ue

P199324 >T°0

AGARD-AG-175
in

<•
Q
cc
<
<

AGARDograph No. 175

on

Active Control Systems for

Load Alleviation, Flutter Suppression
and Ride Control

DISTRIBUTION A N D AVAILABILITY
O N BACK COVER
 AGARD-AG-I75

NORTH ATLANTIC TREATY ORGANIZATION

ADVISORY GROUP FOR AEROSPACE RESEARCH AND DEVELOPMENT

(ORGANISATION DU TRAITE DE L'ATLANTIQUE NORD)

AGARDograph No. 175

ACTIVE CONTROL SYSTEMS FOR LOAD ALLEVIATION,

FLUTTER SUPPRESSION AND RIDE CONTROL

This AGARDograph was sponsored by the Structures and Materials Panel of AGARD.
 THE MISSION OF AGARD

The mission of AGARD is to bring together the leading personalities of the NATO nations in the fields of
science and technology relating to aerospace for the following purposes:

— Exchanging of scientific and technical information;

— Continuously stimulating advances in the aerospace sciences relevant to strengthening the common defence
posture;

— Improving the co-operation among member nations in aerospace research and development;

— Providing scientific and technical advice and assistance to the North Atlantic Military Committee in the
field of aerospace research and development;

— Rendering scientific and technical assistance, as requested, to other NATO bodies and to member nations
in connection with research and development problems in the aerospace field;

— Providing assistance to member nations for the purpose of increasing their scientific and technical potential;

— Recommending effective ways for the member nations to use their research and development capabilities
for the common benefit of the NATO community.

The highest authority within AGARD is the National Delegates Board consisting of officially appointed senior
representatives from each member nation. The mission of AGARD is carried out through the Panels which are
composed of experts appointed by the National Delegates, the Consultant and Exchange Program and the Aerospace
Applications Studies Program. The results of AGARD work are reported to the member nations and the NATO
Authorities through the AGARD series of publications of which this is one.

Participation in AGARD activities is by invitation only and is normally limited to citizens of the NATO nations.

Part of the content of this publication has been reproduced directly

from material supplied by AGARD or the authors; the remainder has
been set by Technical Editing and Reproduction Ltd.

Published March 1974

629.73.062 - 52 : 533.6.013.422 : 533.6.048.1

*
Printed by Technical Editing and Reproduction Lid
Harford House, 7-9 Charlotte St. London. W1P IHD
 PREFACE

In view of the ever increasing performance of modern aircraft, wing safety must be
improved, without any weight penalty, through devices permitting the suppression of
critical flutter speeds, on the one hand, and, on the other, the alleviation of maneuver or
gust induced loads. Such devices are always "active" systems based on the feed-back
principle.

The present AGARDograph collects, in a single volume, several papers on this theme
presented on the occasion of a Specialists' Meeting organized in The Hague in October 1973
under the joint sponsorship of two separate Working Groups of the AGARD Structures and
Materials Panel: one of these groups deals with the interactions and maneuverability of
aircraft in flight, the other with aeroelasticity and unsteady aerodynamics.

Major current problems such as flutter suppression and the reduction of loads induced
by lateral gusts are reviewed by the authors who describe the solutions either developed or
in the course of development in their respective countries: United States, Great Britain,
Germany and France. When reading these texts, one is struck by the ingeniousness and
variety of the solutions presented, as well as by the fact that some devices which have been
used for a long time to improve maneuverability prove efficient in a much broader field;
this would justify, if such a justification was still needed, the necessity of an increasing
collaboration between research workers specialized in closely related subjects.

We express the hope that, by presenting solutions which may still be unknown by
many, this AGARDograph will promote useful thoughts and contribute to the development
of the means used to increase the safety and comfort of aircraft simultaneously with their
speed.

R.MAZET
Chairman of the Working Group on
Aeroelasticity and Unsteady Aerodynamics

iii
 CONTENTS

Page

PREFACE iii

INTRODUCTION v

EFFECT OF YAW DAMPER ON LATERAL GUST LOADS IN DESIGN OF THE

L-l Oil TRANSPORT
by F.M.Hoblit 1
THE EFFECT OF ACTIVE CONTROL SYSTEMS ON STRUCTURAL DESIGN
CRITERIA
by N.F.Harpur 11

STATUS OF TWO STUDIES ON ACTIVE CONTROL OF AEROELASTIC RESPONSE

AT NASA LANGLEY RESEARCH CENTER
by I.Abel and M.C.Sandford 23

CONTROLE DU FLOTTEMENT PAR DEPLACEMENT DUNE VALEUR PROPRE

par J.AngiMini 49

ACTIVE FLUTTER SUPPRESSION ON WINGS WITH EXTERNAL STORES

by (.11.mil. A.Lotze and O.Sensburg 57

iv
 INTRODUCTION

ACTIVE CONTROL SYSTEMS FOR LOAD ALLEVIATION,

FLUTTER SUPPRESSION AND RIDE CONTROL

Modern active or feedback control technology has been applied for perhaps a quarter of a century now to
augment handling qualities and stability and control characteristics of aircraft. Moreover, many important contribu-
tions have extended applications of active control technology in a broader manner and to other areas. Maneuver
and gust load reductions, fatigue damage reductions, load distribution control, ride comfort improvement, and
relaxation or reduction of unaugmented static stability requirements are some present or near-future applications.
For many of the above phenomena, not only the more or less rigid body responses have been controlled, but also
the significant responses of the lowest vibration modes of the flexible aircraft have been reduced. Novel and very
important contributions are now being made to the theory and hardware pertinent to higher frequency ranges of
flexible aircraft. "Active flutter suppression" is a distinct possibility in some applications.

In view of the above progress, the Working Group on Interaction of Handling Qualities, Stability and Control
on Structural Loads together with the Working Group on Aeroelasticity and Unsteady Aerodynamics recommended
a joint conference which was approved by the AGARD Structures and Materials Panel at the April 1973, 36th
meeting in Milan, Italy. Consequently, and in consonance with the mission requirements of AGARD as stated on
page ii, five pilot papers were presented during the Opening Ceremony (afternoon) Session of the 37th
meeting of the Structures and Materials Panel, 8th October 1973, Den Haag, Netherlands. The purposes of the
papers were to describe the status and impact of the rapidly expanding domain of aircraft active control relative to
our Panel activities. In particular, these limited pilot papers were primarily selected to define (a) the present status
of industrial applications of active control technology in reducing loads on modern aircraft, and (b) the future
potential of active control for aircraft flutter suppression.

Both Working Groups and the Panel are grateful to Dr Mazet for competently acting as Chairman of this
joint meeting, to the authors for their high quality papers, and to the audience for the lively and interesting
discussions. In view of the success of this meeting and accelerating progress in this most important technological
domain, further such joint meetings are envisioned.

WALTER J.MYKYTOW Dr JOHN C.HOUBOLT

Vice Chairman Chairman
Working Group on Aeroelasticity Working Group on Interaction of
and Unsteady Aerodynamics Handling Qualities, Stability and Control
EFFECT OF YAW DAMPER ON LATERAL GUST LOADS IN

DESIGN OF THE L-1011 TRANSPORT

by

Frederick M.Hoblit

Commercial Project Loads

Lockheed-California Company, P.O. Box 551
Burbank. California 91520, USA
 EFFECT OF YAW DAMPER ON LATERAL GUST LOADS
IN DESIGN OF THE L-1011 TRANSPORT

by

Frederic M. Hoblit

SUMMARY

In the design of the L-1011 transport, the reduction of lateral gust loads, in continuous turbulence, due to the presence of
a yaw damper was reflected in the limit design loads. The resulting load reduction was about 27 percent. In establishing the limit design
loads, both the mission analysis and design envelope forms of continuous turbulence gust loads criteria were used. Account was taken,
under both forms of criteria, of the fraction of time the damper might be inoperative. The effect of saturation of the damper at the
limit-load level was also taken into account. This effect was determined by means of time-history analyses in which the input was a
random gust velocity and the rudder angle limits (governed by available hinge moment) were included in the simulation.

INTRODUCTION

In the design of the L-1011, it was known from the start that this airplane, like virtually all other modern transports,
would have a yaw damper. It was also known that in actual flight the gust loads - on the vertical tail - would be lower because of the
damper. And it was decided, therefore, at an early stage of design to reflect this reduction in the limit design loads.

The purpose of this paper is to provide further background relative to this decision, to indicate how the presence of the
damper influenced the procedures by which the design loads were obtained, and to comment briefly on what was learned from loads
measurements in flight through turbulence.

For the purpose of gust loads determination, two mathematical idealizations of the atmospheric gust structure have been
in wide use. The first is the discrete gust idealization, in which the gust structure is considered to consist of individual gusts of standard
(usually one-minus-cosine) shape, variable gust length or gradient distance, and a maximum gust velocity independent of gust length.
The other is the continuous turbulence idealization, in which a stationary Gaussian gust structure is assumed. Neither Idealization even
approaches describing accurately all atmospheric turbulence. But there seems to be a growing acceptance of the continuous turbulence
idealization as being by far the more realistic for the purpose of predicting airplane loads at the limit or ultimate load level, as well as at
lower load levels pertinent to structural fatigue. In the design of the L-1011, limit design gust loads were determined by application of
continuous turbulence criteria. Static discrete gust loads were also obtained, but were at no point critical.

Under a continuous turbulence input, any lightly damped mode will tend to resonate, reaching an amplitude, relative to the
input, perhaps many times that possible in response to a single pulse input of the same magnitude. For most transport airplanes, and
especially swept-wing configurations at cruise altitude, the Dutch roll mode is such a mode. In the absence of effective damping action
by means of either a yaw damper or the pilot, yaw oscillations of sufficient amplitude are developed so that the sideslip angle reached
can be several times that associated with the lateral gust velocity directly. Great potential exists, therefore, for utilization of a yaw
damper to reduce the lateral gust loads.

From the initial stages of design of the Lockheed L-1011 transport, provision was made for a yaw damper in order to assure
good handling qualities. Use of a yaw damper for this purpose is, of course, characteristic of large swept-wing aircraft generally. But
for the reason discussed above, a damper also results in a substantial reduction in the structural loads due to lateral gust encounter.
Early in the design of the L-1011, it was decided to reflect this loads reduction in the limit design loads.

Care was taken in the development of the damper to assure its effectiveness from a loads reduction as well as a handling
qualities standpoint. Initially a Dutch roll damping ratio of at least 0.30 was set as a design objective. It was found, however, that
substantially higher values, up to 0.50 or more, could be achieved over most of the climb, cruise, and descent flight regimes. It was
also found, contrary to earlier expectations, that these higher damping ratios provided a further reduction in loads. In addition, in
establishing the damper authority, consideration was given to the need for the damper to remain effective at fairly high turbulence
intensities.

AIRPLANE AND YAW DAMPER DESCRIPTION

The Lockheed L-1011 TriStar, in service since April 1972, is a wide-body three-engine turbo-fan commercial transport,
designed for short to medium range operation. It is of high-subsonic swept-wing design with two of the three engines pylon-mounted
beneath the wing and the third located internally in the extreme aft end of the fuselage. Normal passenger capacity is 256, with mixed
six and eight abreast seating. Wing span is 155 ft and design takeoff gross weight is 430,000 Ib.

The nominal area of the vertical tail is 550 square ft including the rudder. The rudder is hinged at 71 percent chord,
and is capable of 30 degrees travel in either direction; its area is 129 square ft.

The rudder is fully powered, driven by a pair of dual tandem hydraulic actuators, supplied by three hydraulic systems.
Of the four actuator units, two are supplied by one hydraulic system each, and two are supplied jointly by a third hydraulic system.
At speeds less than 164 knots EAS, all three systems are active. At speeds in the range 164 - 260 knots, only system " C " is active.
resulting in a reduction of available hinge moment to 41 percent of that available at low speeds. At speeds above 260 knots, the available
hinge moment is further reduced to 28 percent of the three-system low-speed value by means of pressure reduction through a pressure
regulator. Separately, a rudder deflection limit of 8 degrees is imposed when the flaps are retracted and the speed exceeds 164 knots.

A servo valve control module is located close to the actuators. It accepts mechanical inputs from the pilot and electrical
inputs from the SAS (stability augmentation system) and autopilot,from which it commands hydraulic flow to the actuators. The SAS
input is the source of the yaw damper action. (The terms SAS and yaw damper will be used interchangeably throughout this paper).
Pilot and SAS inputs act in series - that is, the rudder angle is the sum of the angles commanded by the pilot and the SAS. The autopilot
which affects rudder motion in the automatic landing mode, includes SAS inputs as well as the landing maneuver inputs.

The yaw damper action provides a rudder angle, 5 r , that is essentially proportional to and in phase with the yaw velocity, + •
The exact characteristic is shown in Figure 1. This characteristic applies to flight with flaps retracted; with flaps extended, higher gains
are used. The washout below the Dutch roll frequency is provided to improve the airplane handling under pilot control in turns. The
dropoff above the Dutch roll frequency is provided in order to minimize the effect of the SASonelastic mode response. It was found,
incidentally, that the (ateral gust loads were about 8 percent higher for the damper as defined by Figure 1 than for a damper defined
simply by 6 r = 1.30 t .

GUST LOADS CRITERIA

The continuous-turbulence gust loads criteria to which the L-1011 was designed were basically as developed in Reference 1 and
summarized in Reference 2. These criteria involve the combined use of mission analysis and design envelope approaches. For the L-1011,
the mission analysis loads were generally the more severe, and the primary effort, therefore, involved use of that criterion.

Under the mission analysis criterion, a set of typical flight profiles is first established. The gust structure is assumed to be
characterized by the Von Karman shape of power spectral density, with L = 2500 ft. The probability density of the rms gust velocity
is defined in a form such that frequency of exceedance of each load quantity is given by the expression.

N(y) " No[piexP(-£f) • P2 «*(-«£)] (1)

In this expression, y can be any load quantity — for example, bending moment at a particular wing station. N(y) is the number of
exceedances of y per unit time. A is the ratio of the rms value of y to the rms gust velocity, and N is a characteristic frequency of y,
obtained as the radius of gyration of the power spectral density of y about zero frequency. Both A and N Q are obtained by appropriate
dynamic analysis. The constants P., Po, b-t, and b2 are parameters defining the gust environment and are specified in Reference 1 as
functions of altitude. Exceedances are calculated by means of the above equation for each mission segment and added to give a total
for over-all operation of the airplane. For the vertical gust analysis, the above equation for N(y) is modified to include the contribution
of the one-g steady flight loads, so that the exceedances obtained are of net loads. Curves of N(y) vs y are obtained in this way for
shears, bending moments, and torsions at as many locations on the wing, fuselage, horizontal tail, and vertical tail as considered necessary.
Each curve is then entered at a limit design frequency of exceedance of 2 x 10'° cycles per hour to give the limit design value of the load.

leo
M O D U L U S 1 UVIPLITUDE
1_>ll R A T I O , 6,7") l | _

§0.4
_E
I V •(ASE ANGLE BY WHICH
/ h . LEADS+1

<
ul
VI
i 30

1?0
A API>ROX. DUTCH R O L L
•W NA T U R A L
l (il i l JENCY
150 .RUISE

180

FREQUENCY, CPS

VERTICAL TAIL SHEAR

FIGURE I. FREQUENCY RESPONSE CHARACTERISTICS
FIGURE 2. FREQUENCY OF EXCEEDANCE OF VERTICAL TAIL
OF L-1011 YAW DAMPER
SHEAR WITH AND WITHOUT YAW DAMPER
 Under the design envelope criterion, design is to the most critical points on the specified design envelope of speed, altitude,
gross weight, fuel weight, zero-fuel weight, and CG position. In this respect, the criterion is similar to the past discrete gust criteria, as
well as to limit design maneuver loads criteria. A quantity ° ' w r l d . analogous to a design gust velocity, is specified as a function of altitude.
In the expression <r w n . ( j , <r w isadesign rms gust intensity and I j is a factor representing the ratio of design load to rms load. The design
load at any point is then given by multiplying ^ ^ . H H by A :

(i
V.I r^d " (A<rw)ld = A
(<rw1d) <2)

In applying these criteria to obtain design loads for the L-1011, A and N Q values were obtained by means of digital computer
programs as described in Reference 2.

EFFECT OF YAW DAMPER - MISSION A N A L Y S I S CRITERION

Basic A p p r o a c h ; S A S R e l i a b i l i t y

A key concern in any attempt to reduce l i m i t design loads by accounting for the beneficial effects of a stability augmenta-
t i o n system is that the system will not be 100 percent reliable - i.e., it will not be in operation 100 percent of the time.

Under the mission analysis concept, the fraction of time that the system is inoperative can be accounted for easily and
quantitatively.

Typical frequency-of exceedance curves for a representative load quantity - fin shear at a location near the root — are
shown in Figure 2. These curves were obtained by means of Eq (1) and represent the total exceedances contributed by all segments of
all profiles. Curve A is obtained on the assumption of no yaw damper. The design load, under this assumption, would be as indicated
by F on the figure. Curve C reflects the load reduction due to the presence of the yaw damper, under the assumption that the yaw
damper is always operating. The design load w i t h a fully reliable yaw damper would be as indicated by G on the figure.

