574 J. AIRCRAFT VOL. 19, NO.

AIAA 81-0604R

Effect of Store Aerodynamics on Wing/Store Flutter

C.D. Turner*
North Carolina State University, Raleigh, N.C.

Owing to the high cost of doing flutter analysis for aircraft carrying large numbers and types of stores, it is not
economically feasible to include store aerodynamics when there will be little change in the flutter results. But
store aerodynamics should be included if it will change the results of the flutter analysis. This study represents
the first systematic analytical study of the effect of store aerodynamics on wing/store flutter. A large number of
wing/store single carriage configurations and parameters were included in the study; multivariate analysis
techniques were used for the first time to analyze wing/store configurations, modal data, and flutter results. The
results of the multivariate analysis indicate that it may not be possible to develop general guidelines, but it is
possible to develop specific guidelines for use with a particular aircraft.
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

Nomenclature I. Introduction
= store body area projected in horizontal plane
= store fin area projected in horizontal plane
= store canard area projected in horizontal plane
W HEN large external bodies or stores, such as engine
nacelles, fuel tanks, or pods, are added to the wing of
an aircraft, the dynamic characteristics of the aircraft will be
= total store area =AB+AF]+AF2 changed. In particular, the flutter speed of the aircraft may be
Aw - wing area adversely affected because of the inertia, elastic, and
C = reference chord of wing aerodynamic coupling between the wing and its stores. It has
= chord of wing at store location been possible for the last several years to account for the
= distance from wing to top of store, defined in Fig. wing/store aerodynamic coupling in flutter analysis. Even so,
2 store aerodynamics are usually not included in a standard
Ls = length of store flutter analysis, except for a few cases of tip pods and tip
Ls = distance from E.A. to store nose, defined in Fig. 2 missiles on fighter aircraft and engine nacelles on the larger
L, = distance from E.A. to store canard, defined in Fig. transport aircraft. This situation is the result of both
2 economic considerations and the hopeful assumption that
= distance from E.A. to store fin, defined in Fig. 2 store aerodynamics have little effect on the calculated flutter
- distance from aircraft centerline to wing tip, speeds.
defined in Fig. 2 In the case of wing/store flutter analysis, the economics of
'W2 = distance from aircraft centerline to store, defined including the store aerodynamics must be considered,
in Fig. 2 together with other technical aspects of the flutter analysis.
N = number of configurations used in multivariate The computation of the unsteady aerodynamic coefficients
analysis for the aircraft is the single-most costly item in any flutter
r = radius of store, defined in Fig. 2 analysis. Owing to the variety of stores that can be carried on
= correlation coefficients, defined in Eq. 3 a typical modern tactical fighter, the total number of possible
= velocity aircraft/store configurations is in the thousands. On military
= clean wing flutter velocity aircraft, the effect of unsteady aerodynamics is computed for
= wing/store flutter velocity with no store both the clean wing and the wing with tip missiles. Using these
aerodynamics two aerodynamic configurations, the flutter analysis is done
'FA = wing/store flutter velocity with store on between 300 and 400 selected wing/store configurations. If
aerodynamics wing/store aerodynamics were also considered for each of
= weight of store these configurations, the cost of the analysis would increase
= weight of wing by several orders of magnitude. In many cases, this increased
X = independent variable in factor analysis cost is not justified because store aerodynamics usually has a
X = mean value small effect on the flutter speed. However, there are a few
= variation from the mean important instances where neglecting store aerodynamics will
= angular pitch of store at store center of gravity for lead to the overestimation of the flutter speed. Determining
mode / when the store aerodynamics should be included in the flutter
= displacement of E.A. at store location, mode / analysis is a major problem.
= number of parameters used in analysis Early wind-tunnel tests left the impression that store
K = reduced frequency aerodynamics were not needed to accurately compute flutter.
X = transient decay rate coefficient However, several instances of wing/store flutter during flight
= wing/store flutter frequency, no store flutter tests changed this. From these flight flutter tests and
aerodynamics subsequent wind-tunnel tests, general "rules" have evolved as
= wing/store flutter frequency with store to when store aerodynamics should be included in the flutter
aerodynamics analysis. These rules are not only based on limited data but
= natural frequency for wing/store, mode i sometimes conflict. Some experimental efforts have deter-
mined the effect of including the store on the wing of an
Presented as Paper 81-0604 at the AIAA Dynamics Specialists aircraft with various shapes, but no real effort has been made
Conference, Atlanta, Ga., April 9-10, 1981; submitted April 24, 1981; to isolate the effect of the store aerodynamics on {he
revision received Sept. 25, 1981. Copyright © American Institute of
Aeronautics and Astronautics, Inc., 1981. All rights reserved. wing/store flutter problem.
*Assistant Professor. Member AIAA. To determine the effect of store aerodynamics on
 JULY 1982 WING/STORE FLUTTER 575

