You are on page 1of 18

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/271366437

Active Alleviation of Gust Loads Using Special Control Surfaces

Conference Paper · May 2006


DOI: 10.2514/6.2006-1833

CITATIONS READS
8 443

6 authors, including:

V. R. Feldgun Fernando Rosich


Technion - Israel Institute of Technology Airbus
101 PUBLICATIONS 1,096 CITATIONS 2 PUBLICATIONS 13 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by V. R. Feldgun on 22 November 2015.

The user has requested enhancement of the downloaded file.


47th AIAA/ASME/ASCE/AHS/ASC Structures, Structural Dynamics, and Materials Confere AIAA 2006-1833
1 - 4 May 2006, Newport, Rhode Island

Active Alleviation of Gust Loads Using Special Control Surfaces

M. Karpel‡, B. Moulin* and V.Feldgun†


Technion - Israel Institute of Technology, Haifa 32000, Israel

and

L.Anguita§, F.Rosich¶ and H. Climent#


EADS-CASA, 28906 Madrid, Spain
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

The alleviation of dynamic gust loads on a transport aircraft using wing-mounted control surfaces is
investigated. Three different wing-mounted control surfaces are considered: symmetrically-actuated
ailerons, under-wing forward-positioned control surfaces at about 0.8 of the wing span, and wing-tip
forward-positioned control surfaces. The main investigation is for a 1:10 cable-mounted wind-tunnel
model of a transport aircraft that was constructed and tested at the TsAGI laboratories. The length of
a one-minus-cosine vertical gust velocity profile is tuned to yield maximal wing-root bending moment.
All the control laws are based on simple low-pass filters for easy and robust application in the wind
tunnel. Each is based on a single input of a wing-tip accelerometer which is shown to react in time to
allow the alleviation of the peak of the wing-root bending moment. All the three control means are
shown to alleviate the extreme wing-root moments by 9-16% and the wing-tip accelerations by 26-33%
at intermediate design flight velocities. The effects on section-loads envelopes at monitoring stations
along the wing are also very favorable. The wing-tip and under-wing controls are found to be more
effective than the existing aileron at these speeds. The ailerons become even less effective at higher
speeds due to reduction in their aeroelastic effectiveness. The application of the same control concept
to the full-scale aircraft model exhibits similar results. The effects of the designed control laws on the
statistical response of the wing loads to continuous gusts are shown to be similar to their effects on
discrete gusts.

I. Introduction
The design of modern flight vehicles requires the evaluation of dynamic loads in response to discrete and
random gust excitations1. Aviation regulations define design discrete gust profiles in a deterministic manner
and continuous gust profiles in a stochastic manner. Control system can be used to reduce the dynamic
response to gusts and by that alleviate critical design loads and improve the ride comfort and equipment
functionality.

Common gust response procedures that do not conveniently include control system effects, such as that of
MSC/NASTRAN,2 are based on second-order frequency-domain formulation. Modal frequency response
functions are used in these procedures to calculate discrete transient gust response, via Fourier and Inverse-
Fourier transforms. The main advantage of the frequency-domain approach is that it is based on oscillatory
generalized aerodynamic force coefficient (AFC) matrices that can be generated by common, well
established, commercially available procedures such as the Doublet-Lattice method implemented in
MSC/NASTRAN2, and the constant pressure panel method implemented in ZAERO3. A disadvantage of
most existing frequency-domain procedures is that they are not conveniently adaptable to the consideration of


Professor, Sanford Kaplan Chair for Aerospace Engineering, Faculty of Aerospace Engineering, Associate
Fellow AIAA
*
Senior Researcher, Faculty of Aerospace Engineering , Member AIAA

Researcher, Faculty of Aerospace Engineering
§
Senior Aeroservoelastic Specialist, Aeroelasticity and Structural Dynamics Department, John Lennon s/n,
Getafe

Aeroservoelastic Specialist, Aeroelasticity and Structural Dynamics Department, John Lennon s/n, Getafe
#
Head, Aeroelasticity and Structural Dynamics Department, John Lennon s/n, Getafe

1
Copyright © 2006 by Moti Karpel and Boris Moulin. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
aeroservoelastic (ASE) control system effects on the structural design gust loads, and to the inclusion
aeroelastic gust-response considerations in the control-system design.