In the development of the L-1011, early estimates indicated that the yaw damper might be inoperative as much as 3 percent
of the time. The exceedances accumulated during that portion of the total flight time during which the SAS is not in operation are then
given by 0.03 of Curve A ; the result is Curve B, obtained by shifting Curve A down by a factor of 0.03. The exceedances accumulated
during the 97 percent of flight time during which the SAS is operating are then given by shifting Curve C down in the ratio 0.97 to give
Curve D. (Curves D and C differ imperceptibly and are shown by a single line.) The total exceedances are then given by the sum of
Curves B and D, or Curve E. This is the curve f r o m which the actual design load is obtained, indicated by H on the figure.

As the design of the airplane progressed, it became evident that the estimate of 3 percent for time inoperative was extremely
conservative. The yaw damper system as finally defined is a two-channel system; there is no degradation of performance and no loss of
authority upon single-channel failure. Further, to preclude the possibility that an airplane might be f l o w n for a protracted period w i t h a
totally inactive yaw damper, certification provisions require that at least one channel be operative for dispatch. Utilizing guaranteed
values of mean-time-until-removal of the various SAS elements, it was calculated that the average failure rate of the second channel,
conservatively assuming the first channel never to be operative, would be 0.00086 per flight hour. Actual service experience to April 1973
showed reliabilities of the individual components that differed considerably f r o m the specification values, but an over-all value that agreed
very closely. With t i m e , the reliability of the system should increase. On the basis of an average flight duration of 2 hours, w i t h the loss
of the SAS occurring — when it does - midway during the flight, it follows that the fraction of time SAS inoperative would be

.00086x1/2x2 = .086 percent

As noted, even this estimate is conservative in that it ignores the likelihood that on most flights both SAS channels will be operative on
takeoff. Normal maintenance practices of the airlines are such that, once a SAS channel becomes inoperative, the equipment will be
replaced within a reasonable time,ordinarily t w o days at the most. Thus the value of 3 percent for time inoperative, used in defining the
design loads, was found to be much higher than necessary. The effect on loads, however, of using a more realistic value — 0.1 percent or
less - would not be great. Even w i t h the fraction of time inoperative reduced to zero, the reduction in load would be only in the ratio
G/H in Figure 2, or 7 percent. An increase in percent of time inoperative above 3 percent, however, would have a much more potent
effect.

Saturation

A second concern is the saturation of the yaw damper at high load levels. The power-spectral analysis as ordinarily performed
requires an assumption that the dynamic system is completely linear. In contrast, the angle to which the rudder can be moved by the SAS
is subject to a well-defined l i m i t governed by the hinge moment that is available f r o m the hydraulic actuators. As a result, in turbulence
of limit-load severity, the yaw damper may be less effective than indicated by the linear analysis.

In order to evaluate the effect on lateral gust loads of a rudder angle l i m i t , time-history analyses were performed using an
electric analog computer. The airplane simulation utilized the three rigid-airplane degrees of freedom of sideslip, yaw, and r o l l , and
included the yaw damper. Provision was made for a rudder angle l i m i t , which could be set at any desired value. The lateral gust input
was provided by a white noise generator in conjunction w i t h a filter t o give approximately the desired shape of power spectral density.

Runs of 600 seconds duration were made, first w i t h no rudder angle l i m i t and then w i t h various rudder angle limits. The
important time history records were of rudder angle, 6 r , and side load on the vertical tail, P „ .

The flight condition represented was one for which the computer was already set up for handling-qualities studies. It was
similar, but not identical, to the condition that predominated in defining the design frequency of exceedance curves.

The intensity of the lateral gust input could be rather arbitrary, inasmuch as the only significant nonlinearity was the limiting
rudder angle, and the results w o u l d be interpreted on a non-dimensional basis. Nevertheless, it was desired to use a realistic gust intensity
in order to facilitate a complete understanding of the results. To determine an appropriate gust intensity, it was noted, first, that the
design loads read f r o m the exceedance curves, if expressed in the f o r m of Eq (2),

yd • ('w^d)*

required a °'VJr\A value of about 107. Here A is the value calculated for the predominant mission segment. Actually the ^ w n , j value thus
obtained differs slightly f r o m one load q u a n t i t y to another, so that 107 is an average value. Assuming a peak-to-rms ratio, ' i j . of 3.0,
f w = 107/3 = 36 fps. Values of approximately this value, therefore, were used.

It may be remarked that for a given value of * j ' v , r \ t j , the " m o s t l i k e l y " combination of r _ and i j can be determined by
appropriate integrations of the j o i n t probability density of the t w o statistically independent quantities, T W and n j . The probability
density of a w is available through its use in the derivation of Eq (1). Inasmuch as the second term in Eq (1) predominates, at the l i m i t
load level, this probability density is defined simply by means of the parameter h ^ The probability density of n,d> or y/cr is Gaussian.
It is f o u n d that the " m o s t l i k e l y " value of ' i d ' * a function only of o - w 1 d / b 2 - For the 28,000 f t altitude of the L-1011 predominant
mission segment, b2 is given by Reference 1 or Reference 2 as 11.8. For c r ^ d = 107 and b2 • 11.8, < r w r ' d / b 2 = 107/11.8 = 9 . 1 .
The resulting value of n ^ is 2.95. This is approximately the same as the value 3.0 assumed above. For other values of ^ w ^ d A ^ , values
of r i d are: " " w ^ d ^ = 5 , ^ d = 2 - 2 0 ' . I T w r i d / b 2 = 10 > ^ d = 3 - 1 ° ; , r w r | d / b 2 = 15, M J - 3.75. These values are consistent w i t h
values that are readily obtained f r o m data given in Reference 3, Figures 75 — 78 and Table 12. The existence of a " m o s t l i k e l y " combi-
nation of o- w and ^ d is plausible. A t high ^ w ' s , t h e given f y ^ d is unlikely to be reached because the airplane is seldom in turbulence
of that intensity. A t low cr w 's, requiring high I d ' s , the high ^ d w o u l d seldom be reached even if the airplane were in turbulence of the
given cr w all the t i m e .

A frequency-of-exceedance curve for rudder angle, obtained w i t h no rudder-angle l i m i t , is shown in Figure 3. This was
obtained from the 600-second time-history record. The horizontal coordinate was taken as 6 r 2 instead of 6 r in order that Rice's
equation for positive slope crossings at a given level (Reference 4 , page 39) w o u l d plot as a straight line. It is seen that the value of 6 r
expected to be reached once in the 600-second run is y/14.3 = 3.78 degrees.

Runs were then made w i t h rudder angle limits of 2.40, 1.95, 1.50 and 1.05 degrees. Exceedance curves of vertical tail load
were obtained for all five runs. Three of these are shown in Figure 4 . It is seen that the value of tail load expected to be reached once
in the 600-second run gradually increases as the available rudder angle is decreased.

The oncein-600-seconds value of Py is then plotted vs. limiting rudder angle in Figure 5. In the non-dimensional f o r m in
which the data are p l o t t e d , it is expected that the curve w o u l d not change significantly w i t h flight condition or even w i t h modest changes
to the yaw damper characteristic.

It might be remarked that entering Figures 3 and 4 at the once-in-600-seconds frequency of exceedance in obtaining the
curve of Figure 5 is arbitrary. A more rational approach for establishing this frequency of exceedance w o u l d involve use of the most

1
RUDDER ANGLE LIMIT
p NO LIMIT
100
y p 1.96 OEG
ft"

5 ?
o •
UJ S
y/r*
(/
DEG

i? 10
3 M
v3^^
k
v •v v^

1
2 4 6 8 10 1? M
5r2.DEG2
FIGURE 3. FREQUENCY OF EXCEEDANCE OF 5 , WITH FIGURE 4. FREQUENCY OF EXCEEDANCE OF VERTICAL TAIL LOAO

NO AUTHORITY LIMIT. CONSTANT O w WITH VARIOUS AUTHORITY LIMITS. CONSTANT O w

* ' H.M. LIMIT / ' MAX NO LIMIT

FIGURE 5. EFFECT OF RUDDER AUTHORITY LIMIT ON VERTICAL T A I L LOAD

 probable value of '_d obtained as discussed earlier in this section. Substitution of this value of I d , or yAr , into Rice's equation,

N(y) = N0exp|-1/2(-I-j2l ,

then gives the frequency of exceedance. It is found that a better value is about 3 exceedances in 600 seconds, instead of one. The effect
on the curve of Figure 5, however, has been found to be negligible.

The frequency of exceedance curves for 6 r obtained in the design loads analysis indicated a value at the limit-load frequency
of exceedance of 5.45 degrees. For the predominant mission segment, the rudder angle that can be reached w i t h the available hinge
moment is 3.0 degrees. (For other segments, the value generally is close to this.) Entering the curve of Figure 5 at a value on the hori-
zontal scale of 3.0/5.45 = 0.55, one finds the increase in tail loads due to saturation to be 5 percent.

All lateral gust loads were increased by this percentage. The percentage increase in loads elsewhere than on the vertical tail
should generally be less than this. For the vertical tail, the response is almost purely static. For other parts of the airplane, however - in
particular for the fuselage in side bending — elastic mode dynamic response is substantial. But the yaw damper is designed to have negli-
gible effect on the dynamic response at elastic-mode frequencies. With saturation increasing the static contribution 5 percent but not
affecting the dynamic response, the total load will increase by some amount between 0 and 5 percent. Thus the treatment of saturation
for loads other than on the vertical tail was slightly conservative.

In view of the care that was taken t o input a realistic limit-design lateral gust intensity, one might have expected closer
agreement between the maximum rudder angle recorded in the analog runs (3.78 degrees) and obtained from the exceedance curves
(5.45 degrees). Part of the difference may be due t o the use of a slightly different flight condition in the analog w o r k . The primary
source of the difference, however, appears to be an increase in the yaw damper gain reflected in the design frequency of exceedance
curves but not in the analog runs, which were made much earlier.

Load Reduction Achieved

From Figure 2, it is seen that a fully reliable linear yaw damper would have reduced the design loads in the ratio G/F • 0.65.
As a result of assuming the yaw damper to be inoperative 3 percent of the time, the reduction is seen to be less — H/F • 0.70. Increasing the
loads 5 percent to account for saturation of the yaw damper results in an actual reduction in the ratio (1.05)10.70) = 0.73.

FLIGHT LOADS MEASUREMENTS

As part of the L-1011 flight test program, an extensive set of flight and ground loads measurements was made. This program
is summarized in Reference 5. Included in the tests was flight through continuous turbulence, w i t h the gust velocities measured by means
of a probe mounted on a b o o m 23 ft ahead of the nose. Sixteen samples of turbulence were obtained, varying from 1 to 5 minutes in
duration. These samples totaled 5 minutes at an altitude of 30,000 f t , representative of normal cruise, and 3 0 minutes at about 5,000 f t .
Flight was included both w i t h and w i t h o u t SAS, and over a range of speeds and fuel weights.

It is of interest to note the load reduction obtained by use of the SAS and to compare this w i t h the theoretical indications.

First, as a matter of background, flight-measured and theoretical power-spectral densities of f i n shear are shown in Figure 6.
Only the frequency range f r o m 0 to 0.6 cps is included, inasmuch as the fin response at the higher frequencies, including elastic mode
resonances, is virtually nonexistent. The theoretical psd's were obtained in the usual way, multiplying the Van Karman gust spectrum by
the square of the modulus of the theoretical frequency-response function. The test psd's were obtained similarly, except that the frequency-
response function was determined from the test data by means of the cross-spectrum method (Reference 6).

The Dutch roll peak is seen to occur at about 0.2 cps. The vast difference in sharpness of peak, for the SAS-off cases, between
the test and theoretical curves is due largely to the smearing effect of the numerical procedures used in determining the test psd's. (The
smearing effect is roughly equivalent to that given by a weighting function having the shape of an isosceles triangle w i t h base equal to
0.20 cps.)

A values for f i n shear are compared in Figure 7. These were obtained by integrating the respective psd's.

The results shown in Figure 7 are seen to be similar for the two flight conditions. With SAS o n , the test and theoretical
values agree very well. (From the standpoint of structural adequacy of the L-1011, the small amount by which the test values exceed
the theoretical was more than offset by conservatisms in other aspects of the analysis.) With SAS off, both test and theory show a load
increase relative to the SAS-on value; but the test data show this increase to be substantially less than indicated by theory.

The substantial differencejpetween test and theory SAS-off is believed to be due, in large part, t o the tendency of the p i l o t
to act as a yaw_damper. Thus the test A SAS-off (pilot as yaw damper) is intermediate between the SAS-on A (test or theory) and the
no-SAS-at-all A (as indicated by theory). This view is supported by the measured coherencies (defined as in Reference 6) between rudder
angle and lateral gust velocity. These are shown in Figure 8. With SAS on, the coherency is generally fairly close to unity, especially in
the key frequency range f r o m 0.1 t o 0.4 cps. This is expected, inasmuch as the rudder angle is linearly related in a simple way t o 4",
which in turn is produced primarily (although not exclusively) by the lateral gust velocity. But even w i t h SAS off the coherency is seen
to be substantial. Thus it is seen that the rudder m o t i o n is, to a substantial degree, linearly related to the lateral gust input. The pilot
clearly is doing something directly in response to the gusts; and it is likely that when the frequencies are as low as 0.20 cps he can be a
fairly effective yaw damper. Further, rudder angle rms values SAS-off are found to be about 4 0 percent of the SAS-on values in the high
altitude cases and 60 percent in the low altitude cases. Thus a potential for substantial damping action is indicated.

In view of the apparent effectiveness of the pilot as a yaw damper, the question arises whether it might not be appropriate
to reflect this damping in the l i m i t design loads, in the event a yaw damper either is not provided or, if provided, is inadequate in either
reliability or saturation characteristics. Indeed, it would appear that the adequacy of some existing aircraft is due in part to the additional
damping provided by the pilot. Reference 1 , for example, notes that t w o of the existing successful airplanes analyzed would not meet the
gust loads criteria recommended, w i t h respect to lateral gust loads, unless such additional damping were considered. On the other hand, it
would appear d i f f i c u l t to establish quantitatively the damping increment that could safely be attributed to the pilot. The pilot's
 SAS OFF SASON SAS OFF SAS ON

THEORY—

K
CL THEORY- —
< ItSI *- TEST
0.2 0.4 0.2 0.4
FREQUENCY, CPS FREQUENCY, CPS TEST
——THEOR v
h - 5000 FT b. h • 30,000 FT
M - 0.53 M > 0.86 -—TEST
—THEORY

FIGURE 6. COMPARISON OF THEORETICAL WITH MEASURED

POWER-SPECTRAL DENSITIES OF VERTICAL TAIL SHEAR

.SASON
h (T
rA , M t OFF
u_
rbkVr v
P\
h - 5,000 FT h - 30,000 FT

7r
" ^ ^ M - 0.53 M - 0.86

\ FIGURE 7. EFFECT OF SAS ON A FOR VERTICAL TAIL

0
0.2 tsX 02 0.4 SHEAR - COMPARISON OF THEORY AND TEST
FREOUENCY, C t t FREQUENCY, CF»
b k • 30.000 FT
M - OH

FIGURE 8. COHERENCIES BETWEEN RUDDER ANGLE ANO LATERAL

GUST VELOCITY, SAS-ON AND SAS-OFF

effectiveness can be expected to vary from day to day and from pilot to pilot. His reliability and especially his saturation characteristics
would be difficult to establish. It would appear preferable by far, in view of the widespread use of yaw dampers in modern airplanes, to
assure that the damper is designed to be adequate from a loads standpoint. The design loads can then reflect not estimates but quantita-
tive, substantiable data.

EFFECT OF YAW DAMPER - DESIGN ENVELOPE CRITERION

Under the design envelope criterion, as under the mission analysis criterion, it is necessary to account quantitatively for the
fraction of time that the yaw damper may be inoperative.

Figure 9 shows the variation with altitude of the design gust velocity, <rw Hd, a t various levels of severity. Each curve of the
family represents a constant frequency of exceedance as measured by N(y)/N 0 . When using the design envelope criterion in combination
with the mission analysis criterion, the recommended level of severity corresponds to N(y)/N 0 = 1.2 x 10" 6 . The corresponding curve in
Figure 9 is one of the two indicated by heavy short-dash lines. At h = 7000 ft, the design o ^ - d is seen to be 62 fps.

To account for the presence of the yaw damper, separate o ^ T d values for the SAS-on and SAS-off cases are needed. These
are obtained by arbitrarily allocating the allowable exceedances between SAS-on and SAS-off operation. This involves selecting any
combination of N(y)/N 0 values SAS-on and SAS-off such as to satisfy the equation,

m SAS-off "7"W SAS-on

/N(y)\
basic value
(3)

L Fraction of time
SAS-off
Fraction of time
SAS-on

FIGURE 9. DESIGN VALUES OF O w i? d AT SPEED V c

The corresponding o ^ l d values SAS-on and SAS-off are then read from the curves of Figure 9 for the two N(y)/N 0 values selected. Any
combination of values satisfying Eq (3) will assure adequate loads; the particular combination of values selected, however, should be such
as to give the lowest loads. To find this combination will call for a trial-and-error, iterative, or similar approach.

