You are on page 1of 15

Aircraft Engineering and Aerospace Technology

Flight flutter testing and aeroelastic stability of aircraft


Altan Kayran

Article information:

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

To cite this document:


Altan Kayran, (2007),"Flight flutter testing and aeroelastic stability of aircraft", Aircraft Engineering and Aerospace Technology,
Vol. 79 Iss 2 pp. 150 - 162
Permanent link to this document:
http://dx.doi.org/10.1108/00022660710732707
Downloaded on: 21 February 2016, At: 05:39 (PT)
References: this document contains references to 33 other documents.
To copy this document: permissions@emeraldinsight.com
The fulltext of this document has been downloaded 792 times since 2007*

Users who downloaded this article also downloaded:


(2007),"Erratum", Aircraft Engineering and Aerospace Technology, Vol. 79 Iss 5 pp. 494-506 http://
dx.doi.org/10.1108/00022660710780623
Achuthan C. Pankaj, G. Shanthini, M.V. Shivaprasad, M. Manjuprasad, (2013),"Aircraft flutter prediction using
experimental modal parameters", Aircraft Engineering and Aerospace Technology, Vol. 85 Iss 2 pp. 87-96 http://
dx.doi.org/10.1108/00022661311302698
W.J. Duncan, (1945),"The Fundamentals of Flutter", Aircraft Engineering and Aerospace Technology, Vol. 17 Iss 1 pp. 16-20
http://dx.doi.org/10.1108/eb031209

Access to this document was granted through an Emerald subscription provided by emerald-srm:427172 []

For Authors
If you would like to write for this, or any other Emerald publication, then please use our Emerald for Authors service
information about how to choose which publication to write for and submission guidelines are available for all. Please visit
www.emeraldinsight.com/authors for more information.

About Emerald www.emeraldinsight.com


Emerald is a global publisher linking research and practice to the benefit of society. The company manages a portfolio of
more than 290 journals and over 2,350 books and book series volumes, as well as providing an extensive range of online
products and additional customer resources and services.
Emerald is both COUNTER 4 and TRANSFER compliant. The organization is a partner of the Committee on Publication Ethics
(COPE) and also works with Portico and the LOCKSS initiative for digital archive preservation.
*Related content and download information correct at time of download.

Flight flutter testing and aeroelastic


stability of aircraft
Altan Kayran

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Department of Aerospace Engineering, Middle East Technical University, Ankara, Turkey


Abstract
Purpose To provide a general review of the flight flutter test techniques utilized in aeroelastic stability flight testing of aircraft, and to highlight the
key items involved in flight flutter testing of aircraft, by emphasizing all the main information processed during the flutter stability verification based on
flight test data.
Design/methodology/approach Flight flutter test requirements are first reviewed by referencing the relevant civil and military specifications.
Excitation systems utilized in flight flutter testing are overviewed by stating the relative advantages and disadvantages of each technique. Flight test
procedures followed in a typical flutter flight testing is described for different air speed regimes. Modal estimation methods, both in frequency and time
domain, used in flutter prediction are surveyed. Most common flight flutter prediction methods are reviewed. Finally, key considerations for successful
flight flutter testing are noted by referencing the related literature.
Findings Online, real time monitoring of flutter stability during flight testing is very crucial, if the flutter character is not known a priori. Techniques
such as modal filtering can be used to uncouple response measurements to produce simplified single degree of freedom responses, which could then be
analyzed with less sophisticated algorithms that are more able to run in real time. Frequency domain subspace identification methods combined with
time-frequency multiscale wavelet techniques are considered as the most promising modal estimation algorithms to be used in flight flutter testing.
Practical implications This study gives concise but relevant information on the flight flutter stability verification of aircraft to the practicing
engineer. The three important steps used in flight flutter testing; structural excitation, structural response measurement and stability prediction are
introduced by presenting different techniques for each of the three important steps. Emphasis has been given to the practical advantages and
disadvantages of each technique.
Originality/value This paper offers a brief practical guide to all key items involved in flight flutter stability verification of aircraft.
Keywords Aerodynamics, Flight performance, Air safety
Paper type General review

flight test speeds. During the flight flutter tests, aircraft is flied
at a range of sub critical speeds and some form of excitation is
applied to the aircraft. The response of the structure is
measured at a number of locations on the aircraft, and the
data obtained is used to determine the stability at the current
test point. Predictions are then made whether it is safe to
proceed to the next test point. This process is repeated at
various flight conditions and aircraft weight, and CG
configurations until the flight envelope is cleared. Common
excitation means used in practice include control surface
pulses, oscillating control surfaces, thrusters, inertia exciters,
aerodynamic vanes, and random atmospheric turbulence
(Kehoe, 1995; Brenner et al., 1997; Meijer, 1995; van Nunen
and Piazzoli, 1979). It is very important that the excitation
system must provide adequate excitation over the desired
frequency range, and must provide adequate force levels to
ensure accurate determination of stability parameter. In
practice, the most common response data analysis procedure
is to track the variation of the modal damping of the aircraft
structure with the air speed. Some examples of other flutter
prediction methods include flutter margin (Zimmerman and
Weissenburger, 1964), envelope function (Cooper et al.,
1993; Nissim and Gilyard, 1989; Nissim, 1993), eigenvector
orientations (Pidaparti et al., 2001). Cooper (1995) gives an
overview of the several modal estimation algorithms, both in
time and frequency domain for extracting stability estimates
and detecting time-varying and nonlinear dynamics.
In the present paper, the effect of different excitation
systems, modal parameter estimation techniques on the
accuracy of the processed flight test data, and other special

Introduction
Aeroelastic flutter is the complex interaction of aerodynamic,
elastic, and inertia forces producing an unstable, usually
divergent oscillation of the aircraft structure or the
component (Fung, 1993). Airworthiness regulations require
that stability within the flight envelope of the aircraft be
demonstrated by flight flutter tests (MIL-A-8870C, 1993). In
spite of the improvements that have taken place in the flutter
test techniques, instrumentation, and response data analysis,
flutter testing is still a risky test for several reasons. One
reason for the risk associated with flutter testing is that sub
critical damping trends cannot be accurately extrapolated to
predict stability at higher airspeeds. In addition, aeroelastic
stability may change suddenly from a stable condition to one
that is unstable with only few differences in airspeed. Thus,
one must fly as close as possible to actual flutter speeds to
detect the instabilities accurately. Therefore, a careful
expansion of the flight envelope is required to avoid possible
hazardous situations. For this purpose, a large number of
flight tests are usually performed with careful increases in
The current issue and full text archive of this journal is available at
www.emeraldinsight.com/1748-8842.htm

Aircraft Engineering and Aerospace Technology: An International Journal


79/2 (2007) 150 162
q Emerald Group Publishing Limited [ISSN 1748-8842]
[DOI 10.1108/00022660710732707]

150

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

Excitation systems

considerations to achieve a successful flight flutter test are


highlighted. Reasons for the high degree of scatter in damping
experienced in flight flutter tests are discussed, and
recommendations towards decreasing the scatter are given.
Reasons for the very common occurrence of non-repeatability
at identical test locations, and its implication on the test
results, and recommendations on improving repeatability are
also elaborated for the success of flight flutter test. Finally,
under the light of the discussions presented in the paper,
concluding remarks, with regard to the needs for flight testbased accurate/safe flutter stability determination, are made.