wing/store flutter a data base had to be generated. This data recommended that tip .tanks be included among those cases
base is limited to single carriage of each store type and does where store aerodynamics are important. Cayman5 conducted
not include transonic flow effects. an experimental and theoretical investigation into the an-
To aid in studying the effects of store aerodynamics on tisymmetric flutter of the Northrop Scorpion F-89. He at-
wing/store flutter, a nonlinear multivariate analysis (factor tempted to cross-correlate the results obtained from six wing
analysis) was made of the data base. The factor analysis used tip pods tested in the wind tunnel in terms of mass
a set of candidate parameters that were selected from flight parameters. This effort was unsuccessful. It was felt that
test, wind-tunnel test, and analytical results. The factor greater insight into the theoretical aspects of the problem was
analysis was used to determine if it is possible to show any needed before a suitable correlation basis could be deter-
correlation between the parameters and the wing/store flutter mined. The results of the study indicated that the effect of tip
speed, thus making it possible to develop guidelines as to store aerodynamics must be included in the flutter analysis for
when to include store aerodynamics in the wing/store flutter wing/store configurations where the store center of gravity is
analysis. on or near the wing elastic axis. Gayman concluded that the
flutter speed would be less affected by store aerodynamics if a
II. Background large store overbalance condition existed. Sewall, Herr, and
The need to consider such items as inertia, elastic, and Igoe6 investigated an F-80 aircraft in which the inertial and
aerodynamic coupling between the wing and a store or a geometric properties of the external stores were systematically
nacelle is extremely important considering the rapid increase varied. Their experimental investigation found that, when the
in speed of fighter and multiengine aircraft. In the early volume of the tip tank increased, the flutter speed decreased.
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

1940s, analytical flutter techniques were restricted to analysis The addition of a horizontal fin to the smaller tip tank in-
of the wing and control surfaces. Thus the primary means of creased the flutter speed, while, for the larger tip tank, this
studying the effects of a store on wing/store flutter was addition decreased the flutter speed. Pollock and Cooley7
through the use of wind-tunnel flutter models. By 1950 recently analyzed a flight flutter test incident that occurred in
analytical techniques allowed the inclusion of the inertia July 1950 on an F-80 aircraft. For this test, the tip tanks were
effect of stores on wing/store flutter. Both wind-tunnel tests balanced with 85 Ib of lead, located 59 in. aft of the wing
and flight tests were still needed to account for the total effect elastic axis. This condition represented 13 gal of fuel in the aft
of the store on wing/store flutter. At this time, it was end of the tank. For their study, flutter analysis was ac-
recognized that the store aerodynamics must be included in complished on three different aerodynamic configurations.
the wing/store flutter analysis for some configurations if one The first was for the wing along, the second included the store
was to obtain results that matched the flight test. By the late body aerodynamics, and the third included the store body and
1960s it became technically possible to include all the possible fin aerodynamics. The first configuration gave flutter results
effects of a store in the wing/store flutter analysis. However, that were 25% above the known flutter speed. The third gave
even today, store aerodynamics are included only for flutter results that were within \% of the known flutter speed,
wing/store flutter analysis for stores located on the wing tip. and the second configuration gave results that were about
The good agreement usually obtained during the wind-tunnel 10% below the known flutter speed. In this particular case,
flutter tests and flight tests of various other store con- accurate representation of the store was found to be im-
figurations usually supports this practice. Today researchers portant. Cross and Albano8 developed a perturbation
are looking for active and passive ways to reduce the effects of technique for the rapid flutter clearance of aircraft carrying
stores on wing/store flutter, but even in these efforts the external stores. The technique allows for perturbations of the
effect of store aerodynamics is usually ignored for underwing store mass and inertia data. For most cases, extreme accuracy
stores because of the complexity of the analysis. was obtained for configurations involving moderate per-
It is useful to summarize the various investigations that turbations. The flutter program discussed in Ref. 8 can
have led to the present research. Goland and Luke 1 did an complete up to 1000 analyses/h and allows the screening of
analytical study on the effect of the chordwise positions of a many configurations and the identification of problem areas.
tip mass on wing flutter. A mass located aft of the elastic axis Van Nunen, Roos, and Meyer9 have developed experimental
was observed to lower the flutter speed. Runyan and Sewall2 techniques for measuring unsteady aerodynamics on
conducted an experimental study of wing flutter in which the wing/store configurations. These configurations include both
chordwise and spanwise positions of several different lumped tip wing and underwing stores. Data obtained in these tests
masses on an unswept wing were varied. In general, they are just now becoming available for the verification of
found that masses located forward of the elastic axis gave existing computer programs. Hwang, Winther, and Mills10
higher flutter speeds than those located aft of the elastic axis, developed both analytical and experimental methods to
except at the wing tip, where results were mixed. For all demonstrate an active wing/store flutter suppression system
chordwise locations of each mass, as the mass was moved on the Northrop YF-17 aircraft. Three configurations were
outboard from the root, there was an initial decrease followed studied, but no attempt was made in the analytical study to
by an increase in flutter speed with span position. Masses include the effect of store aerodynamics for the underwing
located forward of the wing elastic axis caused a minimum stores. Reed, Foughner, and Runyan 11 have developed both
flutter speed at about 25% span, while masses located aft of analytical and experimental methods to demonstrate a simple,
the elastic axis caused a minimum flutter speed at about 50% effective wing/store flutter suppressor known as a decoupler
span. In each instance where the mass was added forward of pylon. The effect of the store aerodynamics was not included
the elastic axis, there was a region along the span where the in their flutter analysis. A comparison of the analytical and
wing/mass flutter speed would exceed the clean wing flutter experimental data is given in the paper presented, with the
speed. Sewall and Woolston3 did an experimental study of variation between experimental and analytical results lying
wing flutter that varied both the inertia and aerodynamic between 4 and 26% for the eight configurations tested.
characteristics of the lumped mass. For the wing model used The present study includes four different fighter aircraft
in this study, it was determined that the flutter speed was and three basic stores. The four-model-data-base study
relatively unaffected by changes in aerodynamic shape. required 364 computer runs to generate the 236 data points
Andropoulos, Chee, and Targoff 4 studied, experimentally, used in the final study.
the effect that two lumped masses have on wing flutter. In- Since it is possible to compute the effects of store
cluded in this study was the effect of the nacelle aerodynamics on wing/store flutter for a given store, the next
aerodynamics. To obtain a satisfactory comparison between logical step would be to determine if it is possible to predict in
theory and experiment, it was concluded that nacelle general, on the basis of a given number of results, the effect of
aerodynamics must be included in the analysis. It was also store aerodynamics on wing/store flutter. A technique of this
 576 C.D. TURNER J. AIRCRAFT

