Download as pdf or txt
Download as pdf or txt
You are on page 1of 17

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/286001833

Real World Drag Coefficient – Is It Wind Averaged


Drag?

Conference Paper · October 2014


DOI: 10.1533/9780081002452.1.3

CITATIONS READS

0 378

1 author:

Steve Windsor
Jaguar Land Rover Limited
5 PUBLICATIONS 17 CITATIONS

SEE PROFILE

All content following this page was uploaded by Steve Windsor on 04 January 2016.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Real World Drag Coefficient – Is It Wind Averaged Drag?
STEVE WINDSOR
Aerodynamics
Jaguar Land Rover

ABSTRACT

Drag reduction of road vehicles continues to be the holy grail of aerodynamicists with
renewed importance of aerodynamics as OEMs strive to reduce CO2 and improve fuel
economy. New legislation such as Worldwide harmonized Light vehicles Test Procedure
(WLTP) is increasing the emphasis on real world boundary conditions (with moving ground
wind tunnel testing) and vehicle configurations, in an effort to obtain more realistic fuel
consumption figures. However, WLTP in common with all other drive cycles used in the
automotive sector is based on zero-yaw aerodynamic measurements. This implicitly
assumes conditions vehicles rarely see: the onset wind vector is at zero yaw, the
atmosphere is still and undisturbed by other road users.

This paper will show that many modern cars (particularly low drag saloon cars) have
noticeable drag coefficient (CD ) minima at zero yaw, with some having up to 10% increase
in CD for relatively small yaw angle changes around zero. The concept of wind averaged
drag ( ) will be discussed as a means of assessing real world aerodynamic performance.
Three methods of calculating CDW are investigated and the MIRA method is preferred for
application to passenger cars. This demonstrates that including yaw angle effects via 
provides a significantly different perspective on the aerodynamic contribution to fuel
economy.

1. INTRODUCTION

Modern wind tunnels, with their moving ground and quasi turbulence generators, are being
designed and built to achieve an experimental set up that is getting closer and closer to the
real world. But the wind tunnel still only partially simulates on road conditions and it is very
difficult to quantify each of the known errors that exist between the wind tunnel and the real
world as in Figure 1. Generally, an automotive wind tunnel provides an environment with
low levels of turbulence, a floor boundary layer approaching 0 mm and most often used at
zero yaw.

However, road vehicles spend very little time being driven in conditions where the ambient
wind is either stationary and/or at 0° yaw to the vehicle’s direction and in low levels of
turbulence, yet the de facto drag coefficient (CD) is measured (and quoted) assuming the
ambient wind is 0 kph and/or at 0° to the vehicle’s direction. Measuring a vehicle’s CD with
this assumption is both repeatable and quick to compute in both the wind tunnel and CFD
codes but the aerodynamicist knows that in the “real world” ambient conditions and vehicle
set-up must, or at least should, be taken into account when calculating the vehicle’s true
CD.
Side wind
Rain
Sun

Dirt, spray

Windage

Figure 1 A vehicle in its real environment (based on Hucho [1] )

Automotive wind tunnels were designed to be able to yaw the vehicle to simulate
crosswinds but even now, nearly 60 years later, it is uncommon for any data of CD at yaw to
be published or quoted. Yet it is the change in CD with yaw angle that is of particular
interest in understanding its effect on real world fuel economy as well as stability etc. The
concept of wind averaged drag takes a range of CD at yaw and calculates the vehicle’s drag
coefficient in the real world environment; essentially it “characterizes” a vehicle’s
aerodynamic drag behaviour.

2. NOTATION

The relative velocity and wind direction of a vehicle travelling at velocity VV with the ambient
wind velocity VW is shown in Figure 2 below. The resultant velocity is termed VRES at a yaw
angle of Ψ.