One of the very crucial items in executing a successful flight


flutter test is the use of the right excitation system for the
particular aircraft to be flight tested. Flight flutter testing
consists of three steps, which are structural excitation,
structural response measurement, and stability analysis.
Without adequate excitation, detection of impending
aeroelastic instabilities cannot be made. The airworthiness
specifications state that the test airplane shall be equipped
with an excitation system that is capable of exciting all
structural vibration modes which contribute to the various
flutter critical conditions. Adequate excitation provides energy
to excite all of the selected vibration modes with sufficient
magnitudes to accurately assess stability from the response
data. The significance of adequate excitation levels is pointed
out by Kehoe (1995), by referencing the flutter test
experience on the Transvia PL12/T-400 airplane. This
airplane was excited on the initial flight tests by control
surface pulses and random atmospheric turbulence, and
flutter incident did not occur during the flight test. In a
subsequent flight, the airplane experienced limited amplitude
flutter of the rudder and tail boom at a frequency around
3.5 Hz, when the airplane was flown in rough weather
condition. Apparently, the weather condition provided higher
levels of excitation than the levels induced in the early flight
flutter test. In subsequent tests, the test pilot found that he
could induce a similar oscillation by applying more than 50
percent rudder actuation in a series of quick jerks in alternate
directions.
The key requirements from the excitation system are
(Kehoe, 1995):
.
adequate force level levels;
.
adequate excitation over the desired frequency range of
interest;
.
lightweight so as not to affect the modal characteristics of
the airplane; and
.
to have power requirements that the airplane can meet.

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Airworthiness requirements
Airworthiness requirements applied to aircraft are mainly
referenced to MIL-A-8870C (1993), and federal airworthiness
requirements, for different category aircraft such as FAR 25
(2000) and JAR-25 (2004). All these documents give a very
broad perspective of the aeroelastic stability requirements. In
essence, according to the specifications, all configurations of the
aircraft shall be free from any aeroelastic instability for all
combinations of altitude and speed encompassed by the limit
speed (VL/ML) versus altitude envelope enlarged at all points by
the airspeed margin of safety. Limit speed is commonly taken as
the dive speed of the aircraft. According to paragraph 3 of MILA-8870C, the aircraft shall meet the following stability design
requirements for both normal and failure conditions.
Airspeed margin
The equivalent airspeed, VE, margin of safety shall not be less
than 15 percent at all points on the VL/ML envelope of the
airplane, both at constant Mach number, and separately, at
constant altitude. This situation is shown in Figure 1.
Damping
The total aerodynamic plus structural damping coefficient g
shall not be less than 3 percent (g 0.03) for any critical
flutter mode for all altitudes and flight speeds from minimum
cruising speed up to VL/ML. Minimum damping requirement
is shown in Figure 1.

In the following, some of the common excitation systems will


be reviewed by emphasizing the advantages and disadvantages
of these systems.

Figure 1 Flutter requirements

151

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

Control surface pulses


The simplest way to apply an aerodynamic force to an aircraft
is by means of controlling surfaces. Sudden displacements
applied to the control surfaces will result in aerodynamic
forces that are of short duration. With the control surface
pulses, depending on the type of the control system, modes up
to 10 Hz can be excited this way (Kehoe, 1995; van Nunen
and Piazzoli, 1979). The response of the aircraft will be of
decaying oscillation, and the analysis of the resonance
frequency and the corresponding damping coefficient will be
simple. Symmetric and anti-symmetric modes of an aircraft
can also be excited by proper selection of the control surfaces.
For instance, symmetric modes can be excited by the
symmetrically operated elevators, whereas anti-symmetrically
operated ailerons are usually preferred to excite the antisymmetric modes. Two main advantages of this method are
that no special excitation equipment is required, and the
transient response of the structure is easy to analyze for
stability. Because test duration is short, many can be applied
at a test point. However, there are major drawbacks of this
method. It is difficult to get repeatable pulses, because the
impulsive control surface force is usually generated by the
pilot. Tests performed with different input signals, and
the very common occurrence of non-repeatability at identical
test conditions, can change damping estimates significantly. In
addition, pilot may not provide a sharp enough input
disturbance to excite any critical flutter modes above 10 Hz,
and this by itself restricts the use of this method in situations
where the critical flutter frequency turns out to be above
10 Hz. Energy input to the aircraft structure may not be of
adequate level to determine the onset of flutter. One reason
for the inadequate level of energy input might be due to the
location of the control surfaces coinciding with nodal lines of
certain critical modes. In such cases, the response will contain
very little information on the mode under concern.
In spite of the limitations, this technique is still used in
practice due to its affordability in certain class of aircraft, such
as small aircraft and sailplanes. Another application may be
the investigation of the effects of small modifications to the
aircraft, as is often the case in the flight flutter testing of
external store modifications. Some examples of the
application of the control surface pulses include the low
speed testing of the DC-10 airplane, some of the early Boeing
747 flutter testing, early testing of F-4 aircraft (Kehoe, 1995).

and response signals available, one can use cross spectral


techniques for calculating the resonance frequencies and
corresponding damping coefficients. Direct activation of the
servos may also lead to larger aerodynamic forces, and thus to
responses that are less affected by noise. Frequencies up
to 35 Hz have been reported to be excited by this system
(van Nunen and Piazzoli, 1979). A disadvantage of this
method is the frequency response limitations of the control
surface actuators. Sometimes actuators for the control
surfaces are modified to provide the required gain and
frequency response characteristics. As was the case for the F-4
flutter testing, often special actuators are required to excite
critical high frequency modes (Meany, 1983). This method of
excitation has been successfully used for the X-31 and YF-22
airplanes, and for the stores clearance work on the F-16,
F-15, and F-18 airplanes (Kehoe, 1995; van Nunen and
Piazzoli, 1979).
Oscillating vanes
Oscillating vane technique is another technique which uses
aerodynamic loading to generate the required excitation. In
principle, such a vane consists of a small airfoil protruding
from the wing or the tail section. Since, the vane is not a
standard part of the aircraft, its location can be chosen in
accordance with the vibration modes of the aircraft. Favorable
locations to install vanes are wing tips, horizontal or vertical
stabilizer tips. Figure 2 shows such vane installations on the
aft fuselage and the wing tip.
The vanes are generally positioned on a shaft which is
oscillated either electrically or hydraulically. The oscillation of
the vane will result in a varying aerodynamic force on the
aircraft structure. The amount of force depends on the size of
the vane, dynamic pressure, and angle of rotation. The most
obvious advantage of this method is the control of the exciting
force is rather simple, and excitation frequency and amplitude
at a given air speed can be controlled making it possible
to generate repeatable excitation. Structural frequencies of
40-50 Hz can be excited by this system. This frequency range
covers almost all of the important resonance frequencies for
most aircraft. Since, vanes are additions to the normal aircraft
structure, their positions can be selected such that the
oscillatory force can act as effectively as possible. A
disadvantage of such an excitation system is the effect it
imposes on the modal characteristics of the aircraft structure
due to the added mass to the aircraft. In addition, the normal
airflow around the region where the vane is installed is
disturbed causing a change in the aerodynamic flow
characteristics of the actual aircraft. Large power
requirement usually needed to operate such a system is also
a disadvantage of this system. Among some of the flight test
programs which utilized oscillating vanes are DC-10, Boeing
747, 757, F-14, F-11, A-10, C-17 (Rosenbaum, 1974).

