Professional Documents
Culture Documents
Formerly with Bombardier Aerospace; Currently Loads and Dynamics Analyst with
L-3 Communications, 7500 Maehr Drive, Waco, Texas 76715, USA
2
Manager, Loads & Dynamics, Bombardier Aerospace
400 Cote-Vertu Road West
Dorval, Quebec, H4S 1Y9, Canada
3
Principal Engineering Specialist, Loads & Dynamics, Bombardier Aerospace
4
Bombardier Aerospace (Retired)
Summary: A DLM-based aerodynamic simulator for flutter is used to identify some of the
most important aerodynamic drivers for the T-Tail flutter mechanism of a complete aircraft.
The simulator is using the Modal Descrambling Factoring Method, which permits individual
variations of each direct and each interference aerodynamic force, moment and hinge
moment independently of any other force or moment. The sensitivity of the flutter solution
to individual variations of very large numbers of direct and interference aerodynamic
derivatives can be studied with ease.
LIST OF SYMBOLS
a1lm , a2lm , a3lm , a4lm = lift correction factors for the modal descrambling factoring method for
direct and interference forces and moments, l , m 1, n
c = reference chord; also denotes streamwise camber deformation; apparent from the context
ChEh = elevator hinge moment due to horizontal stabilizer h (roll/bending)
ChE = elevator hinge moment due to horizontal stabilizer (torsion)
ChE(h+) = elevator hinge moment due to horizontal stabilizer h (roll/bending) AND
(torsion)
ChE = elevator hinge moment due to vertical fin yaw (torsion)
ChEE = elevator hinge moment due to elevator rotation
ChER = elevator hinge moment due to rudder rotation
ChR = rudder hinge moment due to yaw ( or V. Fin torsion)
1
1 INTRODUCTION
The current paper is a revision of the material contained in Reference [1]. The discussion
here applies to all T-Tails analyzed with linear unsteady aerodynamic methods such as the
Doublet Lattice Method (DLM, Reference [2]) with factoring.
Aerodynamic factoring is important; in Reference [3] it is shown that a flutter analysis of an
airplane under development without using factoring for the aileron and tab aerodynamic
hinge moments, when either 2-D strip theory or DLM aerodynamics is used, may not
predict flutter and in this particular case has led to an in-flight flutter incident. Intersecting
(therefore interfering) lifting surfaces pose additional problems for the flutter analyst. FAA
Advisory Circular AC No. 25.629.1A recommends that interference effects be included in
flutter analyses. Parametric variations of calculated aerodynamic forces and moments are
recommended in order to cover uncertainties in calculated values. These parametric
variations are achieved through factoring.
It is known that the T-Tail flutter mechanism (generally referred to as the vertical fin
bending/ torsion mechanism) is strongly influenced by the unsteady aerodynamic forces and
moments present on the horizontal stabilizer. See Reference [4], where the effect of steady
upload on horizontal stabilizer on T-Tail flutter speed is appended to the MSC Nastran
flutter solution (Reference [5]) and an early attempt to separate and factor differently the
horizontal stabilizer direct lift due to horizontal stabilizer pitch/torsion from the interference
lift due to vertical fin torsion on the horizontal stabilizer of a T-Tail is described.
It is also known (Reference [6]) that the unsteady aerodynamic forces and moments present
on the horizontal stabilizer of a T-Tail consist of a superposition of direct forces and
moments resulting from the rigid and elastic motions of the horizontal stabilizer and its
control surfaces and tabs and of interference forces resulting from the oscillating vertical fin
and its control surfaces and tabs. The superimposed direct and interference aerodynamic
forces and moments on the horizontal stabilizer are called stacked forces.
Thus, for general motion of a given arrangement of n lifting surfaces residing in the same
interference group, every lifting surface experiences 1 direct set of forces and moments and
n-1 sets of interference forces and moments. For n>2, each lifting surface experiences more
interference forces than direct forces!
Figure 1 shows a general (or scrambled) mode of vibration of a T-tailed airplane at the
aerodynamic surface.
Clh
Cl
ClhVF
Cl
ClR
ChEE
ChEh
ChE
ChE
ChER
CYVF
CYVFR
ChR
ChRR
For every mode of vibration, rigid or elastic, the Modal Descrambling factoring method
(formerly known as the General Aerodynamic Derivatives Factoring Method, Reference [6])
performs the descrambling of the general modal motion at every aerodynamic strip and
replaces each scrambled mode with 5 simpler and always the same modes: heave (or
bending), pitch (or torsion), control surface rotation, tab rotation and elastic streamwise
camber deformation. Unsteady direct and interference aerodynamic forces and moments
distributions are then calculated for the descrambled modes and they are available for
factoring separately at every aerodynamic strip (Figure 2).