A common approach to ASE analysis is based on the conversion of the second-order aeroelastic equation of
motion into first-order, state-space constant-coefficient time-domain equations of motion, which requires the
approximation of the AFC matrices by rational functions in the Laplace domain. Rational approximation
techniques such as the term-by-term Roger's approximation4 and the more complicated but more efficient
minimum-state method5 are well established and commercially available. However, their application might
involve some problems such as degraded accuracy, excessive number of augmenting states or the need for
careful parameterization, namely loss of process robustness.

Two different formulation approaches to ASE gust-response analyses were developed, evaluated and
compared in Ref. 6, and implemented in the gust-response module recently integrated in the ZAERO
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

package.3 The first is a first-order frequency-domain approach which accommodates control systems in their
most general state-space form, but with the AFC matrices still in their transcendental frequency-domain form.
This option was intended for being used in industrial gust loads evaluation processes. The second approach is
based on state-space time-domain formulation which requires rational approximations of the generalized AFC
matrices. This option was intended for cases where time-domain simulations, possibly with non-linear
effects, are required, and for the inclusion of the aeroelastic system in the control design process.

The work presented in this paper was performed within the Active Aeroelastic Aircraft (3AS) 5-th
Framework project of the European Commission for exploring the possibilities of using aeroelastic properties
for obtaining better design of flight vehicles. A previous work in this project7 was the development of gust-
loads alleviation control systems using specially designed wing control surfaces for demonstration on the 5.3-
meter-span cable-mounted EuRAM aeroelastic wind-tunnel (WT) model. The purpose of the work in this
paper was to compare the performance of the various control options in alleviating wing design loads and
acceleration levels. The comparison was performed for the full-scale (FS) and the WT EuRAM aircraft
model.
.

Figure 1: EuRAM wind-tunnel model at TsAGI facilities

2
II. Frequency-Domain Formulation
The Laplace transform of the open-loop aeroelastic equation of motion in generalized coordinates, excited by
control surface motion and atmospheric gusts, is

M hh s 2 Bhh s+ K hh q Qhh ( s ) { ( s )}
(1)
q
M hc s 2 q Qhc ( s ) c ( s) QhG ( s ) wG ( s )
V
where s is the Laplace variable t, q is the dynamic pressure, V is the air velocity, the left-hand-side matrix
coefficient matrices are the generalized mass, damping, stiffness and AFC matrices associated with modal
displacements { ( s )} , and the right-hand-side matrices are the generalized mass and aerodynamic forcing
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

matrices due to control surface commanded deflections c ( s ) and gust velocity vector wG ( s ) . The
AFC matrices Q hh (s ) , Q hc (s ) and QhG ( s ) can be calculated by unsteady aerodynamic codes at
various tabulated reduced-frequencies k= L/V values along the imaginary axis of the non-dimensional
Laplace variable p sL / V g ik , where L is a reference length and V is the true flight velocity.
To allow control systems of the most general forms, a first-order state-space formulation was suggested in
Ref. 6 where the aerodynamic matrices keep their frequency-dependent form. The frequency-domain version
of these equations of motion with control and gust excitation is
i x ae (i ) A ae (i ) x ae (i ) Bae (i ) u ae (i ) Baw (i ) wG (i ) (2)
where

c
0 I
xae , uae i c , Aae 1 1 ,
i 2
M hh K hh q Qhh (ik ) M hh Bhh
c

0 0 0 0
Bae 1 1 , Baw q 1
q M hh Qhc (ik ) 0 M hh M hc M hh QhG (ik )
V

The augmentation of the actuator states3,5 yields the open-loop “plant” equations excited by the actuator
inputs { c}. With state-space representation of the control system, the uncoupled plant and control equations
can be expressed as:
Xp Ap (i ) 0 X p Bp 0 up Bpw (i )
i wG
Xc 0 Ac Xc 0 Bc uc 0 (3)
yp C p (i ) 0 Xp 0 0 up CGp (i )
wG
yc 0 Cc Xc 0 Dc uc 0
T T T T T
where the plant state vector is X p s s s2 T
, which includes the control actuator states and
does not have aerodynamic lag states, and {Xc} is the vector of states of the control system.

The ASE loop is closed by the application of the gain matrix


up G pp G pc yp
(4)
uc Gcp Gcc yc
The resulting closed-loop vehicle equation is
i Xv Av (i ) Xv Bvw (i ) wG (5)

3
T
where Xv X Tp X cT . Modal response analysis to sinusoidal gust excitation can be performed by
solving Eq. (4) while interpolating for the AFC matrices in Eq. (2) from the tabulated ones. A robust method
for solving the aeroservoelastic response to discrete gusts is by using Fourier transforms6. The input gust
excitation in Eq. (5) is replaced by its Fourier transform, frequency-domain modal response is calculated, and
response parameters of interest are then recovered from the modal response and transformed to the time
domain by inverse Fourier transforms.