Oscillating control surfaces


An extension of the application of the control surface as a
means to excite an aircraft is through the use of commanded
oscillations of the control surfaces. In this case, it is no longer
necessary for the pilot to induce the aerodynamic force. By
means of electronic function generators, signals are provided
to the control surface servos in the autopilot system. In this
application a cockpit controller is needed to adjust the
excitation amplitude, mode selection such as dwell or sweep
and their related frequency. Advances in the electronics made
it possible to generate other commands, such as bandpassfiltered random commands besides sinusoidal signals to the
flight control surfaces. An attractive point in the oscillating
control surfaces technique is that no additional hardware is
required except for the control box in the cockpit. Thus,
vibrational behaviour of the aircraft will not be disturbed.
Another advantage of the method is the good control one has
over the applied force, or the input signal. Having both input

Random atmospheric turbulence


A simple method of excitation is to make use of free air
turbulence. The aircraft will be excited in a very nearly
random manner when flying through turbulent air. The main
advantage of using air turbulence is that no special onboard
exciter hardware is needed. In addition, turbulence excites
both symmetric and anti-symmetric modes simultaneously,
eliminating the need to perform symmetric and antisymmetric sweeps. Although this method is very simple to
use, there are several drawbacks. The energy contained
152

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Figure 2 Examples of aerodynamics vanes

are difficult to calculate accurately. Inadequate excitation


levels are also responsible for large scatter of the estimated
damping values, and estimated damping values often indicate
lower damping than actually exists. Flutter envelopes of many
modern aircraft such as X-29A, advanced fighter technology
integration F-16, F-15 short take-off, and landing maneuver
technology demonstrator have been cleared by atmospheric
turbulence (Kehoe, 1995).

in atmospheric turbulence has been found to be concentrated


at the low side of the frequency range, and decreases
exponentially at high frequencies. Therefore, turbulence
usually excites only lower frequency modes for most
airplanes. Another disadvantage of the method is that long
data records are needed to obtain results with a sufficiently
high statistical confidence level. An improvement to this
method can be obtained by applying random decrement
analysis procedure which was used to clear the flutter
envelope of basic YF-16 (Kehoe, 1995). Random decrement
method has been used to analyze response signal from
randomly excited systems (Cole, 1973). The principles of this
technique is shown in Figure 3.
In the random decrement method, within the response
signal a selection level is chosen and each time the signal
crosses this level, it is shifted towards a common reference
time. By adding all the successive partial time signals, finally
results in a signal resembling the free decay signal from which
the resonance frequency and damping coefficient can be
determined. However, the disadvantage of the random
turbulence method cannot be totally eliminated by the use
of random decrement method. Long time histories are still
required for accurate prediction of stability parameters.
In addition, turbulence may not be of required intensity to
produce sufficient excitation when compared to excitation
with on-board exciters. At lower altitudes, turbulence is more
readily found, and the lack of adequate excitation becomes
less of a problem. At lower altitudes, potentially critical areas
of the flight envelope are cleared. Figure 4 shows the power
spectrum density plots obtained from the F-16XL airplane
excited by random atmospheric turbulence, and by a vane
with a slotted rotating cylinder attached to the trailing edge
(Vernon, 1993).
At the flight conditions turbulence was reported to be light
to moderate by the pilot. Figure 4 shows that atmospheric
turbulence could only excite the 8 Hz mode well. On the other
hand, all the expected structural modes were excited by the
vane. In Figure 4, the width of the peaks, which is
proportional to structural damping, is clearly smaller than
for the forced excitation data. This is attributed to the poor
signal-to-noise ratio in the amplitudes for the modes excited
by atmospheric turbulence. The modes are not well excited,
and are contaminated by noise; therefore, the damping levels

Inertia exciters
Inertia exciters work on the principle of generating excitation
forces by rotating eccentric weight and oscillating weight.
In the rotating eccentric mass excitation system, inertia force
is proportional to the exciter mass multiplied by the square of
the rotating speed. Therefore, the excitation capability of
these exciters may be limited at low frequencies, and excessive
at higher frequencies. The main advantage of these systems is
that their design is simple and several exciters, each producing
different excitation forces can be distributed throughout the
aircraft. Each of these exciters can be controlled individually
allowing to generate specific types of excitation such as
symmetrical or anti-symmetrical. Frequencies up to 65 Hz
can be excited by these systems. Frequency sweeps as well as
frequency dwell followed by abrupt stopping can be applied in
flight flutter tests. It is reasonably simple to determine the
force or a signal proportional to force exerted on the aircraft
structure, cross-spectral techniques can be employed for
calculating resonance frequencies and corresponding
damping coefficients. However, these systems are usually
heavy and they raise the concern about the effect on the
modal characteristics of the aircraft they are installed on.
A further problem may be the generation of sufficiently high
exciting force at low frequencies. Inertia exciter system
adequately excited the modes of interest, and was essential for
safely expanding the flight envelope of B-1A airplane (Dobbs
and Hodson, 1979).
Thrusters
Thrusters are also known as impulse generators, and they are
used for structural excitation. Thrusters are usually
lightweight and they provide transient responses of short
duration, which are important when the aircraft has to dive
from an altitude to reach a test condition. Owing to their
lightweight, they do not affect the modal characteristics of the
153

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Figure 3 Principles of random decrement method

Figure 4 Comparison of excitation responses

154

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

aircraft they are installed on. With proper choice of


combustion time, one has the ability to concentrate on a
selected frequency range. The main disadvantage of thrusters
is their inability to provide a wide frequency band of
excitation. Thrusters with different burn times are necessary
to excite modes in the frequency range of 5-50 Hz. In
addition, repetitive use of thrusters is out of question.
Reloading the exciters with a new charge can only be done
between flights, provided that the thrusters has been built for
that purpose. Among the aircraft on which this type of
exciters have been used are Concorde, A-300B Airbus,
MRCA Tornado, etc. (van Nunen and Piazzoli, 1979).
Thrusters were also used by Douglas Aircraft Company in
1950s on several airplanes to investigate flutter characteristics
(Meany, 1983).

