146 Longitudinal Maneuverability Tests__— Chapter 7 ACh, = Cr,Adt* Cry A Be It follows that the change in hinge moment per g is obtained by using the definition of Crim 2n4 the facts that Chm Pos zm = and Crm, =-#eVit0 get (7.10) Chap Cbeim (Oh, 20VH ( MH (Ch, CL. ‘Tuming our attention to stick-free conditions, we compared the stick-fixed and stick-free neutral points in eqn. 6.14 a 1-8 oa. Ch, where Vi, ~Vjy, and the free elevator factor is defined as F =1 — peace Chee, CO Ch, a va ee {t- Boeee CL aaCL, CL ae which, when substituted into eqn. 7.10, leads to: np, Chay 20VH sett), e th, CL, (7.11) 1-1 aeVy Next, we consider the tail contribution to the damping in pitch in order to eliminate the Xmp term from eqn. 7.11. This increment in pitching moment is by far the most important contribution to the damping in pitch, usually at least 90% of the total Cra,'. This increment in pitching moment is given by: Var, ft Ve and 8m =-VyAc,, =— YH sich implies that (2m) = ame 29 (22 (ls) aoe MihiChapter 7 Longitudinal Maneuverability Tests 147 Since the tail contribution is such a large part of Cm, we will simply allow for the wing ‘and body contributions to this damping derivative through K, where K is on the order of 1.1, and write the damping in pitch for the complete airplane configuration as K2VyCL, fe a (7.42) K2VyCL,, Le ‘Substituting eqn. 7.12 into eqn. 5 and solving: Sera (7.13) Cr, Shy Returning to equation 7.10, substituting for xmp, and using —* = 7 Ch, PalF 7) ACh, Cha, Claim (Xoq _ Xn aeftVs | —1) SerttH 14) at ee ee i we oe ‘The increment in stick force that the pilot must apply to generate the wings-level pull- up we have been analyzing is related to the required change in hinge moment through the gearing ratio G. AF = AC», G7Sec0 Rearranging eqn. 7.12 and substituting AC), into the expression for stick force Ms. Gra, Stam | Xoo 00 4K 4p 1f S0ltVe 2 cas ee |e +(k+F a] 28 (7.15) This expression is one form of a stick force per g expression. However, we can simplify the form somewhat by developing the concept of a maneuver point to include a stick-free case analogous to the neutral point definitions. 74.1.4 Stick-Free Maneuver Point. The stick-free maneuver point is defined as the airplane center of gravity position for which the stick force per g vanishes. In pilot terms, itis that og position at which zero stick force is required to pull up at any normal ac- celeration. Using the definition of C1 py, 89ain, eqn. 7.15 reduces to He iors) (7.16) Equation 7.16 lets us write stick force per g in terms of the stick-free maneuver point:148 Longitudinal Maneuverability Tests. Chapter 7 Cn, \ Xoo % avn bee is an approximate form of stick force per g (strictly speaking, stick OF 5 4-0/9 (7.17) Recognizing that force per g is the gradient thus, in the limit: Sow S304 Xoo 7.18) As we did with the stick-fixed maneuver point, let us close this section with a summary ‘of the physical meaning of these results. First, the preceding analysis is based entirely on a steady state load factor being maintained throughout a wings-level pull-up. The equa- tions shed no light on transient conditions between the initiation of the pull-up and the time that the steady state trajectory is reached. Furthermore, if the stick force per g or the ele- vator angle per g is measured in tuming flight, the equations must be altered as we will see in the next section. Having reminded you of these limitations, we note: + Stick force per g varies linearly with og position and is negative for xq > Xmp' It must be emphasized that this statement has nothing to do with stick travel per 9, only stick force per g. Stick travel per g is related to stick-fixed conditions. + Stick force per g varies directly with wing loading and directly as the "volume" (SZ) of the airplane. High wing loading airplanes like the F-15 require more force to move the controls and almost always use an irreversible control system to help the pilot. Large transport or cargo airplanes with large control surfaces also re- quire a mechanical advantage of some type to fly with acceptable pilot forces while maneuvering. ¢ Inthe absence of Mach number and Reynolds number effects, stick force per g is independent of speed. This result shows up in that eqn. 7.18 contains no speed ‘or speed dependent term other than Cpg,. Of course, this derivative, like most other stability derivatives, is strongly dependent upon Mach number. The relative mass parameter affects the relative distance between the stick-free neutral and maneuver points in just the same ways (altitude, size, and wing load- ing) as it did the difference between the stick-fixed neutral and maneuver points (see section 7.1.2). 7.4.2 Tums As noted above, the theory developed so far does not properly account for the pitch tate contribution to either elevator angle per g or stick force per g in tuming fight. Col- fecting maneuvering stability data in a tum can be a much less time-consuming process than collecting data with a series of wings-level pull-ups. Since time, especially flight test time, is money, the well-informed flight test engineer must understand the differences in maneuvering flight test measurements taken with both test methods. There are two types of tums ordinarily used to collect maneuvering stability information: (1) a steady level tumChapter 7 Longitudinal Maneuverability Tests 149 and (2) a windup tum. A steady level tum implies a constant load factor, constant altitude, and constant true airspeed. A windup tum is a dynamic maneuver in which a constant Joad factor is maintained as airspeed is allowed to decrease until some limiting condition is reached. Consider the forces acting during a constant altitude, constant airspeed tum as sketched in Fig. 7.4. Dividing the vertical component of lift by its horizontal component: H+ wh Sd Fig. 7.4. Load Factor ina Tum Loos® gr, Vo _gtan® _¢ Lsn® y2 6 Vw where © is the angular velocity about the local vertical axis. The components of © can be resolved as shown in Fig. 7.4. Notice that in a steady level tum, there is a component of about the y body-axis and a component about the z body axis. We are interested in the pitch rate q, which is the component about the y-axis. -ahnensto) 0 (1 os a= 2 -4) Va 008 Va, \cos & les 2 m= 250 ease Fee Vu 008 Taking equilibrium (trim) as a steady level tum at ig qo = H{n-2)--2f mg -L oa 76 (0-3) iol ws) ‘Simplifying, q = Lo-noKt +nng). Recalling eqn. 7.3 and modifying it slightly to fit level a, Cm, a+ “aed Cm, he mental angle of attack in a similar fashion for the change in load factor to go with the pitch (n-no)Cy,, - Cy, 55, rato inthe turn, Ace = 0 bam ~ Cy Be Cu We can now express the increment of elevator deflection required to maintain the pitch rate in the level tum. tum dynamics, A5e = , and also modifying the expression for the incre-Longitudinal Maneuverability Tests, (Chapter 7 C, eC, Bebo Mo am ~C1, gba Ser 8 (n-ng\t +079) Cin, Collecting terms that do not have the factor (n - ng) on the left side and those that do on the right side and solving for —“22 =o Cin, Stony + O14Cm, Fep(+ 0) 719 Cmts, ~CL,Cms, Itwe take the limit as n—>ng, —8e__, Se. Using the definitions of the relative mass. n=ng Parameter and Cy pny IPSC isn (4, 1 Bq _ 0M-Cham * CL, Cm, ay | 15) an CeCe, Ce Fi, Then, using Cm, =C1, *2) to simplify the expression slightly, 2h 7.19) Ee (7.19) én CimaCty, CL, Cm, Setting eqn. 7.19 to zero to solve for the stick-fixed maneuver point gives (-3) 72 Xap (7.20) 2h Equation 7.20 clearly shows that the difference between measuring the maneuver points in a tum and in a wings-level pull-up is a factor of 1+. This difference shows up a in each of the maneuver points with no change in the rest of the form for the rest of the equations. Obviously, as n gets large, each pair of measurements is approximately equal. Johnson? has provided a comparison of data for the two types of maneuvers as sketched in Fig. 7.5 (next page). Notice that as n gets larger, the slopes of the two curves become nearly identical, precisely what the theory predicted.Chapter 7 Longitudinal Maneuverability Tests: 151 Steady Symmetric Pullups Difference in slopes comes from the additional damping term pro- portional to 1 1+az Fy + ibs Difference in Fs is due to the damp- ing term being pro- 4 portional to 1 + tt 2. 3 4 5 6 Normal Acceleration - 1 Fig. 7.8. Variation of Control Forces in Different Test Maneuvers 1 7.2 LONGITUDINAL MANEUVERABILITY TEST METHODS It is the flight test engineer's responsibility to prepare @ matrix of test conditions (air- speeds, airplane configurations, power settings, altitudes, and the like) the test team will use to collect maneuvering fiight data. Typically, this matrix will be decided well in ad- vance of the actual fights, perhaps even in the master plan agreed upon before the first flight. One of his primary concems will be how to reduce the size of the matrix and still ‘adequately define the handling characteristics of the vehicle. Obviously, the contractual ‘specifications and the FAR requirements that must be demonstrated will guide this plan- ning process. A brief discussion of the techniques used in flying ‘each maneuver will also be of assistance in test planning. {In each of the test methods, as in all stability and control testing, careful trimming of the airplane to specified speed and altitude is absolutely essential to obtaining good data. ‘The pilot should be held to very close tolerances on trim conditions if precise measure- ments of angular position and control forces are to be made. 7.2.1 Symmetric Pull-up Method ‘The objective in this maneuver is to establish a steady load factor as early in the ma- neuver as possible and maintain it as the airplane pitches through the horizontal. The data recorded includes indicated airspeed, normal acceleration, elevator position, and ap- plied stick force. ‘Automatic recording is almost essential because the pilot must try to achieve the nominal test conditions (airspeed, altitude, and constant acceleration) with the longitudinal reference axis as. nearly horizontal as possible. Figure 7.6 illustrates good and bad data runs for a wings-level pull-up. ‘This test method is very time-consuming and the piloting technique requires practice, skill at estimating speed decay, and aggressive maneuvering to arrive at the required test conditions without delay. After trimming the airplane in level fight at the desired conditions162____ Longitudinal Maneuverabitity Tests_____Chaptor 7 ‘and recording these trim conditions on the data record, a climb is made to start the ma- neuver 200 to 500 feet above the test altitude at some airspeed slower than the trim air- speed. Power is set and a dive angle selected that will accelerate the airplane back to the trim airspeed just before reaching the trim altitude. Ideally the pull-up would be initiated and the normal acceleration established so that the aim load factor, airspeed, and altitude all occur just as the airplane is pitching through level flight. As Fig. 7.6 shows, it is ideal to have the g onset be smooth and rapid with no overshoot and for the load factor to remain constant until the contro! input is released. A common error is sketched in the right part of Fig. 7.6 where the load factor has been allowed to decrease as the airspeed decayed Getting an ideal data trace may take several practice runs to iterate and find the right starting airspeed, altitude, and dive angle combination. It is also very helpful to have a real-time data display with at least normal acceleration and indicated airspeed so that data points which need to be repeated can be identified before trim conditions are changed. In judging quality of a data point, the following conditions are important: © Load factor must be constant (zero slope) at the instant data are read. ‘@ Stick forces must be constant. @ Trim must be set carefully and recorded to insure that equilibrium stick forces are measured precisely. ALTITUDE + 7000 ft je= 0 GOOD MANEUVER: ' POOR MANEUVER Fig. 7.6 Examples of Good and Bad Wings-Level Pull-up Data In sum, the wings-level pull-up is an accurate, though not very time-efficient, method for collecting longitudinal maneuvering stability information. Now, let us consider alterna- tive test methods. 7.2.2 Steady Turn 7.2.2.4 Stabilized Load Factor. Trimming carefully and recording the equilibrium data are just as important for this tuning maneuver as for the wings-level pull-up. After trim conditions are recorded, a climb is made up to about 2000 feet above the test altitude and then the airplane is banked to a fixed angle while maintaining the selected load factor and airspeed precisely. Speed is maintained by carefully adjusting pitch attitude and climbing or descending as necessary. Both airspeed and load factor must be maintained precisely. (Typical limits for acceptable variations during the maneuver are +1 knot and +40. 1g for bank angles up to 30° and 42 knots and +0.1g up to about 60°.) The need for ‘such precision demands that the pilot have sensitive accelerometers and sensitive air-Chapter7 Longitudinal Maneuverabllity Tests 153 speed indicators in the cockpit for continuous references. All data should be collected within a maximum altitude band of +2000 feet from the trim altitude. ‘When a bank angle of 60° is reached, the usual technique is to maintain a constant bank angle while attempting to step up through the load factor range in 0.59 increments. ‘As for the lower bank angles, airspeed is maintained constant by adjusting the pitch atti- tude. To stabilize at each of these increments of load factor becomes increasingly more difficult 2s load factor increases because the load factor goes up exponentially with bank angle. Consequently, this technique is largely applicable for configurations where the limit load factor is not much greater than 2. Transport/cargo airplanes and other airplanes in the power approach configuration allow the use of the stabilized g steady tum. 7.2.2.2 Slowly Varying Load Factor. This variation of the steady tum is most ap- propriate for highly maneuverable airplanes with limit load factors greater than 4 and per- mits @ more economical data collection at the high load factors. It is often combined with the stabilized g method. ‘After the usual careful trim shot is recorded at the nominal test altitude, the working data band is again chosen. Typically, a band of £2000 feet is used for this method as well. ‘The airplane is slowly rolled into a continuously increasing bank while the load factor is si- multaneously increased. ‘The rate of onset of g is held at 0.1g/second or less and indi- cated airspeed is maintained as closely as possible (+1 knot is desired). Care must be exercised to avoid reversing the stick forces during this dynamic maneuver, otherwise, the friction and breakout force deadband will make it very difficult to interpret the data. tt will usually be necessary to "piece together” the data since it is rare to obtain the complete range of load factors without descending outside the £2000 feet allowed for data collec tion. Using the stabliized g method up to about 60° bank angle and then switching to the slowly varying g method is the most common way to collect maneuvering flight test data for fighter and trainer airplanes. 7.2.3 Windup Turn This method is sometimes called the constant g method since the primary goal of the pilot is to maintain a constant load factor. In this case the trim conditions are chosen close to the maximum speed for the test. Since the maneuver is certainly a dynamic flight con- dition, you do not have to insist on trimming exactly on aim airspeed. However, the trim state should be recorded before taking data and there must be no compromise that the equilibrium was stable for at least 10 seconds with hands off. ‘After the trim condition has been established and recorded the airplane is placed in a tum at some constant load factor and the data recorder is started. Pitch attitude is ad- justed to climb or dive the airplane and maintain a 2 to 5 knots/second rate of airspeed decay. Typically, a slight climb will be required to maintain this airspeed decay rate at ‘small bank angles and a descent will be needed as the bank angle increases. Data ‘should only be collected within the +2000 feet specified for all maneuvering fiight tests. Maintaining the load factor constant (not necessarily exactly at the selected nominal load factor) during the maneuver is absolutely essential to acquire repeatable maneuvering flight data. Frequently, it will be necessary to discontinue taking data because the altitude tolerance from the trim altitude has been exceeded. The pilot should note the airspeed departing the altitude band and the next segment of data should be started at an airspeed154 Longitudinal Maneuverability Tests, Chepter 7 slightly above this airspeed once the airplane has been climbed back into the allowable altitude band. This procedure will allow the engineer to reduce the data with a check on the continuity of the data. Airspeed records should also be marked where at the onset of buffet and when the constant load factor can no longer be maintained (often called the "g- break"). Windup tums can then be repeated at the same trim conditions for increasing in- ‘crements of load factor. Typically, 0.59 increments are used at higher altitudes and about ‘1g increments are used at low altitudes. The exact values of load factor are not important; it is important that a constant load factor be maintained throughout any given windup tum and that the entire range of allowable range of load factors is covered. As in all fight test- ing, care must be taken to avoid exceeding established limits, especially when the flight ‘envelope is being explored for the first time. Maneuvering flight tests, especially windup tums, are particularly susceptible to inadvertently overstressing the vehicle. Figures 7.7 and 7.8 show windup tum data?; these piots illustrate typical ways to present data. reakout Force + Friction BUFFET ONSET ui STALL STICK FORCE - Ibs pull 10° 20 30 40 50 60 NORMAL ACCELERATION - n, in g's Fig. 7.7 Stick Force per G from a Windup Tum 7.2.4 Concluding Remarks 7.2.4.1 Instrumentation Required. From these brief descriptions of longitudinal ma- neuvering test methods, the necessity for automatic recording of the pertinent parameters is obvious. The minimum instrumentation required, as suggested by Johnson? but with details and emphasis added, includes the following: ® Normal acceleration must be measured as close to the airplane center of gravity as is feasible, displayed for the pilot, and recorded automatically. As already noted, the pilot's display of load factor must allow him to fly to a resolution of approximately 0.1g if he is to be expected to produce accept- able maneuvering flight test data. A magnetic tape recorder, telemetry, or some other means of continuous automatic data collection is a must for ob- taining accurate samples of continuously varying data like these. Airspeed indications must be displayed on a sensitive indicator for the pilot, especially for those methods that require constant airspeed. Automatic re- cording is also essential to alleviate the pilot's workload and to insure preci- sion.Chapter 7 Longitudinal Maneuverability Tests, © Left Steady Tums —arinydg - 5000 ft 2 Bight Steady TUS Stabiity Augmentation — On Vim - 190 KEAS Virim - 194 KEAS| 807 Ww - 20000 Ibs - 61200 Ibs w Xog __~ 0.2166 Keg 0.2486 1) a. 60}--- i ; a ag 40 gs = 20}-;. o 2 10 <5 0 agi — i it 10. 20.30 10. +20 3010 20 30 NORMAL ACCELERATION - 9 y es 80 8s 7 Lo ‘ 109 : wo2 40 an 4 ge a 1.62.5'9 a i = 010 20 30 40 50 10 20 30 40 50 CENTER OF GRAVITY POSITION - % mac Fig. 7.8 Longitudinal Maneuvering Flight Test Data © Altitude can be recorded from a standard altimeter, set to 29.92 inches of Hg to provide pressure altitude. It is not absolutely necessary to record altitude information on the recorder, though it often is. © The forces typically applied by the pilot must be recorded, but no display is required for the pilot. It is very desirable to have a near real-time display for the flight test engineer to examine so the pilot can be advised when it is necessary to refly a maneuver prior to retrimming the airplane at a new flight condition. Such a quick-look capability usually pays for itself in test time ‘saved in short order. © Positions of control surfaces, of the pilot's controller (stick, wheel, or hand), and of longitudinal trimming devices should also be recorded and displayed on strip charts or other near real-time displays. It is possible to omit the last two of these position measurements, but they are highly desirable and are well used to help interpret the data. © Angle of attack should be recorded continuously during longitudinal maneu- vering. Again, pilot display of angle of attack is not absolutely essential but it is very convenient for the pilot and helps the engineer reduce the data and ‘compare it to other sources of information like wind tunnel tests.158 Longitudinal Maneuverability Tests, Chepter 7 © Accurate tracking of center of gravity movement for the airplane is impera- tive for this type of test. Often, this tracking can be done by hand-recording the fuel counter readings or even fuel gage readings, on a cruder scale. (Production fuel gages are notoriously inaccurate and should only be used asa last resort.) © The recording device should have some means of recording time elapsed and synchronizing all of the data channels to the same time base. For digi- tal systems, the flight test engineer must be very careful in comparing con- tinuously varying data that has been sampled at different times. Even small time differences can introduce artificial phase shifts between important vari- ables. This synchronization of the data is particularly important for control surfaces moving at near their maximum rates. Of course, this factor is more important for the dynamic data to be discussed in Chapter 9, but it can also be important in determining stick force per g variation when the inputs are made rapidly. Example 7.1: The following maneuvering flight data were collected in @ Bell P-63 at 5000 feet on a stan- dard day and @ nominal gross weight of 7780 pounds. The aircraft was in its normal cruise configuration (gear and flaps up, cow! flaps closed). Estimate the stick force gradients from each set of data (considering each direction of tum as a seperate set of data). Use these estimates to find the predict the average stick- free maneuver ooint for these conditions, 78.-0.231 22 -0.262 z z ‘Let windup Tum ____ Right Windup Tum ‘Let Windup Tum ___ Right Windup Turn 7 a Fr fi Fe Bi A 2 25 13 6154.95 20 14 25 14 10028 75 (428 3522818043 16044 20 086 10040 7555 2554 40622053 *8.-0.274 *8 -0.304 @ é Let Windup Tum Right Windup Turn “Left Windup Tum ___ Right Windup Turn fe 2 Fe a Fe 7 Fe a a4 22 30 18 76195 ~~~30 a7 as a7 32 35 20 30 56 87 80 45 9s 53 30047 10360 13066 36 68 We will estimate the stick-ree maneuver point graphically, assuming that all curves are linear (which allows us to use @ linear least-squares curve fit to each set of data). Plotting the points shown in the tables above and fiting each set of points with a straight line gives the charts on the next pege. Clearly, there is ‘enough data scatter that not all the straight lines have the same intercept (as they theoretically would ifthe measurements were perfect and if our simplifying assumptions were all absolutely correct). Nonetheless, ‘we can easily extract the slopes of each curve for the given center of gravity locations. Tabulating these slopes and plotting them against og location, the following table and charts result.Chapter 7, Longitudinal Maneuverability Tests 197 eq Slope for Slope for E__LeftTum Right Tum 02316070 «6263 0262-3289 «3.669 0274 2048 2.226 0304 0825 _ 1.300 Bs Rk 8 ‘As the chart on the lower left shows, the intercepts on the abscissa give ‘estimates of the stick-free maneuver point for each set of tums. The least ‘squares straight-line curve fits give: Fe Stiok Force - Ibs (Pull > 0) forthe left tum and x 86.9432 + 18.205 for the right tum. Solving these two equations: for the two estimates and averaging, we get: Xm 1 (see 12208) on = 343 * 56.343, Xmp__ 0.316 + 0.323 € 2 Xme «0.319 (Of course, the problem could have been solved by ignoring the difference between the left and right tum data. However, by considering the data 2s itis naturally grouped the test team is encouraged to ex plore reasons forthe differences in the data. Often this kind of approach wil lead to better understanding ‘of the underlying physics of the experiment. 7.2.4.2 Estimation of Stability Derivatives. Johnson? also suggests that maneu- vering flight data can provide good estimates of Cis. Ch,,. and Cp, First, plot the perti- nent gradients versus the center of gravity in feet,"as sketched in'Fig. 7.9 (page 158). Elevator effectiveness can then be approximated using the inverse slope of the rate of change of elevator angle with lift coefficient versus og position. og _SE (7.21) Gi. = a te Te \Statee aw nw where Cjyis the airplane's normal force coefficient defined as Cy = “= ‘The elevator hinge moment parameter due to elevator deflection comes fromLongitudinal Maneuverabliity Tests, Chapter 7 Cy, Sel w— (7.22) GbecsW where + is the ratio of elevator lift effectiveness to stabilizer lift effectiveness and Gis the elevator gearing in radiansift. Xeg Fig. 7.9 Control Surface Gradients If tis set equal to 1 and the elevator dimensions are replaced by the corresponding stabilizer dimensions, eqn. 7.22 applies to an all-moving horizontal tail. Also, both eqns. 7.21 and 7.22 apply without modification to data collected in either steady turns or in wings-level pull-ups. Finally, Cp,, can be calculated from bmp = XmpP (7.23) ifthe data were collected in a steady turn. If instead the data were collected in wings-level pullups, the term 1+ is left out of the denominator.Longitudinal Maneuverabiliy Tests 158 Chapter 7, 7.3 SUMMARY ‘This chapter reviewed the basic principles governing maneuvering fight stability and described concisely how to collect fiight test data to measure the necessary stick forces, control surface deflections, pilot input forces, and normal load factor that typically are used as measures of merit for this kind of stabilty. The feel of the airplane (and therefore the pilot's opinion of it) depend rather strongly on these maneuverability messurements. We will have more to say about the pilot's subjective opinion in Chapter 9 when dynamic sta- bility measurements are introduced. Typical flight test measurements for maneuverability may be made either using wings-level pull-ups or in a tum. In either case the dynamical ‘system is not truly in static equilibrium, but a pseudo-equilibrium (constant non-zero accel- eration is assumed to simplify the techniques. Corrections between the two basic types of trajectories (wings-level pull-ups or tums) are necessary if data taken using the two differ- ‘ent maneuvers are to be compared; these corrections are particularly important for ma- neuvering load factors just slightly in excess of one.160 Longitudinal Maneuverability Tests, Chapter 7 REFERENCES ‘ Etkin, B., Dynamics of Flight - Stability and Control, (2nd Edition), John Wiley & ‘Sons, New York, 1982. 2 "Fixed Wing Stability and Control: Theory and Flight Test Techniques,” USNTPS- FTM-No. 103, Naval Air Test Center, Patuxent River, Maryland, January 1975 (Re- vised November 1981). Johnson, H. |, "Flight Testing Aircraft for Longitudinal Maneuvering Characteristics," Chapter 4, Volume Il, AGARD Flight Test Manual, Pergamon Press, New York, 1959. 4 “Stability and Control Flight Test Theory," Vol. |, Chapter 4, AFFTC-TIH-77-1, USAF 5 Test Pilot School, Edwards AFB, California, Revised February 1977. “Stability and Control Flight Test Techniques,” Vol. I, Chapter 5, AFFTC-TIH-77-1, USAF Test Pilot School, Edwards AFB, California, Revised February 1977.161 Chapter 8 STATIC LATERAL-DIRECTIONAL STABILITY TESTS The concept of static stability for the lateral-directional modes of motion cannot be as simply presented as it can for the longitudinal modes. First, there are two moment equa- tions and one force equation involved and the moment equations are coupled kinematt cally through the product of inertia ‘xz, as well as aerodynamically. Second, the lateral mode (roll) has no inherent static stability, no aerodynamic restoring moment is generated directly by rolling. Rather, a secondary moment is generated through the directional axis due to sidesiip and dihedral effect becomes the dominant factor. Third, the controls used to produce moments about either of the axes also produce moments about the other. Ai feron deflections produce yawing moments and the rudder produces a significant rolling moment. In spite of these three facts, itis still instructive to measure the static directional stability and the dihedral effect through steady state tests and to quantify the control authorities about the x and z axes with steady state maneuvers. These steady state test methods are the subject of this chapter, but the reader will do well to keep in mind that lat- eral-directional dynamics (see Chapter 9) are very important because of the strong cou- pling between these two axes. 8.1 THEORETICAL FOUNDATIONS For this introduction to static lateral-directional flight test methods it is sufficient to ig- nore the side force equation and concentrate on the two moment equations. Using the notation of Etkin! and simplifying the equations slightly: Cz = Cz B+Cz,P+Cz,0+C.ry,80 + C5, br (1) Cp =CpB+Cn,P +Cn,F+Cng,80 +Cng,5r 2) Of course, for static conditions, p = r= 0, and these equations simplify to: Cy = Ce h+Czy,80+C25,8r 83) Gq =CngB+Cng,80 +Cng,5r (8.4) Equations 8.3 and 8.4 emphasize that the dominant aerodynamic terms for static condt- tions are Cz, and C,,, with stationary controls. For static directional stability (‘weather- cock” stability), C,, > 0; and for the airplane to have "wing leveling" tendencies, C.z, <0. 8.1.1 Directional Stability and Control As indicated above directional stability depends primarily on the value of C,, and the major contributor to this derivative is the vertical tail. ‘The fuselage and nacelles produce a significant effect, though ‘secondary to the vertical tail in most conventional designs. Di- rectional control is exercised primarily through the rudder, though the ailerons contribute @ significant yawing moment about the z axis also. This coupling of the controls are another important reason that lateral stability and directional stability must be considered as inter- related problems. These two axes cannot be separated for any meaningful stability and control analyses.162 ‘Static Lateral-Directional Stability Tests. Chapter 8 Fig. 8.1 Yawing Moment Produced by Sideslip 8.1.1.1 Weathercock Stability. Weathercock stability refers to the tendency of a vehicle to tum into the relative wind when sideslip is encountered. To have directional stability, the sideslip must produce an aerodynamic yawing moment that tends to restore the airplane to a zero sideslip condition (Fig. 8.1). The relevant stability derivative, which must be positive with the usual sign conventions for positive sideslip and yawing moment, iS Cpa, represented by the slope of the curve in Fig. 8.2. Although Crp is similar to Cm, in that both are the dominant terms for static stability, the wing configuration and the center of gravity position have relatively minor effects on Cys. For a more complete discussion of the contributions of the vertical tail, the fuselage and nacelles, and the wing, see Etkin', Roskam2, or the notes from one of the Test Pilot Schools?.45, | XS Fig. 8.2 Rudder Deflection versus SideslipChapter 8. Static Lateral-Directional Stability Tests 163 8.1.4.2 Directional Control. Aircraft and missiles are, for the most part, symmetric vehicles. If a symmetric vehicle has positive directional stability, it will tend to fly at zero sideslip. Rarely are such machines intentionally flown with the sideslip angle other than zero for long periods of time. However, when engine failure occurs or when some other ‘asymmetric flow field occurs (like slipstream rotation from a propeller or simply turning flight), sideslip must be controlled. In conventional airplanes the rudder is used for this purpose. The stability derivative that characterizes this control authority is called “rudder power" and is represented by Cn, Rudder deflections are necessary to maintain constant sideslip whenever asymmetric power is applied, to hold steady sideslips, and to coordinate tums when aileron deflections produce unwanted sideslip. This derivative is one of those that has different algebraic signs, depending on what scheme is used for denoting positive angular deflections. The most common method, the one used in most NASA publications, is not alvays convenient in fight test. Left rudder, under this convention, leads to a nega- tive yawing moment. Hence, Cn, is negative. The other approach®, preferred in fight teat, makes Cr, positive since positive control deflections are defined as those which pro- ‘duce @ positive moment about the axis in question. The reader is cautioned to take care- ful note of the method used to assign positive signs to the control deflections in any litera- ture studied, ‘Steady sideslips lead to a simple method for measuring rudder power. During any steady sideslip the rudder deflection itself is proportional to the ratio of the weathercock stability Co, to Cr, 88 can be readily deduced from eqn. 8.3 by solving for 6, and then taking the partial derivative with respect to B. The result is ¢, Br rod (85) Cra, Where the "fred" Subscript simply emphasizes that the rudder is not free to float. Recalling that the signs for Cr, and Cp, are both positive using the “fight test con- vention” eqn. 8.5 and Fig. 8.2 illustrate thet the slope of the rudder deflection versus sideslip curve must be positive under this convention for static directional stability to exist. Example 8.1; ‘A small business jet with a conventional (reversible) control system has the following lateral- directional control derivatives: C.z,, .065/rad; Crp, = 0.00005/rad; C.4,, = -0.00001/rad: Cre, = 0.005/rad. CNN NEE ee GEE err Ta For @ series of steady, straight sideslips st —Sideatie ©) _Alleron (°)_ Ruder) 20,000 feet and @ Mach number of 0.55, the fol pea 3.12 3498 lowing set of aileron and rudder defiections were 4.02 278 31.2 measured. Assuming C.z,, and Cng, are much 2.03 141 15.82 larger than C.z,, and Crg,, Calculate approximate oot 0.04 0.02 values of the stick-fixed ‘directional stability and the dihedral effect at these flight conditions. 2.04 1.40 16.48 i oe ‘er eres From the data given, approximate slopes are: &, ES 451 ut 3502 Br = -7.8hedend = = 0.69/red. tC B eB Using equation 8.5 and its counterpart for dihedral effect: (C4, =-0.045 /rad] Cnp = 0.039 rad164 ‘Static Lateral-Directional Stability Tests. Chapter 8 8.1.1.3 Effect of a Freely Floating Rudder. The rudder hinge moment coefficient is approximated by specializing eqn. 6.10 (for a symmetric airfoil and no tab deflection). Ch =Chgp te +Cpy, 8 (88) ‘The rudder float angle is obtained by setting C), to zero and solving eqn. 8.6 FOF Vig Stree = 7) (2) and Cy, =. The verti- Defining a, like ag in section 6.1.2.3; that is, 2, %, ‘trim Cal tail side force coefficient is given by Cyp =Cy,,(-B +0) +0,8,; (8.8) 80, eqn. 8.7 and the definition ¢- -++ can be used to eliminate , for rudder free con- ditions (again using a prime to denote a floating surface). 8rChap | TaeChs, Chap Che, Cy = CL, pF ~8r af Cage (6.9) 8.1.1.4 Yawing Moment Due to Lateral Control. The yawing moment due to ai- leron (for conventional airplanes) is often called a cross-coupling derivative because it produces a moment about an axis other than the one for which it is primarily intended. Aileron deflection is used to produce rolling moment, but inevitably it also produces yawing moment as a secondary effect. Lateral control devices generally cause a rolling moment by creating an unbalanced lift about the plane of symmetry. For conventional ailerons the half of the wing with the higher lit will also have higher induced drag, producing a yawing couple about the z axis. Of course, defiecting a surface like an aileron also produces a change in profile drag. For conventional ailerons at subsonic flight conditions, induced drag usually dominates and the yawing moment produced by deflecting ailerons is oppo- site to the desired direction of roll. The effect is then called adverse yaw; Cis negative for this case. If profile drag predominates, as it usually does at supersonic Speeds for a conventional aileron surface, C,,_ is positive. The effect is then called proverse or com- plementary yaw. It is desirable for C,,_ to be zero or even slightly positive; designers may 90 to considerable lengths to achieve this goal. 8.1.2 Lateral Stability and Controt The rolling moment equation is the one aircraft moment equation for which there is no natural stability. That is, an angular displacement in bank angle produces no tendency to retum to an equilibrium bank angle. This state of affairs is in contrast to the tendency to retum to a static equilibrium angle of attack when displaced in pitch about the y axis or to, the equilibrium sidesiip angle (usually zero) when displaced about the z axis. The natural tendency about the x axis is to retum to a zero roll rate, but not to an equilibrium bank an- gle. It is, however, desirable to return to wings level flight after a small bank disturbanceChapter 8 ‘Static Lateral-Directional Stability Tests 165 for most purposes: this characteristic is the only sense in which we can speak of lateral “stability.” The dominant term in eqn. 8.2 is Cz,, which produces this effect. 8.1.2.1 Dihedral Effect. Dihedral effect or C.z, is affected most by the geometric di- hedral and/or by the sweep angle of the wing. Both Etkin' and Roskam2 (and many other texts) give excellent discussions of the physical mechanisms that cause this "wings-tevel- ing" phenomenon. Wing-body aerodynamic interference also increases or decreases di- hedral effect, depending on the vertical placement of the lifting surfaces on the fuselage. A high wing produces an increment of rolling moment that retums the airplane to wings level when sideslip occurs and a low wing gives the opposite effect. Finally, the vertical tail also contributes an increment in Cz, that depends primarily on the size of the vertical tail and the vertical tail moment arm. This moment arm changes with angle of attack and, for large angles of attack, can reverse the sign of this increment. Secondary factors that may affect C,, include aspect ratio, taper ratio, extemal tanks, and wing flaps. Again, Roskam's discussion of the individual contributions to C.z, and his illustrations of typical variations with Mach number will be valuable to the interested student. 8.1.2.2 Lateral Control Power. The most common lateral control devices are ailer- ons, although spoilers and differential horizontal tail deflection are also common. Differ- ential elevons, seen frequently on tailless designs, are similar in principle to ailerons. Ai- lerons, the most representative of lateral control surfaces, produce a roll command dif- ferent to either the rudder or the elevator. In fact all forms of lateral control mentioned re- sult in a rate of roll (rate command) instead of an angular displacement. Lateral control power, 245 18 negative using either the sign convention suggested by Etkin’ or the flight test approach. That is, a positive aileron deflection (right aileron up) produces a positive rolling moment because of the decreased lift produced over the right wing and the in- ‘creased lift produced over the left wing. The exact value of the lateral contro! power (or aileron effectiveness as used by some texts) is difficult to estimate accurately and wind tunnel testing is usually in order. An expression? relating aileron control power to airfoil section (after aileron deflection), ratio of the area of the control surface to the area of the ‘wing, and the location of the wing center of pressure along the span is given by CL, Sa¥ tage C4, =— $5 (8.10) where C.,, is the lit curve slope with the ailerons deflected and y is the distance from the aifplane x axis to the wing center of pressure, again with ailerons deflected. 8.1.23 Rolling Moment Due to Directional Control. Another of the cross-coupling derivatives is Cz, the rolling moment coefficient produced by deflection of the rudder. ‘Simply stated, this ‘derivative is the non-dimensional rolling moment resulting from the side force generated by the rudder multiplied by the non-dimensional moment arm measured from the reference body axis up to the center of side force on the vertical tail. Because it is measured from the reference x axis (trim condition), this length is a function of the per- turbed angle of attack.‘Static Lateral-Directional Stability Tests, Chapter 8 Cy, 2v a Cg (0.11) Because of this dependence on z,, Cz5, decreases with increasing C,. 8.1.3 Lateral-Directional Matrix Equations The static lateral-directional equations (8.3 and 8.4 plus the side force equation) can be expressed more compactly in matrix form. The aerodynamics are clearly coupled, un- like the static longitudinal case. Of the four control derivatives in equation 8.12, only the side force derivatives, Cy,, and Cy, , are negligible for most airplanes. Cy) (Cv Cvs, Crs, Se é Cel=|Czy Czy, Ces, 9 |] 5? (8.12) Cn) (Crp Sms, Sms, 9 lg ‘Another complicating factor, from the stability and control test engineer's perspective, is that the lateral contro! (54) is of a different type than the directional control (,). The rudder, like the elevator, is @ control that commands displacement; that is, the rudder is used to command an angular position. On the other hand, the ailerons are used to com- mand an angular rate; they are rate controls. This difference means that different sensors are needed to measure the effectiveness of each type of control. in fact, measurement of the primary usefulness of the lateral control cannot be measured by static means. A dy- namic test method (described in Chapter 9) must be used. These facts mean that static lateral-directional flight tests are slightly more difficult to conduct than fongitudinal ones. There are at least twice as many parameters to be re- corded simultaneously for these tests as was the case for static longitudinal measure- ments. Automated data collection, therefore, becomes more desirable. 8.2 STEADY STATE FLIGHT TEST METHODS Requirements for lateral-directional static stability handling qualities vary considerably from civil to military specifications, as do longitudinal requirements. Of course, the prudent flight test engineer should be less concemed with the requirement and more interested in insuring that the design being tested is safe to fly and easily controlled by the typical pilot who will fly the machine. The requirements are merely guides which try to insure that minimum standards are met. Any design is a compromise pitting these good handling qualities against the pragmatic and very real need to produce a cost-effective airplane. This compromise is evaluated in every flight test, but lateral-directional requirements are particularly subjective. FAR requirements®.'0 illustrate this subjectivity: (1) The static directional stability, as shown by the tendency to recover from @ skid with the rudder free, must be positive for any landing gear and flap position appropriate to the takeoff, climb, cruise, and approach configurations. This must be shown with symmetrical power up to @ maximum continuous power, and at speeds from 1.2Vs, up to the maximum allowable speed for the condition beingChapter8____—_—Statie Lateral-Directional Stability Tests 167 investigated. The angle of skid for these tests must be appropriate to the type of airplane. At larger skid angles up to that at which full rudder is used or a control! force limit in Part 23.143 is reached, whichever occurs first, and at speeds from 1.2Ve, to Va, the rudder pedal force must not reverse. (2) The static lateral stability, as shown by the tendency to raise the low wing in @ Slip, must be positive for any landing gear end flap positions. This must be shown with symmetrical power up to 75% of the maximum continuous power at speeds above 1.2Ve,, up to the maximum allowable speed for the configuration being investigated. The static lateral stability may not be negative at 1.2Vs,. The ‘angle of slip for these tests must be appropriate to the type of airplane, but in no case may the slip angle be less than that obtainable with 10° of bank. (3) In straight, steady slips at 1.2Vs, for any landing gear and flap positions, and for any symmetrical power conditions up to 50% maximum continuous power, the aileron and rudder contro! movements and forces must increase steadily (but not necessarily in constant proportion) as the angle of slip is increased up to the maximum appropriate to the type of airplane. At larger slip angles up the angle at which the full rudder or aileron is used or @ contro! force limit contained in Part 23.143 is obtained, the rudder pedal force may not reverse. Enough bank must accompany slipping to hold constant heading. Rapid entry into, or recovery from, @ maximum slip may not result in uncontrollable flight characteristics. FAR Part 23.1774 Static lateral stability must be positive; meaning that dihedral effect Cz, must be negative. ‘There must be @ wing-eveling tendency when sideslip is generated. Observe also that these requirements cover virtually the ‘entire flight envelope, including the speed range just above stall up to maximum allowable speed and reasonable configurations. It ‘should strike you that demonstrating compliance with these requirements can take an ‘enormous amount of flight test time! You should also be impressed with how nonspecific the adjectives are and how much room that leaves for interpretation by individual FAA in- spectors. The requirements for transport category airplanes is rather similar: (@) The static directional stability (as shown by the tendency to recover from @ skid with the rudder free) must be positive for any landing gear and flap position ‘and symmetrical power condition, at speeds from Vs, up to Vee, Vie, or Vec/Vmc (as appropriate). (b) The static lateral stability (as shown by the tendency to raise the low wing in a sidesiip with the aileron controls free and for any landing gear and flap position and symmetrical power condition) must be positive at Vee, Vue, or Vro/Vuc (8s appropriate). () In straight, steady sideslips (unaccelerated forward slips) the aileron and rudder control movements and forces must be substantially proportional to the an- gle of sidesiip, and the factor of proportionality must lie between limits found nec- essary for safe operation throughout the range of sideslip angles appropriate to the operation of the airplane. At greater angles, up to the angle at which full rud- der control is used or a pedel force of 180 pounds is obtained, the rudder pedal forces may not reverse and increased rudder deflection must produce increased‘Static Lateral-Directional Stability Tests, Chapter 8 angles of sideslip. Unless the airplane has a yaw indicator, there must be enough bank accompanying sideslipping to clearly indicate any departure from unyawed flight. FAR Part 25.177 ‘The military specification’? that spells out static lateral-directional requirements is a bit more specific. Like the FAR, rudder deflection or force must produce sidestip opposite to the direction of applied directional control to meet the military specification. In addition the rudder forces and rudder deflections must remain linear for +10° and +15° sideslip, re- spectively. The military document also limits maximum allowable rudder pedal force to 250 pounds for maximum rudder deflection at any point in the envelope. Furthermore, in a steady sideslip, aileron deflection in the same direction as sideslip must accompany the rudder requirements. The wing-leveling tendencies due to dihedral effect shall not be so strong that more than 75% of roli control power nor more than 10 or 20 pounds (depending on the whether the lateral controller is a stick or a wheel) of force are required in normal use. These examples illustrate the quantified requirements incorporated into the military handling qualities requirements, though they are not all the considerations for such lateral- directional tests. The responsible fight test engineer will carefully study and analyze all applicable requirements documents for complete details during the test planning phase. Unstable 7 Fig. 8.3 Aileron Deflection versus Sideslip From this discussion of typical static-lateral directional requirements and from the theoretical considerations discussed in section 8.1, it is clear that the measures of merit used in such flight tests are closely tied to control forces and surface deflections. Van der Maas’ refers to this approach to measuring stability as contro! position stability. Quali- tatively, Figs. 8.2 and 8.3 indicate the relationship between control surface deflections and sideslip that are expected for a subsonic airplane with acceptable controls-fixed lateral- directional characteristics. Similar sketches could be drawn for aileron and rudder forces as a function of sideslip with the controls-free (vehicles utilizing reversible flight control systems). Of course, controis-free stability considerations have no meaning for a fully powered (irreversible) contro! system like the ones on most high performance fighters. 