type would be extremely useful in cases where a large number pressure of the store. This allows the effects of the canards,
of store configurations are involved in the flutter analysis or fins, and body to be independently studied. These parameters
in cases where a perturbation technique is applied for rapid are given in Fig. 2. A set of modal parameters was selected to
flutter clearance of aircraft carrying external stores. The describe the ratio of the natural frequencies and the relative
success of a prediction technique depends on the ability to displacement of the store (Fig. 2).
select a set of parameters that best describes the effect of store One additional area of concern was an apparent discon-
aerodynamics on wing/store flutter. A review of past studies, tinuity in flutter speed due to mass variations as described by
together with the data base generated for the present study, Gayman.5 The discontinuity is due to a change in the flutter
has resulted in the selection of 23 candidate parameters. mechanism as various parameters are changed. In Cayman's
study, the mass and center-of-gravity location were varied for
III. Analytical Data Base a fuel tank. There was no discontinuity in the damp-
To determine the effects of store aerodynamics on ing/velocity plots for various parameters, but there was a
wing/store flutter requires a data base that allows for the discontinuity in the flutter velocity vs mass properties. In an
study of the effect of selected parameters on store attempt to predict when the flutter mechanism will switch for
aerodynamics. The required modal and flutter data needed various aerodynamic parameters, additional modal
for this specific type of data base was not available; therefore parameters have been included in the parameter set. These are
it was necessary to generate the data base analytically. The u3/ult 65, a3 (see Fig. 2).
data base had to be developed with considerations for the
scope of the task and for the effect it would have on the
multivariate analysis. To keep the task manageable, four
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

