Sunteți pe pagina 1din 16

A Quasi-Unsteady Description of Windscreen Wiper

Induced Flow Structures.

A P Gaylard†, A C Wilson† and G S J Bambrook‡


† ‡
Jaguar Land Rover, TASE & Weight and Jaguar Land Rover Research

SYNOPSIS
This paper draws on full scale wind tunnel flow visualisation and CFD simulation,
carried out on a full-sized SUV geometry, to provide a description of windscreen
wiper induced flow structures. The focus of this work is the effect of wipers on the
local flow, rather than the more usual consideration of the aerodynamic forces
exerted on the wipers.

The flow structure was analysed for a series of fixed wiper arm orientations, as well
as a bare windscreen (no wipers) reference case. This enabled the identification of a
number of coherent vortex structures. Evidence is also presented which indicates
that these structures are present in other vehicle types. Further, the vortex structures
associated with the wiper blade and arm are seen to convect downstream,
maintaining their coherence well onto the vehicle roof. It is suggested that during a
dynamic wiper cycle these vortex structures are swept over the screen and roof. This
raises the possibility of aeroacoustic sources remote from the wiper system location
as well as interactions with open sunroofs.

1. INTRODUCTION

The presence of windscreen wipers is, of course, essential for the maintenance of
forward vision during adverse weather conditions. It is also the case that they are
aerodynamically active components, both subject to aerodynamic loading and a
source of aeroacoustic noise.

Initial consideration of wiper aerodynamics was mainly concerned with the forces
experienced by the wiper system and ensuring that they did not degrade wiper
function to an unacceptable degree [1,2].

Latterly they have been considered as an aeroacoustic source [3]; the precise
mechanism of noise generation subjected to scrutiny by fundamental numerical and
experimental investigations [4].

The aim of this paper is to provide an overview of the characteristic vortex structures
generally associated with current wiper systems, by reference to previously published
work, surface flow visualisation and numerical simulation.

2. PREVIOUS STUDIES

2.1 Review
Previous studies have looked at the important issue of maintaining an acceptable
wiper performance in the presence of aerodynamic lifting forces acting on the arm
and blade [5,6].
In an early study Clarke and Lumley [1] provided a comprehensive survey of the
problems associated with windscreen wiper operation. Reduced scale water tank
experiments (1/20th) were used to survey the flow structure over the screen of four
different vehicles. Further, wiper arm pressure was measured both at full scale and
on a wiper mounted in a rectangular duct. Finally, tests were carried out on a half
scale test rig, representing the bonnet and windscreen of a car. The work considered
the influence of vehicle geometry, along with the size and shape of the screen.

Their motivation was increasing vehicle speeds and the advent of "wrap-around"
windscreens. They were able to demonstrate both the typical radial flow pattern
seen on windscreens and flow acceleration due to screen curvature.

The use of an aerofoil to produce "anti-lift [sic] properties" was suggested, along with
some other, less practical, ideas.

Dawley [2], in a review of aerodynamic effects on automotive components, elucidated


the flow pattern and pressure distribution over a 3/8th scale model of a two door hard
top vehicle. This work again demonstrated the typical radial flow over an
unobstructed windscreen. The author made the, now typical, observation that the
passenger's side wiper arm tended to be parallel to the onset flow, whereas the
driver's side wiper tends to be perpendicular to the onset flow, at least through a
significant (perhaps considerable) portion of its travel.

The use of a spoiler formed from a continuous arcuate aerofoil section was
investigated. The relationship between its angle to a perpendicular onset flow and
wiper lift force was investigated for a fixed speed. A "wiper blade wind tunnel" was
used to generate this data; effectively a test wiper mounted in the collector of a wind
tunnel. This demonstrated that a practical wiper-mounted aerofoil could be used to
generate negative lift (referred to as "antilift" and now more commonly as
"downforce".)

Some fundamental work has been undertaken to determine the basic relationships
between wiper geometry and aerodynamic forces. Barth [7], for example, provided
insight into the relationship between the lift and drag forces acting on a wiper blade
and the tilt angle of the blade with respect to the screen.

Latterly, CFD models have been used to provide both quantitative and qualitative
evaluations of wiper performance. Strumolo et al [8], for instance, constructed a CFD
model of a simplified 3-box saloon that included a wiper system in the parked
position and a "leaf screen cavity" (cowl). The main focus of this work was to
understand the aerodynamic mechanisms underlying "water blow back" [sic].