III. State-Space Time-Domain Equations of Motion


The time-domain ASE model for gust response analysis is based on the ASE model of Ref. 5, augmented by
gust states and excited by a gust velocity inputs. The formulation is based on the gust modeling of Ref. 6.
The tabulated aero matrices are first used for approximating the AFC matrix as a rational function of k in the
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

entire frequency domain. An expansion to the entire Laplace domain is performed by replacing ik in the
rational expression by the non-dimensional Laplace variable p. The resulting Laplace-domain approximation
of Qh Qhh Qhc QhG is
1
L L2 V
[Qh ( s )] [ A0 ] [ A1 ]s 2
[ A2 ] s 2 [ D ] [ I ] s [ R ] [ E ]s (6)
V V L !
where the [ Ai ] and [ E ] matrices are column partitioned as
[ Ai ] [ Ahhi Ahci AhGi ] , (i = 0, 1, 2), [ E ] [ Eh Ec EG ]
The resulting vehicle time-domain equation of motion, with the control system augmented, becomes
{xv } [ Av ]{xv } [ Bv ]{uv } [ Bvw ]{wG }
(7)
{ yv } [Cv ]{xv } [ Dv ]{uv } [CGv ]{wG }
wG
where {wG }
wG
The equation of motion of the closed-loop aeroservoelastic system is, similarly to Eq. (5), of the form
{xv } [ Av ]{xv } [ Bvw ]{wG } (8)
Output parameters which are not used by the control system but are of interest for dynamic response analysis
can be defined as a separate vector { yR } and expressed in terms of the plant states and the gust inputs
yR Cv xv CGv wG (9)

The frequency-domain and the time-domain approaches are applied in this study. The first method is more
robust because the AFC matrices are still kept in the transcendental frequency-domain form while the second
one requires rational approximation. Even though the rational approximation techniques are well developed
and usually give excellent results,3,5,6,7 they might cause some inaccuracies in massive industrial applications.
Hence, the frequency-domain approach is used in this study for gust-response analyses of the open-loop ASE
systems, and the state-space approach is applied for the generation of the ASE plant model for control-system
design using Matlab. The gust-response loads analysis with the resulting closed-loop system is performed
using the frequency-domain approach again.

IV. Design Loads


Net loads at extreme gust cases for structural design are often the primary purpose of the gust response
analysis. Two methods are widely used for calculating dynamic loads8, the mode-displacement (MD) method
and the summation-of-forces (SOF) method which is based on the classic mode-acceleration method. The
MD method, that recovers the loads directly from the modal displacements, is more suitable for loads-
alleviation control design. When used carefully in fairly well-distributed loads such as in gust excitation

4
cases and with a sufficient number of modes taken into account, it can be used for calculating the design loads
as well.8 The MD approach assumes that the modal superposition assumption used for the construction of the
generalized aeroelastic equations of motion can also be used to recover the net-load distributions. The modal
superposition assumption is
ug (t ) "gh (t ) (10)
where "gh is the matrix of modal displacements. This assumption implies that the vector of net
aerodynamic plus inertial loads expressed at the structural grid points is
FMD (t ) K gg "gh (t ) (11)
where K gg is the discrete-coordinate stiffness matrix. The net loads can be integrated for obtaining
section loads by the application of integration modes that can be calculated from the model geometry5.
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

Theses modes, "gL , are often referred to as "load modes". The resulting section loads, such as shear force
, bending moment and torsion, Are calculated by
T
LMD (t ) "gL K gg "gh (t ) (12)
T
where the product "gL K gg "gh can be prepared when the model is constructed.

V. Aeroelastic Models
The aircraft model with winglets serves as the baseline configuration for this study. The gust-alleviation
control of the baseline configuration is performed by using the existing main ailerons activated
symmetrically. This configuration is referred to in this paper as the Basic-Main-Aileron (BMA)
configuration. The two other configurations in this study are one with underwing control surfaces (UWC)
located forward of the wing planform of about 80% of the span, and one with wing-tip control surfaces
(WTC) also located forward of the wing (see below).

