Application of active controls technology to aircraft bide smoothing systems

TABLE OF CONTENTS (Continued) Chapter Page III ANALYSIS AND SYNTHESIS .................

3.1 Order 3.2 Demonstration Aircraft ......

3.4 Longitudinal Ride Smoothing Systems ............ ... 28 3.4.1 The Basic JetStar--Longitudinal 3.4.2 Baseline Longitudinal Ride Smoothing System . 33 3.4.3 Effect of Inner Loop Closures .......... ... 35 3.4.4 Basic Multi-Loop Longitudinal Ride 3.4.5 Analytic Model of Longitudinal Ride 3.4.7 Longitudinal Ride Smoothing System II . ... 62 3.5 Lateral Ride Smoothing Systems .... ......... ... 65 3.5.1 The Basic JetStar--Lateral Case ......... 65 ...

3.5.3 Analytic Model of Lateral Ride Smoothing System ....... .................... ... 82 3.5.4 Alternate Lateral Ride Smoothing System . . . 85 3.6 Overall Effectiveness of Combined Axis Ride Smoothing System ........ ................ ... 89 IV SIMULATION EXPERIMENTS ......

.................. ... 91 4.1 Order of Presentation ........

4.3 Digital Computer Program .. ...............

. 93 4.4 Analog Circuits ....... ................... ... 95 vi

TABLE OF CONTENTS (Continued) Chapter Page IV 4.5 Hybrid Simulation Verification ......... 95 ......

4.6 Simulation Evaluation 4.7 Handling Qualities Evaluation 4.7.2 Longitudinal Task .... ........... . ... 05 4.7.4 4.7.5 Smooth Air Evaluation, Conclusions ......... Ill 4.7.6 Instrument Landing System Approach Task . Ill 4.7.7 Simulation of Straight and Level Flight . 122 4.8 Conclusions ........

V FLIGHT TEST PROGRAM ......... ............

5.1 Planned Program ....... .................. ... 129 5.2 Implementation of RSS Aboard the JetStar ........ .. 129 5.4 Data Acquisition and Reduction ... ......... ... 133 5.5 Summary of Flight Test Data .....

........ ... 134 5.6 Conclusions ......... ................ ... 138 VI EXTENSION OF RIDE SMOOTHING SYSTEM CONCEPT TO STOL AIRCRAFT .......... .............. .... ...

6.-1 Selected Aircraft ....... ................. ... 143 6.2 Synthes-is of Ride Smoothing Systems ........... ...

6.2.1 Longitudinal RSS ...... ............. ... 145 6.2.2 Lateral RSS ........ .............. ... 153 6.2.3 Improvement in Passenger Comfort ....... ... 159 vii

TABLE OF CONTENTS (Continued) Chapter Page VI 6.3 Simulator Evaluation of STOL Ride Smoothing Systems ...... ............ ....... . . . . .. 159 6.4 Conclusions ... ....... .................. ... 161 Appendix A DEFINITION OF STABILITY DERIVATIVES ... ........... ... 165 A.1 Axis Systems ......... .................. ... 165 A.2 Definition of Nondimensional Stability Derivatives 166 A.3 Transformation of Stability Axis Derivatives A.3.1 Longitudinal Derivatives ........... . . . .. 166 A.4 Dimensional Stability Derivative Definitions ..... . 168 A.4.l Longitudinal Derivatives ............. . .. 168 A.4.2 Lateral Derivatives . ...... ..... . 169 B TURBULENCE FILTERS AND INPUT-OUTPUT RELATIONSHIPS . ... 173 C JETSTAR DATA ........ ...................... .. 177 D FORMULATION OF TRANSFER FUNCTIONS FOR MULTI-LOOP FEEDBACKCONTROL SYSTEMS ...... ................ ... 181 E STOL DATA ......... ........................ .. 187 ViII LIST OF TABLES Table Page I Longitudinal Competigg Systems ...... ..........

... 23 II Performance of Baseline Longitudinal RSS ...... .....

III Characteristics of Longitudinal Ride Smoothing System I 60 IV Characteristics of Longitudinal Ride Smoothing System II 67 V Effect of Feedbacks on Roll Subsidence and Spiral Modes 74 VI Characteristics of Lateral Ride Smoothing System .... ... 80 VII Comparison of Lateral Ride Smoothing Systems VIII Cooper-Harper Rating Scale ..... .............

IX Average Cooper-Harper Pilot Ratings, Longitudinal Task.

. 107 X Average Cooper-Harper Pilot Ratings, Lateral Task . . . 109 XI Average Cooper-Harper Pilot Ratings, Combined Axis Task . 110 XII Average Cooper-Harper Pilot Ratings, ILS Task ....

XIII Simulation Results, ILS Tracking Task ... ..........

... 123 XIV Ride Smoothing System Flight Test Results .......... ... 136 XV Comparison of JetStar and STOL Longitudinal Ride Smoothing Systems .......... ..........

XVI Comparison of JetStar and STOL Lateral Ride Smoothing Systems ............

.............. . 156 XVII Average Cooper-Harper Pilot Ratings, STOL ILS h Approac -Task .......... ....................

... 160 ix LIST OF FIGURES Page Figure 1 Open-Loop Control System 2 Closed-Loop Control System ........ ................

Passenger Satisfaction Criteria .............. .. . •.•.18 4 NASA of NASA General Purpose Airborne Simulator . . . 27 5 Schematic 6 Handling Qualities Specification for n/ ............ ... 30 Power Spectra of a Due to Turbulence for Basic JetStar . 31 z Power Spectra of a for Basic JetStar . . . . 32 8 Partitioned z ......... 33 System ...

9 Baseline Longitudinal Ride Smoothing Baseline Longitudinal RSS ......... .... .. 36 10 Root Locus for e Loop Closure .............. 37 Root Locus for 6 .....

. 38 Ride Smoothing System .........

12 Basic Longitudinal RSS .. ......... ... 39 13 Root Locus for Basic Longitudinal 14 Performance of Basic Longitudinal RSS; Ga as a Function of K and K. 4o a z z RSS; 15 Performance of Basic Longitudinal aq a e z RSS; 16 Performance of Basic Longitudinal 0 f as a Function of K and K 17 Performance of Basic Longitudinal RSS; C as a Constrained Function of Ka and K . . . . . . ..

z z 18 Comparison of Digitally Calculated a with Analytic Expression ... .... az . ....

... 49 for w Due to A ..... .............

19 Power Spectra Root Locus of Effect on Short-Period Dynamics of Filters in az+ +f Feedback Loop xi PRECE ING PAGE BIANK NOT ifIIvF LIST OF FIGURES (Continued) Figure Page Root Locus of Effect on Short-Period Dynamics of a Lead 'Filter in 6 +e Feedback Loop ..........

Longitudinal Ride Smoothing System I .............

23 Performance of Longitudinal RSS I; aa as a Function of Ka and Ke ......

z z 24 Performance of Longitudinal RSS 1; a as a Function of K and K . ....

e 25 Performance of Longitudinal RSS I; .

26 Performance of Longitudinal RSS I; a a as a Constrained Function of Ka and Ke ..........

z z 27 Root Locus for Longitudinal RSS I .... ............

... 59 28 Comparison of a Power Spectra for Basic and Longitudinal RSS I Augmented JetStar ...... .................

... 61 29 Longitudinal Ride Smoothing System II ...

......... ... 63 30 Bode Magnitude Plot of 31 Root Locus of Effect on Short-Period Dynamics of a Notch Fite i z Filter .

in a f Feedback Loop ... .... .......

32 Comparison of az Power Spectra for Basic and Longitudinal RS$ 11 Augmented JetStar .... .................

33 Power Spectra of ay Due to Turbulence for Basic JetStar 70 34 Lateral Ride Smoothing System 35 Dutch Roll Root Locus for Lateral RSS ... ....... ... 73 36 Performance of Lateral RSS; a as a Function of K and K r.......... ......

a. ay r Performance of Lateral RSS; a as a Function of K and K .........

r a r y Performance of Lateral RSS; o as a Function of K p a r.

Yxi) LIST OF FIGURES (Continued) Page Figure Lateral RSS; Performance of a as a Function of K and K ...... ... .. 78 6 6 r sfg a y Performance of Lateral RSS; a. as a Constrained Function r ya A a y y .1 41 Comparison of a 'Power Spectra for Basic and Lateral -RSS Augmented Y JetStar .........

a a with Analytic 42 Comparison of Digitally Calculated Expression ....

Ride Smoothing System ........... ... 87 43 Alternate Lateral ........... .... .. 94 45 GPAS Test Pilot's Instrumentation Circuit Diagram; 46 Analog Equalization a + 6f Feedback z 47 Analog Equalization Circuit Diagram; . ......... 97 ............

6 0 6 Feedback Loop .e 48 Analog Equalization Circuit Diagram; 51 Power 52 Power 53 Altitude Track; Basic JetStar in Design Turbulence Field (Pilot A) ............ ...................

54 Deviation from Localizer; Basic JetStar in Design Field (Pilot A) .... ..............

Turbulence History; JetStar in Design Turbulence 55 Simulation Time I. . 117 Field (Pilot A) ........ ....................

56 Power Spectra of a Due to Turbulence, Simulation Data . 124 z xiii LIST OF FIGURES (Continued) Figure Page 57 Power Spectra of a Due to Turbulence, Simulation Data . 125 y 58 Airborne Analog Computer ....

59 Comparison of a Power Spectra for Basic and z Longitudinal RSS I Augmented JetStar (Fl'ight Data) . . . 137 6o Time History; Basic and Longitudinal RSS J Augmented JetStar in Turbulence (Flight Data) ... ....... .....

61 Buffalo Longitudinal RSS .... .......... . . .. . .147 62 S-11 Longitudinal RSS ...

63 Comparison of a Power Spectra for Basic and z Longitudinal RSS Augmented 64 Comparison of a Power Spectra for Basic and z Longitudinal RSS Augmented S-il .....

.......... ... 152 65 Buffalo Lateral RSS ....... .................

66 S-11 Lateral RSS ...... ........

67 Comparison of a Power Spectra for Basic and Lateral RSS Augmented Y Buffalo ...........

68 Comparison of a Power Spectra for Basic and Lateral RSS Augmented Y S-l. .... ...........

69 Axis Systems ...........

xiv NOMENCLATURE a speed of sound in air a longitudinal acceleration along the X-body axis at the x center of gravity (positive forward) a lateral acceleration along the Y-body axis at the center Y of gravity (positive out right wing) a' lateral acceleration parallel to the Y-body axis at a y distance I and I from the center of gravity; z a' = 1 r x, - lIzp z x y a normal acceleration along the Z-body axis at the center z of gravity (positive down) a] normal acceleration parallel to the Z-body axis at a z distance I from the center of gravity; a' = a - q x z z x b reference wing span C reference wing chord C comfort rating (Equation 2.3.1)_ - C lift coefficient; steady-state CL L CL0 2 P VT 0 0 D aerodynamic drag force along total velocity vector (positive aft) g acceleration due to gravity G. transfer function of output j due to input i i h altitude to body axes referred of inertia moments x, Iy, I z of inertia referred to body axes I xz product jW imaginary part of complex variable; s = a ± jW K feedback gain particularized by subscript i xv NOMENCLATURE (Continued) I xdistance along the X-body axis from the center of x gravity (positive forward) I zdistance along the Z-body axis from the center of z gravity (positive down) L rolling moment about the X-axis due to aerodynamic torques (positive right wing down) L aerodynamic lift force perpendicular to the total velocity vector in the aircraft's plane of symmetry (positive up) L. characteristic gust length, particularized by subscript m aircraft mass M Mach number M pitching moment about the Y-axis due to aerodynamic torques (positive nose up) n/a handling qualities parameter (Figure 6) aerodynamic normal force along the Z-body axis (positive up) N yawing moment about Z-axis due to aerodynamic torques (positive nose right) NA numerator of transfer function G i i N'i - coupling numerator of first kind: outputs qi and q.

6 6 due to inputs &k and 61 0 1 p roll rate; angular velocity .about X-axis (positive right wing down) q pitch rate; angular velocity about Y-axis (positive nose up) xvi (Continued) NOMENCLATURE Z-axis yaw rate; angular velocity about r '(positive nose right) o + jw s Laplace operator, area- reference wing S T± time to I amplitude time to double amplitude T along X-axis linear perturbed velocity u (positive forward) velocity along the X-axis linear steady-state U forward) (positive velocity along the Y-axis linear perturbed v out right wing) (positive steady-state velocity VT total linear (positive forward) along the Z-axis linear perturbed velocity w (positive down) weight W loading parameter W/S wing it steady-state velocity along the Z-axis W linear (positive down) X-axis aerodynamic force along the X (positive forward) force along the Y-axis V aerodynamic wing) (pos-itive out right the Z-axis aerodynamic force along Z down) (positive xvii NOMENCLATURE (Continued) 1/10 handling qualities parameter: inverse cycles to I/Cl/lO mode amplitude of short-period aperturbation angle of attack a steady-state (trim) angle of attack relative to fuselage reference line; a = Cs-I (Wo/VT ) = sin- (U /VTo) 0 0 path angle y flight 8sideslip angle 6a aileron deflection (positive for positive rolling moment) direct lift flap deflection 6 f (positive for trailing edge down) elevator surface deflection from trim (positive for 6e e aft surface) nose down pitching moment for yawing rudder deflection (positive for nose-left 6 r moment (negative N)) asfg side force generator deflection (positive for trailing edge left) A denominator of free aircraft transfer function denominator of augmented aircraft transfer function A' to fuselage inclination of thrust axis relative reference line second-order mode particularized i damping ratio of linear by subscript Xcomplex root of characteristic equation A white noise angle (positive nose up)

e perturbation pitch

steady-state (trim) pitch angle G .

xviii NOMENCLATURE (Continued) p mass density of air G real portion of complex variable; s = a ± jw root-mean-square intensity of motion quantity, forcing function., or surface deflection particularized by subscript 1.

time constant of first-order mode particularized by subscript 4roll angle (positive right wing down) (i) power spectral density of quantity (i) Pheading angle W spectral frequency Cd.

undamped natural frequency of second-order mode particularized by subscript Special Subscripts c control surface actuator command dr Dutch Roll mode gust ph phugoid mode R roll subsidence mode s spiral mode sp short-period mode ST static Superscripts () indicates differentiation with respect to time xix (Continued) NOMENCLATURE Abbreviations Alleviation System Gust Load GLAS General Purpose Airborne Simulator GPAS System Instrument Landing ILS System Mode Suppression MSS and Space Administration National Aeronautics NASA Induced Oscillation PIO Pilot Ride Smoothing System RSS Augmentation System Stability SAS Take-off and Landing STOL Short xx

CHAPTER I INTRODUCTION 1.1 Problem Statement This dissertation reports on the analysis, synthesis, and experimental evaluation of a Ride Smoothing System for aircraft flying in atmospheric turbulence. Both longitudinal and lateral systems were investigated. Multiple design criteria, intended to satisfy the require ments of all components of the aircraft/pilot/passenger system, were established.

Three Ride Smoothing System designs, two for the longitudinal and one for the lateral case, all of a multiloop feedback type, were developed.

Two sets of unique control surfaces, direct-lift flaps and side-force generators, were used in addition to elevator and rudder for the mechanization.

Predicted system performance was verified in a fixed-base ground simulator. The systems were also mechanized aboard the National Aeronaut'ics and Space Administration (NASA) General Purpose Airborne Simulator (GPAS). Limited flight tests were conducted to evaluate two of the Ride Smoothing Systems.

Before discussing the motivation for this research, it is necessary to define several concepts: Ride Smoothing System (RSS), Gust Load Alleviation System (GLAS), Mode Suppression System (MSS), and Stability Augmentation System (SAS). The first three systems are designed primarily to attenuate aircraft response to atmospheric turbulence, but differ considerably in design criteria.

A Ride Smoothing System can be defined as one which proposes to improve passenger and flight crew comfort. It is generally designed to suppress aircraft motion induced by moderate to heavy continuous turbulence ( = 2.1 m/sec). Attentuation is achieved by damping rigid body modes, changing their natural frequency and/or deflecting control surfaces to counteract transient loads.

A Gust Load Alleviation System is designed to protect the aircraft structure from exceeding load limits. Transport class ai.rcraft are typically stressed to + 2.5 g. At low speeds, lift loads induced by large "sharp-edged" gusts (w = 15 m/sec) can exceed the design limit.

Such aircraft are termed "gust-critical."

Significant extension of the load-factor envelope or an equivalent reduction in structural weight are possible if an active GLAS is incorporated.

A Mode Suppression System is designed to counteract turbulence induced flexible-body mode excitation.

The design objective for a MSS is usually twofold: improvement of ride qualities at the pilot station and improvement of the fatigue life of the airframe.

Both the Gust Load Alleviation System and Mode Suppression System may include the functions of a Ride Smoothing System.

Successful implementation of any of the three, the RSS; GLAS and MSS, may require the addition of a Stability Augmentation System in order to restore or improve the aircraft handling qualities.

Unfortunately, the above terms, and a number of variations, are often used interchangeably in the literature.

Similarly, the terms turbulence (herein considered continuous) and gusts (discrete) have, in the past, been used synonymously.

This report will deal only with the investigation of a Ride Smoothing System designed to operate in continuous turbulence as defined above.

1.2 Historical Perspective Past Ride Smoothing System designs have used two general approaches: open- and closedrloop design philosophies. The.essential difference between the two can be i1llustrated by simple block diagrams: TURBULENCE SENSOR & AIRCRAFT MOTION DYNAMICS FILTER PILOT I CONTROL INPUT COMMAND FIGURE 1. OPEN-LOOP CONTROL SYSTEM TURBULENCE MOTION AIRCRAFT DYNAMICS CONTROL PILOT C OMAND I NPUT SENSOR FILTER FIGURE 2. CLOSED-LOOP CONTROL SYSTEM The open-loop scheme (Figure 1) has one very desirable feature: in principle, the aircraft dynamics remain unchanged as a result of the control. Practical difficulties, however, abound. In order to optimize the control law, a precise mathematical formulation of the turbulence field and aircraft dynamics is required. An adequate gust angle of attack sensor is difficult to mechanize. The most popular sensor has been the nose-boom mounted angle of attack vane. Unfortunately, an angle of attack vane measures not only variations in the remote wind, but responds to aircraft motion as well. Unless the vane measurements are accurately corrected for'aircraft motion, an "aerodynamic feedback" results--and the characteristic equation is modified. Finally, if the overall gain of the system is high; i.e., almost total alleviation of, say, normal acceleration is achieved, the pilot will be unable to command a change in flight path by conventional means. With modern analog circuitry, servosystems and analytic techniques, an open-loop design can be implemented, but the resulting system is quite complex.

The closed-loop RSS is shown as a classical feedback system (Figure 2). As compared to the open-loop scheme, the main advantage of a feed back system is that no explicit knowledge of the turbulence field and its effect on the aircraft response is required.

Careful analysis of the effect of feedback on the characteristic equation roots must, however, be undertaken. The effect of high gain systems on control is, of course, the same as for the open-loop case. The simplification in terms of sensor requirements afforded by the closed-loop system makes this approach more attractive from the practical viewpoint.

aircraft with Not surprisingly, the first attempts at providing gust alleviation capab'ility' ended in failure.

a ride smoothing or several of these pioneering Phillips, in a survey article (1), describes to efforts. Waterman, about 1930, built an airplane with wings attached hinges. Steady lift forces were balanced by the fuselage by skewed thus Unsteady lift loads caused the wings to deflect, pneumatic struts.

the local angle of attack. A modern equivalent of this reducing mechanism is found in the flexible, swept-wing aircraft. The biggest lateral control: drawback in Waterman's design was lack of adequate would cause deflection of the wings in opposition deflection of ailerons to the desired rolling moment.

1953, results of a series of ride smoothing flight tests In conducted with a Lancaster bomber by the British Royal Aircraft were published (2). The Lancaster system was designed Establishment to operate essentially in an open-loop sense: the vertical component mounted on a boom of turbulence was sensed by a "wind incidence meter" nose of the aircraft. The derived electrical analog ahead of the so as to signal was then used to command symmetric aileron deflection indicated reduce the anticipated lift increment. Flight data, however, The preliminary explanation, an amplification of aircraft response.

in a 1961 report (3), blamed the failure on incomplete confirmed effect of analysis: the system design had neglected the adverse aileron-induced pitching moment on system performance.

1950, the Douglas Aircraft Company conducted' flight tests with In feedback control a C-47 aircraft configured for gust alleviation. The aileron deflection as a used a linkage system which caused symmetric the same effort, and for As with the British of wing bending.

function (1).

were inconclusive reasons, flight tests by design, also summarized essentially open-loop GLAS/RSS Another Ren4 Hi,rsch about (1), was developed by the Frenchman Phillips light aboard a specially-fabricated and successfully flight-tested Hi-rsch's clever mechanization the period 1954-1967 (4..

aircraft during mechanically too complex and lateral system is of both a longitudinal The many free aerodynamic surfaces,' cables, to fully discuss here.

success of his design that were critical to the bellcranks, etc., and servosystems if the to be replaced by modern sensors would have aboard a larger aircraft.

were to be implemented design the investigation of NACA/NASA Technical Notes document Numerous Research at the NASA Langley Ride Smoothing System a longitudinal (5), in 1951 by Phillips and Kraft The first of these, published Center.