Jallet et al [5] also provided a numerical simulation of wiper system aerodynamic


behaviour. This work focused on validating the simulation technique for the
prediction of both wiper drag and lift forces. A single wiper blade and arm were
modeled mounted on a flat plane and perpendicular to a uniform onset flow. Two
different wiper blade designs were evaluated, both with and without a spoiler. The
calculated drag and lift forces were seen to compare reasonably well with experiment
(Differences of 1%-7% for drag and 2%-9% for lift with an onset flow of 40m/s).

Billot et al [6] took this validated computational methodology and applied it to a wiper
system at a mid-wipe position (30°), installed on a realistic car geometry. The
mechanical distortion of the blade/arm under the spring load was calculated by an FE
program prior to the aerodynamic simulation. The results of the aerodynamic
simulation were then fed back into the FE model. Thus the authors were able to
determine the effect of aerodynamic lift on the net pressure applied by the wiper
blade, along its length, to the windscreen.

Moving away from purely aerodynamic wiper performance considerations, Sanon


and Jallet [3] examined windscreen wipers as a local aeroacoustic noise source.
This was done by wind tunnel and road based test work, supported by CFD
simulation. Again, realistic car geometry was used with the wipers fixed at a mid-
wipe (30°) position.
2.2 Flow Topology
The work reviewed here also provides insights into the local flow structure around the
wiper systems.

For instance, Dawley [2] proposed a two-dimensional view of the flow structure over
a wiper blade and spoiler. This comprised a recirculation in front of the blade with a
region of separated flow behind it.

The simulations of Stromolo et al [8] showed a time-averaged flow pattern with three
recirculation regions (trapped vortices) around the wiper arm and blades.
(i) Leaf screen cavity (cowl).
(ii) Upstream of the blade/arm.
(iii) Downstream of the blade/arm.

Taking slices through the trapped vortices (ii) and (iii) reveals a flow structure with
some similarities to that suggested by Dawley [2], at least upstream of the wiper
blade. It should be noted though that Dawley's two-dimensional reconstruction
shows a turbulent separated zone downstream of the blade, rather than the coherent
trapped vortex evident in later work.

The simulations of Jallet et al [5] showed both of the trapped vortices upstream and
downstream of the wiper blade/arm along with a wake. This work showed substantial
lateral variation of the downstream vortex, due to interactions with flow through and
over various small geometric features.

The paper by Sanon and Jallet [3] contains experimental and computational flow
visualisation for a mid-wipe (30°) configuration, with and without a spoiler. The
downstream arm/blade vortex is clearly present along with the wake. The lateral
(spanwise) variation in these structures caused by the stacked blade/arm elements is
also evident.
This paper seeks to build on these observations, adding in additional data, to provide
a comprehensive qualitative description of the flow structures induced by typical
wiper systems. This is done by reference to surface flow visualisation obtained in a
full scale wind tunnel, along with some CFD simulations.

3. WIPER LOADS AND DYNAMICS

In common with the published literature, this paper neglects the dynamic aspect of
wiper performance. However, it is important to briefly review the complexities of both
wiper dynamics and the time-varying aerodynamic forces that they are subjected to.

The aerodynamics and mechanics of wiper systems is complex. As the wipers


sweep across the windscreen the blade and arm orientation to the flow changes
dynamically. Typically, on the driver's side, the wiper is almost perpendicular to the
main flow through most of the sweep, as the flow over the windscreen is largely
radial. Additionally, the angle of the blade to the screen surface is different on the
up-sweep and down-sweep. During the upsweep, on the driver's side, the wiper
moves with the prevailing flow; on the down-sweep it moves against it.

Thus both the aerodynamic drag and lift forces on the wiper system (predominantly
the blade and arm) vary substantially during the wiper sweep. The aerodynamic
loads also vary with onset flow velocity.

To sweep the blade across the screen the motor must overcome both surface friction
and aerodynamic drag. Aerodynamic lift forces can tend to pull the blade away from
the screen, degrading wiper performance. This force is opposed by a spring
mounted in the wiper arm which pulls the arm towards the screen applying a load
onto the blade. Historically, many wiper designs have included cantilevers to transfer
the load from the arm out laterally along the blade. Recently, "Beam Blade" designs
have become popular. These do not include cantilever elements, so the spring load
is applied at the centre of the blade only.

