Modern Solutions For Ground Vibration Testing of Large Aircraft
Modern Solutions For Ground Vibration Testing of Large Aircraft
Modern Solutions For Ground Vibration Testing of Large Aircraft
Figure 1. (a) A310 boom demonstrator aircraft ; (b) A330 multirole transport
tanker.
Figure 2. Aircraft development process; GVT takes place shortly before the first flight.
Vibration
Tests
Design
Requirements
Flight
Tests
Physical
Prototype
Model Test
Experimental
Modal Model
Pre-Test
Prototype
Virtual
Prototype
FE Model
Analytical
Modal Model
Calibrated Virtual
Prototype
Response Analysis
Design Modifications
Test Equipment. A GVT campaign typically requires the involvement of different teams from different divisions or companies. In
this case, teams from EADS CASA, lava Ingenieros and the engineering services division of LMS International cooperated. Here
we will review the recently installed GVT solution at EADS CASA
that avoids making compromises on the level of data management
and engineering collaboration.
The excitation was provided by LDS permanent magnet shakers
(Model V450), having a sine force peak using forced-air cooling of
311 N and a peak-to-peak stroke of 19 mm (Figure 5a). Labworks
Model PA-138 shaker amplifiers provided the power. The forces
injected into the aircraft are measured by PCB 208C03 force gages,
and the aircraft vibration response is measured by PCB 333B32 and
PCB 393B04 accelerometers.
The shakers were controlled and the transducer signals recorded
by a 700-channel LMS data acquisition system consisting of four
SCADAS III front-end frames16 connected in master/slave mode
with eight sources and 700 measurement channels (V12 modules).
The LMS Scadas III is a completely digital front end, with one
sigma-delta ADC per channel, 24-bit data transfer and ultra-low
noise floor. The V12 modules provide voltage, ICP and TEDS
signal conditioning. The sensor connectivity was facilitated by
embedding the front ends in two racks with patch panels (Figure
5b). Besides a data acquisition PC that processes and stores the
data, two analysis stations are available that perform an on-site
modal analysis, ensuring optimal use of testing time and engineering resources.
The PCs are equipped with LMS Test.Lab data acquisition and
analysis software.17 Test.Lab covers all GVT test modes in one
user environment; MIMO random, swept sine, stepped sine and
normal modes (phase resonance) are all available. Working with
the exact setup database allows a very fast transition from one test
mode to another. Furthermore, a seamless integration between test
and analysis is realized by offering the analysis capabilities as
add-on functions in the acquisition workbooks, allowing almost
real-time modal parameter estimation that can be considered as an
advanced check of measurement quality and data validity during
the measurement process.
Test Conditions. The new GVT system was deployed, and the test
teams were trained for the demonstration GVT on EADS CASAs
A310 boom demonstrator aircraft. During the test, the aircraft must
be in a condition ready for flight, and in a configuration as close
as possible to the configuration of a test flight. The flight control
systems (control surfaces, spoilers and air brakes) are serviceable
and kept in the 0 position. The flap and slat systems are serviceable retracted and locked. The control systems of the trimmable
horizontal tail are serviceable, and the horizontal tail is kept in
the 0 position. The hydraulic systems were switched on (elevaSOUND & VIBRATION/JANUARY 2009
Random
Swept Sine
Stepped Sine
Normal Modes
[Hz]
AP(F) [N2]
AP(F) [N2]
AP(F) [N2]
AP(F) [N2]
[Hz]
[Hz]
Phase Separation of
Frequency Response Funciton (FRF) Based Methods
Measure FRF
Modal Parameter
Estimator
Experimental Modal
Model [, , ]
resonance [Hz]
Phase Resonance /
Mode Appropriation
Measure / Identify
Mode [, , ]
Figure 4. Dynamic excitation signals and modal parameter estimation strategies in GVT.
Figure 5. (a) Engine shaker excitation; (b) 700-channel LMS data acquisition system.