The of the open-loop system.

the basic design philosophy sets forth control surfaces are of attack vane. Two element is an angle sensing gearing, lift flaps and, through fixed driven by this signal: direct in downwash the flap-induced change In order to counter the elevator.

portion of the flaps it was proposed that a small inboard at the tail, the proposed flaps. In principle, in opposition to the main be driven vertical of turbulence-induced was capable of total alleviation system flight path was moment. Pilot control of and pitching acceleration lift flaps stick to both the direct by connecting the control provided feasibility of using research established the and elevator. Concurrent measure of the vane to provide an adequate a single angle of attack (6).

over the entire wing span angle of attack perturbation average indicated work and analog computer simulation Subsequent analytic could be insured by providing a small that adequate static stability of some alleviation capability (7). Initial static margin at the expehse a C-45 aircraft flying at a single flight tests were conducted aboard 40 to 50% at specific airspeed. A reduction in acceleration of of longitudinal control frequencies was realized (8). Pilot opinion adequacy was reported favorable.

in of a more complex flight test program were reported Results Additional alleviation capability had 1961 by Hunter, et al. (9)(10).

lift flaps. Another been achieved by slaving the ailerons to the direct of a negative feedback loop in the modification was the ifcorporation using a circuit. The feedback was mechanized flap position command integrator. This feature permitted longitudinal mechanical/electrical Performance of the minimized phugoid mode excitation.

trim changes and attenuation of 60% at the system was improved to a maximum acceleration was recorded when frequency. Somewhat lower performance short-peri6d accelerometer command signal was generated by a c.g.-mounted normal the Hunter, et al. do not rather than the angle of attack vane. Curiously, effect on aircraft dynamics of changing from an essentially discuss the of attack vane) to.a close-loop (accelerometer) system, open-loop (angle to approach instability except to state that the latter system was known at high gains.

the C-45 project was completion of these experiments, Following renewed In 1971, Phillips' original design received terminated.

and Sparrow (11) explain the decade-long attention (11 - 16). Barker of the relative insensitivity hiatus in development as being the result of the 1960's generation of aircraft to atmospheric turbulence. It-was the advent of Short Takb-off and Landing (STOL) aircraft that provided motivation for continuation of research in Ride Smoothing Systems.

Several reasons can be cited for the poor ride quality anticipated aboard STOL aircraft. The sensitivity of an aircraft to turbulence is, to first order, inversely proportional to wing-loading (W/S). Yet, a number of STOL designs rely on low wing-loading in order to achieve re quired short field performance. In addition, STOL aircraft are intended to operate at low altitudes where atmospheric turbulence is most severe.

Several other investigations of open-loop RSS/GLAS have been reported in the literature. One of these, a 1957 report by Tobak (17), is particularly interesting in that he was the first to apply the Weiner optimum filter theory to the problem of minimization of aircraft response to turbulence. Tobak's analysis validated some of the classical analysis results of Phillips and Kraft (5), as well as establishing the form of the optimum cbmmand circuit filter. Tobak assumed that a sensor signal proportional purely to fluctuations in angle of attack was available, the turbulence field could be described by the Dryden model, and a single control surface was available.

A very similar analysis, culminating in 1971 flight tests with a Dassault Mirage III delta-wing fighter by the Office National d'Etudes et de Recherches Aerospatiales (ONERA), was reported by Coupry (18).

Initial data indicated that substantial reduction in the normal acceleration levels at the pilot-station was achieved.

A series of studies of closed-loop, longitudinal RSS Smoothing of at the University a group out by carried has been designs System Osaka in Japan (19)(20).

In the first of these papers, three systems were postulated; all depending on feedback of normal acceleration and pitch attitude, rate and acceleration to the elevator and direct lift flaps of a conventional subsonic aircraft. The first system; designated a "Linkage-Control System," summed all feedback signals before generating .a command signal for the two control surfaces. The second, "Noninter acting System," made provision for separate equalization in each feed back path. The last, "Split-Control System," commanded the direct lift flaps in response to vertical acceleration and the elevator in response to pitch rate only. Within the limits of the assumptions of the study, the authors concluded that the "Split-Control System" was not only the simplest, but also the most effective in reducing c.g. acceleration.

Stability of the aircraft system was substantially increased but the short-period frequency was decreased. The authors did not comment on the effect of such a shift on the handling qualities of the vehicle, although the possibility of introducing an integrating circuit in the feedback loops in order to improve control was postulated. The second paper reported on the calculation of an optimal feedback system, and showed the performance of the optimal and simplified ("Split Control") systems to be equivalent.

A closed-loop design approach, almost identical to that of the Osaka group, was adopted by Holloway, et al. of Boeing (21) for a feasibility study of a STOL Ride Smoothing System. Vertical acceleration was fed back to the direct lift flap through a low-pass filter and pitch rate to the elevator through an integrator circuit. Well-defined operating criteria were established, including the design turbulence level, attentuation requirement for passenger acceptance, and a handling qualities specification. In addition, a lateral ride smoothing system was designed based on feedback of filtered yaw rate and lateral acceleration to the rudder. The same general system was adapted for installation aboard a deHavilland DHC-6 Twin Otter aircraft (22).

Several theoretical studies based on the application of optimal control theory to closed-loop Ride Smoothing Systems have also appeared in the.literature. Hess (23) investigated a system that.drove the elevator in response to the sum of three signals: normal acceleration, pitch rate, and angle of attack. One of his major conclusions was that the performance of the optimal controller was insensitive to characteristics of the turbulence field; in particular, the "character istic gust length." In subsequent investigations, the feedforward loop was eliminated because of the difficulty in mechanizing a practical angle of attack sensor. The resulting system, identical in form to an "acceleration autopilot," was shown to have an alleviation capability nearly equivalent to the optimal controller (24)(25).

A similar con figuration had been studied earlier by McClean (26).

Probably the most ambitious study of an aircraft gust alleviation system designed to suppress longitudinal rigid-body response was undertaken by Iliff (27). His research involved the application of stochastic identification theory to a system (the aircraft) contaminated by state noise (turbulence). Not only did lliff's technique successfully extract almost exact values of aircraft stability derivatives, it also yielded a good approximation of the root mean square turbulence intensity.

liff also demonstrated application of stochastic control theory to solving the gust alleviation problem; minimizing either vertical acceleration or pitch rate. Unfortunately, no research aircraft equipped with an onboard digital computer capable of performing the required calculations is available to prove Iliff's concepts in flight.

A great deal of research effort since the early 1960's has dealt with the problem of structural mode alleviation for flexible aircraft.

A good survey of this work is presented in a paper by Swaim (28). " Solutions to this problem are generally attempted through the application of linear optimal control theory. An example of this approach is discussed by Smith, et al. (29). Since this dissertation does not consider the effect of turbulence on non-rigid aircraft, detailed review of Mode Alleviation Systems will be omitted.

1.3 Research Objectives As mentioned previously, the ride quality aboard STOL-class aircraft might be improved by the incorporation of a Ride Smoothing System. In fact, several conceptual studies of STOL designs (e.g., Reference 30) assumed that a Gust Alleviation and/or Ride Smoothing System would be an integral part of the aircraft design. Although several flight investigations of open-loop RSS performance have been conducted, no closed-loop systems have been so tested. It was the ultimate purpose of this research to provide such an evaluation for both a longitudinal and lateral Ride Smoothing System. Furthermore, previous designs often neglected to consider the effect of such systems on aircraft handling qualities, both in terms of stability and control characteristics. Such consideration is most important for STOL aircraft, since they will be expected to maneuver extensively in the airport terminal area. An evaluation of the interaction of the pilot with the RSS-augmented aircraft was, therefore, identified as a critical area in need of investigation. The most critical flight regime for piloted flight is the approach for landing. For this reason, the handling-qualities evaluation was conducted with the aircraft in the approach configura tion. By approaching the analysis and synthesis of a Ride Smoothing System from a comprehensive, systems engineering viewpoint, it was hoped that not onl'y the above major objectives could be accomplished, but a better understanding of inherent engineering trade-offs would be achieved.

CHAPTER II PROBLEM DEFINITION 2.1 Equations of Motion It is assumed that the motion of the aircraft can be adequately described by standard, linearized, separable, small perturbation of feedback equations of motion. In order to simplify the formulation quantities obtained from aircraft sensors (e.g., accelerometers), the respect to body fixed axes. The coefficients equations are written with form (see Appendix of these differential equations are in dimensional A). Derivations of the equations of motion can be found in any standard airplane flight mechanics (e.g., Reference 31). Validity of text on is subject to the following major assumptions: these expressions body; 1. The airframe is a rigid 2. The earth is an inertial reference frame; 3. The mass and mass distribution of the vehicle are constant; 4. The XZ plane is a plane of symmetry; 5. Disturbances from steady flight conditions are small; 6. Initial conditions are straightline flight with forces and moments balanced; perturbations forces and moments due to lateral 7. Longitudinal are negli'gible and vice versa; 8. The flow is quasi-steady; and 9. The effect of engine gyroscopics is negligible.

Furthermore, the airframe may be subject to forces and moments caused by control surface (direct lift flap, elevator, side force generator and rudder) deflections.. Thrust is assumed constant. The effects'of turbulence are included by assuming uniform immersion of the aircraft and applying the disturbances in terms of vertical and lateral velocity perturbations (w and vg) and the related angular 9 9.

velocity increments in pitch rate, roll rate and yaw rate (q , pg, and r ) at the center of gravity through the appropriate aerodynamic coefficients (32). The effect of the longitudinal turbulence, u,9 is neglected.

In matrix form, the resulting Laplace transformed equations of motion for the aircraft are: Longitudinal ( X)s * -(Xs + X) (- X + Wo)s + g cos0 u u. W w q 0 0 -Z.s - Z * (I - Z.)s - Z (-z -Uo)s + g sin 0 w u uW q 0 - M.s - M * - (M s + Mw) s -Ms u u w w q X~*f X e f w e ef f Z e Z f Z w Zq w M H e f w q L qg j q S6 (2.1.2) a sw - U q + (g sin 6o)6 (2.1.3) a' a - s 6 (2.1.4) z z x I Lateral - W S + g cos 0 UoS - y sin 60 0 o ) (s - Vv Vs vV T VT s s-L') - L' p/s -LS s(s Lp r - N- N s (s -N') r

S pr

Y6 r 6 Y6 sfg Yv 0 sfg L6 ' L 'sfg ' L L 9(2.1.5) r 3s fg pI g M N ' N N Np r 6 6 5 r pg v = VT 0 (2.1.6) 4, = p/s +(r/s) tan a (2.1.7) , = (1/Cos ec0)(r/s) (2.1.8) ay sv + U r - W p - gcos 00) (2.1.9) 0 0 = a a + I sr - l sp (2.] 10) z x y y derivatives (X,., X etc.) are defined in Appendix A.

The stability of Turbulence and Calculation of Atmospheric 2.2 Description Aircraft Response both random in time; being turbulence is generally Atmospheric Thus, the input-output and variable in intensity.

intermittent in terms to turbulence is described of aircraft response relationship A concise by random process theory.

quantities defined of statistical this theory as applicable-to of the important concepts of treatment in an article by Pratt (33). Short aircraft problem can be found the to satisfy certain statistical turbulence are assumed 'patches" of ergodicity, and homogeneity, isotropy, properties: stationarity, is Taylor's hypothesis the Gaussian sense. In addition, normality in frozen in space. Thus, turbulent velocity pattern is assumed valid: the the velocity exists between defined on the aircraft a relationship of turbulence.

spatial and spectral frequencies in terms for the input-output relationship Mathematical expressions of turbulence transfer quantities as well as definitions of statistical relationships, needed to given in Appendix B. Additional functions are loops, are developed of closing feedback control include the effect in the discussion.

as required 2.3 Ride Smoothing System Criteria 2.3.1 Passenger Comfort comfort of aircraft generally recognized that the It is factors (34); by numerous physical and psychological passengers is affected the important is believed to be one of these, the motion environment of comfort criteria for predicting Although no comprehensive variables.

is available, several mathematical models of subjective passenger response to aircraft motion have been developed by Jacobson, et al.

(35).

The simplest form, valid for motion dominated by vertical acceleration, predicts a comfort rating: C = 2 + 11.9 azrms + 7.6 a (2.3".1) where C = 1; Very Comfortable C = 2; Comfortable C = 3; Neutral C = 4; Uncomfortable C = 5; Very Uncomfortable and the acceleration levels are expressed in units of acceleration due to gravity (g's). This subjective reaction to an aircraft motion environment has also been correlated to passenger satisfaction with the "quality" of the ride (Figure 3).

2.3.2 Design Level of Turbulence Because all of the work discussed herein is concerned with an aircraft flying in the approach flight regime, the nominal aircraft operating altitude was defined as 365 meters (1000 feet). The corresponding characteristic gust lengths are L = 305 meters (1000 ft) w and L = 442 meters (1450 ft).

v A value of the root-mean-square vertical gust velocity corresponding to a 1% probability of exceedance was chosen as the standard; thus, o = 2.1 m/s (7 ft/sec) (Equation w B.8) and'av = 2.6 m/s (8.4 ft/sec) (Equation B.9).

g hi1 ww o'

a

o"q

MATHEMATICAL FIT TO THE CURVE CR < 3 CR).

(A- %-B-VB2-4C 2C A= -159/11 B= 26/55 C= -0.035/11 %= 162.5- 27.5CR CR k FIGURE 3. PASSENGER SATISFACTION CRITERIA (From Jacobson and Kuhlthau, unpubl ished data, 1973) 2.3.3 Surface Activity that 99% Gaussian process it can be shown For a zero mean to fall within + 2.6c, the time a random variable can be expected of In order not to violate the the standard deviation.

where a is of feedback control loops assumption of linearity, gains mathematical the must be limited such that Ride Smoothing System mechanization in a deflection does not exceed approximately root-mean-square control surface 38% of the available range.

Qualities 2.3.4 Handling standards for handling The current, industry accepted, contained in Military qualities of aircraft in smooth air are As pointed out by Barnes (37), the Specification F-8785B (36).

subject of handling qualities for flight requirements are vague on the applied MIL-F-8785B can, however, be in turbulence. The criteria of aircraft in order to determine baseline and RSS augmented to both the mode acceptable levels of aircraft dynamic compatibility with minimum For this purpose, the aircraft (e.g., wn, sp, TR' etc.).

parameters Smoothing System was for augmentation with a Ride under consideration low to medium assumed to fall in the Class II ("medium weight, adequate") flying category. Level I ("clearly maneuverability") for the category C ("terminal") flight phase.

qualities were sought of a Ride addition to the possible detrimental effect In on control System on the dynamics of an aircraft, the effect Smoothing evaluation of handling of concern (e.g., n/a). Thus, final power is piloted simulation using the Cooper- qualities must be accomplished in Harper criteria (38).

Modes Failure 2.3.5 system is subject to failure. In the Any automatic control argued that system operation of a Ride Smoothing System, it can be case safety. For to the integrity of the airframe or flight is not critical failures in non-self-monitoring mechanizations, however, unrecognized changes in the feedback systems could result in significant multi-loop Thus,, a system of this type must be aircraft stability characteristics.

alter handling to failure modes that do not catastrophically constrained in Compatibility with this requirement is again best tested qualities.

piloted simulation.

2.3.6 Feasibility design must be implementable. Few aircraft are Any system digital computing capability. Thus, equipped with an extensive onboard be met with analog signal processing requirement must any system command and reliability with most engineering solutions, feasibility devices. As of simplicity Smoothing System is to a great degree a function of a Ride criterion, feasibility is and economy of design. As a quantifiable of engineering design.

difficult to describe--it is the art Index 2.3.7 The Optimal Control Performance theory performance index is customarily The optimal control of the weighted sum of squared state variables.

expressed as an integral to either the longitudinal Ride Most optimal control theory solutions design problems have included a Smoothing System or Gust Alleviation integrands in the performance combination of a , 6, or 6f as the z that not the preceding discussion, it should be evident index. From Although minimization of a is all design criteria are so satisfied.

z is not an absolute prerequisite a desirable goal, total alleviation in for satisfactory system performance. For any gaven aircraft flying of alleviation compatible with a given level of turbulence, only alevel Further the passenger comfort (satisfaction) criterion need be provided.

especially in with the handling qualities criteria, more, compatibility cannot be adequately included in the classical a system failure mode, Finally, optimal filters, in the case of performance index formulation.

to all control surfaces, tend to be feedback of all state variables by analog devices. For a Ride Smoothing overly complex for mechanization to attenuate rigid body response to turbulence, System, i.e., one proposed criteria dictates a classical successful design to the above-mentioned control theory, approach.

(suboptimal), rather than optimal Control Surfaces, and Feedbacks 2.4 Selection of Sensors.

the Having, in the interests of design simplicity, chosen to l.imit of feedback loops, the system analyst/designer is faced with the number which control surfaces to use and deciding what signals task of choosing rational are needed to implement a useful feedback control law. A has been proposed by Stapleford, et al (39).

approach to this problem feedbacks.

involves the identification of essential The technique Quoting: or "The essential feedbacks...derive from one both of two basic flight control system purposes: * To establish and maintain certain states of specified equiltbrium vehicle.,motion.

" To remedy aircraft handling quality deficiencies.

The establishment and maintenance of an equilibrium of motion requires an outer control loop state to the vehicle, motion quantity defining pertinent 39, page 8.)

state." (Reference that A Ride Smoothing System essentially fits the above definition. Note that an inner control loop (of a multi-loop feedback the implication required to satisfy handling qualities requirements.

system) may be For the longitudinal Ride Smoothing System problem, i.e., the outer loop closures reduction of a response to vertical gusts, three z are possible (see Table I).

No equivalent guidelines are available for a lateral Ride Smoothing System.design. Lateral autopilot functions have classically involved hold devices the use of yaw dampers (r + - feedback) or roll attitude r + a feedback) to reduce aircraft response to turbulence. The (4 or p (22) recent Ride Smoothing System Feasibility Study by Gordon and Dodson reports on the performance of a lateral system using yaw rate and c.g.

transverse acceleration feedback to the rudder (r and ay 6 ). The encountered with such a mechanization is explained by major difficulty conflicting requirements on the rudder: significant side force cannot yawing moments that counter the be generated without inducing large aircraft's natural tendency to weather-vane into the remote wind. Thus, to turbulence can successfully be lateral acceleration in response the a given fuselage station. Application of suppressed only at concept points to a solution to this dilemma: essential feedback acceleration to a pure side-force generating feedback of transverse control surface (ay - 6 sf). Clearly, any number of other feedback loops might serve to implement a Ride Smoothing System.

I TABLE LONGITUDINAL COMPETING SYSTEMS Primary Function Feedback Performed Equalization Requirements Practical Design Problems a e6 1. Increase Cp and wn I. Gain Ka l 1. Severe gain adjustment with Z es sp VT LCm flight condition.

~ 2 R due ad Sensor location adequate for e 2. all Reduce h and a z flight conditions, S2.

element to gusts 2. Lead/lag response mode feedback.

3. Structural desirable

4. Increase e response to

vertical gusts.

a- 1. Increase and to I. Gain K. c - I. Gain adjustment with flight e sp e condition.

p 2.'Reduce h and a 2. Sensor instrumentation z 2. Lead/lag element response to gusts desirable a. Determination of operating point.

b. Errors due to aerodynamic interference.

c. Elaborate sensor complex gust required to suppress inputs.

a + dl 1. Reduce h and a 1. Gain K = m I. Severe gain adjustment with z 3 C CC flight condition.

response d a 1f to gusts 2.

Sensor location adequate F L for conditions.

flight af aall 2. Crossfeed mode feedback.

3. Structural to adjust desirable effective 14N /Z 4. Probable drag penalty due to _dlc dlc I direct lift control surface, (Reference 39, page 14)

CHAPTER I I I ANALYSIS AND SYNTHESIS Order of Presentation 3.1 and synthesis of Ride In this section of the report, the analysis of Chapter il is presented.

Systems consistent with the criteria Smoothing contained in Section 3.4. Lateral of longitudinal systems is Development 3.5.

are discussed in Section systems 3.2 Demonstration Aircraft Ride Smoothing the flight evaluation of a closed-loop To provide chosen.

Airborne Simulator (GPAS) was System, the NASA General Purpose for light utility transport modified GPAS Is a Lockheed JetStar (C-140) Laboratory, stability experiments by the Cornell Aeronautical variable (Figures 4 and 5).

Inc.

possible: model following GPAS modes of operation are Two basic study, the basic Jetstar feedback (40). For this and response capability Thus, the model-following used as the model aircraft.

was system of the response feedback and only some elements was not required package (accelerometers, attitude used. These Included the sensor were Associates, Inc.

computer (Electronic rate gyros) and onboard analog and surfaces of the aircraft the fully-powered control PC-12). All of and side-force generators) flaps, ailerons, rudder (elevator, direct-lift system.

the response feedback can be coimmanded by the power condition was for the aircraft in The RSS design flight previously, because an Instrument configuration. As mentioned approach N25 * NO? h L~oj) PRFIJ)hNG(PPAGR BLANK

OI

ZI

QI

oI

zI

wI

U, iii 4i =I iii iii iii iamii~i i iili~iiliiii iN~iliIi Cilii 1-J SYSTEM - .. ~ DATA ACQUISITION LIFT IRECT -ANALOG COMPUTER C TEST ENGINEER CONSO SIMULATION SYSTEM b - !SID~ SCs ELECTRON!