For the L-1011, a potentially critical flight condition occurred at design cruise speed at h = 7000 ft. A values for vertical
tail shear at a particular fin station for this condition were

SAS-on 414 Ib/fps

SAS-off 640 Ib/fps

As noted earlier, the SAS was conservatively considered to be inoperative 3 percent of the time.

If the design N(y)/N 0 of 1.2 x 10' 6 were allocated equally to SAS-on and SAS-off operation, the two <r w nd values would
be obtained by use of Eq (3) as follows:

.60 x I D 6 + .60 x ID" 6 1.20x10-6

(.03) (20 x 10-6 ] + (.97) (.618 x 10-6 j 1.20X10 6

4-
( r w n d - 36 • < r w 1 d = 62
( r w T i d = 68

A ((rw1d) " 640x36 A (<rwnd) =414x68

= 23,0001b = 28,100 1b

(With fully reliable SAS, A ( < r w n d ) " 4 1 4 x 6 2 = 25,6001b)

On the second line, the factors .03 and .97 are the p and (1 - p) values. The quantities that these factors multiply are N(y )/N 0 values
such as to give the products appearing on the first line. The t r ^ d values on the third line were read from Figure 9 for these N(y)/N„
values.

The design load would be the higher of the two A (<r w nd) values, 23,000 Ib and 28,100 Ib. It is seen that, for SAS-off
operation, the o- w 1d value of 36 fps is much lower than the basic design value of 62 fps. But the a w n d value for SAS-on operation
must be somewhat higher than 62 fps, in order to "make room" for the exceedances contributed by SAS-off operation.

The equal allocation of exceedances between SAS-on and SAS-off operation is seen to result in different design loads for
the two cases. To equalize these two loads and thus minimize the higher of the two, the allocation will be changed. Noting that the
design load will be less than 28,000 Ib, but greater than 25,600 Ib, say 27,000 Ib, <r w HH for SAS-off can be increased to about
(27,000/23,000) 36 = 42. The corresponding N(y)/N 0 , from Figure 9, is about 12 x 10*. The allocation then becomes,

•36x10-6 + .84x10-6 = 1.20x10"6

(.03) ( l 2 x 10-6 j + (.97) (.87x10-6) = 1.20 x l O " 6

+ + +
c r w n d = 41.5 o - w 1 d = 64.5 a w n d = 62

A ("Tw^d) = 640x41.5 A(o- w r|d) = 414x64.5

= 26,6001b = 26,7001b

The design load is thus found to be 26,700 Ib, about 4 percent higher than the 25,600 Ib value that would be obtained with a 100 percent
reliable SAS.

The reduction in load achieved by use of the yaw damper is in the ratio (1.051(41.5)/(62) = .70, where the 1.05 factor
accounts for the load increase due to SAS saturation. This .70 factor is comparable to the .73 factor attained under the mission analysis
criterion.

CONCLUDING REMARKS

From the experience gained in the design and flight testing ofthe L-1011, it is evident that:

1, For an airplane of this type, the presence of a yaw damper can substantially reduce the lateral-gust-induced loads that the airplane
experiences in flight.

2. This reduction can be reflected, rationally and safely, in the limit loads to which the structure is designed.
10

REFERENCES

1. Hoblit, F.M., Paul, N., Shelton, J.D., and Ashford, F.E., "Development of a Power-Spectral Gust Design Procedure for Civil A i r c r a f t , "
FAA-ADS-53, Washington, D.C., Federal Aviation Agency, Jan. 1966.

2. Stauffer, W.A., and Hoblit, F.M., " D y n a m i c Gust, Landing, and Taxi Loads Determination in Design of the L - 1 0 1 1 , " Journal of
Aircraft, V o l . 10, No. 8, Aug. 1973.

3. Fuller, J.R., Richmond, L.D., Larkins, C D . , and Russell, S.W., "Contributions to the Development of a Power-Spectral Gust Design
Procedure for Civil A i r c r a f t , " FAA-ADS-54, Washington, D.C., Federal Aviation Agency, Jan. 1966.

4. Houbolt, J.C., Steiner, R., and Pratt, K.G., " D y n a m i c Response of Airplanes to Atmospheric Turbulence Including Flight Data on
Input and Response," TR R-199, Washington, D.C., N A S A , June 1964.

5. Stauffer, W.A., Lewolt, J.G., and Hoblit, F.M., " A p p l i c a t i o n of Advanced Methods to Design Loads Determination for the L-1011
Transport," Journal of Aircraft, V o l . 10, No. 8, Aug. 1973.

6. Coleman, T.L., Press, H., and Meadows, M.T., " A n Evaluation of Effects of Flexibility on Wing Strains in Rough Air for a large
Swept-Wing Airplane by Means of Experimentally Determined Frequency-Response Functions w i t h an Assessment of Random-Process
Techniques E m p l o y e d , " T R R-70, Washington, D.C., N A S A , 1960.
 II

THE EFFECT OF ACTIVE CONTROL SYSTEMS ON

STRUCTURAL DESIGN CRITERIA

by

N.F.Harpur

British Aircraft Corporation Ltd

Commercial Aircraft Division, P.O. Box 77
Filton, Bristol BS99 7AR, UK
12
 13

THE EFFECT OF ACTIVE CONTROL SYSTEMS ON

STRUCTURAL DESIGN CRITERIA

N.F.Harpur

1. INTRODUCTION

Interest in the use of active control systems is increasing and they are being considered not only to improve
stability but also structural efficiency. This greater interest is in a large measure due to the improved reliability of
the necessary electronic systems and also to their considerably reduced weight and bulk. This makes it possible
nowadays to envisage quite complex control system requirements and to ensure very high integrity (by duplication,
triplication, etc.) without using unduly large equipment.

Active control systems, for the purpose at least of this paper, may be defined as systems which operate any or
all of the controls on the aircraft (including flying controls, flaps, spoilers, throttles, etc.) to some degree independently
of the action of the pilot.

They can be used for

(a) improving aircraft stability thus enabling the size of the fixed and moveable control surfaces to be reduced,
(b) reducing the static design loads (manoeuvres, gusts, etc.),
(c) reducing the fatigue design loads,
(d) increasing flutter speeds thus reducing the additional stiffness or mass which might otherwise be required,
(e) reducing the margin required between normal operating speed and design diving speed due to upsets, entry
into wind shears, temperature gradients, etc.,
(0 improving the ride comfort and reducing vibration effects on stores, for example.

It is not intended to discuss here the structural weight savings which will obviously accrue from (a) above, nor
the criteria which will determine whether (0 is worthwhile since these must be closely associated with the role of the
aircraft.

However (b) to (e) can all significantly affect the structural design criteria. The extent to which weight can be
saved depends on the particular aircraft but it is probable that the weight savings will be limited unless a number of
the objectives (b) to (e) can be achieved simultaneously. There is little point in reducing structural scantlings in
static design cases unless the fatigue loads can also be reduced and the resulting more flexible structure made flutter
free.

The active control system is therefore likely to be quite complex and consideration must be given to the
consequences of failures in the system when deciding by just how much the structure weight may be safely reduced.

Each of the objectives (b) to (e) will be considered separately. The discussion is confined to fixed wing aircraft
since it appears that active control systems in helicopters will be largely limited to objective (a) and (0 above. The
paper refers mainly to civil aircraft and their airworthiness requirements, chiefly because of the author's greater
familiarity with this field, but the principles are equally applicable to military aircraft.

2. REDUCTION OF STATIC DESIGN LOADS

2.1 Manoeuvre Loads

Autostabilisers have been in use for many years but their prime objective has been to improve the stability or
control of the aircraft and normally the lifting surfaces of the aircraft have been sized on the basis that a minimum
degree of controlability is retained in the event of autostabiliser failure. As far as the structure was concerned the
main concern was to ensure safety in the event of system malfunction (e.g. a runaway) and the structural design
criteria were based on the most severe cases arising from either operative, inoperative or malfunctioning autostabilisers.
Such devices are now being envisaged as means of directly reducing structural loads.
 Autostabilisers operate by sensing a rate of rotation signal (yaw, roll, pitch) and moving the appropriate
control surface. A typical example is the yaw autostabiliser or damper. Figure 1 shows the variation of fin load
with altitude, with yaw autostabiliser both operative and also inoperative, on the BAC/AS Concorde. The fin load
plotted is that arising from the specified pilot induced manoeuvre at design cruising speed Vp i.e. application of
maximum rudder except as limited by pedal force, servo-control power e t c . 1 , 2 .

The variation with altitude is complex due to the considerable variation of VQ with altitude (and hence Mach
number) on this aircraft. Figure 2 shows the side-slip, rudder angle and consequent load variation with time.

It will be seen that the effect of the autostabiliser is to reduce the peak loads by about 20%. This was achieved
even though the autostabiliser had limited authority (±3°), since it was designed primarily to meet stability and
control requirements.

No reduction is seen at altitudes above 30.000 ft because, at this altitude and above, the rudder servo-control
jacks become saturated during the manoeuvre and the autostabiliser is obviously ineffective.

An additional requirement which has to be considered is one where the pilot induces yaw by a sinusoidal
operation of the rudder control at a frequency such that the maximum lateral response of the aircraft is induced.
Usually only one or two complete cycles are required to be considered and the peak fin loads do not. of course,
occur in phase with the achievement of peak rudder angle. In these cases the autostabiliser may well prove
effective in reducing loads even if the demanded rudder angle is so large that the servo-control jacks will stall at
periods during the cycle. Figure 3 shows significant reduction in fin load on the Concorde with autostabiliser
operative even when the peak demand rudder angle is in excess of the sum of the stall angle plus autostabiliser
authority (about 20°).

To realise such load reductions requires that the airworthiness authorities accept that a reduced severity loading
case maj be assumed when the autostabiliser is inoperative: otherwise the inoperative case would always produce the
critical design loads. The concept of reducing safety factors or the severity of the loading cases for combinations of
(light conditions which can be shown to be rare, has been accepted in principle for some time. A typical example
is the reduction of gust and manoeuvre load levels prescribed for flight with landing flaps extended where (except
in cases where these flaps are used en-route as well as on take-off or landing) the design gust intensity for example
is reduced from 50 ft/sec to 25 ft/sec as a crude basis of assessing the combined probability of gust encounter and
flap position 3 . It is obviously a relatively small step to include, under aircraft condition, the probability of auto-
stabiliser failure.

Perhaps the closest analogy lies in the requirements for fail-safe evaluation of structures 4 . Here it is accepted
that safety will be ensured if, after a failure or partial failure of a single principal structural element, the structure
remains capable of carrying a reduced load, typically between 53% and 67% of the design ultimate load. Again this
is a rough and ready way of assessing the combined probability of the structural failure and of the applied load
condition.

11 would seem reasonable therefore to associate the autostabiliser inoperative conditions with a reduced severity
loading case. A simple approach would be to take autostabiliser failure as being equivalent to the structural failure
in ;i fail-safe evaluation and require, say, an ultimate factor of 1.0 on limit loads i.e. reducing the strength requirement
to 67%. However this would imply the assumption that the frequency of autostabiliser failures was roughly the same
as that of structural failures.

Nowadays it is possible to assess system reliability with a fair degree of confidence and a more quantitative
approach might be considered.

Assessment of the overall probability of the event occurring, i.e. the combination of the loading case and the
system failure, requires a knowledge of the probability of the loading case. Acceptance of a figure for the overall
probability of structural failure, taking all loading cases and parts of the structure into consideration, would be
required before an acceptable probability could be established for each such event. It would involve replacing all
the existing structural safety criteria with a single statement such as "the probability of structural failure from any
cause shall be less than 10 l 2 per hour". Such a statement, while quite logical, is not very practical when it comes
to demonstration of compliance, although it must be admitted that some requirements are already heading part way
down this road e.g. the power spectral gust criteria, referred to later.

An alternative is simply to consider the probability of the system failure and permit a reduction in ultimate
strength as a function of this probability (i.e. analogous to the fail-safe rules) taking no quantitative account of the
probability of the design case occurring. A compromise is suggested here by the author, which is to take some note
of the greater proportion of flight time spent within the normal speed envelope i.e. up to design cruising speed
V ( -/M(; und the very much smaller lime spent outside this envelope i.e. up to design diving speed Vp/Mp . The
suggestion is that the loads be calculated assuming the device has failed and then these loads be reduced according
to the failure probability of the device and the design speed associated with the loading case as indicated in Figure 4.
This implies that:
 15

(a) if the device has a failure probability of 10~3/hr or worse, the full strength must be achieved for the
failed case at both V^ and V D .
(b) if the device has a failure probability of 10" 7 /hr, the full strength must still be achieved at V c but this
may be reduced to 67% of full strength at V R .
(c) if the device has a failure probability of 10~"/hr or better, the strength to be achieved is 67% of the loads
d -rived from the device failed case at both V c and V D .

The ado tion generally of such an approach would leave the way clear to utilising systems of proven reliability
to reduce stru 'ural loads by up to ^. For example in the case illustrated in Figure 1 the design loads (which occur
at Vf;) would be
autostabiliser operative 62,000x1.5 = 93,000 1b
autostabiliser inoperative - 78.000x1.0 = 78.0001b
(failure rate 10~" per hour)
autostabiliser inoperative 78.000x1.5 = 117.000 1b
(failure rate 10" 7 per hour)

Thus to achieve the maximum load reduction of 21% (i.e.— 1

V 117,000 1b/

The autostabiliser inoperative ultimate factor should be less than 1.19 and using Figure 4 this would require a total
system failure probabilities factor less than about 3 x 10" 10 per hour of flight. With duplicated systems this is quite
practicable.

Of course the reliability computation must take into account any possible operational considerations such as
the desire to despatch the aircraft with part of the system failed. If it were intended that a duplicated system be
regarded as a "go" item even in the event of failure of one channel, then the remaining channel must itself show a
reliability close to that derived by the above method.

Manoeuvre loads can also be reduced by restricting the control movements of the pilot to the maximum which
he needs to meet the handling requirements of the aircraft. This can be most conveniently done by means of an
artificial feel system but this cannot truly be described as an active control system. However it is often difficult to
restrict manoeuvres this way, particularly when these involve simultaneous use of more than one control e.g. in a
rolling pull-out. Strike aircraft require a high rate of roll combined with high " g " but the resulting rolling power at
low "g" can be excessive. Figure 5 indicates, for a modern strike aircraft, the variation of rolling power with "g"
at high equivalent air speeds. Without use of active control systems, the low " g " rate of roll is over 8 times that
at high "g". An active control system is installed which boosts the high "g" rate of roll, i.e. where it is needed,
while limiting the rate of roll at low "g", thus cutting down the design loads, in this case mainly wing tank pressures,
wing distributions and wing store loads. The controls are active in that, when full lateral stick is used, the rate of
roll is sensed and the feedback system tries to maintain a constant rate of roll regardless of "g" i.e. a roll rate
command system.

The extent to which the load alleviation at low "g" can be exploited depends on the reliability of the system.
If the previously described criterion is adopted (see Figure 4) the maximum alleviation possible achieved with failure
probabilities better than 10""/hr of flight, would be a \ reduction of load in design cases related to \ Q .

2.2 Engine Failure Loads

Fin and rudder loads arising from engine failure cases may also be dealt with in the same way. Figure 6 shows
the fin loads arising from single engine failures on the Concorde. In this case, in addition to the autostabiliser, an
alleviation is obtained at Mach numbers greater than 1.4 by the use of an autorudder which applies rudder angle in
response to a signal from a transducer sensing lateral acceleration. This figure shows also the adverse effect of
corrective rudder applied by the pilot at or about peak sideslip. In this example the failure rates of the autostabiliser
and autorudder systems should be combined with the chance of adverse pilot action, and also with the chance of
an engine failure itself, to arrive at the overall probability of the event. However a conservative approach might be
to ignore any adverse pilot corrective action, off-setting it against the low probability of engine failure, and to use
just the combined system failure rates to derive a reduced ultimate factor based, say, on Figure 4. A critical design
condition is that of 50,000 ft, associated with the higher structural temperature at Mach 2, and therefore the auto-
stabiliser operative case becomes the design case with an ultimate load at Vfg of 37,500 x 1.5 = 56,250 lb. For
the inoperative case to give no greater loads it must have an ultimate factor less than 56,250 lb/41,500 lb i.e. 1.35.

Based on Figure 4 this would mean that the combined system failure rate must be better than 6.5 x 10" 9 per
hour of flight.

The autorudder system was installed primarily to limit yaw angles within the range acceptable for engine intake
flow considerations and the reduction in structural loads might therefore be greater if the system were optimised for
this objective as well.
16

2.3 Engine Surge Loads

The loads arising due to engine surge, or hammershock, can give critical design conditions for supersonic intakes
and, to a lesser extent, for subsonic intakes. The surges can arise due to malfunctions of the engine or air-intake
control systems, both of which can be complex in supersonic aircraft. They may also arise due to external perturba-
tions, such as gusts and manoeuvres, which may affect the flow conditions at the engine face such as to induce
surge. While the design of the intake and engine control systems will be such as to minimise the frequency of such
surges, it is nevertheless necessary to design the intake structure and systems to meet the surge loads.