8.2.1 Test Methods to Determine Dominant Stability Coefficients ‘The basic test method is the steady, straight sideslip — a classical, proven way to de- termine static directional stability and dihedral effect from a single maneuver. It depends, though, on prior knowledge of the control derivatives like C.z,, and Cp, Like any other |Chapter 8 ‘Static Lateral-Directional Stability Tests 169 stability and control testing, the method also demands appropriate instrumentation, careful attention to trim, and precise control of airspeed and altitude. Because data are collected point by stabilized point in the steady, straight sideslip technique, it takes a lot of flying time to completely explore the flight envelope for all configurations of a given airplane. Consequently, slowly varying methods are also used to speed up the process. Both methods will be discussed in the following paragraphs. 8.2.1.1 Steady, Straight Sideslips. The classical steady, state test method is com- monly used to obtain measures of C5, and Cry, for all classes of airplanes. The stan- dard data collection technique is fairly ctficient for’ most airplanes and fight conditions and therefore needs modification. Simplicity is its strength. Setting up for this maneuver is essentially the same as for other stability tests. After eveling off at the nominal test altitude the test pilot must trim the airplane with the usual care for stability and control data. To check the trim prior to taking data, he should release the controls and the airplane's airspeed and altitude should not change for at least 10 sec- ‘onds. The control surface positions and forces at these trim conditions should be re- corded just before starting the maneuver. Sideslip limits for the flight conditions must be meticulously observed; they should be reviewed in the preflight briefing and, if the in- ‘strumentation and ground support include real time read outs, this parameter should be monitored as a flight safety item. Exceeding sideslip limits has caused loss of control and even structural damage in the past. Though steady, straight sidesiips are not normally considered hazardous tests, normal prudence should be exercised. In this test maneuver data are collected at constant, stabilized headings. The pilot should choose a reference point on the distant visual horizon and use it to fly a stabilized heading for each individual point (sideslip angle). Rudder and aileron are applied, essen- tially simultaneously, to set up a stabilized sideslip. This cross-controlled condition is an unnatural piloting technique; the rudder is normally used to maintain zero sideslip in all maneuvers. Using rudder and aileron in opposite directions, particularty at large sideslip angles, gives the pilot a sensation of ‘sliding sideways in the seat" due to the lateral forces that are applied. These unusual sensations and unnatural control applications mean that the pilot must concentrate carefully on setting up the conditions and maintaining them. Crosschecking the point chosen on the visual horizon along with the cockpit instruments for constant heading and bank angle should allow the pilot to determine when he has achieved a steady, stabilized sideslip. When equilibrium conditions have been attained, the data system should be tumed on to record altitude, airspeed, sideslip angle, bank an- gle, control surface deflections (aileron, rudder, and elevator), and control forces for all three axes. From such data, plots like Figs. 8.2 and 8.3 can be constructed and, asstette ing the control pawer derivatives are known, Czy, and Crq, can be calculated from A ir slopes and of the curves. 8.2.1.2 Slowly Varying Sidesiips. As was indicated earlier, the pure steady, straight sideslip method can be time-consuming and is often modified to make the data collection more efficient. ‘The time saving is largely achieved by changing the sideslip conditions continuously rather than stabilizing at each rudderfaileron position. Trimming is done the same as for the stabilized method and the pilot must still pick out a point on the distant ho- rizon and maintain a constant path toward it, But, at this point the technique changes.170 ‘Static Lateral-Diractional Stability Tests, Chapter 8 Rudder and opposite aileron are blended in slowly and in concert so that a constant track is maintained. To acquire the data, automatic recording of some form must be used and the data are collected continuously. The slowly varying method obviously saves time if + does not change too rapidly (no more than about 1 */second is a good rule of thumb) and if the test pilot can maintain the proper integrated rate of change of rudder and aileron to give constant track. Such a dynamic maneuver is harder to fly than the stabilized one and, when conditions are not carefully maintained, the point must be flown again. Natu- rally, such repetitions eliminate the advantage of the method. But, the potential for saving significant amounts of fight test time favors the slowly varying technique and it is probably the most common way to collect static lateral-directional data. 8.2.1.3 Steady Tums with Rudder Fixed. This flight test technique, though no longer widely used since parameter estimation techniques have become popular, offers the advantage of simplicity and relatively uncomplicated instrumentation requirements. For our purposes, it is a useful way to teach the fundamentals of such measurements. As always, the airplane must be carefully trimmed at the desired altitude and airspeed conditions. The pilot then holds the rudder at the trim position and banks the airplane us- ing the lateral control alone. This bank should be entered slowly since a steady sideslip and bank angle is sought and too abrupt an aileron (or lateral control) input may induce a Dutch roll oscillation that hinders data collection. Once equilibrium has been attained, 5g, 8;. Fa. Fr. >, B, and r are recorded, preferably with some form of automatic recording in- strumentation. Sensors for measuring each of these quantities must be provided. The procedure is repeated at several increments in bank angle and the same parameters are recorded at each test point. Care must be taken not to exceed sideslip limits and it may become increasingly difficult to maintain airspeed and remain within 1000 feet of the nomi- nal test altitude at larger bank angles. The elevator should be used to control airspeed but care must be taken to not introduce spurious lateral control movements while making cor- rections with the elevator. Measurements should be made during tums in either direction. Data from these tests are plotted to obtain trim curves of 8, and Fg versus bank angle and plots of B and as functions of nondimensional yaw rate z. These plots can be used to calculate the damping derivative C,,, though we will pursue this data reduction. 8.2.1.4 Steady Tums with Ailerons Fixed. This test method is essentially identical to the one discussed in the preceding section, except that the turns are commanded by rudder alone and information is measured to calculate C_,, rather than C,,. Again, we will not pursue this damping cross derivative further in this introductory text. 8.2.1.5 Use of Steady Sidestip Data. The basic steady sideslip flight test technique is a very useful one. Obviously, it provides a quick, but accurate assessment of the static lateral-directional stability of the airplane. In addition steady sideslip data can be used to ‘measure the rudder control power derivative C,,,, and the aileron control power derivative Cz,,, 88 we wil ilustrate more completely in the next section. To obtain additional infor- mation, like these control derivatives, the flight test planner must often add equipment or devise means of applying known extemal moments, But, the usefulness of such a simple test is remarkable.Chapter 8. Static Lateral-Directional Stability Tests 1m 8.2.2 Test Methods to Determine Control Effectiveness ‘Measuring the effectiveness of the controls is one of the more important purposes of static lateral-directional fight tests. Since the rolling and yawing moment equations are so ‘tightly coupled, these tests are also some of the more difficult to conduct with certainty. ‘There are methods that utilize movement of the vertical position of the og to measure ai- Jeron effectiveness but they are not very accurate and, consequently are not in general use®. We will ignore them, too. However, steady, straight sideslips at differing longitudi- nal c.g. locations can produce rudder control effectiveness data that are usable. A second ‘approach that will be discussed for measuring C,,, involves the use of a parachute at a known location on the wing. If the parachute's drag’ coefficient is also known, the resulting drag force can be calculated. The rudder necessary to overcome this known yawing mo- ment is a direct measure of its effectiveness. Similarly, an asymmetric moment can be applied about the roll axis by positioning a weight at a known spanwise location and measuring the aileron deflection necessary to counteract it in equilibrium flight. This latter method is straightforward and is quite accurate since it can involve no change in drag other than the induced drag due to defiection of the aileron. Obviously, all these tests still take into account the coupiing between lateral and directional modes. Axeg Fig. 8.4 Effect of cg Shift on Coy 8.2.2.1 Steady, Straight Sideslips at Different C.G. Locations. In eqns. 8.12 we can usually ignore the effect of aileron (or other lateral controls) on the side force. If the rudder side force term can be neglected and ¢ is small, the equations reduce to:‘Static Lateral-Directional Stability Tests__Chapter 8 cy) (Cy 0 0 Gli, Ca|=|Cay Cay, Cay, O lI 5e (8.13) Cn Crp Sms, Sms, lo If we now fly @ series of steady, straight sideslips covering a range of 8, equations 8.13 can be written and solved for each of the flight conditions. If we move the c.g. loca- tion between two such series of steady, straight sideslips, the only aerodynamic coeffi- cients that change significantly are Cr, and Cn,. As long as the tall moment arm is large compared to the shift in c.g. location, Cr, is also essentially independent of c.g. position. The change in C,,can be calculated from (using the geometry of Fig. 8.4, page 172): c (8.14) Meg mba Cre +O Using equations 8.13 and the slopes - and a from the series of steady, straight sideslips, Cy, can be calculated: Cy =-Cy 2 (8.15) The slopes St. end & are usually independent of og location, but it is good practice to check this assumption by taking sideslip data at several cg positions. Eliminating the side force equation, any two sideslip conditions can be evaluated using ®, ®, ® Ss) +0n,() =0 and Gm, =n, (33) Defining ACny = Cry, - Cry, and observing that the aileron slope change with sideslip is usually quite small, & % AC +C, (2 -(5) 0 (8.16) a * Cre, (op), CoB), If the moment arm for the vertical tail (distance the center of pressure of the vertical tail is above the c.g. for an aft tal airplane) is established, Cy, can be estimated from the value of Cn, calculated from eqn. 8.16: b Ot, Cre, Fe (8.17) The data reduction for this test is very straightforward, though the flying can be tedi- ous because it is not always easy to alter the longitudinal c.g. location in flight. Rudder defection and bank angle are plotted as functions of sideslip angle as they are for all sideslip tests. At B = 0 the slopes of these curves are obtained and C; is calculated forChapter 8 ‘Static Lateral-Directional Stability Tests 173 the test conditions and the test weight. The slopes Sr and FP are pated for each og location. Then, Cyg, Cng, and Cy, are computed from eqns. 8.15, 8.16, and 8.17. ‘This test method is rather cumbersome and is only moderately accurate. A large number of sideslip points are required to credibly establish the small differences that de- termine the slopes of the measured variables. These differences come from two relatively large numbers so the information is highly uncertain. The technique is not often used be- cause of the test time needed to acquire the data. 8.2.2.2 Yawing Moment Produced by a Wing Parachute. The second method of directly measuring rudder effectiveness is also cumbersome to use. It is, of course, im- perative that the parachute can be deployed in flight and jettisoned after the data are taken. Safety concems can become hazards if care is not taken to ensure that parachute deployment and jettison is completely reliable. Uncertainty added to the process because few parachutes are completely stable after deployment. Typical installations for a wing parachute are sketched in Fig. 8.5 on the next page. Figure 8.5b shows one possible way to provide some stability for the parachute, especially when the wing tip vortex excites the oscillations. The more simplistic (and less practical hardware configuration) is illustrated in Fig. 8.5. The simpler installation gives the in- cremental yawing moment created by the parachute as: AN= Tt (8.18) lB af T T fa) Direct Measurement (b) Indirect Measurement of Parachute Force of Parachute Force Fig. 8.5 Wing Parachute Installations ‘The measurements can be made and the results expressed in terms of Xog, @, and the force P rather than direct measurements like T and ¢. If this arrangement is used, Van der Maas suggests taking data at several different tether cable lengths. ‘The condition for equilibrium at zero sideslip (eqns. 8.13) is simply AC, = Cn Abn, where AC, is the incremental yawing moment coefficient. A.known AC, is provided by the parachute since it can be calculated from the para- chute's drag coefficient and the moment arm in Fig. 8.5a. Of course, the parachute force can be measured directly (further complicating the instrumentation scheme) with either a174 ‘Static Lateral-Directional Stability Tests Chapter 8 force gage in parachute riser or in the tether cable as suggested in Fig. 8.5a. But by defi- nition AC; is also AC, = 7b and a direct calculation of rudder effectiveness is then possible. The rudder side force co- efficient follows from eqn. 8.17 with the required rudder control effectiveness given by: Ong, * a5" The piloting technique for these tests is that described for the steady, straight sideslip method. Care must be taken to avoid oscillations and, if the Dutch roll oscillation is not ‘well damped for the airplane, it may be difficult to stabilize the parachute. Safety proce- dures should be carefully rehearsed in advance. Clearly, this test is one of the those which might benefit enormously from flight simulation of the planned profiles and careful evaluation of possible emergency procedures on an engineering flight simulation. 8.2.2.3 Measurements with Weights at Known Spanwise Locations. A technique similar to the one described above can be used to measure lateral control effectiveness. Like most other static lateral-directional tests, this one also involves fiying a series of steady, straight sideslips. An unbalanced rolling moment is generated by positioning a weight at a known spanwise location. if the airplane can carry fuel in tanks that provide an adjustable and precisely known lateral imbalance, there is no need to install special stores or to add weights than can alter the flutter characteristics of the vehicle. How to ‘achieve the known lateral imbalance and still guarantee adequate handling qualities during landing and takeoff is a practical issue of significance to the test planner. Alleron and rud- der deflections (and the corresponding control forces) required to balance the known ‘asymmetric rolling moment are measured over a range of sideslips. ‘Atzero sideslip the equilibrium condition suggested by eqns. 8.13 reduces to: Cz = CzyNig, giving an expression that allows direct calculation of the aileron control effectiveness from Measurement of AC, and A8y AC, C4 aig At the same time the aileron defection produces an increment in yawing moment that ‘must be counteracted by the rudder. Crs Bat CngA5,=0 From this equilibrium expression it follows that, Abr Cray * Cre, B56 Pilot procedures for this test are no different from those described for other steady, straight sideslip tests. Most of the care to be exercised is associated with fuel transfer or stores jettisoning to attain the desired lateral imbalance. At least two different measurable (8.19) (8.20)Chapter 8 ‘Static Lateral-Directional Stability Tests 175 lateral cg positions must be tested. Three or more such cg locations are necessary to ‘check for nonlinearities in the estimated data. Data reduction for this test is similar to previous ones. First, measured 5, values are plotted against B. The slope of this curve at 6 = 0° is needed, but measurements at other sideslips allow fairing of the 54-6 curves and give a better estimate of the slope at zero sideslip. Then, with the rolling moment known, eqns. 8.19 and 8.20 allow direct calcula- tion of the two aileron control derivatives Cp, and Cys, 8.2.3 Estimation of Stability Derivatives from Steady Lateral-Directional Tests Having detailed tests that provide flight test estimates of the control effectiveness de- rivatives, let us briefly summarize how these static lateral-directional test measurements: are used to confirm wind tunnel data and analytical estimates of these same derivatives. First, the control effectiveness derivatives C,,, Cg C¥gy and Cy, are determined from one or more of the control effectiveness tests ibed in section 8.2.2. With these values in hand the dominant lateral-directional stability derivatives, Cz, and Cy, (along with the less important Cy,,) are determined from plots of control surface nd forces measured st each of the sideslip angles. The resulting functional relationships, hopefully all linear for conventional airplanes, are experimentally determined in steady, straight sidesiip tests or the slowly varying sideslip variation of the test. Next, the yaw rate derivatives, C,, and C,,, are calculated from data recorded in steady tums initiated with aileron or rudder alone. In this fashion, nine of the most important lateral-directional pa~ rameters can be estimated with reasonable accuracy from essentially static tests. 8.3 SUMMARY Static lateral-directional flight tests are slightly more complicated both in measurement techniques and in data reduction requirements than are their longitudinal counterparts. Coupling between the two relevant moment equations accounts for this complication. However, if control effectiveness derivatives are known or can be measured accurately with specially designed tests, simple steady, straight sidesiip tests can reliably determine the dominant lateral-directional derivatives Cz and Cy While they are not as often con ducted nowadays (since parameter estimation identification algorithms have become so widely used), steady tums using aileron or rudder alone also make it possible to straight- forwardly estimate C,,and C,,. All in all, these kinds of fight tests provide a wealth of information for a moderate amount of effort and rather minimal data reduction schemes. Their most serious drawback is that they do require more data collection time than some of the more modem techniques like parameter estimation. Of course, such techniques also need sophisticated instrumentation and they consume large amounts of digital com- puter time to extract the same stability coefficients.176 ‘Static Lateral-Directional Stability Tests, Chapter 8 REFERENCES ' Etkin, B., Dynamics of Ellaht: Stability and Control (2nd Edition), John Wiley & Sons, New York, 1982. 2 Roskam, J., Airplane Fllght Dynamics and Automatic Controts, Part |, Roskam Aviation and Engineering Corporation, Lawrence, Kansas, 1979. “Stability and Control Flight Test Theory,” Vol. 1, Chapter 5, AFFTC-TIH-77-1, USAF Test Pilot School, Edwards AFB, California, Revised February 1977. 4 "Stability and Control Flight Test Techniques," Vol. ll, Chapter 7, AFFTC-TIH-77-1, USAF Test Pilot School, Edwards AFB, California, Revised February 1977. 5 "Fixed Wing Stability and Control: Theory and Flight Test Techniques," USNTPS- FTM-No. 103, Naval Air Test Center, Patuxent River, Maryland, January 1975 (Re- vised November 1981), ® Perkins, C. D., “introduction,” Chapter 1, Volume ll, AGARD Flight Test Manual, Pergamon Press, New York, 1959. 7 Van der Maas, H. J., "Lateral and Directional Control and the Measurement of Aero- dynamic Coefficients in Steady Asymmetric Flight and Flight on Asymmetric Power,” Chapter §, Volume Il, AGARD Flight Test Manual, Pergamon Press, New York, 1959. Hunter, P. A., "Flight Techniques Used to Determine Adequacy of Lateral Control,” Chapter 6, Volume Il, AGARD Elight Test Manual, Pergamon Press, New York, 1959. ° “Airworthiness Standards: Normal, Utility, and Acrobatic Category Airplanes”, FAR Part 23, Federal Aviation Administration, Washington, 1969. 10 “pirworthiness Standards: Large Commercial Transport Category Airplanes", FAR Part 25, Federal Aviation Administration, Washington, 1969. "Military Specification, Flying Qualities of Piloted Airplanes", MIL-F-8785C, November 1980.Chapter 9 DYNAMIC STABILITY TESTS Flight testing to determine the dynamic stability of an airplane is even more compli- cated than the tests outlined in the previous three chapters for static stability tests. Al- though is it usuelly still possible to decouple the equations of motion for the longitudinal and the lateral-directional cases, the analysis is considerably more complicated. Dynamic stability, as the name implies, is concemed with the long-term effects of disturbances on the vehicle. It requires that we study the time-retated behavior of the response of the dynamic system, not just its initial tendency after a disturbance from equilibrium. ‘Anderson! has captured the essence of dynamic stability: [A body is dynamically stable if, out of its own accord, it eventually retums to and} remains at its equilibrium position over a period of time. Static stability measurements are necessarily made with all forces and moments in equi- librium. Conversely, dynamic stability measurements, while they are usually initiated after carefully trimming the vehicle, must be made with the forces and moments not in equilibrium. Furthermore, measurement of these dynamic quantities must include some form of time correlation in order to have meaning. Of course, dynamic stability is closely related to static stability. While an airplane can be statically stable without being dynamically stable, it cannot be dynamically stable without being statically stable. Moreover, dynamic stability testing is concemed with the ability that the vehicle affords the pilot (or controller) to change from one equilibrium condition to another. The speed and precision with which these changes can be made is determined by the characteristic modes of motion of the vehicle. Dynamic stability is not the only concer of the flight test team engaged in this facet of flight testing. The handling qualities (sometimes called controllability) of a manned airplane are also of prime importance and many of the requirements that drive dynamic stability spring from the handling qualities as well. Again, both static and dynamic stability, along with the characteristics of the control system, the kind of airplane, and the task to be ‘accomplished, determine the handling qualities of the airplane. Controllability has been defined in terms of what the pilot is tasked to do: [Controviabity may be defined asthe capability ofthe eplane fo perform, at the] equired in total mission accomplishment? However, controllability means something quite different to the controls engineer and the term handling qualities is to be preferred because of this ambiguity. From this discussion it should be clear that the handling qualities of any manned aircraft depend heavily on subjective assessments. Certainly most handling qualities requirements are based on pilot opinion? Before the test team can measure the characteristics of such dynamic motions, it is essential that they have a thorough understanding of these and other terms, the modes of motion, and how to quantitatively define them. Let us tum now to laying the theoretical foundation for the measurement of these dynamic properties.178 Dynamic Stability Tests Chapter 9 9.1 THEORETICAL FOUNDATIONS To property understand the dynamics of aerospace vehicles requires a brief return to the six degree-of-freedom equations of motion that have been developed in many textbooks45. and to the definitions used to quantify the modes of motion. First, we need to fix the definition of certain terms used in dynamic flight analysis. 9.1.1 Definition of Terms and Assumptions AAs stated in the introductory paragraphs, dynamic stability requires that the motion of 2 body ultimately retum to its equilibrium position. There are at least two ways in which physical systems retum to (or diverge from) equilibrium as shown in Fig. 9.1. The exponential decay (or divergence) is associated with a linear factor in the goveming characteristic equation, while the damped (or divergent) oscillation is associated with a quadratic term in the characteristic equation. Of course it is also possible for oscillatory motion to be undamped and this condition implies neutral dynamic stability. Fortunately, all linear systems can be represented mathematically represented by a combination of ei- ther linear terms or quadratic terms in the characteristic equation; that is, the time-de- Pendent solutions are linear combinations of either exponential decays or divergences and damped, neutrally damped, or divergent oscillations. (c) Undamped Sinusoid (d) Damped Oscillation (Neutral Dynamic Stability) (Dynamically Stable) Fig. 9.1. Possible Responses of Linear Dynamic Systems Since any linear dynamic system can be represented by a combination of no more than second order factors, it is instructive to review the definitions used to describe the dynamic behavior of an airplane. Figure 9.2 depicts a spring-mass-damper system that we will use in making these definitions. Mathematically the solution to the second order spring-mass-damper system is identical to solution of the airplane's longitudinal equations of motion. Only the constants change. The equation describing the spring-mass-damper system with no extemal forcing function is:Chapter 9. Dynamic Stability Tests, 179 mit + ox + kx =0 and the characteristic equation is: BrEaekno @.1) Fig. 9.3. Characteristic Roots in the Complex Plane For the airplane's longitudinal analog of this motion one needs only to define the aerodynamic “spring” that corresponds to k and the damping terms that correspond to c. Typically, the stability derivative Cm, adequately describes the aerodynamic “spring” and the pitch damping Cn, relates to the damping coefficient c in the analogous aerodynamic ‘system. The aircraft moment of inertia is the analog to the mass m in the simpler me- To introduce additional definitions of interest in an airplane's dynamic behavior, let us now consider the roots of eqn. 8.1 and how they are affected by changes in & and x. These roots are: 5 c,fey_k ener —|-= (9.2) 2° 3m * 4) m cS 2 ) >t | the roots lie along the real axis in the complex plane (Fig. 9.3). This situation resuifs in an exponential decay if root is negative or en exponential divergence180 Dynamic Stability Tests Chapter 9 if the root is positive. The time-domain behavior of such a root is illustrated in Fig. 9.1a. Other terms used to describe this type of dynamic behavior are aperiodic motion or deadbeat subsidence (for a negative root). A system that has a decaying exponential mode of motion is often referred to as overdamped. If the damping is reduced or if the spring constant is increased until ( * ) the system is said to be ertically damped. Critical damping is indicated on Fig. 9.3 by the point on the real axis where both real roots coincide: 4,2 -= at point C. The motion is still aperiodic at this point, but it is on a boundary. Any increase in x or decrease in © will produce @ complex conjugete 2 solution to the characteristic equation because (3) - Keo. Physically, an oscillation will result. The characteristic equation can also be written in terms used to define oscillatory behavior and compared to eqn. 9.2, 15g + mq +05 =0 which has the solution yp =p tq yt-C? (9.3) Undamped Imaginary Fig. 9.4 Motion Parameters on the Complex Plane ‘The damping ratio ¢, which compares the actual damping to critical damping (Fig 9.4), is @ measure of how rapidly an oscillation decays or of how rapidly an exponential decay or divergence occurs. For critical damping, ¢=1. The system is said to be un- derdamped if 0<¢<1 and undamped (or neutrally damped) if ¢=0. The undamped natural frequency @, is a measure of how often the peak amplitudes occur. The |Chapter 9, Dynamic Stability Tests, 181 damped frequency of the oscillation (which, strictly speaking, is not a frequency) is simply og = @n\1—G2 . (The time response is not purely periodic unless ¢=0. Consequently, © is sometimes called a conditional frequency’, indicating that it has the units and many of the properties of a frequency.) ‘The period of the oscilaton is T= 2%. Each of these a parameters is used to quantify the characteristics of oscillatory roots. The angle @ is measured from negative real axis to a line drawn from the origin in the complex plane to the characteristic root (or eigenvalue) in question. ‘Therefore, this angle defines a vertical line of constant damping ratio. Clearly, a number of parameters associated with the modes of motion can be represented on the complex plane. The transient response of the dynamic system to inputs also must be characterized in terms of "quickness" parameters and there are terms used that indicate both the speed with which the system responds and whether or not it reaches the ultimate value commanded by the input without exceeding or "overshooting" that value. The steady state (or new equilibrium) response of the system to an input is that value which the out- put will eventually reach. ‘Typical steady state output responses to a step input are shown in Fig. 9.5. Generally, second order systems will show some oscillatory behavior 2s sug- gested by Fig. 9.52. The damping factor Co is the real component of the characteristic root that controls the rate of rise or the rate of decay of the system response after an input. It appears as the constant in the exponential term of the time-domain solution of the system equations. As sketched in Fig. 9.5a, this constant defines the shape of the ‘envelope bounding the peak amplitudes of the oscillation typical of second order dynamic system. The time-to-half-amplitude (or time-to-double-amplitude in the case of an ‘oscillatory divergence) is the time required for the amplitude of the motion variable under consideration to change by a factor of two. The time-to-peak-amplitude f, is the time required to reach the maximum amplitude of the output. rr eas t {e) Second Order System (b) First Order System Fig. 9.5 Responses to a Step Input Maximum overshoot (or peak overshoot) is the largest deviation of the output re- sponse to a step input during the transient motion and this amplitude is a strong indicator of the relative stability of the system. It indicates how precisely the system can move from182_____________ Dynamic Stability Tests__—__—Chapter 9 one equilibrium state to another. It follows that the percentage maximum overshoot is given Percent Maximum Overshoot = ‘max imum overshoot steadystate Settling time f, is another parameter used to measure the transient behavior of a stable linear system. It is defined as the time required for the response to a step input to decrease and stay within a specified percentage of its steady state value. Rise time f is defined as the time required for the output to increase from 10% of its steady state value to 90% of its steady state value when the system is excited by step response. (An alter- native approach sometimes used is to give the rise time the value of the reciprocal of the slope of the step response at the instant when the output is at 50% of its final value.) Delay time ty is the time required for the output response to a step input to reach 60% of its final value. All of these time measures of merit are related to the time constant x. This charac- teristic is defined as the time to reach 63.2 per cent of the steady state value for this first order response (as shown in Fig. 9.5b) or for the envelope of the oscillatory peaks to reach the same value. This latter fact means we can still use our definition of the time constant for a second order response; itis stil the real part of the root (and that is the only part for a first order system) that sets the rate of convergence or divergence. An instability or divergence implies a positive root. Note, though, that the damping factor as shown on the complex plane is a negative number for stable systems. Since o is always positive, the damping ratio ¢ is described as being negative when a positive real eigenvalue exists. Heuristically, this "negative" damping is an appealing way to describe a diverging system. (Again, strictly speaking, damping cannot be negative; but the terminology is widely accepted and understood.) 100%. 9.1.2 Equations of Motion There are two approaches commonly used in the United States to approximately model the linear dynamic motions of an aircraft:. Both are based on Bryan's approach to linearizing the equations by assuming that the aerodynamic forces and moments can be adequately represented with a first order Taylor series expansion. (Etkin4 correctly points out that this approach is mathematically flawed, specifically for cases where the aerodynamic forces and moments change abruptly or when the control surfaces are rapidly displaced, the first order approximation is inadequate. However, the transfer function approximation that Etkin suggests to replace Bryan's method has not been widely accepted and it depends on convergence of an infinite series that is not always mathematically well-behaved.) We will restrict our comments to the two formulations that use Bryan's approach, simply because it fits our purposes in this introductory text better than any other approach for two reasons: (1) the flight test engineer is more likely to be provided data based on Bryan's approach from wind tunnel and computational estimates and (2) the bulk of the experimental measurements he must make can be readily verified with this relatively simple linear model, These reasons are clearly pragmatic, rather than scientifically satisfying. Nonetheless, they are appropriate for introducing the flight test engineer to dynamic stability tests.Chapter 9. Dynamic Stability Tests, 183 ‘The first formulation utilizes dimensional equations that virtually every textbook on the subject of aircraft stability and control. §6. 17 develop, often in excruciating detail. Table 9.1 below summarizes the results, using dimensional stability parameters. Table 9.1 Linearized, Small Perturbation Equations Using Dimensional Parameters Longitudinal Equations SU + Woq + GOCOS@p = X, + Xyw + Xiy SW +Xqq+ X sho sw + Ugg + gosin@g = Zu + Zyw + Ziy WW + Zq9 + Z5,50 (9.4) 8q = Myu+ Myw + Miy sw + Maq+ Ms,B0 irectional Eauatc ‘sv-+ Upr= Wop - 40089 = Y+ YySv+ Yop Yir+ Ya,da+ Yadr spor = hy shiv Lypt Lit + h,be +s, 8, (9.5) die 5p = Nyv-+ NysV+Npp+Nr+N5,5q +N5,8, x a In Table 9.1 lower case letters (except for s, which is the Laplace operator) represent perturbed quantities; for example, U = Up + u, where the upper case symbols with the 0 Subscript are the equillrium (or tim) values for the parameter in question. it is assumed in the equations presented that Po = Qo = Ro = Vo = Oand that all control surfaces are initially in the trail position (86, = Ba = Bro = 2 ‘the trim condition. There is by no Means Consensus about the use of these Symbols; Etkin uses &n,and< to denote the perturbation control surface variables, while Nelson uses ABs, A5,, and A6, to identify the ‘same deflections. ‘McRuer® and his co-authors point out that the terminology most com- monly used (“dimensional stability derivatives") is more properly “dimensional stability parameters" and the common usage is to include either mass or moment of inertia terms the definitions: that is, xy ai % is the form used for force derivative parameters and Mg it is the form used for moment derivative parameters. Nelson'7 calls these iy parameters “derivatives divided by mass (or inertia)". Each of these parameters is ‘evaluated at trim conditions in the Bryan linearization and is therefore a constant in eqns. 9.4 and 9.5. itis important to note that this nondimensional formulation is concerned with direct measurements, while a second approach that still uses Bryan's linearization is concemed with force and moment coefficients and nondimensional velocities. Most of the jc stability derivative data that comes from wind tunnel measurements is presented in this nondimensional form. Table 9.2 lists a common set of equations using this approach. Typically, when the analysis is concemed with the data base iteetf, the flight test engineer will use nondimensional forms. However, when he tums to calculation of transfer functions, ‘actual time histories, or carrying out complicated time domain simulations that are best illuminated with physical engineering units, the flight test engineer usually prefers the dimensional form for the data. In any event, the flight test engineer must be able to readily recognize and convert between the two forms of data.184, Dynamic Stability Tests 0 Chapter 9 Table 9.2 Linearized, Small Perturbation Equations Using Nondimensional Coefficients Longitudinal Equations (2uD - 2Cgtan®, - Cx,)0 - Cx, + CigOq- Cxs,80= 0 (2ig- Cz,)0 + (2uD - Cz, D- Czq)a.- (2u+ Cz)9 + CigtanO,9 - Cz,, (9.6) = Cm) - (Cg D - Cmg)2+ (ip -Crng)4+ Crng,50 = 9 -De=0 Directional Eauat 2uD ~ Cyp)B - Cyp + (2uD - Cy,)P- Cup - Cy,ba - CvySr = 0 + CagB+ (iD - Cap )B- (cD + Ca,)P - Czs,5a - Ces5 ~ CrpB = (ieD + Cyp)B + (icD - Cy,)P- Cns8a ~ Cns5r=0 (9.7) B+? tan@g-Dg=0 P sec@g- Dy=0 ‘Small angle approximations are used to simplify these linearized equations. There are ‘several different forms for these equations, as Etkin points out clearly. The chief differences lie in the definition of the normalizing unit of time (t* defined below), the choice of the characteristic length (in fact, ¢ is defined differently for the longitudinal and for the lateral-directional equations), and the definitions used for the nondimensional stability coefficients themselves. Summarizing, the following definitions apply in Table 9.2: w v é £ m ly rat ~ fe = . = 7. = a aD Pm a, Uo’ **psé ese pene er ce mee eee ee =Fe rao ines ic - Pa 200 "205° | Sees "Fee Derivatives with respect to ? are indicated by D. 9.1.2.1 Longitudinal Modes of Motion. For a stable vehicle, the longitudinal motion characteristic equation results in two different oscillations; the eigenvalues are two ‘complex conjugate pairs. One pair, usually called the phugoid, describes a low frequency interchange of potential energy and kinetic energy, which is very lightly damped with a Period of 30 seconds or longer. It typically takes place with negligible change ine. A second oscillation, the short period, occurs at V,;. Its period is usually an order of magnitude less than that of the phugoid; hence, its name. Fortunately, this oscillation is well-damped throughout most of the flight envelope. 9.1.22 Lateral-Directional Modes of Motion. Solving the quartic lateral-directional characteristic equation typically produces two real roots and a single pair of complex roots for a stable conventional airplane. The real roots correspond to the roll mode and to the spiral mode. Typically, the roll mode has a time constant that is much shorter than the spiral mode. The oscillation described by the complex pair is called the Dutch roll mode of motion and it is moderately damped within the flight envelope for a conventional vehicle.‘Chapter 9 Dynamic Stability Tests. 185 Vj; = 350 knots Bost Tested Boundary Vi = 950,Arats ——— Unsatisfactory Boundary Fs /g = 6.0 lb/g 10) ey £51508 g § ou # Fos BiI5 2718 § 2 [Soal. > lB ere 3 8 is 02 §5)° o 00 02 04 06 08 10 1.2 Damping Ratio Fig. 9.6. Use of Pilot Opinion Ratings'° 9.1.3 Handling Qualities ‘The parameters defined above are excellent tools for engineers and designers; in fact, these individuals must have a solid grasp of such dynamic measures of merit. Unfortunately, few operational pilots are engineers. And it is the pilot alone who has the utimate responsibilty for controlling the airplane. The ease with which the pilot (whether he or she is in the machine or on the ground) can fly the vehicle to accomplish the tasks which it was designed has a very strong influence on the usefulness of the machine. How easy an airplane is to fly for a given operational task comes under the subjective heading of handling qualities. For the design engineer and for the flight test team it then becomes necessary to communicate with the pilot and assess the handling qualities of design before it can be developed to have, not just "good" performance, but also “good” handling qualities. Since "goodness" is a subjective measure, there is a need to relate the engineering parameters defined previously and the pilots’ opinions regarding the airplane. ‘The most useful guidelines that have evolved over the years are the military's handling qualities specifications’.18 for manned aircraft and their substantiating documents. ‘These specifications spell out the ranges of parameters that historically have given acceptable handling qualities. Nonetheless, these documents are historical and often there is no information on how a new design innovation might affect handling qualities.196 Dynamic Stability Tests Chapter 9 For that reason the test pilot and the fight test team must be familiar with pilot rating schemes and how they should be used in the evaluation of handling qualities. Pilot opinion surveys, utilizing pilot rating scales like the Cooper-Harper scale, have been used extensively to quantify this subjective information®. Figure 9.6, adapted from Cooper and Harper'®, illustrates the general principle. A large number of pilots were asked to fly a specific airplane performing a well-defined task and ¢ and «, were altered over a wide range of values. From this statistical base and the opinions recorded, an acceptable and an unacceptable range of values of ¢ and @, were inferred. This rather simplistic ilustration does not tell the whole story, however. 