(a)
FRF WngL:135:+Z / WngR:114:+Z
((m/s2)/N)
dB
Phase
180
180
Frequency, Hz
(b)
N2
dB
20
HTPL:211:-X
HTPL:211:-Z
WngL:135:-Z-f
WngR:114:-X
40
Frequency, Hz
Figure 6. (a) Shaker locations during random tests; (b) power spectra of
some input forces.
Figure 8. (a) Reciprocal FRF: left wing vertical, right wing vertical; (b) PolyMAX stabilization diagram.
((m/s2)/N)
dB
(a)
Phase
180
-180
Frequency, Hz
((m/s2)/N)
dB
(b)
Phase
180
-180
Frequency, Hz
tion combinations that were used for MIMO swept-sine testing, the
phase resonance (normal modes) testing capabilities of the new
GVT solution were verified. The following modes were subjected
to phase resonance testing: all engine modes, horizontal tail plane
(HTP) roll mode, HTP yaw mode, as well as elevator rotation and
SOUND & VIBRATION/JANUARY 2009
11
Figure 12. PolyMAX stabilization diagram: lowest frequency band from wing
vertical sine sweep tests.
(b)
N/V, dB
(a)
180
180
Frequency, Hz
Figure 13. System FRFs [N/V]: (a) Low-frequency band with large shakerstructure interaction; (b) High-frequency, flat system FRFs/low shakerstructure interaction.
Figure 10. Shaker locations during vertical wing testing and longitudinal
horizontal tail plane testing.
((m/s2)/N)
dB
(a)
FRF WngL:115:+Z/WngL:115:+Z
FRF WngR:114:+Z/WngR:114:+Z
Phase
180
180
Frequency, Hz
((m/s2)/N)
dB
(b)
FRF WngL:115:+Z/WngR:114:+Z
FRF WngR:114:+Z/WngL:115:+Z
Phase
180
180
Frequency, Hz
Figure 11. Vertical wing swept-sine tests: (a) driving point FRFs; (b) reciprocal FRFs.
twist modes. In general, the automatic tuning algorithm17 immediately detects the modes under investigation. In cases where the
automatic tuning does not converge, a manual tuning is required.
12
Figure 14. Mode shape visualization typically used in normal mode testing;
coincident and quadrature part.17
Figure 15. Linearity check for HTP first bending, wings second bending (S)
mode excited using the phase resonance method at five different force levels
with 2 vertical HTP shakers.
Figure 17. A310 mode shapes: (a) HTP first bending (A), wings second bending (A - in counter phase with HTP), wings fore-aft (A), boom first lateral
bending; (b) Wings third bending, HTP first vertical bending (in counter
phase with wings), wings fore-aft, ruddervator bending.
Figure 16. A310 mode shapes: (a) Wings first bending, engine lateral in
phase with wings (S); (b) Wings first bending, engine lateral in counter
phase with wings (S).
models during the GVT. The FE technique is the standard tool for
structural analysis. The FE model available prior to the GVT can
be used to make predictions on the aircraft dynamic behavior and
to optimize the test arrangement and duration. As an example, the
FE model of the A330 MRTT is represented in Figure 20.
Pretest simulations make use of:
Preliminary FE model
Estimated aircraft GVT masses
Estimated GVT (close to free-free) boundary conditions
The normal modes obtained from the FE model before the test
represent an accurate enough estimation of the aircraft eigen frequencies and mode shapes. This information is used to plan the
test; that is, to determine excitation conditions, shaker locations
and accelerometer locations.
www.SandV.com
Figure 18. A310 mode shapes: (a) Left wing torsion; (b) right wing torsion.
SOUND & VIBRATION/JANUARY 2009
13
.1
.6
60
.1
.6
.5
57
54
51
.1
46
.6
.0
.8
42
38
34
.0
.5
.4
30
24
18
13
9.
W, Kg
X, m
Tx
Heave
5%
Yaw
Pitch
Roll
Ty
+5%
Pre-test
Post-test
Figure 22. A330 MRTT FEM-GVT comparison of eigenfrequencies of rigidbody modes (Ref 22).