The structural beam-type MSC/NASTRAN model of the full 1:10 wind-tunnel model and the more detailed
model of the full-scale aircraft were generated by TsAGI and updated to match ground vibration test (GVT)
results. The baseline structural MSC/NASTRAN and aerodynamic ZAERO half-models of the FS aircraft
and the WT model are presented in Figures 2-5. The structural and aerodynamic models of the UWC
configuration are shown in Figures 6-9 and the models of the WTC configuration are presented in Figures 10-
13.

Z
X Z
Y X
Figure 2: Baseline WT structural model (BMA) Y Figure 3: Baseline FS structural model (BMA)

Z
X Z
Y
Figure 4: Baseline WT aerodynamic model X
Y
Figure 5: Baseline FS aerodynamic model

5
Y

Z 85
X
Z

X
19
Y

Z Figure 6: UWC WT structural model X


Y
Figure 7: UWC FS structural model
X
Y
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

Figure 8: UWC WT aerodynamic model X


Y
Figure 9: UWC FS aerodynamic model
Z

X
Y

X
19
Y

Z Figure 10: WTC WT structural model X


Y
Figure 11: WTC FS structural model
X
Y

Z
Z Figure 12: WTC WT aerodynamic model
X Figure 13: WTC FS aerodynamic model
Y
X
Y

6
The gust vertical velocity is defined by the "1-cos" profile

wg 2# t
wg (t ) 1 cos (13)
2 Lg !

where wg is the maximal gust velocity, wg =1m/s in this study for the WT model and 25.13m/s for the FS
aircraft, and Lg is the gust length, in terms of the time for a point in the aircraft to travel across it. Open-loop
gust response analysis was first performed for the WT BMA configuration for various gust length values at
V=25m/s, sea level. This analysis was performed using the frequency-domain option of ZAERO with 35
structural modes taken into account and with the control terms omitted. The resulting time histories of the
wing-root bending moment are presented in Figure 14. They demonstrate that the maximal bending moment
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

is obtained with Lg=0.55s which coincides with the period of the first wing bending moment (1.8Hz). Hence,
subsequent analysis and control design are performed with Lg=0.55s.

50
Lg=0.25 s
40 0.40
0.55
30 0.70
Wing root bending moment, N × m

0.85
20

10

−10

−20

−30

−40

−50
−0.5 0 0.5 1 1.5
Time, s

Figure 14: Open-loop wing-root bending moment with various gust lengths

All the driving actuators of the WT model were modeled by a state-space realization of the 3-rd order transfer
function
a0
Tac (14)
s 3
a2 s 2
a1s a0
with a0=1.25*107, a1=1.5625*105 and a2=1144.0, which is based on the 2-nd order transfer function obtained
from TsAGI and an additional low-pass filter with a pole at s=-1000 to yield the minimal order required for
ASE modeling5. The FS aircraft actuators were modeled as follows:
a0 B3 s 3 B2 s 2 B1 s B0
T AC Tac $ Tac ,cs $ 3 (15)
s3 a2 s 2 a1 s a 0 s A2 s 2 A1 s A0
with a0 = 6.251*106, a1 = 9.396*104 and a2 = 762.0, A0 = 5.483*105, A1 = 1.159*104 , A2 = 108.5, and B0 =
6.179*105, B1 = 795.0, B2 = 56.47, B3 = - 0.2385. The transfer function Tac was defined as an actuator in the
ZAERO ASE model, while Tac,cs was included as part of the control model.

A wing-tip accelerometer was found in Ref. 7 to react fast enough to cause significant alleviation of the loads
peak. Even though it was also fast enough for the alternative UWC and WTC configurations, their

7
accelerometers were placed at the respective forward tips of the stick holders. The augmentation of the
actuator model with the aeroelastic model resulted in an 81-state plant models with 3 inputs and 5 outputs. In
each case, the inputs to the state-space model were the command to the actuator, the gust velocity wg (t ) and
wg (t ) . The outputs for the control design model were the accelerometer signal awt , the actuator output ,
the wing-root shear force Fz, bending moment Mx and torsion moment My.