determine damping and frequency trends as dynamic pressure


and airspeed increases during the test. Flight test is repeated
at points of incrementally increasing airspeeds or Mach
number which are flown at either constant altitude or along a
constant dynamic pressure line. If test is performed at
constant altitude, then at first, flight test is executed at high
altitudes where the dynamic pressure will be lowest. The
airspeed increments between points will depend upon the
confidence in the flutter analysis which has to be performed
prior to test execution. In addition, the proximity to the
predicted flutter boundary also affects the decision on the
proper airspeed increments. Small steps are required when
flutter boundary is approached, or when a rapid decrease in
damping is observed. An initial test airspeed and altitude
must be selected to start the flutter test. The choice of the
initial speed and altitude is based on the predicted flutter
modes and flutter margin.
In principle, the test program will differ according to
whether the aircraft is to be cleared for low- subsonic, high
subsonic or transonic/supersonic test regimes. Typical test
procedures for these speed regimes are shown in Figure 6
(Meijer, 1995; van Nunen and Piazzoli, 1979).
In the low subsonic regime, the basic variable is the dynamic
pressure only. At each successive test point, the airspeed, thus
the dynamic pressure, is increased until the maximum value is
reached (Figure 6(a)). Typically, the maximum value is the dive
speed of the aircraft at the altitude. A safe minimum test altitude
has to be selected for safe recovery in case of an accident. At this
altitude the aircraft should be able to reach the dive speed.

Flight test procedures


During the flight flutter test of aircraft, subcritical envelope
expansion procedure is used. In this procedure, less critical
points are flown prior to the more critical ones. The aircraft
response and flight parameters are also monitored in real
time. A typical modern flight test process is shown in Figure 5
(Meijer, 1995).
Test data are telemetered from the aircraft to real time
frequency analyzers, and displayed on strip charts
simultaneously. A computer acquires the data for analysis to
determine frequency and damping estimates. Real time
monitoring of the test data allows the flight test engineer to
Figure 5 Modern flight flutter test process

155

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Figure 6 Flutter test procedures

critical flutter frequency can still be conducted to ensure that


the critical mode is excited.
For the supersonic test flight, Figure 6(c) shows the flight
test sequence of the Concorde. In the first part of the flight
test, both altitude and Mach number were increased after
each test point. This way dynamic pressure can be kept
constant and Mach number is the only variable. To cover
maximum dynamic pressure, a few test points are flown at
lower altitudes. Usually, special attention is given to the
transonic speed regime at high dynamic pressure. Transonic
flight regime (M . 0.7) is considered the most critical from
flutter standpoint due to the coupling of structure response
and the motion of the shock waves (MIL-STD-1763A, 1990).
During the actual execution of the flight test, the processed
telemetry data and comparable information from the earlier
flutter calculations will exist in the ground station where the
whole flight test activity is coordinated. During either the
dynamic pressure increase or Mach number increase, if
significant differences occur, such as sudden drop of
damping, the flight test engineer may be recommended to
decrease the steps, in either the dynamic pressure or the
Mach number, by the flutter engineer who performs the
calculations. In certain situations the test may be terminated,

In the high subsonic flight flutter test, the influence of the


Mach number and dynamic pressure on the flutter
characteristics has to be distinguished. In general, the
influence of Mach number will be less predictable than the
effect of dynamic pressure. Therefore, this type of test is
usually started at a high altitude, and a sequence of Mach
numbers are flown at a reasonably low dynamic pressure. This
test sequence is shown in Figure 6(b). Then, a second set of
test points is flown at a lower altitude to determine the flutter
behaviour at the higher dynamic pressures. Finally, some test
points may be flown at which both parameters reach their
maximum values. If high test speeds cannot be reached in
straight and level flight, dive maneuvers are required to obtain
the required airspeeds until dive speed. For these test points,
an altitude band of approximately 300 m is created around the
desired test altitude at which data are required regardless of
excitation type (Kehoe, 1988). For a dive, the sweep rate may
be changed if necessary so that the sweep is accomplished
within the test altitude band. However, as the dive angle
becomes steeper, the time in the altitude band becomes
shorter and sweeps can no longer be conducted. At these test
conditions, sinusoidal frequency dwells at the predicted
156

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

and a detailed analysis of flight test data is conducted before


continuing the flight test. Especially, in case of explosive
flutter character of the critical mode, utmost attention should
be given to carefully arrange the increments during flutter
testing. A typical flutter diagram showing the variation of
damping and frequency with airspeed is shown in Figure 7 for
a two degree of freedom system (van Nunen and Piazzoli,
1979).
In Figure 7, it is shown that the rate of decrease of damping
with airspeed may be fast depending on 0the characteristics of
the system. This situation is shown by II in Figure 7. In such
cases, flutter diagram, which is based on the results of all
calculations done before flight testing and on the previously
obtained flight test results, should warn the flight test engineer
that he should proceed with very small speed increments.
During the execution of flight test, a quick-look analysis of
the telemetered signals from the aircraft has to be performed
in most cases to make sure that an explosive flutter situation is
not encountered. For prototype aircraft flutter testing, the
post flight procedures may be quite extensive. Usually,
information extracted from all the sensors are examined
during post-flight analysis to obtain a complete view of the
flutter characteristics of the aircraft. In the case of flight
testing during a store certification program, a more limited
post flight analysis may be required, since the flutter
characteristics which are of interest may be restricted to a
few well defined areas of the aircraft such as the wing.

produce accurate frequency and damping values is crucial to


the entire process. This is necessary not only from the safety
point of view but also because, having confidence in the
estimated parameters will enable the flight envelope to be
cleared more quickly.
The sub-critical approach to flutter testing is to perform
vibration test at constant flight conditions, curve-fit the data to
estimate frequencies and dampings, and quantify the variation of
frequency and damping with airspeed or Mach number. The
damping values are then extrapolated in order to determine
whether it is safe to proceed to the next test point. Damping is the
most critical parameter in flutter testing, and therefore, its
accurate estimation is very crucial. However, modal damping can
be a very poor indicator of impending instability as it depends on
critical flight parameters, Mach number and dynamic pressure,
in a highly nonlinear fashion. Tests performed with different
input signals, and the very common occurrence of
nonrepeatability at identical test conditions, and sensitivity of
damping to noise can change damping estimates dramatically
using the state-of-the-art identification techniques.
In general, the response signal may consist of random
response caused by atmospheric turbulence or exciter input,
transient responses caused by either impulse input or exciter
frequency dwell-quick stops, or steady state responses caused
by exciter frequency sweeps. The common technique to
extract the frequency and damping estimates from aircraft
responses has been the curve fitting techniques. However,
there are significant differences between the set-ups for tests
performed on ground as compared to those taken in flight
flutter tests. Fewer transducers are used for flight tests and
there are serious problems involved in putting sufficient
excitation energy into the aircraft structure over the
frequencies of interest. In addition, the amount of noise
interference on ground is small and tends to be primarily due
to measurement noise. However, in flight flutter tests there is
often considerable noise content due to turbulence which
affects the measured responses in a different manner
compared to measurement noise.
Most parameter estimation techniques are formulated to
minimize some error function based on noise corrupting the
output data sequences. Cooper gives very nice review of the
parameter estimation techniques used in flight flutter test
(Cooper, 1995; Cooper et al., 1995). In flight flutter testing,
in general input is a forcing term and acceleration is generally
taken as the measured output. The parameter estimation
methods which will be summarized in the following assume
that the systems are linear and time-invariant. These methods
are usually classified depending upon whether the curve-fit is
performed in the frequency domain or time domain. It should
be noted that in modal analysis community a multiple input
multiple output approach is considered to be superior.
However, such an approach is not recommended in flight
testing, because there is a likelihood of having wide scatter in
damping. Even frequency estimates from individual
transducers may be scattered. Therefore, most of the
methods are intended to be used in a single input single
output formulation.