Finally, the blade is designed to conform to the contours of the windscreen as it


sweeps over the surface. Thus the curvature of the wiper blade/arm system changes
dynamically.

4. SIMPLIFYING ASSUMPTIONS

Having reviewed the complexities of wiper aerodynamics it is understandable that the


work published to date uses a range of simplifying assumptions. These are
summarised in Table 1(below).

The work reported in this paper uses similar simplifying assumptions. In the
experimental work, the wiper sweep is represented by considering fixed positions on
the windscreen. Thus the relative movement between the wiper and onset flow is not
captured. The CFD model is similarly static and does not include the change in
blade angle with respect to the screen seen between the up-sweep and down-sweep.
Further, the wiper geometry has been adjusted and morphed to match the screen
profile, but cannot deform as a result of the aerodynamic load.
However, this quasi-unsteady investigation is based on detailed vehicle and wiper
system geometry.

Reference
Authors
Clarke Dawley Sanon Fischer Jallet Billot Barth Strumolo
Simplifying
and and and et al et al et al
Assumption
Lumley Jallet Zuccini
Reference Number in Text
[1] [2] [3] [4] [5] [6] [7] [8]
Experiment/CFD Y/N Y/N Y/Y Y/Y Y/Y Y/Y Y/N N/Y
Two-
N N* N N** N N N N
Dimensional
Fixed Wiper
Y Y Y Y Y Y Y Y
Position
Flat Inclined
Screen (No N N N Y N N N N
Vehicle)
Flat Horizontal
Screen (No N Y Y N Y N Y N
Vehicle)
Normal Onset
Y Y Y N Y N Y N
Flow
Part vehicle Y N N N N N N N
Simplified
N N N N N N N Y
Vehicle
Realistic Vehicle Y N N Y/N N N
Deformable
Y Y Y/N Y/N Y/N Y n/a N
Wiper
*Three-dimensional experiment with two-dimensional flow structure proposed
**Geometry invariant laterally with finite width
Test rig comprised bonnet and screen.
Full scale and model scale (water tunnel) tests carried out also.

Table 1 Simplifying Assumptions Made In Previous Work.


The following sections use the results of both experiment and CFD simulation to
illustrate the wiper induced flow structures seen on a full sized SUV. First, the
experiments and CFD simulations are described.

5. SIMULATION

5.1 Wind Tunnel Experiments


The experiments were performed in the MIRA Full Scale Wind Tunnel. This is a
closed test section, open return facility with a 4.4m (H) by 7.9m (W) test section. Its
four fans are able to develop a maximum flow velocity of 130 km/h. The empty
tunnel has a background turbulence intensity of 1.8% and a 200mm characteristic
length [9].
The flow visualisation data presented in this paper were obtained by testing either
production level vehicles or aerodynamics simulators (buck) at zero yaw with a
nominal test speed of 100 km/h. Time-averaged surface flow visualisation patterns
were obtained using a proprietary UV fluorescent dye suspended in white spirit,
running the wind tunnel for a period of 5-10 minutes.
5.2 CFD Simulation
Further insight into the flow structures of interest is provided by CFD simulation. It is
presented here as indicative flow visualisation. No local measurements are provided
to indicate its reliability. However, the CFD tool that has been used (Exa
PowerFLOW™) has been previously validated for both aerodynamics [10,11] and
aeroacoustics simulation [12,13].

The CFD model represents a fully-detailed full sized SUV with closed-cooling intakes.
Onset-flow conditions have been matched to MIRA FSWT. Around the windscreen
and wipers the smallest volume element (voxel) has a characteristic dimension of
1.1mm.

Simulation Parameter Bare Screen Parked Mid-Wipe


No. Volume Elements /106 42.03 108
No. Surface Elements /106 6.05 14.39
Time Step Length /106 .s 6.187 6.187 6.187
Averaging Period /s 0.528 0.387 0.187
Reynolds Number /106 4.03 4.03 4.03
Table 2 Basic Details for the CFD (PowerFLOW™) Simulations

The CFD simulations have been interrogated to provide the following insights into the
flow structure.