9.1.3.1 Closed Loop Response versus Open Loop Response. The damping ratio and the undamped natural frequency discussed above are properties determined solely by the airplane and its flight control system, that is, they are open loop parameters. The pilot's inputs close the loop, since he or she acts as an observer (or sensor) and a controller (control law and actuator). The open loop part of the system is shown in the dashed rectangle in Fig. 9.7, which depicts the closed loop system as well. The compo- nent block labeled pilot command is itself a very complex transfer function. A given pilot, performing @ particular task at a specific time will have one set of dynamic response characteristics. Change any of these conditions and the human response is likely to change. This human variability (or, in the best light, adaptability) dictates that flying qual- ities experiments be carried out under carefully controlled conditions. It also means that the real test of an airplane's handling qualities is in providing satisfactory closed loop re- sponse with a large number of pilots. AIRCRAFT EQUATIONS OF MOTION ‘STABILITY, AUGMENTATION YSTEM FOR AN OPEN-LOOP SYSTEM, THIS FEEDBACK PATH IS OPEN Fig. 9.7. Closed and Open Loop Block Diagram 9.1.3.2 Pilot Rating Scales. In the mid-1950s the Society of Experimental Test Pi- lots (SETP) organized a session of technical papers at an annual meeting of the Institute of Aeronautical Sciences, one of the predecessor organizations for the present American Institute of Aeronautics and Astronautics (AIAA). One of the papers'! presented was ti- tied "Understanding and Interpreting Pilot Opinion.” In 1984 Robert P. Harper, Jr., was. invited to give the Wright Brothers Lecture’! at the 25th AIAA Aircraft Systems Design, Operations, and Test Meeting in San Diego. Much of this work was carried out in variable stability airplanes and in ground simulators but perhaps the most significant result was the widely used 10-point Cooper-Harper rating scale (Fig. 9.8). This rating scheme is based on a series of binary decisions in response to the following questions: © Is the configuration controllable or uncontrollable? © Is the airplane acceptable or unacceptable?Dynamic Stability Tests, 187 Chapter 8, ‘© Is the airplane satisfactory? Except for a “no answer to the first question, each of the responses leads to a possibility of three different levels of controllability, acceptability, or satisfaction. If the pilot will lose control of the vehicle during some portion of the mission, then it receives a rating of 10 (the worst qualitative assessment). Now, let's tum our attention to defining the terms that govem these dichotomous decisions. ADEQUACY FOR SELECTED | AIRCRAFT Demands on the Pilot in TASK OR REQUIRED OPERATION* CHARACTERISTICS Selected Task or Required Operation EXCELLENT Pilot Compensation Not HIGHLY DESIRABLE —_ Factor for Desired Performance| GOOD; NEGLt- Pilot Compensation Not a BLE DEFICIENCIES Factor for Desired Pertormence| 2 FAIR: SOME MILDLY UN- Minimal Pilot Compensation Re-| PLEASANT DEFICIENCIES quired for Desired Performance MINOR BUT ANNOY- Desired Performance Requires ING DEFICIENCIES Moderate Pliot Compensation L MODERATELY OBJEC: Adequate Performance Requies| 57 TIONABLE DEFICIENCIES Considerable Pilot Compensation, VERY OBJECTIONABLE BUT Adequate Performance Requires} TOLERABLE DEFICIENCIES " Extensive Pilot Compensation MAJOR DEFICIENCIES * fe if with Maximum Tolerable Plot ‘sation: Controlisbilty not in Question Pilot Compensation is Required for Control cl Fig. 9.8. Cooper-Harper Pilot Rating Scale Controllability is the ability to command a desired response which is dictated by the intended use. In answering the first question, how much attention the pilot must pay to achieve the desired control is not considered. The airplane is controllable if he can command the desired response even if he has concentrate ail his attention on flying the machine to the exclusion of all other matters. However, it is not controllable in the context of the mission if he cannot maintain control of the airplane with the effort and attention188 Dynamic Stability Tests Chapter 9 available over and above his other mission duties. Hence, the airplane can be un- controllable for a given mission, even though it does not crash. Cooper-Harper pilot rat- ings always assess controllability for an intended mission. Therefore, the mission and its individual tasks must be clearly and completely defined before pilot ratings have meaning. Acceptable suggests that the vehicle can accomplish the mission. Acceptable does not say how well the required tasks can be performed. It may require an inordinate ‘amount of the pilot's attention to achieve the required level of performance, but it can be done. If the effort, the concentration, and the workload necessary to complete the mission ‘are too high, the pilot may reject the airplane and declare it unacceptable. Satisfactory implies “adequate for the purpose.” In pilot ratings, satisfactory does not necessarily imply perfection (almost no machine is perfect!). It is good enough that he is, not asking that it be fixed; it can meet ail the requirements of the assigned mission. To infuse as much objectivity as possible into the pilot rating process, definition of the mission is perhaps the single most important factor. Not only must the mission be defined, it must be understood alike by both the evaluation pilot and by the test engineer. What the pilot is required to accomplish with the vehicle and the circumstances under which he must perform the mission are two essential elements in this definition. The circumstances may be part of the system being evaluated (like the cockpit displays, the configuration of the airplane, or the weapon controls, for example) or they may come from extemal influences (like whether turbulence is present or absent, the pilot's level of fatigue, or his proficiency in the task). Since virtually all engineering tests must simulate to some degree, this lack of total realism must be clearly understood by both the test engineer conducting the test and the evaluation subject. It would, for example, be quite impractical to evaluate a close air ‘support airplane in an environment including all the ground-to-air defenses that were found in Central Europe in the 1980s. It is absolutely wrong to presume that any numerical pilot rating totally describes any subject's qualitative assessment. The data collection phase during pilot rating experi- ments must also include these narrative comments. The rating itself is merely a summary of all his subjective feelings with regard to the defined task. Any evaluation pilot should be required and encouraged to make qualitative comments over and above the numerical rating. A good flight test engineer will thoughtfully prepare for the debriefing session, seeking to ask questions that will bring out the subtleties that the numerical ratings frequently gloss over. This debriefing session, where these extra comments are most likely to be elicited, should occur immediately after each evaluation. This need for immediacy suggests strongly that a voice recorder is needed during the flight for complex mission profiles. Communication, as is true throughout such an experimental process, is of the utmost importance. Perhaps the most important single element in the subjective rating process is trust between the evaluation pilot and the engineer conducting the experiment. The pilot must be encouraged to believe that the engineer is vitally interested in the qualitative data, and the engineer must be just as assured that the pilot wants to give him accurate, meaningful data. Even when the pilots' comments oppose the engineer's judgement of how the airplane ought to behave based on his knowledge of its characteristics, both parties must strive to communicate and understand each other. Let us now retum to the linearized longitudinal equations of motions succinctly summarized in Table 9.1Dynamic Stability Tests 189 9.1.4 Longitudinal Dynamics: If we define the state vector for the nondimensional equations as x=[u a @ 0], then the dynamic system can be expressed in conventional state space form. anda (2 ppp f8M Opt Cx) + Cx = Chgin® * Org 50 (Qu-Cz,)Da = (20,,+ Cz,)a + Oza + (21+ Cz) + Caa.Be ipDG -CmyDa. = Cy + Cryo + Omg + Cin Be De =qG (9.8) Equations 9.8 are easily manipulated into matrix form: Cyk = Ayx + By (9.9) where, Bm 0 00 2C inj #290 + Cx, CX, 0 -Chren cya]? He 2 1g, o Cz, 2n+Cz, 0 0 -Cm, ig 0 o Cm, Omg ° ° o o4 0 o 1 ° Cx, @ ¢, By =| Ce |; and, of course, x=|4 Crs, 4 ° 8, In this simplified longitudinal case, where we have assumed that the elevator is the control (that is, power is not changed and there are no other moveable longitudinal controls), u is a scalar — not a vector; that is, u = 8». If Cy-1 exists, eqns. 9.9 can be put into "standard" form by taking C-1Ay = A and C;-1B; = B. Then x= Ax + Bu. Ais called the plant ma- trix and Bis called the control matrix. Operating on eqns. 9.9 with the Laplace operator, they are transformed into algebraic equations which can be solved straightforwardly. if we take Laplace transforms with the usual assumption of zero initial conditions, (218 -2CL pint pimp ~ Cx) 4 (8) ~ Cxg(S) ~ Crp (8) = Cxg,80(8) (2019- Gz,)i (s) + (2H - Crg)8- CzJa(s) ~ (2+ Cre) 4 (5) = Czy,Bols) (9.10) = Cm, (8) - (Cog8 + Cm,)048) + igs - Crn4(S)) 4 (8) = Cing,5e(S) 9 - 4(8) a0] Of course, the free response is obtained by solving the equations with the forcing functions (Bu) set to zero. Since the system is linear, the forced response can be ob- tained separately and added to the solution for the free response. The free response can be readily obtained using Cramer's rule or any similar method for solving linear algebraic equations. Symbolically solving for a due to an elevator input, Cramer's rule gives:190 Dynamic Stability Tests Chapter 9 a(s) _ det NS (s) 5e(s) det A(s) Brin 128i +Cx, Cxs, 9 Clay ie 0 Cy, 2n+Cz, 0 where NG®(s) = i o Cin, Cmq ° 0 1 ° 2Ciin 809m +Cx, Cx, 9 Clam 0 Cz, 2+Cz, 0 A(s)= o ee 0 0 o 1 0 The denominator determinant in the above equation is simply the determinant of the pseudo-plant matrix Ay, which when set to 0, is the characteristic equation of the dynamic. system and its roots are called the eigenvalues. For our simplified longitudinal equations the characteristic equation is a quartic and, for most conventional airplanes, it can be fac- tored into two quadratic terms. These two quadratics describe two oscillations of widely differing frequency; both are illustrated in Fig. 9.9. The lower frequency one is called the phugoid and the higher frequency one is the short period mode of motion. -10 | H ‘ i - ‘ : 010 20304080 60—70-— 8090700 Time - sec Fig. 9.9 Computed Time Histories after an Elevator Input ‘Typical periods for the phugoid are on the order of 30 seconds to a minute and a half, while those for the short period are usually an order of magnitude less. The phugoid is also very lightly damped or even slightly divergent. The pilot's inputs are made often ‘enough that he usually can control a divergent phugoid and not even be aware of a long Period instability. Such handling qualities often occur when the airplane is configured forChapter 9, Dynamic Stability Tests 191 landing (the power approach (PA) configuration in handling qualities requirements®.) The short period oscillation is consequently of greater importance because the pilot must con- trol it immediately if it is not well-damped. If his reactions are too slow or are phased im- property, his commands may even drive the closed loop system unstable. Of course, the dynamics of the control system itself may hamper or even preclude the pilot from making appropriate corrections. ‘Such short period motion occurs essentially at constant true air- ‘speed. This fact leads us to an approximation (section 9.1.4.2) for the short period that can be useful in planning flight tests. 9.1.4.1 Dimensional Form of the Equations. The linearized longitudinal equations ‘of motions have also been written utilizing dimensional derivatives. McRuer, Ashkenas, and Graham® developed these equations with assumptions that essentially parallel the previous development. Taking stability axes that align the x-axis with the relative wind in the trim condition, Whim = 0. Xw =Xq =0 have also proven to be satisfactory assump- tions, Then, eqns. 9.4 can be rewritten in matrix form: Cok = Agx+Bzu (9.11) 1.0 0 Xu Xw 0 = 96088 Him 01-Zy 0 0) Zu Zw tim +Zq o ere C2 =| 9 ~My 10) A=) My My = Mq 0 o 0 04, o 0 1 ° Xbq wu | 280 |. {7 and Bz ine vand, 2=| 4 oO 8, 9.1.4.2 Short Period Approximation. Considering eqns. 9.6 with d = 0 (no pertur- bation in true airspeed), four equations in three unknowns resutt. Since q is simply re- lated to the pitch attitude by taking the derivative with respect to time and since the x force ‘equation represents a balance of momentum change in the direction of u, it is logical to ‘select the z force equation and the pitching moment equation to approximate the two un- knowns of interest, a and @. Sticking to the Laplace operator form and the non-di- mensional stability coefficients and assuming that Cz, and Cz, are negligible, eqns. 9.6 reduce to: (2us - Cz, - (218+ Cipjqf@ Orin)® = C7550 = (Cig 8 + Cm) + (ip Cg = Cmg,S0 For these simplified equations the characteristic determinant yields a cubic in s that factors readily into a trivial s = 0 factor and a quadratic factor if we assume that prim = 0 a78? + 418+ ag=0 (9.13) where @2 = 2uig 84 == 2u(Cma, + Omg) ~ Sxl fa= Cn Co - 200m (9.12)Dynamic Stability Tests Chapter 9 Dividing eqn. 9.13 by az and comparing this result with the standard form for dynamic response given in eqn. 9.3: Png Cr, = 2u(Cmg *Cmg)-Crgi8 2059 and Ssp * ‘These approximations are too complicated to provide ready i derivatives that affect the short period mode's undamped natural frequency and damping ratio. In the expression above for wp,,, the product CrngCz, is often quite small relative to the Cm, term. Thus, Cm, is clearly the dominant derivative in estimating @q,,, though physical configuration and flight conditions also play a part through n and ig. Similarly, in the first term for sp, Cm,, May be small (though , more often, itis simply ignored) in com- Parison to Cy Omitting Cm, from the expression alters the form of the result not at all ‘and often the value of the damping term very little. The damping ratio depends primarily on the stability derivative Cry, Of course, the moment of inertia about the y axis again is important, as are the flight conditions. Taking these assumptions into account, eqns. 9.14 simplify to: ny * (9.15) With dimensional derivatives these approximations take the form: Mg -Z Ong * "Ma and Se "35 (9.16) 8p ‘Care must be exercised in applying these approximations. The primary reason for in- troducing them is to clarify which stability derivatives most influence each of the dynamic figures of merit. Do not use eqns. 9.15 or 9.16 when precise values for wn,, and Cop are needed. Too many assumptions are involved. 9.1.4.3 Phugoid Approximation. As Fig. 9.9 suggests, the phugoid oscillation oo- curs at essentially constant angle of attack (remember a. is the perturbation in angle of at- tack from the trim condition). Thus, the equations of motion can be reduced from 4 to 2 with a line of reasoning similar to that used for the short period approximation. In this case the moment equation is the logical candidate for elimination since the rotational motion has little effect on the exchange of kinetic and potential energy that the phugoid represents. If we neglect C,_, again take Orrin = 0, and also utilize the fact that a ~ 0 during most of the phugoid oscillation, the x and z force equations reduce to: (2S ~ Cy)O + Cri ~ Crs, (2Cigim~ Czy) G - 2480 - Czy,80= 9 (9.17) Naturally, the resulting characteristic equation for these approximate equations is slightly simpler than for the full set. Noticing that the lifts litle different from the weight so long as. ‘we do not perturb the vehicle too far from the equilibrium (trim) conditions, a quadratic equation like (9.13) again results with these coefficients:Chapter 9 Dynamic Stability Tests 193 a= 42 ay=-2uCy, 80 = 2C bin? ~ Chemo Let 2CLim ~ Cru) & On, * aa and p= Fea Ee (9.18) However, except for the transonic flight regime, Cz, is usually small compared to 2C pram: Consequently oy, = vt Oxy ew? 2A2C Lum To interpret eqns. 9.19 physically, we must recall that an, is pressed in nondimen- sional time units; thet is, radians/airsec, where as noted in Table 9.2, "= 77. So to put pp into physical time units, we must divide eqn. 9.17a by this parameter. Carrying out this division and noting that Ci... = and that p= and Gp ~- (9.19) putimS none 2W__ pgSl Utim 20. "pubis VW & Uti In trimmed level fight, stil ignoring transonic effects, C,,=- 2p, Thus, g Sop Do PBC ~ V2Lum ‘The phugoid damping ratio can also be approximated with dimensional derivatives: Xu 20n, Equations 9.20 through 9.22 must be used carefully, but they show that the phugoid natural frequency is inversely proportional to true airspeed. This approximation suggests that the phugoid time constant tp = 0.138up. It also indicates o, and tp are roughly inde- pendent of altitude and gross weight of the sirplane. Blakelock'? shows that both these frequency parameters do depend on variations in density, evidently because of the effects of terms neglected in this analysis. For equilibrium fight eqn. 9.21 implies that Cp is di- rectly proportional to total vehicle drag. As trim airspeed increases at constant altitude, phugoid damping ratio increases as the square of the airspeed. Similarly, if rue airspeed is held constant, Cp decreases in direct proportion to the change in density with attitude. While these approximate equations do not give accurate answers, they do offer insight into the most important stability derivatives affecting the phugoid. ae post" (9.20) (9.21) op (9.22)194 Dynamic Stability Tests Chapter 9 Example 9.1: Several examples of typical sircraft dimensional stability derivative data are given by McRuer, Ashkenas, and Graham®. The data in Table 9.3 summarize such stability derivatives for an attack aimplane flying at low subsonic speeds. Solving the characteristic equation that results from equations 9.11 with the data in Table 9.3 we get: @n,, = 1.5611 rad/sec and Cyp = 0.3588. Using approximations like oqne. 9.20 and 9.22: Ong, 1.5082 af yp ~0.2609 ‘Table 9.3 Longitudinal Data for a Subsonic Fighter (M = 0.2, h = sea level ) = 0.0813 (sec) My = 0.0029 (f-secy? =0.0312 (sec) My 0.0102 (ft-secy4 0.00432 (fY/sec*/rad) My = 0.000846 (ft1) - 0.026 (sect) Mg 1.48 (scr?) + 0.307 (sect) Moe = 2.21 (sec?) -0.001681 Msp (0.000182 (sec?) 07 (fsec?rad) ‘Note: This airplane has both a trimmable horizontal tail and an elevator (denoted by subscripts 5, and 8, respectively, above). Elevator deflection does not affect the X force equation. ‘The approximate undamped natural frequency differs from that calculated from the full characteristic equation by approximately 3.4% while the estimated demping ratio is about 17% higher. A similar calcula tion for the phugoid mode yields considerably less satisfactory results, with the approximation giving an er- ror of approximately 34% in @,, and over 200% in ty. These approximations are useful for quick estimates of the short period but the phugoid estimates are extremely crude. Table 9.4 Stability Derivatives Significantly Affecting Longitudinal Oscillations'2 Aerodynamic Motion Parameter Effect on Motion Parameter Parameter Affected Most Cinq Short period damping ratio, Cs sp increases with increasing Cm, Cin ‘Short period undamped natural rep increases with increasing Cm, frequency, On, Cy, oF Do Phugoid damping ratio, Sp Gp increases with increasing Cy, oF Do 1.) Phugoid undamped natural fre- ©p, increases with increasing Czy Or quency, &ap ( 1 ) C2, or Ve) 9.1.4.4 Sensitivity to Longitudinal Stability Derivatives. Blakelock'? has also ex- amined the sensitivity of the damping and natural frequency of the longitudinal modes of motion to variations in Cyy Cxay CzpCxgr Cys Cmy 294 Cmg Table 9.4 summarizes his findings for a four engine jet transport. 9.1.4.5 Summary of Linear Longitudinal Dynamics. The linearized longitudinal ‘equations of motion used to analyze airplane pitch dynamics illustrate the complexity of dy- namic flight tests. Approximate equations of motion, obtained to help the fight test engi- neer anticipate trends in dynamic figures of merit, must be used with good judgment. They give insight to help you plan dynamic flight tests more efficiently and to help you spot trends in data collected that indicate undesirable longitudinal dynamic characteristics.Chapter 9. Dynamic Stability Tests 195 9.1.5 Lateral-Directional Dynamics The linearized lateral-directional equations (see Tables 9.1 and 9.2) are separated from the longitudinal ones by assuming that the product of inertia /,z is negligible and that there are no other aerodynamic or control surface coupling terms are of significance. With ‘these decoupled equations we can study the asymmetric dynamic response of an airplane or a missile. tf we choose the state vector to be x =[8 p 7 4)", we can put eqns. 9.7 in matrix form: 2uDB = CyB + Cypb =Qn- Cy)? + Ciyigh* — Oye Bat Oya Br ipDB -igD? =CeB+Ce,h + Cz,? + e580 CesSr (9.23) ~igD P+ gD? = CygB + Cop B + Cn,F + CryB0% CngSr De B é ‘These equations can also be put into the matrix form of eqn. 9.9 am 0 0 Cy Cy Cy, ~28 Chum |0 in -ie Ol, |Cz, Ce Cx, 0 C3=)0 -ig ic of As= 1p One 0 0 0 o 4 o 0 ° 1 Cys, Cys, B a A By =| C4 O48r|: and, of course, x= 4 Cre, Cre, o 0 $ ‘Again, if Cyt exists, this control equation can be put into “standard” form by premul- tiplying the plant and the control matrices by Cy"? Cs'7Ag = A and Cy'1Bs = B. 9.1.5.1 Dimensional Form of the Equations. The linearized lateral-directional ‘equations of motions can also be written to utilize dimensional derivatives. Again, see McRuer, Ashkenas, and Graham® for the full development and the complete set of as- sumptions. Cyt = Agx+ Byu (9.24) a Oueoneo Y, Yp Y;—Vn -9C08 rim 0 kk -le of}. 4 4 & 0 wre Ce=lg pg ol!) MATIN, Np OM 0 oo of a) 1 Ys, Yo, v 45, <5, é Banlgt ug] 4 *=/? 0 0 4,196 Dynamic Stability Tests Chapter 9 For either set of lateral-directional equations the control vector includes at least two elements. For conventional aircraft, u7 = [ 8, 6, ]. The plant matrix sets the free re- sponse of the dynamic system: its eigenvalues describe transient behavior. These char- acteristic roots typically include two aperiodic and one oscillatory mode of motion (two lin- ear factors and one quadratic factor). These three modes of motion make up the asym- metric dynamics for a conventional airplane. In Figure 9.10 all three of the modes are su- Perimposed after an aileron input excites them. It is almost impossible to pick out to the roll mode. The oscillatory Dutch roll mode appears to dominate. 4 iain AAD Skyhawk Simulation i © 0.2 (223 fps) Sea Level P - Degrees/sec 1 - Degrees/sec SodwonaohbhrdFovaok 0 5 10 15 20 25 30 Time - seconds Fig. 9.10 Lateral-Directional Dynamic Response 9.1.5.2 Roll Mode. The roll mode, as the name implies, describes the response of the airplane to a lateral or roll command. For a conventional airplane, the control surface input ‘would be the aileron. The expected response is a decaying exponential in roll rate with a small time constant, usually on the order of 1 or 2 seconds. MIL-F-8785 requires that this time constant be less than 3 seconds?. This mode is illustrated later in Fia. 9.17.Chapter 9, Dynamic Stability Tests 197 9.1.5.3 Roll Mode Approximation. Etkin‘, Blakelock'2, and Roskam® ail retain only the rolling moment equation to give an approximate expression that will show which of the stability derivatives most directly influences the roll mode time constant. The rolling mo- ment alone is assumed to contain all needed information. Moreover, the 8 and 7 terms are neglected. Ib - £pb=0 (9.24) 8p ar — : 34 Bowe 1 Bb leben | 7 ~ M = 0.2, Sea Level 49 |}. _i| M2 0.8, 38,000 feet 0 10 20 0 40. 50 60 70 80 90 100 Te Fig. 9.11 Stable and Unstable Spiral Modes of Motion 9.1.5.4 Spiral Mode. ‘The second first order response from the lateral-directional equations of motion has a much longer time constant, typically about 10 times that of the roll mode. This spiral mode relates time to the airplane's tendency to roll and descend if no control action is taken. Bank angle is the most logical dynamic flight test measurement used to describe the spiral mode. Like the phugoid mode discussed under the longitudinal response, the spiral mode can be controlled easily by the pilot under normal circum- ‘stances even if the bank angle is divergent. The time-to-double-amplitude must, however, be long enough so that the pilot does not have to devote an excessive amount of his at- tention to this correction. Figure 9.11 illustrates both a divergent and a convergent spiral mode, from simulations for the A-4 Skyhawk aircraft model given by McRuer, Ashkenas, and Graham®. ‘The spiral mode is typical of many aircraft in that at high altitude and rela~ tively high angle of attack the spiral often becomes divergent. Also, notice that the time constant for the divergence is quite long, suggesting that this instability will ikely be a mi- nor nuisance to the pilot. Both military and civilian requirements documents for this mode of motion allow some instability of the relative ease with which such divergences can be controlled by the pilot. 9.1.5.5 Spiral Mode Approximation. Roskam® suggests an approximate expression for the spiral mode, generated using the same general approach 2s above. However, as ‘we will see shortly, this approximation is often quite inaccurate. The highly coupled nature of lateral and directional modes of motion is largely responsible for making such estimates unacceptably crude. Part of the difficulty lies in deciding which of the four equations best represents this mode of motion. We note that 6 changes little in a spiral, suggesting that elimination of the side force equation may work; further, in the spiral transient, the roll rates are usually quite low. So, following Roskam, we will try taking p= 0 while ignoring199_______________ Dynamic Stability Tests Chapter 9 the side force equation. Simplifying eqns. 9.7 with these approximations, extracting the simplified characteristic equation, and solving for As yields: CayCn, ~ Cry Cu =e er 7 np “Ar (0.28) BCay , baCny ‘Sab ‘Sqb For conventional airplanes and stability axes, /,Cny << /,C, because of the relative sizes of f, and J. Then, eqn. 9.26 simplifies further CaCn, ~ Sng Ca, +t (9.27) As Ta om The inaccuracy of this approximation is highiighted by using Blakelock's example of a jet transport and calculating the spiral mode eigenvalue and its associated time constant from the full set of equations. The results are: 4 = 0.004/sec and T, = 250 sec. By way of comparison, eqn. 9.27 gives: As = 0.099/sec and T, ~ 10.1 sec. Rather obviously, the approximation leaves much to be desired. About all we can say is that it did show the spi- ral to be unstable; but this approximation does a very poor job of predicting the time re- sponse character of the mode. Roskam points out that the asymmetric aerodynamic. forces depend on B, 4, and W, rather than on B, 4, and y. The damping terms associ- ated with § and w are often of the same magnitude as those due to 6. Consequently, they cannot be neglected safely. However, the approximation exercise was not wholly fu- tile, for we can pick out the significant stability derivatives from eqn. 9.27. The numerator term on the right, coupled with the knowledge that dihedral effect is negative for a con- ventional airplane, leads to the following condition for a stable spiral mode. C24Cn, > Ong Cz, (9.28) ‘Again, because the spiral mode has a relatively long time constant, it is not absolutely essential that it be stable. As Blakelock shows, many airplanes have divergent spiral modes at low speeds and convergent ones at high speed. It is often quite within the ca- ability of the pilot to correct for spiral divergences. In fact the spiral mode and the Dutch roll mode are both strongly affected by dihedral effect; increasing dihedral (making C.z, more negative) to drive the spiral mode more stable can reduce Dutch roll damping. Often a slight spiral instability is accepted to improve Dutch roll damping, a parameter that often has a much larger effect on lateral-directional handling qualities. So, except during the performance of certain tasks (instrument approaches, for example) requiring virtually all of the pilot's attention, the handling qualities as perceived by the pilot may be largely unaf- fected by an unstable spiral mode. Moreover, the spiral mode can be controlled fairly eas- ily with stability augmentation; yaw dampers, even in their simplest forms, can be designed to improve spiral stabilty in a closed loop automatic control system. 9.1.5.6 Dutch Roll Mode. This lateral-directional oscillation is a tightly coupled roll ing and yawing motions usually occurring at medium to high frequency and having moder- ate to light damping with no yaw damper installed. Typically, at cruise conditions the pe- Tiod of the oscillation is on the order of 3 seconds. The ratio of bank angle to sideslip an-Dynamic Stability Tests 199 Chapter 8, oie * isa key parameter, it often indicates how pilots will react to such an oscillation. Usually, a low ‘ ratio (yaw dominates the Dutch roll) is more acceptable to pilots than an oscillation in which relatively large bank angle excursions take place. If the Dutch roll is lightly damped and/or the frequency is too high, it may be necessary to add a yaw damper of simitar automatic flight control system component to augment the natural characteristics of the aircraft. 9.1.5.7 Dutch Roll Approximation. Roskam®, Blakelock'2, and Nelson‘? ap- proximate the Dutch roll oscillation by simplifying the equations of motion, though each uses a slightly different approach. They also arrive at similar conclusions. Blakelock sug- gests that examination ofthe rudder input transfer function —, ly emphasizes the direct link of opp, to altitude (through @}). Like the short period, @np, is also directly proportional to V,.. So, increasing airspeed increases on, and increasing altitude decreases Op," ‘Aitematively, we could slightly improve our approximation by solving the reduced or- der set of equations (side force and yawing moment equations only) as both Roskam® and Nelson‘ do. This approach is useful for airplanes with low dihedral effect. The resulting characteristic equation is Yp YpN, NBY, 2 %), (Ney, MOY) {: {uote +MY, }-2 ‘Now, the approximate expression for op, is slightly more complicated.200, Dynamic Stability Tests Chapter 9 (9.31) ‘The middle term in eqn. 9.31 usually dominates the lateral-directional oscillation; so, this approximation is in effect adding small terms to the approximation of eqn. 9.30. From the characteristic equation the damping ratio Cpr is approximated by: Yp + Veo Ny (9.32) Sor= 20m Ve N, Damping ratio is dominated by the second term, which is set by the ratio - Te and itis 's also related to 7, using similar reasoning as we did for np, . Example 9.2: Returning tothe airplane considered eartier to ilustrate longitudinal dynamics to provide ad- ditional insight into tateral-dynamics and our approximations, The data® are summarized in Table 9.5. Table 9.5 Lateral-Directional Data for an Attack Airplane (M = 0.2, h = sea level ) - 22.9 (Wsec/rad) 1.875 (sec'2) Yoq -0.606 (R/sac/rad) 45, 0.1264 (secr2) Ys, - 0.00272 (tYsec2/rad) Np 28 (sec?) 45 +321 (ser?) Np -0.111 (soc) 4p = 0.412 (sect) N, = 0.296 (sect) 4 0.0317 (sect) ~ 0.0242 (sec?) = 1.272 (se?) Table 9.6 Effect of Altitude and Airspeed on the Roll Mode of an Attack Airplane Flight Conditions ‘Time Constant (seconds) ‘Approximation Full Equations Attitude (feet) Airspeed (fps) aS (eqns. 97) ‘Sea Level 223 2.42 1.78 Sea Level 950 0.26 0.26 35,000 487 1.87 178 35,000 at 1.23 4.19 ‘Substituting the derivatives from Table 9.5 into the approximate relations and comparing to the solu- tion of the complete emall perturbation lateral-directional equations gives an approximate en, that is about 85% smallor than comes from the complete equstions: en, it 1.894 radisec from the complete set of ‘equations, while the approximation gives 1.6823 radi/sec. The same comparison for Cp, shows that Dutch ‘oll damping ratio is overpredicted significantly; Cp is approximated as 0.1185, compared to 0.0502 for the complete equations result. This error is over 100%. For this example at least, this Dutch roll approximation Gives a fairly reasonable estimate of Dutch roll frequency, but the demping ratio is unreliable, The ap- Proximation typically gives such results. Roskam applies a similar approximation to a small business jet and finds that the Dutch roll frequency is predicted quite well (1.62 radisec from the approximation compared to 1.618 radisec from the approximation). Again, the damping ratio is not predicted well (0.058 from the ap-Chapter9__ Dynamic Stability Tests 201 ‘Skyhawk Simulation M= 0.2 (223 fps) 5 10 16 E 25 0 Time ~ Seconds Fig. 9.12 Dutch Roll Transient Response (Full 6 DOF Response to Rudder) ‘Table 9.7 Effect of Altitude and Airspeed on the Dutch Roll Mode of an Attack Airpiane Flight Conditions ‘Damping Ratio @p (radisec) Full Equations (eqns. 9.31/9.32) (eqns. 924) | (eqns. 9.31/9.32)_(eqne. 9.24) SeaLevel 223 1.68 1.89 Sea Level 950 858 929 35,000 487 234 255 337 3.47 ‘Table 9.6 summarizes the effects of airspeed and altitude on the roll mode and Table 9.7 illustrates the ef- fect of these variables on the Dutch roll mode. Each tabulation shows the errors introduced in using our ‘epproximations. Clearly, the complete equations are preferable. These compiete equations generated the time histories in Fig. 9.12, showing the character of the Dutch roll. The time histories show the strong damping that develops at high dynamic pressure; there is a stark contrast between the first two time histo- ties in Fig. 9.12 thet illustrate this effect. Also, the decrease in o,,,. 8 altitude increases is clearly seen. Finally, notice the lerge increase in a p,, a8 airspeed increases at either altitude202 ynamic Stability Tests Chapter 9 9.1.5.8 Sensitivity to Lateral Directional Stability Derivatives. Blakelock’? also considers the sensitivity of the lateral-directional modes of motion to variations in the asymmetric derivatives. Table 9.8 summarizes and compares the effects of the most sig- nificant of these derivatives on this type of motion, just as we previously did for the longi- tudinal modes of motion. Notice that our example, using the mathematical model of a rather different airplane, leads to the same conclusions. Table 9.8 Effects of Stability Derivatives on Asymmetric Modes of Motion’? ‘Stability Motion Parameter Effect on Motion Parameter Derivative Affected Most Co, Cor Increase |Cp, | to increase damping Cop ng, Increase Cp, to increase natural frequency C4, Roll Subsidence increase le. ap | to increase roll mode time constant Cz, Spiral Divergence increase |ez,| to improve spiral stability 9.1.5.9 Summary of Linear Lateral-Directional Dynamics. The linearized lateral- directional equations of motion illustrate clearly just how complex asymmetric motions are. ‘The approximate equations are even more inaccurate than they were for the symmetric motions. Though these approximations are useful for planning fight tests, they must be used with discretion. 9.2 DYNAMIC FLIGHT TEST METHODS, Flight tests designed to determine dynamic characteristics of airplanes are some of the most challenging. The flying precision required and the care in data reduction required are primarily responsible for this challenge. Moreover, the subjective character of closed loop handling qualities is tied very closely to open loop dynamic. characteristics of the ve- hicle and its control system; so, cooperation and understanding between the pilot and the flight test engineer are absolutely essential to obtaining meaningful results from such tests. First, consider how open loop dynamics can best be excited. 9.2.1 Types of Control Inputs An infinite number of inputs could be used; indeed, random excitations from atmos- pheric disturbances excite the dynamics when encountered. However, most linear dy- namic theory is built around three ideal inputs — step inputs, pulse inputs, and doublets (Fig. 9.13). Except for the phugoid and the spiral modes, dynamic motions are usually ex- cited by rapid pilot inputs. No control system can provide an infinite rate of surface move- ment; ideal inputs cannot be achieved. Even if the control system could provide such rates, pilots cannot perfectly retum the surface exactly to its trim position for stick-fixed in- puts. Pilot inputs are not repeatable from test point to test point, much less from pilot to pilot. Some fiight test programs now use automatic flight control system inputs to excite dynamic modes. Repeatability is improved when electromechanical inputs are pro- grammed into the fight profile. The Shuttle Orbiter and the X-29 are two test programsChapter 9 Dynamic Stabitity Tests 203 where such programmed inputs were used to advantage. Both the test pilot and the test engineers on the ground simply monitor the resulting motion. Ideal Actual “~~ ~- SINGLET er PULSE Time Time Fig. 9.13 Control Inputs for Dynamic Testing 9.2.1.1 Step Input. Ideally, a step input is a movement of the control surface to the desired position in zero time. The surface then remains in that new position, either be- cause the pilot holds it there or because the surface has no tendency to float with aero- dynamic forces (as in an irreversible control system). Of course, if the stick or rudder is freed and the trim mechanism moves the stick or rudder back to the original position, the input is no longer a step. ‘The final level is not maintained. No real control system can provide an input in zero time; therefore, as Fig. 9.13 shows, there is a finite slope to the actual input. Naturally, a step input has the undesirable effect of setting the contro! sur- face to @ new trim position, which causes the vehicle to take on a new equilibrium state. ‘An elevator step input will cause an airplane to stabilize in a new attitude and at a new trim airspeed some time after the step input is made. In other words, any oscillation that is ex- cited by a step input will not retum to the original equilibrium state. Dealing with these two equilibrium states in response measurements can be tedious. 9.2.1.2 Singlet As Fig. 9.13 also shows, a singlet (often called a pulse) is merely a step input, followed a short time later by a negative step input of the same size but op- posite sense. The surface deflections occur instantaneously in an ideal singlet. Again, such infinite rates of movement of any physical mechanism are impossible. 9.2.1.3 Doublet. A doublet, merely a singlet followed immediately by a second pulse opposite in sense to the first one, is an even more common technique used to excite dy- namic modes of motion. A doublet is ordinarily used when an oscillation exactly (or as close @s possible) about the original equilibrium point is sought. For quick evaluation of dynamic characteristics without mathematical analysis, such a time history is easy to inter- pret. As can be seen from Fig. 9.13, the doublet is a periodic function if performed per- fectly. This picture makes it easy to visualize the frequency content of this type of input excitation. As you might expect, the nature of the response depends on the frequency of this input. We will have more to say about this frequency dependence when we tum our attention to the longitudinal modes of motion. Let us consider first the phugoid, which has little dependence on the nature of the input; in fact, one does not even have to use the primary control surfaces to excite this oscillation.204, Dynamic Stability Tests Chapter 9 9.2.2 Phugoid Test Methods The first step in setting up for virtually any of the dynamic tests is to carefully trim the airplane at the desired test conditions. The engineer should insist that the desired equi- librium conditions are attained before any excitation is applied. Of course, there are some types of tests where the airplane cannot be trimmed completely. For example, the Shuttle Orbiter (or any other glider, for that matter) cannot be trimmed for constant airspeed and altitude. So, data must be collected in a dynamic environment. However, most of the elementary test methods described in this book start from true equilibrium conditions and the pilot must trim the airplane with care. A good rule of thumb is that, if airspeed and air- speed can be maintained within 0.5 KIAS and 20 feet for 10-15 seconds, trim is satisfacto- Tily established. The pilot must make no control inputs during this time. The phugoid test is one of the tests for which such a careful trim point is essential. A singlet of fairly long duration is used to excite the phugoid. Actually, the duration of the singlet is not critical; it is best to simply hold the initial step in long enough to reduce (or increase) the indicated airspeed by some set amount (approximately 5% of the trim in- dicated airspeed is a good starting estimate). Care should be taken to retum the control surface as nearly as possible to the original trim position; otherwise, the oscillation will have a climb or dive superimposed on it and the analysis is a bit more difficult. Depending ‘on which type of measurements are sought, the pilot may fix the stick after retuming it to the neutral position (stick-fixed oscillations) or he may release it after retuming it to the neutral position (stick-free oscillations). Both types of oscillations should be examined if the control system is reversible. Of course, the airplane's free response is sought, the pi- Jot must make no longitudinal inputs after returning the stick to the neutral position. How- ever, small lateral control inputs to keep the wings level are allowed while the oscillation is being recorded. But, bank angles as small as 5°-10° may affect the phugoid damping ratio (Gp) and its undamped natural frequency (aq, ). For some vehicles, it is more convenient to reduce the airspeed using speed brakes or even power than by elevator or horizontal tail inputs alone. Any method that will start the exchange between potential and kinetic ‘energy is satisfactory. As with all dynamic parameters, it is best to record the phugoid oscillation with some form of automatic recording device, though this oscillation is usually of low enough fre- quency that the pilot can time a period with a stop watch. Airspeed, altitude, or pitch atti- tude all give accurate phugoid oscillation parameters with their response character. If you lack instrumentation, a video tape or motion picture camera of the instrument panel will provide an accurate measurement of the phugoid parameters. One trick to obtain repeat- able periods for the phugoid is to mark the time when the vertical velocity indicator passes through zero rate of climb or descent. Of course, the period of the oscillation is between the adjacent times for zero rate of climb when the indicator is moving in the same direc- tion. You should not start timing to obtain this parameter until after the short period (if itis excited by the input) has damped out. Typically, a half cycle of the phugoid is quite ade- quate for the short period mation to be negligible; so timing for the phugoid should start no sooner that one phugoid half-cycle after the input excitation is removed. Calculated phugoid time histories are shown in Fig. 9.14. We will defer discussing how to extract phugoid damping ratios and frequencies until we have reviewed meas-Chapter 9, Dynamic Stability Tests 205 urement of lateral-directional dynamic modes. Notice that the phugoid oscillation is lightly damped in all cases and that ©, increases as altitude increases. 9.2.3 Short Period Test Methods The pilot is quite sensitive to short period frequency and damping; these parameters have a strong influence on pilot ratings for virtually any precision flying task. The pilot ‘senses dynamic parameters through both visual and tactile perceptions. He observes the pitch attitude variations, he feels the normal accelerations, and he visually records infor- mation from all available indicators (angle of attack indicator, accelerometer, or pitch rate indicator). All of these cues are integrated in the human brain to give frequency response information about the airplane's dynamic behavior. The stability and control manuals”? from the test pilot schools have more complete discussions of how each of the dynamic parameters affects pilot opinion of an airplane. Perhaps the most complete document dis- cussing this important aspect of airplane dynamics is Chalk’s work®. This document is “tequired” reading for those regularly engaged in fiying qualities tests. —_— AAD Skyhawk Simulation M = 0.2 (223 fps) Sea Level ] M=07 (681 fps) } ee 35,000 feet essnstcfeneeee 0 20 40 60 80 100 120 140 160 180 200 Time — Seconds Fig. 9.14 Phugoid Oscillation206, Dynamic Stability Tests Chapter 9 The most common input used to excite the short period is the doublet discussed pre- viously (see Fig. 9.13). This type of input is usually quite good for exciting the short period while suppressing the phugoid, simply because the input begins and ends at the trim pitch attitude at very nearly the trim airspeed if performed correctly. Starting the doublet with a nose down pitch rate first is slightly more comfortable for most pilots, but satisfactory data can be obtained with the opposite initial pitch rate. This periodic input causes transient deviations in pitch rate, normal acceleration, angle of attack, and pitch attitude that are as- sociated with the short period motion. typical short period oscillation for a light twin is shown in Fig. 9.15. 