Professional Documents
Culture Documents
Simulation of Rear Glass and Body Side Vehicle Soiling by Road Sprays
Simulation of Rear Glass and Body Side Vehicle Soiling by Road Sprays
Simulation of Rear Glass and Body Side Vehicle Soiling by Road Sprays
Bradley Duncan
Exa Corporation
PREVIOUS RELEVANT WORK door mirror glass. Karbon and Longman [10] demonstrated the
viability of combining Lagrangian particle representation of
Road sprays have been a long-standing concern for the
airborne water droplets with a thin (surface) film model to
commercial vehicle industry. With large numbers of wheels
account for water movement on vehicle surfaces, within a
and axles, tractor-trailer combinations can generate substantial
RANS simulation. An interesting feature of this work was the
road-spray, which then becomes a hazard for other road users.
use of a periodic surface water source on the windscreen to
A useful summary of work in this area has been provided by
model the effect of windscreen wipers driving surface water
Gotz and Mayr [1]. Nevertheless, a review of specific
over the a-pillar.
contributions of workers in this field that includes recent
developments is apposite.
Waki et al [11] investigated door mirror design, assessing
mirror glass contamination. They used a full-scale track test
Clarke [2] explored the reduction of splash and spray from
with photosensitive paper attached to the mirror glass and
commercial vehicles; building on work done during the 1970s.
spraying “developing liquid” into the airflow. In a second
This work looked at road-spray generation mechanisms, the
phase of the work, a water tunnel was used to perform flow
conversion of splash to spray, wheel set interaction, fender and
visualisation experiments. A simple assessment of aeroacoustic
wheel well treatments. The investigation was undertaken using
impact was also made. Bannister [12] provided a seminal
actual vehicles running through water troughs. Olson and Fry
analysis of the relationship between aerodynamic drag, door
[3] reported an investigation of the effect of aerodynamic aids
mirror design, and side glass soiling. This work was based on
on splash and spray generation using water troughs, combined
track testing and the use of a rain grid in a full-scale
with video recording and a laser measurement system.
aerodynamic wind tunnel. Borg and Vevang [13] investigated
droplet size distribution within wakes (foreign-generated
Lajos, Preszler and Finta [4] looked at mud and dust deposits on
contamination), enabling the development of a wind tunnel
bus bodies via scale-model wind tunnel testing. They
based foreign-soiling simulation using a bespoke turbulence
demonstrated the effect of a moving belt road simulation, finding
generator and rain grid. They demonstrated the efficacy of this
that the, “rear wall deposits depend mainly on the structure of the
system in assessing front side-glass contamination. The work
wake and the characteristics of the under body flow.” Fujimoto
also included an assessment of body side self-soiling,
et al [5] combined full scale trials and reduced-scale wind tunnel
introducing water with a UV tracer onto a dynamometer roller.
testing to elucidate the mechanisms of mud accretion on the rear
Bouchet et al [14] used specially designed rain collectors in
surfaces of a “van type truck” and the potential for drag-reducing
the Jules Verne climatic wind tunnel to determine the intensity
devices to mitigate both this and road spray.
spatial gradient in the wake of a vehicle. Their measurements
of the droplet size distribution generated by the rear wheels of
Goetz and Schoch [6] reported on a laser-illumination method
their test vehicle showed more negative skew and positive
for measuring the spray generated by both commercial vehicles
kurtosis (leptokurtic) compared to natural raindrop size
and cars. They also demonstrated a number of modifications to
distributions. Both the negative skew and kurtosis tended to
the wheel arch system that reduced road spray (wheel arch
decrease (slightly) with increasing vehicle speed, the mode
liner grooves and extended mud flaps).
diameter decreasing and thus becoming more centred in the
range. The mode droplet diameter for a vehicle speed of
Kim [7] investigated the ability of a truck-cab corner turning-
80 km/h was 0.2 mm.
vane to reduce cab side dirt contamination. The work used
both Reynolds Averaged Navier Stokes (RANS) CFD
Foucart and Blain [15] reported a multi-stage computational
simulation to study the aerodynamic characteristics, along
method for simulating rain collection on the windscreen and
with full-scale road tests.