The peak surge pressure is a function of both airspeed and engine conditions. An example is given in Figure 7,
which shows the peak engine-face surge pressures for the Concorde at Mach 2 as a function of calibrated airspeed
(CAS) and low pressure compressor speed (N,). There is little variation of the surge pressures with Mach number
which affects primarily the probability of surge occurring. The intake strength is sized on the loads occurring at
VQ , 530 knots at cruise altitudes. In order both to reduce the chances of surges occurring and also to limit their
peak pressures a device has been installed which reduces the engine compressor speed in the event of the airspeed
increasing beyond Vfj towards Vp . The speed increments above V^ , due to upset manoeuvres, sudden
temperature changes or wind shears, are themselves reduced by the consequent reduction in engine thrust.

Failure of the device would result in loads at V D (565 knots) some 14% higher than those at V c and an
ultimate factor of not more than 1.30 would be required to ensure that the failed case did not design. Again using
the criterion of Section 2.1 (Fig.4) the maximum failure probability permissible is 2.5 x 10~5/hr. A system reliability
of this order is practical.

2.4 Gust Loads

2.4.1 Discrete Gusts

Possibly the first attempt to alleviate the loads arising in discrete gust encounters was made in the design of the
Bristol Brabazon long-range transport which first flew in 1949. This aircraft, with a predicted all-up weight of
330,000 lb in the production version, was obviously structure weight sensitive. A gust alleviation system was therefore
designed to reduce wing bending loads by operating the ailerons symmetrically in response to signals from a gust vane
mounted on the nose of the aircraft. Such faith was placed in the practicability of this system, that the wing structure
of the prototype was some 20% weaker than would have been required to meet the design gust cases without gust
alleviation. The gust alleviation system was not proved in flight before the aircraft was scrapped in 1953, although
the response of the vane to turbulence had been measured. A complete system had, however, been flown in an
Avro Lancaster5 in 1952 and some problems had arisen. The main one was that the large pitching moment
contributed by the ailerons resulted in a considerable loss of stability such that at large gust gradient distances the
alleviator effectiveness decreased.

Furthermore it is obvious that, in the limited flying time available (57 hours), it was not possible to find
really severe turbulence although sufficient had been found to indicate non-linearity in alleviation with gust magnitude.

This illustrates a major problem with the use of an active control system for alleviating the large static design
gusts i.e. the difficulty of demonstrating that it works. Assuming that a device can be engineered to cope with the
very high response rates required to cope with the large limit gust velocities (e.g. 50 ft/sec), the behaviour of the
whole system, and in particular that of the power operated controls, should be checked under these limit gust
conditions and these, by definition, will be almost impossible to find. Assumption of extrapolation from lower
level gusts may be difficult to justify.

The failure cases may be dealt with as indicated in Section 2.1 and the criteria suggested in Figure 4 would be
equally appropriate to the gust cases.

2.4.2 Continuous Turbulence

Power spectral gust criteria have been under discussion between the FAA, the European civil airworthiness
authorities and the aircraft manufacturers for some time. A proposal for a standard was put forward by FAA for
the SST and is, in fact, a special condition on Concorde*. In the "mission analysis" approach the overall frequency
of exceedance of load levels is determined taking into account the combinations of flight conditions (e.g. speed,
altitude, payload, etc.) with turbulence intensities. The strength level is set by that load which will have a frequency
of exceedance of 2 x 10"s per hour. Active control systems are specifically referred to, viz:-
"If a stability augmentation system is utilized to reduce the gust loads, consideration must be given to the
fraction of flight time that the system may be inoperative."

With the order of system failure probabilities which can nowadays be assumed e.g. IxlO" 7 per hour or better,
it is clear that consideration of system failures will have virtually no effect on the strengths required to meet the
power spectral gust criteria.
 17

This is simply because the probability of system failure is being combined with the probability of requiring the
system to operate (i.e. of encountering the maximum load level). This is a much less severe criteria than that
considered in previous paragraphs where a certain minimum strength level has been envisaged as a function of system
failure probability alone, regardless of how infrequent the critical design loading conditions may be (except for some
consideration of the low probability of being at Vp, for example).

The whole idea of the "mission analysis" while certainly being logically based, is not consistent with the past
practice of defining a series of static strength cases which must be satisfied regardless of the probability of their
occurrence. For example previous rules required the large discrete gust of 50 ft/sec velocity to be withstood over
the whole flight envelope bounded by the chosen design cruising speed V c at all practical combinations of weight,
altitude and centre of gravity. The critical combination may only occur for a very short period on each flight but
nevertheless the required strength must be provided. It is for such reasons, and because of doubts about the
validity of the linearising assumptions necessary, that the European Airworthiness authorities have so far declined
to accept the power spectral approach, as a replacement of the discrete gust, for static strength cases. If and when
this is done, it would be very easy to incorporate the effects of active controls and their failure probabilities.

3. REDUCTION OF FATIGUE DESIGN LOADS

In contrast to the remarks made in the pevious section, it is certainly true that the power spectral approach may
be used with confidence for assessing the affect of active control systems on the lower levels of gust and manoeuvre
loads i.e. those of greatest significance in causing fatigue defects. Furthermore it is quite simple to carry out the
necessary flight demonstrations. The effects of system failures may be taken into account on the same basis as
indicated in Section 2.4.2, and are likely to be insignificant with the level of system reliabilities now possible.

4. IMPROVEMENT OF FLUTTER CHARACTERISTICS

In the form of autostabilisers, active control systems have been involved in flutter evaluations for some time.
It is rare that they have ever been modified, let alone designed, with the object of improving the flutter behaviour.
The main reason is that the operating frequency range of the autostabiliser is usually well below the critical flutter
frequencies so that there was very little coupling.

However if active control systems are used to improve flutter behaviour the effect of a system failure must be
considered.

This is already covered in the FAR standards7 which say

"It must be shown by analysis or tests, that the airplane is free from such flutter or divergence that would
preclude safe flight, at any speed up to VD , after any other reasonably probable single failure,
malfunction, or adverse condition affecting flutter or divergence."

This is a reduction in standard from the non-failed case where flutter freedom must be shown in principle up
to 1.20 VD .

This rule can be used for active control system failures. It is suggested that a system failure probability not
worse than !CTs/hr is required so that the chance of a failure may be deemed no more than "reasonably probable".

5. REDUCTION OF SPEED MARGINS

A device can be envisaged which will sense increases above cruising speed, V^ , and operate the flying or engine
controls in such a way as to return the aircraft to V^ . The autopilot normally does this via the flying controls.
Autothrottles are often used to aid control on the approach and landing phase, where the aircraft is operating below
the minimum drag speed, and it is not very difficult to contemplate using them on other flight regimes to control
airspeed. A more sophisticated device has already been described under engine surge loads in Section 2.3. Since the
critical static strength cases are often associated with design diving speed, Vp , and this is also critical for flutter,
anything which can be done to reduce VQ in relation to a given cruise speed requirement will have all-round benefits.

In the case of flutter described in Section 4, a system failure probabilities factor of better than 10_5/hr of flight
was suggested as adequate. The same order of reliability should be adequate for the speed control device provided it
is only called upon to operate on rare occasions. However, the more closely the device is called upon to control
speed, the more often it will operate and then lower system failure rates will be required.
18

6 CONCLUDING REMARKS

This paper has suggested a number of ways in which the structural design criteria may be related to the design
of active control systems. It seems desirable to look at all the varieties of active control systems in a consistent way
and one such way has been suggested in this paper; it is not the only way and not necessarily the best. Airworthiness
regulations, both civil and military, should be based on agreed policy, if the full benefits of active control systems are
to be realised.

ACKNOWLEDGEM ENT

The author is grateful for the assistance received from his colleagues in the UK but would emphasise that the
views expressed in this paper are his own.

REFERENCES

1. TSS Standards for Concorde. Certification Authorities of Great Britain and France.
(TSS 4-3 Section 3.5.)

2. Federal Aviation Regulations, Part 25. Airworthiness Standards. Transport Category

Airplanes, Federal Aviation Administration. (Section 25.351(a).)

3. FAR Part 25. (Section 25.345.)

4. FAR Part 25. (Section 25.571(c).)

5. Zbrozek, J. Preliminary Report on a Gust Alleviator Investigation on a Lancaster Aircraft.
et al. Reports and Memoranda No.2972, August 1953.

6. Tentative Airworthiness Standards for Supersonic Transports. Revision 7 (January

1971), Federal Aviation Administration (Section 25.305(e) and Appendix G).

7. FAR Part 25. (Section 25.629(d).)

 19

ALTITUDE ( f e e t )

60,000

50,000

RUDDER JACKS STALLED AT

HIGHER A L T I T U D E S

AUTOSTABILISER
INOPERATIVE

SEA L E V E L
0 50,000 100,000

L I M I T FIN L O A D ( l b )
( D U E TO SIDESLIP O N L Y )

Fig.l Fin loads from pilot induced yawing manoeuvres

r-RUDDER DEMAND A N G L E
RUDDER ANGLE(deqs)
10

AUTOSTABILISER AUTHORITY

1 2
SIDESLIP ANGLE(deqS.) T I M E ( s e c s )
10

FIN L O A D (lb.) T I M E (sees.)

+ 50,000 AUTOSTABILISER OPERATIVE •
(DUE TO SIDESLIP AUTOSTABILISER INOPERATIVE
AND RUDDER
DEFLECTION)

-50,000

Fig.2 Responses during pilot induced yawing manoeuvres

20

RUDDER JACKS STALLED

L I M I T F I N LOAD (lb.)

100,000

50,00 0
ISER OPERATIVE

I 10 20 30

PEAK RUDDER DEMAND ANGLE ( deqs.)

Fig.3 Fin loads from sinusoidal rudder application

PERCENTAGE OF
ULTIMATE STRENGTH
TO BE ACHIEVED
'V-l 100

50

v°
SYSTEM FAILURE PROBABILITY

PER HOUR OF FLIGHT

Fig.4 Relationship between system failure probability and choice of ultimate factor
 21

RATE O F R O L L / C O N T R O L SURFACE ROTATION

1 1 T~
LOADS ALLEVIATED

ACTIVE C O N T R O L , O F F _ j r — ,.

VERTICAL ACCELERATION ( q )

Fig.S Strike aircraft rolling power with and without active control

AUTOSTABILISER INOPERATIVE
WITHOUT PILOT CORRECTIVE ACTION

ALTITUDE (feet)

60,000

AUTOSTABILISER OPERATIVE -
PLUS PILOT CORRECTIVE ACTION

\AUTOSTABILISER INOPERATIVE-
iPLUS PILOT CORRECTIVE ACTION

20,000 40,000 60,000

LIMIT FIN LOAD ( l b )

Fig.6 Fin loads due to single engine failures

22

ENGINE FACE PRESSURE ( p.S.i.Q. )

24|-

23

22

21

20

19

18

17

Fig.7 Variation of engine face surge pressure at M = 2.0

 23

STATUS OF TWO STUDIES ON ACTIVE CONTROL OF

AEROELASTIC RESPONSE AT NASA LANGLEY

RESEARCH CENTER

by

Irving Abel and M.C.Sandford

NASA Langley Research Center

Hampton, Virginia, USA
24
 25

STATUS OF TWO STUDIES ON ACTIVE CONTROL

OF AEROELASTIC RESPONSE AT NASA LANGLEY RESEARCH CENTER

Irving Abel and Maynard C. Sandford

SUMMARY

The application of active control technology to the suppression of flutter has been
successfully demonstrated during two recent studies in the Langley transonic dynamics
tunnel. The first study involved the implementation of an aerodynamic-energy c r i t e r i o n ,
using both leading- and trailing-edge controls, to s u p p r e s s flutter of a simplified delta-
wing model. Use of this technique has resulted in an i n c r e a s e in the flutter dynamic p r e s -
sure of approximately 12 percent for this model at a Mach number of 0.9. Analytical
methods used to predict the open- nnH Hosed-loop behavior of the model a r e also d i s -
cussed. The second study, which is a joint effort with the Air F o r c e Flight Dynamics
Laboratory, was conducted to establish the effect of active flutter suppression on a
model of the Boeing B-52 Control Configured Vehicle (CCV). Some preliminary r e s u l t s
of this study indicate significant improvements in the damping associated with the c r i t i -
cal flutter mode.

INTRODUCTION

Considerable interest has emerged over the last few y e a r s toward applying active
control technology to s u p p r e s s aeroelastic response of present and future aircraft con-
figurations. Potential gains in aerodynamic efficiency and weight savings can be realized
through ride-quality control, reduction of gust and maneuver loads with a consequent
reduction in fatigue damage, reduction of static-stability r e q u i r e m e n t s , and suppression
of flutter. The use of active controls to s u p p r e s s aeroelastic response is not new. It
has already been used, to a limited degree, on such airplanes as the XB-70 and B-52 to
improve ride quality by reducing structural response to turbulence (refs. 1 and 2).

The use of active controls to s u p p r e s s flutter s e e m s further from realization than

other active control concepts, partly because of the lack of a thorough experimental eval-
uation. A review of the recent literature (e.g., refs. 3, 4, and 5) indicates that most of
the work in this area is analytical. In an effort to fill the need for experimental r e s u l t s ,
wind-tunnel model p r o g r a m s a r e underway in the Langley transonic dynamics tunnel to
demonstrate the effectiveness of using active controls for flutter suppression.
The purpose of this paper is to describe some recent activities at Langley Research
Center toward evaluating the use of active controls for flutter suppression. (Some early
p r o g r e s s on these p r o g r a m s was reported in ref. 6.) The application of active controls
to the suppression of flutter at transonic speeds on a simplified delta-wing model is
described. Included is a brief s u m m a r y of the analytical aspects of the problem, a
description of the model, measured and calculated flutter points with and without active
controls, and some experimental techniques used to establish the behavior of the model
at subcritical test conditions.
26

In addition to the delta-wing p r o g r a m , a program is being conducted to evaluate the

use of active controls for flutter suppression, maneuver-load control, ride-quality control,
and reduction of static-stability r e q u i r e m e n t s on a model of the Boeing B-52 CCV airplane.
This p r o g r a m is a cooperative effort by Langley R e s e a r c h Center; the Air F o r c e Flight
Dynamics Laboratory; and The Boeing Company, Wichita Division, under NASA and Air
F o r c e c o n t r a c t s . These studies a r e being conducted in conjunction with a flight r e s e a r c h
p r o g r a m (ref. 7) in an effort to c o r r e l a t e model and flight data. This paper includes a
brief s u m m a r y of preliminary wind-tunnel f l u t t e r - s u p p r e s s i o n studies accomplished to
date.

SYMBOLS

b reference semichord

c reference chord, 2b

f. ,f frequencies at half-power points of forced response

g s t r u c t u r a l damping coefficient

h(x,y,t) vertical displacement

h. ,h„ v e r t i c a l displacement of delta-wing model at 30 and 70 percent of the

reference chord, respectively (fig. 5)

Mi generalized m a s s of ith vibration mode

m(x,y) m a s s distribution

Ap(x,y,t) p r e s s u r e distribution
qj generalized displacement of ith vibration mode

S reference area

s Laplace o p e r a t o r

t time

U aerodynamic-energy m a t r i x defined in equation (2)

U complex conjugate of the matrix U

V f r e e - s t r e a m velocity

x,y s t r e a m w i s e and spanwise coordinates, respectively

Z (x.y) normalized deflection in ith vibration mode

n angle of attack at section A-A (fig. 5)

fi leading-edge-control deflection
 27

6 trailing-edge-control deflection

6 B-52 outboard-aileron deflection

a
6 deflection command signal to B-52 outboard aileron
a,c
6, B-52 flaperon deflection

6t deflection command signal to delta-wing trailing-edge control

t ,c

a. circular frequency

Matrix notation:

( y column matrix

row matrix

square matrix

iT
IT t r a n s p o s e of matrix

Dots above symbols indicate derivatives with respect to time.

FLUTTER SUPPRESSION BASED ON AERODYNAMIC

ENERGY CONSIDERATIONS

Flutter is a self-excited oscillation in which energy is absorbed by the lifting s u r -

face from the a i r s t r e a m . The state of stability of the system is defined by the sign of
the work per cycle when the lifting surface undergoes an a r b i t r a r y oscillatory motion.
The use of energy techniques to investigate the stability of an aeroelastic system is not
new (ref. 8); however, a recent contribution to the area of flutter suppression is the devel-
opment of an aerodynamic energy criterion by Nissim (ref. 3). This criterion states that
a necessary and sufficient condition for the prevention of flutter is that for all allowable
oscillatory motions of an elastic system in an a i r s t r e a m , positive work must be done by
the system on the surrounding a i r s t r e a m . A brief summary of the salient points brought
out in reference 3 is given in the following section.