15 cee yer paces 5t i pi 0 2 4 Time — Seconds Fig. 9.15 Short Period Response to a Doublet for a Light Twin ‘The airplane response must be measured after the control inputs have ceased; but, unlike the phugoid measurements, the important part of the time history is immediately af- ter the doublet is completed. For that reason it is important that control inputs be recorded simultaneously with the other dynamic parameters. It should also be obvious that record- ing devices are essential for obtaining quantitative measures like short period damping ra- tio, undamped natural frequency, and damped natural frequency (C5p. ng, ANd yyy FE spectively). The pilot can obtain a rough estimate of Sep by counting the number of peaks of the oscillation in pitch attitude until he can no longer discem an oscillation. If the damping ratio is 0.1 < Cyp < 0.7, as itis for most aircraft,Chapter 9. Dynamic Stability Tests 207 pa ~ number of peaks counted 10 The singlet can also be used to excite the short period mode of motion. Unfortu- nately, especially for slow-responding airplanes like most commercial airliners and most bombers and tankers, this type of input also excites the phugoid and it is not easy to sepa- rate the two quadratic factors in analyzing the data. It is useful for quick-responding air- planes that also have enough pitch damping to cause the short period to subside before the phugoid oscillation develops. Finally, a form of the step input can also be used to excite the short period when the airplane has a low frequency short period and/or heavy damping of this oscillation. It is especially appropriate for this type of airplane when a larger amplitude motion is needed for accurate data reduction. The disadvantages of this excitation method are that it re- quires more test time and maneuvering by the pilot. It will ikely not be quite as repeatable 2s the simpler doublet form of excitation. This technique is sometimes called the "2g pull up" method, though of course any normal acceleration level consistent with the airplane limitations and the pilot's skill could be used. After carefully trimming (as always), airspeed is traded for altitude and then the pilot pushes over to establish a shallow dive. The dive angle must be adjusted for each configuration and for different points in the operating en- velope. This adjustment is often an iterative process and adds to the amount of test time that this excitation method requires. Trim altitude should be approached in a steep enough nose-down attitude so that the pilot can pull up sharply to the desired normal ac- celeration and establish a constant pitch rate. As the airplane approaches the initial trim attitude in this rotation, the pitch rate should be constant, normal acceleration should be constant at approximately the desired value, and the altitude should be very close to the initial trim altitude. At that point the longitudinal controller should be smartly neutralized and either fixed in the neutral trim position or freed at that position. ‘The use of the traditional second-order dynamic system measures of merit (like damping ratio and undamped natural frequency) alone has led to inexplicable discrepan- cies between pilot opinion ratings and these measures of merit. The military specifica- tions3.8 state requirements in terms of another parameter, a The ratio of maximum pitching acceleration to steady state normal acceleration during maneuvers is roughly 2 © equal to the ratio a . This parameter seems to indicate initial response better than ‘a. Sep and «,, alone. Obtaining this parameter for each test condition is often a simple task if analog recording devices are available or if digital systems have a high enough sample rate. If the normal acceleration and the angle of attack at any desired flight condition can be recorded, obtaining © is a simple division. ‘One way to obtaining such data is a near- sinusoidal pumping of the longitudinal controller while recording normal acceleration and angle of attack. The frequency of the periodic input can be varied to cover the entire range of frequencies of interest. These data can also be used to determine the minimum transient stick force per g ratio. This ratio is usually obtained when the stick is pumped at208, Dynamic Stability Tests Chapter 9 a frequency close to wp, Care must be taken by the test pilot in making such inputs to keep them small enough to avoid overstressing the airplane when the input frequencies are close enough to an,, that they might excite a resonance or lead to pilot-induced oscillations. Use very small amplitude inputs and build up to larger ones only if needed for data analysis when conducting such a test. Having completed our discussion of longitudinal dynamic test methods; we now tum to lateral-directional measurements. 9.2.4 Dutch Roll Test Methods The only remaining oscillatory motion normally encountered in dynamic flight testing is the coupled Iateral-directional oscillation. The other lateral-directional modes are typically aperiodic. For that reason we will defer discussion of the test methods associated with the spiral mode and the roll mode until after discussing the Dutch roll. This oscillation can be quite uncomfortable to both crew and passengers and may seriously affect the utility of an airplane if it is objectionable. Swept-wing commercial air- liners often need yaw dampers to augment the natural damping of the aircraft. Otherwise, Passengers might even become ill from Dutch roll excited by atmospheric turbulence. Military ground attack aircraft may not allow precise tracking of targets at low altitudes if this oscillation is not damped artificially. Dutch roll, ike the short period, is most commonly excited by a doublet, though other forms of inputs can be used. The rudder pedals are usually the pilot controller used to input this doublet. If the pilot input is not quite symmetric, it is very easy to excite the roll mode and/or the spiral mode and make data analysis difficult. Also, the test team must keep the dynamic sideslip limitations of the airplane in mind when performing deliberate Dutch rolls. Finally, on vehicles operated at high dynamic pressures, deliberately inducing Dutch rolls can load the vertical surfaces highly. Several test aircraft have lost parts of the vertical tail or the rudder due to excessive aerodynamic loads during such dynamic ‘overshoots. Appropriate care must be exercised; loads may be monitored in real time if the aircraft is highly instrumented and has telemetry capability. An aileron singlet can also be used to excite the Dutch roll. Ordinarily, such a ma- neuver is started with the airplane at some steady bank angle. Aileron is applied as rap- idly as possible (near a step input) to roll out of the tum. Opposite aileron is abruptly applied as the wings approach level flight. For some aircraft, like those with large mass and high inertias, this technique often provides more realistic amplitudes of Dutch roll ‘oscillation than do rudder doublets. This input, like the rudder doublet, may end with the rudders fixed or free in the neutral position. The Dutch roll can be also be excited by another form of the singlet, though this con- trol input involves the use of two controls simultaneously. The controls are abruptly re- leased from a steady, straight sideslip by simultaneously neutralizing both the rudder and the ailerons (and then either fixing or freeing both of these controls at the neutral point). This technique is applicable for some airplanes for which a single contro! input (either rudder doublet or aileron singlet) does not adequately excite the Dutch roll oscillation. Ordinarily, it is only needed if the Dutch roll is highly damped. Releasing the controls abruptly from a large sideslip can produce unexpected results, from dynamic overshoots ‘of maximum allowable sideslip angle to structural overloads on the vertical tail. Such safety considerations dictate that the pilot should build up cautiously by starting with small values of f and then increasing B incrementally until an oscillation with satisfactory char-Chapter 9, Dynamic Stability Tests 209 ‘acteristics is achieved. Do not attempt this form of excitation from maximum static side- slip angles without such a preliminary buildup. Whatever the type of input, quantitatively analyzing Dutch roll measurements also re- quires recording devices to be on the airplane. Like all the other airplane oscillations, ex- tremely high frequency data recording is not essential. Frequencies of interest in these motions ere usually on the order of 10-25 Hz; 0, digital recording is possible with rela- tively low cost instrumentation. Do not forget, however, that the sampling theorem states that discrete sampling rates must be at least twice that of the highest frequency of inter- est. Most flight test agencies target a sampling rate an order of magnitude greater than the highest motion frequency being studied. ‘A typical Dutch roll oscillation is depicted in Fig. 9.16. This particular test sequence was initiated by a rudder doublet. Notice that the aperiodic lateral-directional modes affect this oscillation. It is virtually impossible to excite the Dutch roll without exciting the other two modes of motion. To extract Cor and @np, from such response time histories, ‘we must first understand how to test for the spiral mode and the roli mode. 15 20 10 Time - Seconds Fig. 9.46 Dutch Roll Oscillation Excited by a Rudder Doubiet 9.2.5 Spiral Mode Test Methods ‘The first aperiodic mode to be considered is a slow divergence or convergence in heading when the airplane is displaced in roll from the wings level equilibrium state. Ex- citing this mode requires no control input. After trimming carefully at the desired test conditions, simply roll the aircraft to a bank angle of approximately 10° and stabilize in the ensuing tum. Then, either fix or free the ailerons and the rudder (the elevator may be used to maintain airspeed constant). If the airplane tends to roll out of the tum, the spiral mode is stable (convergent). The time-to-half-amplitude can be measured by timing how tong it takes to roll from @ 20° bank to a 10° bank. Conversely, if the airplane tends to roll further into the bank, the mode is unstable and the time-to-double-amplitude can be210____________ Dynamic Stability Tests Chapter 9 obtained by noting how long it takes to proceed from the original 10° bank condition to 20° bank. When using this test method, care should be taken to avoid bank angles greater than 20°. If ¢ gets larger than this value, the linearizing assumptions used in the small perturbation equations are no longer valid. For multiengine airplanes, care must be taken to ensure that all engines are producing the same thrust; otherwise, the asymmetric moment produced by the engines will overpower the spiral mode behavior. The spiral mode usually has such a long time constant that it has little effect on the handling qualities of a conventional airplane even if itis unstable. This point is ilustrated in Table 9.9°. Notice that the minimum times-to-double-amplitude are significantly longer than the usual Dutch roll period, but if they are shorter than indicated, pilots object be- cause the spiral mode will require attention during precise maneuvering tasks. Such additional pilot workload is particularly unwelcome during any precision maneuver that ‘occurs over a long time period, like instrument approaches or takeoffs and landings. Table 9.9 Typical Spiral Stability Requirements: Time-to-Double-Amplitude (seconds) Class Flight Phase Category __Level 1 Level 2 Level 3 tand iv A 12 12 4 BandC 20 12 4 Mand it Al 20 12 4 ~Adapted from Chalk? Fig. 9.17 Roll Mode Tests: 45°-45° Bank Change for a Light TwinDynamic Stability Tests 211 Chapter 9. 9.2.6 Roll Mode Test Methods The roll mode is most often excited by a step or a singlet aileron input. Figure 9.17 ilustrates the results of a roll mode test excited by a step input for a general aviation light twin. Procedurally, the most efficient way to collect roll rate measurements is to roll from a chosen bank angle to the same bank angle of the opposite sign; once stabilized at this opposite bank angle, roll back in the other direction with an identical step aileron input of the same magnitude. Temporary stops are often used to keep the aileron inputs to ap- proximately the same size and to make them precise and repeatable. Of course, the usual necessity for trimming precisely at the desired fight conditions applies to this test ‘also. The choice of the initial bank angle to use largely depends on the type of aircraft being tested. Typically, fighters, trainers, and other smaller airplanes are more maneu- verable and 45° is used. This bank angle change was used for the light twin in the ma- neuver depicted in Fig. 9.17. For larger aircraft, 30° is more appropriate. Some require- ments documents specify times to roll through 60° or 90°; if these data are also sought, be sure the pilot carries the roll slightly beyond the bank angle of opposite sign before stopping the roll. Otherwise, the times will be inaccurate. For timing such rolls, recording bank angle and surface position time histories is highly desirable, though stop watch tim- ing of such rolls gives approximate information if instrumentation is meager. In any case, you must record such time histories to extract the dynamic information; the roll mode time constant is almost always too short to be observed manually. It is also useful to record variations in airspeed during roll mode testing, especially as you build up to maximum aileron defections. It is permissible to use the longitudinal control (elevator) to help maintain airspeed during rapid rolls. Such roll tests should be carried out in both direc- tions, since there are often factors that produce different roll dynamics in each direction. 9.2.7 Basic Data Reduction Methods ‘As promised earlier, we now tum our attention to data reduction techniques. In this introductory text we will confine our study to simple methods, techniques that would have been called "hand-calculations”. Of course these basic tools could be implemented on @ ‘computer, but before you automate a process, you need to understand it. So, we will de- scribe the data manipulations in graphical terms to make the material easy to understand. Most of the calculations can be readily automated; in fact, homework exercises used by the author in formal classes often ask students to write simple programs to obtain perti- nent dynamic parameters from response data. Much of the material in this section is adapted from Appendices III and VB of Chalk’s report®. Table 9.10 Applicability of Basic Graphical Techniques For Determining ¢ and @p eee ee eee ee Name of Method Range of Applicable Damping Ratio Transient Peak Ratio (TPR) =05<6<05 Modified TPR (MTPR) -05<6<05 Time-Ratio (TR) 05 = qs If the ratios are not 4 identical (and they typically are not, for actual flight test measurements), an average of the measurements is often used. Whatever the procedure, once the TPR is selected, the damping ratio can be read from a chart like Fig. 9.19. Finally, op comes from the meas- urements, since the damped period, T, is scaled as the time between the local maxima or minima or is the measured time between the steady state crossings (Fig. 9.188). Then, 2n he 10 a 08 07 06 Transient Peak Ratio— TPR 0.01 0.02 0.03 0.04 0.08 0.08 0.1 02 03 04 06 0810 Damping Ratio Fig. 9.19 Transient Peak Ratio Versus Damping Ratio 9.2.7.2. Time-Ratio Method. To use the time-ratio (TR) method, you must be able to ascertain steady state equilibrium after the excitation has subsided. This method, de- scribed by Dolbin'S, ufilzes three specific values along the output response curve to calculate three estimates of ¢, at selected fractions of the change in the output variable. ‘As shown in Fig. 9.20, these times are measured where x, the output response, is 0.796Ax, 0.406Ax, and 0.1994x below the final steady state value, respectively, for At;, At, and Af, Then the ratios Af,/At,, Afy/At,, and (At; - Af,V(At, ~ At,) are calculated and used to enter the chart shown in Fig. 9.21.214, Dynamic Stability Tests Chapter 9 4 0.194% | $0406 6x9 736 ax 1 - at ; lool | |—ats—e} Fig. 9.20 Measurements for TR Method [Frequency Time Product Ate 14 Damping Ratio ¢ Fig. 9.21 Chart for Determining ¢ and @, by TR Method Figure 9.21 yields three estimates of ¢. We could simply average these three numbers, though it is a good idea to weight the ¢ according to how much the oscillation measurements are distorted from a pure second order system. Generally, At, is the least accurate of the time measurements because it is quite sensitive to our choice of fp Moreover, the response is often affected significantly in this region by contro! system dynamics, which introduce higher order responses. On the other hand Af, can also be in error due to uncertainty in choosing the final steady state value. These factors mean that you should use some engineering judgment in considering which damping ratio to choose. But, the TR method does provide you with three checks. Finally, the undamped naturalChapter 9. Dynamic Stability Tests 215 can also be determined straight from the chart in Fig. 9.21, just as the values for ¢ were obtained. The only difference is that o,f, yf, and o,f, are read from the ordinate of the chart and then divided by the appropriate time to obtain three estimates of @,. Naturally, the remarks about use of engineering judgment in either averaging or choosing the best estimate of o, are still applicable. Maximum Slope ‘Ax Tangent Line A — 8 I Lt 0.08 0.12 0.16 0.20 0.24 0.28 032 Ax ax Fig. 9.23 Chart for Determining ¢ and @, by the Maximum Stope Method 9.2.7.3 Maximum Slope Method. The maximum slope (MS) method was devel- ‘oped'S to reduce the inaccuracies in the TR method that are associated with determining the steady state response. Figure 9.22 shows the basic parameters to be taken from the response time history for the MS caiculation. In this approach, the peak amplitude of the response and the maximum slope of the tangent to the transient rise are used instead of intermediate times (as in the Time Ratio method). However, the MS technique has its own216. Dynamic Stability Tests Chapter 9 inherent error source: the amplitude Ax, is usually very small and difficult to measure ac- curately. This error effects ¢ mostly; @, is largely unaffected by this uncertainty. The ra- tio cam is then calculated and used to enter Fig. 9.23, where both ¢ and @,AT can be obtained graphically. Then, rather simplistically, @q = onl 9.2.7.4 Separated-Real-Roots Method. As Table 9.10 indicates, the separated-real- roots (SRR) technique was developed to evaluate responses composed of two unequal real roots. It is useful only if the damping ratio is large enough to ensure strongly con- vergent aperiodic modes of motion. Typically, the roll mode from the lateral-directional quartic is characterized by such behavior. Therefore, it is often analyzed using the SRR technique, especially if the spiral mode is also convergent. Figure 9.24 sketches out the parameters to be taken from the output response measurements. Once again, it is nec- essary to select the steady state output and the method is moderately sensitive to how ac- curately this value can be determined. Moreover, with this approach you must also plot ‘Ax(t) on semilog paper. After the faster root (f2) has decayed, the semilog paper will be a straight fine, whose slope determines the lower root (ty). Then, f2 =H, where Axo 0 is_the intercept of the straight line on the semilog graph. Altematively, t2 can be deter- mined from the slope of a semilog plot of Ax(f) versus f. Fig. 9.24 Measurements for the Separated-Real-Roots Method Whichever of these two graphical methods is used to obtain ty and tz, the damping ratio and the undamped natural frequency come from: afi ttedon eee get? and on = Vee To help reduce the sensitivity of the SRR procedure to errors in graphically estimating the steady state response, a modification (due to Dolbin'§) may be helpful. The values ofChapter 9. Dynamic Stability Tests 27 ‘Ax(0 are simply measured in such a way that determining the steady state response level is unnecessary. They are measured at equal time increments as suggested in Fig. 9.258. Then, each value of is Ax(f) simply: x(t + AT) - x(P), where AT= (ty to) = (la-ty) = *+* = (tet ~ ta) " Sx2 368Ax, (0) : Fig. 9.25 Measurements for Dolbin's Modified ‘Separated-Real-Roots Method Plotting Ax() on the semilog scale as suggested for the basic SRR method leads to the same result as before, except that no steady state value for the response had to be graphically determined. There is usually more scatter in the data points on the semilog plot with this modified approach, but the slope of the faired line is often more accurate. If you are using the unmodified SRR technique and assuming that ¢ > 1, but it is actually less than 1, you may not be able to find a part of the semilog plot that can be adequately fitted with a straight line. It will probably be obvious that some sort of error has occurred. With the modified method, this sort of faulty assumption is not as likely to be noticed. You can usually fit a straight jine with reasonable scatter to these points. However, with the two real roots that you extract from these modified calculations, ¢ and , will be grossly in error. If you suspect that the damping ratio is only slightly greater than 1, check the re- sults of the SRR method using the TR or the MS methods. Example 9.3: ‘The time histories on the next page were generated from a simulation of an F-89 Scorpion flying at 20,000 feet on a standard day at a Mach number of 0.638 (660 fps). ‘The airplane weight was 30,500 pounds. Graphically estimate the short period damping ratio and the undamped natural frequency the damped. “To graphically estimate the required damping ratio and undamped natural frequency, we must greatly expend the scales to have any hope of reasonable accuracy. We will choose the region where the re- sponse trace just drops into the negative range end analyze the overshoot characteristics. However, for the time ratio method we also must know the steady stale value and we notice from the above chert that itDegrees (Elevator Deflection) or Degrees/second (Pitch Rate) Time - seconds ‘Next, we measure from this steady state value for pitch rate down to the minimum pitch rate (as shown in the figure below) to get a Ax of approximately 0.568 degrees/second. 06 Toto 7 0.4064 x ~= 0.231 ‘/sec : Fi 0.736Ax = 0.418 ‘/sec Time - seconds Next, draw the three horizontal lines in that corespond tothe three levels of overshoot from the steady state value of q suggested in Fig. 920 (0.199 Ax, 0.406 Ax, and 0.736 Ax), Where these overshoot levels intersect the response curve, drop three vertical lines to the time axis and graphically moasure the ind ccated time increments from the time when the minimum q was achieved (the peak overshoot time). For this ‘example those values are shown on the chart above as: At, ~ 0.196 seconds, tz = 0.338 seconds, and ‘ts 0.498 seconds. Forming the ratios necessary to enter Fig. 9.21 M2 172. which gives ¢ ~0.50 fom Fig. 921; Gia 22 which gives (0.51 from Fig. 9.21; and i Als ~ Atty Alz—Aty ‘= 070 which gives ¢~ 0.55 rom Fig. 9.21. Averaging these values, Caug = O52Chapter 9, Dynamic Stabiilty Tests 219 Reading from the other solid lines in Fig. 9.21 (the Frequency Time Product lines), we obtain the fol. owing values for a7 Af and then @,: op Aty » 0.865, oy Aty » 1.46, and on Aty ~ 1.90. Dividing by the appropriate times and averaging agein: Saag ~ 455 120/90) ‘Now we turn to estimating the same parameters using the maximum slope method. We stil need the ‘expended scale drawing of the response curve, which is repeated below. For this procedure start by 0.10} ---- eevee B&Bs (Normalized Roll Rate) 20.05 plp 2 z 0085 6 7 8 9 10 + 2 3 4 5 Time — seconds 30 Semilog Plot to Estimate Dutch Roll Mode Damping Ratio Finally, the estimated Dutch roll envelope is transferred back to the combined roll- mode-Dutch roll mode time history (Fig. 9.29) to estimate the roll mode time constant. Often, itis useful to enlarge the first few seconds of the oscillation, as shown in Fig. 9.31. ‘The Dutch roll mode period and damping ratio are now known; they are used to re- construct the initial part of the Dutch roll component (enlarged scale of Fig. 9.31). This reconstructed oscillation is subtracted graphically from the combined roll-Dutch roll mode ‘component time history to yield the roll mode time history trace. This estimate for the roll mode component of the time history is also illustrated in Fig. 9.31. For the interested stu-24 Dynamic Stability Tests Chapter 8 dent, an altemate method of determining the initial Dutch roll mode that utilizes potar plots is given by Chalk. In the interest of brevity, that alternative will not be covered. 4 2 8 z & 8 Pl, (Normalized Roll Rate) 8 8 yaa 6 x <8 Fig. 9.31 Estimation of the Roll Mode Component The roll mode time constant comes from Fig. 9.31. Calculating 0.368K, and drawing a horizontal line at that value of normalized roll rate, the time where this horizontal line inter- sects the estimated roll rate time history is ,. We could also go back to the semilog plot and find K; as the intersection of the straight line at f = 0 seconds, just as we did for the spiral mode. Another way to obtain +r, is to simply calculate it from a time increment and that —t) the natural logarithm of the corresponding amplitude ratio: t, “ait Both al- nt 7 Pk temative methods are illustrated in Fig.9.32. 0.30 0.0 O41 02 03 (04 05 06 07 Fia. 9.32 Alternative Calculation of the Roll Mode Time Constant | |Chapter 9, 9.3 SUMMARY Dynamic fight tests are one of the most challenging types of tests you will encounter as a fight test engineer. They require a thorough understanding of stability and control principles, a moderate knowledge instrumentation that can record the necessary parame- ters automatically, insight into handling qualities and how pilot ratings were developed, knowledge of the types of standard inputs (pilot-induced or not) that produce responses adequate for analysis, and at least an introductory understanding of how to extract dy- namic measures of mert from response data. This chapter has explained each of these fundamentals. The serious student of flight test dynamics should continue his study with more training in experimental instrumentation, sampling theory, and in parameter estima- tion. Each of these topics offers the fight test engineer more advanced tools. However, the goal of this chapter has been met in introducing the subject of dynamic measurements.226 ynamic Stability Tests Chapter 9 REFERENCES, ‘Anderson, J. D., Introduction to Flight (3rd Edition), McGraw-Hill Book Com- pany, New York, . “Fixed Wing Stability and Control: Theory and Flight Test Techniques," USNTPS- FTM-No. 103, Naval Air Test Center, Patuxent River, Maryland, Jan. 1975 (Revised Nov. 1981). "Military Specification, Flying Qualities of Piloted Airplanes,” MIL-F-8785C, Nov. 1980. Etkin, B., Dynamics of Flight - Stability and Control, (2nd Edition), John Wiley & ‘Sons, New York, 1982. Roskam, J., Airplane Flight Dynamics and Automatic Control, Part 1, Roskam Aviation and Engineering Corporation, Lawrence, Kansas, 1979. McRuer, D., Ashkenas, nd Graham, D., Aircraft Dynamics and Automatic Control, Princeton University Press, Princeton, New Jersey, 1973. Kuo, B. C., Automatic Control Systems (6th Edition), Prentice-Hall Inc., Englewood Cliffs, New Jersey, 1991. Chalk, C. R., et al, “Background Information and User Guide for MIL-F-8785B (ASG), Military Specification - Flying Qualities of Piloted Airplanes," AFFDL-TR-60-72, Air Force Flight Dynamics Laboratory, Wright-Patterson AFB, Ohio, Aug. 1969. "Stability and Control Flight Test Theory," Vol. |, Chapter 6, AFFTC-TIH-77-1 USAF Test Pilot School, Edwards AFB, Califomia, Revised Feb. 1977. Cooper, G. E. and Harper, R. Jr., "The Use of Pilot Rating in the Evaluation of Aircraft Handling Qualities," NASA TN-D5153, Apr. 1958. Harper, R. P., Jr. and Cooper, G. E., "Handling Qualities and Pilot Evaluation,” Jour- nal of Guidance, Control, and Dynamics, Vol. 9, Sept-Oct. 1986, pp. 515-529. Blakelock, J. H., Automatic Control of Aircraft and Missiles (2nd Edition), John Wiley & Sons, Inc., New York, 1991. lif, K. W, "Parameter Estimation for Flight Vehicles," Journal of Guidance, Control, and Dynamics, Vol. 12, Sept.-Oct. 1989, pp. 609-622. Lung, L., System Identification: Theory for the User, Prentice-Hall Inc., Englewood Cliffs, New Jersey, 1987. *5 Dolbin, B.H., "Study of Some Hand-Computing Techniques to Determine the Ap- proximate Short Period Mode from Airplane Responses,” FDM No. 371, Comell Aeronautical Laboratory, Inc., Mar. 1966. Neal, T. P., "Frequency and Damping from Time Histories: Maximum Slope Method," Journal of Aircraft, Vol. 4, Jan.-Feb. 1967, pp. 76-78. Nelson, R. C., Flight Stability and Control and Automatic Control, McGraw-Hill Book Company, New York, 1989. “Military Standard, Flying Qualities of Piloted Aircraft," MIL-STD-1797A, Jan. 1990. 2Chapter 10 PRINCIPLES OF AEROELASTICITY Aeroelasticity'? is the study of the interaction between fluid forces, structural (elastic) forces, and inertial forces. The three-ring diagram shown in Figure 10.1 helps to under- score the interdisciplinary nature of aeroelasticity. MECHANICAL VIBRATIONS Fig. 10.1 Interdisciplinary Nature of Aeroelasticity ‘The combination of any two rings represents @ discipline of aerospace engineering. ‘The combination of elastic and inertial forces forms the study of mechanical vibrations. The combination of inertial and aerodynamic forces forms the study of rigid-body vehicle stability and control. ‘The combination of aerodynamic and elastic forces forms the study of static aeroelasticity. The combination of all three rings forms the study of dynamic aeroelasticity. Other disciplines such as control, thermal, and damage (aging aircraft) is- ‘sues may be considered as additional overlapping rings. 10.1 ANALYTICAL FOUNDATIONS 10.1.1 Aeroelastic Definitions As in most specialties within aerospace engineering, aeroelasticity has its own jargon ‘and special use language. To begin, we need to establish a common set of terminology for the phenomena unique to aeroelastic analysis. 10.1.1.1. Agroelastic Divergence.28 i rinciples of Aeroelasticity Chapter 10 10.1.1.2. Control Reversal Control reversal is the loss of aerodynamic effectiveness for a control surface |due to deformation of the primary structure to which the surface is attached. 10.1.1.3 Lift Effectiveness. [Lift Effectiveness is the change in eerodynamic loads due to wing flexibility, which typically leads to a change in the load distribution and an increase in lift. ‘Two characteristics of aeroelasticity are evident in the preceding definitions: (1) aero- elasticity is dependent upon the flexibility of the vehicle's structure, and (2) aeroelasticity involves a feedback between aerodynamic loads and structural deformations. 10.1.1.4 Flutter. Flutter is a dangerous structural dynamics phenomenon. From a flight test perspective it is definitely a hazardous type of flight envelope expansion. Flutter is a dynamic instability in which the structural modes that are affected “| |erodynamic loads coalesce to form a single aeroelastic mode that rapidly grows with diverging oscillatory motion to structural failure. Aerodynamic loads, especially those that are unsteady in character, are the source of excitation energy for this destructive type of structural dynamics. Two or more interacting modes of motion merge — dependent upon flight conditions — to form a single aervelastic frequency and mode of motion that resembles a combination of the participating modes. From an energy perspective, energy is extracted from the aerodynamic flow field and is absorbed by the structure. For subcritical conditions, structural and aerodynamic damp- ing dissipates this energy; for critical conditions, system damping is lost thereby leading to a growing oscillatory motion that results in catastrophic flutter. 10.1.2 Agroslastic Equations of Motion In general, the equations of motion for the aeroelastic system may be cast in a form similar to that of the forced mass-spring-damper system discussed in Chapter 9. Writing them in matrix form Md +Cq+Kq=Fa (10.1) where M, C, and K are the matrices that represent the structural mass, damping, and stiffness, respectively, and q is the vector of generalized coordinates. In aeroelastic sys- tems, Fa denotes the time-dependent, motion-dependent (that is, the "unsteady") aerody- namic loads, Fa=Mad+Cag+Kaq+F (10.2) where Ma, Ca, and Kq depict aerodynamic inertial, damping, and stiffness contributions which depend upon fiowfield conditions (Mach number, altitude, and velocity) as well as the motion of the vehicle's structure, and F denotes the extemal loads such as those arising from gusts, turbulence, and buffeting. Equations 10.1 and 10.2 may be combined,Chapter 10 Principles of Aeroetasticity, 29 (M-Ma)d+(C-Ca)d+(K-Ka)a=F (10.3) This expression may be solved in either the frequency domain or the time domain. For the frequency domain approach, harmonic motion is assumed so that q=Be™; thus, substitution into eqns. 10.3 gives: (2 (w-Ma)+2€-Ca)+(K-Ka))B=0 (10.4) F is not considered in this homogeneous solution. Equations 10.4 are solved as eigenvalue problems in which the eigenvalues are complex numbers; hence, 1=8+ jo, where & represents the aeroelastic damping and w represents the aeroelastic frequency. B is the aeroelastic mode shape associated with the complex eigenvalue. Thus, the re- ‘sponse of the structure of the vehicle is q = Bee**, where o* represents a decaying ‘oF growing motion depending upon the sign and e* represents a constant amplitude os- cillatory motion. We note that the possibility of negative damping comes from the pres- ence of aerodynamic forces. Three important trends must be monitored during fight tests: (1) changes in frequency characteristics for changes in fight conditions, (2) changes in damping characteristics, and (3) changes in amplitude of response. ‘These changes are related to the motion described by e” and e*. — 8 | 0 ~ t OE (a) Exponential Rise (b) Exponential Divergence —_(c) Undamped Sinusoid to Steady-State (Unbounded) (Neutral Dynamic Stability) Fig. 10.2 Oscillatory Motion ‘The eigenvalues indicate damping characteristics dependent upon different dynamic pressures (or other flight conditions). In traditional mechanical vibration analysis the natural frequency is given by the imaginary part of the root. The real part, which repre- sents the damping, was typically zero. However, the presence of unsteady, motion- dependent aerodynamic forces creates damping and, although naturally occurring struc- tural damping will lead to a decaying motion, aerodynamic sources may lead to loss of damping. ‘The concept of complex aeroelastic modes may be illustrated by considering the clas- sical bending-torsion aeroelastic response of a high aspect ratio wing. Experiments (or analysis) on wings show that all spanwise sections of the wing have bending and torsionat components. The modes associated with bending and torsion have distinct frequencies and phase relationships. However, unsteady aerodynamic loads contribute to these fre- quencies and phase relationships; and, the knowledge of these phase relationships are valuable in flutter identification. At flutter, frequencies, mode shapes, and phasing co-230 Principles of Aeroslasticity Chapter 10 lesce to form a single aeroelastic flutter mode. Hence, the determination of the complex eigensolution (frequencies and mode shapes) is of value. It is important to note that the aerodynamic matrices in eqns. 10.4 depend upon the frequency of motion (as well as Mach number and dynamic pressure) and, consequently, the solutions of eqns. 10.4 are typically obtained with an iterative algorithm. Popular ‘methods for finding the solution to the flutter equations include the k-method and the p-k method? in which velocity-damping (.c., “V-g") diagrams and velocity-frequency dia- grams* are constructed from the roots of the eigenvalue solutions ("k" is the reduced fre- quency to be discussed in the next section). In the k-method, the damping term is multi- plied by (1 + jg) in which a fictitious damping, g, is introduced which permits the solution to the complex eigenvalue problem. Essentially, damping is assumed to be proportional to stiffness but in phase with velocity. Consequently, frequency and damping is found for increasing velocity. As indicated in Figure 10.3, a V-g diagram displays the roots of the solution. The exact solution of the flutter conditions occurs when g = 0 (note: that the "V- gf terminology is an unfortunate choice since the same symbology is also used to de- scribe aircraft load diagrams). in aeroelasticity, V is a velocity (in many cases, V is a nondimensional velocity referred to as the flutter speed index) and g represents the over- all damping of the aeroelastic system. 7h i i Mode: Fig. 10.3 vet bung Diagram The roots of eqn. 10.4 are examined for stability characteristics. The number of roots is equal to the number of degrees of freedom in the system. For typical aeroelastic analy- ses, such as that for a wing structure or a complete aircraft, these degrees of freedom are associated with the structural mode shapes used in the analysis. These roots are traced for increasing values of velocity. The roots will naturally begin as those associated with a Positively-damped system since the motion will decay with naturally occurring structural damping (consider the case of no aerodynamic effects). However, as velocity increases the path of one or more of these roots may advance into the negatively-damped regime. The crossing (sign change) represents the location of the neutral stability and indicates the flutter velocity. In certain cases, one might find that for higher velocities, a root will again become positively damped, indicating thet the aeroelastic system is again stable. It is noted that the "Hump" mode illustrated in Fig. 10.3 may be eliminated with additional structural damping and, consequently, this instability may be suppressed. Regardless of structural damping, the hard crossing is present. The instability associated with the low- est velocity will aways be of interest.Fig. 10.4 Frequency-Velocity Diagram In addition to examining the change in damping for increasing velocities, analysis of frequency behavior for changes in velocity as illustrated in Fig. 10.4 reveals the sensitivity of aeroelastic modal frequencies to flight conditions. The frequencies are the natural fre- quencies of the structure when the velocity is equal to zero. These frequencies change as the velocity increases since the aerodynamic loads couple the structural dynamics of the system. More importantly, these “aeroelastic’ frequencies converge toward a com- mon frequency 2s the velocity increases. The velocity at the point where any two (or more) frequencies coalesce is one indicator of the instability boundary for the aeroelastic system; however, damping must also be examined since the loss of system damping is the more significant measure. During fiight tests both damping and frequency are moni- tored to identify their respective sensitivity to flight conditions — zero damping and coa- lesced frequencies are of concern. In addition, changes in the amplitude of response for changes in fight conditions should be monitored. Mode 2 ies g Divergence Dynamic Pressure —q Fig. 10.5 Frequency-Velocity Diagram indicating Divergence ‘A root associated with system damping may also vanish as it approaches zero and, simultaneously, a frequency of the system approaches zero. This zero frequency root represents aeroelastic divergence since a characteristic of divergence is that the fre- quency of the diverging mode approaches zero at the divergence speed. This character- istic provides an approach to identify potential aeroelastic divergence behavior in real time during fight tests. For example, the primary frequency of a beam with an serodynamic load on the tip is sketched in Fig. 10.5. In actuality, this solution is representative of a slender beam-like structure such as a wind tunnel sting or aircraft sensor probe with an aerodynamically sensitive device attached to the tip. The aerodynamic loading is de- pendent upon the displacement and rotation of the tip, as well as the velocity of the flow; consequently, the deformation-dependent loads lead to seroelastic divergence.232 Principles of Aeroelasticity Chapter 10 ‘As an altemative approach to the frequency domain solution, time domain solutions (Le., numerical integration) of the governing equations (eqns. 10.3) provide a simulation of the physical response. However, in such a time domain solution the task of modeling the aerodynamic loads may be more complicated. Although physical properties associated with the structure are fixed, the flowfield characteristics such as density, velocity, dynamic Pressure, etc. (or, more appropriately, the nondimensional similarity parameters) are changed to examine various flight conditions. The motion for three different values of dynamic pressure is shown in Figs. 10.6-10.8. bending Strona jummation| hie © Time Time Fig. 10.6 Decaying Motion in the Time Domain Figure 10.6 shows a decaying motion for a value of dynamic pressure below flutter conditions and illustrates that the response is, in reality, the superposition of multiple re- sponses such as individual bending and torsion contributions. a bending = torsion = flutter Fig. 10.7 Divergent Motion in the Time Domain Figure 10.7 shows a growing motion for a dynamic pressure that exceeds the conditions for flutter and, as suggested by the figures, the modes of motion have coalesced to a common frequency. i Fig. 10.8 Undamped Motion in the Time Domain Figure 10.8 illustrates a motion with constant maximum amplitude; that is, it is neu- trally damped (although the motion may appear similar, this response should not be con- fused with limit cycle oscillations found only in nonlinear systems). This case suggests the dynamic pressure (and/or Mach number) for the aeroelastic instability has been bendingChapter 10, Principles of Aeroelasticity. 233 reached. Lower values of dynamic pressure show damped motion. Higher values of dy- namic pressure show growing motion. A difficulty associated with the time-domain ap- proach is the identification of the precise point where the behavior is neutral. 10.1.3 Aeroolastic Similarity Parameters Itis both convenient and efficient to eliminate unnecessary free variables in analyses or experiments by identifying similarity parameters. For example, dimensional analysis {introduced in Chapter 4) reveals that only two nondimensional parameters — Mach num- ber and Reynolds number - are necessary to characterize the flow and flowfield effects on lifting surfaces. These two parameters address fluid density, viscosity, velocity, speed of sound, temperature, physical dimensions, and retated variables. In @ comparable fashion, dimensional analysis for the aeroelestic equations reveals additional nondimensional terms that must be considered. The reduced frequency is defined as oo ka (10.5) where cis a reference length (typically the mean aerodynamic chord), is the frequency of vibration, and Vz is the freestream velocity. The reduced frequency is the ratio of the vibratory motion to the velocity of flow over the vehicle's surface, and represents a meas- ure of how fast unsteady disturbances are moved downstream relative to the transverse (vibratory) motion. The mass ratio is defined as (10.6) =| onc? where m is the mass of the vehicle, p is the density of the fluid, and b is the wing span. The mass ratio is a measure of the ratio of the vehicle's mass to the mass of the air that the vehicle displaces as it moves relative to the surrounding fluid. (A a Fig. 10.9 Mass Ratio Geometry ‘An additional term, the flutter speed index (alluded to earlier) is defined as We) coy (10.7)24 Principles of Aeroelasticity Chapter 10 This parameter contains both the mass ratio and reduced frequency. As the analyst com- pares experimental and analytical results in aeroelastic investigations, not only the ge- ometry but also the mass distribution and structural stiffness must be similar in order for the comparisons to be valid. Results are often expressed in terms of flutter speed index and mass ratio. If two systems have identical flutter speed indices and mass ratios, then the reduced frequency is identical, since eqn. 10.7 is an obvious combination of eqns. 10.5 and 10.6. Clearly, the equality of these parameters insures dynamic similarity be- tween the two systems. 10.1.4 Static Aeroelastic Phenomena Static aeroelastic phenomena involve the combination of aerodynamic loads and structural fiexibilty. Static aeroelastic phenomena include aeroelastic divergence, control surface reversal, and lift effectiveness. An example of static aeroelasticity is the effect of aerodynamic loads on the lift distribution of a wing. The aerodynamic loads cause the wing to deform (twist and bend) and, in tum, this new shape will result in new aerody- namic loads. Naturally, one should expect a balance of forces to occur about this new deformed shape. One also observes that, as a result of deformations, the load distribu- tion is attered and the center of pressure shifts. As a consequence, control and stability characteristics are affected. 