subsequent movement of a water film over the a-post onto the
side glass. The initial computation of the flow field used a
The literature for cars follows a similar pattern, starting with
Lattice Boltzmann solver. This was used to provide the
full-scale track testing, moving into wind tunnel and ultimately
boundary conditions for a local unsteady RANS model
computational simulations. Early work by Dawley [8] focussed
focussing on the windscreen and front side glass. Rain droplets
on the use of a roof trailing edge spoiler to keep a station
were represented via Lagrangian particle tracking, with surface
wagon backlight clear of contamination, using a test track
water simulated by an Eulerian model.
based experiment. In a similar vein, Costelli [9] published on
the aerodynamic development of the Fiat UNO; part of this
Experimental validation was presented, using wind tunnel
work demonstrated the effectiveness of a variable-attitude
experiments with a UV tracer in the water. Borg and Vevang
spoiler for reducing soiling on the backlight. This work appears
[16] examined the use of Lagrangian particle tracking methods
to have used wind tunnel based test methods.
with and without a wall film model to examine front side-glass
water contamination, within the context of a RANS simulation.
Currently, the major theme in the published literature is the
Kruse and Chen [17] reported the development of an Euler-
management of water and soiling on the front side glass and
Lagrange method, to address the same issue. This work also
included experimental verification using a water spray with a a pre-launch field evaluation unit (FEU); numerical simulation
UV tracer in the FKFS thermal wind tunnel. Although the was based on a production-representative CAD model.
paper concentrated on front side glass, it is clear that the
experimental method was more widely applicable. The typical concern with road sprays for this type of vehicle is
soiling of the rear surfaces. However, body side soiling is still
Road sprays, the topic of this paper, have been explored using a relevant, particularly with reference to door handle cleanliness.
combination of unsteady Lattice Boltzmann flow simulation
and Lagrangian particle tracking, starting with Roettger et al
[18, 19]. Kuthada and Cyr [20] provided a comprehensive
SALOON
overview of vehicle soiling issues and explored the use of this The 2.7L V6 diesel variant of the 2008MY Jaguar XF was chosen
combination of modelling techniques for road spray driven off for the saloon car geometry. Its aerodynamics has been discussed
the vehicle’s tyres. In addition, this work provides an in detail by Gaylard [29]. This full-sized saloon aims to fuse the
experimental validation for the road spray generated by a wheel style and performance of a sports car with the space and
in isolation. Finally, Borg [21] reported the use of RANS flow refinement of a luxury saloon. The sports car heritage is seen in
simulation plus Lagrangian particle tracking to analyze body the generally aggressive stance; epitomised by large ‘haunches’
side and chassis soiling from vehicle generated road sprays. over the rear wheels along with an upright and aggressive front
grille. Alongside this it incorporates a number of (potentially)
Other surface contamination issues reported in the literature aerodynamically beneficial features: a well raked screen,
include the nature of the rainwater film that forms over the cambered roofline, glass house tumblehome, reasonable front
bonnet of a road vehicle [22] (building on the earlier tyre-face coverage, squared-sills and a ‘fastback’ style rear with
development of a bespoke test facility [23]) and the deposition a sharp boot deck trailing edge condition.
of brake dust on the front wheels of a saloon car [24].
The XF is 4.961m long, with a 2.909m wheelbase. Its height is
Experimental methods continue to be developed; Aguinaga and 1.460m; vehicle width (mirrors deployed) is 1.877m with front
Bouchet [25] illustrate the ‘state of the art’ in test techniques and rear track of 1.559m and 1.571m, respectively. The
for side-glass contamination measurement. Bouchet [26] has projected frontal area is 2.33m2. This is combined with a drag
explored ‘rain’ intensities due to splash and spray from a coefficient that is relatively representative of contemporary
vehicle running on a wet road. Future developments in the full-sized saloon vehicles, of 0.29.
numerical simulation of soiling deposition may well include the
use of unsteady Large Eddy Simulation (LES) [27] and The experimental work was completed using a full-scale
the adoption of open-source CFD code as the simulation aerodynamic test property (‘aerobuck’). This married a new
framework [28]. fibreglass upper body with an updated donor ‘go-cart’,
providing a fully representative test property with complete
This work explores the extension of a combination of under body and engine bay. The numerical simulations use an
Lagrangian particle tracking and flow field simulation using a equivalent CAD model.