Flight flutter prediction


One of the most important steps of a successful flight flutter
test is the proper use of the stability parameter estimation
methods to accurately predict the flutter. In a typical flight
test, response data are usually measured by accelerometers
which are mounted on critical points on the aircraft structure.
Owing to the risks involved in flight flutter testing, a safety
first attitude to testing must be taken. Therefore, the ability to
Figure 7 Flutter diagram for a two degree of freedom system

Stability parameter estimation methods


Frequency domain methods
Frequency domain methods are the traditional approach to
modal parameter estimation. Calculation of the frequency
157

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

response function is a parameter estimation procedure, and


random noise is removed through the use of averaging. Any
errors in the frequency response function estimation is passed
onto the estimated modal parameters. In flutter testing, it can
be difficult to produce smooth frequency response functions
due to the levels of noise and also the relatively small amount
of data that is obtained. In addition, due to the difficulties in
maintaining exactly the same flight condition few averages can
be taken.
For a multi degree of freedom system with well separated
modes, the parameters of each mode on the frequency
response function can be found by examining each circle on
the Nyquist plot, or peak on the Bode plot, individually. The
damping values can be found from either the half power
points, or a number of points around the circle, or by curve
fitting. Use of single degree of freedom methods becomes
useless if the modes become close, and half power bandwidths
overlap. This will result in poor damping estimates. However,
the use of single degree of freedom methods is still common
practice, particularly for military aircraft, because they are
easy to use, and the response of a mode that is close to flutter
mode will dominate the frequency response function, and will
be estimated accurately. A wide variety of different
approaches to filter out each of the individual modes appear
in literature (Cooper, 1995; Ruhlin et al., 1983). Often a
number of transformations between the time and frequency
domain are required in order to isolate all of the modes.
When multi degree of freedom frequency domain methods
are employed, frequency response function is modeled as
some form of rational fraction expansion that include
contributions from all of the modes. Therefore, no problem
arises when close modes are analyzed. The majority of
techniques in this category employ an iterative nonlinear least
squares curve-fit to the frequency response function model
over the frequency range of interest. The alternative approach
is to produce a linearized model and to find the estimates
using a direct procedure. Such methods include the rational
fraction polynomial method. Modal analysis methods are
reviewed very nicely by Jimin and Zhi-Fang (2001).

high signal to noise ratio from the measured data. The data
inaccuracy and presence of noise may lead to computational
modes in the calculation.
Time domain modal analysis methods also use single
degree of freedom and multi degree of freedom models.
Among the single degree of freedom models, the simplest but
most easy to use model is the logarithmic decrement method
which uses a single exponentially damped sinusoid model.
The model will give incorrect estimates, if there is more than
one mode in the response even if the resonances are well
separated.
Among some of the multi degree of freedom time domain
methods, we can name least-squares time domain method,
Ibrahim time domain method, autoregressive moving average
(ARMA) series method, and least squares complex
exponential method.
In the least squares method, the vibration response data are
used for a curve-fitting effort against a defined mathematical
model to derive the modal parameters. For a multi degree of
freedom dynamic system, its free vibration response is known
to be in the form:
xt

N
X

e2ni t ai sin vdi t bi cos vdi t

i1

where, the coefficients ai and bi are the components of the


vibration amplitude, and vdi and ni are the damped natural
frequency and related damping ratio, respectively. In this
method the response is sampled at a time resolution, and a
least squares minimization of the difference equation
formulation results in a linear equation set after inputting
initial estimates for the frequency and damping. The whole
process heavily relies on the initial estimates and the number
of modes used.
Ibrahim time domain method uses the impulse response
function data to identify modal parameters. It constructs an
eigenvalue problem from the impulse response function data
and solves the problem to derive the natural frequencies and
damping factors (Pappa and Ibrahim, 1981). Ibrahim time
domain method has been applied to flutter data from the
space shuttle (Cooper, 1995).
Least-squares complex exponential method is another time
domain modal analysis method. It explores the relationship
between the impulse response function of a multi degree of
freedom system and its complex poles and residues through a
complex exponential. By establishing the analytical links
between the two, one can construct an auto regression model.
The solution of this model leads to the establishment of a
polynomial whose roots are the complex roots of the system.
Natural frequencies and damping ratios can be estimated
from these complex roots. Having estimated the roots, the
residues can be derived from the AR model for mode shapes.

Time domain methods


Time domain methods belong to another category of modal
analysis methods which uses time response data. These
methods were developed on the advances of modern control
theory and computer technology. Time domain methods
generally use the procedure of curve fitting the impulse
response function calculated from, either the inverse Fourier
transform of the frequency response function, time response
to an impulse, or generated from response to an unknown
random input. The data of known inputs enable correlation
analysis and usually assure better accuracy of the analysis.
The data from ambient excitation, however, grant a special
appeal to the time domain approach. The independence of
some of the time domain methods on the excitation data
enables them to be used for special applications (Jimin and
Zhi-Fang, 2001).
Time domain modal analysis has several advantages over
frequency domain approaches. It does not depend on
excitation and equipment needed to obtain designed
excitation forces. Generally, it needs fewer response data. As
the analysis is not based on the frequency response function
data, it can be an alternative for analyzing close vibration
modes. However, time domain modal analysis requires a very

Parameter estimation using random atmospheric


turbulence
As it was discussed in excitation systems, because of the
difficulties and expense of installing excitation systems for flight
flutter testing, frequency and damping values are sometimes
extracted from the response of the aircraft to turbulence.
Common applications of atmospheric turbulence are the testing
of light aircraft, and aircraft which have undergone minor
modifications such as external store installations. Random
decrement method is the most commonly used technique to
158