aircraft were selected. The F-54/B14 and the F-5E12'13 were

selected because they were similar, but still have variations in
wing planform. In both cases, both the analytical and ex-
perimental data were available for checking the modal
analysis and the flutter analysis. The F-1710 was selected
because of the availability of the wind-tunnel flutter test data.
The F-807 was selected because it represented one of the first
aircraft to have a known flutter problem that required the
store aerodynamics to be included in the analysis to obtain a
valid wing/store flutter prediction. Therefore wind-tunnel
and flight test data were available for it. A wing planform
view for each aircraft is given in Fig. 1.
A review of experimental and analytical flutter data in- PARAMETERS.
dicate that the parameter groupings that are most likely to L H L
represent the effect of store aerodynamics on wing/store POSITION : W2/L W , w/y CG/CS
flutter are store motion, store location, store size, and store
center-of-pressure location. Studies by Goland and Luke, 1 SIZE : LSI/CS A
S/A W
Runyan and Sewall,2 and Gayman5 indicate that store mass,
chordwise, and spanwise position have an effect on the CENTER OF PRESSURE 1 L S2/c s Ls
^/LS| Ls4
/L s ,
wing/store flutter speed. The first two studies did not include
store aerodynamics, therefore the changes in flutter speed A
F|/A B A
F2/A B r/Ls|
were due to changes in the natural frequencies and mode
shapes of the wing itself. The last study included store MODAL PARAMETER
aerodynamics so that the change in wing/store flutter speed
was a combination of modal and aerodynamic effects. Later "Vw, w3/aj| 5 ( a, 82 a2 S3
studies by Sewall, Herr, ana Igoe6 and Pollock and Cooley7
indicate that accurate representation of the store center-of- FLUTTER PARAMETERS
pressure location is needed to obtain valid flutter results for VVFA ^F/OJ, WFA/CO(
some wing/store configurations. A set of 20 parameters was
selected to represent store motion, store location, store size, Fig. 2 Definition of parameters.
and store center-of-pressure location. Three additional
parameters were selected to represent the effect of store
aerodynamics on the wing/store flutter speed. Store position
is given by three parameters; store size is given by two
parameters; and, instead of making the center of pressure one
parameter, it was decided to include a set of geometric
parameters that can be used to determine the center of
a) AIM Missile

b) Fuel Tank
a) F-17 b) F-5A/B

JI\

c) F-5E d) F-80 c) MK-84

Fig. 1 Wing planforms. Fig. 3 Stores used in analyses.

 JULY 1982 WING/STORE FLUTTER 577

For the multivariate analysis to yield useful results, each of It is standard U.S. Air Force practice to use analysis to
the parameters must be adequately represented in the data certify or "clear" stores for flight only after a sufficient
base. With this in mind, three basic stores were selected with amount of theoretical analysis, wind-tunnel test, and flight
three different span positions being used with each store. The flutter test experience is obtained for a particular aircraft. The
stores selected were the AIM missile10'12'14 and launcher, the maximum allowable wing/store operation velocity of the
fuel tank,14 and the MK-8414 bomb (Fig. 3). aircraft is in most cases 85% of the wing/store analytical
Each wing model consists of an elastic axis structural flutter velocity obtained using wing aerodynamics.
dynamics model restrained at the root by pitch and plunge To present the effect of store aerodynamics on wing/store
springs while fixed in the x,y roll and yaw directions. The flutter taking into account the 15% margin, three arbitrary
elastic axis models used for each aircraft are the same as the regions have been established. They are the following:
ones presented in Refs. 7, 10, and 12-14. In each case the 1) 0 to ±7% change in wing/store flutter speed;
structural model was checked against analytical and ex- 2) ±7% to ±15% change in wing/store flutter speed;
perimental natural frequencies and mode shapes given in these 3) ± 15% or greater change in wing/store flutter speed.
references. The nonplanar doublet lattice method, the method In region 1, there is little need to include store
of images, and slender body theory as incorporated in MSC- aerodynamics in the wing/store flutter analysis since the
50A NASTRAN15'21 were used to model the wing/store change in speed is less than half of 15%. About 60% of the
configurations aerodynamically. The aerodynamic model of configurations fall within this region. Region 2 is considered
the F-17 with tip missile is presented in Fig. 4.
The addition of store aerodynamics to the wing/store
flutter analysis can have one of three effects:
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

0.6 First Bending

First Torsion
1) The aeroelastic instability begins as flutter without store \ Second Bending
aerodynamics, and the flutter mechanism is unchanged by
store aerodynamics, but the flutter speed is changed.
2) The aeroelastic instability begins as flutter, but the flutter
mechanism changes because the store aerodynamics causes 0.0
different modes to couple together. This effect normally
causes large variation in flutter speed.
3) The aeroelastic instability changes from flutter to -0.4 <• \
divergence, a pseudostatic instability.
a) No Store Aerodynamics
In the case of divergence, the addition of the store
aerodynamics has increased the flutter speed beyond the wing
divergence speed. This type of change occurred for 2% of the
configurations studied. In each instance where this type of 0.6 First Bending
First Torsion
behavior was noted, the flutter velocity was first increased by X Second Bending