transient Lattice Boltzmann solver, where additional particle
sources are added to account for splashing and dripping from Generally, for a vehicle of this type, the principal road-spray
the wheel house and rear under body. concern is body side soiling as it relates to door handle
cleanliness and the soiling of customers’ clothes as they enter
and egress from the vehicle. For this reason, the experimental
VEHICLES investigation did not consider soiling of the rear surfaces.
EXPERIMENTAL
Two experimental techniques are reported here. For the SUV a
test track experiment was undertaken. A vehicle was driven at
80 km/h over a graded and wetted gravel track at Volvo’s
Hällered complex, to obtain a rear soiling pattern.
SUV Figure 4. (a) Deposition Intensity Analysis and (b) raw data
The results of the SUV road-spray experiments are shown in picture for SUV tailgate soiling.
Figures 4 and 5. The wind-tunnel experiments (Figure 4) show
high levels of soiling over the rear glass, with a central soiling
band narrowing onto the rear bumper. The rear glass has two
SUV
First, the aerodynamic simulation results for the SUV were
used to compute the body side soiling from water spray from
the front left wheel, including wheel house and rocker panel.
The effects of splashing and dripping were included as shown
in Figures 1, 2, 3. Figure 7 shows a visualization of one frame
of the transient turbulent flow using isosurfaces of negative λ2
values [31] for visualizing the vortex structures. This indicates
the most energetic vortex structures, representing the fine-
Figure 5. (a) Before and (b) after tailgate soiling test at scale turbulent flow. Vortices are created in the regions of
Hällered. strong shear in separated flow regions, and then these vortices
are carried downstream into the vehicle wake. The trajectory
of the turbulent wheel wake is also evident. The wheel wake
contains small-scale vortex structures which, due to strong
SALOON spatial gradients in velocity, are capable of pushing small
Only front wheel-driven soiling was experimentally simulated particles travelling near the vehicle surface toward the surface
for the Jaguar XF saloon. The results are shown in Figure 6 where they can impact the surface and deposit dirt and other
(below). This shows water being driven out of the wheel house materials.
from a high angle onto the downstream body side. The highest
levels of soiling are seen immediately behind the front wheel,
decreasing with both height and distance downstream. There is
some door handle contamination.
In Figure 8, the trajectories of particles emitted from the tyre Then, in Figures 10, 11, 12, 13, the final analysis of material
are shown, including the effect of splashing of particles in the deposition pattern is shown. The particle trajectories are
wheel house. The trajectories follow and fill up the turbulent simulated over 120 frames of data, from (time) t=0.53 to
wake behind the wheel. The trajectory of particles closely t=0.99 seconds (t=0.616 to t=0.99 seconds shown). The
follows the advection of turbulent flow structures in the particle accumulation pattern shows soiling of the body side
wheel wake. surface up to about 50% of the height of the door. The sequence
of image shows the accumulation of the soiling pattern using
successive frames of transient flow data, using the frozen-flow
approximation for advection of droplets in the flow.
The trajectories of low-velocity particles that are dripped from (a) Frame 30 (t=0.616)
either the wheel house or rocker panel are shown in Figure 9.
These behave similarly to the particles ejected from the wheel
and fill in the wake of the wheel and will add to deposition of
material on the body side.
Figure 16. Rear face soiling pattern for spray from rear-left Figure 17. Rear face soiling pattern for spray from rear-left
wheel including the effect of droplets splashing in wheel wheel including splashing droplets in wheel house and
house, but no dripping particles from the wheel house and dripping droplets from wheel house and rear under body.
rear under body. (a) Frame 50 (t=0.699 sec) and (b) Frame The dripping particles create more coverage of the lower
120 (t=0.990 sec). portion of the back face. (a) Frame 50 (t=0.699 sec) and (b)
Frame 120 (t=0.990 sec).