Energy Concept
Consider the equations of motion for a system with n degrees of freedom:

(F} = •«2 [M] * »pb4s[JAR] + { A A q ) + [Kj{q> (1)

where, at flutter, the generalized force { F ) = 0 and w is the circular frequency of

oscillation; [ M J is the m a s s matrix (called B in ref. 3); [A^l and TAJI a r e the
real and imaginary unsteady aerodynamic-force m a t r i c e s , respectively; | K] is the
structural stiffness matrix (called E in ref. 3); p is the fluid density; S and b
a r e a reference a r e a and length, respectively; and {q\ is the generalized displace-
ment vector.
28

Nissim shows that the work per cycle W (called P in ref. 3) done by the system
on the airstream can be written as

W - i ^pb 4 Sa, 2 |q R - iqi|[u][q u + iqjj (2)

where
icot
(q) = (q R + iqj}

and

[u] = M^N^CM- M*\

A positive value for W indicates a transfer of energy from the system to the airstream,
and hence stability. The matrix U is Hermitian (i.e., U = u) and therefore possesses
real eigenvalues. By use of these eigenvalues it is shown in reference 3 that the energy
input per cycle into the airstream can be reduced to a principal quadratic form as

W = i 7.2pb4w2s
_B (4. + V R
2
+ 2
h ^Rn+i (3)

where Ap Ag An are the eigenvalues of the matrix U and i; denotes generalized

coordinates associated with the aerodynamic energy. It can be seen from equation (3) that
the work W will always be positive if all the eigenvalues A are positive. Therefore, a
sufficient condition for flutter stability is that all the A terms are positive. A notable
characteristic of the energy method is that the criterion for flutter stability is determined
by the characteristics of the aerodynamic-force matrices alone. Therefore, if a particu-
lar system has undesirable flutter characteristics (i.e., too low a flutter speed), the flutter
characteristics can be improved if a mechanism can be found which changes the U matrix
in an appropriate manner. One such mechanism is the addition of control surfaces to the
basic system. The motions of these surfaces generate aerodynamic forces which modify
the aerodynamic terms in the U matrix for the basic system. For flutter suppression
the control-surface deflections are related by a "control law" to the plunging and pitching
motion of the main surface. Nissim points out in reference 3 that a suitable configuration
is one employing both leading- and trailing-edge controls since the two working together
provide independent control of lift and pitching moment.

Delta-Wing Flutter Suppression Model

To evaluate the practical aspects of the aerodynamic-energy concept, The Boeing
Company, Wichita Division, under contract to NASA, performed an analytical study of the
application of this concept to an early supersonic transport (SST) configuration. Some
results of this study, as described in reference 9, indicate increases in the flutter speed
from 11 to 29 percent for several spanwise locations of leading- and trailing-edge controls.
Because of these positive results, an experimental program aimed at providing evaluation
and validation of the energy concept was initiated by using a wall-mounted 1/17-size sim-
plified semispan model of a recent SST configuration. The Boeing Company, under con-
tract to the Langley Research Center, is providing general support for this program in
the area of controls implementation and analysis. A photograph of the model installed
in the Langley transonic dynamics tunnel is shown in figure 1, and a sketch of the model
is presented in figure 2.
 29

The wing has a clipped-delta planform without twist or c a m b e r , a s y m m e t r i c

c i r c u l a r - a r c airfoil .section with a maximum thickness-to-cluud ratio of 0.03, and hydrau-
lically actuated leading- and trailing-edge controls. The trailing-edge control was approx-
imately 20 percent of the local chord, while the leading-edge control varied from about
15 percent of the chord inboard to 20 percent of the local chord outboard. Both controls
were located between 73 and 84 percent of the wing span. These locations a r e approxi-
mately those r e f e r r e d to in reference 9 that resulted in the largest increase in flutter
speed. Simulated engine nacelles a r e mounted on the underside of the wing. The model
construction consists of an internal aluminum alloy plate that was tapered in thickness
in the spanwise direction and had cutouts to simulate s p a r s and r i b s . The plate was
covered with balsa wood to provide the proper aerodynamic contour.

Because of the large hinge moments required and the necessity of keeping the
control-surface actuation system within the physical constraints imposed by the model,
that i s , small and light, it was n e c e s s a r y to design and fabricate an electrohydraulic
actuation s y s t e m . Within these constraints an actuator was designed that weighs only
56.7 g r a m s yet can produce approximately 4.52 N-m of torque throughout the operating
range of interest (approximately from 0 to 25 Hz). Because of the limited thickness of
the wing, it was also necessary to design and fabricate special control-surface position
indicators. This was accomplished by mounting silicon solar cells to the actuator control
shaft and illuminating them with a stationary light s o u r c e . As the control shaft r o t a t e s , a
voltage proportional to the angular position of the surface is produced. A photograph of
the model showing the actuator and position indicator is presented in figure 3. A complete
description of the design and fabrication of the control actuation system for the model is
given in reference 10.

In order to perform analytical calculations for the model, it was necessary to s p e c -

ify a set of generalized m a s s e s , mode shapes, and natural frequencies. These properties
w e r e determined experimentally for the first nine s t r u c t u r a l modes of the model by using
methods s i m i l a r to those described in reference 11. The measured modal contours, nat-
ural frequencies, and generalized m a s s e s a r e given in figure 4.
Control Law
A simplified block diagram of the delta-wing flutter-suppression system is p r e -
sented in figure 5. The control law used is of the form

c c G G
ll 12 ll 12
(4)
c C G
21 22 21 '22

where fi is the leading-edge control deflection; 6 is the trailing-edge control deflec-

tion; hj and a a r e the plunging and pitching motions, respectively, of a representative
streamwise section of the wing (section A-A in fig. 5); b is a reference length; and C
and G a r e constant coefficients which were evaluated from an aerodynamic-energy
analysis.
The motions of hj and hg a r e measured by a c c e l e r o m e t e r s located at 30 and
70 percent of the local chord c. The control law is mechanized on an analog computer
which has been programed to perform the operations indicated in figure 5 to determine
h p h p a , and a , and pass the proper command signal as expressed in equation (4) to
the control s u r f a c e s . Figure 5 indicates that the period of oscillation l/o> must be
determined. However, reference 3 showed that essentially the same r e s u l t s can be
obtained if the value of 1/cu i s taken to be constant and equal to the open-loop flutter
30
period. This result was confirmed by preliminary wind-tunnel investigations.

Flutter Analysis
In order to illustrate the mechanism of flutter suppression, a flutter analysis,
both with and without active controls, is presented. The flutter equations for a three-
dimensional lifting surface are obtained from Lagrange's equation of motion by assuming
that the unknown mode of motion is described by a linear combination of orthogonal modes,
that is, the undamped natural modes of the system, in the following manner:

h(x,y,t) = £ q.WZ^x.y) (5)

i=l

If structural damping is neglected, then the equations of motion become

M ^ t ) + WjZMjq^t) = Q^t) (6)

where

M t = J J m(x,y)Z i 2 (x,y)dx dy
S

is the generalized mass and

Qj(t) = J j Apfr.y.OZ^x.yJdx dy

is the generalized aerodynamic force. The total pressure distribution Ap(x,y,t) is com-
posed of contributions due to each flexible mode plus those due to the leading- and trailing-
edge controls. Therefore,

n
Ap(x,y,t) = ) Ap.(x,y)q (t) + Ap 6 + Ap /3
is j J o p
j=l

where Ap. is the pressure distribution due to each flexible mode, and Ap and Ap.
J p o
are the pressure distributions due to leading- and trailing-edge controls, respectively.
Substituting this expression for the pressures into equation (6) and expanding results in
the following form of the equations of motion:
n
(-w^ + a^M^t) = J /q.(t) J J Ap.Zi dx dyj

+ p J J A p Z j dx dy + 5 j J Ap 6 Z i dx dy (7)
S S

From equation (5), the nondimensionalized deflection of the wing for the responses
h. and hg can be written as

n n
q (t)z x y =
v'hl i i( i' i) F b 1 %^HV2)
i=l i=l
 31

Assuming that a straight line between the locations of the two s e n s o r s gives a reasonable
approximation to the angle of attack at the reference station and noting that the s e n s o r s
a r e 0.8b apart lead to the following equation for angle of attack:

a ta
- o ( b - t l ' oib 1 (zi(x2'*2) - H*vyi})\ (t)
i=l

Substituting the above r e s u l t s into the control law (eq. (4)) r e s u l t s in a matrix equation
relating the control-surface motions to the generalized coordinates in the following form:

W
m Aj + iBj Ag + iBg An + i B n
(8)
C l + IDj Cg + iDg C
n +iD
n

'n

where the p a r a m e t e r s Aj, B i t Cj, and Dj a r e constant coefficients defined a s follows:

C c
A =z ll i; 12
i i( i'yi)l-F--(5iE/ + z i( x 2'y2)o: .8b
x

G G
B ll 12 'JJ.
i = i( ryi)l-_r-OE; + z i( x 2'y2)o
z x
.8b

C C
c 21 22 -22
i'H i*i)[h-mrHx2*2)o .8b
x

/O,21 u
22 + z x
D z x '22
i = i{ i-yi)i— - ro i( 2'y 2 )o .8b

Substituting equation (8) into equation (7) r e s u l t s in the final form of the equations of
motion:

n
(-W2Mj + wi Mi)qi(t) = Y q / J J A Pj Z i ^ dy + fA. + iB^ J J Ap Z. dx dy
2
J
J-l \s s

+ C +
( i ^ j ) I I A P 5 Z i ** d y ) (9)

It should be noted from the form of the equations presented here that the active controls
s e r v e only to modify the aerodynamic forces of the wing alone. The Hermitian matrix U
described e a r l i e r can be derived directly from the aerodynamic t e r m s appearing in this
equation and the effect of active controls on this matrix determined. Flutter calculations
without active controls a r e performed by setting the coefficients A, B, C, and D
equal to z e r o .
32

Flutter calculations were performed for the delta-wing model at Mach numbers of
0.6, 0.7, 0.8, and 0.9. The generalized aerodynamic forces appearing in equation (9) were
formulated through the use of doublet-lattice a^rodvnamics as described in reference 12.
This method requires the subdivision of the lilting surface into an array of trapezoidal
boxes arranged in streamwise columns with a line of pulsating doublets located at the
quarter chord of each box. The geometric boundary condition of tangential flow is satis-
fied at the 3/4-chord location for each box. The delta-wing model was divided into
160 boxes arranged in 16 streamwise strips with 10 boxes per strip. This arrangement
provided six boxes on each control surface. All flutter calculations were made using the
first nine measured structural modes, generalized masses, and natural frequencies. It
should be noted that the equations of motion did not include control-surface dynamics
since the natural frequency of rotation for each surface was considerably above the fre-
quency of interest.

Results
Flutter.- Flutter characteristics of the model without active controls were experi-
mentally determined in the Langley transonic dynamics tunnel at Mach numbers of 0.6,
0.7, 0.8, and 0.9. For these tests the control surfaces were kept at 0° deflection by
applying hydraulic pressure to the actuators. The pressurized system acted as a very
stiff spring and kept the rotational frequencies of the controls many times higher than the
flutter frequency. Once the flutter boundary of the wing was established, an evaluation of
the effect of active controls on raising the boundary was begun. However, these studies
were conducted only at a Mach number of 0.9 because of an unexplained high-frequency,
large-amplitude oscillation of the leading-edge control above a certain range of dynamic
pressure at the lower Mach numbers. This phenomenon occurred around 65 Hz, whereas
the flutter frequency was 11 to 12.5 Hz. This problem is not believed to be a result of
the control law, since this motion is also observed with the control loop open, but has
been introduced in some manner by the mechanization of the controls on the model.
The experimental flutter results are presented in figure 6. At a Mach number of
0.9 the basic wing model fluttered at a dynamic pressure of 5.879 kN/m2. With active
controls the flutter point was raised to 6.607 kN/m2_ reflecting an increase of approxi-
mately 12 percent in dynamic pressure. The degree of confidence in the control system
was such that when open-loop flutter was encountered, the active control loop was closed
to suppress the motion. The observed flutter motions for both open- and closed-loop
operation were similar in nature and closely resembled the second natural vibration
mode with some slight primary bending.
A comparison of calculated and experimental data is also presented in figure 6.
The calculations for the basic wing show excellent agreement at all Mach numbers; how-
ever, the calculations with active controls predict a higher flutter point than was meas-
ured. At a Mach number of 0.9 the calculated increase in flutter dynamic pressure was
approximately 21 percent compared with the measured increase of 12 percent. This dif-
ference is believed to be due to the inability of the aerodynamic theory to predict ade-
quately the pressure distributions resulting from actuating the control surfaces, the lack
of control-surface dynamics in the equations of motion, and the amplitude and phase lags
incurred between the desired and actual control-surface deflections introduced by imple-
menting the control loop on the model.
On the basis of these calculated results, it was decided to investigate analytically
the sensitivity of the system to phase lag between the desired and actual control-surface
deflections. A separate set of calculations were made which included a phase lag for both
 33

leading- and trailing-edge controls. The phase angle was experimentally determined for
the model by measuring the frequency response of the actuator systems. At the flutter
frequency of about 12 Hz, both surfaces had a phase lag of approximately 18°, and this
value was used in the calculations. The results of these calculations are presented in
figure 6. At a Mach number of 0.9 the phase angle reduced the increase in flutter
dynamic pressure from 21 percent to 16 percent and resulted in a more favorable com-
parison with experiment.
Subcritical response.- In order to explore fully the behavior of the model below the
flutter boundaries, two techniques for estimating the damping associated with the flutter
mode were used. The first of these techniques (described in ref. 13) involves measuring
the forced response of the model to an input generated by the trailing-edge controls as
indicated in sketch 1. A measure of the damping in each mode can be obtained for both
open- and closed-loop operation if the transfer function relating the forced response to the
command signal \ b . \ / \ c ) is determined a.s a function of frequency. During the wind-tunnel
test an electronic signal analyzer was used to determine the in-phase and out-of-phase
components of the response hj with respect to the trailing-edge command signal. Fig-
ure 7 presents a typical plot of this response at a Mach number of 0.9 and a dynamic pres-
sure of 5.429 kN/m2. The curves in the upper portion of this figure represent the response
of the basic wing; the lower curves, the response with the control loop closed. The damp-
ing in the modes can be estimated from the out-of-phase component by the frequencies
labeled fA and fR. For an equivalent system with a single degree of freedom, these

•yy Trailing-
edge
aciuator

6^
Leadirm-
edge
actuator

Leading-edge
.\_ control law
/ A

Trailing-edge
control law

Sketch 1

are the frequencies at the half-power points, and the damping can be expressed in terms
of these frequencies:

(V'B) 2 - 1
(VB)2 *
The data shown in figure 7 are for a 3-minute logarithmic sweep from 5 to 25 Hz.
A qualitative measure of the effect of active controls in reducing the forced response
of the system is evident from figure 7. The closed-loop system significantly alters the
response by adding appreciable damping to the model. However, a quantitative measure
of the damping is quite difficult to estimate because of the noise in the signal resulting
from the model responding to tunnel turbulence. As the dynamic pressure is further
increased, the signal-to-noise ratio becomes extremely high and results in very poor
34

response plots. For this model the forced-response technique did not provide very use-
ful information, but it will be shown later that this procedure can be an extremely useful
tool in estimating the subcritical response.
The second technique that was used is referred to as "randomdec" and is described
in reference 14. Basically, the technique extracts the damped sinusoidal response of the
model vibration modes from the response of the model to tunnel turbulence. This is
accomplished by assuming that the measured response is composed of the response to a
step, an impulse, and a random force. By averaging the measured response over a num-
ber of time sweeps, the response of the system to a step is determined, since the response
to an impulse and to a random force average to zero. Damping is then obtained in the
same manner as from a free-vibration decay which would be obtained if the model were
given an initial displacement in the critical mode and then released. For a system with
multiple degrees of freedom a filter is required to isolate the modes of interest. It
should be pointed out that when the frequencies of the structural modes are closely
spaced, both the randomdec and forced-response techniques suffer from the problem of
accurately determining the damping value.
The randomdec technique was used to obtain the plot in figure 8, which is the meas-
ured system damping in the critical flutter-mode frequency range as a function of dynamic
pressure at a Mach number of 0.9. The hatched area represents the experimental scatter
of the data. Typical randomdec signatures are presented at a dynamic pressure of
5.841 kN/m2, which is within 1 percent of the flutter dynamic pressure. The open-loop
damping is about 0.008; the closed loop, about 0.075. Also shown are the measured open-
and closed-loop flutter points. Because of the frequency spectrum of the structural modes
of interest, this technique proved to be quite valuable in establishing the subcritical behav-
ior of the model. For test conditions at which the forced-response technique described
earlier gave meaningful results, these results also fell within the scatter indicated in
figure 8.

B-52 FLUTTER-SUPPRESSION PROGRAM

In addition to the delta-wing program, the Langley Research Center is engaged in

a cooperative program with the Air Force Flight Dynamics Laboratory to study symmet-
ric flutter suppression on a model of the B-52 CCV airplane. As mentioned in the Intro-
duction, the B-52 program will include studies of the application of other active control
systems including maneuver-load control, ride-quality control, and relaxed static stability.