the addition of the store and then increased beyond the wing
divergence velocity by the addition of the store aerodynamics.
These data were not included in the factor analysis since there
was a change in the basic instability mode. This is a discon-
tinuity in the data. For the remaining 98% of the con-
figurations, the aeroelastic instability remained flutter, with
92% of the configurations displaying a continuous behavior -0.4
in that the flutter mechanism was unchanged. However in 6%
of the configurations, the flutter mechanism changed, with
b) With Store Aerodynamics
different modes causing the instability. This mode switching
or discontinuous behavior was observed by Gayman,5 and as Fig. 5 F-5 A/B, damping vs velocity.
noted previously, parameters were selected in an attempt to
model this type of behavior. The velocity-damping curves for 0.6 First Bending
the F-5A/B with a forward mounted underwing tip missile, First Torsion
Second Bending
shown in Fig. 5, illustrates this type of behavior. When the X

aerodynamic center of pressure is moved forward by the

addition of canards, the flutter mechanism changes from first
bending to second bending/torsion with a lower flutter 0.5
0.0 ——I—
velocity. But in the majority of the configurations the flutter
mechanism remained the same with only a shift in the flutter
velocity, as shown for the F-5A/B with a forward mounted tip
missile (Fig. 6). The ratio of the flutter velocity of the -0.4
wing/store configurations vs the flutter velocity of the a) No Store Aerodynamics
wing/store with store aerodynamics included is given for each
configuration in Ref. 23.

0.6 First Bending

First Torsion
Second Bending
X

1.0

• ~ - - - - - vfFA/VC
-0.4 \
b) Store Aerodynamics
Fig. 4 F-17 with tip missile. Fig. 6 F-5 A/B, damping vs velocity.
 578 C.D. TURNER J. AIRCRAFT

the crossover region. Therefore there is a need to carefully configurations; the largest possible value is unity. To aid in
evaluate the requirements for aerodynamic modeling of the determining the key relationships among the parameters, the
store. About 17% of the configurations fall within this eigenvalues of the correlation matrix are found. It is then
region, with 10% of these configurations being non- assumed that the dimension of the solution space, the
conservative. Those configurations falling within region 3 minimum number of parameters necessary to describe the
would require the aerodynamic modeling of the store to data base, is equal to the number of principal components for
obtain valid flutter results. About 23% of the configurations which the eigenvalues are greater than unity. The eigenvectors
fall within this region, with 70% of these configurations being of the eigenvalues which are greater than unity contain the
nonconservative. In summary, the data generated show that parameter correlations. The larger the eigenvalue, the better
18-30% of the total configurations studied would yield the correlation is among the parameters represented in the
questionable flutter results if the store aerodynamics were not eigenvector.
included in the flutter analysis. Assuming a well-behaved-type Owing to the nonlinear influence of system parameters on
wing/store flutter problem, a few general observations flutter velocity, a nonlinear factor analysis was also done on
concerning the effect of store aerodynamics on wing/store the data base to determine which of the parameters showed
flutter could be made from examination of the data base highly nonlinear behavior.
itself. If the mechanism of flutter is second bending, which To accomplish this task, a series of factor analyses were
includes a significant amount of wing torsion motion (first made on the data base. The first set of analyses included all
torsion), then the location of the center of pressure of the the aircraft, but looked at the various stores and store
store is a key parameter. In this particular case, the addition combinations. The eigenvalue/eigenvector of the correlation
of fins to the store would increase the flutter velocity and the matrix was selected in which VF/VFA was a maximum. The
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

