Umi Ku 2513 - 1
Umi Ku 2513 - 1
Umi Ku 2513 - 1
Submitted to the graduate degree program in Aerospace Engineering and the Graduate Faculty of the University of Kansas in partial fulfillment of the requirements of the degree of Masters of Science.
________________________ David Downing, Chairperson Committee Members ________________________ Richard Colgren ________________________ Richard Hale
The Thesis Committee of Rylan Jager certifies that this is the approved version of the following thesis:
Test and Evaluation of the Piccolo II Autopilot System on a One-Third Scale Yak-54
________________________ David Downing, Chairperson Committee Members ________________________ Richard Colgren ________________________ Richard Hale
Abstract
To gain a better understanding of the dynamics of the great ice sheets the National Science Foundation established the Center for Remote Sensing of Ice Sheets (CReSIS) to develop technologies that would improve data gathering of said ice sheets. CReSIS was tasked with the development of an unmanned aerial vehicle, named the Meridian, which would have the ability to make use of advanced radar systems that could be used to gather data on the ice sheets of remote Polar Regions. CReSIS decided to use commercial-off-the-shelf autopilot systems on the Meridian, selecting the Cloud Cap Technologies Piccolo II UAV autopilot system as the initial system to be tested and evaluated. A process for test and evaluation of modeling, simulation and control systems is presented. Three dynamic models for a one-third scale Yak-54 are developed. A deliberate and methodical flight test program is developed to evaluate the Piccolo II flight control system. Parameter identification flight tests are performed to evaluate the three modeling and simulation techniques. Closed loop flight testing is performed to evaluate the flight control systems ability to control an aircraft and the ability of the gains to be performance optimized. Finally flaws are found in the communication system architecture of the Piccolo II autopilot system which causes the system to go pilot-in-loop unstable and to be rejected from further consideration by the CReSIS team.
ii
Acknowledgements
First I would like to thank Dr. Shah Keshmiri for giving me the inspiration to stick through all the rough times. Your support cannot be overstated. Thanks to Edmond Leong for being my partner through this project and teaching me a little bit about Chinese culture along the way. Thanks to Dr. Dave Downing for giving some sage advice to a sometimes rash young engineer. Thanks to Dr. Richard Hale for believing we could pull off such a crazy feat like building our own airplane. Thanks to Dr. Richard Colgren for lending your years of experience in industry to young minds like I. Thanks to Dr. Ron Barrett for getting me to come out to a wonderful school like the University of Kansas. Thanks to Lance Holly for helping us on all our questions about flying model airplanes. Thanks to Andy Pritchard for putting up with a bunch of know-it-all engineers, showing us how to use a wrench and brightening all of our days out at the airport. Finally thanks to all my friends in the aerospace engineering community, this job is too hard to do without your camaraderie to make it all worth it.
iii
Table of Contents
Abstract ......................................................................................................................... ii Acknowledgements...................................................................................................... iii Table of Contents......................................................................................................... iv List of Figures ............................................................................................................. vii List of Tables ................................................................................................................ x List of Symbols ........................................................................................................... xii 1 Introduction........................................................................................................... 1 2 Modeling, Simulation and Autopilot Test and Evaluation Process ...................... 7 2.1 Step 1: Select System.................................................................................... 7 2.2 Step 2: Select Flight Test Platform ............................................................... 8 2.3 Step 3: Develop First Principles Aircraft Dynamic Models ......................... 9 2.4 Step 4: Integrate Dynamic Model into Simulation Platform ..................... 11 2.5 Step 5: Refine Dynamic Models using Flight Test Data ............................ 11 2.6 Step 6: Tune Flight Control System in Simulation..................................... 13 2.7 Step 7: Flight Test Tuned Control System.................................................. 13 2.8 Step 8: Evaluate the Performance of the Control System and Reject or Certify 14 3 The Cloud Cap Piccolo II Autopilot System ....................................................... 15 3.1 Piccolo II Avionics Hardware..................................................................... 16 3.1.1 Gyros and Accelerometers.................................................................. 17 3.1.2 GPS ..................................................................................................... 18 3.1.3 Pressure System .................................................................................. 18 3.1.4 Kalman Filter ...................................................................................... 18 3.2 Communication System .............................................................................. 18 3.3 Ground Station ............................................................................................ 19 3.3.1 Main Ground Station Electronics........................................................ 19 3.3.2 Operator Interface PC ......................................................................... 20 3.3.3 Futaba Console.................................................................................... 21 3.4 Operator Interface ....................................................................................... 22 3.5 Piccolo II Flight Control Laws ................................................................... 24 3.6 Piccolo II Guidance and Navigation System .............................................. 26 3.7 Piccolo II Modeling and Simulation Environment ..................................... 26 3.7.1 Standard Cloud Cap Simulator ........................................................... 27 3.7.2 Athena Vortex Lattice Based Simulator ............................................. 29 3.8 Hardware-in-Loop Simulation .................................................................... 30 4 Flight Test Platform............................................................................................ 33 4.1 Yak-54 Geometry........................................................................................ 34 4.2 Yak-54 Mass Properties.............................................................................. 35 4.3 Yak-54 Engine and Propeller...................................................................... 35 4.3.1 Engine Calibration .............................................................................. 36 4.4 Yak-54 Servos............................................................................................. 36
iv
7 8
4.5 Piccolo II Installation.................................................................................. 37 4.6 Antennas ..................................................................................................... 37 4.7 Batteries ...................................................................................................... 37 4.8 Engine Kill Switch...................................................................................... 38 Yak-54 Dynamic Model Development ................................................................ 39 5.1 Advance Aircraft Analysis Based Model ................................................... 39 5.1.1 Lateral State Space Model and Analysis ............................................ 40 5.1.2 Longitudinal State Space Model and Analysis ................................... 41 5.2 Athena Vortex Lattice Based Model........................................................... 42 5.2.1 AVL Generated Stability Derivatives: Longitudinal .......................... 43 5.2.2 AVL Generated Stability Derivatives: Lateral-Directional ................ 50 5.3 AVL-Piccolo Hardware-in-Loop Simulation and Dynamic Analysis ........ 55 5.3.1 Modified Transient Peak Ratio Method.............................................. 55 5.3.2 Maximum Slope Method .................................................................... 57 5.3.3 Open Loop Modal Analysis: Phugoid Mode ...................................... 58 5.3.4 Open Loop Modal Analysis: Short Period Mode ............................... 59 5.3.5 Open Loop Modal Analysis: Dutch Roll Mode.................................. 60 5.4 Standard Cloud Cap Aircraft Modeling and Simulation............................. 61 5.5 Simulation Model Dynamics Comparison.................................................. 62 Flight Test Planning and Preparation................................................................ 64 6.1 Preflight Checkout ...................................................................................... 64 6.2 Flight Test Team ......................................................................................... 65 6.2.1 Flight Test Engineer (FTE)/Command and Control Engineer............ 65 6.2.2 Flight Test Pilot in Command (PIC)................................................... 66 6.2.3 Vehicle Engineer (VE)........................................................................ 66 6.2.4 Safety Officer (SO) ............................................................................. 66 6.2.5 Data Processing Engineer (DPE) ........................................................ 67 6.2.6 Pilot Assistant (PA)............................................................................. 67 6.2.7 Video Operator (VO) .......................................................................... 67 6.2.8 Flight Test Organizational and Communications Flow Charts........... 68 6.3 Emergency Procedures................................................................................ 69 6.4 Flight Test Area .......................................................................................... 70 Manual Control Flight Test ................................................................................ 71 7.1 Manual Control Tests Results..................................................................... 72 Parameter Identification Flight Testing ............................................................. 75 8.1 Dutch Roll Mode Flight Test ...................................................................... 75 8.1.1 Dutch Roll Mode Flight Test Results ................................................. 76 8.2 Short Period Mode Flight Test.................................................................... 78 8.2.1 Short Period Mode Flight Test Results............................................... 78 8.2.2 Short Period Data Processing Using the Time-Ratio Method ............ 80 8.3 Phugoid Mode Flight Test .......................................................................... 82 8.4 Flight Test vs. Simulation Model Dynamics .............................................. 83 8.5 Open-Loop Flight Test Conclusions and Recommendations ..................... 87 Closed-Loop Piccolo II Flight Control System Tuning Using Simulation ......... 89
Closed-Loop Gain Tuning Using AVL Hardware-in-Loop Simulation ..... 89 Lateral-Directional Response with Default Gains ...................................... 91 Longitudinal Command Performance with Default Gains ......................... 95 Lateral Gain Tuning Using AVL-Piccolo HIL Simulation......................... 97 Longitudinal Gain Tuning Using AVL-Piccolo HIL Simulation ............. 101 Closed Loop Gain Margins....................................................................... 105 Closed Loop Phase Margins ..................................................................... 106 10 Closed Loop Piccolo II Gain Tuning in Flight Test ..................................... 108 10.1 Bank Angle Control Flight Test................................................................ 108 10.2 Heading Angle Control Flight Test........................................................... 112 10.3 Tuned Gains for the Lateral-Directional Control System......................... 115 10.4 Airspeed Control Flight Test .................................................................... 115 10.5 Piccolo II Closed Loop Flight Test Conclusions ...................................... 119 11 Yak-54 Flight Incident and Investigation ..................................................... 121 11.1 First Flight................................................................................................. 121 11.2 Second Flight ............................................................................................ 122 11.3 Third Flight ............................................................................................... 123 11.4 Accident Investigation .............................................................................. 126 11.5 Communication System Laboratory Test ................................................. 130 11.6 Effect of Discrete Drop of Pilot Commands on Handling Qualities ........ 137 11.7 Incident Investigation Conclusions........................................................... 139 12 Conclusions, Recommendations and Future Work....................................... 140 12.1 Piccolo II Flight Control System Test and Evaluation Conclusions......... 140 12.2 Piccolo II Flight Control System Test and Evaluation Recommendations141 12.3 Future Work .............................................................................................. 142 13 References ..................................................................................................... 143 Appendix A: SCCS Yak-54 Model (ASCII Format)................................................ 147 Appendix B: SCCS Wing File (ASCII Format) ....................................................... 153 Appendix C: SCCS Horizontal Tail File (ASCII Format)........................................ 154 Appendix D: SCCS Vertical Tail File (ASCII Format)............................................ 155 Appendix E: SCCS Engine File (ASCII Format) ..................................................... 156 Appendix F: SCCS Propeller File (ASCII Format) .................................................. 157 Appendix G: AVL Input Geometry File (ASCII Format) ........................................ 161 Appendix H: AVL Simulator File (ASCII Format).................................................. 166 Appendix I: AVL Modal Data Reduction................................................................. 169 Appendix J: Yak-54/Piccolo Ground Testing Procedures ........................................ 170 Appendix K: Yak-54/Piccolo Pre-Flight Checklist .................................................. 172 Appendix L: Handling Qualities Flight Test Dance Card ........................................ 175 Appendix M: Open Loop Dynamics Flight Test Dance Card .................................. 176 Appendix N: Dutch Roll Flight Test Data Reduction............................................... 178 Appendix O: Bank Angle Control Flight Test Dance Card...................................... 179 Appendix P: Heading Angle Control Flight Test Dance Card ................................. 181 Appendix Q: Airspeed Control Flight Test Dance Card........................................... 183 Appendix R: Longitudinal Gain Change Log Sheet ................................................. 185
vi
List of Figures
Figure 1: Meridian Three View .................................................................................... 6 Figure 2: Modeling, Simulation and Autopilot Test and Evaluation Process............... 8 Figure 3: Aircraft Dynamic Model Development Flow Chart...................................... 9 Figure 4: Flight Control Tuning Flow Chart............................................................... 10 Figure 5: Piccolo II System Block Diagram ............................................................... 16 Figure 6: Cloud Cap Piccolo II Autopilot [26] ........................................................... 16 Figure 7: Piccolo Hardware Block Diagram [26] ....................................................... 17 Figure 8: Ground Station Unit Front Panel [26] ......................................................... 20 Figure 9: Ground Station Unit Back Panel [26].......................................................... 20 Figure 10: Itronix GOBook II [27] ............................................................................. 21 Figure 11: Futaba Pilot Console [26].......................................................................... 22 Figure 12: Operator Interface Telemetry Page [26].................................................... 23 Figure 13: Piccolo Navigation Screen [26]................................................................. 24 Figure 14: Aerosonde AVL Model [30] ..................................................................... 30 Figure 15: Hardware-in-Loop Simulation [29]........................................................... 31 Figure 16: Flight Gear Visualization [29]................................................................... 32 Figure 17: Aero-Works 75cc Yak-54 [31].................................................................. 33 Figure 18: 3W-80 XI CS Engine [34]......................................................................... 35 Figure 19: Normalized Thrust and RPM Curves for Engine Calibration ................... 36 Figure 20: Engine Kill Switch Schematic [35] ........................................................... 38 Figure 21: Yak-54 AVL Model .................................................................................. 43 Figure 22: Modified Transient Peak Ratio [14].......................................................... 55 Figure 23: Transient Peak Ratio vs. Damping Ratio [14]........................................... 56 Figure 24: Maximum Slope Method [14] ................................................................... 57 Figure 25: Maximum Slope Method for Determining and n [14].......................... 58 Figure 26: AVL-Piccolo HIL Phugoid Simulation Response..................................... 59 Figure 27: AVL-Piccolo HIL Short Period Simulation Response.............................. 60 Figure 28: AVL-Piccolo HIL Dutch Roll Mode Simulation Response...................... 61 Figure 29: Flight Test Organizational Chart............................................................... 68 Figure 30: Flight Test Communication Flow Chart.................................................... 69 Figure 31: Foley Airfield and Flight Test Area .......................................................... 70 Figure 32: Race Track Pattern .................................................................................... 72 Figure 33: Pilot-in-loop Pitch Instability .................................................................... 73 Figure 34: Stabilized Pilot-in-Loop Pitch Control...................................................... 74 Figure 35: Flight Test Dutch Roll Excitation #1 [33]................................................. 76 Figure 36: Flight Test Dutch Roll Excitation #2 [33]................................................. 77 Figure 37: Flight Test Dutch Roll Excitation #3 [33]................................................. 77 Figure 38: Flight Test Short Period Excitation #1 [33] .............................................. 79 Figure 39: Flight Test Short Period Excitation #2 [33] .............................................. 79 Figure 40: Flight Test Short Period Excitation #3 [33] .............................................. 80 Figure 41: Time-Ratio Method [14] ........................................................................... 80
vii
Figure 42: Time-Ratio Method Data Reduction Plot [14] .......................................... 81 Figure 43: Flight Test Excitation of the Phugoid Mode #1 [33]................................. 82 Figure 44: Flight Test Excitation of the Phugoid Mode #2 [33]................................. 83 Figure 45: AAA Yak-54 Phugoid Simulation ........................................................... 85 Figure 46: AVL-HIL Yak-54 Phugoid Simulation..................................................... 86 Figure 47: AVL-HIL Bank Angle Response with Default Gains............................... 93 Figure 48: AVL-HIL Heading Angle Response with Default Gains:......................... 94 Figure 49: AVL-HIL Airspeed Response with Default Gains.................................... 95 Figure 50: AVL-HIL Altitude Response with Default Gains..................................... 96 Figure 51: AVL-HIL Bank Angle Control Response with Simulator Tuned Gains... 99 Figure 52: AVL-HIL Heading Angle Control Loop Performance with Simulator Tuned Gains .............................................................................................................. 100 Figure 53: AVL-HIL Airspeed Control Response with Simulator Tuned Gains ..... 103 Figure 54: AVL-HIL Altitude Control Response with Simulator Tuned Gains....... 104 Figure 55: Initial Bank Angle Control Flight Test with AVL-HIL Tuned Gains..... 110 Figure 56: 20 Degree Bank Angle Command Flight Test with Simulator Tuned Gains ................................................................................................................................... 110 Figure 57: Final Roll Control Flight Test with Tuned Gains.................................... 111 Figure 58: Initial Heading Command Flight Test with Simulator Tuned Gains ...... 113 Figure 59: Final Heading Angle Command Flight Test with Tuned Gains.............. 114 Figure 60: Initial Airspeed Regulator Flight Test with Simulator Tuned Gains ...... 116 Figure 61: Second Airspeed Command Flight Test Response with Simulator Tuned Gains ......................................................................................................................... 117 Figure 62: Final Airspeed Command Flight Test Response with Tuned Gains ....... 118 Figure 63: RSSI Levels during First Flight Pilot Control Loss ................................ 121 Figure 64: RSSI during First Flight Test .................................................................. 122 Figure 65: RSSI during Second Flight Test Pilot Control Loss................................ 123 Figure 66: RSSI during Second Flight Test.............................................................. 123 Figure 67: RSSI during Third Flight Pilot Control Loss .......................................... 124 Figure 68: RSSI during Crash................................................................................... 125 Figure 69: Engine Data during Crash ....................................................................... 125 Figure 70: RSSI and Time between Packets during Ground Testing ....................... 127 Figure 71: RSSI during Uncommanded Autopilot Activation ................................. 128 Figure 72: RSSI Signal Strength Levels and Percentage of Data Lost..................... 128 Figure 73: Data Loss Magnitudes ............................................................................. 129 Figure 74: Data Dropout Rates during All Flight Tests............................................ 130 Figure 75: RSSI Signal Strength and Percentage of Data Loss during Laboratory Test ................................................................................................................................... 131 Figure 76: RSSI and Packet Time at Autopilot Activation during Lab Test ............ 132 Figure 77: RSSI Signal Strength and Percentage of Data Loss during Autopilot ON Lab Test .................................................................................................................... 133 Figure 78: RSSI Levels and Data Dropouts during Various Lab Tests.................... 134 Figure 79: Autopilot Activation during Laboratory Test.......................................... 134 Figure 80: Dropout Rates for Laboratory Tests........................................................ 135
viii
Figure 81: Autopilot Activation during Flight and Lab Test Activities ................... 135 Figure 82: RSSI and Data Dropout Times during 900 MHz Lab Tests.................... 136 Figure 83: Poor Pilot-in-loop Pitch Stability Due to Time-Lag and Data Drop....... 137
ix
List of Tables
Table 1: Piccolo Gain Scaling Parameters.................................................................. 25 Table 2: 75cc Yak-54 Lifting Surfaces [33] ............................................................... 34 Table 3: Vehicle Mass Properties [33]........................................................................ 35 Table 4: Trim Conditions [33] .................................................................................... 39 Table 5: Yak-54 Stability Derivatives Generated Using AAA................................... 40 Table 6: AAA Steady State Coefficients .................................................................... 40 Table 7: AAA Yak-54 Lateral Modal Analysis [33] .................................................. 41 Table 8: AAA Yak-54 Longitudinal Modal Analysis [33]......................................... 42 Table 9: AAA vs. AVL Longitudinal Stability Derivatives ....................................... 43 Table 10: AVL Steady State Coefficients................................................................... 44 Table 11: AAA vs. AVL Longitudinal Gain Scaling Terms ...................................... 44 Table 12: AAA vs. AVL Lateral-Directional Derivatives.......................................... 51 Table 13: AAA vs. AVL Lateral-Directional Gain Scaling Terms ............................ 52 Table 14: AVL Generated Lateral-Directional Dimensional Stability Derivatives.... 53 Table 15: AVL Yak-54 Modal Analysis..................................................................... 54 Table 16: SCCS Modal Analysis [33] ........................................................................ 62 Table 17: Simulator Model Modal Comparison ......................................................... 62 Table 18: Dutch Roll Flight Test Results ................................................................... 78 Table 19: Short Period Mode Flight Test Results....................................................... 81 Table 20: Flight Test vs. Simulation Model Dynamics .............................................. 84 Table 21: Yak-54/Piccolo Gain Scaling Terms .......................................................... 90 Table 22: Piccolo Default Lateral-Directional Gains ................................................. 91 Table 23: Piccolo Default Longitudinal Gains ........................................................... 91 Table 24: Bank Angle Control Loop Performance with Default Gains...................... 93 Table 25: Heading Angle Control Loop Performance with Default Gains ................ 94 Table 26: Airspeed Control Loop Performance with Default Gains .......................... 95 Table 27: Altitude Control Loop Performance with Default Gains............................ 96 Table 28: AVL-HIL Tuned Lateral-Directional Gains............................................... 98 Table 29: AVL-HIL Bank Angle Control Loop Performance with Simulator Tuned Gains ........................................................................................................................... 98 Table 30: Heading Angle Control Loop Performance with AVL-HIL Tuned Gains100 Table 31: AVL-HIL Tuned Longitudinal Gains....................................................... 101 Table 32: AVL-HIL Airspeed Control Performance with Simulator Tuned Gains.. 103 Table 33: AVL-HIL Altitude Control Performance with Simulator Tuned Gains... 104 Table 34: Piccolo II AVL-HIL Lateral-Directional Gain Margins........................... 106 Table 35: Piccolo II AVL-HIL Longitudinal Gain Margins..................................... 107 Table 36: Roll Control Loop Performance with Simulator Tuned Gains................. 111 Table 37: Final Roll Control Performance................................................................ 112 Table 38: Final Heading Angle Control Performance with Tuned Gains ................ 114 Table 39: Final Lateral-Directional Gains ................................................................ 115
Table 40: Second Airspeed Command Flight Test Performance with Simulator Tuned Gains .............................................................................................................. 117 Table 41: Final Airspeed Command Response Performance with Tuned Gains...... 118 Table 42: Final Longitudinal Gains .......................................................................... 119
xi
List of Symbols
Symbol
a
Definition Acceleration Aspect ratio Span Chord Mean geometric chord Airplane drag coefficient Airplane trim drag coefficient Variation of airplane rolling moment coefficient with angle of sideslip
Units ft/sec2 ~ ft ft ft ~ ~ ~
AR
b
c
c
CD
C D1
Cl
Cl a
1/rad
Cl
1/rad
Variation of airplane rolling moment coefficient with dimensionless rate of change of roll rate
1/rad
Clr
Variation of airplane rolling moment coefficient with dimensionless rate of change of yaw rate
1/rad
CL
Lift coefficient
xii
Symbols (cont.)
