the flow field of automobile add-ons — with particular reference to the vibration of external...

*Corresponding author.E-mail address: [email protected] (S. Watkins)

Journal of Wind Engineeringand Industrial Aerodynamics 83 (1999) 541}554

The #ow "eld of automobile add-ons * withparticular reference to the vibration of

external mirrors

Simon Watkins*, Greg OswaldDepartment of Mechanical and Manufacturing Engineering, RMIT University, Level 3, Building 251,

264 Plenty Road, Mill Park 3082, Australia

Abstract

The #ow"eld of externally mounted car mirrors is discussed and a wind-tunnel study of thelocal #ow"eld is presented. Mean velocities, pitch and yaw angles and turbulence intensities andspectra were measured using hot-wire anemometers. The #ow is complex and turbulent due toa vortex that is associated with the car A pillar and local turbulence intensities were found toreach 40%. The vehicle yaw angle was varied from !123 to #123 and was shown to havea strong in#uence on the #ow characteristics. Translational and rotational vibrations of themirror glass were also measured in the wind tunnel and on-road. At frequencies above about20 Hz there was very close agreement between the on-road and wind-tunnel vibration spectraindicating that the main forcing functions were aerodynamic rather than arising from mechan-ical sources such as road-induced vibrations or rotational imbalances. ( 1999 ElsevierScience Ltd. All rights reserved.

Keywords: Flow "eld; Automobile add-ons; Vibration; External mirrors

1. Introduction and objectives

Due to the e!orts of the passenger vehicle manufacturers, levels of noise, vibrationand harshness (NVH) have been reduced for car mechanical components (e.g. engines,drivetrains and exhausts) such that aerodynamically generated noise and vibration isnow signi"cant [1].

0167-6105/99/$ - see front matter ( 1999 Elsevier Science Ltd. All rights reserved.PII: S 0 1 6 7 - 6 1 0 5 ( 9 9 ) 0 0 1 0 0 - 2

Modern cars are characterised by streamlined forms and have large areas ofattached #ow (i.e. there are few #ow separations from the car surfaces). This is true forthe body shape until at least until aft of the C pillar, with the exception of the local#ow in the vicinity of the A pillar. However, various blu! shapes are added to thebody, such as external mirrors, radio aerials and roofracks. These blu! shapes exhibitaerodynamically generated noise and vibration, which raises the NVH level of cars.Measures to reduce NVH, such as extra sealing and double glazing are costly, hence itis desirable to minimise NVH at the source.

In addition to the annoyance caused, vibration can interfere with a components'primary role, such as blurring the images in rear view mirrors. In order to understandthe aerodynamically induced noise and vibration, an understanding of the local#ow"eld is useful. The objective of the work presented here was to document the local#ow"eld of external mirrors and to measure the vibration of the mirror glass in orderto understand the causes of image blurring.

2. The relative air6ow as experienced by a moving vehicle and the e4ect of atmosphericwinds

Under calm conditions and no tra$c, vehicles travel through still air, hence therelative air#ow they experience has no turbulence, is unyawed relative to the vehicleand has the same magnitude as the vehicle speed relative to the road. If an atmo-spheric wind is present, generally a yaw angle is created (thus the #ow is not alignedwith the centreline of the vehicle) and the air speed a vehicle experiences is not thesame as the road speed.

It is well known that the atmospheric wind is turbulent and its mean velocityincreases with height, thus the #ow"eld experienced by the moving vehicle is turbulentand changes slightly with height. Driving in average winds, the turbulence intensity(de"ned as the standard deviation of velocity #uctuations divided by the mean relativevelocity) is typically a few percent and has been measured by Watkins and Saunders[2]. The change in approach #ow up the height of the vehicle is relatively minor andreviews of mean and #uctuating wind conditions experienced by a moving vehiclehave been given by Cooper [3] and Watkins [4]. The temporal variation of theatmospheric wind and the mean spatial variations with height and terrain give rise toa range of `reala approach #ows which generally are not reproduced by car com-panies during most commercial wind-tunnel tests (e.g. measurement of drag coe$-cients or engine cooling) as their e!ects are often minor.