addition of canards to the store would decrease the flutter wing/store configurations along with the eigenvalues are
velocity. If the mechanism of flutter is primarily first wing given in Table 1. As stated earlier, the closer the eigenvalue is
bending motion, then the addition of either fins or canards to one, the weaker the correlation. For the eigenvalues in
would decrease the flutter velocity. The discontinuous type of Table 1, the correlation is fairly weak, with the tip missile and
flutter problem is encountered when there is a close underwing store configurations showing the strongest
relationship between first bending and first torsion/second correlation. The results of this set of analyses would indicate
bending. The addition of a fin or canard will shift the center that it is not possible to develop a good general method of
of pressure enough to change the flutter mechanism with determining the effect of store aerodynamics on wing/store
either a large increase or decrease in flutter velocity. flutter using the selected parameters. The next step was to do
a series of factor analyses on each aircraft for all stores.
IV. Multivariate Analysis Again, the eigenvalue/eigenvector of the correlation matrix
was selected in which VF/ VFA was a maximum. The results
A multivariate analysis was performed on the previously are given in Table 2 along with one of the earlier results from
described data base using factor analysis24'25 with selected Table 1. For each of the specific aircraft, the correlation is
parameters. Factor analysis seeks to explain observations in very strong. These results indicate that it is possible to develop
terms of a few important variables and to aid in making
specific guidelines or a prediction technique for each type of
predictions concerning these observations. For these two aircraft.
reasons factor analysis was used to obtain information To develop specific guidelines or a prediction technique, the
concerning the strength of the correlation among the various
eigenvector in which VF/VFA is a maximum would then be
selected parameters, eigenvalues of the correlations matrix,
and the relationship of the parameters for the various
considered. An example of this for F-17 is given in Table 3.
The key parameters are a2, AF2/AB, L CG /C S , 62, AS/AW,
strengths of correlation, eigenvectors of the correlation o> 5 /co 7 as determined from the linear eigenvector. When both
matrix. Thus to be able to make predictions the correlation
the linear and nonlinear eigenvectors are used, the key
must be good between the wing/store flutter speed and a set of
known parameters. parameters are a2, AF2/AB,LCG/CS, 62, AS/AW, w 5 /a) 7 ,
Hw/y, LS3/LS]. For this aircraft the analytical flutter
Having selected the parameters to be used in the factor mechanism is first wing torsion. Thus the effect that the store
analysis, the mean value for each of the selected parameters is aerodynamics has on wing/store flutter is most closely related
determined. to the selected parameters that represent store pitch, store
center-of-pressure movement, fore/aft location of store and
0)

Table 1 Factor analysis of all aircraft for various configurations

where rj equals the parameters w 2 /co 7 ,<5 7 ,a 7 , etc.; and TV are
the observations, wing/store configurations to be included in Wing/store Eigenvalue for which
configurations VF/VFA is a maximum
the analysis.
The difference between each observation and the mean for Tip missile 2.61
each parameter is computed. Missile 1.64
Fuel tank 1.48
x
ji — Xji ~ Xj j = 7,2,...,17 MK-84 1.48
Underwing stores 2.04
i= 1,2,. ..,7V (2) All stores 1.37
The final step is the development of the correlation matrix.
The coefficients of the matrix are given by
Table 2 Factor analysis for each aircraft, all stores
i\
LJ XjiXKi Aircraft Eigenvalue for which
type VFIVFA is a maximum
(3)
F-5A/B 11.77
F-5E 13.29
F-17 38.39
F-80 14.57
The correlation matrix shows the strength of the relationship All aircraft 1.37
between any two parameters for a given number of wing/store
 JULY 1982 WING/STORE FLUTTER 579

Table 3 Factor analysis results of wing model No. 1—all stores flutter analysis. About 23% of the configurations fall within
v
this region.
Linear Quadratic Cubic
Parameter eigenvector eigenvector eigenvector About 75% of the configurations represented in regions 2
and 3 yielded nonconservative flutter results when store
a>2/o> ; 0.312 0.687 0.753a aerodynamics were not included in the analysis.
OJ 5 /CO ; -0.583 -0.776a -0.767 To aid in studying the effects of store aerodynamics on
5; 0.387 0.665a 0.646 wing/store flutter, a factor analysis was made on the data
«7 -0.405 -0.620 0.763a base. Flight test results, wind-tunnel test results, and
*2 -0.663a -0.626 -0.542
0.812 0.994a 0.902 analytical results were used to determine a set of parameters
<*2
-0.283 -0.488 -0.294 to be used in the factor analysis. The parameters represented
*3
«3 0.488 0.574a 0.393 wing/store geometry, mass, modal, flutter, and aerodynamic
ws/ww -0.167 0.686 -0.830a configuration data. The factor analysis was used to determine
OJ F /W ; 0.039 0.009 0.020 if there was any correlation between the parameters and the
coF/4 /co; 0.066 -0.005 -0.062 wing/store flutter speed, thus aiding in developing guidelines
^ W2 IL Wi -0.191 -0.065 0.068 to determine when store aerodynamics would have to be
A$/AW -0.589 -0.379 -0.679a included in the wing/store flutter analysis to obtain valid
LCG1'CS 0.716a 0.294 -0.496 flutter results. The results of the factor analysis indicate that it
Hw/y 0.578 0.849 0.939a
LS1/CS -0.300 -0.617 -0.627a may not be possible to develop general guidelines, but it is
0.174 -0.442 0.454 possible to develop specific guidelines for use with a particular
^S2 I^Sl
-0.869a aircraft.
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

