Vous êtes sur la page 1sur 9

Aerospace Science and Technology 8 (2004) 175–183

www.elsevier.com/locate/aescte

Investigation of engine jet/wing-tip vortex interference ✩

Untersuchung der Interaktion von Flügelrandwirbel


und Triebwerksstrahl
Guido Huppertz ∗ , Ehab Fares, Ronald Abstiens, Wolfgang Schröder
Aerodynamisches Institut, RWTH Aachen, Wüllnerstraße zw. 5 und 7, D-52062 Aachen, Germany
Received 24 February 2003; received in revised form 6 October 2003; accepted 31 October 2003

Abstract
The wing-tip vortex of a rectangular wing half-model with a model engine is studied experimentally and numerically. The airfoil has
a supercritical BAC 3-11/RES/30/21 geometry with a chord length of c = 150 mm. The investigations include three different spanwise
positions of the engine and two jet velocities with a constant free stream velocity of u∞ = 27 m/s yielding Rec = 2.8 × 105 . The numerical
simulation is divided into two parts. First the flow around the airfoil is calculated. In a second step the data is taken as an input to calculate
the wake downstream of the trailing edge. Experiments include 2C- and 3C-PIV-measurements for the velocity distribution of the jet and the
wing-tip vortices. The experimental and numerical results show the significance of the streamwise velocity component for the analysis of the
jet/wing-tip vortex interaction.
 2003 Elsevier SAS. All rights reserved.
Zusammenfassung
Der Randwirbel des Halbmodells eines Rechteckflügels mit Modelltriebwerk wird numerisch und experimentell untersucht. Der Flügel
hat ein superkritisches BAC 3-11/RES/30/21-Profil mit einer Sehnenlänge von c = 150 mm. Die Untersuchung umfasst drei spannweitig
verschiedene Positionen des Triebwerks und zwei verschiedene Geschwindigkeiten des Triebwerksstrahls bei konstanter Geschwindigkeit
der Anströmung u∞ = 27 m/s, entsprechend Rec = 2.8 × 105 .
Die numerische Simulation ist untergliedert in zwei Abschnitte. Zunächst wird die Strömung um den Flügel berechnet. Im zweiten
Schritt werden die so gewonnenen Daten als Eingangswerte zur Berechnung des Strömungsfeldes im Nachlauf des Flügels verwendet. Im
Experiment werden mit Hilfe von 2C- und 3C-PIV-Messungen Verteilungen der Geschwindigkeit von Triebwerksstrahl und Flügelrandwirbel
ermittelt.
Die experimentellen und numerischen Ergebnisse zeigen die Bedeutung der Geschwindigkeitskomponente in Strömungsrichtung für die
Beurteilung der Triebwerksstrahl-Wirbel-Interaktion.
 2003 Elsevier SAS. All rights reserved.

Keywords: Vortex; Wake; Engine; Jet; PIV; Numerical simulation

Schlüsselwörter: Wirbel; Nachlauf; Triebwerk; Strahl; PIV; Numerische Simulation

1. Introduction Particularly in the vicinity of airports this is of real concern,


where the planes fly in high-lift configuration. The trailing
Heavy aircrafts generate vortex wakes that are a serious vortices cause a reduction of starting and landing frequency
danger to the following air traffic, especially in the case for the airports and an increase in flight-time and costs for
of a smaller aircraft crossing the wake of a large airplane. passengers as well as for airlines. The strength of these
vortex wakes slowly reduces by diffusion and dissipation.