35
Real Cp
30
Reference Plane
25
20
15
10
5
0
0
0.2
0.4
x/c
0.6
0.8
=
Real 2-D Theodorsen Cp Due to Main Surface Heave;
k=0.372; Calculate Lh, Mh, Th, Qh
5
Real Cp
Reference Plane
a1*Lh
3
2
1
0
Reference Axis
-1
0
0.2
x/c
0.4
0.6
0.8
+
Real 2-D Theodorsen Cp Due to Main Surface Pitch;
k=0.372; Calculate L, M, T, Q
35
Real Cp
30
Reference Plane
25
a2*L
20
15
10
5
0
0
0.2
0.4
x/c
0.6
0.8
+
Real 2-D Theodorsen Cp Due to Control Surface Rotation;
Hinge Line at 70% Chord; No Aerodynamic Balance
Surface; k=0.372; Calculate L, M, T, Q
14
12
Real Cp
Reference Plane
a3*L
10
8
6
4
2
0
0
0.2
x/c
0.4
0.6
0.8
+
Real 2-D Theodorsen Cp Due to Tab Rotation; Hinge Line
at 85% Chord; No Aerodynamic Balance Surface; k=0.372;
Calculate L, M, T, Q
15
Real Cp
Reference Plane
12
a4*L
9
6
3
0
0.2
0.4
x/c
0.6
0.8
Figure 2. Modal Descrambling at Typical Strip of Aerodynamic Surface. Factoring Shown Only for Direct
Lift. Elastic Streamwise Camber Deformation not Shown.
Experimental evidence shows that the factors required for direct and interference
aerodynamic forces and moments are different from each other. For a DLM aerodynamic
analysis without modal descrambling the user has access only to the stacked forces and
moments for factoring. As shown in this paper, this may lead to inaccurately calculated
flutter speeds.
5
w w w w w w
j
hj
cj
(1)
with the scrambled downwash w j being the sum of the descrambled downwash vectors as
defined for the modal descrambling factoring scheme (Figure 2). The pressure coefficients
and integrated forces and moments can then be obtained for the scrambled or descrambled
modes:
Cpi Ai j w j
1
(2)
with i,j=1,NBOX.
If Equation 2 operates on the scrambled modes, we have the typical aerodynamic analysis
which calculates stacked aerodynamic forces and moments and only the stacked forces will
be available for factoring; it will be later seen that this is incorrect. If Equation (2) operates
on the descrambled modes of Equation (1), we have access to the descrambled aerodynamic
forces and moments, but not yet to the interference aerodynamic forces and moments.
In order to calculate the interference aerodynamic forces and moments, the entire aircraft
descrambled downwash vectors are then partitioned into n components. This number is
generally fairly small, such as 4 for the airplane analyzed: (1) wing with associated control
surfaces and tabs, (2) horizontal stabilizer with associated control surfaces and tabs, (3)
vertical fin with associated control surfaces and tabs and (4) engines, pylons, interference
body and ventral fins.
(3)
where l , m 1, n .
The stacked factored lift at any aerodynamic strip of component l is:
L' (h c)lstacked, factored
n
m1
L(h)
lm
(4)
where as before, l , m 1, n and a1lm , a2lm , a3lm , a4lm are the direct and interference factors on
the lift, respectively. For the time being, the forces and moments due to camber deformation
are not factored, due to a lack of experimental or CFD guidance. Also for the time being,
the real and imaginary forces and moment distributions are factored by the same real factors
derived from steady state wind tunnel and CFD calculations. If unsteady experimental or
CFD guidance is available, the imaginary part can easily be factored differently from the
real part.
Every one of the direct and interference aerodynamic forces and moments for all the lifting
surfaces can be independently factored with pin-point accuracy, including far-fetched
interference effects such as the vertical fin side force due to wing tip aileron tab deflection
or the elevator hinge moment due to wing camber, etc.
Thus, the Modal Descrambling factoring method and simulator program gives the user
unprecedented control over all direct and all interference forces for factoring, on any lifting
surface of the aircraft.
(5)
(6)
However, is the following true for the factored DLM including interference effects?
(7)
3 NUMERICAL RESULTS
Three configurations of a Bombardier stick model of a twin engine airplane with manual
control surfaces are analyzed: Model No. 1 has nominal vertical fin bending and torsional
stiffnesses; Model No. 2 has 25% higher vertical fin bending and torsional stiffnesses than
the nominal and Model No. 3 has 50% higher vertical fin bending and torsional stiffnesses
than the nominal configuration. As such, all three models are versions with much reduced
vertical fin stiffnesses of a real airplane model.
Each aileron has a geared tab, the left aileron has a trim tab and the rudder also has a trim
tab. The ailerons, elevators and the rudder all have aerodynamic balance surfaces. Mass
balance is present on all three control surfaces. The control surfaces mechanical circuits are
modeled to include the mass and inertia properties of the control wheels, control columns,
quadrants, cables and pushrods and the elastic properties of the cables and pushrods.