• Surface streamlines, to facilitate direct surface flow structure comparison with


the wind tunnel experiments.
• Flow streamlines, to capture the off-surface flow structures.
• Contour plots of vorticity magnitude ( ×u) on a vertical plane through the roof
header, enabling the identification of regions of high vorticity.
• Vortex core detection using closed iso-surfaces of the λ2 parameter. (This is
the second largest eigenvalue of the symmetric tensor S2+Ω Ω2; where S and
Ω are the symmetric and antisymmetric parts of the velocity gradient tensor
∇u. Jeong and Hussain [14] proposed this approach and demonstrated that
λ2 is negative in vortex cores. A physical interpretation of λ2 is that it identifies
vortex cores as pressure minima due purely to vortical motion. Therefore, in
theory, λ2 < 0 implies a vortex core. However, in practice a larger negative
value is required to provide meaningful detection, thus a value of -200 was
selected for this work†,‡. In addition, vortex core detection was improved by
using the final data frame of the calculation (averaged over 2000 time steps),
rather than an average over the complete simulation period.)

† th
Freed, D. M. Letter to author. Dated 12 July 2006.

Zaehring, E., Letter to author. Dated 12th July 2006.
6. FLOW STRUCTURES

6.1 Bare Windscreen – No Wipers.


In the absence of the disruption caused by the presence of the windscreen wipers
the flow over a typical automotive windscreen is largely radial. This can be clearly
seen in the surface flow visualisation presented in Figure 1. The lower part of the
screen also shows the influence of a trapped lateral (“cowl”) vortex. The shear layer
reattachment line is clearly visible, along with the reversed flow region associated
with this trapped vortex.

Shear layer reattachment line

Figure 1 Surface Flow Visualisation On A Bare Windscreen.


The CFD simulation shows the same surface flow features (Figure 2). Some of the
additional field data available in the numerical simulation is presented in Figures 3
and 4. Both of these show the vorticity present in the plane of the roof header.
Vorticity in both the boundary layer and a-pillar vortex is evident.

Figure 2 Surface Streamlines On A Bare Windscreen.


Figure 4 attempts to provide detection of the vortex cores present in the flow via iso-
surfaces of λ2. Multiple vortex cores are evident within the cowl region. A small
trapped vortex is also seen at the front of the roof header, caused by the forward
facing step formed by the windscreen glazing inset. This metric also seems to be
sensitive to vorticity in the a-pillar and roof header boundary layers.

Figure 3 Flow Streamlines (Blue: Cowl Vortex; Red: Reattaching Shear Layer
And Attached Flow) On A Bare Windscreen.

Figure 4 Iso-Surfaces Of λ2 On A Bare Windscreen.


6.2 Parked Wipers
Introducing windscreen wipers, in their parked position, clearly disrupts a region of
the previously radial flow (Figure 5). The surface flow visualisation clearly shows a
region of separated flow consistent with the blade/arm wake. However, the reduced
energy in the flow downstream of the arm/blade has reduced the effectiveness of the
flow visualisation technique.

Figure 5 Surface Flow Visualisation On A Windscreen With Parked Wipers.


More detailed insights into the flow structure are provided by the CFD simulation.
Figure 6 shows a near-wiper flow structure similar to that described by Stromolo et al
[8], i.e. a cowl vortex along with vortices upstream and downstream of the wiper
blade/arm, the latter forming the wiper/arm wake.

Figure 6 Surface Streamlines On A Windscreen With Parked Wipers.


The flow streamlines shown in Figure 7 show some fluid being drawn from behind the
wiper arm by entrainment into the cowl vortex. Some disruption of the vortices
upstream of the blade and arm by the lower arm on the driver's wiper system (B) is
also evident. The vortex core plot (Figure 8) confirms this assessment.
B

Figure 7 Flow Streamlines Over A Windscreen With Parked Wipers. (Blue:


Cowl And Windward Wiper Vortex; Green: Leeside Wiper Vortex; Red:
Reattaching Shear Layer And Attached Flow).

Figure 8 Iso-Surfaces Of λ2 Over A Windscreen With Parked Wipers.


6.3 Mid-Wipe
Moving the wipers to a mid-wipe position changes the orientation and position of
some of the vortex structures previously identified. Figure 9 shows that the vortices
on either side of the wiper blade/arm now convect downstream over the roof header
(A). The vortex upstream of the driver's side wiper arm/blade is very clearly seen (ii).
A A

(iii) C

D
(ii)

Figure 9 Surface Flow Visualisation On A Windscreen With Wipers At The Mid-


Wipe Position.