P ILOT SBETTEST SFORCE GENERATOR PURPOSE AIRBORNE SIMULATOR FIGURE 5. SCHEMATIC OF NASA GENERAL landing approach is the most difficult flight phase from a pilot's point of view, it is the best condition for evaluation of aircraft stability Operational parameters, aircraft handling qualities.

dynamics for this configuration derivatives, and control surface actuator Appendix C.

and flight condition are summarized in 3.3 Method of Analysis of this research Throughout the analysis and synthesis portion digital computer program "CONTROL" written extensive use was made of the by J. W. Edwards of NASA Flight Research Center. "CONTROL" permits response, analysis of open- and closed-loop continuous systems by frequency transient response, and root locus techniques. The plant is specified loops and equalization may be in state variable form, but the feedback diagram (frequency domain) form.

specified in block 3.4 Longitudinal Ride Smoothing Systems 3.4.1 The Basic JetStar--Longitudinal Case In the power approach configuration, the longitudinal of the basic JetStar are characterized by the following dynamics parameters: J 1.3948 Short Period Mode: sp = -0.9123 + (0.35) sp = 0.546 W = 0.266 Hz (0.11) n sp Phugoid mode: Xph = -0.00923 + j 0.1714 (0.04) = 0.054 Cph P = 36.6 sec T, = 74.8 sec.

Control Authority (see Figure 6): = 6.22 g /rad. (2.0) values of the given param The numbers in parenthesis refer to minimum as specified in MIL-F-8785B (36). The basic aircraft clearly eter handling qualities specifications. Only the meets all longitudinal mode damping is marginal.

phugoid i At the design turbulence condition (a = 2.1 m/sec), the w g was computed to be Ya = 0.1178 g.

root-mean-square vertical acceleration z Throughout this report, calculation of root-mean-square values is accomplished by integrating the appropriate power spectra over the frequency range of interest: 0.01 < w < 100.0 rad/sec. The mean-square (power spectra) is depicted in acceleration distribution by frequency Figure 7.

Although the comfort model (Equation 2.3.1) is given only in terms of total a , it is known that, depending on the frequency band z which oscillatory excitation occurs, the effect on human comfort over is quite different (34). Low frequency oscillations tend to cause organs, leading to annoyance and motion sickness. Resonance of body frequency range between 2 and 8 Hertz. For pain, is possible in the percentage of the total mean-square the JetStar, a significant acceleration is in the low frequency range. Consider the partitioned power spectrum for the basic JetStar (Figure 8). The "power" in To L U 1.0 0.1 1.0 n/a (9's/rad) FIGURE 6. HANDLING QUALITIES SPECIFICATION FOR n/, - I - 2 _ 10_ u - 'U IO-5 - 6 1O 10-7 1. 0 .1 0.0 1 0. 0 1 0 (Hertz) SPECTRA OF a DUE TO TURBULENCE FIGURE 7. POWER z FOR BASIC JETSTAR - 2 "I 1-3,_ I-) - 4 1o N , - 5 SHORT PHUGOID PERIOD PEAK PEAK - 6 i0 10-7L I0 1.0 0.I 0.01 0.001 en (Hertz) FIGURE 8.

PARTITIONED POWER SPECTRA OF a FOR BASIC JETSTAR z the lowest frequency band (phugoid peak) is approximately 38% of the total. Only 9% of the total mean-square acceleration occurs at frequency above I Hz; the remainder is concentrated in the short-period peak.

3.4.2 Baseline Longitudinal Ride Smoothing System a baseline Based on the concept of an essential feedback, employing feedback of vertical.

longitudinal Ride Smoothing System acceleration to the direct-lift flaps was analyzed. In simplified block diagram form: TURBULENCE c,,,.. ACTUATOR 'e z

i DYNAMICS

AIRCRAFT <" DYNAMICS c 6f -f oa DYNAMICS FIGURE 9. BASELINE LONGITUDINAL RIDE SMOOTHING SYSTEM a z 6 f The feedback loop has associated with it only a gain: (no K a equalization). Performance of this system at the design turbulence is summarized in Table II.

condition TABLE II PERFORMANCE OF BASELINE LONGITUDINAL RSS a 4 z K az 2 az z lf n/a a 2 2 (radlm/sec ) (rad/ft/sec ) (9) % Alleviation (0) (g/rad) 0 0.1178 o 0 6.22 0.1 0.03 0.1024 13 5.6 3.94 0.2 O.06 0.0938 20 10.2 2.91 0.3 0.09 0.0892 24 14.5 2.28 o.4 0.12 0.0903 23 19.6 1.88 The locus of roots of the aircraft characteristic equation for this system is presented as a function of feedback gain K in a z Figure 10.

Several deficiencies in the simple a 6f system are immediately apparent. At reasonable levels of flap activity (a6f Z 100), the degree of vertical acceleration alleviation is small. Both the natural frequency of the short-period mode and the magnitude of the handling qualities parameter n/ are rapidly reduced to marginal values as K is increased. Phugoid damping remains low.

a z Consequently, an "inner" loop closure to augment the essentlal feedback is indicated.

3.4.3 Effect of Inner Loop Closure A number of feedbacks will serve to increase short-period frequency: angle of attack to elevator (at 6 ), pitch attitude to elevator (0 + ), or normal acceleration to elevator (a or a ' + 6 ).

e z e The first of these, a 6e, can be eliminated from consideration because of the diff.iculty in providing an adequate sensor. Feedback of a * z e has a minor effect on phugoid damping and tends to increase pitch response to turbulence (31)(39). The best compromise appears to be incorporation of the classical "pitch damper" or 0 + 6 e feedback. The root locus for this closure is depicted in Figure 11.

3.4.4 Basic Multi-loop Longitudinal Ride Smoothing System The basic multi-loop longitudinal Ride Smoothing System in simplified block diagram form is depicted below: jo) -1.6 -1.4 1.2 0.1 1.0 0.2 K a Z 2 0.8 (rad/m/sec ) 0.3 -0.6

(a

F.2 G .4 i.O 1.4 R.8 0.8 FIGURE 10. ROOT LOCUS FOR BASELINE LONGITUDINAL RSS j-JO 2.6 1.4 2.4 1.2 2.2 1.0 0(0) 2.0 0.8 o.6 1.8 o.4 1.6 0.2 1.4 1.2 1.0 0.8 ----------- 0.2 0.1 1.4 1.2 0.8 0.2 0.1 11. ROOT LOCUS FOR 0 6 LOOP CLOSURE FIGURE e TURBULENCE _ _ _ Iot e ACTUATOR e _ < DYNAMICS -AIRCRAFT - DYNAMICS, -C) z e a k c ACTUATOR 6f ,Z'. 0 DYNAMICS C'

'S

K2> FIGURE 12. BASIC LONGITUDINAL RIDE SMOOTHING SYSTEM and K are pure gains. K e a z The effect on aircraft short period and phugoid dynamics is presented as a function of the two feedback gains in Figure 13. Note that the phugoid mode is rapidly stabilized by feedback of 0 for any value of K Any desired value of short-period frequency can also be a z attained. Some degradation in the short-period damping ratio, however, results..

Performance of the Ride Smoothing System, in terms of percent reduction in ra and q, is depicted as a function of the feedback gains z in Figures 14 and 15. Root-mean-square direct-lift flap activity is similarly presented in Figure 16. Maximum permissible root-mean-square flap deflection, consistent with the constraint of Section 2.3.3, is 100.

No plot is presented for elevator activity since 0 < 1.20 for all e levels of feedback gains considered and is thus well within available jw 2.6 2.4 1.6 2.2

Ke

/

1.4

2.0

/ 1.2 1.8

1.0 1.6 o.8 1.4 0.6 .1.2 0.1 KE o.4 ! j.o Ka 2 0.2 (rad/m/sec ) 0.2 0.8 0.3 o.4 0.2 0.1

-,a

1.2 1.o o.8 0.2 0.1 FIGURE 13. ROOT LOCUS FOR BASIC LONGITUDINAL RSS K K o.4 o.8 1.0 1.6 2.0 20 bu ao (rO/rns)c) 3 o 0.15 'IN (rad/m/se /0.2 - 40-o 6o -50 0.4 L60

o.4

zz FIGURE 14. PERFORMANCE OF BASIC LONGITUDINAL RSS; or AS A FUNCTION a OF K ANDK 8 a 4o K o.4 0.8/ 1.2 1.6 2.0 - -10 K0.

zb a

-20 g

(rad/m/sec 0.230

-

-4o OF BASIC LONGITUDINAL RSS; FIGURE 15. PERFORMANCE a AS A FUNCTION OF K ANDl 10 K 20 .4 e 0O20 1.6 2.0 o.4 0.8 1.2 5' K a 8 b z (rad/m/seC ) 0.1 FIGURE 16. 'PERFORMANCE OF BASIC LONGITUDINAL RSS; AS A FUNCTION OF K AND K JetStar limits (-200 <6< + 160). Figure 17 was constructed by superimposing the flap deflection criteria and lines of constant short (w n ) on Figure 14. The period damping ratio ( sp) and frequency sp sp resulting surface can be interpreted as a rudimentary graphical representation of a RSS performance index. By referring to this plot, the system designer can choose any combination of gains K and K to a e z minimize a and simultaneously satisfy a handling-qualities criterion a z mode characteristics. Note that no combination based on short-period of feedback gains will permit a return to the free aircraft short-period characteristics. The limit on permissible K as established by surface a z activity considerations is also shown.

In order to maximize the performance of this system, K a z should be chosen so as to take complete advantage of the available direct-lift'flap authority. Choice of K is then limited to a narrow band 0.4 < K < 0.5; the lower limit being based on system performance e considerations, the upper, on handling qualities criteria (Csp > 0.35).

K = 0.26 rad/m/sec , A typical design point might then be chosen at a a z r, = 0.4 rad/rad/sec resulting in a 41% reduction in oa .

z Longitudinal Ride Smoothing System 3.4.5 Analytic Model of the Significant insight into the mechanisms underlying the baseline longitudinal RSS can be gained by examining performance of for aircraft root-mean-square vertical a simplified analytic expression acceleration due to turbulence. As the first step in the derivation, D).

the appropriate aircraft transfer function is required (see Appendix is conceptually straightforward, the resulting Although the development Considerable simplification results equations are extremely lengthy.

= Csp Constant 3 = Constant ....

nsp (O/O0 -10 N 4 2.0 2 1.2 1'.

0.8 0.10.4 K0 a z 30 .0 (rad/m/sec L .

-40 -50- L 60 0.4.1 0.4 FIGURE 17. PERFORMANCE OF BASIC LONGITUDINAL RSS; AS A CONSTRAINED FUNCTION OF K AND K a z z assumptions are made: if several acceleration, pitch gust, q , on vertical I. Effect of vertical compared to the effect of az, is small as gust, w ; can be approximated of the aircraft 2. The dynamics equal to the phugoid mode frequency by setting zero; and perfect.

3. All actuators are in aerodynamics only) terms (based on JetStar If only the highest order the transfer function variable are retained, each power of the Laplace of the aircraft stability can be written in terms for a due to w z 9 derivatives as K aM6) _ Mq s - s2(s a (3.4.1) q 2} KST 2 G Gw =KS s( 2~ m + 2)p~ sp sp static gain, KST' is where the Z (3.4.2) (I - K Z ) KST is given by the short-period damping (3.4.3) - Mq -T -2asp = spwn sp natural frequency and the short-period (3.4.4) Kaz U ) = _U-M _ KoM + KST(Mq n 0 w 8 f45 e z nsp With the additional assumption that the transfer function of w due to turbulence, A, can be approximated by a first-order filter GA - (3.4.5) we can write s -M s -KoM63 a Iq 06e GAZ :ST s (s + 2 sp n sp S + W0n s

2 GA KS l ~s+~~hi~2) ' (3.4.6)

p and (from Equation B.2)

2 = FIGZI

Ga (Adw (3.4.7) A z The integration can be performed in the complex plane to yield: KST (KOM 6)2 a m

{

nsp n + r (KsM )2 (4 2 )- 2 + M + 2KM 4sp 0 e sp ns 5p q e] (3.4.

where O is the lower limit for a truncated input turbulence (white noise) power spectra.

Several comments about the deficiencies of this approximation are in order. Note that the expression for asp (Equation 3.3.3) does not properly account for the reduction in short-period damping with increasing K. (Figure 13).

A value of wo > 0 is clearly required for the integration to be bounded.

If t0 is arbitrarily chosen.as w0 = 0.56 rad/sec, the approximation predicts system performance in good agreement with the digitally-calculated results (Figure 18).

The critical parameters affecting the acceleration alleviation capability of this system are the constants outside the radical: K ST and W)n 2 From the definition of KST (Equation 3.4.2), sp it is clear that, for a given lift curve slope Z, the overall system performance is determined by the flap effectiveness term Zf' Conceptually, this conclusion is intuitively obvious. The fact that system performance is improved as wP is increased can be explained n sp by considering the exact input power spectra (Figure 19). Above the break frequency, ' = 0.236 rad/sec, the input power decays at the rate of 40 dB/decade; thus, the higher the aircraft effective short-period resonant frequency, the lower the magnitude of response to turbulence.

Finally, the effect of changing the damping ratio of the short-period mode, Csp, is contained within the second term inside the radical. At =sp 0.5, this term contributes nothing to a ; for sp < 0.5, however, z the magnitude of a is increased as K is increased.

z I Note that the performance of the baseline longitudinal RSS, to first order, depends only on a single dynamic derivative: Mq.

q Generally, dynamic derivatives are more difficult to estimate or measure than static stability derivatives. In order to be most successful, any RSS design should be minimally sensitive to errors in estimation of the plant parameters. Calculations, based on the simplified model, showed that variations of + 25% in the magnitude of Mq resulted in less than 1% change in the performance of the baseline longitudinal RSS.

, 2.0 1.

.

0.1 .2 0.0.8 K a z (rad/m/sec )/ " / 0.2' o.4 • - ANALYT IC DIG I TAL-- CALCULATED a FIGURE 18. COMPARISON OF DIGITALLY z WITH ANALYTIC EXPRESSION 00 0 < E 10"3 1-01 10 0101 10-2 w (rad/sec) FOR w DUE TO A 19. POWER SPECTRA FIGURE 1 4 I System Ride Smoothing Longitudinal 3.4.6 a measure longitudinal RSS provides Although the baseline criteria, several while meeting design acceleration alleviation of results First, the simple mechanization are desirable.

improvements of feedback of re. Secondly, no choice limits for the choice in narrow aircraft short-period characteristics.

recovery of the basic gain permits param as represented by the control of flight path, Finally, the pilots is degraded.

eter n/, in the feedback paths of proper equalization The inclusion improving system perform shortcomings while can eliminate all of these For the two feedback + feedback of a 6 f.

Consider first the ance.

, (see Appendix D): it can be shown that

systems (a + af, e + 6e)

loop z n~U0 -Z n 0 w (g/rad) (3.4.9) + f (washout) is included in the a If a filter of the form w basic will be the same as for the the steady state n/ feedback path, a pure gain in a with K than for A more rapid decrease aircraft. sp a Z s +a (s+a a < b) Introduction of a lag filter feedback, however, results.

s + b trend. A to offset this undesirable with the washout tends in series is shown the effect of these filters root locus depicting short-period as to permit the were chosen so 20. The filter parameters in Figure of lag circuits, analog circuits. In the case construction of feasible or equal to be greater than of a/b is customarily rest.ricted the ratio considerations.

to 0.1 by circuit noise az 1.3 0.)0 (rad/m/sec 1.2 WAS HOUT sec 2.0= -1.0 3.0 0.20 2.0 0.9 LAG l+s/l -0.8 0.3 -0.7 3.0 _o.6 PURE GAIN -0.5 LAG AND WASHOUT - ~ l 1.0 0.9 0.8 0.7 0.6 0.5 FIGURE 20. ROOT LOCUS.OF EFFECT ON SHORT-PERIOD DYNAMICS OF FILTERS IN a + f FEEDBACK:'LOOP Additional short-period damping can be provided-by incorpor ating a lead filter (- d > c) in the 0 6e feedback loop. In essence, this filter provides a pseudo-differentiation or feedback of a component of pitch rate at very low frequencies (Figure 21) The to maximum ratio c/d is again limited by feasibility considerations values equal to or less than 10.0.

break For both the lead and lag filters, the desired frequencies were determined by inspecting a number of root locus plots.

from the incorporation of these filters is depicted The system resulting in Figure 22 and designated Longitudinal Ride in block diagram form Smoothing System I. System performance surfaces, equivalent to those are presented in Figures 23 through 26.

shown for the baseline system The complete root locus carpet plot is given in Figure 27.

Based on the procedures outlined in Section 3.3.4, the system design point was chosen at K = 3.3, K6 = 0.14. The aircraft dynamics a z and system performance parameters for this configuration at a = 2.1 w m/sec are summarized in Table 11.

Figure 28 compares the power spectral density of vertical basic and longitudinal RSS I augmented aircraft.

acceleration for the Note that the response to turbulence is heavily attenuated at both the as in the range above the phugoid and short-period frequencies, as well in a narrow frequency band, short-period peak. A slight amplification however, results.

Compared to the baseline system, Longitudinal RSS I is both in terms of acceleration alleviation capability clearly superior and closed-loop short-period mode characteristics. As pointed out it'] LEAD LEAD + / ( 1 +s I PURE l+s1_2 "T...l. GAIN 1+-2.2 .10 0.12 1.0 0.10 0.08 - 2.0 0.8 0.08 .0o6 K o 0.4 .

0.04.

1.6 0.2 0.02 1.4 I I I I I I 1.2 1.0 0.8 0.6 0.4 0.2 OF A FIGURE 21. ROOT LOCUS OF EFFECT ON SHORT-PERIOD DYNAMICS LEAD FILTER INO -+-S e FEEDBACK LOOP TURBULENCE epl e e 6 < epilot G e AIRCRAFT w.

L ec DYNAMICS > z a + I- s (l+s/.I) s +- ' FIGURE 22.

LONGITUDINAL RIDE SMOOTHING SYSTEM I 2 , (Design Point: Ka = 3.3 rad/m/sec K 0.14 0/0) a ' 0 0 0.04 20 oao K 0.12 z 2 (rad/m/sec 2 4.060 LONGITUDINAL RSS I; FIGURE 23. PERFORMANCE OF AS A FUNCTION OF K AND K C~ a a z K 0 K a 1.0 0.08 z 20 (rad/n/sec2 0.12 4t -40-a -60 5.0 70 FIGURE 24. PERFORMANCE OF LONGITUDINAL RSS I; OF Ka AND K aq AS A FUNCTION z 5.0 4.o .64.2 2 6 0 ' 0 8 0 .1 o.1 / .

K az z (rad/m/sec b 2.0 FIGURE 25. PERFORMANCE OF LONGITUDINAL RSS I; r f AS A FUNCTION OF Kaz AND K.

-s = Constant = Constant 0 0 o -20 ..... msp N 0 Design Point 1.( 02 ",.0.o8 . 30 ._ -4o 2.0 Ka z -50 (rad/m/sec) L60 RSS 1; FIGURE 26. PERFORMANCE OF LONGITUDINAL OF K AND K AS A CONSTRAINED FUNCTION z z 2.6 0.20 0.16 K, 2.2 (0/0) 0.12 0.08 1.8 O.04 1.4 1.0 1.0 K 2.0 ) (rad/m/sec 3.0 0.8 4.0-- - - - - - - - - -Kaz 0.20 -0.15 -0.I0 0.05 I I J I I I " - - I I 1.4 1.0 o.8 0.15 0.1 0.05 FIGURE 27. ROOT LOCUS FOR LONGITUDINAL RSS I TABLE III I SYSTEM RIDE SMOOTHING LONGITUDINAL OF CHARACTERISTICS ,4 - Longitudinal RSS I Basic JetStar 0.567 0.546 p Hz 0.356 Hz 0.266 Wn sp 0.054 0.522 ph 53.2 sec 36.6 sec ph P sec 9.6 74.8 sec Ti ph Ca 0.1178 0.0572 g z 0.0040 g g 0.0112 Fa x 0.70 /sec 1.44 /sec aq 9.9 0 S-- 0.40 6e ar 51.8% reduction % a z 64.6% aa % reduction x 51.3% % reduction aq 1 0 ° - 2 i0 " -3_ BASC JETSTAR '5 RSS I - 6 .

Io -5_ 10-7,1 10 0.001 0.01 0.1 1.0 (Hertz) COMPARISON OF a POWER SPECTRA FOR BASIC AND FIGURE 28.

LONGITUDINAL RSS I AUGMENTED JETSTAR previously, the presence of a washout circuit in the az - 6f feedback path prevents degradation of the handling qualities parameter n/a. From the pilot's point of view, the only noticeable effects of the RSS might be some reduction in the speed of normal acceleration response to a stick input and the slight ly greater stick deflection required to produce a given change in pitch attitude. The degree of these potential handling qualities problems was left to be considered in the ground-based simulation phase of this research.

Also to be evaluated in simulation was the effect of a failure of the stabilizing 0 6 e feedback on the controllability of the augmented e vehicle. From the root locus diagram (Figure 27), it is clear that with only the acceleration feedback operational, the short-period natural frequency would drop to marginal values ((n = 0.14 Hz).

sp 3.4.7 Longitudinal Ride Smoothing System II An alternate mechanization, designated Longitudinal RSS I, is depicted in block diagram form in Figure 29. It differs from the previous system only in the form of equalization in the a + 6f feedback z path. A Bode magnitude plot of this filter is given in Figure 30. At the phugoid frequency, this circuit acts to heavily attenuate the feedback signal (notch filter). For all other frequencies, the magnitude response characteristics are similar to that of the lag filter used in System I. The lightly damped quadratic numerator of the filter introduces a pair of stable, very low-frequency roots which help delay the onset of short-period instability as K is increased. When the inner, 0 e' a z loop is closed, this artificially-introduced mode as well as the phugoid TURBULENCE epilot e-- G OGe DNMC ec AIRCRAFT ' C , 6f c z 2) 6S+34S (1+0 K--- az (z+9.0s+340s K Jl+s/O'l5)1 K0 (l+s/l.5) FIGURE 29. LONGITUDINAL RIDE SMOOTHING SYSTEM II 2 , (Design Point: Kaz = 3.3 rad/m/sec K = 0.1 °/0) LAG FILTER----- 0% -16 -24 -32 -4o 0.01 0.I 1.0 10 l00 (r ad/sec) FIGURE 30. BODE MAGNITUDE PLOT OF NOTCH-FILTER are rapidly stabilized. The effect of the inner loop closure on the short-period roots is almost identical as for System I (Figure 31).