C L
1/rad
C L&
Variation of airplane lift coefficient with dimensionless rate of change of angle of attack
1/rad
C Le
1/rad
C Lq
1/rad
C Lu
Cm Cm1 Cm
Pitching moment coefficient Airplane trim pitching moment coefficient Variation of airplane pitching moment coefficient with angle of attack
~ ~ 1/rad
Cm&
Variation of airplane pitching moment coefficient with dimensionless rate of change of angle of attack
1/rad
Cme
1/rad
CmT
CmT
1
~ ~
xiii
Symbols (cont.)
C mq
1/rad
Cmu
CmT
Variation of airplane pitching moment coefficient due to thrust with angle of attack
1/rad
CmTu
Variation of airplane pitching moment coefficient due to thrust with dimensionless speed
1/rad
Cn
1/rad
C n a
1/rad
C n
1/rad
Variation of airplane yawing moment coefficient with dimensionless rate of change of roll rate
1/rad
Cn r
Variation of airplane yawing moment coefficient with dimensionless rate of change of yaw rate
1/rad
CnT
Variation of airplane yawing moment coefficient due to thrust with sideslip angle
1/rad
CP
Power coefficient
xiv
Symbols (cont.)
CT CT x1 CT xu
Thrust coefficient Trim thrust coefficient in the X-axis direction Variation of airplane thrust coefficient in the X-axis direction with dimensionless speed
~ ~ ~
CT
Variation of airplane thrust coefficient in the X-axis direction with angle of attack
1/rad
Cy
Cy
Airplane side force coefficient Variation of airplane side force coefficient with sideslip angle
~ 1/rad
C y a
1/rad
C y
1/rad
rudder angle
Cy p
Variation of airplane side force coefficient with dimensionless rate of change of roll rate
1/rad
Cyr
Variation of airplane side force coefficient with dimensionless rate of change of yaw rate
1/rad
ft lbs varies
D
e
xv
Symbols (cont.)
h
Altitude Acceleration due to gravity Airplane moments of inertia about XYZ Propeller advance ratio Gain Lift Roll angular acceleration per unit sideslip angle Roll angular acceleration per unit roll rate Roll angular acceleration per unit yaw rate Roll angular acceleration per unit aileron angle Roll angular acceleration per unit rudder angle Pitch angular acceleration per unit angle of attack Pitch angular acceleration per unit angle of attack due to thrust
ft ft/sec2 slug ft2 ~ varies lbs rad/sec2/rad 1/sec 1/sec rad/sec2/rad rad/sec2/rad 1/sec2 1/sec2
g I xx , I yy , I zz
J
K L L Lp
Lr
L a L r M M T
Mu M Tu
Pitch angular acceleration per unit change in speed Pitch angular acceleration per unit change in speed due to thrust
rad/sec/ft rad/sec/ft
M &
1/sec
xvi
Symbols (cont.)
n n also:
Propeller RPM Load factor Yaw angular acceleration per unit sideslip angle Yaw angular acceleration per unit sideslip angle due to thrust
N
NT
Np
Nr
Yaw angular acceleration per unit roll rate Yaw angular acceleration per unit yaw rate Yaw angular acceleration per unit aileron deflection Yaw angular acceleration per unit rudder angle Perturbed values of P, Q and R
N a N r
p, q, r
P, Q, R P also: q
s S
Airplane angular velocity components about XYZ Power Dynamic pressure Laplace domain variable Area Thickness Time Rise time Settling time
t t also:
tr
ts
xvii
Symbols (cont.)
T
Tlag Tlead Tn u , v, w U,V,W Va Vg
W
Thrust Pilot lag time constant Pilot lead time constant Pilot neuromuscular time constant Perturbed values of U, V, and W
Components of Va along XYZ True airspeed Ground speed Weight Airplane aerodynamic center location along the X axis
xac
xcg X Xu X Tu
Airplane center of gravity location along the X axis Forward acceleration per unit angle of attack Forward acceleration per unit change in speed Forward acceleration per unit change in speed due to thrust
X e Y Yp
Forward acceleration per unit elevator angle Lateral acceleration per unit sideslip angle Lateral acceleration per unit roll rate
xviii
Symbols (cont.)
Yp also:
Yr
Human pilot transfer function Lateral acceleration per unit yaw rate Lateral acceleration per unit aileron angle Lateral acceleration per unit rudder angle Vertical acceleration per unit angle of attack Vertical acceleration per unit change in speed Vertical acceleration per unit rate change of angle of attack
Y a Y r Z Zu Z &
Zq Za
Vertical acceleration per unit pitch rate Vertical acceleration per unit elevator angle
ft/sec/rad ft/sec2/rad
Greek
&
Angle of attack Rate of change of angle of attack Trim angle of attack Angle of sideslip Flight path angle Ground track heading Control surface deflection angle
deg or rad rad/sec deg or rad deg or rad deg deg or rad deg or rad
xix
Greek (cont.)
Increment of a parameter Perturbed value of Airplane pitch attitude angle Sweep angle Damping ratio Air density Pilot reaction time delay constant Perturbed value of Airplane bank angle Perturbed value of Airplane heading angle Natural frequency Damped frequency
~ rad rad ~ slugs/ft2 sec rad rad rad rad rad/sec rad/sec
n d
Subscripts
xx
Subscripts (cont.)
Critical Relative to the quarter chord Damping Dutch roll Elevator Horizontal tail Integrator Maximum Minimum Peak Phugoid Predicted Rudder Required Spiral Steady State Throttle Vertical tail Wing Wing + fuselage In the x, y or z direction
xxi
Acronyms
ac AGL ASL AVL c.g. CReSIS DPE FTE GPS HIL IAS IPCC PA PIC SCCS SFC SO UAS UAV VO
Aerodynamic center Above ground level Above sea level Athena Vortex Lattice Center of gravity Center for Remote Sensing of Ice Sheets Data Processing Engineer Flight Test Engineer Global positioning system Hardware-in-loop Indicated airspeed Intergovernmental Panel on Climate Change Pilot Assistant Pilot in Command Standard Cloud Cap Simulator Specific fuel consumption Safety Officer Unmanned aerial system Unmanned aerial vehicle Video Operator
xxii
1 Introduction
In November of 2007 the Intergovernmental Panel on Climate Change (IPCC) released a summary of their findings with regard to climate change [1]. In this summary the IPCC states that the average sea level change has risen to a rate of 3.1 mm/yr (the average from 1961 is 1.8 mm/yr). The report also stated that the average sea ice extent in the Arctic has shrunk by 2.7% per decade and a correlation is clearly drawn between these two events. The IPCC predicted that the Earths sea level will rise 18 to 59 centimeters due to climate change but does not include in these predictions changes in the mass balance of the great ice sheets, only the thermal expansion of the oceans. Currently glacial data sets do not exist to such an extent that the dynamics of the glaciers under rapidly warming conditions can be understood with any high level of certainty. To combat this lack of knowledge the National Science Foundation, in 2005, established the Center for Remote Sensing of Ice Sheets (CReSIS). CReSIS is a multi-university organization headquartered at the University of Kansas (KU). The main goal of CReSIS is to conduct and foster multi-disciplinary research that will result in technology, new datasets, and models necessary to achieve a better understanding of the mass balance of the polar ice sheets (e.g., Greenland and Antarctica) and their contributions to sea level rise. For some years the Electrical Engineering Department at KU has been developing radar systems that allow data on the great ice sheets to be collected. The radar gives scientists the ability to measure the depth, surface elevation, and
accumulation rate of the great ice sheets and determine their basal conditions. Knowing the basal conditions of the glaciers allows modelers to better predict their dynamic behavior. The radar has been carried by both ground based and manned air based vehicles and part of the initial goals of the center was to develop an autonomous aerial platform on which to carry the radar sensors. The Aerospace Engineering Department at the University of Kansas was given the primary task of developing an unmanned aerial vehicle (UAV) to carry the radar system. The final design, named the Meridian, is a low wing, V-tail, 1,000 lb, diesel driven, tail dragger airplane that will operate autonomously during its data collecting missions (Figure 1). The aircraft has been designed to carry up to eight radar antennas along the wing that will be used to collect the data on the glaciers. A detailed discussion of the design of the Meridian can be found in Reference [2] and a detailed discussion of the Meridians avionics systems can be found in Reference [3]. For the Meridian to operate autonomously during its missions an autopilot must be selected that can safely and effectively control the aircraft within the mission parameters. The Meridian development team at CReSIS chose to use a commercial off-the-shelf (COTS) autopilot system rather than developing a system in-house because of a short time to flight schedule. The initial autopilot of choice was the Cloud Cap Technologies Piccolo II flight control system. To determine if the Piccolo could safely control the aircraft and successfully perform the science mission a detailed development, test and evaluation process was needed. The Piccolo requires user-end modeling and simulation of the aircraft for user-end gain tuning. These
modeling and simulation packages were included with the Piccolo system and their accuracy is unknown and their validity must be determined. A one-third scale Yak54 was chosen as the platform to develop and verify modeling and simulation tools, test the systems ability to perform the desired mission and also establish a flight test program structure that could be easily transitioned to Meridian or other UAV flight testing. The research presented in this thesis was a team effort and proper recognition should be performed before proceeding. Hou In Edmond Leong and Shariar
Keshmiri were the primary collaborators to the work presented herein. The direct contributions of each team with regards to this thesis were be broken down as follows.
1. Rylan Jager
Developed flight test plans for all flight test activities Lead flight test engineer for all flight test activities Setup the Piccolo system in the Yak-54 Developed Standard Cloud Cap Simulation models Developed Athena Vortex Lattice models Analyzed piloted responses Performed some parameter identification data processing Analyzed parameter identification flight test data Analyzed simulation models Performed gain tuning activities in simulation and flight test
Analyzed closed-loop performance Developed communications lab tests and performed tests Analyzed processed communications data
2. Edmond Leong
Estimated Yak-54 moments of inertia Developed telemetry plotting software Performed open-loop analysis of the Standard Cloud Cap Simulator Assisted in the construction of the Yak-54 Advanced Aircraft Analysis dynamic model
Developed Bolly 26 x 10 propeller model Performed some parameter identification data processing Plotted open-loop response flight test data Developed communications data loss analysis techniques Processed and plotted communications data
3. Shahriar Keshmirir
Chapter 2 of this document presents the high level test and evaluation process that was developed for modeling, simulation and control system development, test and evaluation. Chapter 3 details the Piccolo II UAS autoflight system that was tested and evaluated. Chapter 4 presents the Yak-54 flight test platform for the Piccolo II test and evaluation program. Chapter 5 presents modeling and simulation platforms
that were used with the Piccolo II system. Chapter 6 details the flight test planning and preparation process. Chapter 7 presents the manual control flight test of the Yak54/Piccolo system. Chapter 8 details the parameter identification flight testing and the evaluation of the various modeling and simulation platforms using the data gathered in those flight tests. Chapter 9 presents the gain tuning of the Piccolo II flight control system using hardware-in-loop simulation. Chapter 10 details the closed-loop Piccolo II flight control system test and evaluation. Chapter 11 presents a flight incident caused by a poor communication system setup in the Piccolo II system that was revealed during the test and evaluation process. Finally Chapter 12 presents the conclusions and recommendations of the Piccolo II test and evaluation process.
Develop Aircraft Dynamic Models Integrate Dynamic Models into Simulation Packages
Perform Open Loop Flight Tests Vehicle Modeling and Simulation Acceptable Tune Flight Control System Perform Closed Loop Flight Tests
Yes
No
Certify System
Reject System
payload capacity for mounting autopilot avionics and the structural integrity of the platform to handle the increased weight of the avionics.
Develop Parametric Aircraft Model (Ex: AAA, DATCOM) Increase Aircraft Model Fidelity (Ex: Wind Tunnel, CFD) Refine Aircraft Model Integrate Aircraft Model into Simulation Platform Design Parameter or System Identification Flight Tests (Ex: Time History Analysis, Time-Domain Sys ID, Frequency Domain Sys ID) Compare Flight Test Data to Simulation (Ex: Time History Comparison, Stability Derivative Comparison, Frequency and Damping Comparison)
No
Does Model Match Flight Data Within Requirements? Yes Final Aircraft Model
Figure 3: Aircraft Dynamic Model Development Flow Chart
simplest method for developing an initial first principles parametric aircraft dynamic model is to use geometric information. These parameters are used to approximate the aircraft stability derivatives using past aircraft information databases, such as the US Air Forces digital DATCOM or DARcorporations Advanced Aircraft Analysis. References [5], [6] and [7] used geometric techniques to estimate UAV stability derivatives. These techniques tend to be low fidelity.
Integrate Flight Control System into Modeling and Simulation Platform (Software-in-Loop or Hardware-in-Loop) Tune Gains in Simulation to Meet Performance Specifications (Ex: Time Response, Frequency Response, Gain and Phase Margin, Actuator RMS) Retune Gains No Flight Test Inner Loops
Retune Gains
No
10
Wind tunnel data can also be used to determine a UAVs stability derivatives, as presented in Reference [8]. Wind tunnel testing is more expensive and labor intensive than geometric techniques but will produce a more accurate model. Computational fluid dynamics can also be used to develop aerodynamic parameters.
2.4
models. 3 DoF linear and 6 DoF nonlinear models can be simulated software-in-loop in platforms like MATLAB/Simulink. 3 DoF techniques are presented thoroughly in References [9], [10] and [11]. The more advanced 6 DoF techniques a used in References [12] and [13]. Reference [6] provides a nice comparison between the various simulation techniques. Hardware-in-loop (HIL) simulation is the ideal as it is the closest to reality. HIL simulation is often platform specific, as with the Piccolo II autopilot system, or is performed using real-time C code generated from MATLAB/Simulink using RealTime Workshop. This is because HIL simulation has to be performed real-time and MATLAB/Simulink does not operate in real-time. The test and evaluation engineer should select the most appropriate simulator for his or her platform.
11
estimate static stability derivatives. Flight data can also be placed directly into simulator to give side by side comparison of responses. More modern system identification techniques are gaining popularity in enhancing UAV dynamic models. This is because wind tunnel and CFD analysis are time consuming and expensive and classical flight test techniques tend to be less accurate at estimating specific stability derivatives. developing aircraft dynamic models from flight test data. Time domain system identification methods were developed at the German Aerospace Research Establishment (References [16] and [17]). These methods utilize the 3-2-1-1 input to excite the aircraft dynamics. 3-2-1-1 refers to step inputs lasting, in sequence, three seconds, two seconds, one second and one second. These Several methods exist for
techniques require an initial parametric model, such as a digital DATCOM model, to work. Reference [18] used these techniques to model a UAV. Frequency domain system identification can be utilized as well. Reference [19] provides detailed information on how dynamic models can be developed using frequency domain methods. Reference [20] used frequency domain system identification on a UAV helicopter and Reference [21] presents how frequency domain techniques can be used to enhance UAV flight controllers. Frequency
domain techniques can be used without parametric models of the system dynamics, can determine high frequency structural modes, and can identify unstable dynamics. They do, however, require more flight time to complete (30 seconds to a few minutes per maneuver). This makes them well suited to helicopters but also makes them
12
difficult to perform on fixed wing UAVs using a remote pilot as the aircraft will leave visual range before the sweep can be finished. Automated frequency sweeps should be used if frequency domain techniques are used on fixed wing UAVs.
13
safe flight test plans prior to first flight. References [9], [10] and [24] provide solid foundations on how to setup a flight test program. Tuning a UAV flight control system in flight should be performed loop by loop if possible. It is important to isolate which each parameter so that problems can be isolated. Starting with the lateral-directional system can be useful as it allows for roll and heading stabilization during longitudinal control testing. Gain tuning can be performed in flight to meet performance specifications but it is important to remember that gain tuning in flight can be expensive, difficult, stressful and dangerous and should be avoided as much as is possible.
2.8 Step 8: Evaluate the Performance of the Control System and Reject or Certify
After flight test of the control system is complete the system should be analyzed to determine of the closed loop performance meets the requirements. If the system meets the requirements and can safely be used then the system can be certified. If not, problem areas can be addressed or the system can be rejected.
14
15
16
The Piccolo II has sixteen digital I/O lines, five serial ports and four analog inputs. The digital I/O lines are used both for payload I/O and servo actuation. A 44pin connector and a 24-pin microdot connector are used to connect the I/O, serial, analog and power lines to the Piccolo.
3.1.1 Gyros and Accelerometers
The Piccolo II uses three Analog Devices, Inc. ADXRS300 MEMS rate gyros and two two-axis ADXL210e MEMS accelerometers to determine the rates and accelerations in all three axes. measurements. The rate gyros give 300 /sec resolution rate
17
3.1.2 GPS
An uBlox TIM-LP GPS receiver is used to provide position data as well as groundspeed determination. The system has an update rate of 4 Hz.
3.1.3 Pressure System
A Freescale Semiconductor 115 KPa mpx4115 barometric pressure sensor, a Freescale Semiconductor 20 KPa differential pressure sensor and a board temperature sensor are used to determine the true airspeed and altitude. The system compares the true airspeed to GPS ground speed to determine winds aloft.
3.1.4 Kalman Filter
The Piccolo uses a Kalman filter to estimate the aircraft attitude in flight as well as determine the sensor biases.
18
The main Piccolo ground station electronics unit manages communication with the avionics, interfaces with the pilot console and streams telemetry and command and control data to and from the operator interface PC. The ground station interface connects to the operator interface using a 9-pin serial connection. To communicate with the avionics the ground station interface utilizes a 1 Watt 2.4 GHz transceiver. This gives a maximum communication range of approximately 6 miles [3]. Figure 8 and Figure 9 give a front and back views of the Piccolo ground station unit.