The arm/blade vortices, (ii) and (iii) clearly convect downstream off the screen and
over the roof header (A). This could give rise to an aeroacoustic noise source
relatively remote from the wiper system.

A A
(iii)
C

(i)

Figure 10 Surface Streamlines On A Windscreen With Wipers At The Mid-Wipe


Position.
Figure 10 provides a very clear visualisation of the cowl vortex (i) and downstream
wiper blade/arm vortices (iii). (However, the upstream vortex (ii) is not seen in the
simulation, likely due to limitations imposed by the spatial resolution of the CFD
model.) The contributions of the wiper knuckle (C) and blade/arm gap (D) to the
driver's side wiper system wake are also evident.

The downstream convection of the arm/blade vortices is shown very clearly in


Figures 11 and 12. Vorticity peaks are seen over the roof header that correlate with
the rotating fluid shown by the flow streamlines and the vortex cores highlighted by
the λ2 plot. An additional (though less coherent) vortex structure is generated by the
driver's side knuckle (iv).
A A

(ii) (iv)
(ii)
(iii)
(iii)
C

(i) (i)

Figure 11 Flow Streamlines Over A Windscreen With Wipers At The Mid-Wipe


Position. (Blue: Cowl And Windward Wiper Vortex; Green: Leeside Wiper
Vortex; Red: Reattaching Shear Layer And Attached Flow).

Figure 12 Iso-Surfaces Of λ2 Over A Windscreen With Wipers At Mid-Wipe


(Zero Degrees Yaw).

6.4 Complete Sweep

Having explored the relevant flow structures in some detail experimental, surface
flow visualisation is presented in for five fixed positions representing a complete wipe
cycle is presented in Figure 13. (If wiper movement were considered this would only
represent a half cycle.)
The individual images
follow the sequence below. (a)

(a) Parked.
(b) 30 degrees.
(c) Mid-wipe.
(d) Passenger's side
wiper at 90 degrees.
(e) End of wipe.

As has been frequently (b)


observed, the angle of the
passenger's side wiper
system to the local flow
results in relatively little
disruption of the flow over
the screen. However,
vortices still form parallel to
the blade and arm on the
upstream and downstream (c)
sides. The lower arm
(close to the spindle) also
provides a limited amount
of local flow distrubance.

The driver's side arm and


blade spend much of the
sweep cycle nearly
perpendicular to the local (d)
onset flow. Hence they
perturb the flow more
strongly. Coherent
upstream and downstream
vortices form along the
blade/arm which convect
downstream.

A quasi-unsteady
(e)
interpretation of these
images suggests that the
wiper blade/arm vortices,
(ii) and (iii), persist over
most of the wipe cycle and,
convecting downstream,
are repeatedly swept over
the screen and roof. In
addtion, for part of the wipe
the driver's wiper arm
knuckle generates a wake Figure 13 Five Fixed Wiper Positions
vortex (iv) which is swept over the screen and roof.

Finally, the cowl vortex (i) remains in a fixed postion, though when the wiper system
approaches the parked position there is interaction between the cowl vortex and
wiper arm/blade vortices.

7 OBSERVATIONS FOR OTHER VEHICLES

The flow structures described for this particular SUV geometry are commonly seen
for other, very different, vehicles. Figure 14, for example, shows the wiper-induced
flow structure for a luxury saloon (a) and sports car (b).

These observations indicate that the description presented is reasonably general. In


particular, the convection of the arm/blade vortices over the roof header onto the roof
is very clear in both cases. In the case of the luxury saloon (a) they pass over the
closed sunroof opening, raising the possibility of an interaction between these
vortices and open sunroofs. Of course, the relative size and strength of the
structures may well vary with screen rake and wiper system design.

(a) (b)

Figure 14 Surface Flow Visualisation On The Windscreen Of A Luxury Saloon


(a) And Sports Car (b), With The Wipers At Mid-Wipe.

8 CONCLUSIONS

A description of the flow topology associated with windscreen wiper systems has
been proposed. Although there are many small features associated with the
geometric complexity of these systems, large coherent vortex structures can be
identified. The main elements in the flow structure are:

(i) cowl vortex;


(ii) upstream wiper arm/blade vortex;
(iii) downstream wiper arm/blade vortex;
(iv) driver's side knuckle wake vortex.