VI. Gust-Alleviation Control Laws


Full details on the WT control design process are given in Ref. 7. The goal of the control design was to
reduce the wing-root bending moment in response to discrete-gust excitation as much as possible without
exceeding the maximal control-surface deflection (actuator output) of 8deg, without degrading the wing-root
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

torsion moment and shear force, and without causing instability within the tunnel velocity range of 0-40m/s.
Simplistic design approaches was taken in order to numerically and experimentally demonstrate the basic
features of gust-response alleviation using wing control surfaces. Since the gust loads of interest are
dominated by the response of the first wing bending mode (1.8 Hz in our case), the main two questions were
whether the sensor can detect this response in time to activate the control surfaces before the loads reach their
peaks, and whether the control surface can provide a significant relief. As will be demonstrated, the answers
to both questions were positive. Based on these targets and features, the control design process was aimed at
defining for each configuration the parameters of a low pass filter that rolls off well below the respective
natural frequency of the control system,
Tc= kc/ %cs+1) (16)

The design was performed with respect to the defined discrete gust profile at V=25m/s, while checking the
ASE stability (flutter) over the entire velocity range. The designed filters were then applied to closed-loop
discrete and continuous gust response analyses at 25 and 35m/s, for comparisons with the respective open-
loop analyses. The state-space aeroelastic plant model described above was exported by ZAERO for control
design using Matlab for each control-surface configuration separately.

Best alleviation which provides acceptable stability margins was obtained with kc=0.09s2/m and %c=0.56s for
the BMA configuration, and kc=0.17s2/m, %c=0.33s for both the UWC and the WTC configurations. It should
be noticed that the surface area are 0.045m2 of the aileron, 0.03m2 of the underwing surface and 0.015m2 of
the wing tip surface. While the BMA and WTC controllers were used at 35 m/s as well, the UWC
configuration required a different control law at 35 m/s due to closed-loop instability problems. The low pass
filter (16) was used for gust-alleviation control of the FS aircraft. The values of the parameter %c remained the
same as for the WT model, while the filter gain kc was tuned to ensure stability and maximal control-surface
deflection of 15deg. It was set to 0.013 for the BMA configuration and to 0.007 for the UWC and WTC
configurations.

VII. Comparison between the Control Laws

A. Response of the wind-tunnel model to the control-design gust profile


The time histories of the open- and closed-loop acceleration signals for the BMA, UWC and WTC
configurations in response to the discrete gust excitation that was used for the control design, at V=25m/s, are
compared in Figure 15. It can be observed that there are no significant differences between the three open-
loop signal and between the three closed-loop ones, which means that the variation in accelerometer locations
did not make much difference.

The time histories of the actuator outputs in the three closed-loop cases are compared in Figure 16. It can be
observed that there is a delay of about 0.15s between the acceleration signals and the actuator response. If we
add some delay in the aerodynamic response, we can deduce that the first peak of the control aerodynamic
forces in each case occurs at about the same time as the first bending moment peak shown in Figure 8 for the
open-loop BMA case (with Lg=0.55) and in Figure 11 below. Also, the positive and negative peaks of the

8
aileron rotations occur at about the same time as the velocity peaks (the acceleration zero crossings), which
provides significant damping. Consequently, the maximal acceleration responses, which occur in the second
positive peak, are reduced by about 38% when the loop is closed.

6 6
main aileron WTC
UWC UWC
4 Open−loop 4 Open−loop
Closed−loop Closed−loop
Wing tip Z acceleration, m/s2

Wing tip Z acceleration, m/s2


2 2

0 0
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

−2 −2

−4 −4

−6 −6
−0.5 0 0.5 1 1.5 2 −0.5 0 0.5 1 1.5 2
Time, s Time, s

Figure 15: Time histories of the acceleration signals of BMA, UWC and TWC, V=25m/s

Main aileron
6
UWC
WTC
4

2
Aileron angle, deg

−2

−4

−6

−8
−0.5 0 0.5 1 1.5 2 2.5 3
Time, s
Figure 16: Time histories of the closed-loop actuator output signals, V=25m/s

The time histories of the open- and closed-loop wing-root bending moments for the three configurations at
V=25m/s are compared in Figure 17. The presence of an alternative control surface affect the open-loop
response in two ways: a) loads are increased due to increased wing area, and b) loads are reduced due to
aeroelastic damping caused by the forward position of the added surfaces. These cause 1.5% increase in the
maximal UWC bending moment and 4.2% reduction in the WTC case. The BMA, UWC and WTC control
laws reduce the maximal bending moment by 9.4, 13.3 and 15.1% respectively relative to the BMA open-
loop case.