In order to provide a comparison of Systems I and I, the same value, K. = 3.3, was selected for the design point. By setting z the pitch attitude feedback gain at Ke = 0.1, the short-period damping ratio is made approximately equivalent to that of the basic aircraft.

Table IV compares the key metrics of the basic and RSS augmented JetStat.

Although System ii appears, from Table IV, to be somewhat inferior to System I in all respects, an examination of the power spectral density plots shows that the alleviation capability of System II is almost identical to that of System I for frequencies above the phugoid peak (Figure 32). Thus, the only major difference between the two mechanizations is in the handling qualities parameter n/. Since handling qualities criteria were postulated as an important consider ation in the design of Ride Smoothing Systems, both System I and II were retained for simulation experiments where pilot opinion was solicited.

As with System I, failure of the 6e feedback loop will cause m sp to be reduced to a marginal value, and simulator studies were carried out to evaluate the severity of this deficiency.

3.5 Lateral Ride Smooth'ing Systems 3.5.1 The Basic JetStar--Lateral Case The lateral dynamics of the basic JetStar in the approach configuration are characterized by the following parameters: 1.4 0.07 1.2 0.13 0.16 K 1.0 a z 0.20 (rad/m/sec2) 3.3 0.26 0.8 0.33 5.0 - 0.6 NOTCH FILTER PURE - o.4 GAIN _0.2 I I I I I 0.8 o.6 o.4 0.2 1.0 FIGURE 31. ROOT LOCUS OF EFFECT ON SHORT-PERIOD DYNAMICS OF A NOTCH FILTER IN a 6 f FEEDBACK LOOP TABLE IV RIDE SMOOTHING SYSTEM II CHARACTERISTICS OF LONGITUDINAL Longitudinal Longitudinal Basic JetStar RSS I RSS II 0.546 0.567 0.534 sp 0.312 Hz w0 0.266 Hz 0.356 Hz n sp 0.158 0.522 " o.o54 ph 52.9 sec 36.6 sec 53.2 sec pph 9.6 sec 36.8 sec T_ 74.8 sec ph g/rad 6.22 g/rad 4.03 6.22 g/rad n/a

a 0.1178 § 0.0572 g D.0607 g

a g 0.0040 g 0.00454 G 0.0112 g a xoo 12.20 97 97 -- Sf0 a -- 0.4 0 0.4 o e 49.4% % reduction a -- 51.8% - z % reduction a -- 64.6% 59.5% x 10 -3 -- BASIC JETSTAR - N 10 ' I0-5- _ 0-7 RSS II I I°-6-_ - 0.001 0.01 0.1 1.0 10 c (Hertz) FIGURE 32. COMPARISON OF a POWER SPECTRA FOR BASIC AND z LONGITUDINAL RSS II AUGMENTED JETSTAR

Dutch Roll Mode: Xdr = -0.0615 ± 1 1.36

o.o45 (C > 0.08) Cdr = W = 1.36 rad/sec (n > 0.4) nr Cdr ndr = 0.061 rad/sec (ga > 0.15) Roll Subsidence: Ti = 0.87 sec (TR < 1.0).

Ti = 0.61 sec '27 R Spiral Mode: Ti = 418 sec (T > 20) . -f where the inequalities in brackets are criteria of MIL-F-8785B (36).

Note that the Dutch Roll mode damping fails to meet these requirements.

At the design turbulence level a = 2.65 m/sec (8.45 ft/sec), V g a = 0.0312 g. As in the longitudinal case, the transverse acceleration a y power spectral densi.ty was integrated over the frequency range 0.01 < W < 100.0 rad/sec (Figure 33).

Lateral Ride Smoothing System 3.5.2 Compared to the longitudinal case, mechanization of a lateral ride smoothing system is considerably easier. The essential, feedback is transverse acceleration., The obvious control surface is a pure transverse force control; i.e., the outer loop becomes lateral acceleration to side force generator deflection ay Asfg number of inner loop closures are possible,,' but since the aircraft exhibits insufficient Dutch Roll damping, a yaw damper (r + 6r) is the conventional solution. Also customary is the inclusion of a washout circuit in the r + 6r feedback path so that pilot commands to the rudder are not suppressed.

- _ i0 i0-35 10-4 10-7 -8 io 0.1 0.01 0.001 w(Hertz) TO TURBULENCE yDUE OF a SPECTRA POWER 33.

FIGURE JETSTAR FOR BASIC The resulting system is depicted in block diagram form in Figure 34. Note that the washout time constant was chosen as w =1 sec.

Increasing To tends to increase Cdr at the expense of ZR without significantly altering system alleviation performance.

The locus of Dutch Roll roots is plotted as a function of K and K in Figure 35.

Note that the a . 8s feedback has almost R a y R 01 y sfg no effect on either Wnd or Cdr' whereas r 6r increases dr while r slightly lowering m . The effect of the two feedbacks on the roll nd r subsidence and spiral modes is summarized in Table V.

The effectiveness of the Lateral Ride Smoothing System in ) terms of reduction of root-mean-square lateral acceleration (a a yaw Y rate (or), and roll rate (cp ) is presented graphically as a function of feedback gains K and K in Figures 36 through 38. Root-mean-square a r Y side-force generator activity is similarly presented in Figure 39. The limit on permissible side-force generator activity, determined by linearity considerations, is o sfg L 9 . A system performance surface, with the limits o6sfg = 90, ( Con)dr=0.1 superimposed, is presented as Figure 40. As in the case of the Longitudinal RSS, this surface allows the designer to choose feedback gains that satisfy all design criteria.

For this study, the selected design point was for K = -3.3 a rad/m/sec (1.0 rad/ft/sec ), K = I rad/rad/sec. The aircraft dynamics r and system performance parameters for this choice of feedback gains are summarized in Table VI.

A comparison of the power spectral density of lateral acceleration in response to turbulence with the RSS on and off is shown in Figure 41. Alleviation is provided over the entire range of TURBULENCE 6r 6 r < pi lot c Gr r AIRCRAFT r DYNAMICS ZC c G sfg X 6sfg c ay0p ...

Kaa C Ks Kr s+l SYSTEM FIGURE 34. LATERAL RIDE SMOOTHING (Design Point: Ka -3.3 rad/m/sec2, K = 1 °/°/sec) Y jt 1.35 - 0.3 K 1.30 - r (0/0/sec) o 0 .9 ', - 1 2 5 11.2 - 1.20 1.5 -1.15 K y -1, (rad/m/sec 2) - 1.10 -5.0 - 1.05 0.05 0,25 0.20 0.15 0.10 RSS LOCUS FOR LATERAL 35. DUTCH ROLL ROOT FIGURE TABLE V EFFECT OF FEEDBACKS ON ROLL SUBSIDENCE AND SPIRAL MODES Variable Fixed Ka0 Variable 0 Fixed K \ >0 >0 >0 a a 0 .

y \ 0 + Fixed + Fixed >0 >0 Variable Variable + >0 >0 where ++ stabilizing effect - + destabilizing effect hlAn + nn effrct K r 0 (o/o/see) 0.03 - 20 9 b -1l . 2 1 .5 - 4 0 .0 -3.(0 -4.w -5.00 LATERAL RSS; FIGURE 36. PERFORMANCE OF OF K AND K a AS A FUNCTION a a r Y Y -- lO K a -0

y -1.0

(rad/m/sec - K / //ec r ) -3.

6 20 -5.0 .2 30 .5 L-50 .FIGURE 37. PERFORMANCE OF LATERAL RSS; AS A FUNCTION OF K AND K r a r V K0 r (°/°/sec) 0.3 ,20 a Kb .6 a o-3.0 80 -5.0 FIGURE 38. PERFORMANCE OF LATERAL RSS; a AS A FUNCTION OF K AND K -16 K KK ay -4.0 -12 (rad/m/sec .5 -3.

-2.04 -1.0 -9 FIGURE 39. PERFORMANCE OF LATERAL RSS; a AS A FUNCTION OF K AND K 6 r 8sfg - r 78y r * DESIGN POINT (°/°/sec) -20 0.16 -(wnddr= ~b 40 r- ~-1.0 Ks L K a y (rad/m/sec ) 2.0 ]00 -3.0 -4.o FIUEk.PEFRAC50.AEA0RS SA OSRIE A CONSTRAINED RSS; a AS :OF LATERAL FIGURE 4. PERFORMANCE Y K AND K FUNCTION OF a r y TABLE VI OF LATERAL RIDE SMOOTHING SYSTEM CHARACTERISTICS Lateral Basic JetStar RSS 0.045 0.155 dr 1.36 rad/sec 1.195 rad/sec n dr sec sec 0.61 0.87 TR 0.42 sec Ti 0.61 sec '2 R (37.5) sec T, (T 2) 418 sec O 0.0312 g 0.0047 y a 1.56 °/sec Y 2.3.5 /sec r 1.95 °/sec a 5.01 °/sec p a -- - 7.8 o 0.92 0 a r % reduction a 84.5% a Y % reduction a Y 43.5% r % reduction a 61.0% 10-3.

BASIC JETSTAR - 4 - 0-5- RSS LTRL U\ I0-7_ _, 0-8 0.1 1.0 0.001 0.01 w(Hertz) FOR BASIC a yPOWER SPECTRA OF 41. COMPARISON FIGURE JETSTAR AUGMENTED RSS AND LATERAL of a small resonance peak frequencies of interest with the exception damped natural frequency.

at the side-force generator of additional equalization No investigation of the effect highly since the system appears the feedback loops was undertaken in potential problem with the as mechanized. The greatest effective required be achievement of the very high gain system was expected to acceleromet, loop. If the lateral acceleration feedback in the lateral action of on a structural member that can be excited by the is mounted become unstable. In force generators, the control system may the side mounting extensive equalization or a change in sensor such an event, would be required.

location that the degree of acceleratioi Finally, it should be noted center of gravity by use of the alleviation obtainable at the aircraft less than when a side-force rudder alone (yaw damper) is considerably Figure 36).

generator is employed (see of Lateral Ride Smoothing System 3.5.3 Analytic Model insight into the effect of lateral In order to gain some mode the performance of the RSS, a simplified stability derivatives on for the longitudinal case was sought.

comparable to the one developed assumptions were made: The following small as compared of r and p on a are I. Effects to the effect of 0g; can be approximated 2. The dynamics of the aircraft the spiral mode root equal to zero; by setting 3. All actuators are perfect; time constant TW 0 0; 4. The washout 5. The transfer function of 13 due to turbulence (A) can be approximated by a first-order filter I/s.

if only the highest order terms (based on JetStar data) are retained, the transfer function of a due to A can be writtenas: y a s2(s2 L's) n L- GA Y K ST +

)(

s(+ +2drndr ndr

where the static gain K ST I is defined as: I

VoY KST' (1 + Ka T

)6f (3.5.2)

y T0 sfg and R I/T (3.5.3) R From Table V (page 74), it is clear that the value of the roll subsidence root (R) is a complex function of the gains K. and K .

r Y The same is true of the Dutch Roll mode damping.

In fact, both modes are, as was pointed out above, also sensitive to the choice of the washout filter time constantT . The following expressions were w derived for the Lateral Ride Smoothing System with t = I.

w W 2 =L'N' -N'L' + cos 6 NI - sin 6 L (3.5.4)

ndr p r p r o 1

2d r Yv -K N' f (3.5.5) V83 r V N 11 .

Y r K K v 2 rT 6 6 r 0 sfq + Y f) (I _ - K N -L N R 6 p r r r v (I + K VT Ye a y 0 sfg (3.5.6) The factor f is numerically where f was empirically chosen to be 0.375.

G r functions the transfer of of the numerators ratio to the equivalent pA this factor, then, G at the steady state (s = 0). Inclusion of and between the roll essentially irorates the yaw damper effectiveness subsidence and Dutch Roll modes. The expressions for , 2 drn , of interest.

accurate to within 22% over the range and R are Evaluation of the integral a

SfF IGAYl Adw (3.5.7)

y 0 2 2 2 . _ _ KST R(R _ L 2 yields: p I _ 412rdr + W 2)2 ay (R2 y ( ndr d) (3.5.8 drnd (nd nd expression and its A number of similarities between this for the longitudinal case (Equation 3.4.8) are apparent. The analog acceleration alleviation capability critical parameters affecting the Lateral RSS are the constants outside the brackets. System of the by: effectiveness can be increased I. Increasing K or, alternatively, increasing the a y (Y * ); generator effectiveness side-force 6 '84 sfg 2. Intreusing the damping of tnie roll subsidence mode (l/1R) ; 3. Increasing the frequency of the Dutch Roll mode (w nd).

Increasing Dutch Roll damping (Cdr) without simultaneously increasing the Dutch Roll natural frequency would appear to degrade system performance.

Terms inside the brackets have little effect on c a Figure 42 compares the RSS performance as calculated by the simplified expression (Equation 3.5.8) tp the digitally calculated results. Agreement is seen to be excellent at fairly high levels of K a As in the case of the Longitudinal Ride Smoothing Systems, failure of the stabilizing feedback loop (r 6 r) can be expected to degrade the handling qualities of the aircraft.

The degree to which this was the case was left to be examined in the simulation phase.

3.5.4 Alternate Lateral Ride Smoothim System Several authors cited in Chapter I (References 21 and 22) proposed the use of rudder alone to provide lateral ride smoothing.

For purposes of comparison with the performance of the system developed above, a calculation was carried out for such a mechanization adapted to the JetStar (Figure 43).

The feedback 2 gains were set at Ka -0.26 rad/m/sec (0.08 ay I rad/ft/sec ) and K = 4.0 rad/rad/sec r so as to yield Dutch Roll dynamics rA approximately comparable to those with the baseline Lateral RSS. Note that the r 6 r feedback signal is filtered by some washout T = I sec).

r~wo0 A comparison of the performance of the two systems is given in Table VII.

r /ec) 0.

.3

• o.6 .5 K a -1 .0 y (rad/m/se ) -2.DIGITAL -3.0 FIGURE 42. COMPARISON OF DIGITALLY CALCULATED a Y EXPRESSION WITH ANALYTIC oe TURBULENCE AIRCRAFT _ a I K s_______________ Kr s+ FIGURE 43. ALTERNATE LATERAL RIDE SMOOTHING SYSTEM Point: K = -0.26 rad/m/sec , Kr = 4 °/°/sec) (Design Y TABLE VII COMPARISON OF-LATERAL RIDE SMOOTHING SYSTEMS Basic JetStar Baseline RSS . Rudder RSS 0.045 0.155 0.131 dr w 1.36 rad/sec 1.195 rad/sec 0.86 red/sec ndr TR 0.87 sec 0.61 sec 0.44 sec sec 23.0 sec T, (T2) 418 sec (37.5) 2s ,f 0.0145 g SO.0312 g 0.0047 g a 0 0 0 a 2.35 /sec 1.56 /sec 0.68 /sec r /sec O 5.01 °/sec 1.95 °/sec 3.98 P 6 -- 7.8 0 - 06 -- 0.92 3.12 r % reduction G -- 84.5% 53.4% a y % reductiono -- 43.5% 71.1% % reduction a -- 61.0% 20.6% Although substantial acceleration alleviation can be obtained with the single control surface Lateral RSS, several practical considerations would make it difficult to mechanize the system aboard the JetStar. First, operation at the design feedback gain levels places severe demands on the rudder servo-actuator.. The servo in the aircraft would be operating at a..damping ratio 6r = 0.17 (cr = 0.24 for the r r baseline Lateral RSS). Secondly, failure of the yaw damper (r-*6 r feedback .loop) would result in a marginally stable Dutch Roll oscillation.

Any attempt to improve the acceleration alleviation capability of the by increasing the K feedback gain, would, under the failure system a y condition, drive Cdr negative. These reasons alone were sufficient to reject the single control Lateral RSS in favor of the baseline mechanization.

3.6 Overall Effectiveness of Combined Axis Ride Smoothing System The prototype Longitudinal and Lateral Ride Smoothing Systems synthesized in the preceding sections meet, with the possible exception of failure mode and structural resonance (feasibility) criteria, all the conditions for a successful design as set forth in Chapter II. The command signals that are required are readily available from typical aircraft instruments. The equalization circuits are all easily mechanized on an analog computer. Minimal handling qualities specifications are satisfied.

But what of the passenger and his comfort? For locations at or the near the center of gravity, under the design turbulence conditions, comfort model (Equation 2.3.1) predicts: C 3.6; the basic JetStar: 1. For 2.7; I and Lateral RSS: C Z For Longitudinal RSS 2.

RSS: C 2.8; RSS II and Lateral For Longitudinal 3.

the RSS in the comfort rating with a 1-point increase or approximately of passenger satisfaction the overall level More important, operating.

In to -85% (Figure 3).

to increase from 63.5% can be expected relatively small the Jetstar, only the of the model aircraft, the case substantial improvement flaps prevent an even more size of direct-lift in ride quality.

CHAPTER IV SIMULATION EXPERIMENTS 4.1 Order of Presentation the report deals with the ground-based simulation This section of of the basic and the dynamics and evaluation of handling qualities of experimental augmented JetStar. A brief description of the RSS verification is facility, simulation mechanization, and operational evaluation pilot's experience is followed presented. A summary of the description of the evaluation tasks. Results of the by a detailed air are presented In terms of handling qualities evaluations in smooth Both subjective and objective measures of subjective pilot opinion.

conducted in simulated qualities are presented for evaluations handling turbulence.

4.2 The Simulator Facility facility used in this study was the NASA Flight Research The ground controlled six-degree-of-freedom, hybrid computer Center fixed-based, of motion were aircraft simulator. The aircraft equations transport the Ride SmQothing a Xerox Model 9300 digital computer and mechanized on on an Electronics Associates, Inc. Model 231 R-V Systems were programmed analog computer.

the following shown in Figure 44, contained The simulation cockpit, instruments (from left to right): Top row: Sideslip (a) meter, Angle of attack (a) meter, i Yii i iiiiill iilii iii iQii Iiiiiii iii iililii

a

a aia

aia

a aii"sit"

a m aiaia

=================== = ====== === === = ==== ==== = === iiiiiilii~liiilili~ i i iiI iii il = ii ii~liJil ili ai ll i!

= iiii i i~ ii l i! = iii iii ii iiliiiiiiiiliiliili ~ ..- i= .......... ii=w~i w iii SIMULATION. COCKPIT FIGUE 4.

FIUR 44 SIUAINOKI m looman m m w m m mso mso m mso Second row: Clock, Airspeed indicator, Flight director (Collins FD-108G) Altimeter, Instantaneous vertical speed indicator, meters, Two engine power level Bottom row: Horizontal situation indicator (Collins 331-6A) Normal acceleration meter.

This instrument panel mockup is almost identical to that provided the command pilot in the GPAS (Figure 45).

Both the yoke and the rudder pedals were provided with a feel system that permitted adjustment of apparent linear control force, breakout force, friction, and damping (41). A four-way trim button on pitch and roll trim. Rudder trim was the yoke allowed adjustment of controlled by a console-mounted switch. Although four throttle levers were mechanized, an asymmetric thrust condition could not be simulated.

Selected cockpit control characteristics, gains, and trim rates were chosen by one of the pilots to be representative of the JetStar.

4.3 Digital Computer Program The real-time digital computer program was based on the six-degree of-freedom routine (SIM II) of Myers and Evans (42). This program solved the aircraft equations of motion (including the control surface turbulence quantities ag, actuator dynamics) as well as generating the were calculated time. The cockpit display signals

3 g, and pg In real

digitally with a repetition rate of 25 calculions per second.

I9

tok IT

mi a a m

a a ai a a a

a a a - a a

ali

FIGURE 45. GPAS TEST PILOT'S INSTRUMENTATION to calculate statistical properties The basic program was modified history of any one of data and to store the sampled time of 20 channels rigid body variables consisted of angular variable of interest. The aerodynamic angles inertial orientation angles ($,8,44, rates (p, q, r) 6 6 (V), control surface deflections (1a e (C,), total velocity , , r , (a , a ) setting, vertical and lateral acceleration 6f, 6sf), power z mean, variance, probability (at , a 1 p ). The and turbulence intensity quantities were probability histogram of these distribution, and of the stored variable real time. Power spectral density calculated in a simulation run.

be calculated following time history could Circuits 4.4 Analog the interface between the digitally The analog computer provided Cockpit control computed motion quantities and cockpit displays.

by the simulated Ride with signals generated commands were summed to the digital computer.

Systems before being transmitted Smoothing are Smoothing System analog mechanization Schematics of the Ride given In Figure 46 through 48.

of analog data could be to digital data, 16 channels In addition task assigned varied with the simulation recorded. Variables monitored check on systems during chosen to provide a the pilots, but were generally approach Landing System (LS) During simulation of an Instrument a run.

on dual X-Y izer tracking were monitored task, glideslope and local Figure 49.

and recorders are shown in The analog computer plotters.