It is asserted that (ii) and (iii) are swept laterally across the screen and roof during
wipe cycles, with (iv) being present for a part thereof. However, the authors freely
acknowledge the limitations of this work. In particular the use of static wiper
positions in an attempt to understand a dynamic process entails a degree of risk.

Finally, one of the more interesting features of this work has been the identification of
the extent to which the wiper arm/blade vortices convect downstream. This raises
the possibility of an aeroacoustic noise source relatively remote from the wiper
system. Further, the possibility exists of an Interaction between these vortices and
sunroof openings.

ACKNOWLEDGEMENTS

The authors would like to thank Jaguar and Land Rover for permission to publish this
paper and their colleagues for many helpful suggestions.

REFERENCES

1 Clarke, J. S. and Lumley, R. R. Problems Associated With Windscreen


Wiping, Paper 600134, SAE Summer Meeting, Edgewater Beach Hotel,
Chicago, Ill., June 5-10, 1960.
2 Dawley, M. W., Aerodynamic Effects on Automotive Components. Paper
650134, International Automotive Engineering Congress, Detroit, Michigan,
January 11-15, 1965. Society of Automotive Engineers Inc.
3 Sanon, A., Jallet, S., Acoustic sources localization: Application to wiper
aerodynamic noises. SAE Paper 2003-01-1700. Noise and Vibration
Conference and Exhibition, Traverse City, Michigan, USA, 05/05/2003 -
05/08/2003. Society of Automotive Engineers, Inc., Warrendale, Pennsylvania,
USA.
4 Fischer and Zuccini, Experimental and Numerical Investigation of Flow
Induced Sound for Impinging Jets in Simplified 3D Geometries. Acta
Acoustica (in press).
5 Jallet, S., Devos, S., Maubray, D., Sortais, J-L, Marmonier, F., and Dreher, T.,
Numerical simulation of wiper system aerodynamic behaviour. SAE Paper
2001-01-0036. SAE 2001 World Congress, Detroit, Michigan, USA,
03/05/2001 - 03/08/2001. Society of Automotive Engineers, Inc., Warrendale,
Pennsylvania, USA .
6 Billot, P., Jallet, S., and Marmonier, F. Simulation of aerodynamic uplift
consequences on pressure repartition~Application on an innovative wiper
blade design. SAE Paper 2001-01-0043. SAE 2001 World Congress, Detroit,
Michigan, USA, 03/05/2001 - 03/08/2001. Society of Automotive Engineers,
Inc., Warrendale, Pennsylvania, USA .
7 Barth, R. in Aerodynamics of Road Vehicles, Hucho, W-H (ed.), 4th Edition, pp
326-327. Society of Automotive Engineers Inc., 1998.
8 Strumolo, G.S., Journal of Engineering Mathematics, 43 (2-4), August 2002,
pp.173-187.
9 Newnham, P., Passmore, M., Howell, J., Baxendale, A. On the Optimisation
of Road Vehicle Leading Edge Radius in Varying Levels of Freestream
Turbulence. SAE International Congress and Exposition, Detroit, Michigan,
USA. 2006-01-1029, SP-1991.
10 Lietz, R., Pien, W., Remondi, S., A CFD Validation Study For Automotive
Aerodynamics. SAE Paper 2000-01-0129, ISSN 0148-7191, Society of
Automotive Engineers, Inc. Warrendale PA. 2000.
11 Amodeo, J The Development of CFD as a Primary Design Tool at Jaguar
Cars. Fifth MIRA International Conference On Vehicle Aerodynamics, 13-14
October 2004, Heritage Motor Centre, Gaydon, UK
12 Gaylard, A. P., Simulation of A-Pillar/Side Glass Flows for Bluff SUV
Geometries. Fifth MIRA International Conference On Vehicle Aerodynamics,
13-14 October 2004, Heritage Motor Centre, Warwick, UK.
13 Gaylard A. P., CFD Simulation of Side Glass Surface Noise Spectra For A
Bluff SUV . SAE International Congress and Exposition, Detroit, Michigan,
USA. 2006-01-0137 SP-1991.
14 Jeong, J. and Hussain, F. On the identification of a vortex. Journal of Fluid
Mechanics, (1995), 285 , pp. 69-94.

S-ar putea să vă placă și