Beruflich Dokumente
Kultur Dokumente
Bruno Wiesler
FH JOANNEUM GmbH, Degree Programme Luftfahrt/Aviation, Alte Poststr. 149,
A-8010 Graz, Austria
Peter Tibaut
AVL List GmbH, Hans List Platz 1, A-8010 Graz, Austria
is achieved by regarding the turbulence as a The noise sources due to the presence of the
pseudo-stochastic phenomenon with many solid boundaries are included via the solid
different length scales and time scales. Snellen boundary conditions in the LEE solver. This
argues that for locally homogeneous turbulence accounts for effects like reflection and diffraction.
the turbulent velocity distribution can be There is no 'direct' interaction between the solid
synthesized, by assuming that the turbulent boundary and the volume source terms: the
velocity distribution can be expressed as a Fourier source terms are computed from local quantities,
integral: e.g. local turbulence velocities, mean flow
r r r r r r r r r
ut ( x ) = ò éëu% (k ) eiy (k )s (k ) ùûei (k × x ) d k , velocities and their gradients. There is thus no
r double counting. There is of course an 'indirect'
with k denoting the wave-vector, ψ a random interaction between the boundary and the volume
r
phase, and s the direction of the turbulent source terms, via the CFD-input: close to a wall,
velocity vector. The turbulent velocity distribution large gradients in flow velocities, TKE and
can be regarded to be locally homogeneous; this dissipation levels may occur, which leads
means that the variation of the TKE k and of the generally to high volume source terms.
Turbulent Length Scale (TLS) Λ is small across
distances of the order of Λ. In this manner, the According to Blom [1], the propagation of sound
above equation can be rewritten as can be described by the Linearized Euler
Equations (LEE) under the assumption that there
is no significant feedback (or back-reaction) of the
r r
{ r r
}
ut ( x ) = ò éëu% (k ) s (k ) ùû × sound to the mean flow. In the linearization
process the possibility of representing feedback is
{ ë
r r
r
r r r r
û }r
écos (k × x + y (k ) ) + i sin (k × x +y (k ) ) ù d k lost. The Linearized Euler Equations are
considered as
and since u should be a real number r r ¶ r ¶ r r r
( )
L U = U+
¶t ¶x j
Fj U = S , ( )
r r
{ r r r r r r
} r
ut ( x ) = ò éëu% (k ) s (k ) ùû cos (k × x + y (k ) ) d k .
describing the perturbation
r
Following Snellen et al. [7], the integral is then U = ( r , u , v, w, p )
replaced by a finite sum of modes, each with its
own wave vector and its own velocity amplitude with regard to the mean flow and the source terms
r
and direction, respectively. Employing N Fourier S.
modes for discretizing the above integral and As a third step the LEE are solved with a
considering only a half-sphere, the following Discontinuous Galerkin Solver (DIGS). The
expression for the turbulent velocity is obtained numerical method for solving the equations is
N
r r r r based on a quadrature-free implementation of the
ut ( x ) = 2å u%n cos (k n × x + y n ) s n ,
n =1
Discontinuous Galerkin finite-element method [8]. plates in order to avoid interaction of the wind
Further details on the DIGS can be found in [1]. tunnel’s shear layer and the cylinder tips.
To predict the pressure signal outside of the CAA
domain, DIGS is coupled with the FWH module.
The simple integration of the surface source terms
on a porous FWH surface does account for the
quadrupole sources enclosing within the surface
[9]. Thus only surface integrals are needed for the
calculation of the far field sound, instead of the
volume integrals required by the traditional
acoustic analogy method. Figure 2: The original model of the half cylinder in
The generalized conservation equations (mass, the wind tunnel, the CAD model for the CFD
momentum, energy) can be manipulated into the simulation.
form of an inhomogeneous wave equation. The The experiments and the CFD simulations are
solution to the FWH equation is given in an performed with four different inlet velocities, 26.4,
integral representation: 31.6, 36.8 and 41.8 m/s. The correlating Reynolds
r é Q& + L& / c ù numbers based on the diameter of the half
4p p ' ( x, t ) = ò ê ú dS + cylinder are 5.33E+04, 6.38E+04, 7.43E+04 and
êë r (1- Mr ) úûret
f =0 2
8.44E+04 respectively.