During the first days of the GVT, the aircraft is weighted, and
the rigid body eigen frequencies are measured. With these results,
a second loop of normal modes calculations is performed. Figure
21 shows an example of the mass corrections to the nominal FE
model, which are needed to match the masses of the actual aircraft
configuration being tested during the GVT. Typically, this correction is a reduction to account for missing elements like passenger
seats, galleys, etc. Figure 22 shows how tuning the FE model to
accurately reflect the aircraft mass distribution and test boundary conditions leads to an improved prediction of the rigid-body
modes. Blue dots show the position of the estimated rigid-body
eigen frequencies before the GVT, while the green dots show the
final rigid-body eigen frequencies right in the 45-degree line when
comparing numerical values (FE) with measured values (GVT).
Once the model has the correct test-measured masses and testmeasured boundary conditions, the right comparison between all
flexible modes can be made to know whether the FE model needs
to be updated or not. If the frequencies of all modes lie inside a
reduced margin (lower than 5%), the model does not need to be
updated. But for larger differences, the model must be corrected.
The correction needs to account for local effects20 or absent components.21 A good approach for model updating is the delta stick
approach, where the missing stiffness is introduced via external
14
Conclusions
We showed how a new 700-channel ground vibration test system
was successfully deployed during a demonstration test on the A310
and later successfully used during a certification test on the A330
MRTT. The new hardware and software solution is designed to
cope with the main challenge of todays GVT campaigns; that is,
to reduce the testing and analysis time without compromising the
accuracy of the results. Highly efficient testing is made possible by
the integration of complementary excitation techniques in a single
www.SandV.com
Acknowledgement
Part of this work was carried out under the EUREKA project E!
3341 FLITE2. We gratefully acknowledge the financial support
of the Institute for the Promotion of Innovation by Science and
Technology in Flanders (IWT).
References
1. Fargette, P., Fllekrug, U., Gloth, G., Levadoux, B., Lubrina, P., Schaak,
H., and Sinapius. M., Tasks for Improvement in Ground Vibration Testing of Large Aircraft, Proc. IFASD 2001 Int. Forum on Aeroelasticity
and Struct. Dyn., Madrid, June 2001.
2. Lubrina, P., Ground Vibration Experiments on Large Civil Aircraft for
Engine Imbalance Purpose, Proc. IFASD 2003 Int. Forum on Aeroelasticity and Struct. Dyn., Amsterdam, June 2003.
3. Gloth, G., Degener, M.. Fllekrug, U., Gschwilm, J., Sinapius M., Fargette,
P., Levadoux, B., and Lubrina. P., Experimental Investigation of New
GVT Concepts for Large Aircraft, Proc. IMAC 19 Int. Modal Analysis
Conf., Kissimmee, FL, Feb. 2001.
4. Climent. H., Aeroelastic and Structural Dynamics Tests at EADS
CASA, Proc. of the 18th Annual Symposium of the Society of Flight
Test Engineers SFTE, Madrid. June 11-13, 2007.
5. Pickrel, C. R., Foss, G. C., Phillips, A., Allemang, R. J., and Brown, D.
L., New Concepts GVT, Proc. IMAC 24 Int. Modal Analysis Conf., St.
Louis, MO, Jan.-Feb. 2006.
6. Gge, D., Bswald, M., Fllekrug, U., Lubrina. P., Ground Vibration
Testing of Large Aircraft State-of-the-Art and Future Perspectives,
Proc. IMAC 25 Int. Modal Analysis Conf., Orlando, FL, Feb. 2007.
7. Schaak, H., Hybrid Dynamic Modal Adjustment of Very Large Aircraft,
Proc. LMS Conf., Paris, Nov. 2001.
8. Pickrel, C. R., Airplane Ground Vibration Testing Correlation with
Nominal Modal Model, Proc. IMAC 20 Int. Modal Analysis Conf.,
949956, Los Angeles, CA, Feb. 4-7, 2002.
9. Van der Auweraer, H., Requirements and Opportunities for Structural
Testing in View of Hybrid and Virtual Modeling, Proc. of ISMA 2002, the
Int. Conf. on Noise and Vibration Eng., Leuven, Belgium, Sept. 2002.
www.SandV.com
15