The maximal open- and closed-loop values of the wing-root bending moment, torsion and shear force, the
acceleration signal and the actuator output are given in Table 1 for the two design velocities. The percentage
differences in this table are with respect to the open-loop values in the same line. The aileron (BMA) is not
effective at the higher speed due to static aeroelastic effects, while the alternative forward control surfaces
become even more effective. In all cases, the actuator outputs are smaller than the limit (8 deg) because of
stability constraints. Considering the ratio of the BMA, UWC and WTC surface area, which it about 3:2:1, it
is clear that the alternative control surfaces are generally more efficient than the main ailerons, and the wing-

9
tip surface is more effective than the underwing one. WTC also has less stability limitations in our study. As
indicated by the control gain and phase margins summarized in Table 2.

60 60
main aileron WTC
UWC UWC
40 Open−loop 40 Open−loop

Wing root bending moment, N×m


Wing root bending moment, N×m

Closed−loop Closed−loop

20 20

0 0

−20 −20
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

−40 −40

−60 −60
−0.5 0 0.5 1 1.5 2 −0.5 0 0.5 1 1.5 2
Time, s Time, s

Figure 17: Time histories of the wing-root bending moment for BMA, UWC and TWC, V=25m/s

Config. Parameter Parameter maximal value


V=25m/s V=35m/s
Open Closed Diff. % Open Closed Diff. %
BMA Mx, Nm 49.06 44.46 -9.38 50.69 49.73 -1.89
My, Nm 31.60 27.54 -12.8 32.59 30.08 -7.70
Fz, N 33.11 33.14 0.09 38.11 40.32 5.80
awt, m/s2 5.126 3.386 -33.9 4.855 3.453 -28.9
, deg - 3.11 - - 3.20 -
UWC Mx, Nm 50.69 42.50 -16.2 52.72 42.33 -19.7
My, Nm 31.92 27.95 -12.4 33.08 28.12 -15.0
Fz, N 32.78 30.34 -7.44 37.67 35.63 -5.42
awt, m/s2 5.367 3.974 -26.0 5.210 3.315 -36.4
, deg - 6.46 - - 4.67 -
WTC Mx, Nm 46.92 41.66 -11.2 48.77 42.88 -12.1
My, Nm 29.24 26.67 -8.79 30.45 27.99 -8.08
Fz, N 31.51 31.66 0.48 37.00 37.86 2.32
awt, m/s2 5.422 3.648 -32.7 5.307 2.831 -46.7
, deg - 7.51 - - 6.56 -
Table 1: Maximal values of discrete gust response

V Config. Gain Margins, db Phase Margins, deg


[m/s] UGM fc, Hz LGM fc, Hz UPM fc, Hz LPM fc, Hz
25 BMA 5.07 23.7 - - 103. 6.7 -113. 2.6
UWC 4.77 25.8 - - 42. 13.5 -104. 2.6
WTC 9.77 21.4 - - 61. 9.2 -84. 2.6
35 BMA 7.19 18.0 - - 79. 7.9 -40. 3.16
UWC 4.27 2.5 - - 42. 13.1 -61. 2.51
WTC 5.36 19.3 - - 37. 13.1 -40. 2.49
Table 2: Gain and phase margins at the actuator inputs

10
B. Response of the full-scale model to the control-design gust profile

The time histories of the open- and closed-loop wing-root bending moments for the three FS configurations at
V = 187.2 m/s, sea level, are compared in Figure 18. The BMA, UWC and WTC control laws reduce the
maximal bending moment by 7.6, 6.0 and 9.5% respectively relative to the BMA open-loop case.

7 7
x 10 x 10
1.5 1.5
main aileron WTC
UWC UWC
1 1
Wing root bending moment, N×m

Wing root bending moment, N×m


Open−loop Open−loop
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

Closed−loop Closed−loop
0.5 0.5

0 0

−0.5 −0.5

−1 −1

−1.5 −1.5
−0.5 0 0.5 1 1.5 2 −0.5 0 0.5 1 1.5 2
Time, s Time, s

Figure 18: Time histories of the wing-root bending moment for BMA, UWC and WTC FS aircraft,
V=187.2 m/s

The time histories of the actuator outputs in the three close-loop cases are compared in Figure 19. It can be
observed that the UWC performs better than the other control means because it requires significantly smaller
control-surface deflections.
15
Main aileron
UWC
10 WTC

5
Aileron angle, deg

−5

−10

−15
−1 0 1 2 3 4 5
Time, s
Figure 19: Time histories of the closed-loop actuator output signals, FS aircraft, V=187.18 m/s

C. Wing loads due to discrete regulation gusts


The discrete gust values defined by the JAR 25 regulations for structural design were scaled to the wind-
tunnel dimensions and used for calculation of envelopes of section loads combinations along the wing span.