Four Microphones, flush mounted in the plates,
é L -L ù
are placed in the near field and six microphones
ò f =0
ê 2 r M 2 ú dS +
êë r (1- Mr ) úûret in a distance of one meter from the half cylinder’s
centerline capture the noise in the far field.
( (
é ( Q + L / c) rM& + c M - M 2 ) ) ùú
òf =0 êê
r r r
dS,
r (1- Mr ) ú
2 3
ë ûret
with
Q = r un - r vn ,
Li = Pij + r ui ( un - vn ) ,
and where LM denotes the inproduct of Li with Mi
[9]. The above equation can be solved Figure 3: Microphone positions for the
analytically, as a consequence the distance of the measurements in the near field.
observer location from the source does not
influence the simulation time.
5. RESULTS
The CAA approach as described in the previous
section is used for determining the flow-induced
noise for a number of practical applications. The
CFD software package SWIFT is used to solve
the three-dimensional RANS equations. In the first
part of this section, simulations to predict the
noise from a basic half cylinder placed between
two plates are described. The second part
describes the investigations performed on the
Figure 4: Microphone positions for the
landing gear.
measurements in the far field.
5.1 Aero acoustics of a basic geometry The simulations monitoring locations are in
The experiments are performed in the aero correspondence with the experimental
acoustic wind tunnel of TNO, equipped with microphone positions.
silencers in the ducting and an anechoic test The computational grids were built with care. For
chamber [3]. capturing the details of the turbulent flow motion a
The geometry used for the basic investigations suitable grid with about 2 Mio. mesh cells and with
can be seen in Figure 2. The half cylinder with a max. y+ values in the regions of interest of 100 is
diameter of 31 mm is mounted between two generated.
The turbulent flow features are accounted for by
the k-e turbulence model and the log-law is
applied at the solid walls.
6. CONCLUSIONS
A hybrid CFD-CAA method is employed for the
assessment of actual prospects in noise
predictions for landing gears within industrial time
and resource constrains. Two cases are carried
out, the first one aiming to predict the noise
radiation from a basic geometry i.e. a half
cylinder, placed in a low turbulent flow, the
second one aiming to perform an aero acoustic
simulation on a real scale (simplified) landing
gear.
The results of the simulations are compared with
experimental data for the half cylinder case. The
simulation results show satisfying agreement with
experimental data for both, the predictions in the
near and the far field, in view of trend analyses.
The simulations with the simplified Boeing 747
nose landing gear show that the used method is
capable of handling complex, full scale
geometries. The impact on the characteristic
noise field, when changes in the geometry are
made, is shown. Removing a small part of the
8. REFERENCES [9] Souliez, F.J., Long, L.N., Morris, P.J.,
Sharma A., “Landing gear aerodynamic
[1] Blom, C. P. A., Verhaar, B.T., van der noise prediction using unstructured grids”,
Heijden, J.C., Soemarwoto, B.I., “A International Journal of aeroacoustics,
Linearized Euler Method Based Prediction volume 2, 2002, pp. 115-135
of Turbulence Induced Noise using Time-
averaged Flow Properties”, AIAA paper [10] Védy E., L. Van Lier, R.R. Parchen, J.
2001–1100 Golliard, P.J.G. van Beek, P. Tibaut, W.
Baier, Application of Statistical Noise
[2] Drage P., Wiesler B. van Beek P., van Lier Generation and Radiation for aeroacoustic
L., Parchen R., Tibaut P., “Application of predictions in the automotive industry,
the SNGR Method to Predict Noise 2005 SAE NVH Congress, Traverse City
Radiation from Basic Configurations of MI, paper no. 2005-01-2494
Landing Gears”, Int. User Meeting, AVL,
Graz, Austria,2005, [11] Wiesler B., Tibaut P., van Lier L., Goilliard
J., Vedy E., “Prediction of Turbulence
[3] Golliard, J. Noise of Helmhotz-resonator Induced Noise within Product
like cacities excted by a low Mach-number Development Processes“, Proceedings,
turbulent flow, PhD thesis, École Noise-Con 2004, Baltimore, MD, 2004
Supérieure d’Ingénieurs de Poitiers, 2002