Simulation Verification 4.5 Hybrid of the accuracy of the hybrid simulation Qualitative verification WASHOUT LAG FILTER a A +IOOV BIAS NOTCH FILTER CIRCUIT DIAGRAM; 46. ANALOG EQUALIZATION FIGURE FEEDBACK LOOP a f I I I III I n N I I I II 2 I I I I I II I l II II l II I I I

mC

m f

BIAS LEAD FILTER FIGURE 47. ANALOG EQUALIZATION CIRCUIT DIAGRAM;

o + 6 FEEDBACK LOOP

e o0 WASHOUT DIAGRAM; FIGURE 48. ANALOG EQUALIZATION CIRCUIT r + 6r FEEDBACK LOOP

m mIaIll- mam mfman a i

FIGURE 49. SIMULATION LABORATORY to the control performed by applying inputs of JetStar dynamics was response. Frequency the time history of aircraft surfaces and observing compared favorably oscillatory modes and damping of the longitudinal of N r' was required A 25% increase in the value with calculated values.

frequency and achieve the numerically-calculated to approximately simulation mode. Also, pilot A evaluated the damping of the Dutch Roll representa aircraft dynamics to be generally and reported the simulated the approach configuration.

tive of the JetStar in for the turbulence fields spectral densities were calculated Power (Figure 50 through 52) are reasonable

a , 8 , and p . These spectra

spectra asymptotes shown.

approximations to the Dryden and notch) compared well in Feedback loop filters (lead, lag, of interest over the frequency range amplitude and phase characteristics by Washout circuits were verified with digitally-calculated values.

measuring the decay time for step inputs.

Pilots 4.6 Simulation Evaluation Pilots A, pilots participated in the simulation experiments.

Five and 12,000 research pilots with 9000, 6500, B, and C are professional Pilots A and B have logged of flight time, respectively.

hours hours the JetStar. Pilot D has more than 10,500 considerable time in and Pilot E is a military aviator with of airline transport experience, as approximately 100 hours in total experience of 3500 hours as well evaluation time.

handling-qualities simulator -DRYDEN 1 ASYMPTOTE

0\

NI 4-i I0 1 o.,Ol 0.1 1.0 o (Hertz) a FIGURE 50. POWER SPECTRA OF SIMULATED ASYMPTOTE DRYDEN I0 1°-2 1.0 0.1 0.01 w(Hertz) g OF SIMULATED POWER SPECTRA 51.

FIGURE DRYDEN ASYMPTOTE N U - -2 w (Hertz) 10"3 1.0 10 0.01 0.1 POWER SPECTRA OF SIMULATED FIGURE 52.

Evaluation 4.7 Handling Qualities General Instructions 4.7.1 handling by all five pilots in the problems were flown Four a longitu and RSS augmented JetStar: evaluation of the basic qualities and an Instru task, combined axes task, axis task, lateral axis dinal the General instructions to (ILS) approach task.

ment Landing System a aircraft is to be assumed as follows: "The simulated pilot were airline consistent with be flown in a manner transport type and should are comfort considerations i.e., passenger operational procedures, are to be kept small; tight bank angle, etc., paramount. Load factor, aircraft For all problems, the be maintained."

control, however, should at the gear down and flaps in the landing approach configuration: was setting.

approach for all problems were: Initial conditions (2000 feet) 610 meters Altitude knots) 260 kilometers/hr (140 Indicated airspeed attack 11 degrees Angle of 7 degrees pitch attitude Displayed 0 degrees Heading Power setting for level 48%.

flight was provided: additional information task, the following For the ILS (0 feet) 0 meters Field elevation 0 degrees Runway heading meters (10,000 feet) length 3050 Runway meters (300 feet) width 92 Runway Initial distance to 15.2 kilometers threshold (8.25 nautical'miles) Initial offset from 0.61 kilometers runway centerline (0.33 nautical miles) Time to threshold 3:45 minutes Glideslope 3 degrees Required rate of sink 213 m/min (700 ft/min) Required power setting 32% Breakout altitude 61 meters (200 feet) Pilots evaluated handling qualities on the basis of the'Cooper-Harper Rating Scale (38) depicted in Table VIII.

4.7.2 Longitudinal Task The longitudinal axis task, repeated five times, was a timed, smooth air problem defined as follows: End Time I. Stabilize aircraft at initial conditions.

0:30 2. Climb to 3000 feet in 60 seconds.

1:30 3. Stabilize aircraft at 3000 feet and hold altitude for 30 seconds. 2:00 4. Descend to 2000 feet in 60 seconds.

3:00 5. Stabilize aircraft at 2000 feet.

3:30 Throughout the manuever heading and airspeed were to be held constant.

One run each was made for the basic JetStar, the two longitudinal RSSs engaged, and each longitudinal RSS with the stabilizing (e + 6e) feed back loop open to simulate a system failure condition.

Failure was initiated approximately 60 seconds after problem initiation. Pilots ftYf.g., In addition ,4 er9C were not informed of the configuration they per": I to Cooper-Harper ratings, pilot commentstwere sblicited on: TABLE VII I SCALE RATING COOPER-HARPER A tE'f ~ f 'Q,{DC AhN AIRCRAFT CHARACTERISTICS •I FR SJLECTED TASK =0R ' OEOUCY RKOUIRED OPERATION desired pocioormonce desirable Ifighly fOClOr for Pilot compensation no, a GO~d pellrooce desired Negligible deficenc-es required IV* riot compensotlon mildly mmal aom, pettownronce desired dehicuencios ufnleosont pilot comtpensation ,ie$ deficienc re$ Adequate pedrtonio(ice requ Moderately ObjeCtiOnOble enoe s satsfator i exn sv nreu s pilotVr perom considerable Adqute e ioa l but dei•n Obe L I fon 11nle Yes ut:ovmn No pilot compensation t olerable deficiencies wiVth 7 performance no10atailnable SAdequate tolefoble pilot compensation.

mo.,mum Mao, defioerncles nt in Qu~sllOn Conttrollability s ar D f e ci e s -- M"xd e"el s o d eto ble pilo t €o m p en so h o n is equ red If for co"Ifl~o impf(dvemenm wldkoad?

pilot to 9 delcerK~es Intense plt compensaion is requied Mt~o control Yesrelolrn opera ion m andatory M ojo rdefc encles • 0 orollob ll ?

and/or subphoses with involves d{esignation of flight phase * Definition of required opertihon ccompanying conditions.

Pioomsoi 1. Ease of establishing trim conditions; 2. Ease of initiating desi'red climb and descent gradients; and airspeed; of maintaining 3. Ease 4. Presence of undesirable pitch or rate of climb/sink excursions.

A summary of pilot ratings for this task is presented in Table IX, below.

IX TABLE AVERAGE COOPER-HARPER PILOT RATINGS, LONGITUDINAL TASK (Smooth Air) Standard Rating Deviation Case 1 Basic JetStar 2.4 0.33 2 Longitudinal RSS I 2.5 0.45 3 'Longitudinal RSS I 2.1 0.20 loop failed) 2.3 0.24 4 Longitudinal RSS I (8 + e 5 Longitudinal RSS I (6 + e loop failed) 2.7 0.3- pilots found no significant differences between the Generally, the reported no problems in performing the first three configurations and assigned task. Surprisingly, Longitudinal RSS II, with a value of n/, and RSS I configuration, -was lower than that of the basic aircraft rated equally good. Since the. configurations were- not presented in the was not a same order for each pilot, the "learning curve" phenomenon factor in the average ratings. Although numerically the simulated failure conditions were not significantly penalized, all pilots due to slight pitch a higher work load had resulted indicated that pilot-induced reported a tendency toward excursions. One pilot the and accurately identified (PIo) 'in pitch response oscillations mode as an excessively low short-period source of the problem frequency.

Lateral Task 4.7.3 times, was a evaluation task, repeated three The lateral follows:' End Time timed problem flown in smooth air and defined as 0:30 1. Stabilize aircraft at initial in 60 seconds. 1:30 Execute 90-degree right turn 2.

heading and hold 3. Stabilize aircraft on new 2:00 for 30 seconds.

left turn in 60 seconds. 3:00 4. Execute 90-degree 3:30 at initial conditions.

5. Stabilize aircraft to be held constant for the basic JetStar, Airspeed and altitude were of the RSS yaw damper RSS engaged, and a simulated failure the Lateral asked 60 seconds into the problem. Pilots were occurring approximately comment on the following: to rudder in order to coordinate the turns; 1. Use of 2. Ease of turn coordination; rate; initiating and maintaining desired turn 3. Ease of and of undesirable Dutch Roll characteristics; 4. Presence heading.

5. Ease of maintaining evaluations is given in A summary of the subjective pilot X, below.

Table TABLE X AVERAGE COOPER-HARPER PILOT RATINGS, LATERAL TASK (Smooth Air) Standard Case Rating Deviation I Basic JetStar 3.1 0.77 2 Lateral RSS 2.5 0.45 3 Lateral RSS (r r loop failed) 3.0 0.35 An examination of the root locus for the Lateral RSS (Figure 35, page 72) indicates that the Dutch Roll characteristics of the aircraft with of the the yaw damper (r - 'r) failed are almost identical to those basic JetStar. Thus it is not surprising that the pilot ratings for the two cases are almost identical. Not all of the pilots attempted to coordinate their turns by use of rudder. All, however, agreed that the turns were essentially coordinated with the RSS engaged. Only one be pilot, using the rudder, reported that coordinated turns could maintained even with the yaw damper failed. All five evaluation'pilots improved Dutch Roll characteristics with the RSS engaged recognized the characteristics.

and reported improved turn-entry and heading-hold 4.7.4 Combined Axes Task The combined axes task, repeated three times, was a timed climbing/descending turn in smooth air defined as follows: End Time 1. Stabilize aircraft at initial conditions. 0:30 2. Descend to 1000 feet while turning right 90 degrees in 1 minute. 1:30 End Time on new heading and altitude and 3. Stabilize hold for 30 seconds. 2:00 Climb to 2000 feet while turning left 90 4.

degrees in I minute. 3:00 at initial conditions. 3:30 5. Stabilize aircraft was to be held constant. The runs were for the basic JetStar Airspeed Lateral and for the Longitudinal Systems I and II with the configuration RSS engaged. Pilots were asked to give an overall Cooper-Harper rating comments regarding handling qualities as for the task and make any appropriate. The evaluation results are summarized in Table XI.

TABLE XI COOPER-HARPER RATINGS, COMBINED, AXES TASK AVERAGE (Smooth Air) Standard Deviation Case Rating 1 Basic Jetstar 3.1 0.71 RSS I and Lateral RSS 2.5 0.45 2 Longitudinal 0.57 3 Longitudinal RSS II and Lateral RSS 2.9 increase in In verbal comments, three pilots remarked on the obvious particular workload due to the more difficult task, but none found any with the Longitudinal RSS I plus Lateral RSS configuration.

difficulty among the pilots about the cause of the reported There was no agreement Longitudinal RSS II of handling qual'ities of the relative degradation plus Lateral RSS configuration. Although the lateral/directional the aircraft were identical for Cases 2 and 3, two characteristics for pilots reported control of heading to be more difficult. Two other pilots found pitch control to be somewhat too sensitive. One pilot preferred the configuration of Case 3 to that of Case 2. According to the average pilot opinion ratings, the basic JetStar would appear to possess the poorest handling qualities of the three simulated configurations. Three pilots reported control of vertical speed and pitch attitude as more difficult, one pilot noticed a slightly annoying Dutch Roll oscillation, but one pilot felt the basic JetStar to be slightly superior to the RSS-augmented configurations.

4.7.5 Smooth Air Evaluations, Conclusions Examination of the results of the first three evaluations leads to the conclusion that the incorporation of the Ride Smoothing Systems makes little difference in the handling qualities of the JetStar for manuevering flight in smooth air. For the lateral axis control task some improvement in Dutch Roll characteristics was detected by the pilots. During the combined axes task, a subtle improvement in pitch characteristics with the RSS engaged resulted in the augmented aircraft configurations being rated better than the basic aircraft. The numer ical differences in ratings, however, are so slight that statistically they are insignificant. More important is the conclusion that even with the stabilizing loops (8 6 e' r 6r ) failed for the RSS-augmented cases, the average pilot opinion rating is approximately three (3).

According to the Cooper-Harper scale, a rating of three (3) represents an aircraft with satisfactory handling qualities requiring no improvement.

4.7.6 Instrument Landing System Approach Task The final simulation evaluation task was an Instrument Ill Landing System approach problem. The pilots were asked to capture and track the localizer and glideslope to a 61 meter (200 foot) breakout altitude. A total of four runs were made by each pilot. The first run was with the basic JetStar configuration in smooth air. During the next three runs (basic JetStar, Longitudinal RSS I plus Lateral RSS, RSS) simulated turbulence was Longitudinal RSS II plus Lateral of crw = 1.2 introduced with components scaled to a vertical gust field g meters/sec (4 ft/sec). The simulation turbulence level was chosen below the design condition after a preliminary evaluation at a = 2.1 m/s w (7 ft/sec) resulted in a pilot opini'on rating of seven (7) for the basic JetStar. Ratings of seven (7) or greater imply a workload level that precludes the pilot from devoting attention to detailed evaluation of handling qualities.

Pilots were requested to comment on the following specific handling qualities considerations: 1. Ability to maintain desired airspeed and attitude; 2. Ability to acquire and track the glideslope; 3. Tendency to PIO in pitch/airspeed; 4. Adequacy of roll control; 5. Precision of heading control; 6. Ability to acquire and track the localizer; and 7. Tendency to PIO in roll/heading.

In addition, a separate Cooper-Harper rating was recorded for the longitudinal and lateral control aspects of the task. The subjective evaluations are summarized in Table XII.

TABLE XII AVERAGE COOPER-HARPER PILOT RATINGS, ILS TASK Longitudinal Lateral 'Standard Standard Case Rating Deviation Rating Deviation Smooth Air I Basic JetStar 3.2 0.3 2.6 0.4 Turbulent Air 2 Basic JetStar 4.3 0.9 5.5 1.4 Longitudinal RSS I + Lateral RSS 2.8 0.9 3.1 1.7 4 Longitudinal RSS II + Lateral RSS 2.9 0.7 3.6 2.3 Whereas no significant effect on handling qualities in smooth air could be attributed to the incorporation of a Ride Smoothing System, the effect of such systems for flight in turbulence is beneficial.

Although the standard deviations of Pilot Opinion Ratings are large, ratings by individual pilots were all improved when the RSSs were engaged. Note especially that at the simulated turbulence level the longitudinal handling qualities of the aircraft with a RSS in turbulence are rated equivalent to those of the basic aircraft in smooth air.

The improvement in the lateral axis is not quite as great.

Verbal comments by the pilots generally indicated few problems with longitudinal axis control for the RSS-augmented configur ations. With the basic aircraft, however, all pilots reported some tendency toward PIO in pitch. It was in the lateral-directional task that a tendency of the aircraft to "wander" in heading and oscillate in cases, such oscillation For the RSS-augmented angle was observed.

roll as typical of flight-in by three of the pilots was characterized were however, these. motions the RSSs were disengaged, turbulence. When desired heading difficulties in holding to yesult in serious reported pilots indicated that localizer. In all cases, the and maintaining the for the approach moderate to heavy level of turbulence appeared the were engaged, Ride Smoothing Systems basic JetStar. When the with the light to light to judged to be from very level of turbulence was the moderate.

qualities of the effect on handling Perhaps the best summary was given by Pilot A. After System for the JetStar of a Ride Smoothing with a in the unaugmented aircraft a simulated approach having flown the experience (7 ft/sec), he compared level of o = 2.1 m/s turbulence w had I and the Lateral RSS Longitudinal RSS the previous run where to engaged: been previous run] [compared to the "General comment: awful condition to fly--laterally, this is an not maintain directionally, and in pitch. Could [the Had to keep adding power because airspeed.

around so much. Attitude: aircraft] was sashaying the best I could...

herding it around I was just because of the [large] roll Could not hold heading like the ship didn't have excursions... Looked Roll tendency to PIO...

much stability.... Definite yaw. Initial poor due to adverse control was very was low... [Apparent] level of roll response turbulence compared to the previous run--almost double."

was small tracking error comments, the pilot's Despite these however, are apparent in workload, 54). The differences (Figure 53 and activity 55a-e) of aileron recordings (Figure in the strip-chart il4

; 300

9000 6000 12000 15000 (m) Distance to Threshold BASIC JETSTAR IN DESIGN 53. ALTITUDE TRACK; FIGURE FIELD (PILOT A) TURBULENCE 600- E r 7 -.- E 300 0 0) 4-S C1€ 0( 12000 9000 15000 Distance to Threshold (m) FROM LOCALIZER; BASIC JETSTAR IN FIGURE 54. DEVIATION FIELD (PILOT A) DESIGN TURBULENCE RSS ON RSS OFF I I -, +7~ , I~t . . . L ..,- .. ... "' , .! : ,; 9 1 . v ,: :. . , ;" . .:,'"L 701I ... . . -' . . ,i , I I ' I ' " : ,i I ' o 20 ho seconds -'--7 CHANNEL 8, iS ACTIVITY FIGURE 55. SIMULATION TIME HISTORY; JETSTAR IN DESIGN TURBULENCE FIELD (PILOT A) RSS ON RSS OFF I, ' I I -I ' ,I, , .

0 20 seconds CHANNEL 4,epio FIGURE 55. CONTINUED RSS ON RSS OFF 4 8 t - .

;~4. J '"~~ i'~h

II44' .,- 1If 4i.'.

L,'- , , i . 7.

.. . .7-"- -" . ..

- .

. .. . ,'r ' " *' * ' !'j, -I L . ' , ,I gI i ,

1 ; 1. '

_ _ rI' 'I I.

T,,, 4o seconds CHANNEL 5, AG FIGURE 55. CONTINUED RSS OFF RSS ON +.25g I -o I I -I I I 4o 0 20 seconds

CHANNEL l, a J

FIGURE 55. CONTINUED RSS OFF RSS ON -i + ,I J, T , ,,i

L 10< I V

0 20 4o seconds CHANNEL 3, a Y 55. CONCLUDED N3FIGURE (Channel 8 ) and pilot inputs to the elevator (Channel 4 ). Note also that the pitch attitude trace (Channel 5 ) was considerably smoother when the RSS was engaged. Vertical acceleration at the aircraft center of gravity is displayed on Channel 1. With the RSS operating, the sharp acceleration spikes were suppressed.

The effectiveness of the Lateral RSS is displayed in Channel 3, the transverse acceleration at the aircraft center of gravity.

With the system engaged, the lateral acceleration was reduced to very small amplitude.

'Digitally-calculated data for these two runs are summarized in Table XIII. The calculated root-mean-square turbulence levels for both runs were as follows: aw = 1.89 m/s (6.21 ft/sec), a = 1.32°, g g = = 2.23 /sec.

Pg Despite the fact that the measured quantities include manuevering loads, the agreement between theoretically-calculated param eters and their experimental values is reasonably good. Only the measured performance of the Longitudinal RSS is considerably inferior to the predicted value. Several additional runs were made to investigate the reason for this discrepancy.

4.7.7 Simulation of Straight and Level Flight Several data runs were made for a straight and level flight condition. Pilot control was "loose." At this condition, a 32.5% reduction "in c and a 80.5% reduction in a were measured when a a z y Longitudinal RSS I and the Lateral RSS were engaged. Power spectral density plots for these experiments are shown in Figure 56 and 57. Note that the power spectral density for the basic aircraft does not show the sharp peak at the phugoid frequency that was predicted by the theoretical .

TABLE XIII SIMULATION RESULTS, ILS TRACKING TASK Longitudinal RSS I + Basic Aircraft Lateral RSS Experimental Calculated Experimental Calculated 0 0 /sec 0.62 /sec aq 1.00 /sec 1.28 0/sec 0.48 0 0 0 /sec 2.68 /sec 1.29 /sec ap 5.07 /sec 3.32 0 0 0 0 1.27 /sec 1.04 /sec Cr 1.59 /sec 1.56 /sec 0 - 1.34 1.60 o a 1.87 0 2.20 0

as

4.69 0 - -- 4.870 " a 0.84 0 -- 0.44 0 0.35 0 10.59 0 8.8 o a -- 'Sf 0.80 0.61 o a 6r 'Sr 5.18 0 7.59 0 -- a6sfg 2.40 0 -- - 2.980 6a

0.0963 9 o.1o4 g 0.0634 g 0.0508 g

Ca z g 0.0061 g 0.0031 g aa 0.0273 g 0.0207 y Comfort Rating 3.4 3.4 2.8 2.6 % reduction aa 33.2 % 51.8% z aa 77.7 % 84.5 % % reduction y % reduction Uq 52.0 % 51.3 % % reduction ap 47.2% 61.0 % reduction a 20.2% 43.5 % % 10-I I0"2 BASIC JETSTAR 10-3 RSS I N "-104 N - 5 - 6 , 10-7 I0 1.0 0. 1 0 .01 . 00l (Hertz) TO TURBULENCE, OF a DUE SPECTRA 56. POWER FIGURE z DATA SIMULATION - 2 i0-3_ BASIC JETSTAR U--LATERAL RSS .

10 6 - 7

1o

I iI 0.01 0.1 (Hertz) SPECTRA OF a DUE TO TURBULENCE, FIGURE 57. POWER y SIMULATION DATA calculations (see Figure 28, page 60). In Section 3.4.1 it was shown that a considerable amount of energy is associated with this peak.

system Thus it can be concluded that the apparent loss of longitudinal effectiveness is the result of calculation errors at low frequencies in data. At higher frequencies, the shapes of the power the simulation spectral density plots closely match the theoretically calculated curves.