Figure 2 Relative Wind Velocity and Direction

VV = vehicle velocity (kph)


VW = wind velocity (kph)
VRES = relative wind velocity (kph)
Ѳ = wind direction relative to the vehicle
Ψ = relative yaw angle
CDΨ = drag coefficient at yaw angle Ψ
CDW = wind averaged drag coefficient for wind velocity VW
3. WIND TUNNEL TEST RESULTS

A range of SUV, saloon and sports cars (as shown in Table 1) were tested in the MIRA Full
Scale Wind Tunnel (FSWT) to measure their aerodynamic characteristics. All the vehicles
were tested at standard loading conditions, and measurements were taken over the yaw
range ±5° in 1° increments and ±30° in 5° increments. The test results were left unedited
and so positive and negative yaw data may not be symmetrical.

SUV and 2 Box Vehicles


Audi Q3 SUV Saloons and Sports Cars
Audi Q5 SUV Audi A3 5-Door Wagon
Audi Q7 SUV Audi A5 Sportback Saloon
BMW X1 SAV Audi A6 Saloon
BMW X3 SUV Audi A8 Saloon
BMW X5 SUV BMW 320d Eff Dyn Saloon
BMW X6 SAV BMW 535 Saloon
Ford Grand C-Max (for shape) MPV BMW 740 Saloon
Ford S-Max MPV Ford Mondeo 4-Door Saloon
Honda CR-V SAV Jaguar F-Type Coupe Sports
Hyundai iX35 SUV Jaguar F-Type Convertible Sports
Jeep Compass SUV Jaguar XF 10MY Saloon
Kia Sorrento SUV Jaguar XF Sportbrake Wagon
Land Rover Discovery 3 SUV Jaguar XJ 10MY Saloon
Land Rover Freelander 2 SUV Jaguar XK Coupe 10MY Sports
Lexus RX-450h SUV Jaguar XK Convertible 10MY Sports
Mazda CX-5 SUV Lexus GS-450h Saloon
Mercedes B-Class (for shape) 2-Box Mercedes C250 Saloon
Mercedes ML SUV Mercedes E350 Saloon
Nissan Qashqai SUV Mercedes S350 Saloon
Porsche Cayenne GTS SUV Porsche Panamera Saloon
Range Rover Evoque 5-Door SUV Vauxhall Insignia 08MY Saloon
Range Rover 10MY SUV Vauxhall Insignia 12MY Saloon
Range Rover 12MY SUV
Range Rover LWB 14MY SUV
Range Rover Sport 10MY SUV
Range Rover Sport 14MY SUV
Volvo XC60 SUV
VW Touareg SUV
Table 1 Vehicle Data Sets

The test results from these vehicles are anonymised in all of the subsequent Figures and
Tables.

It was noted that many vehicles, particularly low drag saloons, had a noticeable drag
coefficient minima at 0° yaw (see Figure 3). This yaw response is not as common on SUVs
but still evident on some vehicles. The percentage increase in CD with yaw from straight
ahead for a selection of saloons and SUVs is shown in Figures 3 and 4 respectively.
In many cases the vehicles which have low CD at 0° yaw are far more likely to have a
greater increase in CD with yaw. What is unclear though is whether this yaw response is as
a result of overall vehicle drag reduction or a particular body feature, for example front
bumper planform. Front bumper planform can be used very effectively to control flow
separation at small yaw angles [1] but this can be disadvantageous at large yaw angles.

Figure 3 shows that for Saloons A to F, drag coefficient increases between 5 and 11% over
the range 0-5°. Saloon G shows that this trend is not typical for all vehicles in this sector,
with an increase of less than 5% over the same yaw angle.

Figure 3 Saloon Car Yaw Response

Results in Figure 4 show that SUVs are less likely to show the exaggerated minima of the
saloon vehicles, but where this response is evident the magnitude is much less. In Figure 4,
SUVs A and B show marked minima at 0° but only have a 6-7% rise in CD at 5° yaw with
the other vehicles having more parabolic yaw responses. In the case of SUVs A and B, they
are not vehicles with class leading low CD at 0° yaw.
Figure 4 SUV Yaw Response

4. WIND AVERAGED DRAG METHODS

Having identified that some vehicles have large increases in CD at small yaw angles it
would be useful to be able to quantify the effect of the wind on CD for a given road speed
and to compare vehicle to vehicle. Wind averaged drag (CDW) is one approach that allows
this comparison to be made.