This article was presented at the German Aerospace Congress 2002.
Under still uncertain conditions strong and nearly abrupt
* Corresponding author. changes in the wake structure have been reported, that
E-mail address: guidoh@aia.rwth-aachen.de (G. Huppertz). even can lead to the collapse of the vortex system [23].
1270-9638/$ – see front matter  2003 Elsevier SAS. All rights reserved.
doi:10.1016/j.ast.2003.10.008
176 G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183

Although these phenomena occur in the far field of the wake, The temporal integration is based on an explicit 5-step
investigations of the near wake have to be performed to Runge–Kutta time stepping scheme [12] using the coeffi-
detect the mechanisms that determine the structure of the cients αi = (0.059, 0.145, 0.273, 0.5, 1.). The coefficients
aircraft wake. are optimized to achieve maximum stability for upwind
Recent measurements [1] and numerical studies of the schemes [16]. The scheme is second-order accurate and al-
isolated turbulent wing-tip vortex [2] were successfully car- lows a maximum Courant number of approximately 3.5. To
ried out for a rectangular wing with small aspect ratio. It was further accelerate the solution process to the steady state lo-
shown in the numerical investigations that good agreement cal time stepping and multigrid [16] are used.
with experimental results could only be achieved with ap-
propriate grid resolution, turbulence models, and boundary 2.2. Turbulence modeling
conditions. Other authors [9,11,21] investigated the dynam-
ics of engine jets and vortices in the far field using simplified The turbulent flow is modeled based on the standard con-
models to predict the inflow conditions from the near wake. cept of Reynolds-averaging. Several simple models from al-
These investigations included a wide range of parameters gebraic to transport equation turbulence models of varying
that influence the wake, such as stratification, wind shear, complexity have been developed and investigated for aero-
buoyancy, and geometrical parameters. Numerical and ex- dynamic wall-bounded flows. The Spalart–Allmaras one-
perimental investigations with regard to engine jet location equation model proved to be very efficient and robust for a
[8,10,18] confirm the influence of the jet on wake properties wide variety of aerodynamic flows and was already used in
and decay mechanisms. similar calculations [3,4,6,13]. This model is slightly modi-
Therefore the vortex generating wing geometry and the fied to account for free vortical flows. The original transport
effects, which engine jets in the wake of the plane can equation is reformulated in a “conservative” manner to de-
have on the vortex structure, are of major concern for termine the turbulent viscosity νt
the investigation of the vortex wake. This analysis of the
D(ρ ν̃) ρ     
interaction of engine jets and vortices in the wake of a wing = ∇ · νl + (1 + cb2 )ν̃ ∇ ν̃ − cb2 ν̃∇ 2 ν̃
is done numerically and experimentally. The experimental Dt σ
 2
tests are limited to the immediate near field of the wing due ν̃
+ ρcb1
S ν̃ − ρcw1 fw
to the size of the test section of the wind tunnel. d
with
 ν̃
2. Numerical method S =S+ fv2 , νt = ν̃ · fv1 .
κ 2d 2
Even when locally refined and automatically adapted A more detailed description can be found in [7]. The
grids are used it is fair to say that the computational costs for choice of the turbulence model and its implementation is
simultaneous numerical simulation of the flow field around still a challenge in any turbulent flow simulation. Based on
the wing and the far wake are extremely high. For this the findings in former analyses [3,4] the slightly modified
reason the numerical method used in this study consists of Spalart–Allmaras model which contains a constraint for the
a two step approach: first, the flow field around the wing vorticity through the strain rate tensor
is determined and subsequently, the data of this simulation
S = Ωij + 2 min(0, sij − Ωij )
is used to provide the inflow condition for a separate wake
simulation. is used in this investigation. The convective and dissipative
terms of the turbulence model are discretized to second-
2.1. Flow solver order accuracy. The upwind approximation of the convective
terms follows the AUSM method using second-order recon-
The method of solution has been successfully applied struction of the left- and right-hand values.
to predict a wide variety of internal and external flow No tripping corrections are used since the calculations are
problems [5,17]. The version used in this paper solves performed for fully turbulent flows. This corresponds to the
the Favre-averaged steady threedimensional Navier–Stokes experiments, where the flow field is tripped at the leading
equations of a compressible ideal gas on the basis of a edge of the wings.
finite-volume discretization on block-structured grids. The
viscous stresses are discretized using standard second-order 2.3. Boundary conditions
accurate central schemes. The spatial discretization of the
convective terms follows the AUSM scheme proposed by Investigations have shown that special attention should be
Liou and Steffen [14,15]. The reconstruction of the left and given to boundaries where inflow and outflow can occur. Es-
right values at the cell interfaces was extended from the pecially simple extrapolations have proven to be the worst
classical second-order MUSCL-interpolation [25] to a third- choice physically and numerically. A new formulation for
order accurate scheme. the free boundaries based on the weak-reflecting boundary
G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183 177