For the work present here a #at velocity pro"le was used in a full-size wind tunneland the freestream turbulence intensity was 2%. Yawing the vehicle in the range!153 to #153, reproduced the primary e!ect of atmospheric crosswinds.

3. The 6ow in the mirror region

In addition to the e!ects caused by atmospheric winds, the velocity "eld close to thevehicle is strongly in#uenced by the presence of the vehicle body. This in#uence

542 S. Watkins, G. Oswald / J. Wind Eng. Ind. Aerodyn. 83 (1999) 541}554

Fig. 1. Flow on the side window. Car at !15 yaw.

changes as the yaw angle changes. Generally, the #ow accelerates around the sides andtop of the vehicle, up to the point of maximum frontal area, then the #ow deceleratesas it closes around the vehicle. Hence the location of any aerodynamic add-ondetermines the local #ow speed and direction.

Local #ow"elds have been measured in the RMIT/Monash Wind Tunnel, whichhas been developed to enable the aerodynamics of full-size vehicles to be assessed,including measurement of wind noise. It is a 3

4open jet, closed return tunnel with

a maximum speed of approximately 180 km/h, an exit nozzle size of 10.55 m2 and hasvery low levels of background noise. The background noise level in the tunnel istypically greater than 10 dB(A) below the interior car noise, and is minor compared tothe car, mirror and aerial wind noise.

Most external mirrors are located in an area that is strongly in#uenced by a vortexthat originates from the junction of the car A pillar and bonnet. Thus the air#ow in thevicinity of an outside rear view mirror is complex and three dimensional. Originalwork in this area by Watanabe [5] showed the existence of a vortex originating fromthe A pillar of the vehicle. Further work by Haruna [6] documented the e!ect of 03and 103 yaw angle on the A-pillar vortex for a sports car and the subsequent increasein size and movement of the vortex.

To give a better understanding of the aerodynamic contribution to the mirror'snoise and vibration under a wider range of operating conditions, the air#ow in thevicinity of a production mirror was investigated for a typical passenger car. Flowvisualisation was performed using wool tufts and the car was yawed (via the wind-tunnel turntable) through $153 to reproduce the primary e!ects of atmosphericcrosswinds. The resulting #ow patterns can be seen in Figs. 1 and 2.

On the windward side the vortex was greatly diminished, and at #153 yaw angle (apositive yaw indicates that the vehicle side being considered is on the windward side)the #ow became more horizontal and considerably less turbulent (see Fig. 2). With the

S. Watkins, G. Oswald / J. Wind Eng. Ind. Aerodyn. 83 (1999) 541}554 543

Fig. 2. Flow on the side window. Car at #15 yaw.

Fig. 3. Longitudinal airspeed * Position 1.

side window in a leeward situation (negative yaw angles), the #ow close to the sideglass exhibited considerable upwash, until at !153, some of the wool tufts at the topof the side glass were at an angle of 303 to the horizontal or greater (see Fig. 1). The#ow was noticeably unsteady under this condition.


Fig. 4. Longitudinal airspeed * Position 2.

To quantify the #ow the local airspeed was measured with TSI two-componenthot-wire anemometers. The mirrors were removed to enable the probe to be placed atthe mirror location. The probe was attached to a traversing mechanism and movedvertically parallel to the door sail area (i.e., the mounting point for the mirror) ata "xed longitudinal position. Each run was repeated with the probe rotated axiallythrough 903 to enable the three-dimensional parameters to be measured. The datapresented here are for two vertical traverses originating at the position below theinnermost and outermost lower edge of the mirror glass and starting at the base of thesail area. The innermost starting position was denoted Position 1 and the outermostposition, Position 2 with 100 mm between them. The probe was traversed 200 mm upthe door over a time period of approximately 3 min. The typical measurementaccuracy for the hot-wire measurements was 1 m/s for mean velocity and 23 for #owangles.