10.1.4.1. Aeroelastic Divergence. Of course, the possibility exists that the forces will not balance since the structural deformations may continue to grow as a consequence of feedback from the deformation-dependent aerodynamic loads. These loads may give rise to even greater deformations. Eventually the surface will fail if this type of feedback interaction continues. [Aeroelastic divergence is a static aeroelastic instability leading to catastrophic| failure of the structure, and occurs when the aerodynamic forces exceed the tic forces of the structural sratior ‘Aeroelastic divergence occurs when the elastic restoring forces associated with the structural stiffness are exceeded by the aerodynamic forces. As stated, aerodynamic loads are not constant; rather, these loads are dependent upon the displacement of the structure. The equations that describe static aeroelasticity may be derived from eqn. 10.3 (K-Kalq=0 (10.7) in which velocity and acceleration forces are not considered. Kais the aerodynamic stiff- ness matrix, which is composed of aerodynamic derivatives and flow properties. The dy- namic pressure, qo, may be isolated such that the expression may be rewritten as qaKa (2 where qa = SY Therefore, eqn. 10.7 may be written as (K- qakka)q = 0 (10.8) Equation 10.8 is an eigenvalue problem in which q. ere the eigenvalues associated with aeroelastic divergence and q are the eigenvectors associated with the shape at di-Chapter 10 Principles of Aeroetasticity, 235 vergence. The term (K - ges) represents the aeroelastic stiffness that approaches zero ‘8 q.. increases. ‘Altitude and compressibility effects may be considered. Since Ve= Me, the dynamic pressure may be written as go = Mata) 1aM2 where qe is a reference dynamic pressure associated with the speed of sound and density at a specific altitude. It is im- portant to note that the aerodynamic properties in Ka depend upon Mach number. Fig. 10.10 Free-Body Diagram for Aeroelastic Divergence Example ‘Aeroelastic divergence is illustrated through the following example. In Fig. 10.10, a rigid wing section is mounted to an elastic support that permits one degree of freedom motion. The wing section has a center of gravity, which is not necessarily coincident with the spring support representative of the elastic axis. Aerodynamic loads act through the aerodynamic center. The equation of static equilibrium for the wing deflected from the balanced system appears as follow: Le- Ko=0 (10.9) where K is the spring stiffness, 0 is the rotation degree of freedom, L is the lift, and @ is the distance between the elastic axis and the aerodynamic center (@ is a measure of aerody- namic eccentricity). The weight of the wing is not at issue since @ is the displacement from the static equilibrium position. These parameters are all constant except for the lift. Lift is a function of dynamic pressure, which depends upon the velocity and fiuid den- sity. Furthermore, lift is @ function of the total angle of attack, a+ 6, where ais the angle of attack at equilibrium. (Be careful; again, the symbology is confusing, since @ has been previously used — in Chapter 6, especially — to denote the Euler pitch attitude angle. Re- member that in this context, 6 is the perturbation from the static equilibrium position). Thus, lft is expressed as L=qeSC,, (a+0) (10.10) where Sis the reference area, and C;,, is the aerodynamic derivative defined by the slope of the lift curve. We define the aerodynamic derivative as the load per unit deflection or per unit rotation. Substituting eqn. 10.10 into equation 10.9, (K~qnS0C1, )0 = 9nS0C,,,0 (10.11) which is @ scalar form of the multiple degree-of-freedom system (eqn. 10.8) and solving eqn. 10.14 directly for 0 leads to236 Principles of Aeroelasticity Chapter 10 qoS0C},, R= qe50C,, The aerodynamic contribution in eqns. 10.11 and 10.12 appears with the structural stifiness. In fact, one observes that the aerodynamic contribution is a negative stifiness, ‘As the dynamic pressure is increased, the term in the brackets approaches zero; conse- quently, @ approaches infinity. The terms in the denominator form the expression for the critical dynamic pressure at which aeroelastic divergence occurs. When the bracketed terms in the denominator approach zero, the result is therefore of special interest; thus, solving this term for qoleads to (10.12) K [QDI * Sec, | (10.13) ‘One observes that the aerodynamic eccentricity, e, may be negative (i.e, the aerody- namic center is behind the elastic axis) which suggests a negative divergence pressure exists. Obviously, this is not physically possible which means that divergence is not pos- sible. Thus, design considerations that minimize the effect of eccentricity will alleviate aeroelastic divergence. Graphically this feature can be illustrated by superimposing and comparing the force- displacement behavior of the structure with the lift-displacement behavior. The structural and aerodynamic stiffness is the slope — the differences in the stiffness represent the aeroelastic stifiness. The force-displacement curve is fixed for a given structure; how- ‘ever, the lift-displacement curve depends upon the dynamic pressure. At a critical dy- namic pressure, qe, ov, the two curves coincide, indicating neutral static stability. In physi- al terms, at this (and higher) dynamic pressure no structural restoring capability exists. This instability is catastrophic, as deformations grow rapidly, and without oscillation, to structural failure. DISPLACEMENT Fig. 10.11 Boundary for Static Aeroelastic Divergence Although this illustration is quite simple, the concept is essentially the same for large degree of freedom systems as described by eqn. 10.8. In fact, it is noted that geov de- scribed by eqn. 10.13 is the eigenvalue of eqn. 10.11. Equation 10.13 may be expanded to illustrate altitude and compressibility effects. The dynamic pressure may be related to Mach number by qo.Dv = Ja (Me, ov)? (10.14)Chapter 10s Principles of Aeroelasticity 287 Also, as previously stated, the aerodynamic derivatives are dependent upon flow con- ditions. The Prandtl-Glauert rule may be used to illustrate the effect of compressibility on tthe lift, thus, C= (10.15) where Cy, 9 represents the derivative for incompressible flow. Substituting eqns. 10.14 and 10.15 into eqn. 10.13 yields KJ1- M2 oy, Go(Ma pv)? = Y= 2 = go,piv tM pry (10.18) SoCo where ga ow represents a reference divergence pressure for the incompressible assump- tion and q, is a reference dynamic pressure associated with the speed of sound at a par- ticular altitude. Equation 10.16 may be solved for the divergence Mach number (M.,ov ) and compared to the reference conditions. We observe that divergence is improved with an increase in altitude and we note that compressibility effects must be included to pro- vide the real boundary. We aiso note that, although not considered in this model, the aerodynamic center moves reerward for increases in Mach number; thus, the aerody- namic eccentricity, @, is affected. x Sy DIVERGENCE ‘DIVERGENCE: ‘MACH NUMBER VELOCITY Fig. 10.12 Boundary for Static Aeroelastic Divergence (Mach Number Effects) Divergence is a static aeroelastic phenomenon, one which is often confused with yield oF utimate load failures. However, we show through eqn. 10.13 and Fig. 10.11 that aero- elastic divergence is independent of the yield or ultimate loads; rather, divergence de- pends upon the stiffness of the structure and the aerodynamic derivatives. ‘We could extend our development to @ wing, which is stiff in bending but comparably responsive in torsion. In this case, we simplify the structural model by discretizing the ‘wing into separate spanwise sections and we introduce the concept of influence coeffi- cients. Influence coefficients represent the inverse of the stiffness; these coefficients re- late the deformation at a specific point due to loads at all sections. For torsional behavior, these coefficients describe the torsional rigidity. We would soive the matrix of this multiple degree-of-freedom system for the critical dynamic pressure that leads to divergence. ALTITUDE238 Principles of Aeroelasticity Chapter 10 10.1.4.2. Control Reversal. Control surface reversal occurs due to the deformations of the structure. As the structure deforms, the ability of the control surface to produce a desired reaction approaches zero. The surface is said to lose effectiveness. Further de- formations result in the reversal of control surface performance; that is, deflecting the surface produces the opposite effect from that which was intended. Fig. 10.13 Control Surface Reversal Our single degree-of-freedom example is now extended to include control surface loads. An important premise for understanding control reversal is that the purpose of the control surface is to create a lift load that leads to a moment for maneuver. Thus, if the wing section represents the outboard portion of a wing, then the new lift leads to a rolling maneuver. On a vertical tail section the rudder is meant to produce a side force that causes a yawing moment. However, if no lift or side force is generated for any control in- ut due to flexibility of the fixed structure, then no maneuvering moments are created. ‘Summing moments about the elastic axis of the wing-control surface model: DMea=0 => Le+Mgo=Ko => 0 bo+ Mac (10.17) in which the aerodynamic tif is L=Lwing + Leontrot = FSC, 8+ GSC. (10.18) and the aerodynamic moment at the aerodynamic center is Mac = 9SCCm,3 (10.19) where & is the contro! input. Substituting eqns. 10.18 and 10.19 into eqn. 10.17 yields Le+Mac _(9801,,0+ 9SC1, 3) 0+qS0Cm,5 K K (10.20) ‘or, solving eqn. 10.20 explicitly for @ geSC1,5 + gSCCm,5 K-ae8C,, a The creation of lift is of primary interest, thus eqn. 10.21 is substituted into eqn. 10.18 GSC,,8+GSCCm, 3 KC1, +SCCm,C., tt = ——1@_—_—. (10.; +0158 |=49 th esGe = |® (10.22)239 Chapter 10, Principles of Aeroslasticity, Control reversal occurs when no lift is created, regardless of the control input. From the numerator of eqn. 10.22, we obtain the dynamic pressure that leads to zero (or negative) values (note: by the NASA convention, Crs is negative; see Chapter 6). = KS LsO of Sv = SoC, oa (10.23) ‘Also, divergence conditions appear through eqn. 10.22 as the denominator approaches zero for increasing dynamic pressure. Lo or on (10.24) 10.1.5 Dynamic Aeroelastic Phenomena Dynamic aeroelastic phenomena include flutter, buzz, buffet, and dynamic response. With these phenomena, airframe motion, aerodynamic loads, and structural deformations are dependent upon one another. Flutter is an instability in which the structural response is excited by the presence of the aerodynamic loads as the structure extracts energy from the flow field. The interested reader is referred to the works of Theodorsen and Garrick’, Bisplinghoff, et al. ', and Fung?, as well as the descriptions provided in references 6 and 7. ‘Types of flutter include: coupled bending-torsion flutter associated with high aspect ratio lifting surfaces, panel flutter associated with low aspect ratio lifting surfaces, control surface flutter, flutter of plate and shelHike structures, propeller-whirl flutter, and stall flut- ter. All of these flutter instabilities, with the exception of stall flutter, are characterized by the coalescence of modes. Stall flutter occurs due to flow separation that occurs during cyclic motion. In this case, the transfer of energy from the flow to the wing does not rely on the coalescence of modes; rather, the wing is “forced” due to energy lost due to stall. Buffet is an instability that occurs due to flow separation or the wake produced from the structure. Buzz is an instability that occurs due to shock wave oscillations. In general, dy- namic response of the aircraft due to gusts and atmospheric disturbances is important to characterize since, even in the absence of instabilities, it is necessary to understand how the dynamic response will affect load distributions and ride qualities. [Flutter is a dynamic instability in which the structural modes that are affected by} aerodynamic loads coalesce to form a single aeroelastic mode that rapidly grows} |with diverging oscillatory motion to structural failure. 10.1.5.1. Bending-Torsion Flutter. Of these dynamic aeroelastic phenomena, the bending-torsion flutter mechanism provides the insight into the nature of these insta- bilities. As previously stated, flutter occurs when the structure extracts energy from the flow field, This structure may be the primary lifting surface, @ control surface, the fuse- lage, or the entire vehicle. As the structure vibrates in a subcritical flow field, the aerody- namic forces damp the vibration of the structure; however, as the speed (or dynamic pressure) increases, a critical condition is reached in which the energy from the flow field240 Principles of Aeroelasticity__—___Chapter 10 feeds the vibration of the structure. This critical point represents the onset of flutter. Flutter is destructive in nature; but, fortunately, flutter boundaries may be identified. Flutter results from the interaction of inertial, structural, and aerodynamic forces and, typically, flutter is the result of coupled modes of motion. The pitch and plunge motion is representative of a wing with bending and torsional response. Although these equations of motion are straightforward, they apply to more representative aerospace structures. A minimum of two degrees of freedom is used since a wing limited only to bending motion will not flutter and a wing limited only to torsional motion will exhibit only stall flutter at high angles of attack. Although unlikely, it is possible for a single degree-of-freedom to flutter in torsion for very specific center-of-gravity and elastic-axis locations (see reference 1). Fig. 10.14 Bending and Torsional Flutter The wing in Fig. 10.14 is limited to motion in two degrees of freedom. It has a “plunging” motion, which is translational motion perpendicular to the flow, and a “pitching” motion, which is rotational motion about the elastic axis. The linearized equations of mo- tion for the system are: m m9) (Ky 9\(y)_(-L (mT) C3 e)-G) (1028) where m is the mass of the wing, ris the distance between the center of mass and the elastic axis, | is the moment of inertia referenced to the elastic axis, Ky and K, are spring constants associated with the plunge and pitch motion, respectively, L is the aerodynamic lift and M is the aerodynamic moment referenced to the elastic axis. ‘The form of this equation is identical to eqn. 10.1. Note that in this general form, the discussion of aeroelasticity is not limited to a system of two degrees of freedom. The mass and stiffness matrices may be representative for any aeroelastic system. Structural damping, C, may also be considered. In real problems, these matrices may be very large. Analytically, the system is often represented through finite element methods to accurately model the structure and the as- ‘sociated modal characteristics. Experimentally, these matrices may be derived using measurements and employing modal methods. The modes and frequencies of such analyses are used in the solution process. Although a wing may possess a very high number of vibration modes, only the lower modes are typically used since the flutter mechanism is driven by the fundamental modes. For example, the lowest ten to fifteen modes usually capture the primary behavior of the ‘structure (the reader is referred to reference 7 for specific case studies). This number may increase as interaction with higher modes such as those that are associated with contro! surfaces becomes important. Typically, the nature of classical flutter employs the primary torsion mode with the first or second bending mode. Of importance now is to un-(Chapter 10, Principles of Aeroelasticity. 2a derstand the content of these equations, the solution of these equations, and interpreta- tion of results. Examination of eqns. 10.25 in closer detail reveals that the equations are coupled through the mass matrix, es evidenced by the presence of off-diagonal terms in the ma- trix. We could have chosen a set of coordinates in which the system is coupled through the stifiness matrix, but the stability characteristics would be identical. The choice of co- ordinates is for convenience in the solution. This ‘inertial’ coupling occurs since the center of mass does not coincide with the elastic axis; the relative location of the center of mass plays an important role in aeroelastic design as mass-balancing inherently improves sta- bility margins. ‘The aerodynamic loads, L and M, are unsteady, which means these loads are motion dependent. For example, Theodorsen® (also see Fung’) provided a very rigorous devel- ‘opment of the unsteady potential flow equation that led to the classical solution for incom- pressible unsteady aerodynamic loads = apb?(7+Veei— Lely £(1_6)\ L=npb’ (F+ve8 Sei) +200 Va Sc{y+van+S(5 oe (10.26) 3 2 Me w5(-ar-($ -1)\aa+(j ?) 2a} 2M Fohroly +Vgo+ (10.27) ‘where C(k) is Theodorsen's Function and is a complex function of the reduced frequency, , and ais the nondimensional distance (in terms of c/2) from the elastic axis to the mid- chord and is a measure of aerodynamic eccentricity with respect to the elastic axis. ‘A quasi-steady aerodynamic model, appropriate for low frequency motion (k < 0.1), is found from eqns. 10.26 and 10.27 as = oy a L200 a5{i+Vase+ $5 +s) (10.28) ot e(1_.)e M=20pVa%(S+e)(7+van+S(5-a)s} However, a source of aerodynamic damping is present in either form. Harmonic motion for y and a. is assumed for the solution and, as a result, an eigen- value problem is formulated as previously discussed. In the simplest form, all terms are grouped in a single matrix. The roots to the equation resulting from the expansion of the determinant of the matrix provide the stability properties. The solution is complex such that the roots contain damping and frequency information. Again, it is noted that the ‘aerodynamic properties depend upon the velocity, which is also contained within the e- genvalue. The solution requires an iterative process (see Hassig’). 10.1.5.2 Example; Quasi-Static Aeroelasticity. Stability characteristics are obvi- ously affected by the matrices found in the governing eqns. 10.25. The designer of the airplane has some control over inertial, elastic, or aerodynamic coupling. Typically, the only parameters modified are associated with the flowfield (that is, velocity, dynamic pres- (+ et (10.2)242 Principles of Aeroelasticity Chapter 10 sure, and altitude) and solving these equations in the frequency domain permits one to examine damping and frequency characteristics. To further understand the flutter mechanism, an illustration based upon the two de- gree-of-freedom system is presented in which the aerodynamic loads are modeled by the static terms in eqns. 10.28 and 10.29. Thus, eqns. 10.25 become, m mr\(y\ (Ky 0\(y)_(~qSCL,@ Sol "\e)* 0 Ka)a} (qoSC_, eo (10.30) or, with rearrangement m my) (Ky oC, \/y)_(0 Gif a+ 0 Ka-GaSC.,e}la) (0 (10.31) Harmonic motion is assumed such that y= Ye and a= Ae 42m mr) (Ky WeSC, VY) _(0 a (m 1 (3 Ka -qwSC1,¢ Ila) “lo (10.32) These equations are simplified by dividing through the mass and introducing similarity parameters previously defined, 7 2C,,V2 1 r) |°Y Y)_(0 -o7| une = 10. i. ( a)" 0 2s? -2ObaVa® 4 (2) em 12 — Cato? ala unc where 1 is the mass ratio, r, is the radius of gyration, and a, and @, are the uncoupled natural frequencies of the system. 2 1 oy iat Y)_(0 ar uae 7 40, “) a} 0 o2y? 20tavae ( () Cae 12 lato? ala axe To put the equations into a more convenient nondimensional form, the first equation is divided by the reference length and the second equation is divided by the square of the reference length, and both equations are divided by w”. Thus, (2y 2C1,V3 7 o pnc2@? (") (3) |. -2|+ = (10.35) ( z) D a 22 20,,V2e A) \9, (10.35) oO) eo? BKC Oo where the overbar signifies a nondimensional length with respect to the reference length.Chapter 10, Principles of Aeroetasticity, 243 ‘The flutter speed index is identified within the aerodynamic terms, thus, in final form, 2 ©, a -14(22) ~740? . ? ()-(0) (10.38) -7 ~12 +(22) 2-6 o where the slope of the lift curve is assumed to be C,, = 2x (the value from theoretical aerodynamics). However, this “quasi-static” assumption is an extremely simplified and misleading "model" for the aerodynamic loads as it is important to note that aerodynamic terms such as C,, depend upon the eigenvalue, a, in addition to the Mach number. In more representative formulations of eqns. 10.36, an iterative process is necessary be- cause the flutter speed index also contains o. Several comments are made in reference to eqns. 10.36: (4) Equations 10.36 are in the eigenvalue form for the aeroelastic system. The de- terminant of the matrix forms @ quadratic equation for the roots, @”, which repre- sent the seroelastic frequencies as a function of the flutter speed index, (2) For the case of 7 = 0, the roots of eqns. 10.36 provide the natural frequencies as a function of the mass imbalance, 7 . The uncoupled natural frequencies are found for the case 7=0. (8) For the case of V+ 0, Fig. 10.15 illustrates the characteristic behavior of the ‘aeroelastic frequencies for increasing velocities. The flutter region is found as the two roots coalesce to a common frequency. Equations 10.36 may be solved as a standard quadratic 2 2 _-B+VB?-4AC B+/D 05,02=—— Fa a (10.37) Dynamic Fig. 10.15 Coupling of Flutter Modes Solving the (flutter) determinant of eqns. 10.36, reveals that the roots become complex conjugates with the single real root providing the frequency at flutter. Thus, from ‘eqn. 10.37, flutter occurs when D = 0. Also, aercelastic divergence - the “zero frequency” instability — is found by examining the case of C = 0 since terms A and B are motion dependent.244 Principles of Aeroelasticity Chapter 10 (4) Aerodynamic toads introduce coupling in the system through two mechanisms. Most obvious is the fact that the loads will not usually act through the elastic axis; thus, lift leads to a moment. The second source of coupling exists since the un- steady aerodynamic loads are dependent upon the displacements and velocities. (5) The formulation illustrates the use of the similarity parameters. (6) An extremely simplified formulation of the aerodynamics is used. In reality, the ‘aerodynamics depend upon the Mach number and reduced frequency. Standard aeroelastic methodologies, such as the "k method" or "p-k method" require an it- erative solution since the aerodynamic properties depend upon the reduced fre- quency. The aervelastic frequency, a, that is the eigenvalue also appears in the reduced frequency and flutter speed index. 10.1.6 Matched Point Analysis V-g Diagrams (and other solution methods) represent the aeroelastic damping of the system for changes in flowfield conditions. The complexity of determining the aerody- namic loads often necessitates solutions at specific altitudes and Mach numbers. Thus, the critical velocity for aeroelastic instability is found for different altitudes (mass ratios) ‘and different Mach numbers. Not immediately obvious here is an inherent paradox in the solution. That is, if one selects an altitude and Mach number to perform analysis, then the velocity is predetermined. Why ? In the atmosphere a speed of sound is associated with each altitude (and the temperature of the air at that altitude); and, together, the Mach number and speed of sound specify a velocity, Ven= Mes Qe (10.38) {s this velocity the same as the velocity from analyses? No, typically, the velocities are quite different except for fortuitous circumstances. Therefore, we must extend the analysis to include the “matched point” solution. ‘My from 4 [Pat toi 4 Matched | i Q Point uit nat ae - footy a Velocity ‘Mach Number Fig. 10.16 Solution for the Matched Point In this approach, as illustrated in the left-hand view of Fig. 10.16, the flutter velocity (obtained from the V-g diagram) is plotted for a series of altitudes at constant Mach num- ber. The intersection of a curve through these results (the curve given by the velocity found from eqn. 10.38) leads to a “matched point" in which the altitude, velocity, and Mach ‘fumber are consistent. This process is repeated over a range of Mach numbers to com- plete a set of data — the Mach numbers, altitudes, and velocities - for which flutter occurs and from which the flutter boundaries, as shown in the middle and right views of Fig. 10.16, are generated. It is noted that flutter is primarily a low altitude, transonic instability as suggested by these figures.Chapter 10, Principles of Aeroelasticity, 245 10.1.7 FAR Advisory Circulars: Parts 23 and 25. ‘As suggested in the Advisory Circulars**, velocity-damping (V-g) and velocity- frequency (V-«) diagrams are widely-accepted approaches used to examine the results of flutter analyses. ALTITUDE/DENSITY 200 MACH NUMBER = xx 6 Fig. 10.17 V-g Diagram for Five-Mode Analysis (adapted from AC 26.629-1*) In Fig, 10.17, a velocity-damping diagram is sketched for an aircraft represented by 5 aeroelastic modes. The roots are found for increasing values of velocity and reduced fre- quency. Of particular interest is the root that leads to the negatively damped system (root 4 in the sketch). Structural damping affects the onset of flutter. In the absence of meas- ured damping, a damping factor of 0.03 for structural damping is assumed (Method 1) as ‘a conservative value. Method 2 uses actual damping. Either way, structural damping manifests itsetf as a vertical shift of the velocity axis in the V-g diagram ~ @ consequence of the solution methodology. ‘The presence of structural damping eliminates the apparent instability indicated by the soft flutter crossing (‘hump mode") associated with root 1. Aeroelastic divergence is characterized by a root that abruptly terminates at zero damping — root 3 in Fig. 10.17. It is noted that civilian aircraft require a 20% margin of safety for flutter clearance, wherees military aircraft require @ 15% margin of safety for flutter clearance. Thus, in Fig. 10.17, Vpis the design velocity upon which a 20% margin as applied. The flutter crossing of root 4 represents 1.2Vp. Other than the exact crossing from stable to unstable, the path of the roots should not be used to infer stability behavior as the solution strategy is only correct at zero damping (sinusoidal behavior was assumed). Also, since the velocity as- sociate with the flutter crossing is associated with an altitude and Mach number, consid- ‘eration must be given to the aforementioned “matched-point” analysis. In Fig. 10.18 a velocity-frequency diagram illustrates an aircraft represented by 5 aeroelastic modes. The natural frequencies (such as those measured through a ground246 Principles of Aeroelasticity Chapter 10 vibration test) of the aircraft are represented at zero velocity, and the affect of aerody- namic loads on these frequencies is seen as @ function of velocity. Although flutter can not be determined solely from this figure, aerodynamic-sensitive modes tend to coalesce as the flutter velocity is approached. Aeroelastic divergence is noted for the mode that approaches zero frequency. Accelerometers mounted to the aircraft should be used to ‘monitor frequencies during flutter tests. VELOCITY Vp 1.2Vp Fig. 10.18 Velocity-Frequency Diagram for Five-Mode Analysis {adapted from AC 25.629-1*) 20% Ve @ CONSTANT MACH NUMBER Fig. 10.19 Flutter Envelopes from V-g / Matched Point Solutions. (adapted from AC 25.629-1*) Flutter conditions — velocities, densities, and Mach numbers - are used to construct flutter clearance envelopes according to FAR standards. Figure 10.19 illustrates two standard envelopes. A 20% margin is required at constant altitude (A-A’, B-B', C-C’), and @ 20% margin is formed for constant Mach number at sea level. This latter case results in clearance below sea level and, although fiight at this altitude is obviously not feasible, thisChapter 10 Principles of Aeroelasticity, 287 boundary is necessitated by the low-altitude, high dynamic pressure variables that most affect flutter. 10.1.8 Wind Tunnel Tests Wind tunnel tests provide the intermediate information necessary to qualify design concepts prior to full scale testing. Earlier, the requirement to satisfy similarity parameters identified through dimensional analysis was addressed. These similarity parameters in- cluded the Mach number, reduced frequency, mass ratio, and the flutter speed index. If the similarity parameters are identical, then one may compare experimental and analytical results with @ measured degree of confidence. ‘A major difference between wind tunnel test procedures and atmospheric fight test procedures is the controlled test environment. The temperature, Mach number, velocity, and density may be controlled in the wind tunnel. In atmospheric tests, nature dictates the dependency between the altitude (density) and air temperature (speed of sound). This dependency creates the need for the Matched Point solution to property interpret the analytical results. Since environmental contro! is possible in the wind tunnel test scenario, all but one of the test parameters may be held constant when identifying the flutter boundary. Typically, one of two methods is employed to determine a point on the flutter boundary. In the first method, the Mach number is maintained and the total pressure is increased in the wind tunnel (this is analogous to maintaining Mach number while decreasing altitude during flight tests). In the second method, the dynamic pressure is increased through the use of the fan speed to increase the flow velocity (this is analogous to maintaining constant alti- tude while increasing aircraft velocity during flight tests). Mach number and dynamic pressure both increase with this second method. Fig. 10.20 The Flutter Boundary ‘The typical approach to establishing the flutter boundary — normally critical dynamic pressure versus Mach number — in the wind tunnel is illustrated in Fig. 10.20. The flutter boundary is shown with the classical “flutter bucket” at which the minimum dynamic pres-248, Principles of Aeroelasticity Chapter 10 ‘sure (or “transonic dip") for flutter is typically found in the transonic regime. The subsonic. and supersonic regimes routinely have nearly constant flutter dynamic pressures. An initial test is planned to identify a flutter-free path throughout the Mach number range. This test is accomplished by increasing the speed of the fiow to well beyond the anticipated transonic dip of the flutter bucket. As stated previously, this method increases both the dynamic pressure and the Mach number, and this method resembles the con- stant altitude flight test. (Note: Conducting a constant altitude flight test at the maximum flight altitude typically identifies this flutter-free path.) Next, the speed is decreased and the tunnel pressure (stagnation) is increased to provide a higher dynamic pressure for this, same Mach number, and this method resembles a constant Mach dive maneuver. Then, the speed of the flow is increased unti the instability is detected establishing a point on flutter boundary. This process is repeated several times to complete the description of the flutter bucket on the subsonic side of the transonic dip. The fiat regions of the flutter boundary ~ the subsonic and supersonic regime — are found by increasing pressure at constant Mach number. The region of the flutter bound- ary beyond the transonic dip is found by significantly decreasing the tunnel pressure (re- turing to the conditions associated with the original flutter-free path) and, then, increasing velocity to obtain higher Mach numbers. Beyond the transonic dip, the pressure in- creases for a constant Mach number. We observe that, at points on the flutter boundary beyond the transonic dip, al decrease in velocity may cause onset of flutter. Therefore, in wind tunnel tests, instead of reducing fan speed, the total pressure is dropped to avoid triggering the instability by entering the transonic dip from supersonic conditions. In fight tests, this drop is analogous to an increase in altitude at constant Mach number. ‘Although these methods establish points on the flutter boundary, it is not desirable to actually test on the instability boundary (.e., "hit" a hard flutter point) due to the highly de- structive nature of flutter. Therefore, subcritical response techniques are used in both wind tunnel and fight tests. As suggested by the name, these methods examine the dy- namic response of the structure prior to the onset of the instability. This examination in- ‘cludes frequency, damping, and/or amplitude trends for increases in airspeed, and an ex- trapolation to actual flutter conditions. The structure or aircraft must be spared. These techniques will now be described in further detail. 10.1.9 Ground Vibration Tests Ground vibration tests (GVTs) are used to identify natural frequencies and mode shapes of the aircraft structure. Kehoe® and Norton? provide a thorough discussion of a GVT program in support of flight tests. Also, albeit a general theory of mechanical vibra- tions is not presented herein, the interested reader is referred to Craig'®. As previously presented, the equations of motion for the general mechanical system with mass, damping, and stiffness may be represented as ME +Cx+ Kx =F (10.39) where M, C, and K are parameters associated with the physical system, F is external forcing, and x is the structural response. If two of three components — physical parame- ters, external forcing, or response -- are known, then the third component may be found.‘Chapter 10. Principles of Aeroelasticity, 249 In analyses, the physical parameters and extemal forcing are known, and the response is simulated. In GVTs, the response due to a known extemal forcing is measured, from which the physical parameters are derived. ‘As depicted in Fig. 10.21, in GVT strategies the accelerometer (or strain gage, if ap- propriately located) is used to measure response due to an input. Typically, this input is from a mechanical shaker or impact hammer. The response determines vibration char- ‘acteristics — namely, the natural frequency, natural mode shape, and structural damping. Fig. 10.21 Wing Mode Shape ‘The fundamental wing modes found with a modal survey are illustrated in Fig. 10.22. 1st 2nd Deformed Position 3rd ~=Node Lines Fig. 10.22 Wing Modes Several supporting comments are in order: (1) A GVT may use a single accelerometer from which measurements are obtained at many locations on the structure, or the GVT may use an array of many acceler- ‘ometers to provide simultaneous measurements. In either case, a family of mode ‘shapes is found. (2) The “reciprocity” principle suggests that the source of excitation may be placed at ‘any location leading to equivalent modal characteristics. However, node lines (as illustrated in Fig. 10.22) are lines of zero displacement and should be avoided as modal excitation is difficult at these locations. The GVT should be conducted with250 Principles of Aeroelasticity ‘Chapter 10 more than one input source to confirm identification of all modes. For example, the second mode of Fig 10.22 would be missed if the shaker were positioned along the node line. (8) The GVT identifies modes that should be tracked during fight tests. The GVT will identify the optimal placement of accelerometers during fiight tests. Often, a sin- gle accelerometer may be used during fiight tests to track a mode identified during the GVT. Strain gages may be present from prior tests (airframe loads, etc) and may be used to measure vibratory response; however, the location may not be optimal and signal to noise ratios may be too low. (4) The underlying theory associated with the GVT assumes linear behavior. Thus, ‘sources of nonlinearities such as freeplay, sources of damping, and excess com- pliance in the structure must be identified and/or minimized. Linearity may be veri- fied through static load-deflection tests. Preloading of the structure will eliminate freeplay. Partial defiation of tires and landing gear struts during the GVT will help distinguish rigid-body modes from flexible-body modes. (©) Mode shapes exhibit the property of orthogonality (see reference 10). Mathemati- cally speaking, measurements shouid behave as OTMD =1 and OTKD=A (10.40) where ® is a transformation matrix whose columns are comprised of the normal- ized eigenvectors (mode shapes), / is the identity matrix, and A is a diagonal matrix consisting of the natural frequencies, a”. Orthogonality should be verified, with the existence and relative magnitude of the off-diagonal terms in the matrices resulting from the operation @®'M® and 7K providing a validation of the quality of measurements. Again, the GVT measures the output response through an accelerometer (or strain gage) from an input normally provided by a shaker or impact hammer. A modal survey, conducted as part of the GVT, determines the modal characteristics of the primary modes for the different configurations. These measured modes are used to validate analysis or may be used to develop a model for predictions. 10.2 FLIGHT TEST METHODS At the stage when fiight tests are required, aeroelastic analyses have identified the flutter boundary and wind tunnel tests have verified this boundary. In addition, math mod- els of the structure have predicted the vibrational characteristics and measurements have Verified the actual modal characteristics. Now, we extend our theory and ground-based test experience to fight tests. Several supporting references are suggested which include the efforts of Norton®, Kehoe", and van Nunen and Piazzol, as well as material ad- dressing flutter test techniques provided in references'* Expansion ofthe flutter envelope is an iterative process from inception of a new con- figuration to fight testing of the full scale aircraft. As the process continues end updatedChapter 10, Principles of Aeroelasticity. 251 flutter calculations improve with new knowledge, a foundation is created for flight tests that confirm an understanding of aeroelastic frequency and damping characteristics. Fig. 10.23. Flutter Expansion Flow Diagram (Adapted from Norton®) 10.2.1 Flutter Excitation In Flight Structural vibrations must be excited at flight conditions to examine aeroelastic sensi- tivity and validate predictions. As illustrated in the Fig. 10.24, several approaches are available. Techniques include: © natural atmospheric turbulence «mechanical excitation devices (mass eccentric devices) «aerodynamic excitation devices (oscillating vanes) ‘© pre-programmed loads from flight control systems © pyrotechnic devices ("bonkers") * pilot-induced control surface pulses or "stick raps‘) ‘Attached Oscillating Vanes Pyrotechnic Bonker Fig. 10.24 Flutter Excitation Schemes252 Principles of Aeroelasticity Chapter 10 Input loads may take several forms, and include: sudden impulses intended to simu- late a ‘step’ input (stick raps and bonkers), frequency sweep and dwell loads (shakers), and random loads (turbulence). The purpose of these loads is to excite the structure at a specific frequency, or range of frequencies, and observe the response characteristics — in particular, damping and frequency characteristics — for changes in flight conditions. The advantages and disadvantages of each form of possible excitation are discussed briefly in the following sections. 10.2.1.1 Natural Atmospheric Turbulence. (including wake and buffet induced vi- brations): The advantages are obvious simplicity and low hardware or aircraft modifica tion cost. Disadvantages include: (1) excitation parameters (frequencies swept, dwell times) are not controllable; (2) data reduction requires more sophistication due to low sig- nal-to-noise ratios; (3) quality of data collection depends on atmospheric conditions; (4) all modes are excited simultaneously and closely spaced modes may require additional data collection; and (5) excitation levels will vary with altitude. 10.2.1.2 Mechanical Devices. Advantages include: (1) efficient use of test time; (2) Precise control of frequency sweeps and dwell times; (3) short duration at test points; (4) similar excitation at all altitudes; (5) data reduction is more dependable; (6) signal-to-noise ratios are higher, and (7) specific modal frequencies can be targeted. The disadvantages of using such devices include: (1) systems may be bulky and adversely affect certain ‘modes; (2) added complexity increases chances of test aborts; (3) external actuators may affect the aerodynamics (4) data storage requirements are increased; and (5) short dura- tion at test points requires high quality signal processing and high data rates. In lieu of additional flight test hardware, the presence of existing on-board flight con- trol systems (FCS) in many aircraft configurations provides in-fight excitation. Pre- programmed inputs of sine-sweep, sine-dwell, and targeted frequencies are often used ‘successfully with automatic fight control systems. Concerns with the use of FCS actuation include notch filters or other such limitations that prevent excitation at specific frequen- cies, or undesirable feedback between the pre-programmed inputs and vehicle re- sponse/performance. 10.2.1.3 Pilot Induced Oscillations and Stick Raps. Allowing the pilot to excite the structural dynamics is a classical fight test approach. Advantages are: (1) no alteration ‘of mass or aerodynamics since no additional hardware is used; (2) pilot exercises some control of excitation; and (3) reasonable time spent at each test point. The disadvantages of manual excitation include: (1) input frequency is limited by human response time; (2) only the lower modal frequencies (up to about 5 Hz) may be adequately excited; (3) con- siderable test time may be required if pulses are required in both directions and about each of the axes. 10.2.2 Flutter Envelope Expansion The flight envelope, including the required margin of safety, must be free of flutter. The test program must consider all store configurations, fuel management scenarios, and altemative vehicle configurations. As previously stated, flutter depends on flight condi- tions such as velocity, altitude, and Mach number; therefore, the test program must con- sider these parameters. Furthermore, an ordered priority must be used in exploring the effects of each parameter.Chapter 10, 10.2.2.1 Time Histories Versus Frequency Dependence. Like wind tunnel tests, flight tests are initiated at the lowest dynamic pressures (Fig. 10.15). Hence, aircraft are normally tested at higher altitudes first since lower densities dictate both lower dynamic pressures and mass ratio effects. Higher speeds are evaluated by making shallow dives fo achieve test speeds that cannot be attained in straight and level flight. However, higher speeds alone may not necessarily illuminate the flutter speed boundaries. The frequency of the input also affects the response of the modes of interest. ‘At a specific test point, output data are obtained and analyzed for frequency and damping characteristics. Measurements from devices such as accelerometers are often in the form of a response time history. Typically, 60 seconds of data are obtained at a specific test point. Given an adequate sampling rate and sufficient measurements, the data are transformed (normally by Fourier transforms) to the frequency domain. Figure 10.25 provides a simplified illustration of the frequency response derived from transforma- tion of time domain measurements, and shows the amplitude of response as a function of Principles of Aeroslasticity, 253 Fig. 10.25 Typical Frequency Response Several prefatory comments are appropriate as one considers the frequency re- sponse behavior depicted in Fig. 10.25. (1) The peaks associated with the system frequency response indicate the aeroelastic frequencies of the system, and these peaks may be observed to shift with changes in test conditions, For the zero velocity case, these peaks represent the natural vibration frequencies of the aircraft. As flutter is approached, two or more of these peaks will merge toward a common peak. As aercelastic divergence is approached, one of these peaks will migrate toward zero frequency. (2) Forced excitation from any extemal sources (such as vortex shedding, actuator movement, oscillating shocks, and other sources with an embedded frequency content) at an gercelastic (natural) frequency will result in large amplitude re- sponse. (8) As depicted in Fig 10.25, the response of the structure is the superposition of all modes. The magnitude of response is the additive affect of each mode — for ex- ample, although the first mode (lowest frequency) dominates the response at that frequency, contributions from higher modes are present. (4) During tests using mechanical excitation methods, sinusoidal (‘sine") sweep through a range of frequencies will identify aeroelastic frequencies indicated by peak amplitudes; the "sine dwell” at specific frequencies will permit examination of specific modes.254 Principles of Aeroolasticity Chapter 10 (6) The random input — from turbulence or mechanical excitation — will excite all modes and frequencies simultaneously. The aeroelastic frequencies will be iden- tified by the peak response. (©) Structural (no-fiow) and aeroelastic damping can be identified through the sharp- ‘ness of the frequency response. The “half-power" method, for example, is one technique used to derive damping from the frequency response spectrum. (7) Modal characteristics must be understood (almost always based on GVTS) prior to. flight tests. ‘The presence of rigid body motion and external vibratory sources will be observed in the frequency response, and these effects must be identified. The accelerometer output time history is often used for test points far-removed from the flutter boundary. For example, frequency trends are tracked by counting cycles and damping trends are tracked by using the half-ampiitude method"? in which the number of cycles needed for the amplitude to halve is directly related to damping. Also, the "log-dec” method" (logarithmic-decrement) is often used to derive damping from the decay rate. 10.2.2.2 Procedures Requiring Dives. Dives may be required to achieve higher airspeeds and dynamic pressures for envelope clearance. An altitude band of approxi- mately #1000 feet near the targeted test altitude is usually acceptable to obtain data at an altitude (mass ratio) assumed constant. Obviously, descent rates associated with large dive angles give large altitude changes and may not allow completion of the sweep ~ re- peated dives may be necessary to complete a test point. Care must be taken in dive pro- cedures to guarantee that airspeed increments are controlled. The sweep rate available with mechanical excitation may be changed to complete the sweep within this altitude band. Often, the “burst and decay" method is used in which a targeted frequency is excited (“burst”) for a few cycles, and the decay from this excitation measured. Frequency dwells may be conducted in a dive if a suitable forced excitation system is installed and averages of data may be needed if atmospheric excitation is used. Test points flown in a dive with control surface excitation done by the pilot are usually Pulses applied about one axis at a time and several passes through the altitude band may be required for a statistically significant data sample. 