19
The command and control interface functions from a PC using the Piccolo operator interface. For this research program the computer of choice is an Itronix GOBook III rugged notebook PC of the following specifications:
20
1.8 GHz Intel Centrino Pentium M 745 1 GB SDRAM Integrated GPS and Wireless Communication Fully-Rugged to Mil-STD 810F Cold Tested to -30 C Heated Hard Drive Water-Proofed Keyboard
Pilot commands are relayed to the aircraft when in manual mode using a Futaba T9CAP Super Console. The console is connected to the ground station interface unit using the Futaba compatible buddy-box 6 pin DIN connector. Using the pilot console
21
the pilot is able to activate or deactivate the autopilot and has full control authority over the aircraft including trim settings.
22
In addition the position of the aircraft can be real-time monitored on the navigation page, where a map can be used for geographic referencing to the GPS coordinates (Figure 13). Flight plans are displayed on the maps and can be updated at any time from the operator interface.
23
24
3. Altitude 4. Airspeed 5. Vertical Rate 6. Waypoint Navigation The Piccolo uses total energy methods to control altitude and airspeed. This means that both elevator and aileron are used simultaneously to control altitude and airspeed.
Table 1: Piccolo Gain Scaling Parameters
Geometry
Wing Area ( S w ) Wing Span ( bw ) Vertical Tail Arm ( xacvt
xcg )
Mass
Gross Mass Empty Mass
C L0
Elevator Effectiveness (
dCL ) d e
Lateral Aerodynamics
Aileron Effectiveness (
p ) a
r
) r
Sideslip Effect ( C y )
Engine
Power
25
algorithms are not publicly available. The waypoint monitoring and entry screen on the operator interface is presented in Figure 13.
26
The foundation of the Standard Cloud Cap Simulator (SCCS) resides in ASCII look- up tables for the various lifting surfaces of the aircraft to be modeled. For each lifting surface an ASCII table containing lift coefficient (CL), drag coefficient (CD) and pitching moment coefficient (CM) for various angles of attack was created (or various sideslip angles for the vertical stabilizer). The aerodynamic parameters were generated internally in the simulator using geometric data for the lifting surfaces (span, incidence, dihedral, area, span efficiency, etc). Control
surfaces were modeled by declaring the inboard and outboard control surface locations as well as the average chord length of the control surface. Fuselage information was also declared, including pitching moment coefficient with respect to angle of attack, yawing moment with respect to sideslip, parasite drag, etc. The propulsion model used a simple power vs. RPM look up table to determine the engine power. The simulator assumes a linear relationship between throttle movement and power generation. From the table the current engine RPM was determined. The advance ratio of the aircraft was then determined as:
J=
Va nd
3.2
27
A propeller model input by the user was then used to lookup the thrust coefficient for the propeller in question and calculates the aircraft thrust. The thrust was determined as:
T = n 2 d 4CT
The power absorbed by the propeller is calculated from the same lookup table as:
Pabsorbed = n 3d 5C p
3.3
3.4
The mismatch between the propeller power and the engine power causes a change in the propeller RPM based on the moment of inertia of the propeller. The RPM will change until the power absorbed and the power generated match. The final set of items in the SCCS are the mass properties of the aircraft. The empty mass, gross mass, and moments of inertia are input to generate the dimensional stability derivatives. As with the other items this data is provided by the user. Reference [29] gives a detailed description of the SCCS and Appendices A through F can be referenced for the SCCS models used during this research program. There are several negatives associated with the Standard Cloud Cap Simulator. Probably the most significant negative is the extreme black box nature of the simulator. It is very difficult to troubleshoot any mistakes in the modeling as only a few stability derivatives are output by the simulator. The other problem is that modeling of a V-Tail cannot be done directly, but must be done using the projected surface on the vertical and horizontal planes. This is not an especially accurate representation, especially as no vortex interactions are taken into account. Another slight disadvantage of the simulator is the lack of a graphical projection of the tabular
28
surface data that would assist in troubleshooting errors in the geometry input to the simulator file.
3.7.2 Athena Vortex Lattice Based Simulator
Cloud Cap Technology offers a second modeling technique that addresses some of the issues associated with the SCCS. This simulator makes use of Athena Vortex Lattice (AVL), a vortex lattice code developed by Mark Drela and Harold Youngren at the Massachusetts Institute of Technology. The basic logic behind AVL is that the aircraft stability derivatives generated by its lifting surfaces can be estimated based on the steady vortex shedding of the surfaces at small angles of attack and sideslip. In AVL the fuselage effects are ignored. In the model each major airfoil section can be defined. Using an aircrafts wing as an example, the center airfoil section, the airfoil section where the flap or aileron starts and ends, the airfoil section where the sweep angle changes, or the airfoil section where the dihedral changes, can all be defined so the geometry will exactly match the actual aircraft. This also allows for the creation of v-tail geometry models directly. A significant advantage of AVL is that the geometry can be graphically represented in three dimensions, allowing for increased troubleshooting ability. Figure 14 shows a 3-D model of the Aerosonde that was included with the simulator. The stability derivatives about the center of gravity are then calculated using the lifting surface geometry. The angle of attack of the aircraft is varied based on intervals set by the user and the stability derivatives are determined for each angular
29
position. The stability derivatives are output to an .XML file for use by the Cloud Cap Simulator. The simulator uses the same propulsion and inertia models as the SCCS, substituting in the AVL aerodynamics for the lifting surface tables. The .XML file allows greater insight into what goes into the Cloud Cap Simulator, allowing for better troubleshooting of the dynamics model. Reference [30] contains a more detailed discussion of the AVL simulator.
30
simulating the sensor data that would be generated by the Piccolo in flight. The Piccolo then calculates the control surface positions using its control algorithms. These control surface commands are sent back to the simulator and the system recalculates the dynamics. The hardware-in-loop simulation can also be performed pilot-in-loop using the Futaba console, allowing for manual piloting to be performed and static stability to be analyzed. Visualization of the aircraft can be performed using a second PC using Flight Gear. A UDP network is created that sends the simulation data to the PC running the flight gear software. This allows for better understanding of the aircraft states.
Piccolo Avionics
CAN Bus
Visualization PC
31
32
33
34
The details of
W I XX IYY I ZZ I XZ
Parameter Takeoff Weight Moment of Inertia About the X Axis Moment of Inertia About the Y Axis Moment of Inertia About the Z Axis Moment of Inertia About the X-Z Axis
Units lbs Slug ft2 Slug ft2 Slug ft2 Slug ft2
35
The Piccolo issues throttle commands as a percentage of maximum power. Thus it was important to calibrate the throttle servo position based on that positions output power. A power measuring device was not available for the engine and thus an approximation was used that assumed thrust varied linearly with power. Thrust was linearized as:
T =
T Tmax
3.1
This revealed the throttle arm position pulse curve shown in Figure 19.
1.00 0.90 0.80 0.70 % Max Throttle 0.60 0.50 0.40 0.30 0.20 0.10 0.00 1100 1200 1300 1400 1500 1600 1700 1800 1900 2000
Normalized RPM Normalized Thrust Normalized Thrust Used
Figure 19: Normalized Thrust and RPM Curves for Engine Calibration
36
4.6 Antennas
The Yak-54/Piccolo system requires two antennas for flight operations. The first was a Maxrad low profile 2.4 GHz antenna that was used for UHF communications with the ground station and pilot console. The 2.4 GHz antenna was mounted to a 3.25 x 4 aluminum ground plane to the rear of the aircraft. The second antenna was a MK-4 GPS antenna used for guidance and position information. The MK-4 was mounted in the Yak-54 canopy on a small 2.5 x 2.5 aluminum ground plane.
4.7 Batteries
Five batteries were used on the Yak-54: 1. Two Fromeco Relion 5200 mAh 7.4 Volt lithium ion Batteries with two Fromeco The Regulator 7.4 Volt to 6 Volt regulators for servo power. The Piccolo had two servo power inputs so no external wiring was required. 2. Two Thunder Power Pro Lite 2100 mAh 11.1 Volt lithium polymer batteries for Piccolo II main power. These batteries were wired in parallel to provide 4200 mAh of capacity.
37
The kill switch operates with a standard RC receiver (72 MHz) and a second Futaba pilot console. Should the Yak-54 become uncontrollable and communication cannot be reestablished the kill switch can be triggered only by the secondary pilot console, thus eliminating the possibility of a powered and careening aircraft. The engine kill switch can NOT be triggered because of a power loss to any part of the system. The use of a fiber optic system also greatly reduced the possibility of interference causing the engine to be killed inadvertently.
38
q
1
The dimensionless derivatives calculated by AAA for these trim conditions are shown in Table 5.
39
Longitudinal Derivatives
Lateral Derivatives
C y Cyp
C Du C D CTX u C Lu CL CL&
CLq
0.0011 0.0863 -0.1546 0.0017 4.5465 1.8918 5.5046 0.0002 -0.3937 -4.3787 -8.7532 0.0000 0.0275 0.0000 0.3762 -0.8778
rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1
-0.3462 0.0080 0.2324 -0.0255 -0.3819 0.0504 0.0955 -0.0041 -0.0171 -0.1161 0.0000 0.2189 0.3490 0.0178 -0.0080 -0.1129
rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1 rad-1
C yr
Cl Cl p
Clr
Cn
Cmu Cm Cm&
C mq
CnT
Cn p
Cnr C y C y
a
Cl a Cl r Cn a Cn r
CL1 CD1
CTx
1
Cm1
CmT
1
From the lateral dimensionless stability derivatives and the trim conditions the
40
lateral state space model was constructed. Please see Reference [33] for the details of the state space model construction.
& p 16.6423 1.0000 & = & 0.0004 & r 0.0958
0.0000 35.7516 2.3092 p 454.0359 22.8520 0.0000 0.0000 0.0000 0.0000 0.0000 a + 0.2722 0.6017 0.9865 0.0000 0.3353 r 0.0000 43.4267 1.8709 r 14.0214 46.9710
From the state space models the eigenvalues were used to analyze the modes of the Yak-54.
Table 7: AAA Yak-54 Lateral Modal Analysis [33]
Eigenvalues
Damping Ratio,
~ ~ 0.18
Period, T (sec)
~ ~ 0.95
Mode
From the eigenvalue analysis it can be seen that the Yak-54 model exhibited the standard three lateral dynamic modes. The spiral mode was slightly unstable, which was to be expected as the Yak-54 wing has no dihedral. The roll mode was quite fast, which would be typical of a very small aerobatic aircraft. The Dutch Roll mode was lower damped than is typical.
5.1.2 Longitudinal State Space Model and Analysis
As with the lateral modes, the longitudinal state space model was formed from the trim conditions and the dimensionless stability derivatives. The details of this calculation can also be found in Reference [33].
41
& 0.0000 32.1554 u 0.0000 u 0.2751 12.9976 0.0043 7.8357 0.9229 & 0.0091 0.6409 + [ e ] = & q 0.0167 19.1618 9.6043 0.0299 q 105.5632 & 0.0000 1.0000 0.0000 0.0000 0.0000
From the state space model the eigenvalues of the A matrix can be found and a modal analysis performed.
Table 8: AAA Yak-54 Longitudinal Modal Analysis [33]
Eigenvalues
Mode
The Yak-54 longitudinal dynamic analysis showed somewhat atypical behavior. The Phugoid mode has higher damping than is typical and the short period mode is very highly damped.
42
Table 9 shows the stability derivatives generated by the vortex lattice simulation.
Table 9: AAA vs. AVL Longitudinal Stability Derivatives
Stability Derivative CL
AAA
AVL
% Difference
4.5465 rad-1 -0.3937 rad-1 5.5046 rad-1 -8.7532 rad-1 0.3762 rad-1 -0.8778 rad-1
3.7007 rad-1 -0.2850 rad-1 3.5927 rad-1 -4.3732 rad-1 0.3976 rad-1 -0.7572 rad-1
Cm
C Lq Cm q
10 20 20 ~ ~
CLe Cme
As can be seen in the table the percentage difference between the two derivative predictions was about twice as large as the margin of error given for AAA. Some of this may be due to the lack of good fuselage modeling in AVL. The AVL model trimmed at an angle of attack of = 1 and the trim
43
Another modeling comparison item is the gain scaling terms. longitudinal gain scaling terms were Cme (elevator power) and
dCL (elevator d e
effectiveness). Elevator effectiveness describes a steady state change in lift with a trim change in elevator and can be calculated as:
CL Cm CLe Cm dCL = e d e Cm
Table 11 shows a comparison of these terms as calculated by AAA and AVL.
Table 11: AAA vs. AVL Longitudinal Gain Scaling Terms
5.1
AAA
AVL
-1
% Difference
-1
-0.8778 rad
-0.7570 rad
13.70 3.92
dCL d e
-9.7608 rad-1
-9.3782 rad-1
The AVL generated gain scaling terms, especially elevator effectiveness, show a fairly good match with the AAA model.
5.2.1.1 Phugoid Approximation
From the derivatives output by the AVL simulator a modal analysis was performed using the Phugoid and short period approximations given in Reference [10]. The Phugoid mode was approximated as motion involving a constant angle of
44
attack and constant pitching moment. Thus these terms can be neglected in a modal analysis. The linearized longitudinal equations of motion can then be approximated as:
u ( s) ( s X u X Tu ) ( g cos 1 ) e (s) = X e ( s) Z e Zu ( Z q + U1 ) s + g sin 1 e (s)
5.2
Further approximation were performed by assuming that Z q << U1 and 1 0 . Roskam [10] states that it is appropriate to consider the thrust contribution to the speed damping term as part of the total speed damping effect of the aircraft and thus
X u a X u + X Tu . This was a useful approximation for our purposes as AVL does not
take into account thrust effects, however with such a high thrust to weight ratio this approximation may be inappropriate which will be elaborated on further later on in this section. These approximations yielded a new system of equations:
u ( s) g ( s X u ) X e ( s) = e Z ( s) e U1s Zu e ( s)
5.3
The speed and pitch transfer functions were then approximated as: ( X e U1s + gZ e ) u ( s) = e ( s) U ( s 2 X s gZ u ) u 1 U1 5.4
( s ) ( Z s X u Z X Z u ) = gZ e (s) U (s2 X s u )
e e e
5.5
U1
45
The denominator of these two transfer functions were used as the approximate quadratic of the Phugoid mode, with the damping ratio and frequency being written as:
n
ph
gZ u U1
5.6
ph
Xu 2n ph
5.7
The speed dimensional derivatives were found from the AVL generated derivatives: Zu = q1 S w (C Lu + 2C L1 ) mU1 q1 S w c (C Du + 2C D1 ) mU 1 speed derivatives at low Mach 5.8
5.9 number
( CLu CDu Cmu 0 ) the dimensional speed derivatives were calculated as: Z u = 0.4938 rad / sec X u = 0.1124 rad / sec
Finally with these derivatives the Phugoid mode damping and frequency were calculated:
n 0.367rad/sec
ph
ph 0.15
Comparing these to the values in Table 8 it was seen that there is a significant
46
difference in natural frequency and a significant difference in the damping ratio of the Phugoid mode. One reason for this discrepancy was the removal of X Tu from the Phugoid approximation. If the Phugoid approximation is used on the AAA based data it was seen that the Phugoid mode damping ratio and natural frequency are approximated as:
n 0.377rad/sec
ph
ph 0.15
However, if the X Tu term is added to the Phugoid approximation such that:
ph
(Xu + XTu ) 2 n ph
5.10
The Phugoid mode damping ratio was recalculated taking into account the propulsion system:
ph 0.365
While this approximation of the Phugoid damping ratio is closer to the eigenvalue analysis, it is still off by greater than 40%. The frequency is also not corrected by the addition of X Tu to the analysis. Clearly the use of conventional analytical techniques did allow a correct analysis of the Phugoid mode using only the aerodynamic data derived from AVL. Additionally the Phugoid approximation only yields accurate representations of the vehicle dynamics in a limited number of cases. The lack of a good thrust model to produce the thrust data significantly degrades our ability to perform any analytical
47
analysis on the Phugoid mode and thus another way to analyze the Cloud Cap simulator must be used.
5.2.1.2 Short Period Approximation
In order to better understand the short period motion described by the AVL analysis it was useful to perform a short period approximation, as described in Reference [10]. In the short period approximation the speed degree-of-freedom terms in the longitudinal state space model were neglected, as short period motions occur at constant speed. Thus the new equations of motion were approximated as:
5.11
The following additional approximations were made: Z& << U1 , Z q << U1 and 1 0 . Additionally the thrust contribution were neglected as part of the total static longitudinal stability of the aircraft such that X u a X u + X Tu . These approximations yielded the following angle of attack and pitch attitude transfer functions:
(s) = e ( s)
Z e + M eU1 M q Z e Z M q Z U1 s 2 M q + + M & s + U M U1 1
5.12
( s) = e ( s)
(U M
1
+ Z e M & s + M Z e Z M e
) (
Z M Z sU1 s 2 M q + + M & s + q M U U1 1
5.13
48
From the denominator quadratic the short period mode damping ratio and frequency were found as:
n
sp
Z M q U1
5.14
sp
Z M q + + M & U1 2 nsp
5.15
M =
q1S w c (Cm ) I yy
q1S w c (Cm& ) 2 I yyU1
5.18
Mq =
5.17
M & =
5.19
AVL does not compute angular rate terms, Cm& = 0 , and thus M & = 0 . The rest of the dimensional derivatives are calculated as: Z = 766.5399 M q = 3.2816 M = 34.9539 M & = 0
From the stability derivatives the short period approximation was used to determine the AVL determined short period mode natural frequency and damping:
n 7.5 rad/sec
sp
sp 0.6513
49
Using these same approximations with the AAA derivatives the short period mode damping and frequency are:
n 10.0406 rad/sec
sp
sp 0.89
This shows that the use of the short period approximation was a valid assumption in this case. It is interesting to note the effect of neglecting the angular rate term from the short period damping approximation:
sp ( M & = 0) 0.7253
Neglecting the rate damping terms had a significant effect on short period damping, reducing it by 0.165.
50
Stability Derivative Cl
Cn C y
AAA
AVL
% Difference
-0.0255 rad-1 0.0955 rad-1 -0.3462 rad-1 0.0504 rad-1 -0.1153 rad-1 0.2324 rad-1 -0.3819 rad-1 -0.0171 rad-1 0.0080 rad-1 0.3490 rad-1 0.0154 rad
-1
-0.0314 rad-1 0.1052 rad-1 -0.2727 rad-1 0.07427 rad-1 -0.1156 rad-1 0.2531 rad-1 -0.5858 rad-1 -0.03874 rad-1 0.0194 rad-1 0.3707 rad-1 0.02194 rad
-1
-23.23 -10.19 21.24 -47.37 0.57 -8.91 -53.38 -126.57 -142.98 -6.05 -43.96 -0.70 -15.5
20 15 20 60 60 60 40 25 30 ~ ~ ~ ~
Cl r Cn r C yr
Cl p Cn p Cy p
Cl a
Cl
r
Cn
C y
The important gain scaling terms in the lateral-directional control system were: aileron effectiveness (
p ) , rudder power ( Cn r ), rudder effectiveness ( ) a r
5.20
5.21
Table 13 shows the gain scaling terms produced by AAA and AVL.
51
Scaling Term
AAA
AVL
% Difference
p a
C n
r
-0.1003 rad-1
rad -1.0429 rad
-0.0996 rad-1
rad -0.9467 rad
r
C y
-0.3462 rad-1
-0.2727 rad-1
As the table shows, AAA predicts greater rolling and yawing performance with aileron and rudder deflections than does AVL. This would result in a lower gain on the respective feedforward and feedback terms.