11
The gust profiles are shown in Figure 20 and the monitoring stations with their local coordinate systems are
defined in Figure 21.
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

Figure 20: Gust profiles for calculations of envelopes of wing section loads

Figure 21: Monitoring stations for section loads

Comparisons of the open-loop (black) and closed-loop (red) envelopes of section shear forces and bending
moments versus torsion moments are shown in Figure 22 for the BMA configuration and in Figure 23 for
UWC configuration. The envelopes were calculated at V=25 m/s, sea level. They include incremental loads
only, that should be added to the corresponding 1-g level flight loads. The envelopes demonstrate that the
gust-alleviation control systems reduce the bending moments and shear forces at all the monitoring stations
and that this reduction is more effective while using the UWC surface, similarly to what shown above for the
control-design profile. It can be also observed that the extreme closed-loop torsion moments exceed the
open-loop ones. This is due to the control-surface gust-alleviation activities. However, the increased
tensional loads are still small compared to typical maneuver loads that involve aileron excitation.

12
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

13
configuration
Figure 22: Envelopes of section shear forces and bending moments vs. torsion moments for the BMA
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

14
configuration
Figure 23: Envelopes of section shear forces and bending moments vs. torsion moments for the UWC
D. Continuous gust response
The three control systems were also compared in continuous gust response performance. Analyses were
performed using the frequency-domain option of ZAERO5 at sea level for V=25m/s and V=35m/s with the
root-mean-square (RMS) value of the vertical gust velocity equals 1m/s and a gust scale of 100m. Power-
spectral-density (PSD) plots of the measured acceleration signal and the wing-root bending moment in the
BMA and UWC cases are shown in Figure 16 for V=25m/s and in Figure 17 for V=35m/s. The sharp peak at
2.4Hz, V=25m/s (Figure 24) indicates a flutter mechanism that involved the wing bending and the inboard
engine yaw modes and becomes unstable at VF=48m/s. This flutter mechanism is suppressed when either the
BMA or the UWC loops are closed. However, a new ASE flutter mechanism is created at 3.15Hz that
involves the outboard engine yaw and becomes unstable at VF=40m/s. This mechanism, that becomes
dominant at V=35m/s (Figure 25), is the instability caused the control gains limits discussed in the control
design section. The large values at zero frequency in the bending-moment plots stem from a singularity that
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

does not affect the results.

Apart from the flutter mechanisms, it is clear from Figures 24 and 25 that the gust alleviation control reduces
the gust response at most frequencies, especially around 2Hz, for which it was designed. The loads around
the short-period peak, at 0.3Hz, are also suppressed, but it can be seen that the aileron becomes ineffective at
V=35m/s, and even causes some bending moment increase at 2.3 Hz.

The RMS values of the response parameters for the three control configurations (integrals of the PSD
function) are summarized in Table 3. Here again, the presence of the under-wing surface increases the
bending-moment response while the wing-tip surface reduces it. It is also obvious that the aileron is
ineffective at V=35m/s. The UWC performance appears to be slightly better than the WTC. However, if we
consider the smaller area of the wing-tip surface, it is still somewhat favorable in the analyzed configuration.

9 2500
main aileron
PSD of wing root bending moment, (N×m)2/Hz

main aileron
PSD of wing tip Z acceleration, (m/s2)2/Hz

8 UWC UWC
Open−loop Open−loop
7 Closed−loop 2000 Closed−loop

6
1500
5

4
1000
3

2 500
1

0 0
0 2 4 6 8 10 0 1 2 3 4 5
Frequency, Hz Frequency, Hz

Figure 24: PSD of the accelerations and bending moments of BMA and UWC, V=25m/s

15
140 4500
main aileron

PSD of wing root bending moment, (N×m)2/Hz


PSD of wing tip Z acceleration, (m/s2)2/Hz main aileron
UWC 4000 UWC
120 Open−loop Open−loop
Closed−loop 3500 Closed−loop
100
3000
80 2500