conditions as introduced by Thompson [24] for the Euler


equations and extended by Poinsot and Lele [19] for the
Navier–Stokes equations was presented in [6]. The proposed
boundary condition smoothly combines inflow and outflow
formulations depending on the velocity vectors at the bound-
aries. A further methodology has been also presented to cor-
rect the residuals at the boundaries using known values from
the interior of the computational domain [6].
The engine jet is simulated numerically by imposing
appropriate boundary conditions as described in [3]. The
jet flow is achieved by reducing the pressure at the inflow
section within the nacelle that results in an increase in the
flow rate that in turn is imposed on the outflow boundary
of the engine. No bypass flow, temperature increase or
swirl of the jet is considered, which is consistent with the
experimental setup.

Fig. 1. Sketch of the rectangular wing, the various used engine positions are
3. Experimental setup denoted by A, C and E.

3.1. Wind tunnel


3.3. Engine model
The experiments with wing/engine configuration are
carried out in the low-speed wind tunnel of the Aerodynami-
sches Institut Aachen. It is a closed circuit wind tunnel with The engine model consists of three parts, shown in
an open test section, which is 1.80 m long and has a diameter Fig. 2. It is a jet apparatus using pressurized air of 8 ×
of 1.2 m. The maximum wind speeds are up to 60 m/s. 105 Pa absolute pressure. The outer geometry corresponds
The test setup consists of a horizontal plate with an elliptic to a Rolls-Royce Trent 900 engine. Its typical dimensions
nose profile that cuts off the boundary layer of the wind agree with real engines used for large aircrafts such as the
tunnel nozzle. The wing model is mounted vertically on the Airbus A380. The engine is made out of aluminium and
plate. For the measurements the wind tunnel is operated at weighs less than 0.2 kg. To minimize reflections of the
a constant flow speed of u∞ = 27 m/s yielding a Reynolds laser light sheet a mat surface was generated by anodizing
number based on the chord length of Rec = 2.8 × 105 . the model. From the leading edge of the wing a small
flexible tube is introduced into the pylon. The engine is
3.2. Wing model mounted onto the wing using screws such that the tube also
is fixed.
The nozzle is put into the nacelle leaving a small gap
For the tests a half-wing model with rectangular shape
between them, the width of which is used to modify the
is used, which represents the cruise configuration, i.e., it
engine jet. Due to technical limitations the measurements are
is not equipped with any high-lift devices. It possesses a
performed at jet speeds of approximately uengine = 47 m/s,
BAC 3-11/RES/30/21 profile with a constant chord length
yielding uengine/u∞ = 1.74.
of c = 150 mm and a threedimensional wing-tip geometry.
The wing is built with an outer shell made out of glass fiber
sheets and a steely core plate. The interspace is filled with
plastics. At five different positions located in the spanwise
direction an engine model can be mounted to the wing.
The mounting points are denoted by A to E, with A
being the nearest position to the wing-tip. Fig. 1 shows a
sketch of the wing, indicating the various engine positions.
To reduce reflections of the laser light sheet the wing
has a black surface. During the experiments the wing-tip
is located 680 mm above the measurement plate. With a
semi-span of s = 680 mm the aspect ratio is λrw = 9.06.
The flow field was tripped at 5% chord length by a zig-
zag tape. Throughout the measurements the angle of attack Fig. 2. Sketch and photo of the engine model and its three components
was α = 8◦ . pylon, nozzle, nacelle.
178 G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183