5.2.2.1 AVL Based State Space Model Generation and Lateral-Directional Modal Analysis
AVL generated all the necessary derivatives, save N T , to generate a complete linear state space of the lateral-directional dynamics. Prior to generating a state space model the lateral-directional dimensional derivatives were constructed. Using the definitions found in Reference [10], the lateral-directional derivatives are shown in Table 14.
52
Y Yp
Yr
ft / sec 2 rad ft / sec 2 rad/sec ft / sec 2 rad/sec ft / sec 2 rad ft / sec 2 rad rad / sec 2 rad rad / sec 2 rad/sec rad / sec 2 rad/sec
L a L r N
N T
rad / sec 2 rad rad / sec 2 rad rad / sec 2 rad rad / sec 2 rad rad / sec 2 rad/sec rad / sec 2 rad/sec rad / sec 2 rad rad / sec 2 rad
Y a Y r L Lp
Lr
Np
Nr
Na Nr
The lateral-directional state space model was constructed using the linearized equations of motion:
& & & & U 1 + U1 = g cos 1 + Y + Y p + Yr + Y a a + Y r r & & & & & A1& = L + L p + Lr + L a + L r
a r
A1 =
I xz I xx
B1 =
I xz I zz
53
& & & & & & In straight and level flight ( p = , p = & , r = and r = ) the equations of
Using the values given in Table 14 and performing the necessary matrix algebra the lateral-directional state space model was constructed:
& p 25.5465 1.0000 & = & 0.0011 & r 0.3200
0.0000 42.4133 3.2982 p 481.4849 30.2026 0.0000 a 0.0000 0.0000 0.0000 0.0000 + 0.3872 r 0.2723 0.4739 0.9852 0.0000 0.0000 49.5492 1.8461 r 5.3924 47.1123
An eigenvalue analysis was performed on the lateral-directional state space model in order to characterize the lateral-directional dynamics described by the AVL modeling. These results are given in Table 15.
Table 15: AVL Yak-54 Modal Analysis
Eigenvalues
Damping Ratio
Frequency (rad/sec)
Period (sec)
~ ~ 0.16
~ ~ 7.13
~ ~ 0.88
Mode
This eigenvalue analysis shows similar dynamics to the AAA model, with the biggest difference being in the spiral mode dynamics. A more thorough comparison will be made at the end of this chapter.
54
transient peak ratio method was used and for the higher damped short period mode the maximum slope method was used. The spiral mode and roll mode were not estimated, the spiral mode because of the very slow time constant that is difficult to confirm in flight test and the roll mode because of a time constant so fast that the dynamics cannot be evaluated at the Piccolos 20 Hz sampling rate.
5.3.1 Modified Transient Peak Ratio Method
The modified transient peak ratio method (MTPR), as described in Reference [14], was a convenient way to estimate the frequency and damping of an oscillatory response with a damping ratio range of -0.5 < < 0.5. parameters of interest for the MTPR. Figure 22 shows the
55
d =
The transient peak ratios were measured as:
2 T
5.27
x x2 x3 x4 = = = .... = i +1 x1 x2 x3 xi
5.28
Typically these ratios do not match exactly so an average was used. Figure 23 was used to determine the damping ratio of the oscillatory response. From the damping ratio the natural frequency was calculated as:
n =
d 1 2
5.29
56
The maximum slope method (MS), as described in References [14] and [15], was a useful approach for determining the natural frequency and damping ratio when the damping ratio of a time response is between 0.5 < < 1.2. Figure 24 shows the parameters used in the MS method.
In the MS method the maximum slope tangent line and the half-cycle peak lines were drawn and from there the intersection of the maximum slope tangent line and the horizontal half-cycle peaks t is defined. The second vertical lines
intersection with the response curve defines x1. x was defined by the distance between the half-cycle peaks. Finally the damping ratio and natural frequency were determined using Figure 25.
57
n =
n T
T
5.30
Using the AVL-Piccolo hardware-in-loop simulation the Phugoid mode dynamics were analyzed using the modified transient peak ratio method. The
Phugoid mode dynamics were excited using a step elevator input that lasted until the airspeed had left trim ( 70 knots) by approximately 5 knots. At that point the elevator is released and the aircraft is allowed to oscillate freely. Figure 26 shows a time history of this test.
58
80
75
70
65
380
390
400
410
420
430
440
Time (sec)
Elevator (deg)
0.7 0.6 0.5 0.4 0.3 0.2 370 380 390 400 410 420 430 440
Time (sec)
The reader is referenced to Appendix H for the calculation tables used to determine the damping and frequency which yielded the following results:
n = 0.3 rad/sec
ph
ph = 0.17
5.3.4 Open Loop Modal Analysis: Short Period Mode
Using the AVL-Piccolo hardware-in-loop simulation the short period mode dynamics were analyzed using the maximum slope method. The short period mode was excited using an elevator doublet. Figure 27 shows the short period time response.
59
100
50 0 -50
224
224.5
225
225.5
226
226.5
227
Time (sec)
6 4
Elevator (deg)
Time (sec)
The reader is referenced to Appendix I for the calculation tables used to determine the damping and frequency of the short period response which yielded the following results:
n = 12.89 rad/sec
sp
sp = 0.85
5.3.5 Open Loop Modal Analysis: Dutch Roll Mode
Using the AVL-Piccolo hardware-in-loop simulation the Dutch roll mode dynamics were analyzed using the maximum slope method. The Dutch roll mode was excited using a rudder doublet. Figure 28 shows the Dutch roll time response.
60
60
40 20 0 -20 -40 -60 -80 467.5 468 468.5 469 469.5 470 470.5 471
Time (sec)
10 5 0 -5 -10 467.5
Rudder (deg)
468
468.5
469
469.5
470
470.5
471
Time (sec)
The reader is referenced to Appendix I for the calculation tables used to determine the damping and frequency of the Dutch roll response which yielded the following results:
n = 7.06 rad/sec
dr
dr = 0.15
61
analyze the validity of the simulator therefore was to perform time response modal excitation and then reduce the data to approximate the damping ratio and natural frequency of the oscillatory modes and the time constant of the first order modes. Using the same data analysis techniques used in the AVL-HIL model the SCCS simulation model can be analyzed. From Reference [33] we see that the modes exhibited the characteristics shown in Table 16.
Table 16: SCCS Modal Analysis [33]
Mode
Roll (sec)
Phugoid n (rad/sec)
81.3 ~ 55.87 ~
0.06 ~ 0.04 ~
Table 17 shows the Dutch roll dynamics were fairly similar between the AVL
62
and AAA simulators, with the most variation being in frequency. It was interesting to note that the AVL-HIL model shows a large difference in the short period mode as opposed to using the short period approximation in the AVL aerodynamics only model. This is most likely due to some power effects in the Cloud Cap simulator, but this cannot be known for certain as the simulator acts as a black box with the input data. The accuracy of the frequency estimation from simulation for a highly damped mode like the short period mode is not especially accurate and thus the short period frequency data was not reliable. The Phugoid mode showed the biggest variation among the modeling techniques. The AVL vs. SCCS show minimal differences, and the AVL-HIL
showed slightly higher damping and lower frequency than the Phugoid approximated AVL terms. The AAA model showed significant differences in Phugoid from the other models. The question remained, which model is closest to the truth? To determine this open-loop flight tests were performed to compare to the modeling techniques. Chapter 8 will present this evaluation.
63
64
and the Vehicle Engineer, ensures that the vehicle and ground station are ready for flight. Appendix K can be referenced for the pre-flight checklist in its entirety.
The FTE was responsible for the following: 1. Responsible for overall test and crew members 2. Briefed crew and pilot on responsibilities 3. Hade go/no-go decision 4. Performed pre-flight checks with the Vehicle Engineer 5. Communicated with the pilot during flight test 6. Communicated with engineering team during flight test 7. Instructed the pilot on which flight test point to follow 8. Responsible for managing the Piccolo ground station 9. Relayed all gain changes determined by the engineering team to the Piccolo 10. Ensured ground station is in proper working order 11. Responsible for all autonomous flight commands to the Piccolo 12. Made Piccolo autopilot activation/deactivation calls 13. Made gain tuning decisions
65
The PIC had the following responsibilities: 1. Safety was the PICs highest responsibility 2. Made independent per-flight checks 3. Performed all manual flight activities 4. Activated/Deactivated autopilot 5. Deactivated the autopilot if the PIC judged the aircraft to be entering an unsafe flight condition 6. Assisted engineers with aircraft performance evaluation
6.2.3 Vehicle Engineer (VE)
The VE had the following responsibilities: 1. Ensured aircraft is in proper working order 2. Performed Weight and Balance 3. Confirmed avionics and power systems were installed appropriately and are in working order 4. Ensured all control surface actuators are properly installed and calibrated 5. Performed pre-flight checks with the FTE 6. Ensured engine is tuned and in working order
6.2.4 Safety Officer (SO)
The SO had the following responsibilities: 1. Ensured all procedures detailed in this document are followed
66
2. Ensured only maneuvers specified in the dance cards are performed 3. Had go/no-go authority with regards to safety 4. Observed that general safe practices are followed 5. Made 5, 10 and 15 minute engine ON time calls 6. Checked FTE gain entries
6.2.5 Data Processing Engineer (DPE)
The DPE had the following responsibilities 1. Evaluated in flight data 2. Evaluated maneuver performance acceptability 3. Communicated flight test point maneuver that is desired to the FTE 4. Made go/no-go recommendations to the FTE with regards to flight data quality
6.2.6 Pilot Assistant (PA)
The PA had responsible for the following: 1. Assisted the PIC in intra-team communications 2. Read test card points to the PIC 3. Assisted the PIC in vehicle trimming 4. Performed air traffic scanning for the pilot 5. Generally assisted the PIC during the entire test phase
6.2.7 Video Operator (VO)
The VO was responsible for filming all flight activities during the test.
67
The flight test communications and organizational charts are displayed in Figures 34 and 35. It was important that communications and decision making flow in the order dictated by the flow charts, minimizing confusion that could have resulted in longer testing times, poor data, or at worse loss of the aircraft.
Safety Officer
Flight Team
Engineering Team
68
69
would be attempted while the aircraft is orbiting. The pilot would continue to attempt control reacquisition until the aircraft is terminated or control could be reestablished and an emergency landing attempted with zero power. If communication could not be reestablished the engine would be killed using the 72 MHz emergency engine kill switch and the aircraft should descend and crash. In the event that the aircraft could not maintain the lost communications orbit and appeared to be careening out of the flight test area the pilot would call KILL KILL KILL and the safety officer would hit the engine kill switch on the 72 MHz Futaba transmitter, killing the engine.
70
actuator movement that were caused by 10 Hz sampling of the Futaba pilot console. These discrete actuator movements were not present when a standard RC controller and receiver were used. The Hitec HS-5985MG servos were high performance hobby grade digital servos, with an operating speed of 0.13 sec/60, and thus it was anticipated that the aircraft dynamics would damp out any discrete servo operation because the servos operated so quickly. Discussions with Cloud Cap Technology about the discrete servo operation caused by the 10 Hz pilot console sampling revealed that the pilot console sampling rate could not be increased and because no improvements could be made on the user-end with the resources available it was decided to accept these limitations for the Yak-54/Piccolo test program and proceed. The discrete time-lag of the Piccolo system under manual control remained the biggest concern to the flight test team and the effects on handling qualities needed to be determined. The standard test maneuver was a race track (Figure 32) as well as full stop takeoffs and landings. The pilot will then give an analysis of the handling qualities of the aircraft in general and as compared with those of the same aircraft under standard RC control. If adjustments could be made with the pilot analysis they would be
71
performed, otherwise the flight test program would proceed to the next phase. The reader is referenced to Appendix L for the handling qualities flight test dance card.
72
950 ASL (ft) 900 850 800 24 24.1 24.2 24.3 Time (min) 24.4 24.5 24.6
20 10 0 -10 24 0 24.1
24.2
24.4
24.5
24.6
Elevator (deg)
-5 -10 -15 24
Figure 33 shows poor pitch damping during approach to landing that goes unstable when there are indications of data dropouts. It should be noted that uplink data drops (e.g. pilot commands) are NOT shown in the stored telemetry data. Thus there could be additional loss of pilot commands not shown in the telemetry. The data uplink issues were not known at this point of the flight test but several measures were taken to increase pitch stability. First the C.G. was moved forward 0.5 inches. It was shown in the parameter identification flight tests that this resulted in a short period mode damping of sp = 1.0. The pilot also made a conscious effort to lead the aircraft, thus counteracting some of the discrete time lags associated with the 10 Hz sampling. Figure 34 shows improved pitch stability, although the pilot still reported poor handling qualities as compared with a standard RC controlled Yak-54.
73
1100 ASL (ft) 1000 900 800 10 20 10.05 10.1 10.15 10.2 10.25 Time (min) 10.3 10.35 10.4 10.45 10.5
Elevator (deg)
-4 -6 -8 10 10.05 10.1 10.15 10.2 10.25 Time (min) 10.3 10.35 10.4 10.45 10.5
It should be noted that in this telemetry there was no indication of downlink data loss. At this point, as the data uplink issue had not been discovered, it was decided that the handling qualities, although poor, were good enough to continue on to the open-loop Yak-54 flight testing program.
74
comparison with the simulator dynamics will be presented. Appendix M can be referenced for the open loop flight testing dance card.
75
One Dutch roll excitation with a rudder singlet and two Dutch roll excitations with a rudder doublet were performed on the Yak-54. The three Dutch roll responses are presented in Figure 35, Figure 36 and Figure 37.
76
77
Using the MTPR method the frequency and damping ratio can be extracted from the flight test data. The data analysis tables can be found in Appendix N and the results of this analysis are presented in Table 18.
Table 18: Dutch Roll Flight Test Results
Test II
Test III
Average
0.23 6.05
0.22 5.68
0.26 5.92
0.24 5.88
An analysis of these results versus the simulation models is presented at the end of this chapter.
Three short period data sets were used for dynamic analysis and are presented in Figure 38, Figure 39, and Figure 40.
78
79
Reference [33] used the time-ratio (TR) method to analyze the short period data presented above. This method, as described in Reference [14], is useful when the damping ratio is near = 1.0. The TR utilizes three specific values with respect to the steady state value along the response curve. The time along the response curve for 0.736x, 0.406x and 0.199x, as shown in Figure 41, was taken.
80
From these values the ratios t2/t1, t3/t1, and (t3 t2)/(t2 t1) were used in Figure 42 to obtain three estimates of . These three estimates are then averaged together to obtain the final value of . The results of this analysis technique are presented in Table 19.
Table 19: Short Period Mode Flight Test Results
Test II
Test III
Average
0.97 13.47
1.0 19.96
1.03 16.44
1.0 16.62
From the plots it was easy to see the short period response is more or less first order. This was easy to anticipate, as the forward movement of the C.G. would result in a highly damped short period mode. This highly damped short period mode helps
81
maintain pitch stability during loss of pilot data packets during manual operations. A comparison with the simulation models will be presented at the end of this chapter.
82
As can be seen by the figures, the Phugoid mode was extremely slow, coming close to reaching the half-period peak at about 10 seconds into the test in the second plot. The airspeed also drops significantly, all the way to around 40 knots IAS. Classical data reduction techniques, like the MTPR method, cannot be used to analyze the dynamics presented in these data. However, it can be useful to make a graphical comparison with the various modeling and simulation packages analyzed before. This will be done in the following section.
83
Source
Phugoid n (rad/sec)
Table 20 shows the Dutch roll mode shows higher damping and lower frequency than all the simulation models predict. The AAA model is the closest to the real dynamics. As with the Dutch roll mode, the short period mode has higher damping than the any of the simulation models predict. The short period mode also had much higher frequency than any of the simulation models predict. The AVL-HIL simulation shows the closest short period frequency while the AAA shows the closes damping ratio. None of the modeling methods show major differences in either short period or Dutch roll. It should be noted that the time ratio method and the maximum
slope method of data reduction can be highly inaccurate in their estimation of natural frequency. This was most likely the cause of the discrepancy between the AAA and AVL short period natural frequencies and the SCCS, AVL-HIL and flight test short period natural frequencies. Damping ratio is generally much more accurately
calculated with the flight test data reduction techniques. There were, however, large differences between the models in Phugoid mode damping. The frequency predicted by the AAA model and the AVL-HIL simulator was very close but the damping ratio of the AAA model Phugoid mode is a full 0.32
84
higher. Because of the incomplete flight data, direct comparison with the Phugoid mode could not be made. However, plots of the various modal predictions were a useful tool in evaluating the Phugoid dynamics. Simulink was used to simulate the state space model presented in Section 5.1.2. Using similar elevator inputs to those shown in Figure 44 a 10 second time period plot was made and is shown in Figure 45.
70 60 50 40
10
Time (sec)
0.5
Elevator (deg)
0 -0.5 -1 -1.5 -2 0 1 2 3 4 5 6 7 8 9 10
Time (sec)
Figure 45: AAA Yak-54 Phugoid Simulation
The damped airspeed response in Figure 45 was slightly faster than the results of the flight test, with a half period peak occurring about 6 seconds into the response. It should also be noted that with these large airspeed changes the validity of the
85
model tends to degrade. The AVL-HIL simulation shows a faster initial response (Figure 46).
70
60 50 40 30
401
402
403
404
405
406
407
408
409
410
Time (sec)
2 1.5
Elevator (deg)
1 0.5 0 -0.5 -1 -1.5 401 402 403 404 405 406 407 408 409 410
Time (sec)
The response in Figure 46 was less damped, with the aircraft both decelerating more and pitching up faster once the elevator is released. A noticeable time-delay was also present in Figure 46 which was most likely due to the sampling of the inherent lag in the Piccolo system. In any event, the Phugoid response in flight test appeared to be slower than either simulator. This indicates that the actual dynamics fit closer to the higher damped Phugoid predicted by AAA model.
86
87
closest possible to the predicted response. Once first flight has taken place system identification will greatly improve the closed loop performance, as system identification will allow for the enhancement of high fidelity models even further and allow for gain tuning to continue in a laboratory environment where the vehicle is not at risk. It is also recommended that autonomous Phugoid mode flight tests be
88
89
Geometry
Wing Area ( S w ) Wing Span ( bw ) Vertical Tail Arm ( xacvt 10.90 7.90 ft2 ft ft slugs slugs slug ft2 slug ft2 slug ft2 rad-1 ~ rad-1
xcg )
Mass
Gross Mass Empty Mass
C L0
Elevator Effectiveness (
dCL ) d e p ) a
r
-9.76
Lateral Aerodynamics
Aileron Effectiveness ( -0.9139 -0.0996 -1.0429 -0.3462 rad-1 rad-1 rad/rad rad-1
) r
Sideslip Effect ( C y )
Engine
Power 11.5 Hp
The Piccolo II was designed to have stable response with these gain scaling terms and the default gain settings. Further tuning of the gains resulted in enhanced performance. The default gain settings for the Piccolo II autopilot are given in Table 22 and Table 23.
90
Table 23: Piccolo Default Longitudinal Gains Table 22: Piccolo Default LateralDirectional Gains
Default Gain
Gain Parameter Roll Command Inner Loop Roll Error to Roll Rate Roll Rate Error to Aileron Roll Rate Error Integral to Aileron Yaw Inner Loop Yaw Rate Error to Rudder Side Force Error Integral to Rudder
Default Gain
0.00 1.50
Z Acceleration Error Integral to Elevator Pitch Damper Pitch Error to Elevator Pitch Rate Error to Elevator Airspeed Control Outer Loop TAS Error to TAS Rate TAS Rate Error to Z Acceleration Throttle Prediction Trust Energy Rate Error Integral to Throttle Altitude Control Outer Loop Altitude Error to Altitude Rate Command Altitude Error to Z Acceleration Command
0.00 0.00
1.00 0.00
Heading Control Outer Loop Heading Error to Turn Rate Heading Error Derivative to Turn Rate Waypoint Navigation Tracker Convergence
0.40 0.10
0.35
0.20 0.50
91
control loops but the gain was set solely based on the gain scaling term Performance parameters were computed in the following manner:
All Responses:
. r
1. Rise Time ( t r ): The time it takes for the system to go from the initial value to
Mp =
4. Settling Time:
y (t p ) y ss y ss
9.1
maintain a value within 2% of the difference between the initial value and the command value. The closed loop bank angle response was the first item analyzed. Figure 47 presents the closed loop response to a ten degree bank angle command. Table 24 shows the performance parameters of the system. The bank angle control loop with the default gains showed an overdamped response. The steady state error was low and would be acceptable for a final system. Gain tuning focused on improving the rise time of the system (i.e. decreasing
92
damping).