10.2.3 Suberitical Response Techniques Wind tunnel tests and fiight tests are required to establish precise flutter boundaries for the vehicle or component. Yet, since flutter is destructive by nature, testing at, or even near, the boundary is highly undesirable. Therefore, subcritical response techniques are employed to predict the onset of instabilities by extrapolating damping and frequency trends. These techniques require the use of accelerometers and/or strain gages. These instruments are mounted on the critical components so measurements of the modal char- ‘acteristics can be recorded and monitored in “real time’, Modal parameters of primary interest include the frequency, damping, and amplitude characteristics. Our development of aeroelastic theory in Section 10.1 shows that shifts in frequency and damping parameters identify the presence of aeroelastic instabilities. We must take advantage of these characteristics to predict where the instability exists based upon measured response at subcritical conditions. Several subcritical response tech- niques have been developed; one such method is referred to as the "Peak-hold” method as used by Cole'®. The peak-hold method requires measurement of the maximum am-255 Chapter 10 Principles of Aeroslasticity, plitude of the response of a contributing mode. This measurement must be made at sta- bilized flight conditions — constant dynamic pressure, altitude, and Mach number — and measurements are repeated for several flight conditions. Using the results of Cole to il- lustrate, responses measured at several dynamic pressures are illustrated in Fig. 10.26. Frequency — Hz Dynamic Pressure Fig. 10.26 Results of Peak-Hold Method (Adapted from Cole'®) ‘An inspection of the responses in Fig. 10.26 reveals a broad peak for the vibration of particular modes. This broad peak suggests well-damped modes. As the dynamic pres- sure is increased, the response appears as @ narrow peak. The inverse of the maximum amplitude is plotted versus the dynamic pressure (right side of Fig. 10.26). As the dy- namic pressure is increased towards the ciitical dynamic pressure, the inverse of the maximum amplitude approaches zero. This inverse is extrapolated to this zero value which corresponds to the flutter dynamic pressure. Changes in frequency as noted by the shift in the peak, and changes in damping as noted by the width of the peak, also provide indicators that may be used to predict the onset of the instability. Example 10.1: X-29. The significant efforis of Kehoe and Rivera'® and others (such a6 Freud: inger"”) are cited to provide a detailed discussion of flutter clearance flight test programs. As an illustra- tion, certain results of the X-29 flutter clearance project are briefly discussed. “The ground vibration test (GVT) is used to identity frequencies and modes during fight tests. Table 410.1 describes the aircraft mode, model frequency, and potential aeroelastic inetabilty in which the mode may participate (according to analytical predictions). Several forms of instability are predicted for the X- 28, end include: (1) flutter, (2) divergence — aggravated by the forward sweep design of the X-28; (3) ‘seroservoolastc instablties — beyond the scope ofthis text, but an instability that involves an interaction ‘ofthe fight control system with flexible vehicle modes; and (4) body freedom flutter ~ an instabilty that is ‘caused by the interaction of flexible modes with rigid body modes). In the following table, “D" indicates a potential Divergence Instablity, °F" indicates a potential Fitter instabilty, and "ASE" indicates a potential ‘AecoServoElestc instability ‘The GVT also identifies the optimal ocetion of accelerometers to track modes. Although the GVT may have used hundreds of accelerometers to cheracterize a specific mode shape, the fight test will only require a few judiciously place accelerometers. Fig. 10.27 (next page) shows the relative placement of few accelerometers on the X-29,256 Principles of Aeroelasticity Chapter 10 ‘Table 10-1: Mode Type, Frequency, and Nature of instability hp REQUENCY [POTENTIAL MECHANISM Hes FOR INSTABILITY a6 D, ASE 113 ASE 716 ASE First Fussiage Lateral Bonding 125 ASE First Fin Bending 762 F ‘Canard Pitch 210 DF ‘Second Fuselage Vertical Bending 243 ‘ASE ‘Second Symmetric Wind Bending 263 ASE, F ‘Second Antisymmetic Wing Bending 268 ‘ASE inboard Fiap Rotation 313 F Wing Torsion-Outboard Flap Torsion 35.370 F Micfiap Torsion 7 F Canard Pitch-Bending 420 F First Fin Bending 500 F Inboard Fiap Torsion 515) F ‘Adapted from Kehoe and Rivera’® Fig. 10.27 Location of Accelerometers for the X-29 (Adapted from Kehoe and Rivera'®) tis well worth noting that placement of a single accelerometer on the wing tip permits tracking of the bending mode. A second accelerometer on the other wing permits tracking of symmetric vs. anti- symmetric bending modes. And, a second accelerometer mounted on each wing — with one placed near the leading edge and the other placed near the trailing edge — permits tracking of torsional modes. Figure 10.28 (on the following page) superimposes predicted flutter boundaries onto the flight en- velope — displayed in the standard format as described previously. Symmetric modes and antisymmetric. modes are shown. A 20% margin of safety on the limit velocity (V,) is used. Several modes are found to ‘epproach the required cleerance envelope; most notably, a midlaperon flutter mechanism that pene- trates deeply into the planned flight envelope. These predictions used an extremely conservative value of structural damping (0.02 - 0.03) that leads to conservative (low) predictions of flutter speed. However, {ests are required to clear the fight envelope.Oo a4 08 12 16 20 24 0 04 08 12 46 20 24 ‘Mach Number ‘Mach Number. a. Predicted X-29 Symmetric Flutter Boundaries b. Predicted X-29 Antisymmetric Flutter Boundaries Fig. 10.28 Predicted X-29 Flutter Boundaries (Adapted from Kehoe and Rivera'®) ‘Scale model (16% dynemicelly scaled model) wind tunnel tests were conducted to investigate the rmidfaperon flutier mechanism. As depicted in Fig. 10.29, several wind tunnel tests were conducted and, in particular, tests were conducted well within the regime of predicted flutier. No instabilities wore ex- perimentally measured — differences with the enelytical predictions of Fig. 10.28 are attributed to struc- {ural damping being greeter than assumed. Structural damping delayed flutter onset. 18 % Scale ° os 06 08 40 12 14 16 18 ‘Mach Number Fig. 10.29 X-29 Midfiaperon Fiutier Boundary Prediction (Adapted from Kehoe and Rivera's) Following extensive analysis, ground vibration tests, and wind tunnel investigations, a flight test program was initiated that uses @ build-up approach. That is, fight tests were conducted in increments from lowest risk to highest risk, and proceeded from low speeds to high speeds beginning at highest al- titudes and finishing at lowest altitudes. Thus, confidence was established prior to approaching potential flutter boundaries. Figure 10.30 shows the flutter test matrix flown. It is noted thet no instabilities were found within the flight envelope, and it is noted that the maximum Mach number for the vehicle was €s- tablished et M,.=1.4 for reasons other than seroelastic concems.258 Principles of Aeroolasticity ‘Chapter 10 centre EEERaaa ee Fig. 10.30 X-29 Flutter Flight Test Matrix (Adapted from Kehoe and Rivera's) 10.2.4 Rules of Thumb In light of the previous discussion, it is appropriate to identify methods of raising the flutter velocity or possibly eliminating flutter altogether. All of these suppression ap- proaches have one purpose: to break the coupling, whether it be aerodynamic, inertial (mass), or elastic. Furthermore, today's flight control systems may be used to suppress the onset of aeroelastic instabilities by affecting the phase relationships of the coupled ‘system or create offsetting loads. One commonly used approach is to mass balance the system. Mass balancing re- ‘quires the addition or redistribution of mass so that the center of gravity is moved forward. Contro! surfaces and rotor blades often use ballast. An increase in mass lowers the natu- ral frequency; mass positioned away from the center of mass lowers the natural frequency in torsion; and, an increase in structural stiffness raises the natural frequency. The presence of flutter and divergence is influenced heavily by the character of the primary torsional mode. The vibration characteristics of the structure may be adjusted to eliminate or delay the onset of flutter. An increase in the spread between the participating modal frequencies will delay the onset of flutter. Typically, the primary torsional mode is higher than the primary bending mode; therefore, increasing the torsional stiffness or re- ducing the moment of inertia (redistribution of mass) is usually advantageous. The addi- tion of structural damping also provides flutter alleviation. Other than the aeroolastic modes, “aerodynamic stiffening’ is typical, as most system frequencies will rise slightly in the presence of increasing aerodynamic loads. Wing sweep has a significant effect on certain aeroelastic instabilities. As suggested in Fig. 10.31, forward sweep adversely affects the divergence boundary and aft sweepChapter 10 Principles of Aeroslasticity, 259 adversely affects control reversal. Relative to these other instabilities, flutter is affected fittle by sweep. ‘Sweep Forward: ‘Sweep Back Fig. 10.31 Sweep Effects (Adapted from Bisplinghoff") Angle of attack effects are minimal except for vortex-shedding phenomena. However, wind-up tums are often used at the completion of certain test points to demonstrate in- sensitivity to angle of attack (see reference 17). Controt surface flutter modes are most sensitive to altitude (mass ratio) effects. Also, control surface “buzz” will be eliminated with reduced free-play of the control surface actuator. The GVT must identify all modes of interest and flutter analysis demands the use of all participating modes. For example, a large transport could be modeled with the first 3 bending modes and first 2 torsional modes’, since bending-torsion flutter is typical. How- ‘ever, higher-order modes are necessary for multi-control surface aircraft; for example, ‘analysis of the F-18 High Angle-of-Attack Research Vehicle (HARV)" and X-29 Forward ‘Sweep Wing (FSW) Research Aircraft'® required in excess of 25 modes. 10.3 SUMMARY “Some fear flutter because they do not understand it, others fear flutter because they do” —_.... Theodore von Karman In this chapter we have examined both static and dynamic aeroelastic instabilities such as divergence and flutter. The highly-destructive nature of these instabilities requires a judicious approach to identify boundaries and clear the flight envelope. Ground-based tests and studies that include ground vibration tests, wind tunnel tests, and flutter analy- 808 are necessary to provide a strong foundation of understanding prior to fight tests. Flight teste at flutter or divergence conditions are not performed directly; rather, subcritical methods that include a continual examination of frequency, damping, and amplitude of response are used to identify instabilities via extrapolation approaches. A build-up ap- proach in the flight test program that permits the lowest risk test points (low velocity, high altitude) to be conducted first is advised. The fight test team should be formed with an ‘emphasis on appropriate experience and expertise, and one should proceed cautiously and “expect the unexpected”.260 Principles of Aeroelasticity Chapter 10 REFERENCES u Bisplinghoff, R. L., Ashley, H., and Halfman, R. L., Aeroelasticity, Addison-Wesley, Massachusetts, 1955. ? Fung, Y. C., An Introduction te the Theory of Aeroslasticity, John Wiley and Sons, New York, 1955. 3 Hassig, H. J., "An Approximate True Damping Solution of the Flutter Equation by De- terminant Iteration", AIAA Journal of Aircraft, Vol. 3, No. 11, Nov. 1971, pp. 885-889. ANM-110, “Flutter Substantiation of Transport Category Airplanes,” Advisory Circular of U. S. Department of Transportation FAA, 1985. Theodorsen, T. and Garrick, |. E., “Mechanism of Flutter — A Theoretical and Experi- mental Investigation of the Flutter Problem,” NACA TR-685, 1940. ce ACE-100, “Means of Compliance with Section 23.629, Flutter,” Advisory Circular of U. S. Department of Transportation FAA, 1985. “Introduction to Flutter of Winged Aircraft,” von Karman Institute for Fluid Dynamics, Lecture Series 1992-01, von Karman Institute, 1992. Kehoe, M. W., “Aircraft Ground Vibration Testing At The NASA Ames-Dryden Flight Research Fac /," NASA TM-88272, NASA Ames Research Center, Dryden Flight Research Center, California, July 1987. Norton, W. ‘Structures Flight Test Handbook,” AFFTC-TIH-90-001, Air Force Flight Test Center, Edwards AFB, California, Nov. 1990. "© Craig, R. R., Jr, Structural Dynamics - An Introduction to Computer Methods, John Wiley and Sons, New York, 1981. Kehoe, M. W., “Aircraft Flight Flutter Testing at the NASA Ames-Dryden Flight Re- ‘search Facility," NASA Technical Memorandum 100417, NASA Ames Research Center, Dryden Flight Research Facility, Califomia, May 1988. van Nunen, J. W. G. and Piazzoli, G., “Aeroelastic Flight Test Techniques and instru- mentation,” AGARD Flight Test Instrumentation Series, AGARD-AG-160, Vol. 9, AGARD, London, England, Feb. 1979. “Flutter Testing Techniques,” NASA SP-415, NASA, 1976. “Flutter Flight Testing Symposium,” NASA SP-385, NASA, 1958 (reprint 1975). Cole, S. R., “Exploratory Flutter Test in a Cryogenic Wind Tunnel,” AIAA Paper No. 85-0736, Florida, American Institute of Aeronautics and Astronautics, 1985. Kehoe, M. W. and Rivera, J. A. , “Flutter and Aeroservoelastic Clearance of the X-29A Forward-Swept Wing Airplane,” NASA Technical Memorandum 100447, California: NASA Ames Research Center, Dryden Flight Research Facility, Sept. 1989. 17 Fruedinger, L. C., “Flutter Clearance of the F-18 High Angle-of-Attack Research Vehi- cle With Experimental Wingtip Instrumentation Pods,” NASA Technical Memorandum 4148, NASA Ames Research Center, Dryden Flight Research Facility, California, 1989, "1 2 13 4 15 16281 Chapter 11 POST-STALL GYRATIONS AND SPINS So far in this introductory survey of flight test, the motions under discussion have been well-behaved small perturbations from some equilibrium condition. The angle of attack has been less than the stall angle of attack. Classically, this angle of attack is thought of as that angle where the C,-a curve has a sharp break as shown in Fig. 11.1. If the trajectory proceeds on to higher angles of attack, the vehicle may go out of control and go through completely uncommanded maneuvers. This chapter briefly summarizes the theory for these kinds of post-stall motions and outlines some of the precautions that must bbe taken when tests are planned that deliberately place the vehicle in this hazardous flight regime outside of the normal operating envelope. o ' Fig. 11.1. Classical Aerodynamic Stall ‘The hazardous nature of post-stall fight tests underscores the importance of conduct- ing them. During the 1970s post-stall accidents generated a considerable amount of re- ‘search aimed at preventing loss of control. Through careful tailoring of the aerodynamic configuration and skilful use of automatic control systems, designers have been able to create highly spin-resistant airplanes. The payoff in this design emphasis became evident during the flight test programs of the current generation of operational fighters. The F-14, F-15, F-16, and F-18 all have remarkably good high angle of attack characteristics compared to their predecessors. The latter three airplanes all completed rigorous high an- gle of attack programs without the loss of a single airplane to an out-of-control incident: ‘an unprecedented safety record in fighter development! But it is not enough to simply design departure- and spin-resistant airplanes. Maneuverabilty in the post-stall arena is important for a significant number of tactical situations’. At the time of this writing there are three major research projects in fight test2.24 meant to help understand and exploit any tactical advantages that come from maneuvering st angles of attack well above the stall. The YF-22 and YF-23 Advanced Tactical Fighter prototypes have demonstrated significant capability in this part of the flight envelope. With this sort of thrust in design re- quirements (driven, of course, by perceived operational advantages) and the inherent hazards associated with high angle of attack testing, it is imperative that an introduction to {fight test engineering include a discussion of post-stall testing.282 Post-Stall Gyrations and Spins Chepter 11 11.1 THEORETICAL FOUNDATIONS 11.1.1 Post-Stall Definitions The terminology used to discuss the post-stall flight regime is unique and an expanded vocabulary is a necessity. Our first step, then, is to understand what a stall, a departure, the incipient spin, a post-stall gyration (PSG), and a developed spin are and how they differ from one another. These definitions are the subject of this section. 11.1.1.1. Stall, No two source documents define stall in exactly the same way. The FAA Flight Training Handbook® defines this term about as simply as possible: A stall occurs when the smooth airflow over the airplane's the lift degenerates disrupted and ‘These words describe the classical stall as depicted in Fig. 11.1. On the other hand, military specifications define the stall in subtly different terms: angle of attack for maximum usable lift of attack’. ‘The same document explains "maximum usable lift" by spelling out conditions other than reaching the peak of the C;-a curve that may limit available lift. Possibilities include: Uncontroliable pitch, roll, or yaw Excessive sink rates Intolerable buffet Inability to perform altitude corrections ‘Stall occurs when a maximum usable angle of attack or airspeed is reached. It is not always an uncontrollable event in itself; rather, it merely describes a critical condition reached by the lift-generating surfaces of the vehicle. Exceeding this critical condition may cause drastic changes in the trajectory followed by the vehicle and in its response to control inputs. The airplane may become uncontrollable. Nonetheless, a stall does not inevitably lead to loss of control. 11.1.1.2 Departure. If a stall does not always produce uncontrolled flight, what event does mark the transition from controlled to uncontrolied flight? The military has coined the term "departure" for that event. [Departure is that event in the post-stall flight regime which precipitates a PS( or deep stall Notice two things about this definition. First, departure occurs in the post-stall fight regime; that is, the stall always precedes or is at least coincident with the departure event. It can be inferred, then, that the angle of attack for maximum usable lift is aways less than or equal to the departure angle of attack. Secondly, only one of three motions may result from a departure: the airplane either enters a PSG, a spin, or a deep stall. Implicit in this definition of departure is the idea that an immediate recovery cannot be made. For example, a general aviation airplane whose stall is defined in the classical sense illustrated | |Chapter 11 Post-Stail Gyrations and Spine 263 by Fig. 11.1, may recover immediately if the longitudinal contro! pressures are relaxed when the stall occurs. However, the same airplane might go out of control, or depart, if ailerons were deflected significantly at the time stall occurred. With pro-spin ailerons applied, the same docile general aviation airplane might readily spin. By the definition of the term, it would then have departed. This latter fact suggests that one must specify sus- ceptibility or resistance to departure for given control positions, along with other config- urational parameters. The departure event is usually a large amplitude, uncommanded, and divergent mo- tion. Such descriptive terms as nose slice or pitch-up are commonly used to portray de- parture, By uncommanded motions, it is meant that the pilot or controller did not intend the control movements to produce the actual motions which resulted. The airplane may not follow the pilot's commands for a number of perfectly valid aerodynamic reasons: High local a. may render control surfaces ineffective. The pilot may be unable to position the control stick properly due to lateral or transverse accelerations at his cockpit position. In either of these cases, the motion is uncommanded. A divergent motion is one which either continuously or periodically increases without bound. Some airplanes, like the T-33, periodically "buck" when forced past the stall; that is, the nose periodically rises. and falls. However, there is no divergence unless full aft stick, ailerons, or some other pro-spin control is held. The T-38 will occasionally produce a non-divergent lateral oscillation near the stall angle of attack. Neither of these motions is usually counted as a departure, though their presence does wam of impending departure if the controls are further abused. In sum, a departure is a very difficult event to precisely describe, but there is little doubt in the pilot's mind when it has occurred! 11.1.1.3 Post-Stall Gyration. A post-stall gyration is an uncontrolled motion about ‘one or more axes following departure®. It is also a difficult term to define concisely be- cause it occurs in 80 many different forms. Frequently, the motions appear to be com- pletely random about all three axes and no more descriptive adjective or noun can be applied than PSG. A snap roll or a tumble are motions that fit the definition of a PSG, and these more common (if less precise) words are sometimes used. The primary difficulty lies in distinguishing between a PSG and the incipient phase of a spin, especially an oscil- latory spin. The chief distinguishing feature is that a PSG may involve angles of attack that are intermittently below the stall angle of attack, whereas the angle of attack must be greater than the stall angle of attack for a spin to exist. 11.1.1.4 Spin. Finally, a spin must be defined®. ‘Sometimes the spin is defined as a yawing rotation rather than a yawing and rolling rotation, but such a definition would only be valid for a perfectly flat spin with a = 90°. Typically, the yaw rate about the z body axis is the dominating rotation. The pilot of an airplane will likely consider a sustained yaw rate one of the most important visual cues that264, Post-Stall Gyrations and Spins Chapter 11 he is ina spin. In fact, if the spin is steep (that is, the angle of attack is relatively low, near Gg), pilots often have difficulty in recognizing the motion as a spin, Such a trajectory may be perceived as a roll rather than a spin. A steep inverted spin is another kind of spin that can be even more confusing and disorienting because the body axis roll and yaw rates are in opposite directions. Even in an inverted spin, the yaw rate dictates the direction of rotation and the control manipulations necessary to recover. 11.1.1.5 Deep Stall. The last out-of-control flight condition to be discussed is called the deep stall: The deep stall is distinguished from a post-stall gyration by the lack of significant an- guiar rates. Some maneuvers with graphic names like “falling leaf" are categorized as deep stalls, but so are others that have large amplitude angle of attack oscillations. In a deep stall, there is almost always a very high rate of descent. It is fairly common for the lateral and directional controls to be ineffective and sometimes even the longitudinal con- trol loses its authority. The latter condition, of course, is to be avoided in any design. The motion during a deep stall may not be steady. Oscillations in angle of attack may be quite large in amplitude. These oscillations sometimes are helpful in thet longitudinal control inputs of the appropriate frequency have been used to dynamically generate the moments necessary to recover from the deep stall 1 2 Spin Modes ‘Three types of adjectives are used to describe the general characteristics of a given spin and they identify the spin mode. The sign of the angle of attack, for example, classi- fies a spin as either upright (a. > 0) or inverted (a. < 0). Average values of angle of attack also classify a spin as either flat or steep, meaning that flat spins occur at high a and ‘steep spins occur with lower a. Finally, the peak amplitudes of the oscillations compared with the average value of rate of rotation spells out the oscillatory character of the spin mode. One descriptive modifier from each of these adjective groups may be used to char- acterize the spinning motion. Table 11.1 summarizes these modifiers. Table 11.1. Spin Mode Modifiers Group 1 Group 2 Upright Steep Smooth Mildly Oscillatory Inverted Flat Oscillatory Highly Oscillatory Violently OscillatoryChapter 14 Post-Stall Gyrations and Spins 285 Mode identification terminology can be confusing. The most confusing facet of ad- jective identifier usage is associated with groups 2 and 3 in Table 11.1. The example” in Table 11.2 should help clarify usage of these modifiers. Table 11.2 F-4E Spin Modes Mode Descrip- | Average |AOAOsci-| YawRate | Roll Rate | Pitch Rate tion AOA (*)_| lations (°) (*/sec) (cisec) | _(*Isec) Steep-Smooth | 42 5 40-50 50 15 Steep-Mildly | 45-60 #10 45-60 Oscillatory Steep- 50-60 20 | 50-60 (with large | Same as Oscillatory oscillations) | yaw rate Flat-Smooth | 77-80__| Negligible 80-90 25 7 BOUNDARIES BETWEEN ‘SPIN PHASES ARE OFTEN INDISTINCT Fig. 11.2 Spin Phases 11.1.3 Spin Evolution A fully developed spin is an equilibrium dynamic state. However, this state of equilib- rium is not reached by every trajectory that progresses into this nonlinear post-stall flight region. Moreover, equilibrium for the strongly coupled, large amplitude governing equa- tions is not necessarily a steady state solution in the classical linear sense; it is more often286. Post-Stall Gyrations and Spins Chapter 11 2 periodic solution (like a limit cycle). Such a complicated trajectory does not occur in- stantaneously; the transient motions are quite observable and are especially important to the fiight test planner (as well as to the operational pilot!). We will discuss three stages or phases (Fig. 11.2) of spin evolution; but, keep in mind that not all spins will reach equi- librium. Thus, a given spin trajectory may pass through all or only one of the phases. it does not always progress through the complete evolution to the fully developed equilibrium state, 14.1.3.1 Incipient Phase. The incipient phase of a spin is the initial, transitory part of ‘the motion during which it is impossible to identify the spin mode. The yawing and rolling rotation (usually perceived by the pilot as yaw rate) begins as the incipient phase begins. The cue to the pilot of an incipient is that of a sustained, though not necessarily steady, yyawing rotation. Even during this incipient phase, the angle of attack is greater than the stalled angle of attack. This characteristic, stalled angle of attack in the spin is one of the distinctions between the incipient phase of the spin and any post-stall gyration that occurs after departure. The incipient phase of the spin ends when a recognizable mode can be identified. Sometimes, this mode is not recognizable in flight but must be identified by examining data traces. To summarize, the incipient phase of the spin is a transitory motion easily confused with a PSG, but distinct from either a PSG or the developed phase ofa spin. 11.1.3.2. Developing Phase. The developing phase of a spin is that stage of the motion in which the spin mode can be ascertained. Oscillations may be present, but the ‘mean periodic motion is discemible. The aerodynamic forces and moments may not be completely balanced by the corresponding linear and angular accelerations, but equilib- rium conditions are being approached. Generally, it is evident when the developed phase is in progress, though the exact point at which it began may be unclear. Since the motion is approaching an equilibrium state, it is frequently advisable to initiate recovery before ‘rue equilibrium is achieved. For example, during the T-38 test program, waming lights were installed to signal a buildup of yaw rate. Recoveries were initiated when these lights ‘came on at approximately 85°/sec. In the fiat spin mode of this basic trainer, a peak yaw rate of 165°/sec was achieved, even when recovery controls were applied when the lights came on. The longitudinal acceleration at the pilot's station was approximately 3.5g and the spin could only be terminated by deployment of the spin chute®. The developing spin, while it may be more comfortable because it is usually less oscillatory than the incipient spin, can be deceptively dangerous. Careful preflight planning and continuous monitoring of the testing is absolutely essential when you plan to stay in a spin while it is developing. 11.1.3.3 Fully Developed Phase. A spin is fully developed when it reaches the final equilibrium state. in this stage of its evolution the motion is typically a repetitive periodic trajectory. Theoretically, all the states would be constants and all linear and angular ac- celerations would be zero on average over a complete period. instantaneously, at least ‘some of the state rates of change would be nonzero if the motion is oscillatory in any fashion. Practically, attaining this state may take a very long time. It may also be hard to recognize in the cockpit or from real-time telemetry data if the periodic motion is oscilla- tory. One very important type of fully developed spin is, however, quite recognizable. The flat spin, in which the angle of attack is theoretically 90°, is just such a case. The difficulty in recognizing a post-stall equilibrium condition is one of the main dangers inherent inChapter 11 Post Stall Gyrations and Spins 267 post-stalispin fiight tests. It is one of the main reasons this kind of testing must be considered hazardous and treated with special attention to detail throughout the test planning cycle and during the conduct of such tests. All systems, whether they are unique to the test effort or are simply production systems, must be capable of and allow the test pilot to safely recover from the worst case of dynamic equilibrium in a fully developed spin. ‘Thus, the fully developed phase of the spin deserves considerable attention from the flight test engineer. Consequently, we will shortly develop six degree-of-freedom, fully coupled, nonlinear equations of motion for a rigid aircraft in this kind of trajectory. But before we proceed to these equations, we have to understand what kind of flight path to anticipate and the underlying aerodynamic forcing functions so we can make intelligent approximations. Fig. 11.3 Helical Flight Path in a Fully Developed Spin 11.1.4 Elight Path in a Spin A spin is a coupled motion that requires all six equations of motion for satisfactory analysis. Extreme attitudes ‘and large angular rates are often encountered in such mo- tions. In the general case the airplane center of gravity describes a helical path while it rotates about an axis of rotation at the center of the helix. As sketched in Fig. 11.3, the helical path may be curved and the total angular velocity vector is continually changing in both magnitude and direction. In general the wings are not level and sideslip is not zero.28_______—Post-Stalt Gyrations and Spins Chapter 11 This general motion is very complex. However, if we make the following simplifying as- ‘sumptions, some of the basic characteristics of spin trajectories can be clarified. 14.1.4. Assumptions. The following assumptions simplify the analysis while allow- ing us to keep the more important features of spin trajectories in focus. Assume: 2 fully developed spin in which the helical axis is vertical (w is constant) no sideslip, which infers that the wings are horizontal the side force is negligible compared to the lift and the drag These assumptions mean that the angular velocity vector a is approximately parallel to the helical fight path (and, consequently, to the relative wind). AXIS_OF ROTATION RELATIVE WIND Fig. 11.4 Changes in Lift and Drag Coefficients at High Angles of Attack 11.1.4.2 Balance of Forces. Next, consider the balance of forces in a such a steady ‘spin without sideslip as shown in Fig. 11.4. Because of assumption 3, the resultant aerodynamic force acts within the plane of symmetry and is approximately normal to the wing chord. Summing the vertical forces gives: (2 w-p~2¥ Sop (11.4) Similarly, the forces in the horizontal direction just balance the centripetal acceleration and (2 mmo? =L = 2¥! eo (11.2) The physical implications of these two seemingly trivial equations are extremely impor- tant, First, eqn. 11.1 says that, as a. increases, Cp increases and, consequently, true air- speed decreases. Since the trajectory is vertical, true airspeed is approximately equal to rate of descent. In other words, as a increases, the airplane becomes more and more like‘Chapter 11 Post-Stall Gyrations and Spins 269 rotating bluff body in the descent and the descent rate slows in a fiat spin. Furthermore, for a. > Gs, C, decreases as a increases. With these facts in mind, the left side of ‘equation 11.2 must decrease as a.increases. It will be shown shortly that the rotation rate @ tends to increase with increasing a. In tum, then, the radius of the helix r must de- crease rapidly as a increases. These observations suggest that in a fully developed, fiat spin, @ and V are nearly coincident. The angular difference between the flight path and vertical is approximated by Babister” as: tan = 2 Atypical variation of 2 with angle of attack is about 5.5° at an a of 50° and about 1° at an a of 80°. Hence, it is justified to assume in fully developed spins that « is nearly parallel to Vand, for flat spins, the two vectors are essentially coincident. ‘These assumptions are valid only if the wings are horizontal and sidesiip is zero. This complexity is beyond the scope of this introductory text; both Babister® and Kerr® provide insight into these topics for the interested reader. Remember, these comments are valid only for the fully developed spin. The underlying physical phenomena are clarified by considering this simplified case, and hopefully, these simplifications will give the student a better appreciation of the more complex components of this type of motion. 11.1.5 Aerodynamic Factors In a Spin In the post-stall fight regime aerodynamic forces acting upon the vehicle are very different than those acting upon it during flight at low angle of attack. Aerodynamic de- rivatives, usually described as linearized approximations with first order partial derivatives, are highly dependent on both a and B and second order derivatives may be important. Of ten these derivatives change sign rapidly as flows separate and reattach and vortex lift patterns break down. ‘Terms which are not particularly significant at low angle of attack may become very important. Stough'© and his associates at Langley have concluded that, at least for straight-wing, wing-oaded airplanes (general aviation aircraft), the ‘aerodynamic behavior of the wing is often the single most important factor in how the airplane itself enters and recovers from a post-stall maneuver. Though these conclusions ‘cannot be blindly applied to all airplanes, they do suggest that looking that the wing's be- havior is a good place to start our consideration of post-stall aerodynamics. 14.1.5.1 Autorotative Couple of the Wing. The wing is often the major contributor to the aerodynamic mechanism that initiates and sustains autorotation. If a wing is oper- ‘ting at a low a and a sudden local change in occurs due to a rapid roll, there is a re- storing moment developed because of the increase in lift generated by the higher aon the downgoing half of the wing and the decrease in lift on the half of the wing that is moving up (s90 the left side of Fig. 11.5, page 270, where AC, and ACp, occur). On the other hand, when a wing operating at o> ag (see the right side of Fig. 11.5, where AC), and Cp, occur) is rolling rapidly, the downgoing wing experiences a loss in lift and an in- crease in drag. This loss of ft creates an even larger rolling moment and the increase in ‘drag produces a yawing moment in the direction of the roll. This latter situation is clearly unstable and self-perpetuating; it leads to an autorotation.Chapter 11 Fig. 11.5 Aerodynamic Mechanisms for Autorotative Moments Xa Xp —— aes: - A a ADVANCING RETREATING WING PANEL WING PANEL ADVANCING WING SECTION Fig. 11.6 Lift and Drag on Advancing and Retreating Wing Panels RETREATING WING SECTIONChapter 11 Post-Stall Gyrations and Spins am ‘A more complete understanding of the aerodynamic mechanism that is driven by the wing comes from Fig. 11.6 (page 270), where the orientation of the resultant forces is shown, as well as the components of the resultant forces resolved along the inertial x and zaxes. The net difference in the X forces at corresponding sections in the advancing wing panel and in the retreating wing panel produces a couple (Xa - Xp)2y1. This couple is ‘shown at the top of this plan view of the wing. The yewing moment produced by the term (%a- Xp)2V4 is positive for a positive yaw rate. The rolling moment produced by the cou- ple (2, - Zp)2V4 is also positive. Thus, the incremental moments generated as the angular velocity becomes significant lead to a moment contribution from the wing that propels the airplane into a spin. Of course, eventually the damping in yaw and damping in roll will become large enough to substantially oppose the autorotative motion, but itis quite clear that a fully developed spin can be sustained indefinitely. It is an equilibrium dynamic state. FAVORS AUTOROTATIVE OPPOSES AUTOROTATIVE YAWING MOMENT YAWING MOMENT Fig. 11.7 Aerodynamic Contributions from the Forebody Shape 11.1.5.2 Effect of Forebody Shape. The fuselage, especially the nose, may also contribute significant aerodynamic moments with a > ag. The pressure field around the nose depends strongly on its cross-sectional shape. Since the flow character is highly dy- namic, often dominated by vortex shedding, prediction of the aerodynamic contribution of the forward fuselage components is rather difficult. Small geometric changes in this region can make dominating changes in rolling and yawing moment coefficient in the post-stall flight regime. One possible fuselage effect is illustrated in Fig. 11.7. In these sketches we are looking at a roughly elliptical cross-section for a fuselage nose as an observer facing forward in the airplane. Naturally, this shape acts like an airfoil, producing both aerodynamic forces perpendicular (like lif) and parallel (like drag) to the local flow direction. This local flow direction is strongly influenced by the rotational rate of the air- plane. As shown on the left side of Fig. 11.7, if the aircraft is rolling and yawing to the left and the net aerodynamic force vector is to the left, the local effect is to reinforce the au-272, Post-Stali Gyrations and Spins Captor 11 torotative rolling and yawing moments. The right side of Fig. 11.7 shows the same cross- section modified with a strake. With this modification the nose produces an aerodynamic moment that opposes autorotation. The spin characteristics of the T-37 trainer were considerably improved by such a modification, incorporated during its flight tests. 11.1.5.3 Effect of Damping Derivatives. Aerodynamic damping derivatives are significantly affected at high angles of attack, especially roll damping (Ci,). In fact, often both G,, and Cp, become positive at or near cis. These aerodynamic effects can be deduced by considering moments produced by the local lift and drag forces depicted in Figs. 11.4 and 11.6. Several authors consider this loss of damping to be one of the main aerodynamic factors resulting in autorotative behavior®.11.11, Obviously, a complete discussion of these aerodynamic coefficients and how they behave in the post-stall fight regime is beyond the scope of this introductory text. But, these references suggest additional reading for the interested student. Now, we tum our attention to the other side of the equation: the inertia terms. 11.1.6 Effect of Mass Distribution ‘The moments of inertia are the most obvious and meaningful measures of mass dis- tribution. The inertial forces and moments, which depend directly on the mass distribution, are particularly important in the post-stall motion of a vehicle because they are of the same. ‘order of magnitude as the aerodynamic forces and moments. This relative sizes of the inertial forces and moments and the aerodynamic forces and moments are features pecu- liar to flight at post-stall angles of attack. Fig. 11.8 Relationship Between Body and Principal Axes 11.1.6.1 Principal Axes. Orthogonal body axes for which all three products of inertia are zero are called the principal axes for the body. Principal axes can be found for any rigid body. If the vehicle has a plane of symmetry, as most aircraft and missiles do, it is fairly easy to choose a coordinate system that approximates the principal axes. The xz plane is simply chosen to lie in the plane of symmetry, thus guaranteeing that both /, = 0. The direction of the x axis is usually chosen so it can be easily identified with configuration geometry. (Thus, the term fuselage reference line (FRL) is often seen in airplane layout drawings.) For brevity this reference set of body axes is usually called simply the body axes. (Strictly speaking, the principal axes and the stability axes used with linearized, small perturbation equations of motion are also body axes.) To simplify our later discussion and to make the large amplitude equations of motion easier toChapter 11 Post Stall Gyrations and Spins 23a understand, we will take the body axes and the principal axes to coincide. An illustration of the difference between these two body axes is shown in Fig. 11.8. In the principal axis system, one of the moments of inertia is the maximum possible for the body. Choosing body axes that are close to principal axes means the product of inertia J, is quite small, A small value of /,, reduces the cross coupling between the large amplitude moment equations that govem the symmetric (longitudinal) motion and the asymmetric (lateral-directional) motion. For the linearized small perturbation equations, this term had to be negligible in order to separate the longitudinal equations from the lateral-directional ones. That simplification gave us the lwxury of solving two quartic characteristic equation sets rather than a higher order characteristic equation. Even though Ixz is typically small due to our choice of body axes, its coupling effects can rarely be ignored in post-stall maneuvers. We will specifically show in a later section that pitch rate can have a significant effect on w, even if /,z can be safely ignored. 11.1.6.2 Relative Density. The relative density parameter p is a measure of the ratio of the airplane's density to the density of the air its volume displaces. It is a very useful nondimensional parameter that appears in the equations we need. 2m 2m. (11.3) 11.1.6.3 Center and Radius of Gyration. Two other definitions are used in dis- cussing mass distribution: center of gyration and radius of gyration. The center of gyra~ tion of @ body with respect to an axis is that point at which the mass of the body could be concentrated and the resulting mass moment of inertia about the specified axis would be unchanged. Similarty, the radius of gyration k; with respect to an axis is the distance from the center of gyration to that reference line. Recalling the basic definitions of the mass ‘moments of inertia, we can write expressions for each of the three radii of gyration: Jo? + 2? )arm = 1, =K7m fo? 2? jdm = ly = km (11.4) foe? + y? Jam = 1, = k2m 11.1.6.4 Relative Magnitude of Airplane Moments of inertia. The relative sizes of moments of inertia for an airplane are a convenient means of classification (as suggested in Fig. 11.9). The moment of inertia about the z body axis is always the largest of these inertias; this fact is assured because airplanes have very little mass outside the xy plane. From a mass-distribution point of view, the vehicle is “flattened” into this plane. A wing- loaded airplane is defined 2s one that has more mass concentrated in the wings. The wings may be structurally heavy; fuel may be carried inside integral tanks in the wing; ‘engines may be slung in nacelles from the wing; or stores may be loaded externally on pylons attached to the wings. Any one or all these factors cause more mass to be located in the wings. /, > I> ly for wing-4oaded airplanes. Figure 11.98 ilustrates such an airplane. A neutrally loaded airplane is one in which the wing and the fuselage have274, Post-Stall Gyrations and Spins Chapter 11 roughly equal mass, that is, J, = J, (Fig. 11.9b). Finally, a fuselage-loaded airplane is one in which most of the mass is concentrated in the fuselage (Fig. 11.9c). Most modem fighter airplanes are fuselage-loaded, which helps to give them very distinctive spin characteristics. We can gain some insight into just what kinds of spin modes might develop for different mass distributions (or “loadings") by careful inspection of the equations of motion. Zz REMEMBER: |, IS NEUTRALLY LOADED ALWAYS LARGEST OF ha THE THREE INERTIAS <4 iz FUSELAGE LOADED ly > Ix () Fig. 11.9 Relative Magnitude of Aircraft Moments of Inertia We have been referring to the large amplitude equations of motion freely and dis- cussing terms within these equations; it is now time to examine these equations directly. 11.1.7 Simplification of the Post-Stall Equations of Motion The equations of motions describing aircraft dynamics in the post-stall fight regime are more complicated than the linearized, small perturbation ones we have discussed and used in previous chapters. The post-stall equations differ in the following ways: + all six rigid body degrees of freedom must be considered in any solution — pitching motions do affect lateral-directional response and vice versa; the changes in states are often quite large — small perturbations in angular positions, rates, and accelerations are simply not warranted; and ‘@ the aerodynamic coefficients are not constants — the stability derivatives are functions (and seldom analytic ones at that) of both angle of attack and sideslip angle at least. 11.1.7.1 Simplifying Assumptions. The basic simplification we will use is that only a fully developed spin with the wings horizontal will be considered. This motion state is an ‘equilibrium condition, in which the aerodynamic forces and moments and the inertial forces,Chapter11____—_—Post-Stall Gyrations and Spins 25 ‘and moments are balanced. To further simplify the discussion, we will further require that all applied moments are aerodynamic ones; in fact, we will not allow even propulsive moments. This latter essumption will avoid a lengthy and confusing discussion of ‘moments generated by gyroscopic couples due to rotating parts in the engines. Also, any coupling between the propulsion system and the controls can be ignored with this simplification, Summarizing these simplifications: @ Because the wings are horizontal, the angular velocity vector o lies entirely within the xz plane. Its only components are P and R; Qis zero. © Furthermore, the velocity vector and @ are parallel at all times. The velocity vector (and «) is vertical for a fully developed spin, that is V= rate of descent. The angular velocity vector is constant, making P = Q = R = Qand the time per tum constant. @ The body axes are assumed to coincide with the principal axes, allowing us to take be = 0. The last assumption above simply allows us to ignore the terms containing the product of inertia [xz in each of the moment equations. The force balance takes on one of the two following forms. In an inertial frame of reference or a body frame of reference, the vector equations are: F=mZ& forthe inertial frame or ) forthe rotating body ai rane of reference. Since we will only look at steady state conditions, SE = in either ret- erence system. Also, @ and V are parallel (assumption 2), so F = 0 in body axis coordinates. 