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

obtain a transient response from random response data. Then,


by using single and multi degree of freedom time or frequency
domain methods, modal parameters are extracted from the
transient response obtained.
A number of other methods exist for estimating modal
parameters from the response of the aircraft to an unknown
random input. The autocorrelation technique and maximum
entropy algorithm are among the techniques used for this
purpose (Cooper, 1995).
In a NASA research conducted in response to the concerns
of the aeroelastic community, the main problems associated
with stability parameter estimation were listed as:
.
problems with estimation of closely spaced modes;
.
assumption of time invariance, linearity;
.
sensitivity to type of input or noise levels;
.
lack of technique for estimating error bounds; and
.
computational complexity.

limit cycle oscillation started. As it can be seen in Figure 8,


a linear extrapolation of the data trend provided an instability
airspeed prediction that agreed closely with the actual
instability onset airspeed encountered.
Flutter margin method
The idea behind the flutter margin method is the quest for a
more fundamental stability criterion than just tracking the
damping present in the system. The original form of the method
is due Zimmerman and Weissenburger (1964), but it only
covers the flutter of a system having two degrees of freedom.
Extension of the technique to trinary flutter is presented by
Price and Lee (1993). With the flutter margin method, it was
intended to avoid the deficiencies of the velocity-damping
technique. It is known that in case of explosive flutter behaviour,
damping values can change abruptly with no prior warning.
However, in case of flutter margin method, a quest is made for a
different stability parameter such that when the new stability
parameter is plotted against airspeed, the curve behaves in a
civilized manner and is not prone to sudden or unexpected
reversals, such as would be the case if damping alone were
plotted. For a two degree of freedom system which represents
the bending-torsion idealization, the characteristic equation
turns out to be quartic:

A survey of modal estimation methods that was made in


NASA Dryden, for aeroelastic applications, revealed that
frequency domain subspace identification methods combined
with time-frequency multiscale wavelet techniques were the
most promising algorithms (Brenner and Feron, 1997).

Flutter prediction

l4 A3 l3 A2 l2 A1 l A0 0:

Damping ratio variation with airspeed


The most widely used indicator of the stability is the variation
of modal damping with the airspeed. At the flutter point, the
modal damping of at least one of the modes is zero, thus
causing self-excited oscillations. Beyond this speed, the
system becomes unstable (Figure 7). In typical flight flutter
test, the damping ratios for all of the significant modes are
evaluated at a number of subcritical airspeeds using system
identification methods. The damping ratio trends are then
curve fitted by a polynomial, and extrapolated to estimate the
flutter velocity and to examine the stability at the next
proposed flight test point. As an example, Figure 8 shows the
damping trend obtained in near real time for an F-16 airplane
configured with AIM-9J missiles, GBU-8 stores, and
370 gallon external fuel tanks (Kehoe, 1988).
Flight of the airplane with this store configuration was
characterized by a limit cycle oscillation. Excitation was
created by the flaperons utilizing frequency dwells and quick
stops, and damping was estimated by using FFT algorithms.
The damping data provided a unique opportunity to validate
the accuracy of this algorithm, because this configuration
could be safely flown to condition of zero damping, where

Flutter margin method then employs the well known Rouths


stability criterion and defines the flutter margin as:
F A2 A1 =A3 2 A1 =A3 2 2 A0 . 0:

In the above definition, one can define the flutter margin in


terms of system frequencies and negative of the system decay
rates, and eventually express the flutter margin in terms of
the configuration constants and dynamic pressure as in
equation (4) (Zimmerman and Weissenburger, 1964).
F B2 q2 B1 q B0 :

In this form, flutter margin is a quadratic function of the


dynamic pressure. Hence, if the flutter margin is known at three
different airspeeds, it can be fitted by a second order
polynomial, and subsequently extrapolated to zero flutter
margin where flutter occurs. In practice, flutter margin is
estimated at a wider range of subcritical airspeeds and then
fitted in a least-squares sense. Flutter margin method can be
used with systems more than two degrees of freedom, if it is
known beforehand which two modes will cause flutter. In that
case flutter margin method can be applied successfully;
otherwise, all possible pairs of modes must be examined.

Figure 8 Damping trend for limited amplitude instability

Envelope function method


The basis of the method is that the impulse response of any
stable damped system is decaying, with the shape of the decay
in the time domain being described by the decay envelope
(Cooper et al., 1993). As the damping in a given aeroelastic
system decreases, the decay envelope grows wider, eventually
becoming a rectangle as the damping becomes zero. When the
position of the centroid of the decay envelope, and the way it
shifts on the time axis as the damping decreases are evaluated,
it is possible to assess the stability of the system (Dimitridiadis
and Cooper, 2001). Time centroid of the decay envelope is
given by:
159

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

motion or by calculating the systems eigenvalues. Finally, the


flutter speed can be obtained by means of a suitable iteration
process, for example, evaluating the systems damping at
increasing speeds until flutter is reached (Nissim and Gilyard,
1989).

R t max

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

t

envttdt
R0t max
:
0 envtdt

The upper limit of the integration serves to define the


rectangle, within which the integration takes place. For a
single degree of freedom system, when the damping is zero,
the time centroid lies at t max =2: For a multi degree of freedom
system, it is suggested that t < t max =2 is an adequate
approximation for the position of the time centroid. Since, t
tends to increase as the damping drops, its inverse 1=
t is a
significant stability parameter. At the flutter condition, the
stability parameter will be 2=t max for a single degree of
freedom system. For a multi degree of freedom system it will
be close to 2=t max :
The envelope function flutter testing procedure is to
evaluate the stability parameter at a number of subcritical air
speeds. The variation of the stability parameter with air speed
is then curve fitted using a polynomial, as with the damping
method, and extrapolated to the point where the stability
parameter becomes 2=t max ; thus yielding the flutter velocity.

Autoregressive moving average-based method


The ARMA method is a system identification technique.
Matsuzaki and Ando (1981) suggest a procedure for using an
ARMA representation of an aeroelastic system as a means of
predicting the flutter velocity. The basis of the approach is the
Jury determinant method for evaluating the stability of a
discrete time system, which is very similar to the RouthHurwitz criterion for continuous time systems. The Jury
stability criterion applies to the characteristic polynomial such
that the system is stable if any of the three different criteria
given by Matsuzaki are above zero. The flutter condition is
defined as the first air speed at which any of the three criteria
specified by Matsuzaki becomes zero. Hence, the system can
be plotted against air speed, and then curve fitted by a
polynomial to yield the points where they intersect the velocity
axis and, hence, the flutter velocities. However, the particular
implementation of the Jury stability criterion is
disadvantageous compared to the flutter margin method in
that the criteria do not vary in a predictable manner with
velocity.