B a n k A n g le (d e g )
15 10 5 0 -5 96 97 98 99 100 101 102
-5 96
97
98
99
100
101
102
Time (sec)
0.8
L H S A ile ro n (d e g )
97
98
99
100
101
102
Time (sec)
Table 24: Bank Angle Control Loop Performance with Default Gains
3.65 0.2
sec deg
The heading command outer-loop response to a 10 degree heading change is shown in Figure 48 and the performance parameters are presented in Table 25.
93
231
232
233
234
235
236
237
238
Time (sec)
231
232
233
234
235
236
237
238
-0.5 230
231
232
233
234
235
236
237
238
Time (sec)
Figure 48: AVL-HIL Heading Angle Response with Default Gains: Table 25: Heading Angle Control Loop Performance with Default Gains
4.1 1.5
sec deg
The closed loop heading command response showed a slow, overdamped response with only a small amount of bank used of the 30 degree bang angle limit. There was also a significant steady state error of 1.5 degrees. Gain tuning focused on increasing rise time and reducing steady state error.
94
-3.5 -3.6 -3.7 -3.8 -3.9 -4 -4.1 250 75 260 270 280 290 300 310 Time (sec) 320 330 340 350 360 370
Throttle (% Max)
70 65 60 55 250
260
270
280
290
300
320
330
340
350
360
370
Figure 49: AVL-HIL Airspeed Response with Default Gains Table 26: Airspeed Control Loop Performance with Default Gains
43.0 0.25
sec knots
The closed loop response showed very good steady state tracking but exhibits a highly over damped response. The elevator response is very slow, winding up type response, and there is little throttle change. Gain tuning will focus on increasing the
95
rise time of the default response. The altitude command loop performance to a 50 foot altitude change with default gains is presented in Figure 50 and the performance parameters are given in Table 27.
260 A ltitude A S L(ft) 250 240 230 220 210 200 190 185 190 195 200 205 Time (sec) 210 215 220
190
195
200
210
215
220
-4 E levator (deg) -4.1 -4.2 -4.3 -4.4 -4.5 185 190 195 200 205 Time (sec) 210 215 220
Rise Time ( t r ) Settling Time ( t s ) Maximum Overshoot ( M p ) Steady State Error ( ess )
sec sec % ft
96
The altitude response closed loop had underdamped behavior. The altitude response should be an overdamped response, as overshoot of altitude commands is to be avoided. Closed loop gain tuning focused on increasing damping and reducing steady state error, while trying to maintain good rise time performance.
97
Gain Parameter Roll Command Inner Loop Roll Error to Roll Rate Roll Rate Error to Aileron Roll Rate Error Integral to Aileron Yaw Inner Loop Yaw Rate Error to Rudder Side Force Error Integral to Rudder Heading Control Outer Loop Heading Error to Turn Rate Heading Error Derivative to Turn Rate Waypoint Navigation Tracker Convergence
Default Gain
1.00 0.00
0.50 0.00
0.40 0.10
0.50 0.00
0.35
0.25
Table 29: AVL-HIL Bank Angle Control Loop Performance with Simulator Tuned Gains
Rise Time ( t r ) Settling Time ( t s ) Maximum Overshoot ( M p ) Steady State Error ( ess )
98
12 Bank Angle (deg) 10 8 6 4 2 0 -2 782 783 784 785 Time (min) 786 787 788
783
784
786
787
788
0.5
0 782
783
784
786
787
788
Figure 51: AVL-HIL Bank Angle Control Response with Simulator Tuned Gains
The new closed loop bank angle response was much improved from the bank angle performance with default gains. The steady state error was reduced by half and the rise time was decreased significantly. This type of bank angle response should allow for rapid response to atmospheric turbulence and faster heading changes. The heading response with the new lateral gains is shown in Figure 52 and the new performance statistics are shown in Table 30.
99
242 Heading Angle (deg) 240 238 236 234 232 230 910 15 Bank Angle (deg/s) 10 5 0 -5 910 1.5 LHS Aileron (deg) 1 0.5 0 -0.5 910 911 912 913 914 915 Time (sec) 916 917 918 919 920
911
912
913
914
916
917
918
919
920
911
912
913
914
916
917
918
919
920
Figure 52: AVL-HIL Heading Angle Control Loop Performance with Simulator Tuned Gains
Table 30: Heading Angle Control Loop Performance with AVL-HIL Tuned Gains
4.8 0.4
sec deg
The heading controller was difficult to tune, as increasing the heading error to turn rate gain too much caused large steady state error and decreasing it too much resulted in high rise times. Strange bank angle auto command generation (large, step like bank angle commands) was also observed when this gain was too high. The Piccolo suffers some performance here because of the lack of a heading tracker integral. These problems are somewhat corrected when the outer waypoint tracking
100
loops are turned on, as the steady state error is driven to zero.
Gain Parameter Elevator Inner Loop Elevator Prediction Trust Z Acceleration Error Integral to Elevator Pitch Damper Pitch Error to Elevator Pitch Rate Error to Elevator Airspeed Control Outer Loop TAS Error to TAS Rate TAS Rate Error to Z Acceleration Throttle Inner Loop Throttle Prediction Trust Energy Rate Error Integral to Throttle Altitude Control Outer Loop Altitude Error to Altitude Rate Command Altitude Error to Z Acceleration Command
Default Gain
0.00 1.50
0.50 1.50
0.00 0.00
0.10 0.10
0.15 1.00
0.60 1.00
0.00 1.00
0.50 1.80
0.20 0.50
0.40 1.20
101
The elevator
prediction trust is the innermost gain that is placed on the scaled feedforward elevator controller. This gain had a significant impact on the amount and aggressiveness of elevator responses to outer-loop commands.
pred
= Kp
9.2
Without the elevator prediction trust all of the elevator inner-loop commands came from the Z acceleration integral to aileron gain. Pitch damping (both rate and angle) was added, although with very low gains. These may (and most likely will not) be needed in flight test as both the short period and Phugoid modes are more well damped than the AVL-Piccolo HIL simulations predict. The throttle prediction trust was also been raised from a zero value. This increased the amount of throttle commanded from outer loop inputs. Finally the altitude error to Z-acceleration command gain has been increased so that there is more elevator input (elevator changes were issued from acceleration commands) for precisions regulation of altitude. The final closed loop airspeed simulation response is given in Figure 53 and the performance characteristics are given in Table 32. The airspeed command rise time increased dramatically. There is slightly more steady state error, but it is still less than a knot which is acceptable.
102
77 76 75 74 73 72 71 70 69 1022
1024
1026
1030
1032
1034
1036
0.95 Elevator (deg) 0.9 0.85 0.8 0.75 0.7 1022 1024 1026 1028 Time (sec) 1030 1032 1034 1036
Throttle (% Max)
80 60 40 20 0 1022 1024 1026 1028 Time (sec) 1030 1032 1034 1036
Figure 53: AVL-HIL Airspeed Control Response with Simulator Tuned Gains
Table 32: AVL-HIL Airspeed Control Performance with Simulator Tuned Gains
Rise Time ( t r ) Settling Time ( t s ) Maximum Overshoot ( M p ) Steady State Error ( ess )
The altitude command response to the new gains is shown in Figure 54 and the performance characteristics are shown in Table 33.
103
260 250 Altitude ASL(ft) 240 230 220 210 200 190 590 600 610 620 Time (sec) 630 640 650
600
610
630
640
650
1.4 Elevator (deg) 1.2 1 0.8 0.6 0.4 590 600 610 620 Time (sec) 630 640 650
Figure 54: AVL-HIL Altitude Control Response with Simulator Tuned Gains
Table 33: AVL-HIL Altitude Control Performance with Simulator Tuned Gains
Rise Time ( t r ) Settling Time ( t s ) Maximum Overshoot ( M p ) Steady State Error ( ess )
sec sec % ft
As can be seen in Table 33 the rise time performance of the closed loop altitude controller was slightly slower than the default gains, however the settling time was greatly reduced. The maximum overshoot and steady state error, which can be dangerous in altitude controllers, was also reduced significantly.
104
maximizing gain performance is extremely difficult to perform unless the block diagram structure is known. The system at this point was ready for flight testing, but first a gain margin analysis was performed so that the flight engineer had a better understanding of which gains are the most critical with regards to stability. To
determine the gain margins of the lateral and directional systems each gain was increased from the tuned value until it drove the system unstable or it reached the maximum value possible. System stability was also checked when the gains were set to zero (i.e. control loop off). Table 34 and Table 35 present the gain margins for the longitudinal and lateral controllers. In nearly every case the there was substantial gain margin in the system. The lowest margin is in the roll rate error to aileron gain. Unpublished documentation was available to the team that showed that this gain mixed Euler and body axis roll angles. Driving the error between the Euler and body axis angles to zero can cause the system to go unstable as the angles should not be equal during maneuver, and thus it was decided that this term would be kept at zero. The instability of the system at a roll rate error to aileron gain of 1.25 reinforced this decision. The typical flight test increase or decrease in gain value is 20%. This would make the possibility of driving the system unstable low. It is also gives the flight test engineer some level of confidence that the system can be tested without undue risk to
105
Gain Parameter Roll Command Inner Loop Roll Error to Roll Rate Roll Rate Error to Aileron Roll Rate Error Integral to Aileron Yaw Inner Loop Yaw Rate Error to Rudder Side Force Error Integral to Rudder Heading Control Outer Loop Heading Error to Turn Rate Heading Error Derivative to Turn Rate Waypoint Navigation Tracker Convergence
Initial Gain
Destabilizing Gain
Yes No No
0.50 0.00
No No
0.50 0.00
N/A N/A
No No
0.25
600
No
technique would require significant engineering effort, as the CAN signal would have
106
had to have been hacked, and thus it was not performed. Thus no closed loop phase margins were determined prior to flight test.
Table 35: Piccolo II AVL-HIL Longitudinal Gain Margins
Gain Parameter Elevator Inner Loop Elevator Prediction Trust Z Acceleration Error Integral to Elevator Pitch Damper Pitch Error to Elevator Pitch Rate Error to Elevator Airspeed Control Outer Loop TAS Error to TAS Rate TAS Rate Error to Z Acceleration Throttle Inner Loop Throttle Prediction Trust Energy Rate Error Integral to Throttle Altitude Control Outer Loop Altitude Error to Altitude Rate Command Altitude Error to Z Acceleration Command
Initial Gain
Destabilizing Gain
0.50 2.50
N/A 50.00
No Yes
0.10 0.10
N/A 60.00
No No
0.60 1.40
N/A N/A
No No
0.50 1.80
N/A N/A
No Yes
0.40 1.20
9999999 N/A
No No
107
10
II avionics on the one-third scale Yak-54 using the initial gains set in Chapter 8. Three control loops were successfully tuned: the bank angle command loop, the heading angle command loop, and the airspeed command loop. Flight test planning and preparation were made according to the system described in Chapter 6.
108
when the aircraft was trimmed and then noted the trim control surfaces. These control surface positions were then be sent to the Piccolo II avionics unit. When the autopilot was activated all of the control surfaces that were not being commanded by the controller (i.e. elevator and throttle) were be set to the trim position. This allowed the aircraft to more or less maintain altitude and airspeed during the lateral-directional flight tests. Once the aircraft was trimmed and the Piccolo had received the trim control surface positions the flight test engineer, after confirming good telemetry, set the bank angle command to zero, with all other control loops off. This in effect regulated bank angle. Appendix O can be referenced for the bank angle control dance card. The initial response to this test is given in Figure 55. Figure 55 shows that getting the aircraft to an initial bank angle of zero can be difficult, thus the response was actually a 10 degree bank angle command. The initial transient was good, but the tracking was fairly poor, with steady state errors as high as 2 degrees. Before any gain tuning was performed however, a larger bank angle command was attempted. The results of this test are presented in Figure 56 and Table 36.
109
2 0 -2 -4 -6 -8 -10 1552 1553 1554 1555 1556 1557 1558 1559 1560
Time (sec)
2
1.5 1 0.5 0 -0.5 1552 1553 1554 1555 1556 1557 1558 1559 1560
Time (sec)
Figure 55: Initial Bank Angle Control Flight Test with AVL-HIL Tuned Gains
20 15
10 5 0 -5 -10 -15 -20 1598 2 1.5 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
Time (sec)
1 0.5 0 -0.5 -1 -1.5 1598 1599 1600 1601 1602 1603 1604 1605 1606 1607 1608
Time (sec)
Figure 56: 20 Degree Bank Angle Command Flight Test with Simulator Tuned Gains
110
Table 36: Roll Control Loop Performance with Simulator Tuned Gains
Rise Time ( t r ) Settling Time ( t s ) Maximum Overshoot ( M p ) Steady State Error ( ess )
sec sec % ft
The closed loop response shows a large amount of overshoot and poor regulation, with the response never settling. It was decided to decrease the roll error to roll rate gain by 20% to 4.0. This resulted in the response shown in Figure 57 and the performance parameters presented in Table 37.
Time (sec)
2
1768
1769
1770
1771
1772
1773
Time (sec)
Figure 57: Final Roll Control Flight Test with Tuned Gains
111
Rise Time ( t r ) Settling Time ( t s ) Maximum Overshoot ( M p ) Steady State Error ( ess )
As can be seen by the performance parameters, the closed loop bank angle response was much improved. The rise time was slightly worse than simulation, but the settling time was much better than the first flight test and the overshoot was much lower. It is difficult to determine what the actual steady state error was as the low test times made it difficult to observe periodic behavior that would indicate poor controller behavior as opposed to random disturbances. At this point the performance was acceptable and the flight team decided to move on to heading angle control testing.
engineer used this trim heading as the initial heading command. The trim heading functioned as the base heading for follow up heading changes (Appendix P) of 10 degrees, 20 degrees and 30 degrees. The initial heading response is shown in Figure 58.
112
10 9 8 7 6 5 4 3 2 1 0 536
537
538
539
540
541
542
537
538
539
540
541
542
Time (sec)
1
0 -1 -2 -3 536
537
538
539
540
541
542
Time (sec)
Figure 58: Initial Heading Command Flight Test with Simulator Tuned Gains
113
742
744
746
748
750
752
742
744
746
748
750
752
Time (sec)
Figure 59: Final Heading Angle Command Flight Test with Tuned Gains
Table 38: Final Heading Angle Control Performance with Tuned Gains
5.7 -1.6
sec deg
114
Gain Parameter Roll Command Inner Loop Roll Error to Roll Rate Roll Rate Error to Aileron Roll Rate Error Integral to Aileron Yaw Inner Loop Yaw Rate Error to Rudder Side Force Error Integral to Rudder Heading Control Outer Loop Heading Error to Turn Rate Heading Error Derivative to Turn Rate Waypoint Navigation Tracker Convergence
Default Gain
1.00 0.00
0.50 0.00
0.50 0.00
0.40 0.10
0.50 0.00
0.50 0.00
0.35
0.25
TBD
115
was set to OFF simulation has shown that the system would be driven unstable. The airspeed controller thus had to be iterated with the altitude controller to find the ideal configuration, giving more weight to altitude than airspeed as exact altitude tracking is more crucial to successful mission performance. Airspeed control testing was performed with the heading controller set to ON and a command value of 8 degrees. This prevented the aircraft from spiraling out of control during the test. The initial trim airspeed, and also the first command airspeed, was 70 knots at an altitude of 200 ft AGL (1300 ft ASL). Appendix Q can be referenced for the airspeed control flight test dance card. The results of the first airspeed flight test are given in Figure 60.
Indicated Air Speed (Knots)
74 72 70 68 66 333
334
335
336
337
338
339
340
341
Time (sec)
70
Throttle (%)
65 60 55 50 333
334
335
336
337
338
339
340
341
Time (sec)
-1.5
Elevator (deg)
334
335
336
337
338
339
340
341
Time (sec)
Figure 60: Initial Airspeed Regulator Flight Test with Simulator Tuned Gains
10.2
116
The airspeed regulator response was very good, with the error never getting above one knot and the steady state error settling out to about 0.3 knots. The next test was to command an airspeed change of 5 knots. The response to this command is given in Figure 61 and Table 40.
Indicated Air Speed (Knots)
80 75 70 65 428 429 430 431 432 433 434 435 436 437
Time (sec)
100
Throttle (%)
90 80 70 60 50 428 429 430 431 432 433 434 435 436 437
Time (sec)
-2.4
Elevator (deg)
-2.6 -2.8 -3 -3.2 -3.4 428 429 430 431 432 433 434 435 436 437
Time (sec)
Figure 61: Second Airspeed Command Flight Test Response with Simulator Tuned Gains Table 40: Second Airspeed Command Flight Test Performance with Simulator Tuned Gains
6.2 3.2
sec knots
The response to an approximately 5 knot airspeed change is slow and first order, with a fairly large steady state error. Gain tuning in flight focused on
improving the steady state response. Twenty-six iterations were performed with
117
different gain sets to get the final gains. Appendix R shows the gain change log sheet with the different gain value used. The final response to the gain values set in flight is given in Figure 62 with the performance parameters shown in Table 41.
Indicated Air Speed (Knots)
80
70
60
50 966
967
968
969
970
971
972
973
974
975
976
Time (sec)
80
Throttle (%)
60 40 20 0 966
967
968
969
970
971
972
973
974
975
976
Time (sec)
-2.5
Elevator (deg)
-3
-3.5
-4 966
967
968
969
970
971
972
973
974
975
976
Time (sec)
Figure 62: Final Airspeed Command Flight Test Response with Tuned Gains Table 41: Final Airspeed Command Response Performance with Tuned Gains
5.4 -1.2
sec knots
Figure 62 shows, the closed loop response has increased rise time and reduced steady state error over the response to the initial gains. The response is actually quite similar to the response seen in simulation (Figure 53). The airspeed controller is now working in an acceptable manner and the altitude controller can be tuned.
118
Table 42 shows the new gains after the airspeed controller tuning.
Table 42: Final Longitudinal Gains
Gain Parameter Elevator Inner Loop Elevator Prediction Trust Z Acceleration Error Integral to Elevator Pitch Damper Pitch Error to Elevator Pitch Rate Error to Elevator Airspeed Control Outer Loop TAS Error to TAS Rate TAS Rate Error to Z Acceleration Throttle Inner Loop Throttle Prediction Trust Energy Rate Error Integral to Throttle Altitude Control Outer Loop Altitude Error to Altitude Rate Command Altitude Error to Z Acceleration Command
Default Gain
0.00 1.50
0.50 1.50
0.30 1.65
0.00 0.00
0.10 0.10
0.10 0.10
0.15 1.00
0.60 1.00
0.70 1.00
0.00 1.00
0.50 1.80
0.50 1.80
0.20 0.50
0.40 1.20
TBD TBD
119
test runs were made to get the gains to the final values shown in Table 42. This would be very difficult to perform on the Meridian UAS, given the difficulty in maintaining manual control. Flight test is an expensive, time-consuming and
inefficient arena for gain tuning as the airspeed test shows. This would reinforce the conclusion of Chapter 8 that a hardware-in-loop simulation model be developed that will produce high fidelity closed loop simulations, thus minimizing any in-flight gain tuning.