SALOON
Similar analysis of body side soiling from the front wheel
spray was performed for the saloon. The wheel wake is shown
in Figure 18 first using particle traces from the wheel spray
sources on the tyre. The wheel wake fills the wheel house and
covers the body side. Additional particle traces from dripping
particles are shown in Figures 19(a) and 19(b); they do not
add significantly to the region of coverage of particles. The
accumulation of material on the surface of the body side is
shown in Figure 20. The deposition of material occurs on the
body side on the front and rear side doors in the region of the
wheel wake.
Figure 20. Body side soiling pattern for spray from front-
left wheel including splashing droplets in wheel house and
dripping droplets from wheel house and rocker panel. (a)
Figure 18. Particle traces from wheel spray for one frame Frame 50 (t=0.699 sec) and (b) Frame 120 (t=0.990 sec)
of transient data, showing the path of motion of droplets
emitted from the wheel, including splashing of those
droplets in the wheel house (Frame 1, t=0.535 sec) DISCUSSION
Comparing the road spray simulations to the experiments
provides some useful insights into the mechanism of soiling.
For the back face of the SUV, in Figure 17(b), the largest
concentration of spray forms on the rear glass and the centre of
the lower tailgate door. This is consistent with the experimental
photographs in Figures 4 and 5. From the analysis of particle
traces in Figures 14 and 15, the mechanism of soiling of the
back face seems to be indirectly related to the spinning rear
wheel. The wheel spray particles are for the most part carried
backward in the wheel wake. Then, some spray particles from
the rear edge of the wake slow down and reverse direction and
then are carried to the back face through large-scale
recirculation. However, Figure 15(b) shows that the dominant
mechanism for soiling of the rear face is from any droplets
exiting the rear under body under the rear bumper, rather than
in the wheel wake. These particles can come from road spray
from either front or rear tyres that impinges on, or wets, the
under body components. Dripping and splashing from these
components produces the source of droplets that are carried by
the more dominant recirculation in the wake that exists
between the rear wheel wakes.
Figure 19. Particle traces from dripping particles in (a) the
wheel house and (b) rocker panel.
For both the SUV and saloon cases the mechanism of body 2. Clarke, R.M., “Heavy Truck Splash and Spray
side soiling from the front wheel is more direct. First, the Suppression: Near and Long Term Solutions,” SAE Technical
source of particles is the high-speed spray tangential to the Paper 831178, 1983, doi: 10.4271/831178.
tyre, but considering for that spray to splash off of the surfaces
3. Olson, M.E. and Fry, P.R., “Highway Tractor-Trailer
of the wheel house so that a large mist of particles is released
Splash and Spray Reduction Through Aerodynamic Design,”
in the entire wheel wake. The downstream velocity in wheel
SAE Technical Paper, 1988, doi: 10.4271/881875.
wake carries both the wake flow structures and the particles.
The particles entrained into the turbulent flow environment 4. Lajos, T.; Preszler, L.; Finta, L., “Wind tunnel
of the wheel wake are subject to strong spatial gradients in investigation of mud deposits on the body of a bus,”
velocity as can be seen in the vortex visualization of Figure 7. International Journal of Vehicle Design 5(6):693-703, 1984.
The spatial gradients include some contribution in the span-
5. Fujimoto, T., Miyake, N., Watanabe, Y., and Takeyama, T.,
wise direction and push the same particles toward the surface.
“Suppression of Mud Adhesion to the Rear Surface of a
Figures 8 and 9 show that this direct wheel spray with splashing
Van-Type Truck,” SAE Technical Paper 920203, 1992,
is the dominant effect, and additional particles dripping from
doi: 10.4271/920203.
wetted surfaces do not provide any additional coverage of the
body surface. 6. Goetz, H., and Schoch, R., “Reducing Splash and Spray of
Trucks and Passenger Cars,” SAE Technical Paper 950631,
For the saloon case, the coverage pattern from the wheel spray 1995, doi: 10.4271/950631.
in Figure 20(b) is indicative of the experimental measurement 7. Kim, M.-H., “Numerical Simulation on the Aerodynamic
shown in Figure 6, which shows the largest concentration on Characteristics Around Corner Vane of a Heavy-Duty Truck,”
the side door in the near wake of the front wheel, up to the SAE Technical Paper 2000-01-3499, 2000, doi:10.4271/2000-
location of the top of the wheel arch. The particle traces in 01-3499.