The local #ow velocity and yaw and pitch angles show marked changes with thecar yaw angle, as shown in Figs. 3}8. The e!ects were consistent with the probetraversing increasingly closer to the A-pillar vortex for the unyawed and positivelyyawed runs and possibly into the vortex for the negatively yawed runs. The resultsin Fig. 9 show that as the car was yawed from 03 towards the windward direction(positive yaw) the turbulence level decreased rapidly. However, yawing towardsthe leeward direction increased the turbulence level to a maximum of 40% at !123with a peak of 20% present for 03. The turbulence intensity 100 mm further intothe freestream (i.e., at Position 2) exhibits characteristics that are close to freestream(Fig. 10).


Fig. 5. Local yaw angles * Position 1.

Fig. 6. Local #ow yaw angles * Position 2.


Fig. 7. Pitch angles * Position 1.

Fig. 8. Pitch angles * Position 2.

The spectral characteristics in Figs. 11 and 12 indicate that most of the tur-bulent energy is concentrated close to the glass (i.e., Position 1) and peaks at about60 Hz.


Fig. 9. Turbulence intensity * Position 1.

Fig. 10. Turbulence Intensity * Position 2.


Fig. 11. Velocity spectra * Position 1.

Fig. 12. Velocity spectra * Position 2.

4. Measurements of mirror vibration

Three single degree-of-freedom microchip accelerometers were "xed to the mirrorglass to enable the rotational vibration of the glass to be found. The signals wererecorded on a digital instrumentation tape recorder and data processing was per-formed on a personal computer. The accelerometer signals were double integrated to


Fig. 13. On-road de#ection at 80 km/h.

Fig. 14. On-road de#ection at 100 km/h.

give displacements and then transformed into a polar coordinate system. The veracityof the system was independently checked via a laser-based system and the accuracywas found to be $0.2 arcmin. Translational vibration was measured independentlyvia a triaxial accelerometer on the mirror glass. Testing was performed in theRMIT/Monash Tunnel and also on-road in order to examine how much of therotational vibration (causing image blurring) arose from aerodynamic inputs com-pared to inputs from other sources (e.g. road roughness, drivetrain or wheel imbalan-ces). Wind-tunnel testing permitted reproduction of the aerodynamic excitations ofthe mirror and also the vehicle, without the road-induced vibrations, whereas on-roadtesting had both.


Fig. 15. On-road rotational spectra.

Wind-tunnel tests were carried out at 80}130 km/h and the road tests were under-taken on smooth roads at similar road speeds in two directions. The atmosphericwind generated a yaw angle of $6.53 for the data presented here and the orientationof the wind was such that the road speed was close to the relative windspeed (i.e., thespeed replicated in the wind tunnel was close to the relative windspeed during theroad tests).

Typical polar plots of the rotation of the mirror can be seen in Figs. 13 and 14.Spectra of rotational vibration obtained both on-road and in the wind tunnel aregiven for 130 km/h in Figs. 15 and 16 and translational spectra are given in Fig. 17.

5. Discussion

The measured responses of vibration on the mirror glass are the result of the inputsfrom mechanical and aerodynamic sources as well as the structural dynamics of themirror and supporting structure. A range of possible mechanical sources exists arisingfrom road roughness and imbalances from the many rotating components on a motorvehicle. Several possible aerodynamic sources also exist; vortex shedding from the(blu! ) mirror casing; turbulent bu!eting on the casing arising from unsteadiness inthe approach #ow and also #uctuations in the pressure di!erence between either sidesof the mirror glass. The structural dynamics of the assembly is complex and is notconsidered here. However some comments can be made regarding the nature andcause of glass vibration.

The polar plots show the existence of a preferred axis of vibration along the linefrom 120 to 3003. This is due to the method of mounting (and electrically adjusting)the glass, further details can be found in Oswald [7]. The two areas of large de#ection


Fig. 16. Wind-tunnel rotational spectra.