120
Figure 63: RSSI Levels during First Flight Pilot Control Loss
The drop out lasted approximately 2.5 seconds, with the RSSI changing from -71 dB to -108 dB. Because the RSSI drop was temporary (the RSSI at most other
121
times appeared to be satisfactory) and the pilot was able to successfully regain manual control it was decided that the flight test would continue.
-70 -75 -80 -85 RSSI (dBm) -90 -95 -100 -105 -110 -115 -120 0 100 200 300 400 500 600 Time (sec) 700 800 900 1000
122
-75 -80 -85 -90 -95 -100 -105 729 730 731 Time (sec) 732 733 734
RSSI (dBm)
Figure 65: RSSI during Second Flight Test Pilot Control Loss
-70 -75 -80 -85 RSSI (dBm) -90 -95 -100 -105 -110 -115 -120 0 200 400 600 Time (sec) 800 1000 1200
The RSSI data on the second flight was similar to that of the first flight, with one large drop to -108 dB with more than 2 seconds of missing data in-between.
123
RSSI (dBm)
523
524
525
528
529
530
As in the second flight the flight test proceeded as planned despite the RSSI drops causing brief losses of pilot control. During final approach communication was lost again with the Yak-54. After communication with the ground station timed out the Piccolo activated the autopilot. As the aircraft was immediately commanded by the autopilot to accelerate to 70 knots the engine went to full power and the elevator deflected positively. This caused the aircraft to impact the runway. Figure 68 shows the RSSI drop during final approach that caused the autopilot activation and Figure 69 shows the engine RPM and throttle position at crash. The RPM is zero when communication is reacquired after the crash.
124
-75 -80 -85 -90 -95 -100 -105 1095 1095.5 1096 1096.5 1097 1097.5 1098 Time (sec) 1098.5 1099 1099.5 1100
RSSI (dBm)
3.5
2.5 1095
1095.5
1096
1096.5
1097
1098.5
1099
1099.5
1100
Engine RPM
1095.5
1096
1096.5
1097
1098.5
1099
1099.5
1100
Figure 69 shows that at no point was the increase in RPM that was heard by the ground crew recorded by the ground station.
125
analysis method, developed by Edmond Leong, computes the time between the current data packet and the last data packet and this value was displayed graphically. Figure 70 shows a plot of RSSI using this method. The top plot on Figure 70 shows the previous method of analyzing RSSI. This was the RSSI plot for the taxi test before the March 13th flight test. This plot to the flight test team indicated excellent signal integrity, with the signal only dropping below the maximum level of -71 dB for a few moments during taxi, and then only to 86 dB. The bottom plot however showed a different story, with data drop outs higher than 1.5 seconds, even when RSSI is at maximum. It was seen in the bottom plot of Figure 70 that there appears to be a random autopilot ON instance at about 500 seconds into ground testing. Figure 71 shows a zoomed in view of this point.
126
RSSI Signal - Flight Test Date : Thu Mar 13 2008; Time : 08-09-13
RSSI (dBm)
-90 -95 -100 -105 -110 -115 -120 0 200 400 600 800 1000 1200 1400 1600
Time (sec)
2.5
1.5
0.5
200
400
600
800
1000
1200
1400
1600
Time (sec)
Figure 70: RSSI and Time between Packets during Ground Testing
Figure 71 shows that the autopilot has activated for about 0.5 seconds. This autopilot ON action was not done manually, but automatically. The only way it could happen automatically is if the pilot timeout limit was reach. The pilot timeout is defined as when uplink data (i.e. pilot commands) are no longer being received by the avionics. The timeout limit was set to 0.2 seconds. Thus the loss of pilot control
was at least 0.9 seconds, as the autopilot ON telemetry was received for 0.7 seconds. There was an additional 0.7 seconds of lost telemetry where the autopilot may or may not have been activated.
127
RSSI Signal - Flight Test Date : Thu Mar 13 2008; Time : 08-09-13
RSSI (dBm)
-90 -95 -100 -105 -110 -115 -120 504 504.5 505 505.5 506 506.5 507 507.5 508
Time (sec)
2.5
1.5
0.5
0 504
504.5
505
505.5
506
506.5
507
507.5
508
Time (sec)
After these revelations of pilot timeout were made, a statistical analysis of the flight test was made.
RSSI Signals Distribution (%) - (2.4GHz at 20Hz Update Rate)
100 90 Signals Distribution (%) 80 70 60 50 40 30 20 10 0
Taxi Test 1st Flight 2nd Flight 3rd Flight -71 96.5844 54.3841 54.1862 56.8031 -79 0.6489 38.3038 36.7397 32.3086 -86 0 2.1347 2.5871 3.5631 -93 0 0.3758 2.0748 2.0675 -101 0 0 0 0 -108 0 0.4645 0.1507 0.58 -115 0.0035 0.0731 0.0043 0.0083 Signal Lost 2.7631 4.2641 4.2572 4.6694
Figure 72: RSSI Signal Strength Levels and Percentage of Data Lost
128
Figure 72 shows there was a significant percentage of a data drop during the flight test, during flight averaging about 4.5%. Figure 73 shows the drop out time distribution for the flight test.
Signal Drop Out Statistics - (2.4GHz at 20Hz Update Rate)
800 Number of Signal Drop Outs 700 600 500 400 300 200 100 0
Taxi Test 1st Flight 2nd Flight 3rd Flight 0.5 432 476 732 592 0.5~1.0 21 14 7 16 1.0~1.5 16 5 6 12 1.5~2.0 2 2 0 1 2.0~2.5 0 1 1 0 2.5~.3.0 0 0 0 0 > 3.0 0 0 0 2
As can be seen in Figure 73 the communication dropouts last for significant amounts of time, some as long as 3.0 seconds. An analysis of the previous flight tests was also performed to examine if this phenomena was limited to the March 13th, 2008 flight test. Figure 74 shows the average dropout time per minute for all the previous Yak-54 Piccolo flight tests.
129
Comparison of Average Signal Drop Out (sec/min) on Different Flight Test Dates (2.4GHz at 20Hz Rate)
4 Average Drop Out Time (sec/min) 3.5 3 2.5 2 1.5 1 0.5 0
Mar-13-08 Feb-29-08 Dec-20-07 Aug-31-07 Jun-19-07 Taxi Test 1.68 2.88 1.72 1st Flight 2.57 2.82 3.59 1.1 4.34 2.94 3.04 2nd Flight 2.58 3rd Flight 2.82
Test Sequence
Figure 74 reveals that the average dropout per minute was actually lower for the March 13th, 2008 flight test than all other flight tests save the August 31st, 2007 flight test.
130
6. Ground Station Networking: ON vs. OFF 7. Ground Station Power: Battery Powered vs. Wall Power 8. Transmission Frequency: 2.4 GHz vs. 900 MHz The first lab test, 20 Hz vs. 1 Hz yielded some interesting results. Figure 75 shows the signal loss percentages for the test.
RSSI Signals Distribution (%) LAB Test on Piccolo 1024 & 1027 - 2.4 GHz
Signal Distribution (%) 100 90 80 70 60 50 40 30 20 10 0
1027 20Hz 1024 20Hz 1027 1Hz 1024 1Hz -71 96.0481 96.165 99.6143 100 -79 0 0 0 0 -86 0 0 0 0 -93 0 0 0 0 -101 0 0 0 0 -108 0 0 0.3857 0 -115 0 0 0 0 Signal Lost 3.9519 3.835 0 0
Figure 75: RSSI Signal Strength and Percentage of Data Loss during Laboratory Test
The lab test shows that there was only a small difference in the performance of Piccolo 1024 and 1027 at both 20 Hz and 1 Hz. The lab test showed that there were no dropouts during the 1 Hz test for either Piccolo unit. There was however a small percentage of the time where the signal strength dropped to -108 dB on Piccolo 1027. A closer look at the data revealed some interesting data, which is presented in Figure 76.
131
RSSI Signal - Lab Test on Piccolo 1027 : Tue Mar 18 2008; Time : 12-38-21
-70 -80 RSSI (dBm) -90 -100 -110 -120 554 556 558 560 562 Time (sec) 564 566 568
4 3 2 1 0 554
556
558
562
564
566
568
Figure 76: RSSI and Packet Time at Autopilot Activation during Lab Test
Figure 76 clearly shows that the autopilot was activated for at least 1 second. This could only have occurred due to the pilot timeout limit having been reached and thus that the autopilot can activate even in situations where no downlink data is missing. The second laboratory test also revealed some very interesting information. The second test tested the system autopilot ON and OFF at 1 Hz and 20 Hz. The results are presented in Figure 77.
132
RSSI Signals Distribution (%) LAB Test on Piccolo 1024 & 1027 - 2.4 GHz
Signal Distribution (%) 100 90 80 70 60 50 40 30 20 10 0
1027 20Hz 1024 20Hz 1027 1Hz 1024 1Hz -71 99.9957 99.995 99.82 1000 -79 0 0 0 0 -86 0 0 0 0 -93 0 0 0 0 -101 0 0 0 0 -108 0 0 0.18 0 -115 0 0 0 0 Signal Lost 0.0043 0.005 0 0
Figure 77: RSSI Signal Strength and Percentage of Data Loss during Autopilot ON Lab Test
Figure 77 shows that on autopilot mode ON there are virtually no dropouts. It should be noted that when the autopilot is on the only uplink data being sent to the Piccolo is the command to the Piccolo to send telemetry information to the ground station. The other tests showed little statistical difference between the flight test data or the original laboratory manual tests at 20 Hz. Figure 78 shows the results of the other tests. Figure 78 does show another instance of -108 dB RSSI during the Piccolo 1027 bench power test. activation. A closer look at the data once again reveals autopilot
133
RSSI Signals Distribution (%) LAB Test on Piccolo 1024 & 1027 - 20Hz at 2.4GHz
100 Signal Distribution (%) 80 60 40 20 0
1027 No Networking 1024 No Networking 1027 No Servo Power 1024 No Servo Power 1027 Bench Power 1024 Bench Power 1027 GS Wall Power -71 96.1099 96.3423 95.5892 96.3691 96.6772 96.4635 95.6249 -79 0 0 0 0 0 0 0 -86 0 0 0 0 0 0 0 -93 0 0 0 0 0 0 0 -101 0 0 0 0 0 0 0 -108 0 0 0 0 0.2605 0 0 -115 0 0 0 0 0 0 0 Signal Lost 3.8901 3.6577 4.4108 3.6309 3.0623 3.5365 4.3751
1027 No Networking 1024 No Networking 1027 No Servo Power 1024 No Servo Power 1027 Bench Power 1024 Bench Power 1027 GS Wall Power
Figure 78: RSSI Levels and Data Dropouts during Various Lab Tests
RSSI Signal - Lab Test on Piccolo 1027 : Tue Mar 18 2008; Time : 13-51-27
-70 -80 RSSI (dBm) -90 -100 -110 -120 201 202 203 204 Time (sec) 2 1.5 1 0.5 0 201 205 206 207 208
202
203
205
206
207
208
Figure 80 shows that there is almost no difference in the different test scenarios except autopilot ON.
134
Comparsion of Average Signal Drop Out (sec/min) on Different Lab Tests Setup
3 Average Drop Out Time (sec/min) 2.5 2 1.5 1 0.5 0
Manual 2.4GHZ 1027 20Hz 1024 20Hz 1027 1HZ 1024 1HZ 2.38 2.31 0 0 Manual Radio Switch 1.93 2.57 0 0 Piccolo Bench Power 1.85 2.14
Test Sequence
The pilot timeout that occurred during the bench power test prompted an investigation of how many times during all testing (lab and flight) the autopilot activated. Figure 81 shows the number of times the autopilot activated due to the pilot communications timeout.
20 18 Times of AutoPilot ON due to pilot comm out 16 14 12 10 8 6 4 2 0 Jun/Aug - 07 Dec-20-07 Feb-29-08 Mar-13-08 Lab Test Lab Test (Piccolo 1024) (Piccolo 1027) (Piccolo 1027) (Piccolo 1027) (Piccolo 1024) (Piccolo 1027) Test Sequence
Figure 81: Autopilot Activation during Flight and Lab Test Activities
135
Figure 81 clearly shows that a significant number of times during flight test the autopilot activated on its own. It should be noted that during the Jun/Aug 2007 flight tests the pilot comm. timeout was set to 2.0 seconds. It should also be noted that these numbers only show the times when the autopilot activation was detected in telemetry. All of the pilot reported activations occurred did not show up in the telemetry and are thus not included in the number of activations shown in the figure. The final test to be performed was to switch from 2.4 GHz to 900 MHz data transfer frequency. Figure 82 shows once again that the same random data drop occurs at this frequency.
RSSI Signal - Redo Lab Test on Piccolo 1027 : Thu Mar 20 2008; Time : 10-19-12
-70 -80 RSSI (dBm) -90 -100 -110 -120 0 200 400 600 Time (sec) 2.5 delta time (sec) 2 800 1000 1200 1400
1.5 1
0.5 0 0 200 400 600 Time (sec) 800 1000 1200 1400
Figure 82: RSSI and Data Dropout Times during 900 MHz Lab Tests
136
20 10 0 -10 24 0 24.1
24.2
24.4
24.5
24.6
Elevator (deg)
-5 -10 -15 24
Figure 83: Poor Pilot-in-loop Pitch Stability Due to Time-Lag and Data Drop
that extreme maneuvering and flight path angles were possible during aborted landings, thus preventing a crash. Using Roskam (Reference [36]) the effect that
137
these discrete data drops had on the system was analyzed. Roskam used the pilot transfer function approximation of:
Yp (s) = K p e s (Tlead s + 1) (Tlag s + 1)(Tn s + 1)
11.1
The e s term represents the pilot reaction time delay. A pilot is not an instantaneous controller and will have come delay while his senses perceive what is going on. Roskam stated that the typical pilot time delay is 0.12 to 0.2 seconds. Given that a remote pilot cannot perceive things as quickly as a manned pilot, as he only has his sense of vision to rely on, it was reasonable to say that the pilot time delay may be double to 0.24 to 0.4 seconds. As mentioned previously the pilot console was sampled by the ground station at 10 Hz, thus there can be up to a 0.1 second additional delay in the system. Thus the new pilot time lag was between 0.34 to 0.5 seconds. It has also been shown in lab tests that there are clear documented cases of between 0.2 seconds and 3 second data drops. This means that the new pilot lag time was between 0.34 seconds best case and 3.5 seconds worst case. Roskam states that a drunken pilot has a time delay of 0.6 seconds. The analysis shows that the pilot in fact will drift in and out of the stability threshold of the system. The pilot does have the ability to lead known faults in the system, like the 0.1 second time delay. He does not, however, have the ability to lead unexpected lags in the system. The stability of the Yak-54 was improved by the addition of faster servos and improved damping of the short period mode, but clearly the aircraft was still operating in a near unstable or unstable flight regime.
138
139
A thorough and deliberate test and evaluation process for modeling, simulation and flight control systems has been developed
The AAA dynamic model predicts a response that is closer to flight test than the SCCS and AVL-HIL models
A good simulation model allows the flight control engineer to tune the system much more efficiently than in flight test
The Piccolo II flight control systems autonomous performance (from a stability and control standpoint) up to the flight incident was promising and could only be said to be positive
Gain tuning during flight test is difficult, time consuming and expensive and would be very difficult to perform during flight test of the Meridian
An efficient and safe flight test program was set up for flight testing the Piccolo II flight control system that could easily be adapted to Meridian flight testing, regardless of the flight control system used.
Pilot induced oscillations occur due to the normal 0.1 second time lag in the system. The oscillations can be overcome somewhat through pilot training.
Poor communication due to system overload was the cause of the crash.
140
Discrete drops of pilot data cause the Piccolo II flight control system to go pilot-in-loop unstable at random points during flight causing extreme hazards to the safety of flight.
The Piccolo II flight control system should not be used on the Meridian. Real time data plotting would greatly enhance the ability of flight test engineers to tune the flight control system and monitor the health of the system and should be implemented on any flight test program.
The flight test program developed should be duplicated for any other flight control system test program performed by CReSIS.
Wind tunnel testing, system identification, CFD or other methods of enhancing dynamic models should be performed on the Meridian to maximize closed loop performance and thus optimize the Meridians ability to perform the science mission.
Evaluation of the communication system and its impact on safety of flight should be given higher priority in future flight control system evaluations so that problems might be identified sooner.
141
A 6 DoF, nonlinear, hardware-in-loop simulator should be developed. System identification flight tests should be developed and flown on the Yak54 to increase the dynamic model fidelity.
The validity of AAA models of small scale aircraft should be investigated, ideally using system identification.
The nonlinear effects of large flight regimes and the subsequent effect on controller tuning should be investigated
A better engine thrust and power model should be developed. A secondary manual control system should be developed for further Piccolo II controller flight testing.
Develop a good way to estimate phase margins with the Piccolo II An automatic take-off and landing system should be developed.
142
13 References
1. L. Berstien, P Bosch, et. al., Climate Change 2007: Synthesis Report. Summary for Policy Makers, Intergovernmental Panel on Climate Change [online publication] URL: http://www.ipcc.ch/pdf/assessmentreport/ar4/syr/ar4_syr_spm.pdf [cited 12 April 2008]. R. Hale, W. Donovan, M. Ewing, et. al., The Meridian UAS: Detailed Design Review, Technical Report 124, Center for Remote Sensing of Ice Sheets, University of Kansas, Lawrence, KS, 25 June 2007. K. Siegle, Avionics System for the Center for Remote Sensing of Ice Sheets Meridian UAV, Masters Project Paper, University of Kansas, Lawrence, KS, 1 December 2007. D. Borys and R. Colgren, Advances in Intelligent Autopilot Systems for Unmanned Aerial Vehicles, AIAA Guidance, Navigation and Control Conference, San Francisco, CA, 15-18 August 2005. D. Jung and P. Tsiotras, Modeling and Hardware-in-the-Loop Simulation for a Small Unmanned Aerial Vehicle, AIAA Infotech Aerospace Conference, Rohnert Park, CA, 7-10 May 2007. M. Sadraey and R. Colgren, UAV Flight Simulation: Credibility of Linear Decoupled vs. Nonlinear Coupled Equations of Motion, AIAA Modeling and Simulations Technology Conference, San Francisco, CA, 15-18 August 2005. N. Jodeh, P. Blue and A. Waldron, Development of Small Unmanned Aerial Vehicle Research Platform: Modeling and Simulation with Flight Test Validation, AIAA Modeling and Simulation Technologies Conference, Keystone, CO, 21-24 August 2006. U. Ly and S. Higashino, Development of a UAV Flight-Test Vehicle at the University of Washington, 2nd AIAA Unmanned Unlimited Systems, Technologies and Operations Aerospace Conference, San Diego, CA 15-18 September 2003. J. Roskam, Airplane Flight Dynamics and Automatic Flight Controls: Part I, Fourth Printing, DARcorporation, Lawrence, KS, 2003. B. Etkin and L. Ried, Dynamics of Flight: Stability and Control, Third Addition, John Wiley & Sons, Hoboken, NJ, 1996.
2.
3.
4.
5.
6.
7.
8.
9. 10.