3.4. Measurement technique a mean particle shift of ≈ 6 px. which corresponds to


 4.5 m/s. In future investigations a modification of
1 px. ≈
Velocity distributions in the wake of the wing/engine con- the experimental setup will allow equal access to forward
figuration are measured by the particle image velocimetry scattered light for both 3C-PIV cameras, also in the area
(PIV) method. In different test cases velocity components of 1.5 < x/c < 3.5.
in planes orthogonal to the wind tunnel flow are recorded
at several locations downstream of the trailing edge. For 3.5. Data acquisition
the 2C-PIV arrangement a PCO SensiCam with a 1280 ×
1024 px. CCD and 12-bit resolution is installed in the dif- The PIV cameras are connected to I/O-cards of conven-
fuser section of the wind tunnel, approximately 2.10 m tional PC’s by light-wave cable. They are operated using
downstream the wing trailing edge and orientated perpen- the PCO CamWare-Software. Each camera records up to
dicular to the laser light sheet. The Stereo- or 3C-PIV setup 60 images in ∗.b16-format at a frequency of 2 Hz. For the
comprises two SensiCams arranged under the Scheimpflug- 3-camera setup more than 900 MB data are recorded for each
condition with an optimum angle of ±45◦ between the axis wing/engine-configuration such that the total amount of data
of the lenses and the wind tunnel centerline [20]. for this specific measurement campaign exceeds 40 GB.
Since they are mounted outside the air flow the cameras
are looking either up or downstream, depending on the
distance from the measurement plane to the trailing edge. To 4. Results
be more precise planes at x/c  1.5 are investigated by the
3C-PIV setup looking upstream, planes at x/c  3.5 looking 4.1. Numerical results
downstream.
The planes at 0.5  x/c  4.0 are investigated by 2C-PIV Six numerical simulations are run with three different po-
with a spacing x/c = 0.5. The light sheet is generated by a sitions of the engine (Fig. 1) and two different jet velocities,
double pulsed Spectron Nd:YAG-Laser with a power rating representing cruise and take-off situation, respectively. The
of 150 mJ. The cameras and the light sheet are controlled by simulation of the flow field around the wing is performed on
an external ILA synchronizer. Some of the measurements structured multiblock grids consisting of 74 blocks and 1.1
are done by a parallel or superimposed combination of million grid cells. The domain of integration around the wing
2C- and 3C-PIV setup, using three PCO SensiCams and has a size of approx. 15c. The wake flow calculation ranging
an additional signal converter after the synchronizer. Fig. 3 to x/c = 4.0 is performed on a stretched cartesian grid con-
gives an overview of the 3C-PIV arrangement, showing the sisting of 7.0 million grid cells. Qualitatively the pressure
cameras looking upstream. distribution on the wing, not shown here, is not affected by
Considering a third camera, installed in the diffuser, it is the presence of the engine or the jet velocity, since the dis-
evident, that each camera due to the spatial non-uniformity tance of the engine and the main wing is relatively large. This
of lightscattering sees a different amount of laserlight, is a consequence of the defaults given by the pylon geometry
which is scattered by the DEHS seeding-particles [22]. of the experimental model. For all six configurations only a
To partially compensate this effect the grey levels of the minor change of less than 1% in the cL coefficient is deter-
particle images are adapted before the evaluation process, mined. Thus an almost equal strength of the circulation Γ of
which is done using the inhouse written software PIA. The the wing-tip vortex is obtained. On the other hand, a change
typical interrogation window size was 64 × 64 px. with of almost 10% of the cD coefficient is noticed. The smaller
the distance between the engine and the wing-tip, the higher
the drag. This tendency is found at both jet velocities.
The formation of the wing-tip vortex is documented in
Fig. 4 at x/c = 1.0 behind the trailing edge for the six con-
figurations simulated. The velocity distribution u indicates
not only the wing-tip vortex but also the shear layer of the
trailing edge. It is apparent that the position and the form
of the wing-tip vortex is almost identical in all cases. This
result is in agreement with the aforementioned negligible
change in the lift coefficient cA and the circulation. At po-
sition x/c = 2.0 first small differences are visible (Fig. 5).
The position of the wing-tip vortex and the inner core are
still similar. This behaviour extends to position x/c = 3.0 as
shown in Fig. 6. It is interesting to note that the jet veloc-
ity plays a minor role for the engine at position A, whereas
Fig. 3. Sketch of the 3C-PIV setup with the CCD-cameras looking a more evident impact occurs at position C and E. Further-
upstream. more, the formation of the shape of the vortex at engine po-
G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183 179