4.8 Conclusions The ground-based simulation program had, as its primary objective, the evaluation of the effect of the synthesized Ride Smoothing Systems on the handling qualities of the JetStar. It is concluded that, for manuevering flight in smooth air, the incorporation of these systems yields a slight improvement in pilot opinion ratings. Under the postulated system failure conditions, handling qualities are not catastrophically degraded. Thus, the Ride Smoothing Systems meet two of the most important design criteria set forth in Section 2.3: maintenance of adequate handling qualities and insensitivity to system failure.

For precision instrument flight in turbulence, incorporation of a RSS significantly improves the handling qualities of the basic aircraft by reducing pilot workload. Parenthetically, it should be noted that when subject to a severe turbulence environment, the handling qualities of a reasonably "well-behaved" aircraft such as the JetStar may deteriorate to unacceptable levels. Thus, the handling qualities qriteria of MIL-F-8785B (36) appear to be inadequate.

a measure of confidence Finally, the simulator experiments provided improvement provided in the performance estimates for the ride quality in anticipated improvement Systems and, based on the by Ride Smoothing test experiments.

rating, justification for flight comfort

CHAPTER V FLIGHT TEST PROGRAM 5.1 Planned Program JetStar flight tests of the Ride Smoothing Systems were planned to be conducted in three phases.

First, a series of developmental flights during which feedback gains were to be increased incrementally to their nominal levels were to be flown. A rudimentary handling-qualities evaluation and acquisition of baseline system performance data was to be accomplished.

When a reasonable level of confidence in system operation had been achieved, Phase II, a repetition of the ground-based simulation flight, was to be performed.

The final flight test phase was to obtain subjective evaluations of RSS performance.

Following a GPAS system failure unrelated to the RSS operation, the JetStar was grounded. Consequently, only two test flights were made and only some of the objectives of Phase I were accomplished.

Results of these very limited experiments are discussed below.

5.2 Implementation of RSS Aboard the JetStar Implementation of the Longitudinal and Lateral Ride Smoothing Systems aboard the JetStar was a straightforward extension of the ground-based simulator mechanization.

The feedback equalization circuits wired on the airborne PC-12 analog computer were identical to those used on the simulator (Figure 46 through 48, page 96ff).

The airborne analog computer is shown in Figure 58.

System-driving signals were obtained from standard GPAS instru mentation.

Yaw rate and pitch attitude gyro outputs were input to 129

PkNCEDI

G PAGt bAK

MOT FUMf

FIGURE 58. AIRBORNE ANALOG COMPUTER

am u a

a a

a

aia

m m

m

-

mm m

the PC-12 patchboard directly from GPAS signal-conditioning circuits.

attitude signal was Upon engagement of the GPAS mode, the pitch automatically nulled by the Response Feedback System circuitry. The signals for operation of the Ride vertical and lateral acceleration by a pair of accelerometers bolted to Smoothing Systems were provided th&-cabin floor slightly ahead of the nominal aircraft center of gravity. Outputs of these accelerometers were input directly to the circuits, requiring PC-12 board, bypassing the GPAS signal-conditioning prior to RSS the normal accelerometer signal to be nulled manually engagement.

Ride Smoothing System commands to the elevator and rudder were summed with pilot commands from the aircraft left seat controls. RSS were to the direct-lift flaps and side-force generators commands applied directly to the surface servos.

5.3 Ground Tests As with the ground-based simulation, performance of the airborne analog circuits was verified by observing the frequency and magnitude response of the RSS filters to sinusoidal inputs. Response of the PC-12 computer circuits was comparable to those of the ground-based verified analog computer. Proper phasing of the command signals was by pressurizing the GPAS system, tilting individual sensors, and observing the deflection of the appropriate control surface.

Prior to the implementation of the Ride Smmothing Systems, feed of acceleration to the direct-force surfaces had never been back attempted aboard the JetStar. Several experiments were, therefore, determine the stability (structural coupling) of these conducted to feedback loops. With the GPAS system pressurized, the acceleration feedback gains were slowly increased to their nominal values and the surface position transducer signals monitored on a strip-chart recorder.

In the case of the direct lift-flaps, no instability was detected. The flaps would, however, respond to movement by personnel about the air craft cabin. Thus, although the accelerometer-mounting was adequate for the Phase I investigation, ultimately a more suitable accelerometer found. Increasing the lateral location would have had to have been acceleration feedback gain above approximately 20% of the nominal value resulted in limit cycling of the side-force generators. This phenomenon was attributed to significant free play in the side-force generator linkages. The feedback gain of this loop was, therefore, set well below nominal during the flight test program. Although the linkages were readjusted, the flight program was terminated before another ground resonance test could be accomplished.

A final pre-flight operational test of the airborne RSS consisted of operating the system in a closed-loop sense. The aircraft equations of motion were solved on three slaved Electronics Associates, Inc.

(two Model TR 58 and a TR 10) analog computers. Calculated motion parameters were fed to left-seat cockpit displays and the airborne PC-12 analog computer. Pilot control inputs and RSS system commands were fed to the appropriate control surfaces of the aircraft. Surface position transducer outputs were fed back to the auxiliary ground computers to complete the closure. Hydraulic pressure for the control surfaces was supplied by a ground system. Signals proportional to components of actual atmospheric turbulence that had been recorded on analog tape were used to perturb the calculated angle-of-attack and sideslips signals in the ground analog computers.

The aircraft was thqs made a part of a ground-based simulator. Selected system param eters were monitored on strip-chart recorders during, the simulation runs. Since operation of the GPAS in this ground mode adds to utiliza tion time of aircraft ,hydraulic components, the experiment was conducted only long enough to qualitatively verify proper operation of the RSS.

5.4 Data Acquisition and Reduction Acquisition of JetStar flight-test data was by means of a Pulse Code Modulation (PCM) System.

Some 80 channels of data were available for analysis. All of the data presented below were sampled at a rate of 40 samples per second. Power spectral analysis of selected data channels was performed using the same digital computer program (PSDQR) employed in the ground-based simulation studies.

Calculation of the statistical properties of the true vertical gust field (w ) was accomplished by correcting the nose-boom-mounted gust vanesignal (av) for aircraft motion: Wg = Cos [VT0 - VT0' + Pxq + f Az dt v (5.4.1) ] wheie Zx is the distance from the aircraft center of gravity, to the gust vane, and A is the vertical acceleration of the aircraft center z of gravity with respect to inertial space. The value of A was deLermined from the aircraft center of gravity ac.celerometer outpuLs Nx, Ny, and N (in g's) by: z A g {N sin e + sin @ cos O(Ny + sin j cos 0) z x + cos 4f cos e(N - cos cos 6)1 (5.4.2) were monitored on strip-chart recorors Several channels of data In addition to providing a qualitative during the flight tests.

in real time, the time code on the indication of system performance segments for recordings provided identification of data strip-chart analysis.

digital Data 5.5 Summary of Flight Test #349 and #350, Two flight tests of the Ride Smoothing Systems, June and II June 1974, respectively. The aicraft were conducted on 5 #349, the in the approach configuration. During Flight was flown feed Lateral RSS and Longitudinal RSS I were engaged. Acceleration 5% of systems were increased incrementally from back gains for these of nominal for the Longitudinal RSS and their nominal values to 45% rudimentary examination of the of nominal for the Lateral RSS. A 15% in smooth air for this configuration aircraft handling qualities "S' turns) was accomplished. The command (flight path angle changes, The aircraft no objectionable aircraft characteristics.

pilot reported approximately flown in light to moderate natural turbulence for was then engaged.

with the Lateral and Longitudinal RSS I systems 10 minutes reversed, and the same geographical area traversed Heading was then with the systems shut down.

Longitudinal RSS I was operated at nominal During Flight #350, the air for approximately 3 minutes before design feedback gains in turbulent a GPAS system anomaly resulted in system shut down. Approximately three minutes of turbulence data for the basic JetStar was recorded immediately following RSS disengagement.

Results of these experiments are summarized in Table XIV . Experi mental values have been adjusted by multiplying by the ratic; of the design turbulence level of o 2.1 m/sec (7 ft/sec) to the measured w c . The numbers in parenthesis are theoretically-predicted values.

w9 The agreement between theoretical and measured acceleration levels for the baseline case is quite good. The theoretical calculation, however, significantly overestimates aircraft response in pitch rate and yaw rate. The measured performance of Longitudinal RSS I, in terms of percent reduction in a and a at the design feedback gain a q z levels, is in excellent agreement with predicted performance. The acceleration alleviation provided by the Lateral RSS, however, is significantly below the expected level, while the reduction in r is r very close to the predicted value. Had the acceleration feedback loop been open, the yaw damper (r + 6r feedback) alone would have provided a 38.3% reduction in aa and a 29.0% reduction in ar' Thus, it appears y that the side-force generators provided no benefit at the very low level of K realized in the tests.

a y A comparative power spectral density plot (PSD) of the output of the center of gravity normal accelerometer is shown in Figure 59 for the baseline and Longitudinal RSS I nominal gain cases. This plot differs from previously-presented PSD's in that the individual curves have been normalized by their respective mean-square values. Since the areas under both curves are thus identical, the plot displays only relative TABLE XIV SYSTEM FLIGHT TEST RESULTS RIDE SMOOTHING I Longitudinal RSS Flight #349 Flight #349 Flight #350 K = 3.3 rad/m/sec 2 K = 1.4 rad/ni/sec 2 .(0.43 rad/ft/sec az (1.0 rad/ft/sec baseline az K = 0.14O/o Ke = 0.030/0 e g . 0.0518 (0.0572) g aa 0.1047 (0.1178) g. 0.0788 (0.0794) 0.402.(0.700) °/sec 0.933 (1.440) °/sec 0.599 (1.12) /sec aq 0 7.70'(9.96) 0 5.31 (5.95) a 6f 50.5 (51.8) % " 24.7 (32.6) % % reduction aa z % 56.9 (51.3) % aq - 35.8 (22.3) % reduction Lateral RSS Flight #349 Flight #349 rad/m/sec 0.5 = K a y (0.15 rad/ft/sec Baseline Kr = 1.0 /°/sec 0.0307 (0.0350) 9 0.0206 (0.0158) 9 aa °/sec 0.88 (1.65) /sec or 1.32 (2.35) (2.67) °/sec 2.14 a --.

32.9 (54.9) % % reduction aa "" 33.0 .(30.0) % % reduction ar y I I RSS JETSTAR BASIC o10 -

lO-3_

_ I O 0.1 1.0 10 0.01" cu (Hertz) FOR BASIC AND FIGURE 59. COMPARISON OF a POWER SPECTRA z DATA) I AUGMENTED JETSTAR (FLIGHT LONGITUDINAL RSS at particular frequencies. The greatest acceleration RSS effectiveness alleviation can be seen to occur from frequencies somewhat below the short agreement with the period peak to I Hertz. This conclusion is in clear theoretical calculations (see Figure 28, page 60). The "spikes" in at approximately 3 to 4 Hertz are attributable the experimental curves of the accelerometer mounting plate. - to resonance RSS I A qualitative impression of the effectiveness of Longitudinal referring to Figure 60. These strip-chart records can be gained by The traces on the-left side of the figures are taken from Flight #350.

the time segment with the Longitudinal RSS I operating (13:44 are for to 13:45:40 hr) and those on the right for the baseline case (13:48 to hr).

13:49:40 chosen as representative of the turbulence The S vane output was motion in the lateral axis is essentially unaffected level since aircraft by the Longitudinal RSS. Note that the magnitude of the turbulence field is approximately equivalent for both time segments. Excursions in reduced when the vertical acceleration (az), however, were substantially RSS was engaged.

5.6 Conclusions Although the limi'ted amount of available flight data makes categorical statements impossible, the data permit some tentative conclusions. First, the theoretical calculations of aircraft root-mean square acceleration response to turbulence ,agree reasonably well with Longitudinal RSS augmented and experimental values for both the is most important since unaugmented configurations. Such agreement OFF RSS RSS ON - 0

-0-

.. .. ....

O-r- I ,_-'_:I * I I I I I * 10 2 - ' - seconds - ,HANNEL 2% Vi RSS I AUGMENTED AND LONGITUDINAL 6o. TIME HISTORY;'BASIC FIGURE DATA) (FLIGHT IN TURBULENCE JETSTAR .

,!F 17 ' IO F - I OI

-10-l-

0 10 20 seconds' CHANNEL 5,0 FIGURE 60. CONTINUED OFF RSS ON RSS I I~fl' . 5g.---"g . I I TTq, Htz. . .... . .. . .

S. § t-" I it '

* I5I9, ' I • L" I h I ;" " I:4:'.t, i ' m it ! :.

II I I I 10 20 CHANNEL 11," , FIGURE 6o. CONCLUDED a is the dominant term in the evaluation of passenger comfort.

a z Secondly, the theoretical prediction of lateral acceleration is also in good agreement with experiment for the baseiine case. The failure of the experimentally-observed value of a to fall to th& theoretically a y predicted level with the Lateral RSS engaged can probably be attributed to the low acceleration feedback gain level necessitated by mechanical difficulties. Finally, it would appear that mechanization of Longitu dinal Ride Smooihing System I is feasible and that its, incorporation would provide substantial improvement in passenger comfort. Had human subjects been on board the aircraft, the comfort model predicts that the .percentage satisfied would have increased from 66.8% (C = 3.5) for the.basic JetStar case to 84.5% (C = 2.8) when the RSSs.were engaged.

CHAPTER VI RIDE SMOOTHING SYSTEM CONCEPT EXTENSION OF AIRCRAFT TO STOL Aircraft 6.1 Selected systems for synthes-izing ride smoothing The success achieved in investigation a brief theoretical and simulation the JetStar prompted two radically different applicability of these systems to of the The first of JetStar.

of the same size as the STOL-class aircraft Buffalo. The Buffalo relies of Canada DHC-5 these was the deHavilland lb/ft ) to achieve short (W/S 1676 N/m = 35 on low-wing loading in configuration to is otherwise similar field performance, but this' investiga a-ircraft selected for The other conventional aircraft.

studied at the NASA Langley design extensively tion was,,a conceptual Ioading LRC S-Il. The wing Centers and designated and- Flight Research jet transports (W/S to typical modern aircraft is equivalent of this achieved performance is S-il short-field 3830 N/m = 80 lb/ft2).

As in the jet flap (43).

an externally-blown through the operation-of the power selected design condition-was JetStar investigation, the 2.1 m/sec).

heavy turbulence (a = approach in moderate to w g w the STOL aircraft made' of a Longitudinal RSS for The mechanization In the-case of the wing-trailing edge flaps.

use of the, elevator and only for the entire flap data were available Buffalo, aerodynamic more effective these flaps are-considerably system. Thus, although the entire surface than those on the JetStar, at the design condition control. The S-Il configuration as the direct-lift had to be assumed has a more sophisticated system of wing-mounted control surfaces.

These include spoilers, flaperons, symmetrically-deflecting ailerons, and a "direct-drag".fiap'system.- Al'though the direct-drag f'laps have an effective lift to drag ratio of only (L/D)f 1.36, their lift capability is equivalent to that of the JetStar system. Therefore, only the "direct-drag" flapswere mechanized in the RSS design.

. The lateral axis RSS. for the STOL :aircraft was i ni-tially mech anized in -the' same.wa.y as- that. for the JetStar, i.e.,,using, the: rudder and si'de-force generators,.- The-hypothetical side-force generators were scaled -to produce the. same lateral accelteratin, per unit deflec tion at the design velocity,as those on the. JetStar. The projected 2 2 2 2 area for-each-of two surfaceswas 4.9 m .(53

ft ) and 8.0 m ,(86 ft.)

for the Buffa:io. and- S--1l, as compared to -1.3 m 2 04 ft ) fqr, the Jet- Star.. The postulated.

STOL side-force generators are quite large;, for the S-1I, the area is. 1.4 times that of the ai-rcraft's vert-ical tail ..

Incorporation of suchcontrols str.ictly for.imprqvement of-ride qualities would be hard-to justify. One cah,-however, envision additiohal, uses of larges5ide-force, generators, e.g.,, improvement of crosswind landing, capability.

,Furthermore, some reduction in s'ize might be possible if the surfaces are .immersed -in the propeller slip , stream or jet efflux. Such tradeoffs, however, were not evaluated.

Dimensional stabili-ty der-ivatives and aircraft parameters cfor the, Buffalo and.S,11 power approach conditions are summarized 'in Appendix E. The. Buf-falo -parameters- were taken from the NASA Ames Research CenterSTOLAND program documentation; the;S-l data, from-NASA Flight, Research Center sources. Actuator characteristics for the control surfaces were assumed identical to those of the, JetStar.

6.2 Synthesis of Ride Smodthinq"Systems 6:2.1 Longitudinal RSS Whereas the basic JetStar longitudinal dyhamics and control characteristics in the approach configuration (Wih the excepfi6n of Zph clearly meet the handling qualities ) requirements of MI'L-F-8785B (36), those of the Buffalo and S-I do'not. Thus, a direct adaptation of the Longitudinal RSS'deeloped previously was not possible.

In particular, the handling'qualities-,parameter n/ is marginal in the 'case of the Buffal'o (n/ = 2:9 g/rad) and inadequate for the S-11-(n/ = 1.57 g/rad).

Incorporation of Longi'tudinal RSS H! would have further degraded thi's metric. Therefore, the applicability only of Longitudinal RSS I to the STOL configurations was stud'ied.

The effect of the equalized essential feedback (a +f z through ciscaded'washout and, lag f1lters) on the dynamic modes of both STOL ai-craft was substantially'different from the s'hort-period and phugoid-rodt location changes observed for the JetStar. First,'for the range of acceleration feedback gains considered, the phugoid root remained essentia'lly statibnary.'

Secondly, the short-period root locus tended toward the imaginary axis (reduction in Csp ) at an almost constant evel of'damped natural frequency (md) These vaiations, however,'were also small.

Consequently, the stabilizing feedback-loop requirements were different than for the JetStar RSS. In the case of the Buffalo, no increase, in short-period frequency was required, and short-period damping was recovered by feedback of pitch rate to the elevator (q 6e) through a lag filter.

) for the The short-period damped natural frequency d sp basic S-li was calculated to be only a factor of two higher than that Under these conditions, con of the phugoid (see Table XV, below).

between vertical velocity (angle of attack) and siderable coupling modal pitch attitude perturbations occurs at the phugoid frequency; 0 0 the short-period frequency: ratio: u:a:O = 2.4 ft/sec:O.7 :1 . (At .) Consequently, both pitch attitude and u:c:0 = 0.2 ft/sec:l.7 :1 rate were fed. back to the elevator (e,q t 6 to increase the pitch e ) frequency and damping and to achieve a greater separation short-period Lead and lag filters were incorporated in the 6 and q of the modes.

loops, respectively.

of the Longitudinal Ride Smoothing Systems Block diagrams for the Buffalo and S-Il are shown in Figures 61 and 62. Table XV compares the dynamic characteristics and Longi.tudinal RSS performance parameters at the design condition for the JetStar and the two STOL aircraft.

The numerical data of Table XV indicate a number of similarities between the RSS augmented aircraft. The vertical acceleration levels (underlined terms) for flight in the standard turbulence field for the three augmented aircraft (a = 2.1 m/sec) w pilot's viewpoint, the dynamics are essentially the same. From the of the augmented aircraft., as expressed in terms of the parameters (time to half amplitude of the short-period mode, Ti , inverse cycles sp to the 1/10 amplitude i/Cl/i 0 and phugoid time to half amplitude, , TURBULENCE" 6e 8e 6e 6 e pilot c G AIRCRAFT e DYNAMICS > C C a z s s q - - "

_ _ ( I ) I s s 1 , _

FIGURE,61. BUFFALO LONGITUDINAL RIDE SMOOTHING SYSTEM 2 = 0 (Design Poi-nt: K = 0.8 rad/m/sec , Kq 5 //sec), a a 9 oo TURBULENCE < 6 6. pio 6e+ Oe C AIRCRAFTq .

DYNAMICS e 6 o z • a z E (l+s/1) jKq (i+s/.I Tj FIGURE 62.' S-11 LONGITUDINAL RIDE SMOOTHING SYSTEM.

0 0 K = 5- / /sec) (Design Point: K 0.8 rad/m/sec , K = 0.6 o/o a 0 aZ TABLE XV SYSTEMS RIDE SMOOTHING OF JETSTAR AND STOL LONGITUDINAL COMPARISON Basic RSS-Augmented Basic RSS-Augmented Basic RSS-Augmented S-1I S-li Buffalo Buffalo JetStar JetStar 0.698 0.826 0.725 0.628 0.546 0.567 sp Hz 0.136 Hz 0.262 0.310 Hz Hz 6.282 Hz 0.266 Hz 0.356 sp 0.60 sec 0.98 sec 0.62 sec 0.49 sec 0.55 sec 0.76 sec TL sp 3.46 3.77 5.41 2.81' 2.21 2.35 1/C/ 0.371 0.249 .O.OO5 0.102 0.054 0.522 ;ph sec 26.9 sec 31.6 sec 22.8 sec 53.2 sec 20.0 36.6 sec P h sec (658) sec 8.7 10.0 sec 21'.4 sec sec 9.6 sec (T p 74.8 Tph 2 h 0.0561 g g 0.805 9 g 0.1073 y 0.0622 0.1178 g '0.0572 p p a z 0.270 g 0.0209 g g 0.0251 g 0.0040 g 0.0318 0.0112 g a 0 /sec 10.8./sec 0.70 /sec 1.33 /sec /sec 2.28 1.44 /sec 0.70 a 2.40 2.6 0 9.9 0.--

a--

6f ° 0.90 -- 1,2 0.4 0 -- C6 -- 93.0% 42.8% 51.8% -- % reduction' -- z 92.2% -- -- '34.1% 64.6% % reduction a -- x 93.5% 41.5% -- 51.3% --

a --

% reduction q and landing approach, passenger comfort T, ) are equivalent. In the 2ph (with the exception previously-considered handling qualities criteria therefore, be met STOL deficiency in -n/a) can, of the aforementioned the three aircraft by the incorporation of a Ride equally well for Smoothing System.