Model Program
The model program uses a 1/30-size dynamically and aeroelastically scaled model
of the B-52. A photograph of the model installed in the Langley transonic dynamics tunnel
is shown in figure 9. In order to provide a simulation of the free-flight dynamics, the
model is mounted on a modified version of the two-cable mount system described in ref-
erence 15. This mount provides the model with a soft support in that the natural frequen-
cies associated with the mount are well below those of the free-flight and elastic modes.
The active flutter-suppression system designed for the model is indicated in the
simplified block diagram shown in sketch 2. This control system was not designed with
the use of the energy approach discussed earlier. It is a result of previous experience
and analysis of the B-52 which have indicated that aerodynamic forces on the wing are
stabilizing for 360° of the flutter oscillation cycle when the incremental lift generated by
the control surfaces lags the wing motion by 90°. The control law is essentially a shaping
 35

filter which provides the required phase lag between wing lift and displacement at the flut-
ter frequency. A summary of the analysis, synthesis, and hardware implementation being
used for the flight program is presented in reference 16.
6
a "outboard

Mudel

«( h
inboard

Aileron Flaperon
actuator actuator

Shaping
filter

Shaping
filter

Sketch 2

As indicated in sketch 2, the control system incorporates an active flaperon and

outboard aileron. The placement of these controls is indicated in figure 10. An out-
board accelerometer is used to drive the ailerons. A second inboard accelerometer is
used to drive the flaperons. Because of the smaller hinge moments required for this
model and the substantially larger volume available than in the delta-wing model, an
electromechanical system was designed to actuate the controls. This system consisted
of separate dc torque motors mounted within the fuselage to drive the ailerons and flap-
erons. The linkages used to drive the controls were designed to isolate them from the
structure so that they would not change the stiffness characteristics of the wing. A
description of this system is presented in reference 10.

Results
Experimental studies of the B-52 model were performed in the Langley transonic
dynamics tunnel. The prime objectives of these tests were to establish the behavior of
the flutter-suppression system below the flutter boundary. A plot of estimated damping
in the critical flutter mode (approximately 12.8 Hz), using the forced-response technique,
against tunnel dynamic pressure is presented in figure 11. Experimental results for the
open-loop system, the closed-loop system with nominal gains, and the closed-loop system
with double the nominal gains are indicated in this figure. It is readily apparent from these
results that the effect of active controls is appreciable. Even with nominal gains, the
damping at a dynamic pressure 2.42 kN/m2 is more than double that of the open-loop
system. With twice the nominal gains, not only has the level of damping increased but
even the trend with increasing dynamic pressure has reversed direction.
A typical plot of the measured in-phase and out-of-phase response of the model is
presented in figure 12. For this model the ailerons were used to generate the forcing
function. As indicated in figure 12 the damping was estimated by determining the ratio
of the outboard-accelerometer response to the aileron command for a frequency range
of 4 to 24 Hz. The randomdec technique did not provide useful results until the model
was tested near the flutter boundary, at which time most of the wing response was pre-
dominantly in the lowly damped flutter mode.
36

CONCLUDING REMARKS

A description of two wind-tunnel studies used to evaluate active control of flutter

suppression has been presented. A flutter-suppression method based on an aerodynamic-
energy c r i t e r i o n has been d e s c r i b e d , and some r e s u l t s of the application of this method
to a simplified delta-wing model a r e presented. An i n c r e a s e of approximately 12 percent
in the flutter dynamic p r e s s u r e for this model has been achieved at a Mach number of 0.9
through the use of leading- and trailing-edge c o n t r o l s . Analytical calculations have been
compared with experiment and indicate excellent a g r e e m e n t for the open-loop system;
however, calculations for the closed-loop system predicted a l a r g e r i n c r e a s e in dynamic
p r e s s u r e than was m e a s u r e d .
Some p r e l i m i n a r y experimental r e s u l t s of a flutter-suppression study of an a e r o -
elastic model of the B-52 CCV airplane have been presented. A flutter-suppression
method based on the phasing between wing motion and control-surface deflections has
indicated that significant improvements in the subcritical damping of the flutter mode
can be achieved through the use of active controls.
ACKNOWLEDGMENTS

The authors wish to e x p r e s s their appreciation to the Air F o r c e Flight Dynamics

Laboratory for their participation and support during the B-52 model p r o g r a m . In
addition, the authors a r e p a r t i c u l a r l y indebted to Mr. Frank D. Sevart of the
Boeing/Wichita Company for his technical contributions during all of the studies
described in this r e p o r t .

REFERENCES

1. Wykes, John H.; and Kordes, Eldon E.: Analytical Design and Flight T e s t s of a Modal
Suppression System on the XB-70 Airplane. Aeroelastic Effects F r o m a Flight
Mechanics Standpoint, AGARD C P - 4 6 , 1970, pp. 23-1 - 23-18.
2. B u r r i s , P . M.; D e m p s t e r , J . B.; and Johannes, R. P . : Flight Testing Structural P e r -
formance of the LAMS Flight Control System. AIAA Paper No. 68-244, Mar. 1968.
3. N i s s i m , E.: Flutter Suppression Using Active Controls Based on the Concept of A e r o -
dynamic Energy. NASA TN D-6199, 1971.
4. Triplett, William E.; Kappus, H a n s - P e t e r F . ; and Landy, Robert J . : Active Flutter
Control: An Adaptable Application to Wing/Store F l u t t e r . AIAA P a p e r No. 73-194,
Jan.1973.
5. Thompson, Glenn O.; and K a s s , Gerald J . : Active Flutter Suppression - An Emerging
Technology. J . Aircraft, vol. 9, no. 3, Mar. 1972, pp. 230-235.
6. Rainey, A. Gerald; Ruhlin, C h a r l e s L.; and Sandford, Maynard C : Active Control of
Aeroelastic Response. Stability and Control, AGARD C P - 1 1 9 , 1972, pp. 16-1 -
16-5.
7. Kass, Gerald J.; and Johannes, R. P . : B-52 Control Configured Vehicles P r o g r a m .
AIAA Paper No. 72-747, Aug. 1972.
8. Hunt, Gerald L.; and Walberg, Gerald D.: Calculated Mode Shapes and P r e s s u r e Dis-
tributions at Flutter for a Highly Tapered Horizontal Tail in Subsonic Flow. NASA
TN D-1008, 1962.
 37

9. Garrick, I. E.: P e r s p e c t i v e s in Aeroelasticity. Israel J . Techno!., vol. 10, no. 1-2,

1972, pp. 1-22.

10. Bergmann, Gerald E.; and Sevart, F r a n c i s D.: Design and Evaluation of Miniature
Control Surface Actuation Systems for Aeroelastic Models. AIAA Paper
No. 73-323, Mar. 1973.
11. Abel, Irving: A Wind-Tunnel Evaluation of Analytical Techniques for Predicting Static
Stability and Control C h a r a c t e r i s t i c s of Flexible Aircraft. NASA TN D-6656, 1972.
12. Albano, Edward; and Rodden, William P . : A Doublet-Lattice Method for Calculating
Lift Distributions on Oscillating Surfaces in Subsonic Flows. AIAA J . , vol. 7,
no. 2, F e b . 1969, pp. 279-285.
13. Keller, Anton C : Vector Component Techniques: A Modern Way To Measure Modes.
Sound & Vib., vol. 3, no. 3, Mar. 1969, pp. 18-26.
14. Cole, Henry A., J r . : On-Line F a i l u r e Detection and Damping Measurement of A e r o -
space Structures by Random Decrement Signatures. NASA CR-2205, 1973.
15. Reed, Wilmer H., HI; and Abbott, Frank T., J r . : A New " F r e e - F l i g h t " Mount System
for High-Speed Wind-Tunnel Flutter Models. Proceedings of Symposium on A e r o -
elastic & Dynamic Modeling Technology, RTD-TDR-63-4197, P t . 1, U.S. Air F o r c e ,
Mar. 1964, pp. 169-206.
16. Hodges, Garold E.: Active Flutter Suppression - B-52 Controls Configured Vehicle.
AIAA Paper No. 73-322, Mar. 1973.
 'S

Fig.l Delta-wing flutter-suppression model

 39

Leading-edge control

Fig.2 Sketch of delta-wing model. (All linear dimensions are in meters)

 £

Accelerometer

Fig.3 Delta-wing control system

 •attfltumuM

W*is • 1.536 kg

MOM 7 FREQUENCY * 4 5 . 8 6 HERTZ MODE 8 FREQUENCY • 4 8 . 1 5 HERTZ 9 FREOUENCY : M.OT HERTZ

Fig.4 Measured modal contours, generalized masses, and frequencies of natural vibration modes. (Contour interval,
0.1 normalized displacement)
 Section A-A

«*••
r,
Iz
1
p (coramanu r C
12

s <•
L^
C
ll
*w G
ll
Leading-edge
actuator a 1
dynamics 1 u J r /J
V

Q
Model
h
n

*1
2 1
s

1
h
\

h
l
o .
fc--•
1
h,
1 1
0.8 t

1
a
T
1

1
h
l
J
V

6 s s . S
h
l
bo;
Trailing-edge (si
actuator
dynamics
G
, .l
6 (commancl)
* * • "

cV * =
C
21
C
22
-^ G
22

Fig.5 Block diagram of flutter-suppression system

 43

20

Frequency, ,0
IU
Hz

ol V

12

Open loop Closed loop

Unstable O Experiment • Experiment
D Calculations Q Calculations
10
[ Calculations with
18° phase angle

Dynamic p r e s s u r e ,
kN/m 6

Stable

1 1
.6 .9 1.0
Mach number

Fig.6 Measured and calculated model flutter boundaries

44

t.C

m/sec
deg

15
Frequency, Hz

(a) Open-loop response, g = 0.037

10 r

^j_ -10

m/sec
deg

i i
5 10 15 20 25
Frequency. Hz

(b) Closed-loop response, g = 0.107

Fig.7 Measured forced response of model to trailing-edge-control excitation at Mach number of 0.9 and dynamic
pressure of 5.429 kN/m2
 4^

.20

.16 -Closed loop

.12

Damping,
g

.08

.04

A 4.0 4.4 4.8 5.2 5.6 6.0 6.4 6.8

2
Dynamic p r e s s u r e , kN/m

Fig.S Measured subcritical damping at Mach number of 0.9

Fig.9 B-52 CCV model mounted in the Langley transonic dynamics tunnel
46

Outboard aileron

Accelerometers
Fig. 10 B-52 model control surfaces used for flutter suppression

.20 r

.16
O Open loop

D Closed loop - nominal gains

.12 ^ C l o s e d loop - double nominal gains

Damping,
g

,08

.04

y^. 2.0 2.4 2.8

Dynamic pressure, kN/m 2

Fig.l 1 Measured damping for B-52 flutter-suppression model

 47

In-phase

-1.0

-1.5

outboard
-2.01
6
a.c
2
m/sec
deg

12 16 20 24
Frequency, Hz

(a) Open-loop response, g = 0.062

outboard -i o
6
a.c
p
m/sec
deg

(b) Closed-loop response double nominal gains, g = 0.176

Fig. 12 Measured forced response of B-52 model to aileron command at a dynamic pressure of 2.154 kN/m2
48
 4')

CONTROLE DU FLOTTEMENT PAR DEPLACEMENT

DUNE VALEUR PROPRE

par

J.Angelini

ONERA
29, ave de la Division Leclerc
92320 Chatillon-sous-Bagneux
France
su

Resume

Le document propose une nouvelle m^thode en vue du contrflle actif du flottement des avions. Cette
methode est oaract^risee par le fait qu'elle permet de ne modifier que 1'amortissement de la valeur propre
instable, sans que les autrea raoines, ni les vecteurs propres soient modifies.

Une telle approche pennet d'esperer que le systeme actif concr£tis£ presentera le minimum d'inter-
action avec d'autres systfemes C.C.V. susceptibles d'etre installes sur la structure.

FLUTTER CONTROL BY MODIFICATION OF AN SIGEN VALUE

Summary

The technique presented here aims at defining an active control system to stabilize a multiple
degree of freedom coupling of modes on an aircraft.

This technique modifies only the value of the unstable root of the characteristic equation and keeps
unchanged the corresponding eigen vector and the other generalized characteristics.

Considering this property, it is hoped that this method leads to the least possible interaction
with the system.
 51

INTRODUCTION

Le contrOle du flottenent des avions est 1'un des aspects de la doctrine C.C.V. dont la necessite
ne fait & present aucun doute.

Cette necessity ayant 6t6 de'gagee il s'agit maintenant de rendre realisableo les principes qu'elle
impose. Ainsi a. 1'ONERA, s'est poe^ naturellement le probleme de d^finir, dans un cadre donne, le moyen
de materialiser les possibilites de contr61e du flottement.

Le cadre choisi, qui est loin d'etre unique, est le suivant : on envisage un flottement fonda-
aental sur une voilure munie d'une gouveme classique ; est-il possible de definir un moyen de oontrflle
qui, agissant par 1'interm^diaire de la gouverne, supprime le flottement?

L'idee de la construction de oette loi de contrftle est bas£e axis la remarque mathematique suivante

soit une matrioe A de valeurs propres ^ ^ de veoteurs propres a droite [ V j d e veoteurs propres
k gauche U , on a alors la decomposition suivante :

A - CVJTNA, Cul

soit X 0 , U 0 , V une c e r t a i n e v a l e u r propre e t s e s v e c t e u r s propres a s s o c i e s

alors l a matrice A + \> U0 V 0 T

adnet X,. -r V, V 0 U 0 coame nouvelle v a l e u r propre e t v e c t e u r s propres a s s o c i e s . Cela permet

d ' e n v i s a g e r l a p o s s i b i l i t e de modifier une r a c i n e d'un systeme sans toucher a u i v e c t e u r s propres e t a u i
a u t r e s r a c i n e s , done de peu p e r t u r b e r un aystfeme t o u t en l e rendant s t a b l e .

I I - DEFINITION DU KODEI_E REPRESEKTAMT LE FLOTTEKENT ET LA LOI DE COHTROLE

Le systeme s u i v a n t
r •» r 1 1

P j 0
i + _—K 1 + PvAt 0
(1) • \\ v
< 1
y

. S
s
1
' >

donne les Equations des petits mouvements de 1'avion. <"> est la colonne des ooordonnees gen^ralisees
ou u la matrice diagonale des masses generalises.

La coordonnee 6 represente le mouvement relatif de la gouveme et n'est pas en g6n4ral orthogo-

nale aui autres coordonn^es, 1'operation < \ est 1'operateur des forces aerodynaniques assooiees au
mouvement et f' H repr^sente la commande qui est un couple exerce sur la chamifere de la gouveme.

Un certain nombre d'accelerometres places sur structure permettent d'identifier ses mouvements ou
ses deformations. Nous supposerons done qu'_i partir de ces mesures aocelerometriques il est possible de
remonter a la connaissance des coordonnees generalises Q .

Une mesure angulaire foumira par ailleurs la coordonnee C

Effectuant la transformation de Laplace de { 1 ),

on pose D (>) - L| S + k -+ ('v A(s)

( 5> e s t l e v a r i a b l e de L a p l a c e ) .

La s t a b i l i t y du systfeme e s t d&finie p a r l e s r a c i n e s de 1 ' e q u a t i o n :

Di.,- ( D ) - 0
52

Soit X ( 1'ensemble des racines stables, c'est-k-dire a partie reelle negative X leur valeur
conjuguee.

Soit X ^ une racine instable a la vitesse V on a par definition

Si 1'on veut modifier la valeur de X ^ d'une quantity V on aura done :

Appelona V ^ la premiere ligne de la matrioe I_> ( )*. + V J

Choisisaons comme loi de pilotage la relation :
LOI DE CONTROLE

0. -.
(2) m Qfc) V

n v.,. f/A^= re.

: \i \ _ I _S_ A, \ / j>_ _ X n- 1\
8,1 TT = v
',' W ' v ; \ v j \v v J
et Q ( ~ ) est un polynfime arbitraire de degre au moins egal a celui de TT

On remarque que, pour le nouveau systems d'Equation :

1) lea racinea X ] e t lea vecteurs propres associes aont inchang^a

2) on peut choifrir Y, pour q_e X n + Vsoit la nouvelle valeur propre associee aui vecteurs propres
de \ n

Le d^nominateur Q a i t i introduit uniquement pour avoir une loi repr^sentable par une fraction
rationnelle o'est-a-dire n'introduiaant pas les derives des grandeurs mesur^ea.

Enfin il est logique de choiair V de facon a n'agir que aur la partie r^elle de X j, c'est-a-dire
de fa?on a ne modifier que 1'amortissement du systeme ce qui, en general, ne demandera que peu d"energie.

Ill - APPLICATION

A partir des iddes precedentes 1'ONERA a decide de realiser un modele en soufflerie pour etudier
la faisabilite d'un systfeme de oontrBle et en vue d'etudier la sensibilite aui diverses perturbations
et aux erreura d*appreciation des parametres.

Dans une premifere etape le modele choisi est une aile rectangulaire montee a la paroi dans une
soufflerie subsonique, voir figure n° 1.

Cette aile a i t i dimensionnee de facon a presenter un flottement flexion toraion vers soixante
mfetres seconde. Elle possede une gouveme dont la raideur est donnee par un moteur couple.

Ce moteur couple sert egalement k assurer le contrftle.

La realisation materielle de 1'aile est actuellement en oours et le montage oomplet sera experi-
ment i en soufflerie au debut de 1'annee 1974.
 5,!

Des calculs de prevision de flottement pour cette aile sont reunis sur la figure n° 2 suivant la
presentation claasique a 1'ONERA (frequence et amortissement fonction de la vitesse).

Pour definir la loi de oontrflle le denominateur Q ( — ) a ete choisi de la forme

et on a decide de rendre positive la partie reelle de la raoine instable pour une vitesse de 75 m/_>« On
a done refait le calcul de stabilite dans ces nouvelles conditions et les resultats sont resumes dana la
figure n° 3.