143
B. Pamadi, Performance, Stability, Dynamics and Control of Airplanes, AIAA, Reston, VA, 1998. B. Stevens and F. Lewis, Aircraft Control and Simulation, Second Addition, John Wiley & Sons, Hoboken, NJ, 2003. W. Phillips, Mechanics of Flight, John Wiley & Sons, Hoboken, NJ, 2004. D. Ward and T. Strganac, Introduction to Flight Test Engineering, Second Edition, Second Printing, Kendall/Hunt Publishing, U.S.A., 2001. R. Kimberlin, Flight Testing of Fixed-Wing Aircraft, AIAA, Reston, VA, 2003. R. Jategaonkar, Flight Vehicle System Identification: A Time Domain Methodology, AIAA, Reston, VA 2006. P. Hamel and R. Jategaonkar, Evolution of Flight Vehicle System Identification, AIAA Journal of Aircraft, Vol. 33, No. 1, January-February 1996. Y. Lee, S. Kim, J. Suk, H. Koo and J. Kim, System Identification of an Unmanned Aerial Vehicle from Automated Flight Tests, AIAAs 1st Technical Conference and Workshop on Unmanned Aerial Vehicles, Portsmouth, VA, 20-23 May 2002. M. Tischler and R. Remple, Aircraft and Rotorcraft System Identification: Engineering Methods with Flight Test Examples, AIAA, Reston, VA, 2006. S. Bahndari and R. Colgren, 6-DoF Dynamic Model for a Raptor 50 UAV Helicopter Including Stabilizer Bar Dynamics, AIAA Modeling and Simulation Technologies Conference, Keystone, CA, 21-24 August 2006. C. Theodore, M. Tischler and J. Colbourne, Rapid Frequency-Domain Modeling Methods for Unmanned Aerial Vehicle Flight Control Applications, AIAA Journal of Aircraft, Vol. 41, No. 4, July-August 2004. K. Ogata, Modern Control Engineering, Fourth Edition, Prentice Hall, Upper Saddle River, NJ, 2002. M. Tischler, J. Lee and J. Colbourne, Comparison of Flight Control System Design Using the CONDUIT Design Tool, AIAA Journal of Guidance, Control and Dynamics, Vol. 25, No. 3, May-June 2002.
18.
19. 20.
21.
22. 23.
144
24.
Y. Miyazawa and T. Motoda, Stochastic Gain Tuning Method Applied to Unmanned Space Vehicle Flight Control Design, AIAA Guidance, Navigation and Control Conference, Portland, OR, 9-11 August 1999. V. Askue, Flight Testing Homebuilt Aircarft, Iowa State University Press, Ames Iowa, 1992. B. Vaglienti, R. Hoag and M. Niculescu, Piccolo System Users Guide, Version 2.0.5, Cloud Cap Technology, Hood River, OR, 25 June 2007. Raymar Information Solutions Web Page, URL: http://www.raymarinc.com/images2/go%20book%203%20open.jpg, [cited 30 March 30th 2008]. B. Vaglienti, Setup and Tuning of Piccolo Control Laws 2.0.x, Cloud Cap Technology, Hood River, OR, 20 June 2007. B. Vaglienti, M. Niculescu and J. Becker HIL/SIL Simulator for the Piccolo Avionics, Cloud Cap Technology, Hood River OR, 5 April 2007. J. Becker, Creating Vortex Lattice Models for the Piccolo Simulator with AVL, Cloud Cap Technology, Hood River, OR, 18 April 2007. Aeroworks Web Page, URL: http://www.aeroworks.net/store/images/upload/plane_large/75cc_yak03.jpg, [cited 16 March 2008]. S. Keshmiri, E. Leong, R. Jager and R. Hale, Modeling and Simulation of the Yak-54 Scaled Unmanned Aerial Vehicle Using Parameter and System Identification,, accepted to the proceedings of the AIAA Atmospheric Flight Mechanics Conference and Exhibit, 18 21 Aug 2008, Honolulu, Hawaii. E. Leong, R. Jager, S. Keshmiri and R. Colgren, Complete Open Loop Dynamics Evaluation of a Small Unmanned Aerial Vehicle Using Different Modeling and Simulation Platforms with Flight Test Validation, accepted to the proceedings of the AIAA Atmospheric Flight Mechanics Conference and Exhibit, 18-21 Aug 2008, Honolulu, Hawaii. 3W Modelmotoren Web Page, URL: http://www.3wmodellmotoren.com/english/www_3W_Modellmotoren_com.html, [cited 21 March 2008]. Electrodynamics Web Page, URL: http://electrodynam.com/rc/EDR107/images/107s.jpg, [cited 25 March, 2008].
32.
33.
34.
35.
145
36.
J. Roskam, Airplane Flight Dynamics and Automatic Flight Controls: Part II, Third Printing, DARcorporation, Lawrence, KS, 2003.
146
147
Left_Aileron_Outboard=-1.1862 // Average aileron chord, in m Left_Aileron_Chord=0.1247 // Channel number Left_Aileron_Channel=0 //Right aileron // Spanwise location of inboard station, in m Right_Aileron_Inboard=0.1778 // Spanwise location of outboard station, in m Right_Aileron_Outboard=1.1862 // Average aileron chord, in m Right_Aileron_Chord=0.1247 // Channel number Right_Aileron_Channel=5 //-------------------- AERODYNAMICS: HORIZONTAL TAIL --------------------// // Reference area, in m^2 Tail_Area=0.21368 // Span, in m Tail_Span=0.9144 // Taper ratio Tail_Taper=0.81 //Tail Aerodynamics Tail_LUT=Yak_tail.lut // Location of tail ac wrt to aircraft cg, in m Tail_X=-1.0762 Tail_Z=-0.0127 // Oswald efficiency factor Tail_Span_Efficiency=1.0 // Incidence angle wrt to fuselage center line, in deg
148
Tail_Incidence=0 // Dihedral angle, in deg Tail_Dihedral=0 // Sweep angle of the quarter chord line, in deg Tail_Sweep=12.6 // HORIZONTAL TAIL CONTROL SURFACES // (Elevators move in unison, controlled by the same servo channel) // Left elevator // Spanwise location of the inboard station, in m Left_Elevator_Inboard=0 // Spanwise location of the outboard station, in m Left_Elevator_Outboard=-0.4572 // Average chord, in m Left_Elevator_Chord=0.10668 // Channel number Left_Elevator_Channel=1 // Sense of rotation Left_Elevator_Sign=1 // Right elevator // Spanwise location of the inboard station, in m Right_Elevator_Inboard=0 // Spanwise location of the outboard station, in m Right_Elevator_Outboard=0.4572 // Average chord, in m Right_Elevator_Chord=0.10668 // Channel number Right_Elevator_Channel=1 // Sense of rotation
149
Right_Elevator_Sign=1 //-------------------- AERODYNAMICS: VERTICAL TAIL --------------------// // Reference area, in m^2 Left_Fin_Area=0.1486 // Span, in m Left_Fin_Span=0.4318 // Taper ratio Left_Fin_Taper=0.35 //Fin Aerodynamics Left_Fin_LUT = Yak_fin.lut // Oswald efficiency factor Left_Fin_Span_Efficiency=1.0 // Location of vertical tail ac wrt to aircraft cg, in m Left_Fin_X=-1.12649 Left_Fin_Z=-0.231902 Left_Fin_Y=0.0 // VERTICAL TAIL CONTROL SURFACES // Rudder // Position of bottom section wrt to aircraft cg, in m Left_Rudder_Bottom=0.016 // Position of top section, wrt to aircraft cg, in m Left_Rudder_Top=-0.4478 // Average chord, in m Left_Rudder_Chord=0.2159 // Channel number Left_Rudder_Channel=3 // Sense of rotation Left_Rudder_Sign=1 //-------------------- AERODYNAMICS: FUSELAGE --------------------//
150
// Reference area, in m^2 Fuse_Area=0.05587 // Total length, in m Fuse_Length=2.0574 // Parasitic drag Fuse_Parasitic_Drag=0.013 // Slope of lift coefficient Fuse_Lift_Slope=0.0 // Slope of side force coefficient Fuse_SideForce_Slope=0.0 // Slope of pitch moment coefficient Fuse_Pitching_Moment_Slope=0 // Slope of yaw moment coefficient Fuse_Yawing_Moment_Slope=0 // Position of fuselage ac wrt to aircraft cg, in m Fuse_X=0.0 Fuse_Y=0.0
//-------------------- PROPULSION --------------------// // Engine is 3W 75cc, 7.5 hp //Engine Type is Piston Left_Engine_Type=0 // Channel number Left_Engine_Channel=2 // Engine parameters look-up table Left_Engine_LUT=Fuji86ccEngine.lut // Propeller is APC 24x10 // Prop diameter, in m
151
Left_Prop_Diameter=0.60696 // Position of propeller hub wrt to aircraft cg, in m Left_Prop_X=0.5715 //Propellor Pan Angle Left_Prop_Pan=3 // Moment of inertia in kg/m^2 Left_Prop_Inertia=0.0025 // Propeller coefficients look-up table Left_Prop_LUT=apc24x10.prd //-----------------Inertia Model--------------// //Gross Mass in Kg Gross_Mass=12.755 //Empty Mass in Kg Empty_Mass=12.33 //Moments of Inertia Roll_Inertia=1.30592385 Pitch_Inertia=3.92075434 Yaw_Inertia=5.15970078 Roll_Yaw_Coulpled_Inertia = 0
152
-1.072068493 0.086106841 -1.00059726 0.076323293 -0.929126027 0.067214472 -0.857654794 0.058780379 -0.786183562 0.051021013 -0.714712329 0.043936374 -0.643241096 0.037526463 -0.571769863 0.031791279 -0.50029863 0.026730823 -0.428827397 0.022345095 -0.357356164 0.018634093 -0.285884931 0.01559782 -0.214413699 0.013236274 -0.142942466 0.011549455 -0.071471233 0.010537364 0.0000000000 0.010200000 0.071471233 0.010537364 0.142942466 0.011549455 0.214413699 0.013236274 0.285884931 0.01559782 0.357356164 0.018634093 0.428827397 0.022345095 0.50029863 0.026730823 0.571769863 0.031791279 0.643241096 0.037526463 0.714712329 0.043936374 0.786183562 0.051021013 0.857654794 0.058780379 0.929126027 0.067214472 1.00059726 0.076323293 1.072068493 0.086106841
153
CL CD CM -0.952949772 0.076371255 -0.889419787 0.066850071 -0.825889802 0.05798552 -0.762359817 0.049777603 -0.698829832 0.042226319 -0.635299848 0.035331669 -0.571769863 0.029093652 -0.508239878 0.023512268 -0.444709893 0.018587518 -0.381179909 0.014319401 -0.317649924 0.010707917 -0.254119939 0.007753067 -0.190589954 0.00545485 -0.12705997 0.003813267 -0.063529985 0.002828317 0.0000000000 0.002500000 0.063529985 0.002828317 0.12705997 0.003813267 0.190589954 0.00545485 0.254119939 0.007753067 0.317649924 0.010707917 0.381179909 0.014319401 0.444709893 0.018587518 0.508239878 0.023512268 0.571769863 0.029093652 0.635299848 0.035331669 0.698829832 0.042226319 0.762359817 0.049777603 0.825889802 0.05798552 0.889419787 0.066850071 0.952949772 0.076371255
154
-0.952949772 0.075571255 -0.889419787 0.066050071 -0.825889802 0.05718552 -0.762359817 0.048977603 -0.698829832 0.041426319 -0.635299848 0.034531669 -0.571769863 0.028293652 -0.508239878 0.022712268 -0.444709893 0.017787518 -0.381179909 0.013519401 -0.317649924 0.009907917 -0.254119939 0.006953067 -0.190589954 0.00465485 -0.12705997 0.003013267 -0.063529985 0.002028317 0.0000000000 0.001700000 0.063529985 0.002028317 0.12705997 0.003013267 0.190589954 0.00465485 0.254119939 0.006953067 0.317649924 0.009907917 0.381179909 0.013519401 0.444709893 0.017787518 0.508239878 0.022712268 0.571769863 0.028293652 0.635299848 0.034531669 0.698829832 0.041426319 0.762359817 0.048977603 0.825889802 0.05718552 0.889419787 0.066050071 0.952949772 0.075571255
155
156
157
0.91 0.92 0.93 0.94 0.95 0.96 0.97 0.98 0.99 1.00 1.01 1.02 1.03 1.04 1.05 1.06 1.07 1.08 1.09 1.10 1.11 1.12 1.13 1.14 1.15 1.16 1.17 1.18 1.19 1.20 1.21 1.22 1.23 1.24 1.25 1.26 1.27 1.28 1.29 1.30 1.31 1.32 1.33 1.34
0.0353 0.0327 0.0344 0.0314 0.0335 0.0302 0.0325 0.0290 0.0316 0.0277 0.0306 0.0265 0.0296 0.0253 0.0285 0.0240 0.0274 0.0228 0.0264 0.0215 0.0252 0.0203 0.0241 0.0190 0.0229 0.0177 0.0218 0.0165 0.0206 0.0152 0.0193 0.0139 0.0181 0.0127 0.0168 0.0114 0.0155 0.0101 0.0142 0.0088 0.0128 0.0076 0.0115 0.0063 0.0101 0.0050 0.0086 0.0037 0.0072 0.0024 0.0057 0.0010 0.0046 -0.0002 0.0032 -0.0015 0.0019 -0.0028 0.0005 -0.0041 -0.0008 -0.0053 -0.0022 -0.0066 -0.0035 -0.0079 -0.0049 -0.0092 -0.0062 -0.0105 -0.0076 -0.0118 -0.0089 -0.0131 -0.0103 -0.0144 -0.0116 -0.0156 -0.0130 -0.0169 -0.0143 -0.0182 -0.0157 -0.0195 -0.0170 -0.0208 -0.0184 -0.0221
158
1.35 1.36 1.37 1.38 1.39 1.40 1.41 1.42 1.43 1.44 1.45 1.46 1.47 1.48 1.49 1.50 1.51 1.52 1.53 1.54 1.55 1.56 1.57 1.58 1.59 1.60 1.61 1.62 1.63 1.64 1.65 1.66 1.67 1.68 1.69 1.70 1.71 1.72 1.73 1.74 1.75 1.76 1.77 1.78
-0.0197 -0.0234 -0.0211 -0.0247 -0.0224 -0.0259 -0.0238 -0.0272 -0.0252 -0.0285 -0.0265 -0.0298 -0.0279 -0.0311 -0.0292 -0.0324 -0.0306 -0.0337 -0.0319 -0.0350 -0.0333 -0.0362 -0.0346 -0.0375 -0.0360 -0.0388 -0.0373 -0.0401 -0.0387 -0.0414 -0.0400 -0.0427 -0.0414 -0.0440 -0.0427 -0.0453 -0.0441 -0.0465 -0.0454 -0.0478 -0.0468 -0.0491 -0.0481 -0.0504 -0.0495 -0.0517 -0.0508 -0.0530 -0.0522 -0.0543 -0.0535 -0.0556 -0.0549 -0.0568 -0.0562 -0.0581 -0.0576 -0.0594 -0.0589 -0.0607 -0.0603 -0.0620 -0.0616 -0.0633 -0.0630 -0.0646 -0.0643 -0.0659 -0.0657 -0.0671 -0.0670 -0.0684 -0.0684 -0.0697 -0.0697 -0.0710 -0.0711 -0.0723 -0.0725 -0.0736 -0.0738 -0.0749 -0.0752 -0.0762 -0.0765 -0.0774 -0.0779 -0.0787
159
1.79 1.80 1.81 1.82 1.83 1.84 1.85 1.86 1.87 1.88 1.89 1.90 1.91 1.92 1.93 1.94 1.95 1.96 1.97 1.98 1.99 2.00
-0.0792 -0.0800 -0.0806 -0.0813 -0.0819 -0.0826 -0.0833 -0.0839 -0.0846 -0.0852 -0.0860 -0.0865 -0.0873 -0.0877 -0.0887 -0.0890 -0.0900 -0.0903 -0.0914 -0.0916 -0.0927 -0.0929 -0.0941 -0.0942 -0.0954 -0.0955 -0.0968 -0.0968 -0.0981 -0.0980 -0.0995 -0.0993 -0.1008 -0.1006 -0.1022 -0.1019 -0.1035 -0.1032 -0.1049 -0.1045 -0.1062 -0.1058 -0.1076 -0.1071
160
161
0016 #Set dcl/da = 2 pi CLaf, i.e. CLaf = CLalpha/(2*pi) CLAF 0.6517 #Set parabolic drag polar #CL1 CD1 CL2 CD2 CL3 CD3 | CD(CL) function parameters CDCL 0.2142 0.00737 0.4235 0.00741 1.3075 0.01381 #=================Define wingtip section SECTION #Xle Yle Zle Chord AngleOfIncidence -0.1849 1.3811 0. 0.2794 0. 14 -2.0 CONTROL #label gain xfrac vx vy vz sgn Raileron 1.0 0.705 0. 0. 0. -1. NACA 0017 #Set dcl/da = 2 pi CLaf, i.e. CLaf = CLalpha/(2*pi) CLAF 0.6517 #Set parabolic drag polar #CL1 CD1 CL2 CD2 CL3 CD3 | CD(CL) function parameters CDCL 0.2142 0.00737 0.4235 0.00741 1.3075 0.01381 #*********************************** SURFACE LWing #Nchordwise Ccpace Nspanwise Sspace 8 1.0 14 2.0 #=================Define wingtip section SECTION #Xle Yle Zle Chord AngleOfIncidence -0.1849 -1.3811 0.0 0.2794 0. 14 -2.0 CONTROL #label gain xfrac vx vy vz sgn Laileron 1.0 0.705 0. 0. 0. -1. NACA 0017 #Set dcl/da = 2 pi CLaf, i.e. CLaf = CLalpha/(2*pi) CLAF 0.6517 #Set parabolic drag polar #CL1 CD1 CL2 CD2 CL3 CD3 | CD(CL) function parameters CDCL 0.2135 0.00651 0.4310 0.00702 1.0646 0.01161 #=================Define section where aileron starts SECTION #Xle Yle Zle Chord AngleOfIncidence -0.218 -0.197 0. 0.559 0. 14 -2.0 CONTROL #control angle is in degrees (for radians set gain=180/pi=57.29578 #label gain xfrac vx vy vz sgn Laileron 1.0 0.75 0. 0. 0. -1.