Several differences between the STOL aircraft and the noted, however. First, even with the RSS engaged, JetStar, should be ) is significantly root-mean-square longitudinal acceleration (cr the x the values larger for the STOL aircraft than for the JetStar. Also, are much greater. *Theseobservations of the stability derivative Zu* suggest that the effect df the longitudinal component of turbulence (u ) on the STOL aircraft acceleration response might be important and be included in a more complete analysis. Second, the degree should of flap activity (af) required to.-achieve an equivalent level of a z the STOL aircraft is only one-fourth that required for the JetStar.

for the much lower Part of the reason for this difference is, of course, the approach speed of the Buffalo and[S-ll. Finally, because of for unstable phugoid mode of the basic S-l, the values calculat@d all (99,7%) of the total aa , a , and aq are very large. Almost z X the frequency calculated mean-square vertical acceleration occurs in low-frequency mbtion is easily band below 0.05 Hz. In practice, the pilot; the significance of the calculated root-mean suppressed by is, therefore, debatable. A comparison of square values for this case for the three aircraft, however, indicates that the a power spectra z is quite similar (Figures 28, 63, the RSS effect at higher frequencies and 64).

- I -2 _ -3_ BASIC BUFFALO U a) 0 4 04 10"- RSS N - - - _ 0.01 0.1 1.0 10 w (Hertz) FIGURE*63. COMPARISON OF a POWER SPECTRA FOR BASIC AND z LONGITUDINAL RSS AUGMENTED BUFFALO - S IC S l BA 'U (D 10-3 -4 N .

10-5 - 6_ 1o 10-7 0.01 0.1 1..0 10 w (Hertz) a POWER SPECTRA FOR BASIC AND FIGURE 64. COMPARISON OF z LONGITUDINAL RSS AUGMENTED S-11 6.2.2 Lateral RSS As in the longitudinal case, closure of the essential Lateral RSS feedback loop (a y 6 sfg) had a negligible effect on the character'istic modes of the'STOL aircraft.

The previously-employed stabilizing loop, r 6 r, however, was still desirable. Both the, basic Buffalo and S-11 have an unstable spiral'mode (see Table XVI).

In addition, damping of the Dutch Roll-modeof the'basic SZI1 is well below the level specified in MIL-F-8785o The incorporation of-an unequalized yaw damper tends to alleviate both of these undesirable characteristics.

A washout was not incorporated in the r 6 + loop r since it was found to radically reduce both Dutch Roll damping and' frequency. A third feedback loop, roll rate to aileron-(p-* a) was , added to the S-il RSS-mechanization in order to increase roll damping.

Block diagrams of the Lateral Ride Smoothing Systems for the Buffalo and S-il are shown in Figures 65 and 66. Table XVI compares the dynamic characteristics and'Lateral RSS performance parameters for the detStar and the two STOL aircraft.

Comparative power spectral density plots for the lateral acceleration of the Buffalo and S-1l in the baseline and RSS-augmented configurations are given in Figure 67 and 68.

In the case of the Buffalo; the RSS completely suppresses the Dutch Roll response peak in addition to reducing the acceleration level across the entire frequency band. In, the case of the S-li aircraft, the effect of the p- 6 a feedback is clearly evident as a sharp dip at the maximum roll gust (p ) input frequency.

Although the Dutch Roll resonance is still apparent, the magnitude of response is sharply reduced.

At higher TURBULENCE r AIRCRAFT w

0 >

DYNAMICS __ - r 6ss y ay Kr BUFFALO LATERAL RIDE SMOOTHING SYSTEMS FIGURE 65.

0 0 , K . / /sec) (Design Point: K = -3:3 rad/m/sec Y, TURBULENCE a.

a I U a a pilot c G p < P_ AIRCRAFT r .lt r 6r DYNAMICS c a 6sff% sfg 5f

G6 sfgo y=

KY

p> ya

K LATERAL RIDE SMOOTHING SYSTEM FIGURE 66. S-11 = 2 Point: K = -3.3.rad/m/sec , K 2 / /sec, Kp -1 0 0/sec) (Design a r Y ON XVI TABLE SMOOTHING SYSTEMS LATERAL RIDE AND STOL OF JETSTAR COMPARISON RSS-Augmented Basic RSS-Augmented Basic RSS-Augmented Basic S-1i S-1i Buffalo Buffalo JetStar JetStar 0.403 0.710 0.003 0.155 0.266 o.o45 r 0.67 rad/s 1.01 rad/s rad/s 0.83 rad/s rad/s 0.89 1.36 rad/s 1.20 w nr 1.18 sec sec 1.37 sec sec 0.70 sec 0.66 0.87 sec 0.61 TR 14.3 sec (5.4) sec 11.7 sec (20.7) sec (37.5) sec 418 sec T s(T 2a 0.0312 s 0.0047 s 0.0214 g 0.0060 g 0.0330 g 0.0054 g aY 2.35 °/sec r 1.56 O/sec 1.62'°/sec 1.46 °/sec 2.39 /sec 2.44 °/sec 9.30 -- 11.20 -- 7.80 -- s f a ° _ 0.92 -- 1.47 -- %reduction ca -- 84.5% -- 72.0% -- 83.6% -2% 9-- -- 43.5% Cr -- % reduction -02 10-3 BUFFALO x-BASIC U co 10-5 C4 10RSS _ 10.

IO-7 - 8 1o I Ii 1.0 - 10 0.001 0.01 0.1 w (Hertz) BASIC AND OF a POWER SPECTRA FOR FIGURE 67, COMPARISON y RSS AUGMENTED BUFFALO LATERAL 10"3.

S-1l BASIC " 4 _ i0 RSS io_6 - eY - 1.0 I 0.

0.01 0.001 (HU) AND" BASIC FOR a yPOWERSPECTRA OF COMPARISON 68.

FIGURE S-11 AUGMENTED RSS LRTERAL frequencies, a uniform reduction in acceleration response to turbulence was achieved. As with the.Longitudinal RSS, the root-mean-square acceleration response for all three, aircraft with the Lateral"RSS engaged was reduced to comparable levels.

Although effective in suppressing the Dutch Roll mode, the yaw damper fails to provide any alleviation of.a for the Buffalo, and r actually increases r for the S-il. The yaw rate, response of both ai.rcraft remains dominated by a.low-frequency heading,-instability which is unaffected ,by the RSS.

6.2.3 Improvement in Passenger Comfort The improvement in passenger comfort resulting from the incorporation of a Ride Smoothing System aboard the STOL aircraft operating in the design turbulence environment is evident from predicted comfort ratings: Buffalo S-lI Comfort % of Passengers Comfort % of Passengers Rating Satisfied Rating Satisfied Basic 3.4 69%- 5.0 25% RSS Augmented 2.8 84% 2.7 86% 6.3 Simulator Evaluation of STOL'Ride'Smoothing System The simulator handling qualities evaluation of the RSS-augmented STOL aircraft was carrired out in the same facility as used for the Jet- Star evaluation. Only the ILS problem was flown by the five evaluation 'pilots. Three runs were made for each of the aircraft: 'basic and RSS: augmented configuration in smooth air and RSS-augmented configuration 'in moderate turbulence (c s 1.2 m/sec). No evaluat-ion'was made of the w g basic configurations in turbulence since a preliminary run with the S-11 resulted in a Pilot Opinion Rating of 10 (uncontrollable).

was increased from 30 to 70 to The simulated glideslope angle better simulate a typical STOL approach. The only noticeable effect of the steeper approach angle was to increase the pilot lead required to fly the simulated Buffalo, i.e., upon intercepting the glideslope, power had to be reduced to idle and a rapid pitch-over accomplished.

Several of the pilots penalized the Buffalo because.of this power/drag characteristic.

Results of the STOL handling qualities evaluation are summarized in Table XVII.

TABLE XVII AVERAGE COOPER-HARPER PILOT RATINGS STOL ILS APPROACH TASK Longitudinal Lateral Standard Standard Rating Deviation Rating Deviation Basic Buffalo 3.2 0.68 4.9 0.81 RSS-Augmented Buffalo 2.9" 0.66 5.0 1.45 RSS-Augmented Buffalo in Turbulence 3.3 0.83 5.3- 1.15 Basic S-11 * * 8.5 0.74 RSS-Augmented S-11 3.0 0.00 4.7 1.54 RSS-Augmented S-11 in Turbulence 3.25 0.43 5.9 1.24 160 *No rating; task dominated by lateral problem.

In light of the aforementioned deficiencies of the STOL configura tions with respect to the control parameter n/,'it is somewhat sur" prising that none of the evaluation pilots reported serious longitudinal handling qualities'problems. One pilot did report a sllght tendency toward PIO in pitch at the S-1l phugoid frequency.

The simulated lateral characteristics of bbth aircraft, even with the RSS engaged, however, were clearly unsatisfactory for the ILS tracking task. Three pilots stated that the basic S-ll.could be landed only if a visual reference were available. Inadequate heading control and high adverse yaw were cited as the major deficiencies of this aircraft.

Several pilots suggested incorporation-of a heading/roll attitude command autopilot and aileron rudder interconnect. Heading precision was also cited as the major directional control problem with the Buffalo. Whether this characteristic is in fact representative of the operational aircraft-would have to be established in a more exten sive investigation.

6.4 Conclusions Although, incorporation of a Ride Smoothing System aboard the selected STOL aircraft would provide substantial improvementsin ri'de quality, the simple systems investigated failed to meet the qualitative handling qualities criteria in terms of pilot opinion rating as set forth in Section 2.3. As was pointed out by the evaluation pilots, a number of elements normally associated with stability augmentation systems (SAS) would have to be incorporated in order to provide adequate handl-ing qualities. Such an integration should not be difficult. Recall that for both the Buffalo and S-l1, the closure of the essential feedback loops..(a + 6f, ay 6 sfg) had neg-ligible effect z 44 reports a Stability Augmentation on aircraft dynamics. Reference System developed-for the S-li in the landing approach flight phase. An obvious extension of the present research would be an investigation of of the proposed Ride Smoothi-ng System with the SAS the compatibility augmented S-il.

CHAPTER VII CONCLUSIONS AND RECOMMENDATIONS The research reported herein is unique in the sense that the problem of analyzing, synthesizing and evaluating aircraft Ride Smoothing Systems was, for the first time, approached from a comprehensive viewpoint. The multiple criteria that were established, both subjective and objective, precluded the application of optimal control theory. Nevertheless, both Longitudinal and Lateral RSSs were successfully developed and were shown to be applicable to STOL aircraft, suggesting that the solution to the RSS problem is generic.

A significant amount of new information was generated. In particular, the feasibility of employing side force generators to attenuate rigid aircraft response to turbulence was demonstrated theoretically and in simulation.

Such systems were shown to be more effective than systems using rudder control alone. Extensive fixed based simulator experiments provided subjective, qualitative and quantitative data that indicate the improvement in turbulent flight handling qualities made possible by the incorporation of a Ride Smoothing System. The simple analytic models developed for the baseline Ride Smoothing Systems allow significant insight into the effect of individual aerodynamic parameters on the performance of these systems.

The constrained "performance index" contours generated by these models, together with the "comfort model," permit a rational approach to the choice of feedback gains. The limited flight data that were generated generally support the theoretical predictions of RSS performance.

Finally, the data presented herein are sufficiently complete to permit independent evaluation and interpretation, thus contributing to the overall data base on Ride Smoothing System characteristics.

As with any broad scope research project, the results of this study suggest as many questions as may have been answered.

Different forms of equalization for the various prototypeRSS feedback loops should be examined. For comparison purposes, it would be interesting to develop optimal control laws for both the longitudinal and lateral axis control problem.

The effectiveness of the proposed RSS should be examined at fuselage locations other than the center of gravity.

The gain scheduling that would be required for system operation over the entire flight regime should be established. The interfacing of the RSS and SAS for the STOL configurations should be undertaken.

Extension of the simplified analytic models to the STOL configurations should be attempted. Finally, additional flight testing of the pro posed RSS would be most desirable.

APPENDIX A DEFINITION OF STAB-ILITY DERIVATIVES (45) A.]

Axis Systems XBU0,P.

X VT S REFERENCE Zr ZB' W-0'r FIGURE 69. AXIS SYSTEMS XB' YB' Z - The Body Axis System consists of a right-handed, orthogonal axes whose origin is fixed at the nominal aircraft center of gravity.

Its orientation remains fixed with respect to the aircraft,, the XB and Z axes being in -the plane of' symmetry.

The exact alignment of XB is arbitrary.

hereif it Vs taken along the body cenzerline reference.

XS, YS, ZS - The Stability Axis System is that particular body axis system for which the XS. axis is coincident with the projection of the total steady-state velocity vector (Vo) on the aircraft's plane of symmetry. Its orientation remains fixed with respect to the aircraft.

A.2 Definition oflNondimensional Stability Derivatives Nondimensional stability derivatives are defined with respect to body fixed stability axes in standard NASA form (e.g., (46)).

A.3 Transformation of Stability Axis Derivatives to Body Axis A.3.l Longitudinal Derivatives CN + = CL Cos a0 CD sin a0 C = CD cos - L sin a X 0 0 + C C Cs CL sin a + C sin a0 C CsO 0 O Ca Co L 0 Da D 0c Na CL cos aCn Cos CN CL. cosct

a. a

C =0 cos N L 0 q q C C cosa + sinas O CNM = CLM + CDM C =C Cosca sin + a O N L 0 D 0 S6 C = CD cosa -C sin a' - C' sin a - C cos a

D 0 ya 0 L 0

XaU Dai 0 CX.

= -CL.

sin a0 a a CXq -CLq sin a0 L 0 C =C Cos -0 sin a XM M 0 L 0LM 0 C C Cos -0 sin a X 0 0 L0 - unchanged Cm Cm , CM.' Cm C C q' MM A.3.2 Lateral Derivatives (C )B = Cos a0 -C s.in a0 ) sin cos a0 + C n n 2 S0 (C I)B = C1 cos2 a (CI + C r p p r p = ) B I cos2 a0 (C - CI sin a0 cos a0 C sin a (C) 1 0 r r r p p ) C cos - sin a (C O O 6 6

(Cl ) =c

c (Cn )B = n cos t + C sin a O 1 O (Cn)B =C n Cs2 a " (Cn -CI sin cos a- C sin a 0 0 0 p p r p r a a + C sin + C ) sina cos =n cos2 a + (C, (C)B 1p0 0 11 np0 0- nrB nr r r r p p (Cn)B = Cn6 Cosa + Cl1 sin U C - unchanged C ,

y y

Definitions A.4 Dimensional Stability Derivative A.4.1 Longitudinal Derivatives U V cos a o ~ W = sin a .

o 0 wo pSUo 0 - C + 2 X x 2 Cx ) u m X* =X + T cos X psu -C -2 (Cx X 2 J La 0 w 2m V 0 pS T X6 = 2m CX6 SUoM ) C Zu = - ("- N - CN+ N' WO u m 2 N m N 2U 0 N.

*=Z z - T sin 0 w I- N a U0 N 2 2 ps& 4m V CN, w 0 Z4V pST2 N Zf 2m

M fS-jo !! + C W

u Iy 12 mM m 2U a l M* M + th u u u I Y M [ 2Wo PSEUo C +-(C +- CmM)

M 2yI Ma U

m mJ

M. sz2U0

w 1y Mq 41 y VT m PSE U pSF2V y q pSCVT0 MS 21 0 Cm Y &T a = T am TM u A.4.2 Lateral Derivatives pSVT V 2m Y PSVT Y6 = 2m Cy V o pSVT b L= ( 2--) x I b pSVT = L 4( 1) c 'x p b2 pSVT L = (41 0 C r x r pSV 2b 0 -) L ( -2-- x 6 pSVT 2b NS = ( 2- ) Cn pSV b T Np = 41 C n z p pSV b Nr = ) n4--- r z r pSV 2 T 2-1 Cn6 z " Iz N/I)G L = (L + (Lp xz p x)G Lp = (L + Ixz Nr/Ix)G L r r ' N6 /Ix)G L6= (L + Ixz Ls/Iz)G N' =(N + z = Ix LpIz)G Np' (Np + p z ) (Np xz Np Lr/z)G xz N r (Nr + Ixz L /Iz)G (N + N6= 6 - = where G I 1 xx I I x z

APPENDIX B TURBULENCE FILTERS AND INPUT-OUTPUT RELATIONSHIPS Under the assumptions on the statistilcal properties of turbulence cited in Section 2.1, it has been shown (47) that the power spectral density of ( ) any aircraft system output quanti-ty of interest, 0 w , can be re'lated to 2 the input power spectral density y.(m) through IG(W) 1 , the square of the modulus of the appropriate transfer function ) G( ) (2 w (B.1) at a given unit sinusoidal frequency w. The root-mean-square value of the output, a, is then given by the integral of the output power spectral density taken over all spectral frequencies: o = [ 0() d() .

(B.2) The root-mean-square value, identical to the variance for a proces with zero mean, is one of the most useful quantities in describ'ing the magnitude of response. The average frequency of exceeding a peak response level can also be related to the power spectral density.

Formulations for the power spectral density of the components of atmospheric turbulence are given in Reference 32. Two forms are generally used-- the Dryden and Von Karman. Although the Von Karman description has been shown to more closely match actual measured spectra, the Dryden form has the advantage of being spectrally

PflECEDI

PAGE BLAiK

NOT fltM

Laplace notation, in expressed function, the transfer Thus, factorable.

reason, an For this available.

input is a white-noise filtering for The turbulence this ,study.

is used in form to the Dryden approximation follows: as defined are functions transfer wL L L 2 aw rVT G A s) 0 (I + w TO VI s (T.4)

s)o s)(B4

( 4b

A = ~ G(g. A

Gg "V -T'0 ___L V (B.5) 2\.-.

n7

CTI

s) - (1 + T Bg.

r

(B.6)

-_G +9 's

G 9

A A w 'U pg(7 -/08TL 1/3 w ls(B.7) +I 4b

_\/L-- T

G 9 4b~ .~ A w LT where and for vertical lengths gust characteristic L are the Lw, respectively, fields, turbulence lateral and the aircraft, span of wing reference is the b aircraft.

of the velocity steady-state total is the V T til I. r[ itk, :; .j 'I , Jvs.;-.!

JIi .. ?I( ''' 1 $r Note that the expressions for q, r, p9 are strictly valid only for very low frequencies. For clear air turbulence, at altitudes above 533. meters (1750 feet), Lv and L are taken equal to 533 meters (1750 feet).

For lower altitudes, the suggested values are L = h meters and w L = 36.2 hIl/3 meters.

The probability v of exceeding a given a once w turbulence has been encountered is given by ) =ep(2 c2 ( exp W (B.8) where c = 0.7 m/sec (2.3 ft/sec).

Finally, the following similarity relationship is given in Reference 32: v o2 L 2 (B.9) v w

APPENDIX C JETSTAR DATA JetStar Power Approach Configuration ) ft (542.,5

m

50.4

= S = 3.3 m (10.9 ft) b = 16.4 m (53.75 ft) VT = 72.1 m/sec (236.7 ft/sec) h = 305 m (1000 ft) = 142300 N (32,000 Ib) W slug-ft (62400 84900 kg-m = I x2 2 I = 272000 kg-m (200000 slug-ft Y ) slug-ft I =204000 kg-rn (150000 z slug-ft (550 kg-m = 750 I xz 2 2 ) W/S = 2824 N/m (59.0 lb/ft Dimensional Stability Derivatives fuselage reference,.line) (XB axis aligned with = = CLo = 0.88 X * = -0.0058 I/sec U X = 0.1040 I/sec w X. = 0.0 = 0.0 m/sec (0.0 ft/sec) X 2 2 X6 = 1.0298 m/sec /rad (3.3787 ft/sec /rad) (1.2719 ft/sec 2/rad) = 0.3877 m/sec2/rad e ~f

PAGE BLANK NOT FIXm

PRECEDING

= -0.0991 1/sec Z * u 1/sec = -0.9192 Z w =0.0 Z- w (0.0 ft/sec) = 0.0 m/sec Z q (-17.3823 ft/sec2/rad) 2/rad = -5.2981 m/sec Z e22 2/rad)- ft/sec 2/rad (-6.5434 -1.9944 m/sec 2 a = S6f 1/ft-sec) (0.0019 = 0.0062 I/m-sec M * u (-0.0081 I/ft-sec) -0.0266 I/m-sec M = I/m (0.0 1/ft) M. = 0.0 w I/sec = -0.9180 Mq I/sec -2.5798 = M6e 1/sec -0.1131 = M -0.1226 I/sec y = V I/sec y * = -0.0061 6a I/sec y * = 0.0473 a r 1/sec = 0.0167 y * sfg I/sec ' = -4.0765 L 1/sec.