The aerodynamic model of the airplane under consideration is partitioned into four
components as follows: component 1: wing with its ailerons and tabs; component 2:
horizontal stabilizer with its elevators; component 3: vertical fin with its rudder and tab and
the fourth component consists of the pylons, engines and ventral fins. Aerodynamic
correction factors are obtained from steady state wind tunnel data and CFD calculations.
The controls-free configuration is analyzed at one transonic Mach Number. The
aerodynamic model is common for all three structural dynamic models. The effects on
calculated flutter speed of vertical fin bending and torsion stiffness increases and of
variations of aerodynamic derivatives can thus be studied and displayed in a compact format
in parametric fashion.
0.006
CN DUE TO HS TORSION
0.004
Real Cn
-0.008
-0.010
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
semispan
Figure 3. Real Parts of Unfactored Descrambled Direct and Interference C n on Horizontal Stabilizer of the
T-Tail for General Mode of Vibration Shown in Figure 1. k=0.700.
Imaginary C n
0.020
0.015
0.000
-0.005
0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
semispan
Figure 4. Imaginary Parts of Unfactored Descrambled Direct and Interference Cn on Horizontal Stabilizer
of the T-Tail for General Mode of Vibration Shown in Figure 1; k=0.700.
Cn*c/cave
CFD
DLM
-1.0
-0.8
-0.6
-0.4
-0.2
0.0
0.2
0.4
0.6
0.8
1.0
semispan,
Figure 5. Typical DLM vs. CFD Horizontal Stabilizer Spanwise Cl Distribution. k=0.000.
The factor for the Cl at every spanwise strip is obtained by dividing the CFD-calculated
value by the DLM-calculated value. At the Mach Number analyzed, the CFD prediction is
larger than the DLM prediction everywhere along the span.
Figure 6 shows a typical variation with Mach Number of the total horizontal stabilizer CL
(interference) factor and of the CL (direct) factor.
10
Aerodynamic Factors
CL Factor>1.000
1.000
CL Factor<1.000
CL Factor vs. M
CL Factor vs. M
Analysis CL Factor
Analysis CL Factor
CL Factor<1.000
Mach Number M
Figure 6. Typical CL and CL Total Factors vs. Mach Number for the Horizontal Stabilizer; k=0.000
At all Mach Numbers, the CL factor is smaller than unity. At low Mach Numbers, the CL
factor is typically greater than unity (and greater than the CL factor) and it decreases with
increasing Mach Number until it crosses the CL factor and continues to decrease below the
CL factor. For the Mach Number analyzed here, the CL factor is greater than the CL
factor.
Figures 7 and 8 show the factored and unfactored, real and imaginary, total direct, total
interference and the stacked factored and unfactored real and imaginary lifts on the
horizontal stabilizer at k=0.700. Note the different shapes of the related factored and
unfactored lifts, as well as the relative magnitudes of the real and imaginary parts.
TOTAL UNFACTORED AND FACTORED DIRECT, INTERFERENCE AND
STACKED REAL Cn ON HORIZONTAL STABILIZER OF T-TAIL FOR GENERAL
(SCRAMBLED) MODE OF VIBRATION ; k=0.700
0.015
0.010
Real C n
0.000
UNFACTORED STACKED
FACTORED STACKED
-0.005
-0.010
0.0
0.2
0.4
0.6
0.8
1.0
semispan
Figure 7. Real Parts of Total Unfactored and Factored Descrambled Direct and Interference Stacked Cn
on Horizontal Stabilizer of the T-Tail for General Mode of Vibration Consisting of Vertical Fin
Bending/Torsion/HS Roll/Elevator Rotation/Rudder Rotation Shown in Figure 1. k=0.700.
11
0.025
Imaginary C n
0.020
0.015
0.010
UNFACTORED STACKED
0.005
FACTORED STACKED
0.000
-0.005
0.0
0.2
0.4
0.6
0.8
1.0
semispan
Figure 8. Imaginary Parts of Total Unfactored and Factored Descrambled Direct and Interference
Stacked Cn on Horizontal Stabilizer of the T-Tail for General Mode of Vibration Shown in Figure 1.
k=0.700.
Flutter solutions for each stiffness model have been run using the common aerodynamic
model. First, the flutter solutions with no aerodynamic factoring have been performed.
Then, the flutter solutions with nominal aerodynamic factoring have been run. Then,
flutter solutions are run with parametric variations of various aerodynamic derivatives
factors from the nominal values. One derivative factor at a time is varied from nominal and
its effect on the flutter solution is noted.
Table 1 shows all the flutter solutions calculated with and without aerodynamic factoring
and all the variations of the aerodynamic derivatives factoring. The primed derivatives
indicate nominal factored values. The magnitude of the up-or-down excursions from
nominal derivatives values is between 20% and 50% as indicated in Table I.