Ces reaultats am-snent les comment aires suivants :

1) le flottement est bien contrflie k 75 m/s

2) il 1'est egalement entre 0 et 80 m/s, bien que la loi ait ete etablie seulement pour 75 m/a

3) le coefficient du terms en B dans la loi oalcuiee eat extrSmenent petit.

En consequence on a aimplifie la loi de commande en annulant le coefficient de 6 ; les calculs

ont done ete repria dana ces conditions et les resultata sont preaentes sur la figure n° 4.

On peut constater le peu de sensibilite de oes resultats par rapport aux parametres prls en compte.

rub. Le fait que le contrOle etabli a 75 m/a soit efficace sur plage (0 - 80 m/s) est une oircons-
tance heureuae qui n'eat paa une consequence de la theorie. Dans d'autres problfemes il faudrait peut-^tre
envisager une evolution des coefficients de la loi de contrflle en fonction de la pression dynamique et
du nombre de Hach.

CONCLUSION

La methode presentee ici est simplement une technique permettant de stabiliaer un avion ayant un
flottement k une oertaine vitesae.

Cependant dans le cadre de la doctrine C.C.V. elle presente 1'avantage important de ne pas interferer
avec les autres systemes de contrOle.

En effet si d'autres racines ont ete modifiees, pour dea raiaona de pilotage par exemple, la methode
propoaee ne lea modifiera pas, a condition, bien aflr, qu'on les prenne en compte.

Signalons enfin que cette methode peut £tre utilisee pour d'autres contrfiles que celui de flottement,
par exemple pour la stabilisation d'un avion a marge statique negative.
 l
.73f
I H2

to..

orf

Fig.l Dimension du modele de soufflerie prevu pour 1974 Fig.2 Calcul classique
 \*l
^
A '_
/ <_
/ > / — « i V •

/" /
to / /

•o o- o o-
*4 /io

0.0 J • of

* * JL ^

Fig.3 Avec controle Fig.4 Avec controle simplifie

56
 57

ACTIVE FLUTTER SUPPRESSION ON WINGS WITH

EXTERNAL STORES

by

G.Haidl, A.Lotze and O.Sensburg

Messerschmitt-Bolkow-Blohm GmbH
Unternehmensbereich Flugzeuge
Postfach 801160
8 Munchen, 80, Germany
ss

SUMMARY

A control system is described, which is able to suppress flutter of wing-external store

combinations. The aerodynamic flutter suppression forces are generated by movable vanes,
attached to the stores, which are moved by a feedback signal from the store motion in
such a way, that these forces damp the store motion.

By adjusting the phase of the servoloop it is possible to have an active flutter-system

below the flutter-speed of the passive system. This arrangement can be used to excite the
flutter-mode at subcritical speeds. By switching off the servoloop, damping and frequency
can be evaluated.

The active flutter suppression system can also be used for reducing the level of
externally forced vibration on stores which could occur through excitation by buffet or
gusts.

Tests results for an elastic wind-tunnel model are given and compared with analytical
predictions. Correlation is very good, considering the complexity of the problem.
 59

ACTIVE FLUTTER SUPPRESSION ON WINGS WITH

EXTERNAL STORES

G.Haidl, A.Lotze and O.Sensburg

1. INTRODUCTION

External store carriage on wings of variable geometry fighters is posing a huge problem to the flutter analyst.
Due to varying weight and inertia of stores together with changing wing sweeps and increased flight envelopes, no
external store position on the wing nor a suitable store attachment stiffness can be selected that would make all
possible stores flutter free. Stiffness increase or mass balance requirements penalize the aircrafts performance.
Active flutter suppression (AFS) is a possible and promising solution to the problem.

A system which is capable of suppressing wing store flutter was developed and tested on a subsonic wing
tunnel model in the flutter tunnel of the Eidgenossiches Flugzeugwerk in Emmen.

The control system drives a vane, attached to a store, controlled by a feed-back signal in a way so that it
counteracts the store motion. The developed mechanism can also be utilized for conventional flight flutter testing
excitation techniques, such as frequency sweep or harmonic signals, in addition to providing a method for quickly
finding the frequency and damping of the flutter mode. Another application of the system is the reduction of
external store amplitudes created by buffet or air turbulence, thus increasing the fatigue life of wing attachments,
improving the target aiming of weapons and enhancing the clearness of pictures shot by reconnaissance cameras in
wing mounted pods.

This paper concentrates on the development of a flutter suppression system, gives a mathematical description
of it and shows the different applications on a dynamically scaled wind tunnel model. No attention is paid to the
implementation of the system in an already existing flight control augmentation system such as is usually available
on modern fighters. This aspect is covered to some extent in (1) and (2). There are indications from (2), lhat it
might not be possible to use the already existing CSAS because of couplings introduced by the sensor mountings
in the fuselage.

2. FLUTTER CONFIGURATION AND MECHANISM

Several configurations varying the number of stores and the sweep angle were tested. Analytical comparisons
were only performed for a wing leading edge sweep angle of 45°.

Two flutter modes occur containing mainly the properties of the still air modes (Fig.l). One flutter mode
can be described as the classical wing bending-torsion, store pitch flutter. The other flutter mode is a problem
mainly introduced by the relatively low yaw attachment stiffness which is inherent on variable geometry airplanes.

The fore and aft motion of the wing (not producing any considerable unsteady aerodynamic forces), together
with the large mass of the store hanging below the wing, induces a torsional angle on the wing which produces
unsteady aerodynamic forces that can cause flutter together with the wing bending mode unsteady air force.

3. VANE EXCITATION AND CONTROL SYSTEM

Excitation created by oscillating aerodynamic vanes is considered to be the most effective way of exciting low
frequency vibration modes. These vanes, together with a driving and control system, can be easily installed in empty
fuel tanks. The different stages of fuel loading, simulated by masses installed in the tank whose distance from each
other can be varied, can represent the mass and inertia properties of all stores that are considered to be flutter critical.

The vane system must fulfill the following tasks:

1. It must follow any specified input signal given to it. i.e.. fixed frequency sine wave, variable frequency
sine wave, stochastic signals, shocks (open loop operation).
2. It must either excite vibration modes or suppress flutter modes by a suitable control circuit (closed loop
operation).
60

This vane system used on an aircraft would be driven by an electro-hydraulic energy source. For the model tests,
an electro-dynamic system was used. Considering the frequencies involved there seems to be no major difference in
those two systems.

Two electrical torque motors per tank were chosen for driving the vanes. The maximum total moment of the
two motors working together was 480 cm p. The mass moment of inertia is 60 • 10" 4 cm p.sec 2 , the motor weight
being 109 gr per motor. A metal-balsa wood construction of vanes of different shapes and dimensions was used.
The rotation axis was at 26% of the chord to minimize the aerodynamic moment. The angular position of the vane
was controlled by a Hall generator which produces a voltage proportional to the angle. The maximum angle is ±15°.
The vane was located at the forward tank end in order to provide an undisturbed flow. It was also possible to turn
the whole tank around, so that the vanes were at the rear tank end. Weight, radius of gyration and center of gravity
of the tank could be varied independently. Figure 2 shows the tank with different vanes.

In Figure 3 a schematic picture of the AFS is given.

The ideal control law is described in Equation (5). The realized function is depicted in Figure 4.

From this figure it can be readily seen that the ideal control law is fulfilled in the frequency region of interest
from 6 Hz to 8 Hz. It was possible to feed the integrated differential sensor signal (z,/ico — z 2 /ico) into the system,
thus compensating for a translatory motion or to use only the signal of one sensor z,/ico . For the latter case the
gain was halved. The angular position of the vane can only be controlled as long as the available moment of the
servo motor is higher than the externally applied moment. This condition was always fulfilled in the tests. It should
be considered that the vane angle must not be greater than the vane stall angle for the AFS to be effective.

4. ANALYTICAL MODEL

A vibration analysis was performed using component branch modes as generalized coordinates. Three-dimensional
unsteady aerodynamic forces for the wing and the vanes were calculated. No aerodynamic forces on the tank and no
interference air forces between wing, vane and tank were considered. Equation (1) describes in general the motion of
the complete aeroelastic system including a control system. All generalized terms of Equation (1) are dimensionless

(—2 IM q q l + — • - ^ - • £ • T - IC" 1 + [ I C ] + - ^ - . V2 . [C' ]) {q} = [F ] {«} (1)

MM MM MM HH M
\c__7 cor m r co r s k m r co r s /

where: m r co r = reference mass and frequency

s = semispan of reference plane

A = area of reference plane

v = true air speed

cos
k = = reduced frequency
v

a = generalized coordinate of the control mode

q = generalized coordinates of the free airplane modes

p = ico or d/dt respectively

C ' , C " = real and imaginary part of the generalized air force coefficients

M = generalized mass

K = generalized stiffness

F = generalized forces
The right-hand side product of Equation (1) vanishes for the uncontrolled case.
If external forces generated by the control mode a are being introduced into the aeroelastic system, the
right-hand side term F q a is replaced by expression (2).

4 M 4 H
cof cor m r co r s k m r co r s

For the controlled system the generalized coordinate of each control surface mode is a frequency dependent function
of the structural displacements at the sensor location.
 61

Considering only second order transfer functions

( T j + pT* + p 2 T*)a = (IT*) + p l T * l + p 2 lT*J){q} (3)

for the analytical representation of each control loop, Equations (1), (2) and (3) can be combined to form the matrix
system (4)

f-,.1 r-.ll
qq *"qa jp_ I p/2 A _ s_s c
qq *-qCK 0 0
+ ) •
T* T* co Ym co s — k • v• k 0 0 j
r r r
[T* id /
K
qq 0
+
P/2 A
• v* • c ^qot
+
0 0
)
q
= 0 (4)
OJ m r co r s 0 0 a
lo [r* T:J /
Using the QR-Algorithm, a computer program was used to solve Equation (4) for the complex eigenmodes and
the corresponding complex eigenvalues.

Only symmetrical modes were considered because the mode was fluttering symmetrically.

In order to provide sufficient damping in the structural modes picked up by the sensor, the control system was
designed to produce amplitudes of the vane displacement proportional to the velocity of the store oscillations over
the frequency region of interest.

Equation (5) describes this relationship

*(«) , • n
''control "
K • eiA* (5)
''sensor * 1

where ^ sensor =
local displacement at the sensor location in a free airplane mode q ;
= an
^control 8 u ' a r displacement of the vane in the control mode a .

The nominal value of K gives a ratio of angular displacements of vane and store of 8.5 at 8 (Hz). The nominal
value of Axp is zero at 8 (Hz).

The phase setting A*p was provided in the model for corrections of possible phase shifts between the vane
displacements and the unsteady air forces generated by the control mode a .

Using the measured transfer function of the total control loop, the control equation can be approximated by (6).

^ontro. ' « ~ (T, + PT 2 )l0<?Lr) W " 0 . (6)

Using Equation (6) the constants of (4) can be defined:

l ^ = -T.l^sorl

llti = - T . l ^ s o r J
1
4 ^control '
IT?) = 0

It = T5 = 0.

All calculations were performed for a wing sweep position of X\vc = 45°. For the system with AFS, two
vanes (defined as standard vane surface) on each store tank were introduced into the calculations. In general, the
forward stations of the store tank are used for location of sensor and vanes.

Because test results were only available for Ma = 0.2 , most of the calculation was performed for this Mach
number. In order to show the efficacy of the AFS at higher Mach number, unsteady aerodynamic forces for
Ma = 0.9 were also introduced.

In the analysis the effect of the gain and the phase shift of the control system was investigated and depicted in
v — g and root locus plots. The application of the AFS to the classical wing-store pitch bending flutter problem was
demonstrated by eliminating the wing pivot yaw mode.
62

5. WIND-TUNNEL TESTS

Dynamically scaled subsonic wind-tunnel flutter models are used to a large extent at MBB for flutter
investigations3.

These tests are made at the flutter wind tunnel of the Eidgenossisches Flugzeugwerk in Emmen. This tunnel
has a quick stop facility which allows an increase in speed until flutter occurs without destroying the model. At
the flutter point, the frequency is measured and the flutter mode visually inspected and filmed.

For the mild flutter cases, existing when external stores are attached to the wings, there was a strong desire
to also measure the damping trends by exciting the model, thus being able to exactly define the flutter point.
This exact definition is especially needed for comparison of analytical and test results. Since the model is free
flying in the wind tunnel (held by a rod and supported by an air spring), the excitation system had to be built
into the model. The vane excitation system, shown in Figure 3 was installed in the external stores.

Another vane was attached to the forward fuselage, to be able to excite the model with the AFS (Automatic
Flutter Suppression) switched on.

Several ways of damping and frequency evaluation were tested, considering later application to full-scale flight
flutter testing. Producing vector plots (Fig.S) with harmonic excitation or using excitation switch-off decays is too
time consuming, because the system takes too long to get to a steady state response due to the low frequencies and
dampings involved. Evaluation of frequency sweep responses by statistical methods was more economic. Using
transportable special purpose computers the damping trends can be monitored quasi on line by evaluation of the
auto-correlation function of the response. A more detailed description of the method is given in (4).

Figure 6 shows a model response to a frequency sweep.

A very powerful method was used during the test. For this method the phase of the AFS vane forces is
shifted 180° so that essentially an unstable system is created. This AFE (Automatic Flutter Excitation) has two
outstanding advantages:
1. It is automatically tuning the frequency into the store flutter mode providing a suitable sensor signal.
2. Switching off the AFE, one can easily evaluate damping from the logarithmic decrement of the response.
In comparison to the frequency sweep response, which cannot be analysed without the help of computers,
the signals produced by the AFE can be used directly by the engineer.

Figure 7 shows an application of AFE. Because the model is free flying in the tunnel a high angle of attack
could be simulated that caused wing stall and a high noise environment. Figure 8 shows that the AFS reduces the
response of the store considerably. The wing response is not attenuated as much. This is due to the fact that the
wing mainly responds in its bending mode. Very little damping force can be introduced into this mode at the wing
pylon station.

6. AFS APPLICATION WIND-TUNNEL TEST RESULTS

In Figure 9 the measured damping versus velocity is plotted for different phase angles. This picture shows that
about 50% increase on flutter speed can be gained by the AFS-system. It is also shown that the 0° phase is nearly
optimal.

This could be expected, because there is little phase shift at a frequency of 8 Hz between unsteady aerodynamic
force and angle of attack on the vane. Figure 10 shows the measured damping versus velocity for Vi the standard
vane area for different phase angles. The aerodynamic damping produced by the AFS (0° phase) is about halved.
Figure 1 1 depicts the damping behaviour for a gain of the AFS of two times nominal and nominal with vane B
(see Figure 2). All these measurements were performed with a wing leading-edge sweep angle of \<^Q = 25° .

7. ANALYSIS RESULTS AND COMPARISON WITH TEST

All analyses were performed with zero structural damping for a wing leading-edge sweep angle A = 45° .

In Figure 1 2 a comparison of measured and calculated damping values for a configuration with A\VQ = 45° is
given. The calculated damping values are taken from v — g plots, Figure 13 and Figure 15. Two percent of
structural damping was added to mode 2 damping, because the drag force is creating additional damping when the
model is supported on its rod in the tunnel. Considering the complicated flutter mechanism, correlation of test and
analysis is very good. The analysis underestimates the tunnel flutter speed only by about ten percent (AFS off)
and gives the same damping trend (AFS off and on).
 63

Figure 13 and Figure 14 show the calculated damping and frequency versus velocity for Ma = 0.2 and
Ma = 0.9 . These two figures demonstrate that there is no pronounced Mach number effect existent, so that for
the external store configuration used, the wind-tunnel tests would be representative for the full-scale aircraft.
In both figures the relatively mild flutter of mode 2 and the strong flutter of mode 3 can be seen. The v — g
plots for AFS on for Ma = 0.2 and Ma = 0.9 are shown in Figure 15 and Figure 16. Both flutter modes are
damped now. The bending mode 1 is very little affected by the AFS. Flutter speed versus gain K of the AFS
is depicted in Figure 17. Here it can be seen that for a specific gain setting (about 75% of nominal) there is no
more flutter for mode 2. Flutter speed of mode 3 is increased by increasing the gain. For both modes 2%
structural damping was considered.

Figure 18 and Figure 19 shows root loci for a variation of K and A*p respectively. From Figure 18 one can
see that mode 2 and mode 3 are strongly affected by AFS, whereas mode 1 is almost unaffected. Varying the phase
*p in Figure 19 shows that mode I damping can also be changed with varying phase of AFS. Figure 20 shows an
application of AFE. For evaluation of dampings of mode 2 and mode 3, different sensors would be necessary.

Figure 21 shows, in comparison with Figure 15, that there would be no difference in effectiveness of AFS if
the vane and sensor are located at the store rear or forward end. This result stems from using linear aerodynamic
effects. There are indications from test results, that there could be a pronounced effect, depending upon whether
the vane was located in the undisturbed flow stream at the store forward end or at the relatively disturbed flow at
the store rear end.

Figure 22 shows the v — g plot for the classical wing bending/store pitch flutter problem by deleting the wing
pivot yaw mode.

In Figure 23 the damping and frequency with the AFS on is plotted. Figure 24 shows the effect of AFE.
These three figures are only given to show that AFE could be applied directly, without separating different modes
by different sensors for fixed geometry airplanes.