11.1.7.2. Simplified Large Amplitude Moment Equations. The summation of mo- ments is by far the more important balance for equilibrium and it is usually only considered in the rotating frame of reference. In component form the moment equations are: aP aR Fix + ORs -ly)- (¢ e aR) =4 21, + Palsy -1)-(R? -P?)I M (11.5) (11.8)276, Post-Stall Gyrations and Spins Chapter 11 PRN, PAM 'y) dt iy iz The first term on the right side of eqns. 11.6 represents the aerodynamic contribution to the angular acceleration and the second term describes the inertial contribution to the Pertinent acceleration. Thus, simply rewriting these three moment equations in terms of the angular accelerations shows clearly that, for equilibrium to occur (that is, for a spin to become fully developed and all angular accelerations to vanish), an inertial term must balance an aerodynamic term in each equation. Notice tht even with our simplifying as- ‘sumptions we have not eliminated coupling between the equations of motion. These simplified equations also suggest that, to accurately calculate absolute angular accelera- tions, it is important to know the moments of inertia accurately. Obtaining them experi- ‘mentally is a tedious job12, one that fight test personnel are often tasked to complete. Before we expand the aerodynamic terms, it is useful to point out a difference in the usual nondimensionalization of aerodynamic coefficients for use in the coupled equations that govern post-stall trajectories. These equations differ from the small perturbation equations that spawned our original nondimensionalization of stability derivatives. They cannot be separated into uncoupled longitudinal and lateral-directional subsets. Typically, data sets must be modified from our usual concept of stability derivatives for use in modeling post-stall dynamics. We have already seen how pitching moment coefficients (Cm) are based on wing semichord as a reference length, while both rolling and yawing ‘moment coefficients (Cz and C,) are based on wing semispan. For the data to represent the same aerodynamic moments in the large amplitude equations of motion, a common reference length must be chosen. For this kind of analysis it is common practice to base pitching moment on wing span and designate it as Cmp. Numerically, the usual small- perturbation pitching moment coefficient is altered by Cnn = Cm The aerodynamic terms in eqns. 11.6 can be rearranged using the radius of gyration of eqns. 11.4. 4 PV?8C 2 V?7Ce ny WP _ 4, ARly “Ie Ik 2k2m = pk? dt ly kk M _ pV?spc, V2Cmp dQ _M , RP(Iz - ty) i ae be and ——=—— + ——~#__x- 11.7) ly 2K2m Buk? dt “ly 5 oy NL pV?SbC, _ VC, OR ON , PQUx - ty) I, 2k2m yk? dt 1, Tz Expressing the definitions of the aerodynamic moments in nondimensional terms, ap dt 22m dQ _pV?SbCmy , RPCIy - Ix) ad okm wf (11.8) 2kzm ly |Chapter 11 Post-Stall Gyrations and Spins a GR _pV?SbC, , OR(Ix - ly) dt 2kzm iy 11.1.8 Aerodynamic Conditions for Dynamic Equilibrium Combining eqns. 11.8 with the first assumption, gives equilibrium conditions. O=C, _pV?SbCmp = RPUIz 2kym yy (11.8) 0=C, Remember that these equations are greatly simplified; they do not completely rep- resent post-stall motion. But they allow us to introduce the idea that a fully developed spin is an equilibrium or “trim” condition, just like straight and level flight. With no extemal disturbances, the airplane will remain in this dynamic state until it impacts the ground. Now, let's look at each of the component moment equations above and understand their implications for a fully developed spin. We will start with the pitching moment equation, since it appears to have the most interesting form in eqns. 11.9. 11.1.8.1 Pitching Moment Equation. Consider eqn. 11.9 in conjunction with typical pitching moment curves for fighter type aircraft (Fig. 11.10). Also, from our original definition of a spin, a. must be greater than as. For the typical highly maneuverable airplane, our simplified pitching moment equation implies that the term with Cinp in it must be of opposite sign to the inertial term on the right side of eqn. 11-9b. For an upright spin, this condition means that Cm must be negative, since in an upright spin P and R are of the same sign (see Fig. 11.11 on the next page). Combining this observation with the fact that /, > |, for any airplane, clinches the argument that Cmp <0. Conditions of equations 10.9 nly met in shaded regions ‘Aircraft A. (eae i —— z Fig. 11.10 Pitching Moment Coefficient in a Steady Spin The slope of the Crp - a curve must be negative for stability in any flight regime, even post-stall conditions. This slope must be negative for static longitudinal stability at either low oF high angles of attack. Said ancther way, Gn > 0 represents @ divergent278. Post-Stall Gyrations and Spins Chapter 11 situation in which the pitch acceleration is not zero. Having such a nonzero angular ac- celeration about any axis (in our case about the y axis) violates the basic assumption of constant @ that we have used do define a fully developed, steady spin. Said another way, any disturbance in a produces @ ACimp that tends to restore disturbances a to its initial g 2omb ue only if <0, value onty if = Figure 11.10 recapitulates and summarizes the pitching moment constraints for a fully developed spin by directly comparing two different aircraft, each with quite different dy- namic equilibrium possibilities. Aircraft B is clearly capable of sustaining a fully developed Upright spin at any angle above ag, insofar as the pitching moment equation is concemed. For this aircraft, Cmp and Se are always negative. However, airplane A can only meet the two constraints suggested above if a is in the shaded regions in the Fig. 11.10. Airplane A has two possible spin modes indicated in the sketch, while airplane B has only ‘one. To be sure, the pitching moment is not the whole story; the rolling and yawing moment equilibrium equations must also be considered and we should give our attention to how they impact dynamical equilibrium. But first, an estimate of w for each of the potential equilibrium conditions is useful and is easily obtained. WINGS. HORIZONTAL, =] RATES Fig. 11.11 Angular Velocity Components in a Typical Upright Spin Figure 11.11 also emphasizes that P = wcos a and R = asin a. for our idealized, fully developed spin. Substituting these expressions into eqn. 11.6b givesChapter 11 Post-Stail Gyrations and Spins 279 * =k) - 0? sinacose or, solving for ©, y y 24 __ (11.10) (iz ~ Ix)sin 2a, Equation 11.10 suggests that the minimum rotation rate occurs near an angle of attack ‘of 45°, though the nonlinearities in Crnp at these flight conditions may be strong enough to change the angle of attack for minimum rotation rate noticeably. But eqn. 11.10 tells only part of the story; it does not spell out all the conditions for equilibrium that must be met for 2 fully developed spin to occur. 11.1.8.2 Rolling and Yawing Moment Equations. Equations 11.9 suggest other constraints that must be met for a fully developed spin to occur. Although we did not ‘emphasize the point in the preceding paragraph, all the aerodynamic derivatives (even Cb) are functions of a, B, and «. Following Anglin and Scher'3, we have considered Crp to be solely a function of a. in spite of its dependence on other parameters. It is also convenient to take Cz and C, 2s functions of @ alone, even though there is little justification for these assumptions, other than that lateral-directional parameters are more directly linked to rotation rate, while the longitudinal parameters are usually more directly linked to angle of attack. These assumptions are useful to the understanding process, but Gmb. Cz, and Cp are all affected by a, B, and @. Next, consider the constraints imposed by the roling and yawing moment equations. o Ce Sua ae ee or } Cn > o Cz oF Cy< 0 restores Equilibrium @ equilibrium Gz or Cq,> 0 opposes Ce ‘ equilibrium or — Cy o aS = Fig. 11.42 Stabilizing and Destabilizing Slopes for Cz and Cy For a fully developed spin to occur the overall rolling moment and yawing moment coefficients, C,, and Cp, must be zero and their rates of change with respect to @ must be negative. The first pair of these conditions is explicitly stated in eqns. 11.9. The second pair of conditions is a requirement for stability very much like the one that says Crp, must be negative for static longitudinal stability to exist. Figure 11.12 illustrates this idea. If an increase in o produces and increase in either or both rolling and yawing moment, then one ‘or both of these total moments cannot remain in equilibrium for small changes ino. That is, any change in rotation rate will cause the autorotative moments to diverge away from the nominally stable rotation rate. For a true equilibrium state to exist, any deviation from280, Post-Stall Gyrations and Spins Chapter 11 this "trim" or equilibrium @ must cause an increment in both C, and C,, that tends to restore rolling and yawing moment to its equilibrium (zero) state. ‘These aerodynamic constraints are rarely all met exactly for an actual airplane spin. In other words, a true fully developed spin is an idealization and does not often ocour in practice. It is much more common to have oscillatory spins in which these conditions are almost met, but the true equilibrium is never quite reached. From an engineering perspec- tive, while it is still useful to base our concepts on idealizations like the fully developed spin, it is even more important to be able to estimate actual spin behavior from computed and wind tunnel aerodynamic data. Even though we expect less than perfect fulfillment of the assumptions listed, these estimates are quite important to safe and efficient flight test Planning. Consequently, the estimates are often in the purview of the flight test engineer. The following example shows how such estimates can be made. 11.1.8.3 Estimation of Spin Equilibrium States. Anglin and Scher!3 have de- scribed in some detail just how they estimated spin characteristics from both experimental ‘and computational evidence for a model of the McDonnell F3H Demon, a carrier-based fighter of 1960s vintage. Although their original use of this empirical estimation process was intended to help them save computer time by establishing appropriate initial condi- tions for numerical modeling of high angle of attack trajectories, it serves as an excellent ‘example of how to estimate post-stall behavior for flight test planning exercises. Later test Planning refinements likely will be based first on the results of computer modeling of the trajectories and, when they become available, certainly on previous post-stall fight tests. The procedure is summarized below and the interested student is referred to the original document'3 by Anglin and Scher for fuller details and a numerical example. To use this procedure for test planning, one must have some prediction of the vari- ation of rolling and yawing moments as functions of @ or, more commonly the nondimen- sional form e. The rotary balance is one means of obtaining these data experimentally ‘and that is how Anglin and Scher obtained their data. It should be emphasized that each of the aerodynamic derivative variations in this data base were obtained by steadily rotat- ing @ model about an axis parallel to the relative wind direction in the vertical wind tunnel. Therefore, no oscillation in the angular rates are part of these data. If there were aerody- namic hysteresis (and there usually is in high angle of attack aerodynamic coefficients), these data would not account for it. This limitation (and similar shortcomings) on the data should always be kept in mind when making estimations like these. Partly because the rotary balance rolling moment data for the F3H model were not “well-behaved”, Anglin and Scher ignored the rolling moment data and concentrated on the yawing moment. This simplification amounted to discarding two of the limitations listed Previously. The procedure then includes the following steps: ¢ Examine the pitching moment curve and the yawing moment curve from the rotary balance tests to locate nondimensional rotation rate(s) a for which Cp = 0 and for which HE <0. ‘There may be more than one of these potential equilibrium x) Points. This step is meant to identify rotational rates for which the aerodynamicChapter 14 Post-Stall Gyrations and Spins 281 yawing moment is zero. To strictly apply the constraints of the previous section, ‘we would have to repeat this process for rolling moment as well. Moreover, both conditions would have to be met simultaneously for a true developed spin mode to exist. Working with actual numerical data quickly convinces the practicing ‘engineer that such idealized conditions do not exist often. Next, utilize the rotary balance pitching moment data and the rotational rate (or rates if more than one exists) selected from the previous step to find a point on the pitching moment curve where the inertial moment and the aerodynamic moment ‘are identical. Notice that this second step in the estimation procedure assumes that pitching moments were measured in (or can be interpolated from) the rotary balance data sot for the SP at which Gy = 0 (as described in the preceding paragraph). Equation 11.10, rearranged to show the equality between the inertiat pitching moment and the aerodynamic pitching moment, forms the basis for this second step in the procedure. (z= IxJo? sin 20 Coy = atx} 20. 414.11 m pV7Sb ee Cm is @ function of 22 and must be chosen to correspond to the candidate equiio- tium value of 22 that came out of step one. To check for an aerodynamic moment equal to the inertial pitching moment coefficient calculated from eqn. 11.11, a plot of the measured aerodynamic pitching moment coefficients and the calculation from this equation can be made. Figure 11.13 illustrates such a plot. The intersection point represents a potential equilibrium point and, thus, a possible spin mode. Even though it is not illus- trated on the sketch in Fig. 11.13, it is quite possible to have more than one intersection point and, consequently, more than one spin mode. The angles of attack for which a spin ‘mode may exist are likely to occur in the vicinity of the angle of attack indicated by the point of intersection. Moreover, using this angle of attack and the value of Crp indicated, © for the estimated spin mode can be calculated. Then, individual roll and yaw rates can be estimated from this calculated @ and the estimated equilibrium angle of attack. Cm Possible Spi Equilibrium Condition Aerodynamic Component — Fig. 11.13 Comparison of Aerodynamic and Inertial Pitching Moment Coefficients ‘A comparison of the estimates that resulted from this procedure and computer simu- lations for the F3H Demon is shown in Table 3 (on the following page). It is unfortunate282, Post Stall Gyrations and Spins Chapter 11 that flight test results from the spin tests of the actual airplane are not available for comparison to the computational and to the wind tunnel predictions. Table 11.3. F3H Demon Computed Spin Modes Versus Estimated Spin Modes 36.0 1.88 37.0 2.18 F ‘Oscillated out of spin 1.89 453 51.8 2.18 218 620 80.0 4.72 3.50 515 36.5 2.80 2.69 365 11.2 POST-STALLISPIN TEST PREPARATION AND FLIGHT SAFETY Post-stall/spin flight testing is one of the most hazardous forms of testing. By the very nature of the flight regime, the skill of the test pilot and the knowledge of the test engineer will be stretched to new boundaries by such tests. After all, the test airplane and its crew are deliberately put at risk in order to make the vehicle safer for normal operations. Con- ‘sequently, flight test techniques are largely reduced to taking extraordinary preparation measures so that exposure to dangerous conditions is minimized. A large amount of data must be collected during fairly brief events; and, while the pilot's qualitative opinion is still valuable, he cannot adequately function as a sensor in such highly dynamic tests. He can be easily misled during such maneuvers. Test equipment and test procedures that are specifically designed to reduce the risk of catastrophic errors must be installed, verified, and used. As the title of this section differs from the "Test Method” of earlier chapters, so do the planning and conduct of post-stall/spin tests from the usual performance and stability and control tests. 11.2.1 Pre-Test Planning and Preparation Planning for any flight test must start several months before the vehicle rolls off the assembly line. The planning must include decisions like: © What are the objectives of the tests? What critical questions must be answered? ¢ What data are required to substantiate answers to the critical questions? How can these data best be collected? What hardware is essential and what instru- mentation is available? What specialized instrumentation is required? © What special training will be required by the crew? What criteria should be used in selecting the crew? What safety procedures are to be followed? ¢ How will the fight test support (chase aircraft, real-time test monitoring, video or ‘camera coverage, communications links, and the like) be provided? Is radar cov- erage required? What air traffic control coordination is needed?Chapter 14 Poat-Stall Gyrations and Spins 283 In answering such questions (and this list of questions is by no means exhaustive), the flight test team must be guided by two competing principles: ‘Test objectives are the prime drivers behind the test; they must be met. [Safety to the flight test crew, the test resources and to the public are important} (sometimes overriding) constraints to meeting test objectives. 11.2.1.1 Test Oblectives. Although setting clear test objectives is one of the first (both in importance and certainly in chronology) points that a fight test team must ad- dress, it is not always given the high priority is deserves. Engineers often think too much in terms of specific problems and quantified answers, without first being sure of what the purpose of their problem-solving is and determining if the problems they are solving are the ones that should be addressed. That approach to hazardous flight testing is courting disaster! For at least the last twenty years goverment test agencies have been required to produce a master plan for doing test and evaluation'4. Most companies have done similar planning for fight test certifications for at least that long. There is no phase of testing where such an overview of test objectives is more important than in post-stall fight testing. But most embryo flight test engineers ask: Are these objectives not the responsibility of the design team (or the system program office or the company project manager)? The answer is yes, but...1 The manager ultimately responsible for the success of the design (whatever his title) is the decision-maker who sets objectives and their priorities. But you, as his flight test “experts” are his primary adviser on whether or not and how his critical questions can be answered by flight test. As that “expert,” you will be called upon to help insure that all and only pertinent objectives are set, as well as being tasked to indicate how best to go about meeting them. To give professionally sound advice in this role means that you must understand the goals that should be set (you may even have to ex- plain to the project manager why certain critical questions must be asked) and you must appreciate how and why the test objectives are prioritized. You must be both responsible and responsive. The governing principles in setting test objectives are: (Do all demonstrations that are essential to meet the requirements laid down by the customer and/or the certifying ag a [Accept only those _objectives that contribute to meeting the requirements. Research or exploratory objectives are relevant only when the program itself is structured for research purposes or experimental exploration in flight is the best (or only ) way to credibly resolve an unknown. Quite clearly these two principles are paradoxical; they compete, sometimes even contradict! _The compromises they imply must be settled be- fore tests begin. ‘At about this stage you are probably ready for concrete examples — enough of these abstract platitudes! (Abstract they are, platitudes they are — but guiding principles worth remembering they also arel) So, let's look at some representative post-stall fight test objectives for high performance fighter built for USAF service and for a single en- gine general aviation airplane.284, Post Stall Gyrations and Spins Chapter 11 The first step in writing such objectives is to scrutinize all applicable requirements that pertain to the post-stall flight regime. For the fighter, applicable requirements would in- clude at the least the intent of MIL-F-8785C'5 (or MIL STD 1797A, which is not releasable for textbooks such as this onel) and MIL-S-83601 (USAF)S and the specific contractual paragraphs that modify or augment these high angle of attack specification documents. The following excerpts from the latter document spell out typical demonstration requirements that lead to “essential” test objectives. (In these excerpts specific numbered paragraph and table references, as well as most of the notes, are omitted for clarity.) Table 11.4. Progressive Test Phases for Post-Stall Demonstration Maneuvers Maneuver Requirements Test Control Application ‘Smooth Abrupt AOA] Tac- Phases AOARate | Rate | tical One g [Accel | One g | Accel. erated erated Ly Pitch control applied to achieve the spectfied| class: Class: ‘Stalls | AOA rate, lateraV/directional contro! inputs as. normally required for the maneuver task. wa fou | an paw | aw Recovery initiated after the pilot has a positive | Wi, IV | IH, IV | Mil, IV. indication of (6) Adefinite o-break, (b) @ rapid angular divergence, or (c} the aft stick stop has been reached and AOA is not increasing. B Pitch control applied to achieve the specified | cy, | Class: Stalls with [AOA rate, tateral/directional controls as re-| icles: | Crores | Ceeee: | ieee: | Chere Aggravated | quired for the maneuver task. When condition wm ym fa av Control inputs | (2), (b), oF (¢) from above has been attained, | ut, 1v | 1, rv | i, Vv [controls briefly misapplied, intentionelly or in response to unscheduled aircraft mations, be-| tore recovery attomet initiated. c Pitch control applied to achieve the specified Class: | Class: | Class: | Class: | Class: Stalls with }AOA rate, lateralidirectional controls as re- Aggravated | quired for the maneuver task. When condition] HL J hth | kW | kV | LV Control and | (a), (b). or (c) has been attained, controls are | i, tv | a tv | HM, IV ‘Sustained | misepplied, intentionally or in response to un- Control Inputs | scheduled aircraft motions and held for 3 sec- |onds before recovery attempt is initiated. D Pitch control applied sbruptiy, lateral” direc Class: | Clase: | Class: ‘Spin At- | tional controls as required for the maneuver | tempts" | task. When condition (a), (b), or (c) has been wwtauv | iw *"This phase required only for training aircraft which may be intentionally spun and for Class land IV aircraft in which sufficient departures or spins did not result in Test Phases A, B, or C to define cheracteristics, Elight Test Demonstration. Each airplane type shall demonstrate, by fight test according to Table 11.4, the degree of compliance with the stall waming, loss-of- control waming (when required), resistance to loss of control, loss-of-control prevention, out-of-control recover, and spin recovery criteria as specified in MIL-F-Chapter 14 Post Stall Gyrations and Spins 285 8785. The flight test investigation shall include an extensive evaluation of those maneuvers that may potentially result in AOA excursions to and beyond a permissible AOA flight limit, thereby subjecting the airplane to possible loss of con- trol. Unless otherwise specified, the use of prolonged pro-spin controls to sustain a spinning condition shall not be required except for trainer-type aircraft to be cleared for intentional spins. However, reasonably delayed recovery attempts after a stall or departure, or exaggerated misapplication of controls following a stall or departure, to simulate possible incorrect pilot responses, shall be investigated under the least conservative circumstances to ascertain the degree of spin resistance/susceptibility for operational users. When spins do result as a natural consequence of testing through departures from controlled flight, a satisfactory spin recovery technique shall be demonstrated in accordance with MIL-F-8785. The results of these tests shall be used to establish the stall and service flight limit AOA's. ‘StalSpin Eliaht Test Variables. The contractor shall establish, with the ap- proval of the procuring activity, which of the following parameters are variables to be tested for influence on stall and post-stall flight characteristics. (@) Configuration (b) Gross weight (©) Center of gravity (@) Stability and control augmentation system status (e) External store loadings” (f) Stall and departure speed, altitude, and attitude (9) Thrust effects (h) Gyroscopic effects “The external store loadings should include as a minimum: (1) No extemal stores (2) Symmetric, fuselage-heavy [if significantly different from (1)] (3) Symmetric, wing-heavy (4) Asymmetric (maximum allowable asymmetry) ‘The military specification® that most succinctly spells out these demonstration re- ‘quirements lays out a progressively more and more demanding series of test phases. The point in this progression where the airplane departs and/or spins determines whether or ot the vehicle Is departure-resistant or departure-susceptible and spin-resistant or spin- susceptible. Table 11.5 summarizes these demonstration requirements. Table 11.5 Definitions of Departure and Spin Susceptibility and Resistance ‘Test Phase Classification Departures. A- Stalls Extremely Susceptible | Extremely Susceptible B - Stalls with Aggravated Control ‘Susceptible ‘Susceptible Inputs: C= Stalls with Aggravated and Resistant Resistant Sustained Control inputs D- Spin Attempts Extremely Resistant | Extremely Resistant206 Pst-Stall Gyrations and Spins Chapter 14. Tables 11.4 and 11.5 are only a very small part of the whole of the military documents that govern how a post-stall test matrix must be put together for a military airplane. But they serve our purpose here which is to merely introduce the student to some of the re- quirements and point him to those documents that would completely govern a specific test. These documents and, most importantly the contractual specifications themselves, set the test objectives. Similarly, for civilian certification the following paragraphs from FAR Part 2316 drive post-stalll fight test objectives for a general aviation airplane. 23.21 Proof of Compliance (a) Each requirement of this subpart must be met at each appropriate combi- nation of weight and center of gravity within the range of loading conditions for which certification is requested. This must be shown — (1) By tests upon an airplane of the type for which certification is requested, or by calculations based on, and equal in accuracy to, the results of testing; and (2) By systematic investigation of each probable combination of weight and center of gravity location, if compliance cannot be reasonably inferred from combi- nations investigated. (b) The following general tolerances are allowed during fight testing. However, greater tolerances may be allowed in particular tests: Item Tolerances Weight +5%, - 10% Critical items affected by weigt c.G. 23.151 Acrobatic maneuvers Each acrobatic and utility category airplane must be able to perform safely the acrobatic maneuvers for which certification is requested. Safe entry speeds for these maneuvers must be determined. 23.201 Wings level stall (@) For an airplane with independently controlled roll and directional controls it must be possible to produce and to correct roll by unreversed use of the rolling con- trol and to produce and to correct yaw by unreversed use of the directional control, up to the time the airplane pitches. (b) For an airplane with interconnected lateral and directional controls (2 con- trols) and for an airplane with only one of these controls, it must be possible to pro- duce and correct roll by unreversed use of the rolling control without producing ex- cessive yaw, up to the time the airplane pitches. (c) The wings level stall characteristics of the airplane must be demonstrated in flight as follows: The airplane speed must be reduced with elevator contro! until the speed is slightly above the stalling speed, then the elevator control must be pulled back so that the rate of speed reduction will not exceed one knot per second until a stall is produced, as shown by uncontrollable downward pitching motion of the airChapter 11 Post-Stall Gyrations and Spins 287 plane, or until the control reaches the stop. Normal use of the elevator control for recovery is allowed after the pitching motion has unmistakably developed. (@) Except where made inapplicable by the special features of a particular type of airplane, the following apply to the measurement of loss of altitude during a stall: (1) The loss of altitude encountered in the stall (power on or power off) is the change in altitude (as observed on the sensitive altimeter test installation) be- tween the altitude at which the airplane pitches and the altitude at which horizontal flight is regained. (2) If power or thrust is required during stall recovery the power or thrust used must be that which would be used under the normal operating procedures se- lected by the applicant for this maneuver. However, the power used to regain level fight may not be applied until fying control is regained. (@) During the recovery part of the maneuver it must be possible to prevent more than 15 degrees of roll or yaw by the normal use of controls. () Compliance with the requirements of this section must be shown under the following conditions: (1) Wing flaps: Full up, full down, and intermediate, if appropriate. (2) Landing Gear: Retracted and extended. (3) Cow Flaps: Appropriate to configuration. (4) Power: Power or thrust off, and 75% percent maximum ‘continuous power or thrust. (5) Trim: 1.5Vs, oF at the minimum trim speed, whichever is higher. (©) Propeller: Full increase rpm position for the power off condition. 23.203 Turing flight and accelerated stalls Tuming flight and accelerated stalls must be demonstrated in flight as follows: (a) Establish and maintain a coordinated tum in a 30 degree bank. Reduce speed by steadily and progressively tightening the tum with the elevator until the airplane is stalled or until the elevator has reached its stop. The rate of speed re- duction must be constant, and: (1) For a tuming flight stall, may not exceed one knot per second; and (2) For an accelerated stall, be 3 to 5 knots per second with steadily in- creasing normal acceleration. () ‘When the stall has fully developed or the elevator has reached its stop, it must be possible to regain level flight without: (1) Excessive loss of altitude (2) Undue pitchup; (3) Uncontrollable tendency to spin; (4) Exceeding 60 degrees of roll in either direction from the established 30 degree bank; and (5) For accelerated entry stalls, without ‘exceeding the maximum permissi- ble speed or the allowable limit load factor. (©) ‘Compliance with the requirements of this section must be shown with:288. Post-Stall Gyrations and Spins Chapter 11 (1) Wing flaps: Retracted and fully extended for tuming flight and acceler- ated entry stalls, and intermediate, if appropriate, for accelerated entry stalls; (2) Landing Gear: Retracted and extended; (8) Cow! Flaps: Appropriate to configuration; (4) Power. 75% percent maximum continuous power; (©) Trim: 1.5Vg, oF at the minimum trim speed, whichever is higher. 23.205 Critical engine inoperative stalls (a) A muttiengine airplane may not display any undue spinning tendency and must be safely recoverable without applying power to the inoperative engine when stalled. The operating engines may be throttled back during the recovery from stall. (b) Compliance with paragraph (a) of the section must be shown with: (1) Wing flaps: Retracted. (2) Landing Gear: Retracted. (3) Cow! Flaps: Appropriate to level fight critical engine inoperative. (4) Power: Critical engine inoperative and the remaining engine(s) at 75 Percent maximum continuous power or thrust or the power or thrust at which the use of maximum control travel just holds the wings laterally level in the approach to stall, whichever is lesser. (6) Propeller: Normal inoperative position for the inoperative engine. (6) Trim: Level fight, critical engine inoperative, except that for an airplane of 6,000 pounds or less maximum weight that has a stalling speed of 61 knots or less ‘and cannot maintain level flight with the critical engine inoperative, the airplane must be trimmed for straight flight, critical engine inoperative at a speed not greater than 1.5V6,. 23.207 Stall warning (2) There must be a clear and distinctive stall waming, with the flaps and land- ing gear in any normal position, in straight and turing fight. (b) The stall waming may be furnished either through the inherent aerodynamic. qualities of the airplane or by a device that will give clearly distinguishable indications under expected conditions of fight. However, a visual stall waming device that requires the attention of the crew within the cockpit is not acceptable by itself. (©) The stall waming must begin at a speed exceeding the stalling speed by a margin of not less than 5 knots, but not more than the greater of 10 knots or 15 per- cent of the stalling speed, and must continue until the stall occurs, 23.221 Spinning (a) Normal category. A single-engine, normal category airplane must be able to recover from a one-tum spin or a 3-second spin, whichever takes longer, in ‘not more than one additional tum, with the controls used in the manner normal used for recovery. In addition — (1) For both the flaps-retracted and flaps-extended conditions, the appli- cable airspeed limit and positive limit maneuvering load factor may not be exceeded;Chapter 11 Post Stall Gyrations and Spins 289 (2) There may be no excessive back pressure during the spin recovery; it must be impossible to obtain uncontrollable spins with any use of the controls. 3) For the flaps-extended conditions, the flaps may be retracted during recovery. (b) Utility category. A utility category airplane must meet the requirements of paragraph (a) of this section or the requirements of paragraph (c) of this section, (©) Acrobatic category. An acrobatic category sirplane must meet the following requirements: (1) The airplane must recover from any point in a spin, in not more than one and one-half additional tums after normal recovery application of controls. Prior to normal recovery application of the controls, the spin test must proceed for six tums or 3 seconds, whichever takes longer, with flaps retracted, and one tum or 3 seconds, whichever takes longer, with flaps extended. However, beyond 3 seconds, the spin may be discontinued when spiral characteristics appear with flaps retracted. (2) For both the flaps-retracted and flaps-extended conditions, the appli- cable airspeed limit and positive limit maneuvering load factor may not be exceeded. For the flaps-extended conditions, the flaps may be retracted during recovery, if a placard is installed prohibiting intentional spins with flaps extended. (3) It must be impossible to obtain uncontrollable spins with any use of the controls. (@) Airplanes ‘characteristically incapable of spinning’. If it is desired to desig- nate an airplane as “characteristically incapable of spinning’, this character- istic must be shown with — (1) Aweight five percent more than the highest weight for which approval is requested; (2) Accenter of gravity at least three percent aft of the rearmost position for which approval is requested; (8) An available elevator up-travel four degrees in excess of that to which the elevator travel is to be limited for approval; and (4) An available rudder travel seven degrees, in both directions, in excess of that to which the rudder travel is to be limited for approval. ‘These requirements documents are the foundation upon which the flight test planner builds the rest of his preparation. The first step in laying out an overall test plan is to set these objective down and be sure that all interested parties agree to them. But not only must the test objectives be clearly stated, they must also be prioritized. Notice that in setting down objectives and relating them to requirements, no relative prior- ities were suggested. Before the test team can construct an efficient test matrix and do detailed test planning, the relative importance of each of the objectives must be estab- lished. This prioritization governs the order of testing as well as the allocation of test re- sources. Again, the flight test team and the design team must function as one entity in this290, Post-Stall Gyrations and Spine Chapter 11 phase of test preparation. There must be agreement between these two groups, even though complete unanimity on the final decisions is not likely. With the overall test objectives stated and prioritized, the flight test team can focus on how to acquire data best and other aspects (usually safety) of test preparation. 11.2.1.2 Data Requirements. In a sense the engineer reverts back to his profession in the next phase of preparation for post-stall testing. He leans on experience (both his and others) to decide what data are needed to answer the questions posed by the objec- tives. He should seek practical and efficient ways to generate and collect these data and lay out the manner of presenting data so that key questions will be clearly answered with quantitative engineering results wherever possible. He devises detailed test plans from which estimates of time, money, and people required to complete post-stall fight tests can be made. This phase of test preparation will necessarily be an iterative process. Available time and resources usually shrink during the course of a project; requirements that were overlooked in the goal-setting phase will often surface here. There are few general principles that apply to all classes of aircraft, but there are some guidelines to follow. (1) Question all data requirements. “Acid test" questions include: (2) Which objective do these data underpin? (b) What is the relative priority of the objective pursued? (¢) Who will use the data or will the data be used? (2) Is there a simpler, more cost-effective, or safer way to provide the essential data? (@) Did or can component testing/ground simulation provide similar data? (b) If s0, how much flight data are needed to verify component tests? (3) Will the proposed data matrix provide an adequate basis for management deci- sions later? (@) Is the experimental sample size statistically relevant? (b) Will test constraints likely bias or skew the results? (4) Is there a more efficient way to acquire the data in flight? (@) Could extended flight times (like air refueling might provide) allow coverage of more of the test matrix in each fight? (b) Is real-time test monitoring adequate to show when repeated test points are necessary? (c) How much redundancy in test support (radar coverage, chase aircraft, com- munications links, data paths, etc.) is needed? As these questions are considered, the importance of efficient, reliable data acquisi- tion will surface as a relevant issue. 11.2.1.3 Instrumentation for Post-Stall Flight Tests. The instrumentation suite for post-stall testing is usually more extensive and more specialized than is used for conven- tional stability and contro! testing. How does it differ? First, the ranges of the dynamic variables when the airplane is operating at high angle of attack are usually much greater. ‘Angle of attack variations from the trim condition may be +90° or more, instead of +15°, as is typically expected while in the normal operating envelope. Angular rates may be doubled or tripled over those encountered at low angles of attack. Accelerations may beChapter 11 Post Stall Gyrations and Spins 2 ‘about the same in magnitude, but the acceleration directions change much more rapidly during post-stall maneuvers. Table 6 shows typical differences between sensor ranges for post-stall fight testing and for conventional low angle of attack testing. But what kinds of sensors are needed and how do the motions encountered in the post-stall flight regime affect experimental measurements and how must the data be handled? Conventional parameters like those shown in Table 6 are ordinarily what is re- corded for post-stall tests; nothing out of the ordinary is on this list. But the list is not ex- haustive either, there are other measurements that are sometimes needed. We also must keep track of all control surface positions, throttle position, and a number of discrete ‘events like gear position, flap position, speed brake position, and the like. For the control positions, complete time histories sampled at moderately high frequencies are usually necessary. The sample rates for fast-acting surfaces should be an order of magnitude higher than the highest motion frequency of interest, even though the sampling theorem only requires that the sample rate be double the frequency of highest interest. Since terested only in the rigid body motion states. Of course, for analyzing structural modes or other high frequency issues, other sensors may dictate the critical sample rate. In any case the bandwidth of the instrumentation sensors must be carefully tailored to the post- stall fight regime. sors Needed for Stability and Control Data ‘Low Angle of Attack High Angle of Attack Typical Ranges Typical Ranges a Seto 15° 200° B 2° #30" P 270%/sec H00"%sec Q 220°/sec #50 sec, R #20°/sec £100%sec ¢ 230° 200° e 10° 280° yw 220° +360° Ty 40.259 40.59 a, 40.259 3059 fy +49 +49 11.2.1.4 Test Team Training. Typically, the test pilot is the only test team member singled out for special training during the preparation for post-stall training. This practice is a mistake! While the test pilot's background, training, and currency are the most meaningful and deserve considerable attention, the entire team needs to prepare for hazardous tests like these with deliberate focus. The underlying principle is:292, Post Stall Gyrations and Spins Chapter 11 Even when you prepare as fully as time and resources allow, post-stall tests often reveal unexpected characteristics and behavior. But, pre-test preparation that includes both the pilot (or pilots) and the entire engineering support team will do much to eliminate surprises. First, how should the pilot be selected and trained? There is no substitute for recent experience in post-stall testing, though the emphasis should be on overall experience, not just recency. The most valuable foundation is actual ost-stall testing in a broad array of airplanes. If the vehicle to be tested were my personal design, a test pilot who had done at least one other post-stall test program would be my choice. If the size and nature of the project permit (or even require) more than one test pilot, give this experienced pilot the responsibilty for training the other participants. in military programs, the senior pilot for the contractor typically will do most of the exploratory demonstrations. He should normally be tasked with the training and preparation of every pilot who fies a post-stall test flight. ‘What elements should this training and preparation include? One of the most valuable considerations in preparing to conduct high angle of attack testing is a formal training course similar to those given by the USAF or USN Test Pilot Schools, the Empire Test Pilots' School, or by the French Test Pilot School. Of course, not every company or en- trepreneur will have access to such personnel, but this reservoir of talent has the best basic background of formal training for post-stall testing. Once the project pilot is chosen, preparation has only begun. It is imperative that whatever his (or their) experience level is, there should also be recent experience in an aircraft quite similar to the type to be tested. If possible, this “in-fight simulation” should be flown into post-stall maneuvers of the most extreme type anticipated. if there is no aircraft available for actual post-stall practice, project pilots should seek out the most real- istic simulator they can find, preferably one where large amplitude motions can be expe- rienced. The time spent in such a device, especially if it has high fidelity both in motion ‘and in cockpit layout to the test airplane, is an ideal place for rehearsing and modifying both “normat" data collection (if data collection during post-stall maneuvers is ever normal) and emergency procedures. Realistic, repetitious practice of test profiles will give the pilots confidence, recent experience, and efficient procedures for data collection and allow them to make the best use of fight test time. These simulations should also include as much of the test team as will be involved in real-time monitoring and quick-look data reduction. The engineers/pilots who will act as test conductors and chase aircrew need to practice communication procedures and refine operational instructions. They should be just as familiar with safety rules (like minimum altitudes for recovery, altitudes for bailout, expected control sequences, and the like) as are the pilots who will fy the test vehicle. The fight test and systems engineers who will monitor key parameters (control surface positions, angular rates, engine performance, ‘emergency recovery subsystems, ground tracking, and all essential support details) should also rehearse their roles as part of the team. The senior flight test engineer should insist that every team member participate in one or more dry runs ~ including exposure to ‘emergency situations — if the entire team is to function efficiently. Every team member must understand and accept his or her responsibilities, know the procedures, andChapter 14 Post Stall Gyrations and Spins 293 ‘appreciate the consequences if he/she fails to meet his obligations during a test fight. Above all, hazardous flight testing is an exercise in group discipline. The test conductor can best guard against a breakdown in discipline with realistic practice sessions, both before test flights actually begin and periodically during a program. This continuation training is particularly important if there are long breaks between test phases or a significant number of new personnel must be assimilated into the test team. 11.2.2 Safety Precautions and Special Equipment Having discussed some of the general aspects of preparation for post-stall testing, we now tum to more specific ways to make such tests safer. We should have no illusions. Post-stall testing is not inherently safe; the test airplane and the crew will be deliberately put at risk in either marginally controllable or out-of-control portions of the flight envelope. ‘As we have already emphasized, all flight testing should be planned to start with benign flight conditi ind_progress incrementally to more difficult or more dangerous ones. This principle is called the "build-up approach" and is designed to preclude unwittingly pushing past boundaries beyond which there is no recovery. However, post-stall testing involves flights where nonlinear dynamical relationships and the large number of variables that are involved make it impossible to perfectly determine in advance safe increments in test parameters. The only altematives that can reduce risks are (1) reduce the likelihood of potential failures and (2) take extraordinary precautions to mitigate the consequences of events that are still likely to occur. This section provides practical guidance for the flight test engineer planning a high angle of attack test series. It involves both procedural and hardware suggestions; and, though it would be highly presumptuous (and patently false!) to claim that these suggestions are comprehensive, they are a very good starting point. 