Cases Ran
(Derivatives factoring)
No Factoring
Nominal Factoring
Cl factors=Cl factors
ChEh*0.5
ChE(h+)*0.5
ChEE*1.2
ClR Theoretical
ChR*0.5
ChER*0.5
ChRR*1.2
ClhVF Theoretical
ChE*1.2
Cl Theoretical
Nominal, camber discarded
CYVF*1.2
%(Vf -VfREF)/VfREF
Model 1
%(Vf -VfREF)/VfREF
Model 2
%(Vf -VfREF)/VfREF
Model 3
65.979
0.000
65.456
0.826
3.220
1.899
7.570
-1.596
10.322
-1.817
0.000
2.119
2.835
2.009
-1.294
109.716
11.313
137.462
14.368
21.002
14.588
23.066
15.001
36.113
10.542
11.891
16.598
16.846
15.029
11.616
149.243
126.452
163.584
116.020
97.578
92.678
139.994
134.132
131.489
135.701
130.003
128.269
124.993
125.048
125.048
Table 1. Cases Ran; VfREF Is for the Nominal Factoring Case, Model #1; Primed Derivatives Are Factored
Nominal Values; Derivatives Factoring Variations Are from Nominal Values.
12
When a derivative is listed such as ClR Theoretical in Table 1, it means that the
unfactored DLM-calculated value is used.
Figure 9 shows in graphical form the effect of both stiffness (and therefore mode shape)
variation and aerodynamic derivatives variation on the flutter mechanism of the T-Tail of all
three models analyzed. For clarity, not all the cases listed in Table 1 are plotted in Figure 9.
Figure 9. Effect of Vertical Fin Stiffness Variations and of Aerodynamic Factoring Variations, One
Derivative at a Time from Nominal on Flutter Speed of T-Tailed Aircraft. Primed Derivatives Are
Factored Nominal Values. Reference Speed Is for Nominal Factoring, Model 1.
It is immediately apparent that the change in the flutter solution due to changes in the Cl
factors is dramatic; at the Mach Number analyzed, the T-Tail flutter speeds calculated with
the Cl factors = Cl factors are decidedly optimistic (see Table 1 and Figures 6 and 9).
Figures 10 and 11 show the V-g and V-f plots for the complete aircraft for the nominal
factoring case (with the Cl factors > Cl factors) and then for the nominal factors but with
the Cl factors being equal to the Cl factors for Model No. 1. Again, the change in the TTail flutter speed due to changes in the Cl factors is dramatic.
13
0.00
damping, g
-0.05
-0.10
-0.15
-0.20
Vertical Fin Torsion
-0.25
Rudder Rotation
-0.30
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
V (KEAS)/Vmax
25
Frequency (Hz)
20
15
10
5
Vertical Fin Bending
Rudder Rotation
0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
V (KEAS)/Vmax
Figure 10. V-g and V-f Plots of Model No. 1 T-Tail Flutter Solution for Nominal Aerodynamic Factoring;
Cl factors > Cl factors
14
0.05
0.00
damping, g
-0.05
-0.10
-0.15
-0.20
-0.25
Rudder Rotation
-0.30
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
V (KEAS)/Vmax
Frequency (Hz)
20
15
10
0
0.0
0.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1.0
V (KEAS)/Vmax
Figure 11. V-g and V-f Plots of Model No. 1 T-Tail Flutter Solution for Nominal Aerodynamic Factoring;
Cl factors = Cl factors.
15
Better
The Best
Figure 12. A Summary of the Aerodynamic Forces and their Phases on the Horizontal Stabilizer Affecting
T-Tail Flutter; No Control Surfaces; Positive Yaw Is Assumed.
4 CONCLUSIONS
The scrambled vertical fin bending/torsion/horizontal stabilizer roll/elevator rotation/rudder
rotation mode at the aerodynamic surface (Figure 1) illustrates the complexity of the modal
motion participating in the T-Tail flutter mechanism.
The T-Tail flutter mechanism speed is strongly influenced by the bending and torsional
stiffness levels and motions of the vertical fin; this effect is known and is included here for
completeness. From the aerodynamic standpoint, the T-Tail flutter mechanism has been
shown to be dependent on a large number of aerodynamic forces, moments and control
surface hinge moments present on the horizontal stabilizer for the bending and torsional
modes of the vertical fin. Among these aerodynamic forces and moments, the Cl
interference factor is the main aerodynamic driver of the T-Tail flutter mechanism.
16
5 ACKNOWLEDGEMENTS
Messrs. Frederic Bradley and Jason Bensimhon of the Bombardier Dynamics Group have
contributed with the preparation of the structural dynamic model and with the voluminous
aerodynamic data compilation required for this project.
17
18