60 2000

1500
40
1000
20
500
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

0 0
0 2 4 6 8 10 0 1 2 3 4 5
Frequency, Hz Frequency, Hz

Figure 25: PSD of the accelerations and bending moments of BMA and UWC, V=35m/s

Config. Parameter Parameter RMS


V=25m/s V=35m/s
Open Closed Diff. % Open Closed Diff. %
BMA Mx, Nm 29.01 27.08 -6.65 36.29 37.38 3.00
My, Nm 18.92 17.25 -8.83 23.68 22.92 -3.21
Fz, N 21.92 22.29 1.69 28.13 32.75 16.4
awt, m/s2 3.275 2.373 -27.5 5.072 3.650 -28.0
UWC Mx, Nm 29.83 25.77 -13.6 52.72 29.82 -19.8
My, Nm 18.93 17.14 -12.4 37.17 28.12 -13.6
Fz, N 21.81 20.79 -4.68 28.99 28.31 -2.35
awt, m/s2 3.234 2.704 -16.4 5.979 4.108 -31.3
WTC Mx, Nm 28.78 26.08 -9.38 35.76 31.93 -10.7
My, Nm 18.14 16.85 -7.11 30.45 20.96 -6.60
Fz, N 21.64 21.33 -1.43 28.90 30.26 4.71
awt, m/s2 3.035 2.354 -22.4 4.971 3.643 -26.7
Table 3: RMS values of continuous gust response

VIII. Conclusions

Specially designed wing mounted control surfaces have been shown to be capable of significantly reducing
gust response loads acting on aircrafts. Mathematical models of a transport aircraft and its aeroelastic wind-
tunnel model were used for demonstrating the capabilities of symmetrically activated under-wing and wing-
tip, forward mounted control surfaces in comparison with the regular ailerons. Frequency-domain formulation
was used for open- and closed-loop loads analyses. State-space time-domain models were used for the
control design process. It was shown that wing-mounted accelerometers can be used effectively as single
control inputs. The simple low-pass filter that was used in each control case provided physical insight and
allowed robust and easy applications in the subsequent wind-tunnel tests. The maximal wing root bending
moment was reduced by 9% with the ailerons, by 13% with the underwing surfaces and by 16% with the
wing-tip surfaces at intermediate velocities. The main ailerons became ineffective at high speeds while the
special surfaces became even more effective because of static aeroelastic effects. Similar results were
obtained with the control laws applied to the aeroservoelastic model of the full-scale aircraft. The controllers
that were designed for the alleviation of wing-root bending moments due to a single discrete gust profile,
demonstrated similar performances in response to regulation discrete gust profiles and in response to

16
statistical continuous gusts. The clear advantages of the special surfaces call for their evaluation in real-
aircraft designs.

Acknowledgements

The work presented in this paper was supported by the European Union through project GRD-1-2001-40122,
Active Aeroelastic Aircraft Structures (3AS). The wind-tunnel model and the associated structural model
were developed by Dr. Svetlana Kuzmina, Dr. Mikhail Zichenkov, Dr. Fanil Ishmuratov and Dr. Vasily
Chedrik of the Central Aerohydrodynamic Institute (TsAGI) in Russia.

References
Downloaded by TECHNION - ISRAEL INST OF TECH on November 22, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2006-1833

1. Hoblit, F.M., Gust Loads on Aircraft: Concepts and Applications, AIAA Education Series, AIAA,
Washington, DC, 1988.
2. Rodden, W.P. and Johnson, E.H., “MSC/NASTRAN Aeroelastic Analysis User’s Guide,” The MacNeal-
Schwendler Corporation Publication, 1994.
3. ZAERO Version 6.2 Theoretical Manual, ZONA 02-12.4, ZONA Technology Inc., Scottsdale, AZ,
October 2002.
4. Roger, K. L., "Airplane Math Modeling and Active Aeroelastic Control Design,'' AGARD-CP-228, 1977,
pp. 4.1-4.11.
5. Karpel, M., “Time Domain Aeroservoelastic Modeling Using Weighted Unsteady Aerodynamic Forces”,
Journal of Guidance, Control, and Dynamics, Vol. 13, No. 1, 1990, pp. 30-37.
6. Karpel, M., Moulin, B. and Chen, P.C., "Dynamic response of Aeroservoelastic Systems to Gust
Excitation," Journal of Aircraft, Vol. 42, No. 5, 2005, pp. 1264-1272.
7. Moulin, B. and Karpel, M., "Gust Loads Alleviation Using Special Control Surfaces," Proceedings of the
45nd Israel Annual Conference on Aerospace Sciences, February 2005.
8. Karpel, M., Moulin, B., Anguita, L., Maderuelo, C. and Climent, H., "Aeroservoelastic Gust Response
Analysis for the Design of Transport Aircrafts," Proceedings of the 45th AIAA/ASME/ASCE/AHS/ASC
Structures, Structural Dynamics & Materials Conference, Palm Springs, CA, April 2004.

17

View publication stats

You might also like