Nissim and Gilyard method


Nissim and Gilyard (1989) method is based on the
identification of the coefficients of the equations of motion
at low dynamic pressures, followed by the solution of these
equations to compute flutter dynamic pressure. The
equations of motion for a forced aeroelastic system are:
M {q} C {_q} K {q} F eiwt

Eigenvector orientation method (EVO method)


Eigenvector orientation method is another flutter prediction
method which serves as a complement to the existing methods
based on eigenvalues (Afolabi et al., 1998). It is based on the
complex eigenvectors to predict the onset of the flutter
instability (Pidaparti et al., 2001; Afolabi et al., 1998). The
eigenvectors, which are real and orthogonal to each other,
lose their orthogonality at the flutter instability. Thus, when
the angle between any two eigenvectors deviates from 90o,
that indicates the extent to which the aeroelastic system is
close to the flutter boundary. Therefore, in the eigenvector
orientation method, the angle between any two complex
eigenvectors (any modes), or its deviation from 90o, or when
the rate of change of the EVO angle is zero, can be used to
track the flutter condition. For a real time flutter control
process, one can use the EVO method, because it can easily
be monitored, and where the EVO angle gets close to zero it
indicates the flutter condition. Pidaparti et al. (2001), applied
the EVO method together with the common V-g method to an
intermediate complexity wing. A reasonably good comparison
of the flutter speed between the EVO method and the V-g
method was obtained.

where it is assumed that forcing is harmonic, and it can, in


general, be assumed to consist of more than one forcing
vectors. If [F ] is assumed to be unknown, and {q} is
measured, then through a series of matrix operations it
is possible to write down (Nissim and Gilyard, 1989;
Dimitridiadis and Cooper, 2001):
T X  B 

where T and B are complex matrices composed of measured


frequencies (wi) and responses (qi), and X is:
9
8
 T >
>
K
>
>
>
>
=
<
T


C

8
X 
>
>
>
>
>
 T >
;
: F
where:
  M 21 K ;
K

  M 21 C ;
C

  M 21 F  9
F

Special considerations and concluding remarks

In equation (8) [X ] is usually solved in least squares sense. In


aeroelasticity the overall damping and stiffness matrices are
expressed as:

In this study, it is intended to present concise but relevant


information on the flight flutter stability verification of aircraft
by elaborating on the three main topics of special
consideration in flight flutter testing. The main items of
special consideration in flight flutter testing are the structural
excitation, structural response measurement and stability
prediction steps. Besides these three main items,
airworthiness requirements with regard to aeroelastic
stability are also reviewed, and flight test procedures are
explained by referencing the relevant studies reported in the
literature.

  Cs  1 rV 1 A1 t; and K
  K s  1 rV 1 A0  10
C
2
2
where subscript s denotes structural terms. If the
identification process described above is repeated at two
distinct velocities, then aerodynamic stiffness matrix [A0], and
aerodynamic damping matrix [A1] can be evaluated
separately. This way, the behaviour of the system at any air
speed can be predicted either by integrating the equations of
160

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

During flight flutter tests, for gradual decreases in damping


with airspeed, the trace of damping at regularly spaced
airspeed intervals appears to be adequate to warn the onset of
flutter. However, it is doubtful that sudden changes in
damping, which may occur between flutter test points, can be
predicted with required accuracy and in a timeliness manner
to avoid flutter with discrete damping trace method. The most
critical part of expanding flutter envelope is the acceleration
from one test point to the next. During this period response
data are not being quantitatively analyzed. Online, real time
monitoring of flutter stability during flight testing is very
crucial, if the flutter character is not known a priori.
Techniques such as modal filtering can be used to uncouple
response measurements to produce simplified single degree of
freedom responses, which could then be analyzed with less
sophisticated algorithms that are more able to run in real
time. This way, damping can be monitored in a continuous
fashion eliminating the most hazardous part of flutter testing
which is the acceleration from one test point to the next. Such
monitoring also eliminates the need for the very time
consuming stabilized test points. This would in turn help
reduce the scatter in damping, and give better estimation of
flutter speed which is obtained by the extrapolation of flight
test data to zero damping.
Control surface excitation is a very common way of excitation
mechanism used in most aircraft. However, the choice of
reference signal in the calculation of frequency response
function is extremely important in case of control surface
excitation in flutter testing. As noted by Wright et al. (2003), the
accurate determination of the frequency response function can
be possible, if the actuator demand signal is used as the
reference signal for the frequency response function
calculation. The use of actuator force or control surface
rotation angle as a reference signal could yield incorrect modal
data, because these signals contain the essential response
dynamics of the flutter system. One way to check whether the
use of control surface rotation as a reference signal leads to
correct frequency response function or not, is to look at the time
history of the control surface. If the amplitude is essentially
constant, then control surface is not significantly involved in the
flutter mechanism, and test results will be satisfactory.
However, this situation cannot always occur especially, if the
control surface is involved in the flutter mechanism.
Scatter of the damping data gathered during flight can be
attributed to several factors. The very common occurrence of
nonrepeatibility at identical test points such as applying
repeatable excitation, and keeping the same trim test speed is
deemed to be one of the reasons for the scatter. Especially, with
the aircraft which does not possess a fly-by-wire control system,
keeping the same trim speed and altitude, especially at high
speeds, can be difficult during different control surface
excitations. It is noted that damping can be a poor indicator of
impending stability as it depends on critical flight parameters
such as airspeed in a nonlinear fashion. Dynamic pressure
varying with the square of the speed may actually account for
different damping values if the same trim speeds, and altitude
band cannot be maintained during the execution of different
control surface excitations at a preplanned test point. Poor
signal to noise ratio for certain control surface excitation and
accelerometer response pairs also gives rise to inaccurate
determination of damping, besides experimental errors. To
reduce the scatter in damping more repeatable tests need to be
performed at identical test locations to get better average values.

Although time domain modal analysis has several


advantages over frequency domain method, it requires a
very high signal to noise ratio from measured data. The data
inaccuracy and presence of noise may lead to computational
modes in the calculation. In time domain modal analysis
method, there is no luxury of utilizing frequency response
function data around resonances for better signal to
noise ratio. Therefore, high signal to noise ratio is very
critical to ensure accurate calculation of modal data, thus
damping.
An evaluation at NASA Dryden has been performed for the
usefulness of several modal estimation algorithms in flight
flutter testing. According to this evaluation, frequency domain
subspace identification methods combined with timefrequency multiscale wavelet techniques were the most
promising algorithms (Brenner and Feron, 1997; McKelvey
et al., 1996).