engine position A engine position C engine position E


uengine /u∞ = 1.74 uengine /u∞ = 1.74 uengine /u∞ = 1.74

engine position A engine position C engine position E


uengine /u∞ = 3.48 uengine /u∞ = 3.48 uengine /u∞ = 3.48

Fig. 4. Distribution of the streamwise velocity component u at x/c = 1 at various engine positions and different jet speeds.

engine position A engine position C engine position E


uengine /u∞ = 1.74 uengine /u∞ = 1.74 uengine /u∞ = 1.74

engine position A engine position C engine position E


uengine /u∞ = 3.48 uengine /u∞ = 3.48 uengine /u∞ = 3.48

Fig. 5. Distribution of the streamwise velocity component u at x/c = 2 at various engine positions and different jet speeds.

sitions C and E show the same behaviour in the region stud- Concerning the shear layer it can be stated that hardly any
ied. The differences are interpreted as a consequence of the impact of the jet velocity occurs at the most outboard engine
interaction between the engine jet and the wing-tip vortex. position (A) whereas especially at x/c  2.0 the stronger en-
180 G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183

engine position A engine position C engine position E


uengine /u∞ = 1.74 uengine /u∞ = 1.74 uengine /u∞ = 1.74

engine position A engine position C engine position E


uengine /u∞ = 3.48 uengine /u∞ = 3.48 uengine /u∞ = 3.48

Fig. 6. Distribution of the streamwise velocity component u at x/c = 3 at various engine positions and different jet speeds.

gine jet results in a more pronounced weakening of the shear for the engine at position A evidence the development of
layer at positions C and E. the wing-tip vortex (at y/c ≈ 0.0) and the engine jet (at
y/c ≈ −1.1). The jet is inclined at 8◦ to the free stream
4.2. Experimental results flow. It drifts away from the trajectory of the wing-tip vor-
tex and leaves the measurement planes. Obviously, even for
First, 2C-PIV measurements for the rectangular wing the most outboard engine position there is no direct interac-
model are conducted to investigate the wake in five planes tion between wing-tip vortex and engine jet. Consequently
downstream of the trailing edge. In the spanwise direction therefore should be none when the engine is located fur-
the area captured extended from engine position A to ap- ther inboard, i.e., farther away from the wing-tip. If the
proximately 30 cm away from the√wing-tip. In Fig. 7 con- 2C-PIV vortex measurements in Fig. 8 are compared with
tours of the tangential velocity v 2 + w2 in five planes the corresponding 3C-PIV data in Fig. 9, some differences


Fig. 7. Velocity contours V = v 2 + w 2 in the wake of the rectangular wing with engine at position A (at y/c = −1.1).
G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183 181

engine position A engine position C engine position E



Fig. 8. 2C-PIV measurements in the yz-plane at x/c = 3.5, contours of the in plane velocity V = v2 + w2 .

engine position A engine position C engine position E



Fig. 9. 3C-PIV measurements in the yz-plane at x/c = 3.5, contours of the in plane velocity V = v2 + w2 .

engine position A engine position C engine position E

Fig. 10. 3C-PIV measurements in the yz-plane at x/c = 3.5, distribution of the streamwise velocity component u at various engine positions.