Figure 18 show that the source of soiling in this region is spray
from the top of the wheel, splashing in the wheel arch, and then 8. Dawley, M.W., “Aerodynamic Effects on Automotive
travelling downstream in the wheel wake. Components,” SAE Technical Paper 650134, 1965, doi:
10.4271/650134.
CONCLUSIONS 9. Costelli, A.F., “Aerodynamic Characteristics of the
Fiat UNO Car,” SAE Technical Paper 840297, 1984, doi:
In summary, the present results indicate that the simulation
10.4271/840297.
method appears to reproduce results indicative of the physical
test. The use of a Lagrangian particle tracking method with an 10. Karbon, K.J. and Longman, S.E., “Automobile
unsteady flow field, plus a representation of splashing and Exterior Water Flow Analysis Using CFD and Wind Tunnel
dripping effects, has been shown to produce results that are Visualization,” SAE Technical Paper 980035, 1988, doi:
consistent with available experimental data. 10.4271/980035.
11. Waki, T., Tokumaru, K., Nouzawa, T., and Hata, N.,
Further, the results provide indirect evidence that direct spray
[Flow Visualization around Door Mirror Considering the
from the tyre on the body surfaces is not the cause of vehicle
Adhesion of Raindrops and Wind Noise,] presented at 1996
soiling, but rather that the spray will splash in the wheel house
JSAE Autumn Convention, 1997, doi:10.1016/S0389-
to form a mist of droplets that occupy the wheel wakes and
4304(97)84964-8.
lead to deposition of material on the surface.
12. Bannister, M., “Drag and Dirt Deposition Mechanisms
In the case of rear face soiling, dripping from other wetted of External Rear View Mirrors and Techniques Used for
surfaces may be more significant than the spray in the wheel Optimisation,” SAE Technical Paper 2000-01-0486, 2000,
wake due the pattern of recirculating flow in the lower wake of doi:10.4271/2000-01-0486.
the vehicle. Also, the particle traces and turbulent flow
13. Borg, A., and Vevang, R., “On the Development
visualizations are helpful tools to understand the mechanisms
of a Wind Tunnel method for the Prediction of Exterior
and sources of soiling on the vehicle body.
Contamination,” presented at 5th MIRA International
Conference on Vehicle Aerodynamics, UK,
Finally, it must be emphasised that this work does not address
October 13-14, 2004.
the potential influence of droplet break-up or coalescence [12].
14. Bouchet, J. P., Delpech, P., and Palier, P., “Wind
tunnel simulation of road vehicle in driving rain of variable
REFERENCES intensity,” presented at 5th MIRA International Conference
1. Hucho, W.-H., “Aerodynamics of Road Vehicles,” 4th ed., on Vehicle Aerodynamics, UK, October 13-14, 2004.
SAE, Warrendale, PA ISBN: 978-0-7680-0029-0, 1998.
15. Foucart, H. and Blain, E., “Water-flow Simulation 25. Aguinaga, S., and Bouchet, J.-P., “Quantitative
on Vehicle Panels by Taking into Account the Calculated Assessment by UV Fluorescence of Rain Water Flow on
Aerodynamic Field,” SAE Technical Paper 2005-01-3572, Vehicle Body in the Jules Verne Climatic Wind Tunnel” In:
2005, doi:10.4271/2005-01-3572. Wiedemann, J., (editor) “Progress in Vehicle Aerodynamics
and Thermal Management, VII,” Expert-Verlag, ISBN 978-3-
16. Borg, A., and Vevang, R., “On The Prediction of
8169-2944-4:127-142, 2009.