11.221 Emergency Recovery Devices. Whenever anyone lays out a serious post- stall fight test program, inevitably these questions arise: is an emergency recovery subsystem needed? If so, what type and what kind of| design requirements must be spelled out? The first question should almost always be answered yes. Only airplanes with tightly lim- ited flight envelopes or very unusual design characteristics can avoid encounters with post-stall flight. Even then, very few designers are willing to bet their reputations that no post-stall fight tests at all are necessary. Even “spin-proof" airplanes have been involved in post-stall accidents. However, how answering the second question is not so easy. It cannot be answered without consideration of the following additional question: What is the intended use of the design (or modification of an existing design)? It makes no sense to do full-blown spin tests, for example, for a large commercial air- liner. Thus, the “post-stall" program for such a vehicle is almost exclusively aimed at precluding stalls — a very limited "post-stall" (if indeed it can be called by that name) test ‘series is in order. Most business aircraft with near centerline thrust should be treated in the same way. However, past experience has shown that even these kinds of airplanes can have post-stall catastrophes if the configuration is quite unusual or not well-known. For example, at least one early T-tail transport design ran into such difficulties, as have294 Post-Stall Gyrations and Spins Chapter 11 corporate jets'® with similar design features. There is also reason to pay careful attention to post-stall certification for canard and three-surface configurations. Is it responsible to ignore post-stall testing based on handbook limitations in- tended for the design? Now this question is harder to answer and we are treading dangerously close to subjective opinion entirely. But, to offer a design to any segment of the flying public with- out having explored post-stall behavior is neither wise nor responsible. (It may also be legally indefensible in today's litigious judicial environment.) No matter what the contract- ing agency or certification authority requires, a professional design team owes that con- sideration to their customers. Please note that this professional obligation is related to vehicle characteristics and intended use, not certification requirements alone. The design process is simply not complete until predictions and engineering estimates have been validated to some margin slightly beyond the published fight envelope. ‘There are two types of devices usually considered for emergency recovery from sping; those which modify the ratio of aerodynamic and inertial forces and those which add additional contro! authority about one or more axes. Far and away the most commonly used emergency recovery scheme is some form of parachute which alters the pitching and yawing moments rather drastically. They have been used on almost every type of airplane {from sailplanes to 30-ton fighters. ANGULAR RATES ‘AND ACCELERATIONS | DYNAMIC PRES- ‘SURE RANGE MASS DIS- TRIBUTION L__ cockerr RECOVERY PARA. JETTISON CONTROLS CHUTE SUBSYSTEM MECHANISM CANOPY SZE AND DEPLOYMENT| ee MECHANISM SUPPORT. RISER ‘STRUCTURE/ ‘PACKAGING eee Fig. 11.14 Emergency Parachute Recovery Subsystem Design Parameters Design and validation of a spin recovery parachute subsystem’? is peculiar to each configuration and thus often becomes the responsibility of the project flight test engineer. the inertia distribution matrix, the likely equilibrium spin rotation rates, and an estimate of the dynamic pressure range over which the parachute must operate are key design pa-Post-Stall Gyrations and Spins 295 (Chapter 14 rameters. Figure 11.14 suggests the compromises associated with designing such a subsystem and how each of these parameters are affected by one another. Sizing of the canopy and its porosity are fundamental to the success oF failure of this ciitical subsystem. Typically an off-the shetf canopy can be used and canopy character- istics are fairy completely documented20. Drag coefficients and opening shock load overshoots may be estimated empirically, but often ground tests are necessary to validate the assumptions. The numerical values for these parameters are strongly dependent on canopy porosity. The canopy material, its shape, and whether the canopy is solid, slotted, or vented in any way all affect the parachute drag coefficient and the opening shock loads drastically. These short duration loads are often considerably higher than the steady ‘aerodynamic loads. Increasing the porosity and reefing the canopy so that it opens more slowly both tend to lower opening shock and thus reduce the severity of the impact loads on the risers and the attachment structure. A more porous canopy is also less likely to oscillate at the end of the risers afier it is fully inflated. In fact oscillatory instability of the canopy can cause the airplane-parachute combination to be dynamically unstable after recovery from the out-of-control condition. Such dynamic behavior can be very uncomfortable (and quite unnerving!) to the pilot who is depending on this device to recover from this kind of situation. However, since a recovery parachute is typically jetti- ‘soned shortly after it performs its function, dynamic instability of the airplane-parachute combination is often of secondary importance. The riser design and length also are quite important'®. The risers should be long ‘enough to guarantee that the canopy opens well away from the aircraft itself. The likely ‘worst case rotation rates and attitude combinations should be studied, es well as probable low pressure, separated flow regions, so provisions can be made to forcibly propel the parachute canopy into relatively clear air before the opening sequence is allowed to begin. For this reason, canopy bags or reefing, along with drogues or ejectors are typically used to extract the parachute from its storage canister. Once the canopy opens successfully, riser length is another important factor affecting the oscillation frequency. Consideration should be given to keeping this frequency distinct from either the longitudinal short period or the lateral-directional Dutch roll frequencies of the rigid airplane. Rocket thrusters have been used as an altemative to spin parachutes2". This emer- gency recovery subsystem is an example of adding extra control devices to guarantee recovery from out-of-control situations. Potential advantages for such devices include: They can be installed so as to produce controlled moments about each axis. Multiple units can provide redundancy (or capability to make muttiple recoveries) if ‘the weight penalty is not too severe. Obviously, this capability could also be pro- vided by rocket motors that can be restarted. Rarely, however, would this level of complexity be justified. ‘Such thrusters could also be used to drive the airplane into different spin modes (for research purposes), though such usage has never occurred to the author's knowledge. Disadvantages to such devices include: ‘© Usually they involve additional complexity, both from a mechanical perspective and in terms of safe operation. Rocket thrusters almost always multiply safety206. Post Stall Gyrations and Spins Chapter 11 ‘concems because of the larger number of failures and the consequences of some of those failures that are possible, % More often than not, the thrusters are pyrotechnic devices (or are susceptible to explosion at least). Again, safety issues are harder to resolve satisfactorily when multiple pyrotechnics must be cartied for each flight. tis difficult to safely verify the adequacy of recovery thrusters for their intended Purpose. Of course, both rocket thrusters and a spin recovery parachute can be installed, but rarely are. One such exception22 was the F-100F, a version of the Super Saber that was highly susceptible to a flat spin from which aerodynamic controls could not recover the airplane. Both a spin chute on the aff fuselage and rocket thrusters on the wing tips were installed for the post-stall testing. Unfortunately, when the test aircraft entered the ‘expected flat spin mode, the recovery rockets did not stop the rotation and the parachute streamed and wrapped around the vertical tail. The test airplane was lost when both ‘emergency recovery devices failed to perform their intended function. Finally, the flight test team must pay careful attention to how emergency recovery devices are actuated, controlled, monitored, and terminated (jettisoned, in the case of parachutes). First, all controls for this system must be accessible to and easily identified by the pilot even under the most extreme conditions of aircraft loading and stress. The pilot must be able to activate and deactivate (or jettison) this emergency system even ‘when cockpit g levels are high and oscillatory. (This fact may also suggest part of the reason why additional pilot restraints are often used for post-stall testing.) Moreover, there must be reliable and complete warning of any matfunction or other condition that affects pilot actions with regard to the emergency recovery subsystem. Indicator lights are one means for providing these wamings. Parachutelthruster arming, position of jettison mechanism, test limits (angular rates, minimum altitudes, and the like), flight control sys- tem status, and engine health are typical candidates for waming/caution displays during post-stail tests. 11.2.2.2 Chase Aircraft In addition to a reliable emergency recovery subsystem, post-stall test teams often rely heavily on chase aircraft. Typically, these support aircraft are flown by backup pilots for the project and they often carry aerial photographers. The functions of this supporting aircraft and crew include: Visually checking the test aircraft for configuration; for loss of access covers, gear doors, etc.; and for obvious structural or system failures during a maneuver. © Clearing the airspace for the test vehicle. Calling out critical events — minimum altitudes, engine failure, and emergency procedures — to the test aircrew. ¢ Advising the test pilot and the test conductor of any anomalies observed. The chase pilot is a critical supporting player. His piloting task is very demanding and he needs to have powers of observation every bit as keen as those of the test pilot. He must be completely familiar with all planned mission profiles, overall test objectives, as well as detailed test restrictions. To meet all these responsibilities, project pilots who will fly such chase sorties should practice for them -- just as the rest of the test team should rehearse. For the test pilot and the supporting chase aircrew, this practice time is abso- lutely mandatory.Chapter 11 Post-Stall Gyrations and Spine 27 14.2.2.3 Procedural Precautions. The whole of section 11.2 is built around the premise of test team discipline. Any engineer or pilot who is charged with the overall re- sponsibility for this class of flight testing must be a stickler for discipline. Insisting that comprehensive, written procedures are developed, followed, and modified constructively is ‘a mark of professionalism. It is absolutely necessary for the successful post-stall test team leader to demand discipline. Here are some practical suggestions for setting up such procedures: Establish some form of independent review process to examine the planning and procedures. ‘The scope of this review can vary from a single experienced con- sultant to a full-blown, structured “red-blue” team approach. The effort expended ‘on this review will largely be dictated by the size and economy of the project, but it should never be omitted. © Test procedures, especially for hazardous tests like post-stall tests, evolve from safety constraints wrapped around clearly stated test objectives. But safety is overriding. Be sure procedures are communicated. Keep them simple, straightforward, and to the point. Practice and repeat the good ones; revise the bad ones. Fig. 11.15 Post-Stall Flight Test Team Communications Net ° Keep the command lines during a test crisp and clear. The test conductor is the maestro (he carries the responsibilit ); the test pilot is the soloist; and the chase pilot is the harmony. But everyone else on the test team plays an essential back- ground instrument. If that musical metaphor sounds overdone — nobody talks to the pilots on the radio except the test conductor, or at the test conductor's request. ‘Supporting test engineers relay pertinent information to the test conductor, who decides what is essential to the conduct of the test. Flying a productive test sortie is always a demanding task; fying a productive post-stall mission can be downright hectic. It is the test conductors job to see that mission objectives are not298. Post-Stall Gyrations and Spins Chepter 11 destroyed by confusion and indecision. Figure 11.15 illustrates this decision making net for a typical post-stall fight test. ¢ Guarantee that everyone on the test team knows the envelope limits and the rea- sons for them. The senior flight test engineer must take the lead in this training; everyone must be completely aware of each fight's objectives — both primary and secondary — and the steps planned to reach them. © Never go beyond a test plan. This rule should be iron-clad; you simply must not “shoot-from-the-hip” in this environment. True, the detailed flight profiles should always plan more than can be accomplished; but the altematives and the relative Priorities are to be spelled out in the briefing on the ground before takeoff. If these six guidelines are followed, deliberate, disciplined procedures will result and the whole test series will reflect that disciplined philosophy. 11.2.2.4 Real-Time Monitoring of Post-Stall Tests. The engineering process that we suggest needs to be so tightly disciplined must still react efficiently. Adequate infor- mation fed back to the test support engineers during post-stall maneuvers is essential to the process. Larger and adequately funded post-stall programs may have all relevant dynamic time histories telemetered back to the control room during a fight. These fortu- ‘nate engineers may even have the computer power and software sophistication available to cross-plot and partially analyze these real-time data. They might, for example, super- impose dynamic states on a-f axes, rather than just time axes. Graphic images of the airplane's rotation and inertial attitude may be useful. Plots of predicted boundaries and actual motions could be overlaid. For such a rich data environment the possibilities for real-time monitoring are almost infinite At the other extreme, where there may only be radio communications between the test pilot and the chase pilot, the real-time analysis and advice may be quite subjective. But that fact does not make it unimportant. Any outside observer's observations can be useful information (data) in this often confusing test environment. Recording a running commentary from both the test pilot and the chase pilot are wise and inexpensive “real-time” tools in post-stall tests. The point is: real-time information — at any level between the extremes described in this paragraph — is valuable and must not be ignored. Plan to collect it. 11.3 SUMMARY In this concluding chapter we have introduced the student to what is in reality an ad- vanced topic in flight test. Most of the principles that apply to post-stall testing also apply to other forms of hazardous test: flutter envelope expansion, weapon separation tests, low level avionics system (like terrain-following) evaluations, and similar activities. The major themes of careful planning, thorough preparation, and unyielding discipline carry ‘over into these other forms of fight testing as well. Of course, the engineering tasks, measurement techniques, and support required will differ. But these basic principles will still be valid. In that sense, this chapter serves to draw attention to the three most themes of all kinds of flight testing, even as it serves to introduce advanced topics through a specific example. And those themes — planning, preparation, and discipline — pervade all ‘ ;Chapter 14 Post-Stall Gyrations and Spins 209 REFERENCES 1 2 Herbst, W. B., "Future Fighter Technologies," Journal of Aircraft, Vol. 17, No. 8, Aug. 1980, pp. 561-566. Gilbert, W. P. and Gatlin, D. H., "Review of NASA High-Alpha Technology Program," Paper at NASA High-Angle-of-Attack Technology Conference, NASA Langley Re- search Center, Oct. 30-Nov. 1, 1990. Nix, J.,"X-31 EFM: The Experiment Continues,” Paper at NASA High-Angle-of-Attack Technology Conference, NASA Langley Research Center, Oct. 30-Nov. 1, 1990. Trippensee, G. A. and Ishmael, S., "Overview of the X-29 High-Angle-of-Attack Pro- gram," Paper at NASA High-Angle-of-Attack Technology Conference, NASA Langley Research Center, Oct. 30-Nov. 1, 1990. Flight Training Handbook, AC-61-21A, Federal Aviation Administration, 1980. Military Specification, MIL-S-83691(USAF), “Stall/Post-Stall/Spin Flight Test Demon- stration Requirements for Airplanes," Apr. 1972. McElroy, C. E. and Sharp, P. S., "Stall/ Near Stall investigation of the F-4E Aircraft,” FTC-SD-70-20, Air Force Flight Test Center, Edwards Air Force Base, Califomia, Oct. 1970. Babister, A. W., Alrcraft Stability and Control, Macmillan Company, New York: 1961 Kerr, T. H., "General Principles of Spinning," AGARD Flight Test Manual, Volume Il, Chapter 8, Pergamon Press, New York: 1962. Chambers, J. R. and Stough, H. P., Ill, "Summary of NASA Stall/Spin Research for General Aviation Configurations,” AIAA Paper 86-2597, Oct. 1986. Stough, H. P., Ill, DiCarlo, D. J., and Steward, E. C., "Wing Modification for Increased Spin Resistance,” SAE Paper 830720, Apr. 1983. ‘Tumer, H. L., "Measurement of the Moments of Inertia of an Airplane by a Simplified Method," NACA TN 2201, Ames Aeronautical Laboratory, Moffett Field, California, Oct. 1950, Anglin, E.L. and Scher, S. H., "Analytical Study Aircraft-Developed spins and Deter- mination of Moments Required for Satisfactory Spin Recovery,” NASA TN D-2181, Feb. 1964. DOD 5000.3, “Test and Evaluation,” Department of Defense, Dec. 1975. Military Specification MIL-F-8785C, "Flying Qualities of Piloted Airplanes," Nov. 1980. Federal Aviation Regulations, Part 23, Airworthiness Standards: Normal, Utility, ‘Aerobatic and Commuter Category Airplanes, Federal Aviation Administration, June 1974 (Changes 1-26, as of Oct. 1989). "British Aircraft Corp. Stil Aims at Initial BAC 111 Delivery Dates," Aviation Week and Space Technology, Nov. 11, 1963, p.45. "Prototype Canadair Challenger Lost,” Flight International, Apr. 19, 1980, p. 1174. Burk, S. M., "Review of Airplane Spin-Recovery Parachute Systems Technology,” Paper at the AFFDLIASD Stall/Post-Stall/Spin Symposium, Wright-Patterson AFB, Ohio, Dec. 1971.Bg Post-Stall Gyrations and Spins Chapter 14 Bixby, H. W., Ewing, E. G., and Knacke, T. W., "Recovery Systems Design Guide,” AFFDL-TR-78-151, Air Force Flight Dynamics Laboratory, Wright-Patterson AFB, Ohio, Dec. 1978. Knacke, T. W., "Recovery System Design Manual,” NWC TP 6575, Naval Weapons Center, China Lake, Califomia, Jul. 1985. Farmer, G. H., “Flight Test Investigation to Determine the Effectiveness of a Rocket as an Emergency Spin Recovery Device,” WADC Technical Report 53-396 (AD 76422), Wright Air Development Center, Ohio, Aug. 1958. Wykes, J. H., "An Analytical Study of the Dynamics of Spinning Aircraff, Part Ill, Cal- culated and Flight Test Spin Characteristics of an F-100F with Strakes,” WADC Technical Report 58-381 (AD 243600), Wright Air Development Center, Ohio, Feb. 1960,UST OF SYMBOLS CHAPTER 1 None CHAPTER 2 a speed of sound... 0 standard day sea level speed of sound c volume of instrument chamber.. GQ lift coefficient. internal diameter of tubing or calibrated attitude measured at the altimeter ea calibrated altitude measured at the eyepiece Nrcalenp Calibrated attitude messured on the parking ramp.. realpas, calibrated altitude of the pacer sircraft heah test day calibrated altitude... by geometric attitude. . geopotential altitude at the base of the ith layer oF indicated altitude read from the altimeter... indicated nominal altitude for the test point pressut standard day pressure at the base of the ith layer..Index 2 ‘Symbol UIST OF SYMBOLS (Continued) Definition CHAPTER 2 (continued) Ps Pr Po Peo PT _ qog te AMcat, calibrated altitude error measured at the altimeter... standard day sea level pressure of the atmosphere freestream static pressure... freestream total pressure dynamic pressure.. instrument corrected dynamic pressure. . test day instrument corrected dynamic pressure ... sensed dynamic pressure universal gas constant radius of the earth indicated temperature. time at beginning of speed course run. time at end of speed course run. temperature temperature at the base of the ith layer or indicated temperature .. instrument corrected temperature standard day temperature. _ knots or fps knots or fps. calibrated airspeed... knots or fps. calibrated airspeed for the pacer aircraft. knots or fps knots or fps knots or fps knots or fps knots or fps knots or fps equivalent airspeed. indicated airspeed. indicated nominal airspeed for the test point instrument corrected airspeed. : test day instrument corrected airspeed. distance from eyepiece to plane of sighting grid ora variable... a height of airplane as measured on the sighting grid or a variable. distance from eyepiece to plane of tower flyby line. height above the plane of the eyepiece.UST OF SYMBOLS (Continued) ‘Symbol Definition CHAPTER 2 (continued) Alrcalyeece Calibrated altitude error measured at the eyepiece . Area calibrated altitude error for the pacer aircraft Ate altimeter instrument error correction. Messer altimeter instrument error correction for the pacer aircraft Aig, test day altimeter instrument error correction... altimeter instrument error correction for the nominal test point Vie sirepeed instrument error correction knots or fps AVicpucar _ait8pe0d Instrument error correction for the pacer .» Knots oF fps MViewe instrument corrected airspeed forthe nominal test point. . knots oF fps Vp —_airspeed static position error correction. . knots or fps AV pepe airspeed static position error correction for the pacer aircraft. . knots or fps angle of attack : idea ratio of specific heats for air. A).. standard day sea level viscosity slugsift-sec density of the atmosphere. standard day density at the base of the ith layer std standard day density... Prropopause standard day density at the tropopause. er total density... Po standard day sea level density of the’atmosphere.. s density ratio... std standard day density ratio tropopsuse Standard day density ratio at the tropopause.. temperature ratio...ron ‘Symbol ST OF SYMBOLS (Continued) Definition CHAPTER 3 (only those symbols not previously introduced are listed) AF acceleration factor. constant for induced drag expression drag or... engine characteristic length. standard day drag... test day drag. total energy (kinetic plus potential) span efficiency factor : none time rate of change of pressure altitude . fps ratio of specific heats none aerodynamic lif. ‘standard day Mach number. standard day/weight load test day/weight load factor .. vertical component of true ait Specific excess POWEF ace test day specific excess power. radius of curvature indicated freestream temperature. instrument corrected freestream temperature. standard day freestream temperature . test day freestream temperature .. thrust horsepower available thrust horsepower required ... standard day thrust horsepower required... test day thrust horsepower required horizontal component of true airspeed .. Power off stall speed, ianding configuration. Power off stall speed, clean configuration. vertical component of true airspeed ... knots, mph, or fps knots, mph, or fps knots, mph, or fps knots, mph, or fpsIST OF SYMBOLS (Continued) corrected (for weight change) indicated attitude . he energy height or specific energy height he time rate of change of energy height (= wy fuel flow rate AD; increment of induced drag. Ah increment of altitude.... Ahe increment of absolute altitude Ahe increment of energy height or specific energy height. increment of indicated attitude.... AP ou Pan ATp At aw Ven . Ay increment of fight-path angle. 8 test day pressure ratio . Bg standard day pressure ratio. Y flight-path angle . or angle between the thrust line and the fuselage reference line Pt test day density. sium CHAPTER 4 (only those symbols not previously introduced are listed) Aor A; constant for first term in power equation .... Bor By or number of blades for a propeller .. BHP, _ brake horsepower required BHP, __ test day brake horsepower available. BHPsg standard day brake horsepower available.Inder ‘Symbol Definition UST OF SYMBOLS (Continued) CHAPTER 4 (continued) Co, parameter for solving Buckingham Pi equations. drag coefficient . induced drag coefficient. parasite drag coefficient. propeller integrated design lift coefficient aircraft pitching moment coefficient power coefficient. torque coefficient parasite drag coefficient... propeller local chord.. oF parameter for solving Buckingham Pi equations. or specific fuel consumption (SFC). blade section design lift coefficient... thrust specific fuel consumption (TSFC). FAQ eee or propeller diameter... endurance... maximum endurance. propeller advance rati propeller torque meter scale factor. ratio of equivalent airspeed to equivalent airspeed for minimum power. fundamental unit for length. test day manifold pressure standard day manifold pressure .. fundamental unit for mass (slugs propeller rotational rate ‘or number of fundamental variables. generalized power. : number of fundamental units... propelier torque... . instrument corrected propeller torque meter reading . propeller radius... range... radius to a given airfoil section on a propel propeller hub radius ... solidity ratio. temperature in the jth engine component...UIST OF SYMBOLS (Continues) dex? ‘Symbol —_Defintion Units CHAPTER 4 (continued) qt t Ven Vemp vw CHAPTER 5 (only those symbols not previously introduced are listed) a a CFL Chae Chan BFL F Fr Miran a m Py RA Re Sa Sa00 So. SDEC test day temperature... fundamental unit for time. true airspeed for maximum endurance equivalent airspeed for minimum power. generalized true airspeed aircraft weight at the end of a cruise segment aircraft weight at the beginning of a cruise segment. ‘weight flow rate of air through an engin net thrust from the jet exhaust for a turboprop.. ith nondimensional grouping from Buckingham Pi theorem. acceleration during takeoff or landing... average acosleration during takeoff or landing critical field length maximum lift coefficier lift coefficient during transition segment. balanced field length net accelerating force during takeoff roll frictional force during takeoff rot. vertical height at end of takeoff transition lift during landing. mass of the airplane of object studied average excess power during takeof.... runway available (actual runway length less allowance for lineup)... ft radius of curvature for takeoff or landing trajectory. ft horizontal distance to clear an obstacle of specified height (after liftoff for takeoff or before touchdown for landing) horizontal distance to clear an obstacle of specified height (during a single engine takeoff. horizontal distance during acceleration to climb speed. horizontal distance covered between engine failure and initiation of braking during an aborted takeof?..ed ‘Symbol UST OF SYMBOLS (Continued) Definition Units CHAPTER S (continued) Ser Sra $6 So, Sep SLO SLO Siz SREF SsToP horizontal distance from brake release to engine failure... horizontal distance from obstacle clearance to landing flare. horizontal distance from brake release to liftoff. horizontal distance from touchdown to complete sto horizontal distance covered during takeoff rotation horizontal distance from brake release to liftoff. horizontal distance from brake release to attaining Vace horizontal stopping distance after application of brakes. horizontal distance covered during takeoff (or landing) transition . horizontal distance during acceleration to climb speed. thrust thrust during landing . average excess thrust during takeoff time for rotation during takeoff or landing .. true airspeed where average excess thrust occurs true airspeed at initiation of braking during an aborted takeoff. critical engine failure speed... true airspeed during initial climb segment of takeoff. true airspeed at liftoff . true airspeed at liftoff based on 7.2V, refusal speed..... air minimum control speed ground minimum contro! speed .. safe single engine speed feggiggz. ZF wind speed during takeoff or landing. minimum takeoff speed for FAR Part 25, weight of aircraft during landing .. increment of lift coefficient. increment of load factor. change in pitch angle during takeoff transitiorUST OF SYMBOLS (Continued) (decd ‘Symbol —_Defintion Units CHAPTER 6 (only those symbols not previously introduced are listed) A a 4 2 B b Che Cho Sra Cn, Cony constant coefficient in stick force equation... lever arm distance in gearing mechanism for control stick.. lift curve slope produced by deflection of the jth control surface. per * or radian lift curve slope produced by elevator deflection......... per * or radian ‘constant coefficient in stick force equation lever arm distance in gearing mechanism for control stick.. elevator hinge moment coefficient. hinge moment coefficient at zero lift and zero control surface deflection. | rate of change of hinge moment coefficient for the jth contro! surface with changes in local angle of attack .. . per ° of radian rate of change of hinge moment coefficient with elevator deflection : rate of change of hinge moment coefficient for the jth controt surface with changes in surface deflection. rate of change of hinge moment coeficiont forthe th control surface with changes in surface tab deflection lift coefficient with the elevator free to float.. trim lift coefficient produced by the jth control surface... trim lift coefficient for the airplane. lift curve slope for the airplane . lift curve slope for the airplane with elevator floating . lift curve slope produced by the jth control surface. lift curve slope for the horizontal tal lift curve slope for the wing-body.. change in lift coefficient with elevator deflection ...... change in lift coefficient with defection of the th control surface .. zero lift pitching moment coefficient about cg... zero lift pitching moment coefficient about cg with the elevator free to float zero lift pitching moment coeticiont for the wing-body alone... ..none change in pitching moment coefficient with change in «....per radian change in pitching moment coefficient with change in a with the elevator free to float... change in pitching moment coefficient with elevator deflection (elevator effectiveness)... lever arm distance in gearing mechanism for contol stick... .. per radian per * or radian ftIndex ‘Symbol UIST OF SYMBOLS (Continued) Definition Units CHAPTER 6 (continued) meagoa mean aerodynamic chord. elevator chord. center of gravity . lever arm distance in gearing mechanism for control stick. lever arm distance in gearing mechanism for control stick. free elevator factor. stick force gearing ratio (or gain). elevator hinge moment. horizontal tail incidence angle . net aerodynamic rolling moment about the x axis control stick lever arm in gearing expression... distance from aircraft og to horizontal tail aerodynamic center Net aerodynamic pitching moment about the y axis. net aerodynamic yawing moment about the z axis .. x component of angular velocity. y component of angular velocity. dynamic pressure. z component of angular velocity. wing planform area (reference area) stick-fixed static margin. stick-free static margin elevator surface area.......... horizontal tail planform area control stick travel... 7 7 in x component of linear velocity of the aircraft center of gravity ......{ps Component of linear velocity of the aircraft center of gravity ..... fps horizontal tail volume coefficient. = none horizontal tail volume coefficient with elevator free to float........none horizontal tail volume coefficient with the og located at the neutral point. 7 -none horizontal tail volume coefficient with the og located at the neutral Point with the elevator free to float ... Rone vertical velocity corrected for engine thrust... -none Z component of linear velocity of the aircraft center of gravity ..... fps weight of the /th control surface. ibs x component of the net aerodynamic force.. bs y component of the net aerodynamic force.. lbs Z component of the net aerodynamic force....... lbs. | | | j | | | |‘Symbol UST OF SYMBOLS (Continued) Irdentt Definition Units CHAPTER 6 (continued) Xecue Xog x np. b ees woes erey aerodynamic center of the wing-body.. location of the center of gravity along the x axis. location of the jth control surface cg relative to the hinge change in pitching moment coefficient... aircraft angle of attack with the elevator free to float trim angle of attack angle of attack of the wing-body. ical arg of air at te fh con ein elevator angle to trim . control surface deflection of the th contro! surface. free floating angle of the jth control surface.. rudder deflection... control surface tab deflection .... control surface tab deftection to trim... ffight-path angle in the power approach flight condition .. downwash angle downwash angle when wing-body lit is zero.Indon-12 UIST OF SYMBOLS (Continued) ‘Symbol Definition CHAPTER 7 (only those symbols not previously introduced are listed) be span of the elevator Cy pitch damping derivativ Cinay tail contribution to pitch damping derivative.. Cy normal force coefficient. kK correction coefficient accounting for wing-body contributions to the pitch damping derivative Lerim aircraft lift in the trim or wings-level equilibrium state. MM stick-fixed maneuver margin ny normal acceleration of the aircraft og 19 load factor in equilibrium (trimmed) fight.. % a dynamic pressure at the horizontal tail. Xmp maneuver point Xmp" maneuver point with the elevator free to float. Cp, change in elevator hinge moment coefficient AFs change in stick force... — Aq change in pitch rate... ba change in aircraft angle of attack ‘ay change in angle of attack at the horizontal tal. Act Ady pu # e change in angle of attack at the wing change in elevator deflection. aircraft angle of attack during a wings-level pull-up relative mass parameter. ‘change in local angle of attack at the horizontal tail during a wings-level pull-up ... ratio of elevator It effectiveness to stabilizer lift effectiveness none + Q aircraft inertial tum rate. CHAPTER 8 (only those symbols not previously introduced are fisted) a, side force curve slope produced by rudder deflection. Chop change in hinge moment due to change in local angle of attack none Che, change in hinge moment due to rudder deflection... CL lift curve slope with ailerons deflected. Ce rolling moment coefficient. Cy dihedral effect... Ce, roll damping derivative. Ce rolling moment change due to yaw rate. C4, aileron control effectiveness. Ca, change in rolling moment coefficient due to rudder deflection ... - per ° or radian‘Symbol UST OF SYMBOLS (Continued) Index Definition Units: CHAPTER & (continued) Cn Cop Coy yawing moment coefficient.. change myevingronent oaient due orl ae ‘eral al cotton to de fro coeflnt rar eo) : change side force coefficient due to sideslip side force coefficient due to aileron deflection. side force coefficient due to rudder defiection .. side force coefficient due to local angle of attack change. distance from aircraft centerline to tip parachute force distance from aircraft cg to vertical tail aerodynamic center . indirect measurement of tip parachute force. fps or mph or knots .fps or mph or knots fps or mph or knots maximum flap extended speed. maximum landing gear extened speed maximum speed for stability characteristics side force contribution of the vertical tal (fin) . distance from aircraft centerline to wing center of pressu distance from aircraft x body axis to vertical tail center of pressure... increment of yawing moment coefficient.. increment of directional (or weathercock) stability increment of yawing moment... increment of side force due to rudder deflection. change in aileron deftection.UST OF SYMBOLS (Continued) rents ‘Symbol Definition Units CHAPTER 9 (only those symbols not previously introduced are listed) A state matrix. various B control matrix. various A state matrix (not normalized). various B control matrix (not normalized) various CG normalizing matrix (inverse is used to normalize A; and B))... various a general coefficients for second order differential equation Cy trim lift coefficient. x, Nnondimensional stability derivative. ox, nondimensiona! stability derivative xq -—_‘nondimensional control derivative. Cz, nondimensional stability derivative. Cz, nondimensional stability derivative. C2, nondimensional stability derivative Cz, _nondimensional control derivative . Cony nondimensional stability derivative Cine nondimensional stability derivative Cing nondimensional stability derivative Corse, nondimensional control derivative .. c damping coefficient ... D derivative operator with respect tot 7 Ay mass moment of inertia about x stability axis .. ‘slug-ft2 I mass product of inertia about x and z stability axes. . ‘slug-ft2 yy mass moment of inertia about y stability axis L mass moment of inertia about z stability axis. in nondimensional mass moment of inertia about x stability axis...none ig nondimensional mass moment of inertia about y stability axis ...none ig Nondimensional mass moment of inertia about z stability axis ... none. je Nnondimensional product of inertia about x and z stability axes... none Kor Dutch roll mode residue .. -. Various K, roll mode residue Ks spiral mode residue. k ‘spring constant .. 4p dimensional stability derivative 4 dimensionat stability derivative. 4 dimensional stability derivative. Ly dimensional stability derivative. 4 dimensional stability derivative dimensional control derivative.LIST OF SYMBOLS (Continued) Index ‘Symbol Definition Units: CHAPTER 9 (continued) 4, é My Mg My My My dimensional control derivative. characteristic length ocal maxima in response variable oscillation . dimensional stability derivative. dimensional stability derivative. dimensional stability derivative.. dimensional stability derivative.. dimensional control derivative local minima in response variable oscillation. dimensional stability derivative.. dimensional stability derivative . dimensional stability derivative.. dimensional stability derivative.. dimensional stability derivative.. dimensional control derivative dimensional control derivative transfer function matrix (input y, output x) angular velocity about the x stability axis in trim condition. perturbation in angular velocity about the x stability axis .....rad/sec initial peak value of roll rate response *Isec or rad/sec nondimensional angular velocity about the x stability axis angular acceleration about the x stability axis angular velocity about the y stability axis in trim condition perturbation in angular velocity about the y stability axis perturbation in angular velocity about the y stability axis angular velocity about the x stabilty axis in trim condition... perturbation in angular velocity about the x stability axis perturbation in angular velocity about the x stability axis time-to-hatf-amplitude ..UST OF SYMBOLS (Continued) Units none (airsecs) x axis velocity component of aircraft og at trim conditions ... linearized perturbation in x axis velocity of aircraft og. control vector. linearized perturbation in y axis velocity of aircraft og axis velocity component of aircraft og at trim conditions linearized perturbation in z axis velocity of aircraft og. dimensional stability derivative......... dimensional stability derivative dimensional stability derivative. dimensional stability derivative dimensional control derivative...... state vector. dimensional stably derivative. dimensional stability derivative dimensional stability derivative.. dimensional stability derivative dimensional stability derivative. dimensional control derivative . dimensional control derivative. dimensional stability derivative dimensional stability derivative dimensional stability derivative dimensional stability derivative. dimensional control derivative. increment in response variable: increment in time. Perturbation in angle of attack. perturbation in sidestip angle.. characteristic matrix aileron deflection at the trim condition elevator deflection at the trim condition... udder deflection at the trim condition ... perturbation in bank angle or roll angle eigenvalue spiral mode eigenvalve.......... inertial pitch angle at trim conditions perturbation in inertial pitch angle.. various . variousLIST OF SYMBOLS (Continued) indent? ‘Symbol Definition Units CHAPTER 9 (continued) t ‘or v s og ©n no RRR Tt roll mode time constant... spiral mode time constant .. undamped natural frequency. Dutch roll undamped natural frequency...... phugoid undamped natural frequency .. perturbation in heading angle. (only those symbols not previously introduced are listed) distance (fraction of semichord) from midchord to elastic axis...none fps ‘serodynamic damping matrix change in lift per control surface deflection .. change in moment per contro! surface deflection ....... per ° or radian slope of lift curve for incompressible flow... distance from elastic axis to aerodynamic center .. extemal force vector .. motion dependent aerodynamic force vector.edt UST OF SYMBOLS (Continued) ‘Symbol Definition Units CHAPTER 10 (Continued) aerodynamic stiffness matrix with q., removed aerodynamic moment... Mach number. mass matrix. generalized coordinate vector reference dynamic pressure for a. at altitude pitch deflection, or angle of attack.. eigenvector. damping ratio, or control surface deflection. natural frequency in plunge - uncoupled. natural frequency in pitch - uncoupled. K& M Me M q qe r fa Vv Yo uM y a B 6 © A “ ® ® Oy © (only those symbols not previously introduced are listed) Potential equilibrium pitching moment coefficient from rotary balance data..... drag force due to advancing wing panel drag force due to retreating wing panel . force vector .. mass product of inertia about x and y stability axes. mass product of inertia about y and z stability axes. radius of gyration about x stability axis. radius of gyration about y stability axis radius of gyration about z stability axis lift force due to advancing wing panel .. lift force due to retreating wing panel.‘Symbol UST OF SYMBOLS (Continued) Indoxt9 Units: Fagen Hai LSERLSS eS PNLSERODEODOESSESDenEDOTSC EEE CHAPTER 11 (continued) R Ra Cp, ACp2 AC, ACL, resultant aerodynamic force fedvancing wing panel contribution to resultant aerodynamic force retreating wing panel contribution to resultant aerodynamic force snes y component of resultant aerodynamic force z component of resultant aerodynamic force... Ibs helix radius in a fuly developed spin velocity vector for the aircraft og ... aerodynamic force in x direction on advancing wing panel ‘aerodynamic force in x direction on retreating wing panel ‘aerodynamic force in z direction on advancing wing panel ‘aerodynamic force in z direction on retreating wing panel increment in drag coefficient between aavancing and retreating wing at low angle of attack increment in drag coefficient between advancing and retreating wing at high angle of attack.... increment in it coefficient between advancing and retreating wing at low angle of attack. . increment in lft coefficient between advancing and retreating wing at high angle of attack. focal angle of attack of an advancing wing section local angle of attack of a retreating wing section angle of attack of the advancing wing at low angle of attack angle of attack of the advancing wing at high angle of attack.. angle of attack of the retreating wing at low angle of attack angle of attack ofthe retreating wing at high angle o tack stall angle of attack helix angle in a spin .