appear. The tangential velocities in the vortex core of the sented in Fig. 6 both experiment and simulation possess a
2C-PIV evaluation exceed the 3C-PIV values approximately smaller u in the vortex center for engine positions C and E
by 2 m/s. The discrepancy between the measurement tech- than in position A. More precisely the velocity at the center
niques arises from the perspective error due to the out-of- changes from u ≈ 24 m/s (pos. A) to u ≈ 23 m/s (pos. C)
plane component of the fluid, which cannot be accounted and u ≈ 22 m/s (pos. E). This tendency corresponds to the
for in 2C-PIV measurements. According to the numerical numerical findings described in Fig. 6. This behaviour is re-
results the 3C-PIV velocity distributions of engine position garded as an effect of the interaction of the wing-tip vortex
C and E resemble each other more closely than the flow field with the engine jet. The development of the wing-tip vor-
of position A. Fig. 10 shows the color coded velocity distri- tex (Fig. 11) shows an increase in u in the vortex core from
bution u in the yz-plane of the foregoing figure. Compared plane x/c = 1.5 to x/c = 3.5. 2C-PIV measurements for en-
with the numerical results for uengine/u∞ = 1.74 and pre- gine position A also show an increase in the tangential ve-
182 G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183

x/c = 1.0 x/c = 1.5

x/c = 3.5 x/c = 4.0

Fig. 11. 3C-PIV measurements in yz-planes x/c = 1.0 (upper left), x/c = 1.5 (upper right), x/c = 3.5 (lower left), x/c = 4.0 (lower right), engine at position A,
distribution of the streamwise velocity component u.

x/c = 1.0 x/c = 4.5

Fig. 12. 3C-PIV measurements of the engine jet in the yz-plane at x/c = 1.0 (left) and x/c = 4.5 (right), distribution of the streamwise velocity component u.

locity from plane x/c = 2.0 to x/c = 3.0. This is caused by In Fig. 12 the deformation of the engine jet is visible.
the mixing of the vortex flow with the accelerated flow field The pictures show the color coded u-component of the jet.
of the upper side of the wing. Even at plane x/c = 4.0 there Note that various areas are visualized. In the left picture the
still exists an excess of u in the vortex core compared to wake of the wing is still visible. In the right picture the jet
plane x/c = 1.5. possesses a larger diameter and a smaller centerline velocity,
G. Huppertz et al. / Aerospace Science and Technology 8 (2004) 175–183 183