162
NACA 0016 #Set dcl/da = 2 pi CLaf, i.e. CLaf = CLalpha/(2*pi) CLAF 0.6517 #Set parabolic drag polar #CL1 CD1 CL2 CD2 CL3 CD3 | CD(CL) function parameters CDCL 0.2142 0.00737 0.4235 0.00741 1.3075 0.01381 #=================Center section SECTION #Xle Yle Zle Chord AngleOfIncidence -0.224 0. 0. 0.61 0. 14 -2.0 NACA 0016 #Set dcl/da = 2 pi CLaf, i.e. CLaf = CLalpha/(2*pi) CLAF 0.6517 #Set parabolic drag polar #CL1 CD1 CL2 CD2 CL3 CD3 | CD(CL) function parameters CDCL 0.2142 0.00737 0.4235 0.00741 1.3075 0.01381 # #*********************************** SURFACE RHTail #Nchordwise Cspace Nspanwise Sspace 8 1.0 8 1.0 # #================R H Tail Center section #Tail 4.15 m from Datum to LE h tail SECTION #Xle Yle Zle Chord AngleOfIncidence 0.798 0. 0.01 0.164 0. 8 1.0 NACA 0015 CONTROL #label gain xfrac vx vy vz sgn Relevator 1 1.00 0. 0. 0. 1. #=================R H Tail Elevator Starts SECTION #Xle Yle Zle Chord AngleOfIncidence 0.808 0.0254 0.01 0.184 0. 8 1.0 NACA 0015 CONTROL #label gain xfrac vx vy vz sgn Relevator 1 0.862 0. 0. 0. 1. #=================R H Tail Elevator Shifts Forward SECTION #Xle Yle Zle Chord AngleOfIncidence 0.831 0.138 0.01 0.3 0. 8 1.0 NACA 0012
163
CONTROL #label gain xfrac vx vy vz sgn Relevator 1 0.46 0. 0. 0. 1. #=================R H Tail Elevator Goes Big SECTION #Xle Yle Zle Chord AngleOfIncidence 0.887 0.354 0.01 0.2261 0. 8 1.0 NACA 0010 CONTROL #label gain xfrac vx vy vz sgn Relevator 1 0.32 0. 0. 0. 1. #=================R H Tail Tip SECTION #Xle Yle Zle Chord AngleOfIncidence 0.902 0.419 0.01 0.2096 0. 8 1.0 NACA 0009 CONTROL #label gain xfrac vx vy vz sgn Relevator 1 0.0 0. 0. 0. 1. #*********************************** SURFACE LHTail #Nchordwise Cspace Nspanwise Sspace 8 1.0 8 1.0 #=================L H Tail Tip SECTION #Xle Yle Zle Chord AngleOfIncidence 0.887 -0.419 0.01 0.2096 0. 8 1.0 NACA 0009 CONTROL #label gain xfrac vx vy vz sgn Lelevator 1 0.0 0. 0. 0. 1. #=================L H Tail Elevator Goes Big SECTION #Xle Yle Zle Chord AngleOfIncidence 0.887 -0.354 0.01 0.2261 0. 8 1.0 NACA 0010 CONTROL #label gain xfrac vx vy vz sgn Lelevator 1 0.32 0. 0. 0. 1. #=================L H Tail Elevator Shifts Forward SECTION #Xle Yle Zle Chord AngleOfIncidence 0.831 -0.138 0.01 0.3 0. 8 1.0 NACA 0012 CONTROL #label gain xfrac vx vy vz sgn Lelevator 1 0.46 0. 0. 0. 1. #=================L H Tail Elevator Starts
164
SECTION #Xle Yle Zle Chord AngleOfIncidence 0.808 -0.0254 0.01 0.184 0. 8 1.0 NACA 0015 CONTROL #label gain xfrac vx vy vz sgn Lelevator 1 0.862 0. 0. 0. 1. #=================L H Tail Center section SECTION #Xle Yle Zle Chord AngleOfIncidence 0.798 0. 0.01 0.164 0. 8 1.0 NACA 0015 CONTROL #label gain xfrac vx vy vz sgn Lelevator 1 1.00 0. 0. 0. 1. #*********************************** SURFACE VTail #Nchordwise Cspace Nspanwise Sspace 8 1.0 8 1.0 #=================V Tail Center section SECTION #Xle Yle Zle Chord AngleOfIncidence 0.7882 0.0 0.05 0.509 0. 8 1.0 NACA 0009 CONTROL #label gain xfrac vx vy vz sgn Rudder -1 0.415 0. 0. 0. 1. #=================V Tail Tip section SECTION #Xle Yle Zle Chord AngleOfIncidence 0.9872 0.0 0.481 0.1778 0. 8 1.0 NACA 0010 CONTROL #label gain xfrac vx vy vz sgn Rudder -1 0.415 0. 0. 0. 1. #****************************
165
166
Gross_Mass=12.755 //Empty Mass in Kg Empty_Mass=12.33 //-------------------- PROPULSION --------------------// // Engine is 3W 80cc, 7.5 hp //Engine Type is Piston Left_Engine_Type=0 // Channel number Left_Engine_Channel=2 // Engine parameters look-up table Left_Engine_LUT=Fuji86ccEngine.lut // Propeller is APC 24x10 // Prop diameter, in m Left_Prop_Diameter=0.60696 // Position of propeller hub wrt to aircraft cg, in m Left_Prop_X=0.5715 //Propellor Pan Angle Left_Prop_Pan=3 // Moment of inertia in kg/m^2 Left_Prop_Inertia=0.0025 // Propeller coefficients look-up table Left_Prop_LUT=apc26x10.prd //----------------- CONTROL SURFACES, OTHER AERODYNAMICS (only needed if AVL is disabled -----------------// //----------------- AVIONICS MOUNTING -----------------// // Avionics (IMU sensor) orientation with respect to the aircraft body axes // Euler angles in deg
167
IMU_Sensor_Roll_Angle=0.0 IMU_Sensor_Pitch_Angle=0.0 IMU_Sensor_Yaw_Angle=0.0 // Avionics (IMU sensor) position vector with respect to the aircraft CG, in body axes // Vector components in m IMU_Sensor_Position_X=0.0 IMU_Sensor_Position_Y=0.0 IMU_Sensor_Position_Z=0.0
168
0.17
n 0.301323
0.15
n 7.061208
169
Appendix J: Procedures
Test Overview:
Yak-54/Piccolo
Ground
Testing
The tests described in this appendix will lay out the ground testing phases of the Piccolo II avionics system that will be installed in the Yak-54. The purpose of these tests is to give flight test engineers a high level of confidence in the accuracy and robustness of the avionics package.
Test Objectives:
The following ground tests of the Piccolo II avionics system will be performed
Servo Calibration
The Piccolo produces pulse widths from 1103 to 1929 s. The Piccolo provides calibration tables for 9 pulse widths. This allows the system to cancel out nonlinearities in the system, to a certain degree. For each surface a pulse width will be inputted into the system and its deflection measured. This deflection will be placed in the calibration table, allowing the avionics system to know what pulse to command for an inputted deflection.
Range Check
Satisfactory Piccolo II avionics signal strength at one field length at the Model Masters R/C aircraft field will be verified prior to flight-testing.
Telemetry Check
Prior to flight-testing the following telemetry will be verified: 1. GPS: A satisfactory 3-D GPS solution will be verified 2. Air Data: Air data will be zeroed and its accuracy in the calculation of airspeed and altitude will be verified using a Pitot-static tube calibration kit. The maximum leakage of the static system is not to exceed 100 ft/min and the Pitot system is not to exceed 2 knots/min. 3. Attitude: The aircraft attitude solution will be calibrated and verified by rotating the aircraft from the steady level trim state, with the telemetry checked to match that of an inclinometer.
170
4. Rate Gyro: The aircraft attitude rate solution will be verified by rotating the aircraft from the steady level trim state and observing if the subsequent rate telemetry is reasonable 5. Check signal strength: RSSI telemetry should not drop below -79 dB at anytime during range checking and not drop below -71 dB while the aircraft is next to the ground station.
Weight and Balance Check
The weight and balance of the aircraft will be set to between 10.2 and 10.6 inches aft of the firewall
Taxi Testing
The aircraft will be taxied for a minimum of 10 minutes after any vehicle modifications are performed. All structural connections will be checked for security after the minimum of 10 minutes has passed, and that the Piccolo has not vibrated out of position. The controllability of the aircraft with the actuated tail-wheel will also be monitored and adjusted as needed.
Signatures of Completion
____________________
Vehicle Engineer
____________________
Flight Test Engineer
____________________
Pilot in Command
171
This checklist is to be completed prior to every flight, without exception. Place initials in the space provided as each item is checked. Where necessary, write flight specific information in the spaces provided. CHECK LIST MUST BE COMPLETED BY TWO PERSONS USING VERBAL CONFIRMATION OF CHECKS Pre-flight Briefing o Check and set predicted trim requirements
o o o o
Verify suitable weather Discuss appropriate crosswind corrections Decide upon flight path Check radio and monitor local traffic.
Transmitter ready o Correct flight profile selected ____ Control sweeps completed (correct direction and ____ magnitude of deflection, no more than 2 difference in split aileron at max deflection) Left Aileron (12) ____ Right Aileron (12) ____ Left Elevator (+12, -12) ____ Right Elevator (+12, -12) ____ Rudder (25) ____ Tail wheel ____ Throttle (Full Range) ____ o Range check ____
o
Airframe ready o Batteries charged Ignition Receiver o Servo wires connected to correct receiver channel with safety clips in place where appropriate. Tail Wheel
172
o o o
Rudder Throttle Left Elevator Right Elevator Left Aileron Right Aileron Ignition Kill Switch Servo control connections secured and free. Tail Wheel Rudder Left Elevator Right Elevator Throttle Left Aileron Right Aileron Landing gear secured. Wheels, spacers and collars present and secure. All landing gear mounting bolts secure. Payload secure. Ballast secure. Appropriate fuel on board Fuel lines secure. Fuel tank secure. Hatches installed. Service hatches Payload hatch Structural connections secure. Engine mount Wings Propeller and spinner Pitot and Static Tubes Clear Weight and Balance Correct. Specify: C.G.____________
____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____ ____
Avionics and Ground Station Ready o External Power Source Ready ____ o Ground Station Computer and Operator Interface Power Supplied ____ o Deactivate 802.11 Transmitter on Operator Interface PC ____
o GPS Telemetry Check o Set Altimeter o Zero Air Data Sensors
173
Attitude and Rate Telemetry Check ____ Map Page Configured ____ Flight Plans Set ____ Emergency Flight Plan Set ____ Verify Correct Gains ____ Verify Correct Limits ____ Verify Correct Trims ____ Verify Mission Limits ____ RSSI Signal Strength Check (CHECK TIME HISTORY MUST BE -71 or -79 DURING RANGE CHECK AT ALL TIMES) ____ o Set Telemetry to Fast Rate ____
o o o o o o o o o
Post-Shutdown Checklist
Ignition switch off Receiver switch off Transmitter off Post flight walk around o Check for damage Specify:_____________________________________ ____ ____ ____ ____
Signatures of Completion
_____________________________________ Flight Test Engineer
174
1. 2. 3. 4. 5.
Take-off Climb to approximate altitude of 200 ft AGL Remain in the pattern. Perform one full revolution of a left hand race track pattern Perform one full revolution of a right hand race track pattern i. Verify 3-D GPS solution ii. Verify airspeed iii. Verify attitude and rate data iv. Verify control surface telemetry v. Verify altitude 6. Continue with race track patterns until pilot is satisfied that he understands the handling characteristics of the aircraft 7. Begin approach to landing 8. Land
175
1. 2. 3. 4. 5.
Take-off Climb to approximate altitude of 200 ft AGL Remain in the pattern. Perform one full revolution of a left hand race track pattern Perform one full revolution of a right hand race track pattern i. Verify 3-D GPS solution ii. Verify airspeed iii. Verify attitude and rate data iv. Verify control surface telemetry v. Verify altitude 6. Continue with race track patterns until pilot is satisfied that he understands the handling characteristics of the aircraft
7. Place aircraft into steady level flight for trim capture, call TRIMMED on completion 8. Allow the aircraft to fly freely for at least 5 seconds i. Check to make sure that the aircraft holds 1 knot and 20 ft for 5 seconds ii. Capture control surface trim values 9. Fly the aircraft into a comfortable upwind starting position for the Dutch Roll mode test 10. Apply a rudder singlet 11. Allow the aircraft to fly freely 12. Return to the pattern on the KNOCKOFF call or when the aircraft is leaving controllable visual range 13. Fly the aircraft into a comfortable upwind starting position for the Dutch Roll mode test 14. Apply a rudder doublet 15. Allow the aircraft to fly freely 16. Return to the pattern on the KNOCKOFF call or when the aircraft is leaving controllable visual range 17. Fly the aircraft into a comfortable upwind starting position for the Roll mode test 18. Fly the aircraft into an approximately 30 degree banked right turn i. Give bank angle feedback to the pilot 19. On READY bank to a 30 degree bank angle left using constant aileron
176
20. Hold 30 degree left bank for approximately 2 seconds 21. Return to the pattern 22. Fly the aircraft into a comfortable upwind starting position for the Short Period mode test 23. Apply a pull up singlet 24. Allow the aircraft to fly freely longitudinally and apply corrective lateral inputs 25. Return to the pattern on the KNOCKOFF call or when the aircraft is leaving controllable visual range 26. Fly the aircraft into a comfortable upwind starting position for the Short Period mode test 27. Apply a push-down to pull-up elevator doublet 28. Allow the aircraft to fly freely longitudinally and apply corrective lateral inputs 29. Return to the pattern on the KNOCKOFF call or when the aircraft is leaving controllable visual range 30. Return to the pattern and fly the aircraft into a comfortable upwind starting position for the Phugoid mode test 31. Apply up-elevator step and return to trim elevator on RELEASE call i. Call RELEASE when airspeed has decreased by 5 knots 32. Allow the aircraft to fly freely longitudinally and apply corrective lateral inputs as needed 33. Return to the pattern on the next KNOCKOFF call or when the aircraft is leaving controllable visual range 34. Continue in the traffic pattern and commence approach to landing. 35. Land 36. Shut-down
177
0.462118 0.493588
0.23
n 6.052757
0.447214 0.340872
0.22
n 5.683227
178
1. 2. 3. 4. 5.
Take-off Climb to approximate altitude of 200 ft AGL Remain in the pattern. Perform one full revolution of a left hand race track pattern Perform one full revolution of a right hand race track pattern i. Verify 3-D GPS solution ii. Verify airspeed iii. Verify attitude and rate data iv. Verify control surface telemetry v. Verify altitude 6. Continue with race track patterns until pilot is satisfied that he understands the handling characteristics of the aircraft
7. Place aircraft into steady level flight for trim capture, call TRIMMED on completion 8. Allow the aircraft to fly freely for at least 5 seconds i. Check to make sure that the aircraft holds 1 knot and 20 ft for 5 seconds ii. Capture control surface trim values 9. Enter the traffic pattern i. Disable all control loops except bank angle control ii. Set the bank angle command to ON and set the command value to 0.0 deg 10. Fly the aircraft to comfortable upwind starting position 11. Engage autopilot on READY for 5 seconds and disengage 12. Return to the traffic pattern 13. Fly the aircraft to a comfortable upwind starting position for a right 10 deg bank angle i. Set the bank angle command to ON and set the command value to +10.0 deg 14. Engage autopilot on READY 15. On KNOCKOFF call disengage the autopilot 16. Return to the traffic pattern i. Check flight data and tune gains
179
17. Fly the aircraft to a comfortable upwind starting position for a left 10 deg bank angle i. Set the bank angle command to ON and set the command value to -10.0 deg 18. Engage autopilot on READY 19. On KNOCKOFF call disengage the autopilot 20. Return to the traffic pattern i. Check flight data and tune gains 21. Fly aircraft to a comfortable upwind starting position for 20 deg right bank angle i. Set the bank angle command to ON and set the command value to +20.0 deg 22. Engage autopilot on READY 23. On KNOCKOFF call disengage the autopilot 24. Return to the traffic pattern i. Check flight data and tune gains 25. Fly aircraft to a comfortable upwind starting position for 20 deg left bank angle i. Set the bank angle command to ON and set the command value to -20.0 deg 26. Engage autopilot on READY 27. On KNOCKOFF call disengage the autopilot 28. Return to the traffic pattern i. Check flight data and tune gains 29. Fly aircraft to a comfortable upwind starting position for 30 deg right bank angle i. Set the bank angle command to ON and set the command value to +30.0 deg 30. Engage autopilot on READY 31. On KNOCKOFF call disengage the autopilot 32. Return to the traffic pattern i. Check flight data and tune gains 33. Fly aircraft to a comfortable upwind starting position for 30 deg left bank angle i. Set the bank angle command to ON and set the command value to -30.0 deg 34. Engage autopilot on READY 35. On KNOCKOFF call disengage the autopilot 36. Return to the traffic pattern i. Check flight data and tune gains 37. Land
180
1. 2. 3. 4. 5.
Take-off Climb to approximate altitude of 200 ft AGL Remain in the pattern. Perform one full revolution of a left hand race track pattern Perform one full revolution of a right hand race track pattern Verify 3-D GPS solution Verify airspeed Verify attitude and rate data Verify control surface telemetry Verify altitude 6. Continue with race track patterns until pilot is satisfied that he understands the handling characteristics of the aircraft
7. Place aircraft into steady level flight for trim capture, call TRIMMED on completion 8. Allow the aircraft to fly freely for at least 5 seconds Check to make sure that the aircraft holds 1 knot and 20 ft for 5 seconds Capture control surface trim values 9. Enter steady level flight i. Disable all control loops except the bank angle and heading control loop ii. Set bank angle command to AUTO iii. Set the heading command to the upwind trim heading 10. Fly the aircraft to a comfortable starting position 11. Engage the autopilot on READY command from test engineer 12. On KNOCKOFF command from the test engineer disengage the autopilot 13. Return to the traffic pattern i. Check flight data and tune gains 14. Fly the aircraft to a comfortable starting position for a 45 degree right heading change i. Set the heading command to +45 degrees from trim 15. Engage the autopilot on READY 16. On KNOCKOFF call disengage the autopilot 17. Return to the traffic pattern
181
i. Check flight data and tune gains 18. Fly the aircraft to a comfortable starting position for a 45 degree left heading change i. Set the heading command to -45 degrees from trim 19. Engage the autopilot on READY 20. On KNOCKOFF call disengage the autopilot 21. Return to the traffic pattern i. Check flight data and tune gains 22. Fly the aircraft to a comfortable starting position for a 90 degree right heading change i. Set the heading command to +90 degrees from trim 23. Engage the autopilot on READY 24. On KNOCKOFF command from the test engineer disengage the autopilot 25. Return to the traffic pattern i. Check flight data and tune gains 26. Fly the aircraft to a comfortable starting position for a 90 degree left heading change i. Set the heading command to -90 degrees from trim 27. Engage the autopilot on READY 28. On KNOCKOFF command from the test engineer disengage the autopilot 29. Return to the traffic pattern i. Check flight data and tune gains 30. Fly the aircraft to a comfortable starting position for a 150 degree right heading change i. Set the heading command to +150 degrees from trim 31. Engage the autopilot on READY 32. On KNOCKOFF call disengage the autopilot 33. Return to the traffic pattern i. Check flight data and tune gains 34. Fly the aircraft to a comfortable starting position for a 150 degree left heading change i. Set the heading command to -150 degrees from trim 35. Engage the autopilot on READY 36. On KNOCKOFF call disengage the autopilot 37. Return to the traffic pattern i. Check flight data and tune gains 38. Return to manual mode
182
1. 2. 3. 4. 5.
Take-off Climb to approximate altitude of 200 ft AGL Remain in the pattern. Perform one full revolution of a left hand race track pattern Perform one full revolution of a right hand race track pattern i. Verify 3-D GPS solution ii. Verify airspeed iii. Verify attitude and rate data iv. Verify control surface telemetry v. Verify altitude 6. Continue with race track patterns until pilot is satisfied that he understands the handling characteristics of the aircraft
7. Place aircraft into steady level flight for trim capture, call TRIMMED on completion 8. Allow the aircraft to fly freely for at least 5 seconds i. Check to make sure that the aircraft holds 1 knot and 20 ft for 5 seconds ii. Capture control surface trim values 9. Enter the traffic pattern i. Set all control loops except waypoint tracking to ON ii. Set the bank angle command to AUTO iii. Set the heading command to the upwind trim heading iv. Set the altitude command to ON and command 200 ft AGL v. Set the IAS threshold to -1 knots vi. Set the airspeed command to the trim airspeed 10. Fly aircraft to comfortable upwind starting position 11. Engage the autopilot on READY command from test engineer 12. After 5 seconds disengage autopilot 13. Return to the traffic pattern 14. Fly aircraft to comfortable upwind starting position 15. Engage the autopilot on READY i. Increase airspeed command to +5 knots from trim 16. On KNOCKOFF call disengage the autopilot 17. Return to the traffic pattern i. Check flight data and tune gains
183
18. Fly aircraft to comfortable upwind starting position 19. Engage the autopilot on READY i. Increase airspeed command to -5 knots from trim 20. On KNOCKOFF call disengage the autopilot 21. Return to the traffic pattern i. Check flight data and tune gains 22. Fly aircraft to comfortable upwind starting position 23. Engage the autopilot on READY i. Increase airspeed command to +10 knots from trim 24. On KNOCKOFF call disengage the autopilot 25. Return to the traffic pattern i. Check flight data and tune gains 26. Fly aircraft to comfortable upwind starting position 27. Engage the autopilot on READY i. Increase airspeed command to -10 knots from trim 28. On KNOCKOFF call disengage the autopilot 29. Return to the traffic pattern i. Check flight data and tune gains
184
185