References
Afolabi, D., Pidaparti, R.M.V. and Yang, H.T.Y. (1998),
Flutter prediction using eigenvector orientation
approach, AIAA Journal, Vol. 36 No. 1, pp. 69-74.
Brenner, M.J. and Feron, E. (1997), Wavelet analyses of
F/A-18 aeroelastic and aeroservoelastic flight test data,
paper presented at 38th AIAA Structures, Structural
Dynamics, and Materials Conference, April.
Brenner, M.J., Lind, R.C. and Voracek, D.F. (1997),
Overview of recent flight flutter testing research at
NASA Dryden, NASA Technical Memorandum 4792,
Dryden Flight Research Center, Edwards, April.
Cole, H.A. Jr (1973), On-line failure detection and damping
measurement of aerospace structures by random decrement
signatures, NASA CR-2205.
Cooper, J.E. (1995), Parameter estimation methods for
flight flutter testing, AGARD-CP-566, November.
Cooper, J.E., Desforges, M.J., Emmett, P.R. and Wright, J.R.
(1995), Advances in the analysis of flight flutter test data,
AGARD-CP-566, November.
Cooper, J.E., Emmett, P.R., Wright, J.R. and Schofield, M.J.
(1993), Envelope function: a tool for analyzing flutter
data, Journal of Aircraft, Vol. 30 No. 5, pp. 785-90.
Dimitridiadis, G. and Cooper, J.E. (2001), Flutter
prediction from flight flutter test data, Journal of Aircraft,
Vol. 38 No. 2, pp. 355-67.
Dobbs, S.K. and Hodson, C.H. (1979), Determination of
subcritical frequency and damping from B-1 flight flutter
test data, NASA CR-3152, June.
FAR 25 (2000), Airworthiness Standards: transport category
airplanes, Aviation Regulation, Federal Aviation
Authority, Washington, DC, December.
Fung, Y.C. (1993), An Introduction to the Theory of
Aeroelasticity, Dover Publications, NewYork, NY, pp. 160-2.
JAR 25 (2004), JAR-25 large aeroplanes, Aviation
Regulation, Joint Aviation Authority, Hoofddorp.
Jimin, He. and Zhi-Fang, Fu. (2001), Modal Analysis,
Butterworth-Heinemann, Oxford.
Kehoe, M.W. (1988), Aircraft flight flutter testing at the
NASA Ames-Dryden flight research facility, NASA TM100417, May.
Kehoe, M.W. (1995), A historical overview of flight flutter
testing, NASA Technical Memorandum 4720, Dryden
Flight Research Center, Edwards, CA, October.
161

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

Flight flutter testing and aeroelastic stability of aircraft

Aircraft Engineering and Aerospace Technology: An International Journal

Altan Kayran

Volume 79 Number 2 2007 150 162

McKelvey, T., Akcay, H. and Ljung, L. (1996), Subspacebased multivariable system identification from frequency
response data, IEEE Transactions on AutomaticControl,
Vol. 41 No. 7.
Matsuzaki, Y. and Ando, Y. (1981), Estimation of flutter
boundary from random responses due to turbulence at
subcritical speeds, Journal of Sound and Vibration, Vol. 18
No. 10, pp. 862-8.
Meany, J.J. (1983), The evolution of flutter excitation at
McDonnell aircraft, 14th Annual Symposium Proceedings,
Society of Flight Test Engineers, August, pp. 4.6-1-4.6-11.
Meijer, J.J. (1995), Aeroelasticity, AGARDograph 300 Flight
Test Techniques Series Introduction to Flight Test Engineering,
Vol. 14.
MIL-A-8870C (1993), Military specification-airplane
strength and rigidity vibration, flutter, and divergence,
MIL-A-8870C, pp. 3-5.
MIL-STD-1763A (1990), Military standard-aircraft/store
certification procedures, MIL-STD-1763A, pp. 27-8.
Nissim, E. (1993), Improvement of method for experimental
determination of flutter speed by parameter identification,
NASA CR-193555.
Nissim, E. and Gilyard, G.B. (1989), Method for
experimental determination of flutter speed by parameter
identification, NASA TP 2923.
Pappa, R.S. and Ibrahim, S.R. (1981), A parametric study of
the Ibrahim time domain modal identification algorithm,
Shock and Vibration Bulletin, Vol. 51 No. 3, pp. 43-72.
Pidaparti, R.M.V., Tischler, V.A. and Venkayya, V.B.
(2001), Flutter prediction methods for aeroelastic
design optimization, Journal of Aircraft, Vol. 38 No. 3,
pp. 557-9.
Price, S.J. and Lee, B.H.K. (1993), Evaluation and
extension of the flutter margin method for flight
flutter prediction, Journal of Aircraft, Vol. 30 No. 3,
pp. 395-402.
Rosenbaum, R. (1974), Survey of aircraft subcritical flight
flutter testing methods, NASA CR-132479, August.
Ruhlin, C.L., Watson, J.J., Ricketts, R.H. and Doggett, R.Y.
(1983), Evaluation of four subcritical response methods
for on-line prediction of flutter onset in wind tunnel tests,
Journal of Aircraft, Vol. 20 No. 10, pp. 835-40.
van Nunen, J.W.G. and Piazzoli, G. (1979), Aeroelastic flight
test techniques and instrumentation, AGARD-AG-160,
Vol. 9.

Vernon, L. (1993), In-flight investigation of a rotating


cylinder-based structural excitation system for flutter
testing, NASA TM-4512.
Wright, J.R., Wong, J., Cooper, J.E. and Dimitriadis, G.
(2003), On the use of control surface excitation in flutter
testing, Proceedings of the Inst. of Mechanical Engineers, Part
G, Journal of Aerospace Engineering, Vol. 217, pp. 317-32.
Zimmerman, N.H. and Weissenburger, J.T. (1964),
Prediction of flutter onset speed based on flight testing
at subcritical speeds, Journal of Aircraft, Vol. 1 No. 4,
pp. 190-202.

Further reading
Hodson, C.H., Dobbs, S.K., Brosnan, M.J. and Chen, J.B.
(1993), X-31A flight flutter test excitation by control
surfaces, AIAA-93-1538-CP, paper presented at 34th
AIAA Structures, Structural Dynamics and Materials
Conference, La Jolla, CA, April.

About the author


Altan Kayran received his undergraduate
degree in mechanical engineering in Middle
East Technical University in Ankara, Turkey, in
1985. He received his PhD degree in
Mechanical and Aerospace Engineering,
Department of University of Delaware in
1990. His PhD work was on the vibration
analysis of laminated composite shells of revolution including
transverse shear deformation. Upon completion of his PhD,
he has worked in Aerospace Industry (TAI), Turkish
Automobile Industry (TOFAS) and Electronics Industry
(ASELSAN) in various projects in Turkey for a period of 11
years. He has joined the Department of Aerospace
Engineering of Middle East Technical University in Ankara
in 2003. He has been with the department since then. He
lectures on aircraft structures, applied elasticity, mechanics of
materials, mechanical vibrations, mechanics of composite
materials, applications of finite element analysis in aerospace
structures. His research interests are: structural design and
analysis, composite materials, finite element method,
behaviour of thin walled structures: beams, plates and
shells, structural dynamics, and aeroelasticity. Altan Kayran
can be contacted at: akayran@metu.edu.tr

To purchase reprints of this article please e-mail: reprints@emeraldinsight.com


Or visit our web site for further details: www.emeraldinsight.com/reprints

162

This article has been cited by:

Downloaded by Indian Institute of Technology Bombay At 05:39 21 February 2016 (PT)

1. Timme S., Badcock K. J.. 2011. Updating Computational Aeroelastic Models using Flight Flutter Test Data. International
Journal of Aerospace and Lightweight Structures (IJALS) - 01, 157. [CrossRef]

You might also like