due to the entrainment of the flow. The shape of the jet has [4] E. Fares, M. Meinke, W. Schröder, Numerical simulation of the
changed from an almost circular to a nephritic structure. This interaction of flap side-edge vortices and engine jets, in: ICAS 2000
is a consequence of the superposition of the engine jet flow, Congress Proceedings, ICA0212, 2000.
[5] E. Fares, W. Schröder, Zonal approach for the solution of the Navier–
which is characterized by strong velocity gradients for the u-
Stokes equations, ZAMM 2000 80 (1) (2000).
component, and the vortex wake with weaker gradients for [6] E. Fares, W. Schröder, Numerical simulation of wake flows, in:
the tangential velocity. In other words the deformation of ECCOMAS CFD 2001 Congress Proceedings, 2001.
the jet shape evidences the mutual impact of engine jet and [7] E. Fares, G. Huppertz, R. Abstiens, W. Schröder, Numerical and
wing-tip vortex. experimental investigation of the interaction of wing-tip vortices and
engine jets in the near field, AIAA Paper, 2002–0403, 2002.
[8] F. Garnier, S. Brunet, L. Jacquin, Modelling exhaust plume mixing in
5. Summary the near field of an aircraft, Ann. Geophys. 15 (1997) 1468–1477.
[9] T. Gerz, T. Ehret, Wake dynamics and exhaust distribution behind
cruising aircraft, AGARD-CP-584, 35, 1996.
In this paper the very near wake up to five chord lengths
[10] T. Gerz, F. Holzäpfel, D. Darracq, Aircraft wake vortices – a position
behind the trailing edge of a rectangular wing with an paper, WakeNet Position Paper, 2001.
engine model is discussed. The investigation comprises [11] L. Jacquin, F. Garnier, On the dynamics of engine jets behind a
numerical and experimental analysis with variations of transport aircraft, AGARD-CP-584, 35, 1996.
the engine position and the jet speed. It is fair to say [12] A. Jameson, Solution of the Euler equations for two-dimensional
that a good qualitative agreement of the computations transonic flow by a multigrid method, Appl. Math. Comp. 13 (1983)
and the measurement is achieved. The tangential velocity 327–355.
[13] M.R. Khorrami, B.A. Singer, R.H. Radeztsky Jr., Reynolds averaged
distributions show a little impact of the engine position and
Navier–Stokes computations of a flap side-edge flow field, AIAA
exit velocity on the formation and the behaviour of the Paper, 98-0768, 1998.
wing-tip vortex. The velocity distribution of the streamwise [14] M.S. Liou, A sequel to AUSM : AUSM + , J. Comput. Phys. 129 (1996)
component u however reveals to be significantly influenced 164–382.
by the spanwise distance between the engine and the [15] M.S. Liou, Ch.J. Steffen, A new flux splitting scheme, J. Comput.
wing-tip. At smaller distances a lower deficit of the axial Phys. 107 (1993) 23–39.
velocity in the vortex center compared to the free stream [16] M. Meinke, D. Hänel, Time accurate multigrid solutions of the Navier–
Stokes equations, in: W. Hackbusch, U. Trottenberg (Eds.), Multigrid
occurs. Overall the engine jet therefore is relevant for the
Methods III, Birkhäuser, 1991, pp. 64–88.
characteristics of the shear stress in the streamwise direction
[17] M. Meinke, E. Krause, Simulation of incompressible and compressible
of the vortical field. flows on vector-parallel computers, in: Parallel Computational Fluid
Dynamics: Proceedings of the Parallel CFD’98 Conference in Taiwan,
1999, pp. 25–34.
Acknowledgement [18] R. Paoli, F. Laporte, B. Cuenot, T. Poinsot, Dynamics and mixing in
jet/vortex interactions, Phys. Fluids 15 (7) (2003) 1843–1860.
This work is funded by the Deutsche Forschungsgemein- [19] T.J. Poinsot, S.K. Lele, Boundary conditions for direct simulations of
compressible viscous flows, J. Comput. Phys. 101 (1992) 104–129.
schaft within the collaborative research program SFB 401.
[20] A.K. Prasad, Stereoscopic particle image velocimetry, in: Experiments
in Fluids, vol. 29, Springer-Verlag, 2000, pp. 103–116.
[21] T.R. Quackenbush, M.E. Teske, A.J. Bilanin, Dynamics of exhaust
References plume entrainment in aircraft vortex wakes, AIAA Paper, 96-0747,
1996.
[1] J.S. Chow, G.G. Zilliac, P. Bradshaw, Mean and turbulence measure- [22] M. Raffel, Ch. Willert, J. Kompenhans, Particle Image Velocimetry –
ments in the near field of a wingtip vortex, AIAA J. 35 (10) (1997) A Practical Guide, Springer-Verlag, Berlin, 1998.
1561–1567.
[23] P.R. Spalart, Airplane trailing vortices, Ann. Rev. Fluid Mech. 30
[2] J. Dacles-Mariani, G.G. Zilliac, J.S. Chow, P. Bradshaw, Numer-
(1998) 107–138.
ical/experimental study of a wingtip vortex in the near field,
[24] K.W. Thompson, Time dependent boundary conditions for hyperbolic
AIAA J. 33 (9) (1995) 1561–1568.
[3] E. Fares, M. Meinke, W. Schröder, Numerical simulation of the systems II, J. Comput. Phys. 89 (1990) 439–461.
interaction of wingtip vortices and engine jets in the near field, AIAA [25] B. van Leer, Toward the ultimate conservative difference scheme V.
Paper, 2000–2222, 2000. A second-order, J. Comput. Phys. 32 (1979) 101–136.

Vous aimerez peut-être aussi