8. CONCLUSIONS

It was shown, that a relatively simple control system could be developed and tested on a wind-tunnel model
to suppress flutter.

The main reason that the AFS is so effective with relatively small vanes that do not change the flight mechanical
characteristics is that mild flutter, very susceptible to small damping changes, occurs when large masses are involved
in the flutter phenomenon.

This system was also very valuable in exciting model modes. Very interesting aspects of extension of existing
flutter testing techniques were illuminated.

REFERENCES

1. Triplett, W.E. A Feasibility Study of Active Wing/Store Flutter Control. J. Aircraft, June 1972.

2. Triplett, W.E., Active Flutter Control. 11th AIAA Meeting, Washington, D.C., January 1973.
et al.

3. Seidel, W. Flatteranalyse im frithen Entwurfsstatium mit Hilfe eines Unterschallflattermodells

und parallellaufender Rechnungen. Jahrbuch 1967 der WGLR.

4. Otnes R.K., Digital Times Series Analysis. John Wiley & Sons, London.
Enochson, L.
64

f2 • 6,92 Hz Sym,

f3 • 8,0 Hz Sym.

Fig.l Measured modes with inboard tanks Ayy = 45°

 65

i i

--ZZ--M.-—-Z
_E

Vane Surface A
(Standard)
A=20 cm2

Fig.2 Vane-tank

I
c
o
"5
c
Ol
CO
I I
I I
I I JDPSVV
v

KKS Power Amplif.

Fig.3 Block diagram of the vane control system

66

1*1X CO
800

600

AOO _p__~-^0

200

0 U 8 12 16 20 2i
Frequency f [Hz]
A(p

120

GO
^L|-
ya ~
-60

Fig.4 Transferfunction F, x ico

792

Re(z) [ c m /sec]

795

Fig.S Vector plot at V/yRef = 1 1 1 A w i n „ = 45° pickup and vane exciter in rear tank position
 67

frequency sweep 30Hz - 3Hz- 30Hz i n 100 sec

mmmmmmmmmmm
Q.l vane

0Dmmr-MtlMmm9 | ,m, V p.._. «..•..•-•«• * » « * » ' II.

left vane tank fwd.

m**s**mmtm0*>*>**'*t> • mi»»>i'» »»»w——"<_• "

right vane tank fwd

«... .>._.w»»4»<irt*f^+_)^ii.t»w«ii»«^ ' f*w— i <««^HHW>w'MW|iHt *wi> Mi*"'' •*>* **• M't<>""^H^
left wing tip

. . . . u . . .

>MM»M^^|^><i_|>f_f^(<>»_i>)_i,i-i«i»iii i i i w i M i j f ^ t m m , ^ , ^ . ^ ,,i »nnw»j|>i-t#4* < f'W»^* '»**"' *"« *."• »^ *»"i "*i>
»' r right wing tip S

Fig.6 Model responses Z to frequency-sweep system excitation with tank vanes WG = 25°V/vRef = 0.95

Off AFS o n

on AFE off With AFS

+ •«
E
ra
O

AFS off

Fig.7 Application of AFE and AFS for damping measurements (MBB patent application P 23493541)
68

Gyy(f)
Response Z (cm/sec ) Power Spectral Density
G
Ret

0,5
AFS off 0

0 I 5 m 15 20 25 [Hz} :

Left Vane Tank

iqs
AFS on 0

[ ! I" ' '! ^ » — . 1 1 j

:
0 I i9 ' 10 • 15 • • :: 2J0- 25 [ Hi]

AFS off 0

AFS on 0

I M Oil ij 5 10 15: 20 25 THzl

Fig.S Model response at 12° angle of attack AFS off and on V/yR e f = 0.77

2,0
V/
VR

AFS on
AFS Dff
^y> _^-Q1

f
1,0 Flutter 6,6 Hi
Mode .>

Variatio n A(p

0 -ICP
< 0°
0 + 10°
D +20°
+40°

10 8 6 U 2 0 -2 -U -6 -8
Damping g(%)

Fig.9 Measured damping versus velocity, A W Q = 25° , sensor Zj , vane area = standard
 69

2,0

V/
VR

1
AFS on AFS off

/ / $ - ^ c < ^
y 1

1 '
1,0
'Flutter 6.6Hz

Mode 2

1 "
Variation A(jp
0 -10°

0 0°
a * 20°
o •40°

10 8 5 U 2 0 - 2 - 4 -6
Damping 9 (*l

Fig. 10 Measured damping versus velocity, A ^ G = 25° , sensor z, , vane area = 0.5 standard

2,0
V
'VR
i
Senso r 'it O • (Gain Nominal)
_i
• • (Gain 2x Nominal)

AFS on y i rf AFS Off

iA
cr
if
r
'Flutter 6. . Hz Mode 2

'.0

16 12 6 4 0 -4 -8 -12 -16
Damping (g%)

Fig.l 1 Measured damping versus velocity, A^n = 25° , vane area = 0.54 of standard, (surface B)
70

2.0
<

J
i ^w

Mode 2 ^
'plnttpr
6.68 Hz

/ !_____

s
JS

AFS
<yr AFS off

N
on J
(
\ /
1,0

O • TEST

A A ANALYSIS

16 12 8 4 0 -4 -8 -12 -16
Damping g (*)

Fig.l 2 Comparison of measured and calculated damping versus velocity A ^ Q = 45°

 71

• 1
\ \
1. ] ;.

/ '
j f

20 16 12 8 4 -4 -8 -12 -16 0 2 4 6 8 10
Damping g I */• 1 Frequency f I Hz 1

Fig. 13 Damping and frequency versus velocity for Ma = 0.2 (AFS off), AWG = 45°

H /
z,u -

1
Ar
1
j
l,U 1

1 1

20 16 12 8 4 B -12 -16
1) -
0 2 4
1 —1—* —
6 8 10
Damping g 1%) Frequency f I Hz)

=
Fig. 14 Damping and frequency versus velocity for Ma = 0.9 (AFS off), hy/G ^5°
7:

2,0
1
\!

J
in

/
1,0

1
0 2 4 6 8 10 12
Frequency f I Hz 1

Fig. 15 Damping and frequency versus velocity for Ma = 0.2 with AFS (K = 100%, A*p = 0°), A WG = 45

20 16 12 8 4 4 6 8
Damping g I %l Frequency f (Hz)

Fig. 16 Damping and frequency versus velocity for Ma = 0.9 with AFS (K = 100%, A*p = 0°), A W Q = 45
 7.?

V_ 3,0 1
1

\ mode 3
fc
2,0

x
t-XJ mode 2
1.0
1

O Ma = 0,2
A Ma = 0,9

0 25 50 75 100
K (% of nominal 1

Fig. 17 Flutter speed versus K for different Mach numbers, A ^ G = 45°

Jm (p

o o
S~ o-
8_ in o
Y, o- *> 60
mode A

507. 07.
K=1007.
— A -
'*- 50- -mode 3
K-1007.
507. 07.
K=100V..
507 >?0 7 .
•757o
757. A0 ^rnode 2
mode 1
50

Ma = 0,2 20
V/V R = 1,36
*f = 0° -10

1— -r-
6 •5 -A -1 0 2 Re (p

Fig. 18 Root locus of the coupled system with variation of K

74

Jm (P;
70
mode A
•90°
&f- -30°
0° +30° 60
mode 3
90
50
30°
•90°
<df = • 3 0 ° mode 2
40
-90° +
330°
mode 1
30

Ma = 0,2 - 20
V/V R = 1,36
K 100 % <0

12 -10 -8 k Re ( p )

Fig. 19 Root locus of the coupled system with variation of A*p

2.0

v
VD

1,0 •

20 16 12 6 4 12 -16 2 4 6 8 10 12
Damping g 17.1 Frequency f (Hz)

Fig. 20 Damping and frequency versus velocity for Ma = 0.2 with AFE (K = 100%. A*p = 180°), A W G = 45°
 75

28 24 20 16 12 2 4 5 8 10 12
Damping g (%) Frequency f IHzl

Fig. 21 Damping and frequency versus velocity for Ma = 0.2 with AFS (K = 100%, A*fi = 0°), A ^ G = 45° ,
sensor and vane location: tank rear station

2,0
1

A
II i 1 :

:y*s 11

I
(F
<
1.0

i) I11 »
20 16 12 8 4 -8 -12 0 2 4 6 8 10 12
Damping g (%l Frequency f (Hzl

Fig.22 Damping and frequency versus velocity for Ma = 0.2 (AFS off), A ^ G = 45°, wing pivot yaw stiffness rigid
76

r .

r
r i

JL V
:

^ ^ ^
, i .
• ^

- ^

28 24 20 16 12 0 2 - 6 8 10 12
Damping g 17.1 Frequency f IHzl

=
Fig.23 Damping and frequency versus velocity for Ma = 0.2 with AFS (K = 100%, A*p = 0°), Ay^c 45°
wing pivot yaw stiffness rigid

r
1
i

;: 1

;
\

A
'

20 16 12 8 4 0 -4 -8 -12 2 4 6 8 10 12
Damping g 17.1 Frequency f 1Hz)

Fig.24 Damping and frequency versus velocity for Ma = 0.2 with AFE (K = 100%, Aip = 180°), A ^ G = 45° ,
wing pivot yaw stiffness rigid
AGARDograph No. 175 AGARD-AG-175 AGARDograph No. 175 AGARD-AG-175
Advisory Group for Aerospace Research and 629.73.062 - 52 : Advisory Group for Aerospace Research and 629.73.062 - 52 :
Development, NATO 533.6.013.422 : Development, NATO 533.6.013.422 :
ACTIVE CONTROL SYSTEMS FOR LOAD 533.6.048.1 ACTIVE CONTROL SYSTEMS FOR LOAD 533.6.048.1
ALLEVIATION, FLUTTER SUPPRESSION AND ALLEVIATION, FLUTTER SUPPRESSION AND
RIDE CONTROL Aerodynamic loads RIDE CONTROL Aerodynamic loads
Published March 1974 Flutter Published March 1974 Flutter
84 pages Loads (forces) 84 Pages Loads (forces)
Flight maneuvers Flight maneuvers
Active Control Systems offer potential to: reduce Flight control Active Control Systems offer potential to: reduce Flight control
structural loads encountered in maneuvers; improve Control equipment structural loads encountered in maneuvers; improve Control equipment
stability; reduce the size of control surfaces; reduce stability; reduce the size of control surfaces; reduce
static design loads; increase flutter speeds and suppress static design loads; increase flutter speeds and suppress
flutter, thereby reducing stiffness or mass required flutter, thereby reducing stiffness or mass required
of members; reduce the margin between normal of members; reduce the margin between normal

P.T.O. P.T.O.

AGARDograph No.l 75 AGARD-AG-175 AGARDograph No. 175 AGARD-AG-175

Advisory Group for Aerospace Research and 629.73.062 - 52 : Advisory Group for Aerospace Research and 629.73.062 - 52 :
Development, NATO 533.6.013.422 : Development, NATO 533.6.013.422 :
ACTIVE CONTROL SYSTEMS FOR LOAD 533.6.048.1 ACTIVE CONTROL SYSTEMS FOR LOAD 533.6.048.1
ALLEVIATION, FLUTTER SUPPRESSION AND ALLEVIATION, FLUTTER SUPPRESSION AND
RIDE CONTROL Aerodynamic loads RIDE CONTROL Aerodynamic loads
Published March 1974 Flutter Published March 1974 Flutter
84 pages Loads (forces) 84 pages Loads (forces)
Flight maneuvers Flight maneuvers
Active Control Systems offer potential to: reduce Flight control Active Control Systems offer potential to: reduce Flight control
structural loads encountered in maneuvers; improve Control equipment structural loads encountered in maneuvers; improve Control equipment
stability; reduce the size of control surfaces; reduce stability; reduce the size of control surfaces; reduce
static design loads; increase flutter speeds and suppress static design loads; increase flutter speeds and suppress
flutter, thereby reducing stiffness or mass required flutter, thereby reducing stiffness or mass required
of members; reduce the margin between normal of members; reduce the margin between normal

P.T.O. P.T.O.
operating speed and design diving speed due to upsets, wind shears, temperature operating speed and design diving speed due to upsets, wind shears, temperature
gradients, etc., improve ride control; and reduce stores vibration. Essential to gradients, etc., improve ride control; and reduce stores vibration. Essential to
adequate solution of the problem of effective employment of active control adequate solution of the problem of effective employment of active control
devices is the development of systems to detect and counteract disturbing loads devices is the development of systems to detect and counteract disturbing loads
by means of properly phased control forces produced by reliable autostabilizing by means of properly phased control forces produced by reliable autostabilizing
systems. Recent research work and specific applications of active control devices systems. Recent research work and specific applications of active control devices
are dealt with in the five papers that comprise this AGARDograph. are dealt with in the five papers that comprise this AGARDograph.

This AGARDograph was prepared at the request of the Structures and Materials This AGARDograph was prepared at the request of the Structures and Materials
Panel of AGARD. Panel of AGARD.

operating speed and design diving speed due to upsets, wind shears, temperature operating speed and design diving speed due to upsets, wind shears, temperature
gradients, etc., improve ride control; and reduce stores vibration. Essential to gradients, etc., improve ride control; and reduce stores vibration. Essential to
adequate solution of the problem of effective employment of active control adequate solution of the problem of effective employment of active control
devices is the development of systems to detect and counteract disturbing loads devices is the development of systems to detect and counteract disturbing loads
by means of properly phased control forces produced by reliable autostabilizing by means of properly phased control forces produced by reliable autostabilizing
systems. Recent research work and specific applications of active control devices systems. Recent research work and specific applications of active control devices
are dealt with in the five papers that comprise this AGARDograph. are dealt with in the five papers that comprise this AGARDograph.

This AGARDograph was prepared at the request of the Structures and Materials This AGARDograph was prepared at the request of the Structures and Materials
Panel of AGARD. Panel of AGARD.
 DISTRIBUTION OF UNCLASSIFIED AGARD PUBLICATIONS

NOTE: Initial distributions of AGARD unclassified publications are made to NATO Member Nations through the following National
Distribution Centres. Further copies are sometimes available from these Centres, but if not may be purchased in Microfiche
or photocopy form from the Purchase Agencies listed below. THE UNITED STATES NATIONAL DISTRIBUTION
CENTRE (NASA) DOES NOT HOLD STOCKS OF AGARD PUBLICATIONS, AND APPLICATIONS FOR
FURTHER COPIES SHOULD BE MADE DIRECT TO THE APPROPRIATE PURCHASE AGENCY (NTIS).

NATIONAL DISTRIBUTION CENTRES

BELGIUM ITALY
Coordonnateur AGARD - VSL Aeronautica Militare
Etat-Major de la Force Aerienne Ufficio del Delegato Nazionale all'AGARD
Caserne Prince Baudouin 3. Piazzale Adenauer
Place Dailly, 1030 Bruxelles Roma/EUR
CANADA LUXEMBOURG
Defence Scientific Information Service See Belgium
Defence Research Board NETHERLANDS
Department of National Defence Netherlands Delegation to AGARD
Ottawa, Ontario K1A OZ3 National Aerospace Laboratory, NLR
DENMARK P.O. Box 126
Danish Defence Research Board Delft
0sterbrogades Kaserne NORWAY
Copenhagen 0 Norwegian Defence Research Establishment
FRANCE Main Library
O.N.E.R.A. (Direction) P.O. Box 25
29, Avenue de la Division Leclerc N-2007 Kjeller
92, Chatillon sous Bagneux PORTUGAL
GERMANY Direccao do Servico de Material
Zentralstelle fur Luftfahrtdokumentation da Forca Acrea
und Information Rua de Escola Politecnica 42
Maria-Theresia Str. 21 Lisboa
8 Munchen 27 Attn: AGARD National Delegate
GREECE TURKEY
Hellenic Armed Forces Command Turkish General Staff (ARGE)
D Branch, Athens Ankara
ICELAND UNITED KINGDOM
Director of Aviation Defence Research Information Centre
c/o Flugrad Station Square House
Reykjavik St. Mary Cray
Orpington, Kent BRS 3RE
UNITED STATES
National Aeronautics and Space Administration (NASA)
Langley Field, Virginia 23365
Attn: Report Distribution and Storage Unit
(See Note above)

PURCHASE AGENCIES
Microfiche or Photocopy Microfiche Microfiche
National Technical ESRO/ELDO Space Technology Reports
Information Service (NTIS) Documentation Service Centre (DTI)
5285 Port Royal Road European Space Slation Square House
Springfield Research Organization St. Mary Cray
Virginia 22151, USA 114, Avenue Charles de Gaulle Orpington, Kent BRS 3RE
92200 Neuilly sur Seine, France England

Requests for microfiche or photocopies of AGARD documents should include the AGARD serial number, title, author or editor, and
publication date. Requests to NTIS should include the NASA accession report number.

Full bibliographical references and abstracts of AGARD publications are given in the following bi-monthly abstract journals:

Scientific and Technical Aerospace Reports (STAR), Government Reports Announcements (GRA),
published by NASA, published by the National Technical
Scientific and Technical Information Facility Information Services, Springfield
P.O. Box 33, College Park Virginia 22151, USA
Maryland 20740, USA

*
Printed by Technical Editing and Reproduction Ltd
Harford House, 7-9 Charlotte St, London. W1P IHD