Fig. 17. Comparison of translational spectra obtained on-road and in the wind tunnel.

in Fig. 13 are thought to come from a bump in the road or an uncharacteristicallylarge wind gust. Wind-tunnel polar plots (not presented here) were similar to thoseobtained on-road, with the exception of the two areas of large de#ections. Thesimilarity between rotational spectra obtained under a range of on-road conditionsand those obtained in the wind tunnel clearly indicates the considerable in#uence of


aerodynamically induced vibration, especially at frequencies of 20 Hz and above. Thisis further veri"ed by the similarity of translational spectra in Fig. 11 where extremelyclose agreement can be seen between the on-road and tunnel data sets, and also fromresults from a series of commercial tests performed on-road where aerodynamicallyshielding the mirror greatly reduced the vibration levels. The peaks in the spectra at2}5 Hz that were evident in the wind-tunnel and road data are considered to arisefrom solid body roll of the car induced by the wind loading, as the lateral naturalfrequency of the vehicle on its suspension was 2.9 Hz [8].

6. Conclusions and further work

The mean and time-varying characteristics of the #ow change rapidly over thelocation of the mirror and these are strongly in#uenced by the vehicle yaw angle.Local velocities in this region can be upto 60% above, or 40% below the freestreamvelocity and local turbulence intensities can be as high as 40%, even with the vehicle ina relatively smooth approach #ow.

The vibrations of the mirror measured in the tunnel and on road at frequenciesabove about 20 Hz are extremely similar indicating that aerodynamic forcing is thecause, rather than by inputs from mechanical sources, such as road roughness orrotational imbalances. It is considered that turbulent bu!eting is the major cause ofthe vibration, rather than vortex shedding, but further work is needed to con"rm this.

Acknowledgements

The Authors wish to thank the Mechanical Engineering Departments of RMIT andMonash Universities for the provision of testing facilities and equipment; Ford andBritax-Rainsford International for the provision of vehicles and wing mirrors andPeter Mousley for his assistance in the production of this paper. The "nancial supportof the Australian Research Council is greatly acknowledged.

References

[1] A.R. George, J. Callister, Aerodynamic noise of ground vehicles, Society of Automotive Engineers(SAE) America, Technical Paper Series SAE 911027.

[2] S. Watkins, J.W. Saunders, Turbulence experienced by road vehicles under normal driving conditions.Society of Automotive Engineers International Congress, Detroit, Feb 27}Mar 2, 1995 SAE 950997,also in Investigations into Vehicle Aerodynamics, SAE Special Publication 1078.

[3] R.K. Cooper, The wind-tunnel simulation of surface vehicles, J. Wind Eng. Ind. Aerodyn. 17(1984) 167}198.

[4] S. Watkins, J.W. Saunders, A review of the wind conditions experienced by a moving vehicle.International Congress of the Society of Automotive Engineers America, Detroit, February 1998.Developments in Vehicle Aerodynamics, ISBN 0-7680-0138-2 (also a chapter in SP 1318).

[5] M. Watanabe, M. Harita, E. Hayashi, The e!ects of body shapes on wind noise, SAE 780266.


[6] S. Haruna, T. Nouzawa, I. Kamimoto, An experimental analysis and estimation of aerodynamic noiseusing a production vehicle, SAE 900316.

[7] G. Oswald, Aerodynamically induced noise and vibration of automotive external rear view mirrors,Ph.D. Thesis, Mechanical and Manufacturing Engineering, RMIT University, Melbourne, Australia,1999.

[8] T.M. Nguyen, J.W. Saunders, S. Watkins, The sideways dynamic force on passenger cars in turbulentwinds, SAE 970405, Topics in Vehicle Aerodynamics, ISSN0148-7191 (also a chapter in SP-1232).


the flow field of automobile add-ons — with particular reference to the vibration of external...

Documents