' = -0.9763 L P I/sec = 0.3842 L r r 1/sec 1.3736 = ' L 1/sec = 0.6888 L6 r I/sec 0.2681 = L 5 fg ]/sec 0.8736 = N 1/sec N ' = -0.1655 p N ' = -0.1617 I/sec r I/sec 0.0932 N a l/sec -0.6051 = N r l/sec 0.0493 = N sfy Surface Actuator Dynamics = , G e (1 + S/lO0) 6e G6dlf dlf (I + S/40)2 c G a 6a (I + S/50)2 C S2 I1 _ 6r r c 227 (I + 2(0.25) S + ) = G 6sfg sfg (1 + c S/30) Maximum Deflections and Rates for Force Control Surfaces Direct Lift Flaps: 6max = 27;max 52/sec

Side Force Generators: 6max = 240; 6 = ± 37 /sec

1max

APPENDIX D FUNCTIONS FOR FORMULATION OF TRANSFER MULTI-LOOP FEEDBACK CONTROL SYSTEMS expanding the transfer function The theoretical framework for is presented formulation to include multi-loop control feedback loops rules are as McRuer et a]. For negative feedback systems, the by follows: to: .1. The effective numerator is equal a. The open loop numerator; b. Plus the sum of all the feedback transfer by the appropriate functions, each one multiplied coupling numerator; effective denominator is equal to: 2. The a. The open-loop denominator; feedback transfer b. Plus the sum of all the functions, each one multiplied by the appropriate numerator; c. Plus the sum of all the feedback transfer taken two at a time, each pair functions multiplied by the appropriate coupling numerator. (Reference 48, page 95-.)

' q + 62) the effective Thus, for two loops closed (e.g., q, - 61, as: function for output q, due to input of 6. is written transfer

R OEzIzN PAGE Rray NOT

q, 6~ q~q 62 q~ I + G N + G N q 6 ( I qD.N) q 62 a i G q q(D N q1 +C 6.. N 6 6 N 6 6 "1 1 2 q 12 l 1 2 where A is the open-loop characteristic denominator. Numerators of 9i the form N j are formed by simply applying Cramer's rule to the air craft equations of motion written in the Laplace variable s (i.e., replacing the column corresponding to q, by the input vector 6.).

qiqJ Coupling numerators of the form N are formed by computing the determinant of the matrix of the aircraft equations of motion with the two columns corresponding to qi and q. replaced by the control vectors corresponding to 6j and k simultaneously. If 6 6k or q. = q the determinant is defined as zero.

For the Longitudinal Ride Smoothing System, with unequalized feedbacks a - 6f, a 0 6e, the transfer function of interest is: a azG a w Nw6 N z K 6 Gw a a e (D.2) SA-K N z-K N KeN +K a O 1 Kaz f 6 6 Kaz e az f e where K > 0 and Ka > 0.

z For the Lateral Ride Smoothing System, with unequalized feedbacks 6 6 a + sfg' r + r' the transfer function of interest is: a a r Y N - K Ny Ga yg r 1q6r(D3 +K a a r ' (D3) 2 2 r Ka r N6Y ay N6Sfg r 6r y sfg r where K > 0 and K > 0.

a r a .a a Expressions for , NZ N N, A N Y Nr can g f e be found in ,sfg Nr 46.

Reference For the JetStar aerodynamics, the following coupling numerators were derived: a Nw6 - w Nz S 2{M zw Mw ge e e + s{M (X z* - xj!z -. w u e + x (MW u *z -HZ ) e + 2Xa (Mu*ZX - M Z X (.4 ,e +fe 6 (Xu" *X) (D.)

aO ee + s{M.

(X Z - X 6 Z feef u , u + X (M.

z M -Z -) f 6e u ul 6e- Z (Mu*X - M.X X*)} " u e u e, 6f N s V NV 'y - N 'y * 0 6 g r VTo r v 1 r) + syN NL") - Lp',N + Y' - L 'N I1 v'6 p p.

p 0r r r, p (.)183 'Y N *} as {N Nsfg~r T (Sr sfg- sfg (r 'N ') '- *(L 'N s V {Y + 6 T sfg P 0 r. sfg r ')} ' - Lp'N 6 r'N sfg*(L + y 6 * s VTo sin 0 0{Yv(L 6sfgN r - L 6rN 6sg1 r ssf 'sfg r Nr) L ''

+ Y *(L e N sN' - L

r sfg sI) r 'N - LNN rI s* 0(LN 6 + y 6 N' - L N 'N) + ( 6 6 rNs rsf srfg {Y(L 'N ' 6 - L' 'rIN .') rf*(L ' - L 'N') + Y 6 8 6 6 s + y sg*QL r'N I - L8 IN ')} (D.7) 6 6 6 is defined as the steady qualities parameter n/a The handling change in angle of attack acceleration change per unit state normal speed (36). This elevator deflection at constant for an incremental stability derivatives written in terms of the dimensional factor, as: defined previously, is expressed I .a u fl (s) o 6 (D.8) n (g/rad) (s) g N6 e evaluated at s E 0 (45). The notation N implies that the short-period approximation is used to evaluate the above transfer functions. From the rules given, it follows that: a a ^z ^z N =N , (D.9) 6 6 e e -wI w 6f w az N(e =Ne N e=N aGa z Ne~f N6 e6f (D.10) 6 6 where z 9 sin 6o(Mw - M z) 6 6 Ne (D.ll) " e W ^w N = -g sin 8 M6 (D.12) e e a ef sin fe MZ ef (D.13) Thus, ge e (g/rad) (D.14) M6e a z f e r / For the JetStar MwZ << M Zw , and (D.15) 6 6 e e (D.16) M << M Z 6 6 6 6 fe Mef

APPENDIX E STOL DATA Power Approach Configuration Buffalo S-11 (600 ft

55.7 m

ft )

= 87.8 m (945

S c = 3.1 m (10.3 ft) 3.0 m (9.8 ft) b = 29.3 m (96.0 ft) 20.2 m (66.2 ft) V = 38.6 m/sec (126.5 ft/sec) 36.0 m/sec (118.2 ft/sec) h = 305 m (1000 ft) 305 m (1000 ft) W = 145400 N (32683 Ib) 213500 N (48000 Ib) 2 2 I = 375800 kg-m (276300 slug-ft ) 289000 kg-m (213000 slug-ft2) x 2 2 Iy = 303400 kg-m (223100 slug-ft

) 315000 kg-m (232500 slug-ft 2)

2 2 2 2 ) I = 625500 kg-m (459900 slug-ft ) 546000 kg-m (402500 slug-ft z 2 2 I =410k-2 ) 42200 kg-m (31150 slug-ft kg-n (29500 slug-ft

4xzio

lb/ft 3833 N/m (80 ) (34.6 lb/ft = 1656 N/mr WS Dimensional Stability Derivatives (XB axis aligned with fuselage reference line) Buffalo .

S-1l 4.90 = -2.40 a O 4.90 = -2.40 CLo = 1.85 .

4.79 u, = -0.0859 I/sec -0.0200 1/sec X = 0.1396 1/sec 0.0935 I/sec w

xW = -0.00035 0.0

Xq = 0.0 m/sec (0.0 ft/sec) 0.0 m/sec (0.0 ft/sec)

NOT FILMED

BLANK

PAGE

pREYEDING

S-11 Buffalo 0.8869 m/sec /rad /rad 0.0 m/sec2 X e (0.0 ft/sec 2/rad) (2.9098 ft/sec 2/rad).

2/rad -1.8745 m/sec = -2.2855 m/sec2/rad X f (-7.4985 ft/sec2 /rad) (6.1500 ft/sec 2/rad) -0.5055 I/sec I/sec.

Z * = -0.5503 U I/sec -0.4829 1/sec = -0.8216 Z W 0.0 Z. = -0.0083 w ft/sec) 0,0 m/sec (0.0 ft/sec) m/sec (-5.8313 Z = -1.7774 , 2/rad.

-10.5619 m/sec /ad = -3.0533 m/sec2 Z 6e (-10.0175 ft/sec 2/rad) (-34.6519 ft/sec /rad) /rad m/sec -2.5452 /rad -5.7892 m/sec = 2 2 f (-18.9935 ft/sec /rad) (-8.3504 f-t/sec rady 0.00062 1/m-sec M * = 0.0023 1/m-sec (0.90019 1/ft-sec) (0.00069 I/ft-sec) u -0.0073 I/m-sec -0,.0539 l/m-sec M w I/ft-sec) (-0.002238 1/ft-se) (-0.61644 I/ft) 1/m (0.0 I/ft) b.o (-0.00i678 = -0.0055 I/m ,M 0.9014 1/sec " - I/sec- = -1.3817 M I/sec -1:4203 I/sec -2.0152 M6 I/sc2 0.0276 I/sec -0.02612 M6 'f -0.1600 I/sec = -0.1577 I/sec Y v I/sec -0.00551 1/sec Y6* = 0.000194 a 0..0349 I/sec I/sec Y * = 0.0570 r I/sec 0.,03354 I/sec = 0.03133 Y6fg 'sfg22 -0.9411 1/sec = -0.7881 ec L -0.3533 1/sec 1/sec = -1.4553 P 0.6986 I/sec 1/sec Lr = 1.1771 S-I Buffalo /sec 0.7476 /sec = 0.3138 L i/sec 0.2116 1/sec = 0.2776 L 2 1/sec 0.0 1/sec = 0.0 L6 6 sfg22 0.6372 1/sec N = 0.4590 I/sec -0.1389 I/sec N ' = -0.1988 I/sec p N = -0.2985 I/sec -0.0957 I/sec r 0.1583 l/se I/sec = 0.0133 N ' 2 I/sec -0.3647 I/sec = -0.6527 N6 2 1/sec 0.0 I/sec = 0.0 N 6 8fg REFERENCES I. Phi;llips, William H.: Gust Alleviation, in "Performance and Dynamics of Aerospace-Vehicles."- NASA Langley Research Center. NASA SP-258, 1971, pp. 521-530, 548.

,2. Zbr6zek, F.; Smith, K. W.; and White, D.: Prelfminary Report on a Gust Alleviator Investigation on a Lancaster Aircraft. Reports and Memoranda No. 2972, British A.R.C., August 1953.

3. Zbrozek, F. K. Theoretical Analysis of a Gust Alleviator Used on a Lancaster Aircraft and Comparison with Experiment.

RAE Report No. AERO 2645, January 1961.

4. Hirsch, R. L'Absorption des Rafales sur Avions et Resultats des Essais en Vol d'un Appareil Experimental. Doc-Air-Espace, no.

105, July -1967, pp. 41-56.

5. Phill'ips, Wi-lliam H.; and Kraft, C. C., Jr.: Theoretical Study of Some Methods for Increasing the Smoothness of Flight Through Rough Air. NACA Technical Note 2416, July 1951.

6.

Kraft',Christopher C., Jr.;'and Assadourian, A,: Experimental Study of an Angle-of-Attack Vane Mounted Ahead of the Nose of an Airplane for Use as a Sensing Device for an Acceleration Alleviator.

NACA Technical Note 2415, February 1951.

7. Boucher; Robert W.; and Kraft, C. C., Jr.: 'Analysis of a Vane- Controlled Gust-Alleviation System. NACA Technical Note 3597, April 1956.

8. Kraft, Christopher C., Jr': Initial Results of-a Flight Investigation of a Gust-Alleviation System. NACA Technical' Note 3612, April 1956.

9.

Hunter,-Paul A.; Kraft, C. C., Jr.; and Alford, W. L.: A Flight Investigation of an Automatic-Gust-Alleviati6n System'in a Transport Airplane. NASA TN D-532, January 1961.

I,. Schott, Russell L.; and Hamer, H. A.: Flight Inves-tigation of Some Effects of a'Vane-Controlled GustLAlleviation System on the Wing and-Tail Loads of a Transport Airplane.

NASA TN D-643, -January 1961.

11. Barker, L. Kei-th; and Sparrow, G. W..: Analysis of Effects of Spanwise Variations of Gust Velocity on a Vane-Controlled Gus't-Alleviation System.'- NASA TN D-6126, April 1971.

12. Barker, L. Keith: Effects of Spanwise Variation of Gust Velocity on Alleviation System Designed for Uniform Gust Velocity Across Span. NASA TN D-6346, June 1971.

PRECEDING PAGE 3LANK NOT FIM 13. Barker, L. Keith; Crawford, D. J.; and Sparrow, G. W.: Effect of Limited Amplitude and Rate of Flap Motion on Vane-Controlled Gust-Alleviation System.

NASA TN D-6723, March 1972.

14. Oehman, Waldo I.: Analytical Study of the Performance of a Gust Alleviation System. for a STOL Airplane. NASA TN D-7201, 1973.

15. Phillips, William H.: Study of a Control System to Alleviate Air craft Response to Horizontal and Vertical Gusts NASA TN D-7278, December 1973.

16.

Oehman, Waldo I.: Analytica'l Study of the Performance of a Gust Alleviation System with a Vane Sensor. NASA TN D-7431, February 1974.

17. Tobak, Murray: On the Minimization of Airplane Responses to Random Gusts. NACA Technical Note 3290, October 1S57.

18. Coupry, Gabriel: Some Results on Gust Alleviation, in "Proceedings, Royal Aeronautical Society, International Conference on Atmospheric Turbulence, London, Engl'and," May 18-21, 1971.

(A71

29785)

19.

Tsumura, Toshihiro; Nakagawa, K.; and Murotsu, Y. Investigation of Gust-Alleviation System for Transport Airplanes.

Osaka.

Prefecture University Bul.letin, Series A, Engineering- and Natural Sciences, vol. 14, no. 2, 1965-, pp. 59-80.

20. Nakagawa, Kenji; Murotsu, Y.; and Fujiwaca, N.: The Optimization of Airplane Gust-Alleviation System.- Osaka Prefecture- Universi-tyBulletin, Series A, Engineetfing and Natural Sciences, vol. 15, no. 2, 1966, pp. 35-48.

21. Holloway, R. B.; Thompson, G. 0.; and Rohling, W. J.: Prospects.

for Low Wing-Loading STOL Transports with Ride Smoothing.

AIAA J. of Aircraft, vol.

9, no. 8, August 1972, pp. '525-530.

22. Gordon, C. K.; and Dodson, R. 0.: STOL Ride Control Feasibili'ty Study--Technical Report. NASA CR 2276, December 1972.

23.

Hess, Ronald: An Application of Optimal Stochastic Control Theory to the Problem of Aircraft Gust Alleviation. -Ph.D. Dissertation, University of Cincinnati, College-of.Engineering, July 1970.

24. Clcment, Russell Lee: A Parameter Optimization to Aircraft Gust Alleviation. M.S.(A.E.) Thesis, Naval Postgraduate School, March 1972..

25. Hess, Rona-ld A.: Some Results of-Suboptimal Gust Alleviation.

AIAA J. of Aircraft, vol.

9, no. 5, May 1972, pp.

380-381.

26. McClean, R.: The Optimization of an Autopilot for an Airplane Subjected to Random Atmospheric Turbulence. UTIA Technical Note No. 45, University of Toronto, Institute of Aerophysics, November 1960.

27. 1l'iff, K. W.: Identification and Stochastic Control with Application to Flight Control in Turbulence. Ph.D. Dissertation, University of California at Los Angeles; UCLA-ENG-7340, May 1973.

28. Swaim, Robert L.,: Aircraft Elastic Mode Control. AIAA J. of Aircraft, vol. 8, no. 2, February 1971, pp. -65-71.

29. Smith, Ralph E.; Lum, E.. L.; and Yamamoto, T. G.: Application of Linear OptimalTheory to Control of Flexible Aircraft Ride Qualities. AFFDL-TR-67-136, Air Force Flight Dynamics Laboratory, January 1968' 30. Morris, R. L; Hanke, C. R.; Pasley, L. H.; and Rohling, W. J.: * The Influence of Wing Loading-on Turbofan PoweredZSTOL- Transports with and without Externally Bloan Flaps.. -NASA CR-2320, November 1973.

31. McRuer, Duane,; Ashkenas, 'Irving; and Graham,,Dunstan., Aircraft -Dynamics and Automatic Control.

Naval Air Systems Command, -Department of the Navy,; Washington, D.C,., August 1968.

Chapter 4: Vehicle-Equations of Mot-ion.

32.'- ChaIk,>,C. R.; Neal, T. P.; Harris, T. M.; Pr-i.tchard, F. E.; and "Woodcock,'R.

J.: Background-information and User Guide for - MIL-F-8785B'(ASG), Military Specification, Flying Qualities of Piloted Airplanes. AFFDL-_TR-69%72, Air Force F-light Dynamics Laboratory, 1969. Section 3.7, Atmospheric Disturbances.

33. Pratt, Kermit G.: Response of Flexible Airplanes to Atmospheric Turbulence, in "Performance and Dynamics of Aerospace Vehicles"' NASA.Langley Research-Center. NASA-SP-258, 1971, pp. 450-460.

34. Stone, Ralph W., Jr.: Ride Quality Overview, in "Symposium on Vehicle Ride Quality." NASA.TM X-2620, October 1972..

35. Jacobson, Ira D.; ,and Richards, L. G.: Ride Quality Evaluation II: Modelling of-Airline Passenger Comfort. STOL Memorandum Report 403217, Dept. of Engineering Science and Systems, Univ.

1 of V.irginia, Charlottesville, December 1974.

36. Anonymous: Military Specification--Flying Qualities of Piloted Airplanes. MIL-F-8785B(ASG), August 7, 1965.

37. Barnes, A. G.: C.S.A.S. Desi.gn for Good Handling Qualities in Turbulence, in "Flight in Turbulence." AGARD Conference Proceedings No. 140, Paper No. 21, May 1973. 193 38. Cooper, G. E.; and Harper, R. P., Jr.: The Use of Pilot Rating in the Evaluation of Aircraft Handling Qualities. NASA TN D-5153, April 1969.

39. Stapleford, R. L.; McRuer, D. T.; Hoffman, L. G.; and Teper, G. L.: A Practical Optimization Design Procedure for Stability Augmentation Systems. AFFDL-TR-70-11, Air.Force Flight Dynamics Laboratory, 1970.

40. Clark, Daniel C.; and Kroll, John: General Purpose Airborne Simulator--Conceptual Design Report. NASA CR-544, August 1966.

41. Musick, R. 0.; and Wagner, C. A.: A Flight Simulator Control System Using Electric Torque Motors. AIAA Paper 75-105, AIAA 13th Aerospace Sciences Meeting, Pasadena, California, January 20-22, 1975.

42. Myers, Albert; and Evans, Martha: SIM II Users Guide. Programming Branch, NASA Flight Research Center, Edwards, California, May 22, 1974.

43. Washington, Harold P.; and Gibbons, John T.: Analytical Study of Performance for a Jet STOL Transport Takeoff and Landing Configuration with Full-Span, Externally Blown, Triple- Slotted Flaps. NASA TN. D-7441, October 1973.

the 44. Powers, Bruce G.; and Kier, David A.: Simulator Evaluation of Qualities of an Experimental STOL Configura Low-Speed Flying tion with an Externally-Blown Flap Wing or an Augmentor Wing.

NASA TN D-7454, October 1973.

45. Heffley, Robert K.; and Jewell, W. F.: Aircraft Handling Qualities Data. NASA CR-2144, December 1972.

bf a General Purpose Airborne 46. Szalai, Kenneth J.: Validation Simulator for Simulation of Large Transport Aircraft Handling Qualities. NASA TN D-6431, October 1971.

47. Etkin, Bernard: Dynamics of Flight. John Wiley and Sons, Inc., 10.)

1959. (Chapter 48. McRuer, D. T.; Ashkenas, I. L.; and Pass, H. R.: Analysis of Multiloop Vehicular Control Systems. Air Force Flight Report ASD-TDR-62 Dynamics Laboratory Technical Documentary 1014, March 1964.

DISTRIBUTION LIST Copy No.

1 - 2 NASA Scientific and Technical Information Facility

P.O. Box 8757

Baltimore/Washington International Airport Maryland 21240 3 - 5 Life Sciences Division NASA/Flight Research Center P.O. Box 273 Edwards, California 93523 Thomas Wolf Attention: Mr.

6 - 7 I. D. Jacobson 8 - 9 Maris Lapins 10 A. R. Kuhlthau 11 - 12 E. H. Pancake Hall Clark 13 I. A. Fischer RLES Files A. T. Symmers 15 - 30 UNIVERSITY OF VIRGINIA School of Engineering and Applied Science The University of Virginia's School of Engineering and Applied Science has an undergraduate enrollment of approxirfiately 1,000 students with a graduate enrollment of 350. There are approximately 120 faculty members, of whom, about 90% hold a doctorate.

Excellence in graduate education is aided and supplemented by a research program approximating $3 million per year. This relatively high level of participation in sponsored research is one factor which helps our faculty consistently to maintain high quality graduate education at all degree levels.

As research is an integral part of the educational program, research interests parallel academic specialities. These interests range from the traditional engineering departments of Chemical, Civil, Electrical and Mechanical to include departments of Biomedical Engineering, Engineering Science & Systems, Materials Science, Nuclear Engineering, and Applied Mathematics & Computer Science. In addition to these departmental interests, there are interdepartmental groups in the areas of Automatic Controls and Applied Mechanics. All departments are authorized to offer the doctorate while the Biomedical and Materials Science Departments are graduate degree granting departments only.

The School of Engineering and Applied Science, is an integral part of an outstanding University, which has strong professional Schools of Law, Medicine, and Business Administration.

In addition, the College of Arts and Sciences has strong basic science departments in Mathematics, Physics, Chemistry, and other departments relevant to the engineering research program. This not only provides an excellent scholarly climate, but also enhances the school's potential for creating truly interdisciplinary teams in the pursuit of our basic goals of education, research, and public service.

Inside this cover are listed some of the present research activities of the department from which this report originates. For more information on this or other areas, address the department chairman or Dean J. E..Gibson, Commonwealth Professor and Dean, School of Engineering and Applied Science, University of Virginia, Charlottesville, Virginia 22901.