LSS /LS1 0.503 -0.781

LS4/LS1 0.292 -0.373 -0.439 The results of this study are limited to those configurations
AF1/AB -0.133 -0.144 -0.0117 which were used to generate the data base. Future efforts are
AF2/AB 0.812 0.948 0.974a needed to determine the effects on wing/store flutter in-
7/Ls; 0.476 0.130 -0.6423 cluding store aerodynamics for multiple carriage of stores. As
a wing/store flutter data become available for more con-
Selected parameters.
figurations the data base can be updated and the multivariate
analysis can be repeated.

store center of gravity, and overall store size. The selected References
parameters can now serve as a guide or be used in regression !
Goland, M. and Luke, Y.L., ''The Flutter of a Uniform Wing
analysis to develop a technique for predicting when store with Tip Weights," Journal of Applied Mechanics, Vol. 15, No. 1,
aerodynamics should be included in wing/store flutter March 1948, pp. 13-20.
2
analysis for this specific aircraft. Runyan, H.L. and Sewall, J.L., "Experimental Investigation of
the Effects of Concentrated Weights on Flutter Characteristics of a
Straight Cantilever Wing," NACA TN-1594, June 1948.
V. Conclusion 3
Sewall, J.L. and Woolston, D.S., "Preliminary Experimental
This study was the first attempt to do a systematic Investigation of Effects of Aerodynamic Shape of Concentrated
analytical study of the effect of store aerodynamics on Weights on Flutter of a Straight Cantilever Wing," NACA RM
wing/store flutter. To determine this effect, flutter analyses L9E17, July 1949.
4
Andropoulos, T.C., Chee, C.F., and Targoff, W.P., "The Effect
were done on four aircraft with single carriage of three basic of Engine Location on the Antisymmetric Flutter Mode," AF Tech.
store types. In all, 308 configurations were analyzed with and Kept. 6353, Aug. 1951.
without store aerodynamics. These configurations 5
Gayrhan, W.H., "An Investigation of the Effect of a Varying Tip-
represented tip missiles, tip tanks, and underwing stores. Weight Distribution on the Flutter Characteristics of a Straight
Including store aerodynamics in the wing/store flutter Wing," Journal of the Aeronautical Sciences, Vol. 19, May 1952, pp.
analysis can have one of three effects: 289-302.
6
1) The aeroelastic instability begins as flutter without store Sewall, J.L., Herr, R.W., and Igoe, W.B., "Flutter Investigation
aerodynamics, and the flutter mechanism is unchanged by of a True-Speed Dynamics Model with Various Tip-Tank Con-
store aerodynamics, although the flutter speed changes figurations," NACA RM L54119, March 1955.
7
Pollock, S.J. and Cooley, D.E., "Evaluation of Prediction
(continuous behavior, 94% of the configurations). Methods for Aircraft/External Store Flutter Clearance," The Fourth
2) The aeroelastic instability begins as flutter, but the flutter Aircraft/Stores Compatibility Symposium, Fort Walton Beach, Fla.,
mechanism changes because the store aerodynamics causes Oct. 1977.
8
different modes to couple together. This group normally will Cross, A.K. and Albano, E.A., "Computer Techniques for the
have large changes in flutter speed (discontinuous behavior, Rapid Flutter Clearance of Aircraft Carrying External Stores, Part I
4% of the configurations). Perturbation Theory and Application," Air Force Flight Dynamics
3) The aeroelastic instability changes from flutter to Laboratory Rept. AFFDL-TR-72-114, Part I, Feb. 1973.
9
divergence, a pseudostatic instability (discontinuity in the Van Nunen, J.W.G., Roos, R., and Meyer, J.J., "Investigation of
the Unsteady Airloads on Wing-Store Configurations in Subsonic
data, 2% of the configurations). In the case of divergence, the Flow," The Fourth Aircraft/Stores Compatibility Symposium, Fort
addition of the store aerodynamics has increased the flutter Walton Beach, Fla., Oct. 1977.
speed beyond the wing divergence speed. 10
Hwang, C., Winther, B.A., and Ivlills, G.R., "Demonstration of
Three arbitrary regions have been established to present the Active Wing/Store Flutter Suppression Systems," Air Force Flight
effect of store aerodynamics on wing/store flutter taking into Dynamics Laboratory Rept. AFFDL-TR-78-65-, June 1978.
11
account the 15% margin normally used with military aircraft. Reed, W.H., Foughner, J.T., and Runyan, H.L., "Decoupler
1) 0 to ±7% change in flutter speed. There is little need to Pylon: A Simple, Effective Wing/Store Flutter Suppressor,"
include store aerodynamics in the flutter analysis since the AIAA/ASME/ASCE/AHS 20th Structures, Structural Dynamics,
change in speed is less than half of the 15%. About 60% of and Materials Conference, St. Louis, Mo., April 4-6, 1979.
12
"Basic Weight and Stiffness Data for F-5E Modal Vibration
the configurations fall within this region. Analysis," Northrop Corporation Rept. NOR 78-111, Aug. 1978.
2) ±1 to ±15% change in flutter speed. There is a need to 13
Arthurs, T.D., "Parametric Flutter Clearance of Wing Tip
carefully evaluate the requirements for aerodynamic modeling Missiles on the F-5 Series Aircraft," Aerospace Flutter and Dynamics
of the store. About 17% of the configurations fall within this Council, San Antonio, Texas, April 27, 1978.
14
region. Clyburn, R.C., "Certification of AIM-9 Matrix on F-5A/B
3) ± 15% or greater change in flutter speed. To obtain valid Aircraft CAN 1634 Flutter Analysis Report," Northrop Corporation
flutter results the store aerodynamics must be included in the Rept. NOR 78-112, Aug. 1978.
 580 C.D. TURNER J. AIRCRAFT