Exterior Contamination with Numerical Simulations (Simple
Lagrangian Particle Tracking methods with and without 26. Bouchet, J.-P., “Equivalent Rain Intensities Due to
Wall Film Model),” presented at 6th MIRA International Splash and Spray in the Back of a Vehicle Running on Wet
Conference on Vehicle Aerodynamics, UK, October 26-27, Road,” presented at 8th MIRA International Conference on
2006. Vehicle Aerodynamics, UK, October 13-14, 2010.
17. Kruse, N. and Chen, K.-H., “Exterior Water Management 27. Verzicco, R.; Fatica, M.; Iaccarino, G.; Khalighi, B.;
Using a Custom Euler-Lagrange Simulation Approach,” Moin, P., “Large Eddy Simulation of a Road Vehicle with
SAE Transactions Journal of Passenger Cars - Mechanical Drag-Reduction Devices,” AIAA Journal 40(12):2447-2455,
Systems 116, 2008, doi:10.4271/2007-01-0101. 2002, doi: 10.2514/2.1613.
18. Roettger, S., Schulz, S., Bartelheimer, W., and 28. Dietz, A., and Indinger, T. “Numerical Simulation with
Ertl, T. “Flow Visualization on Hierarchical Cartesian Particle Transport to compute self- and third party Soiling of
Grids,” In: Breuer, M., Durst, F., and Zenger, C., (editors) Motor Vehicles,” presented at 5th OpenFOAM® Workshop,
“High Performance Scientific and Engineering Computing: Sweden, June 20, 2010.
Proceedings of the 3rd International FORTWIHR Conference
29. Gaylard, A. P., “Aerodynamic Development of the New
on HPSEC (March 2001)”, Springer, ISBN 978-3-540-42946-
Jaguar XF,” presented at 7th MIRA International Conference
3:139-146, 2001.
on Vehicle Aerodynamics, UK, October 22-23, 2008.
19. Roettger, S., Schulz, S., Bartelheimer, W., and Ertl, T.,
30. Samples, M., Gaylard, A. P., and Windsor, S., “The
“Automotive Soiling Simulation Based On Massive Particle
Aerodynamic Development of the Range Rover Evoque,”
Tracing,” In: Proceedings of EG/IEEE TCVG Symposium
presented at 8th MIRA International Conference on Vehicle
on Visualization VisSym ‘01, http://www.vis.uni-stuttgart.
Aerodynamics, UK, October 13-14, 2010.
de/∼roettger/data/Papers/PARTICLES.PDF, 2001.
31. Jeong, J.; Hussain, F., “On the identification of a vortex.”
20. Kuthada, T., and Cyr, S., “Approaches to Vehicle
Journal of Fluid Mechanics 285:69-94, 1995,
Soiling,” In: Wiedemann, J, Hucho, W.-H., (editors)
doi: 10.1017/S0022112095000462.
“Progress in Vehicle Aerodynamics, IV, Numerical methods”,
Expert-Verlag, ISBN 3-8169-2623-1:111-123, 2006.
21. Borg, A., “Status and Trends in Large Scale CFD
CONTACT INFORMATION
Applications for Vehicle Design at Volvo Car Corporation,” Adrian Gaylard
presented at Workshop on State-of-the-art in Scientific and agaylar1@jaguarlandrover.com
Parallel Computing, Sweden, June 18-21, 2006.
22. Aroussi, A., Brewer, D., Lee, K.L., and AbdulNour, B.S., ACKNOWLEDGMENTS
“Analysis of the Progression of Rainwater Film Over the The authors would like to thank Jaguar Land Rover for
Bonnet of a Road Vehicle,” SAE Technical Paper 2005-01- permission to publish this paper. Thanks are also due to former
1513, 2005, doi: 10.4271/2005-01-1513. colleagues Joe Edge and Andy Sheppard for leading the XF
23. Abdul Ghani, S. A. A.; Aroussi, A.; Rice, E., “Simulation and Discovery experimental activities, respectively. Finally,
of road vehicle natural environment in a climatic wind they would also like to thank Mark Bannister for his helpful
tunnel,” Simulation Practice and Theory 8(6-7):359-375, suggestions.
SAE
2001, doi:10.1016/S0928-4869(00)00028-8.
24. Gaylard, A. P., Lynch, D., Amodeo, J., and
Amunugama, R., “The Simulation of Brake Dust
Deposition,” presented at 8th MIRA International Conference
on Vehicle Aerodynamics, UK, October 13-14, 2010.