15 22
MacNeal, R.H., ed., "MSC/NASTRAN Theoretical Manual," Rodden, W.P., Harder, R.L., and Bellinger, E.D., "Aeroelastic
The MacNeal-Schwendler Corporation. Addition to NASTRAN," NASA Contractor Rept. CR-3094, March
16
McCormick, C.W., ed., "MCS/NASTRAN User's Manual," 1979.
The MacNeal-Schwendler Corporation, May 1976. 23
17
"MSC/NASTRAN Aeroelastic Supplement," The MacNeal- Turner, C.D., "The Effect of Store Aerodynamics on
Schwendler Corporation. Wing/Store Flutter," The Fifth Aircraft/Stores Compatibility
18
Wall, S.E., ed., "MSC/NASTRAN Programmer's Manual," Symposium, St. Louis, Mo., Sept. 9-11, 1980.
The MacNeal-Schwendler Corporation. 24
Harman, H.H., Modern Factor Analysis, The University of
19
"The NASTRAN Theoretical Manual," NASA SP-221(04), Chicago Press, Chicago, 1976.
1978. 25
20 Gregg, L.W. and Pearson, R.G., "Factor Analysis of Lightplane
"The NASTRAN User's Manual," NASA SP-222(04), 1978.
21 Accident Impact and Damage Variables," TREC-TR-61-122, Aug.
"The NASTRAN Programmer's Manual," NASA SP-223(04),
1978. 1961.
Downloaded by RMIT UNIV BUNDOORA on September 21, 2013 | http://arc.aiaa.org | DOI: 10.2514/3.57431

From theAIAA Progress in Astronautics and Aeronautics Series

ALTERNATIVE HYDROCARBON FUELS:
COMBUSTION AND CHEMICAL KINETICS—v. 62
A Project SQUID Workshop

Edited by Craig T. Bowman, Stanford University

and J&rgen Birkeland, Department of Energy

The current generation of internal combustion engines is the result of an extended period of simultaneous evolution of
engines and fuels. During this period, the engine designer was relatively free to specify fuel properties to meet engine per-
formance requirements, and the petroleum industry responded by producing fuels with the desired specifications. However,
today's rising cost of petroleum, coupled with the realization that petroleum supplies will not be able to meet the long-term
demand, has stimulated an interest in alternative liquid fuels, particularly those that can be derived from coal. A wide
variety of liquid fuels can be produced from coal, and from other hydrocarbon and carbohydrate sources as well, ranging
from methanol to high molecular weight, low volatility oils. This volume is based on a set of original papers delivered at a
special workshop called by the Department of Energy and the Department of Defense for the purpose of discussing the
problems of switching to fuels producible from such nonpetroleum sources for use in automotive engines, aircraft gas
turbines, and stationary power plants. The authors were asked also to indicate how research in the areas of combustion, fuel
chemistry, and chemical kinetics can be directed toward achieving a timely transition to such fuels, should it become
necessary. Research scientists in those fields, as well as development engineers concerned with engines and power plants, will
find this volume a useful up-to-date analysis of the changing fuels picture.

463pp., 6x9illus., $20.00Mem., $35.00List

TO ORDER WRITE: Publications Dept., AIAA, 1290 Avenue of the Americas, New York, N. Y. 10019