There are 3 commonly known and documented methods for calculating ‘wind averaged’
drag coefficient: MIRA [2], SAE J1252 [3] and TRRL Report 392 [4] and these documents
should be referred to for full derivation of their methods. Each of these methods is very
similar in their general calculation but they differ in how a weighting is applied to the results
as outlined below.

The MIRA method (see Appendix 1) assumes a fixed vehicle velocity and then uses 7
different ambient wind velocities from 2kph - 26kph (1.2mph – 16.2mph) with a weighting
factor applied to each velocity. The wind averaged drag coefficient is then calculated using
133 data points and averaged over the range ±180°. A worked example of the MIRA
method can be found in Appendix 4, and in this case study the vehicle measured CD= 0.272
in the wind tunnel and CDW = 0.289 by calculation.

The SAE J1252 method (see Appendix 2) assumes a fixed vehicle velocity and a fixed
ambient wind velocity of 11.3kph (7mph). In this method it is assumed that the wind
approaches the vehicle with equal probability from any direction. CDW is then calculated
using 12 data points, 6 in both yaw directions.

The TRRL method (see Appendix 3) is similar to the SAE method in that it too assumes a
fixed vehicle velocity and a fixed ambient wind velocity of 11.3kph (7mph). However the
TRRL method applies a weighting factor to the wind direction over the range ±180°. CDW is
then calculated using 12 data points, 6 in both yaw directions.

All of the methods use the following simplified derivation in their calculation:

The relative wind velocity is given by equation (1):

2 2
VRES = VV + VW + 2VVVW Cosθ (1)

And the yaw angle by equation (2):

 VW sin θ 
ψ = tan −1   (2)
 VV + VW cos θ 

The wind averaged drag coefficient CDW can now be defined, assuming longitudinal vehicle
symmetry, by equation (3) where CDΨ is determined using wind tunnel data.

2
1 π V 
CDw = ∫ CDψ  RES  dθ (3)
π 0 V 
 Vψ 

The integral in equation (3) cannot obviously, or simply, be evaluated by finding an


indefinite integral so numerical techniques were employed in its calculation, namely the
mid-point and trapezium rules. A synopsis of the calculations can be found in the Appendix.

5. WIND AVERAGED DRAG RESULTS

The wind averaged drag coefficient (CDW) for all of the vehicles was calculated using the
three methods described above and compared to the CD as measured at the MIRA FSWT,
and the results are shown in Table 2. A fixed vehicle speed of 70 mph, maximum speed for
cars on UK motorways, was used on the calculations. It should be noted that many of the
vehicles were not “eco” or low drag variants; many were fitted with high powered engines
and sports kit with wide tyres.
Table 2 Difference Between the Wind Averaged Drag Methods and MIRA FSWT Results

The SAE method [3] for calculating CDW is the simplest of the three methods, in that a
notional average wind speed is assumed to be equally probable from all directions relative
to the vehicle. The MIRA method [2] gives a weighting to the wind velocity and the TRRL
method [4] gives a weighting to the wind direction distribution for the UK.

It is of no surprise that the SAE and TRRL methods give essentially the same result as they
use a similar approach. The SAE method averages 12 data points to calculate the CDW , and
the TRRL method uses the same 12 data points but then uses a weighted average of 6
averages and thereby avoiding Simpson’s Paradox. The MIRA method is the more
comprehensive of the three methods as it uses a much larger data set of 19 yaw angles
and 7 wind speeds, and therefore is the preferred method in this paper to calculate CDW.

6. DISCUSSION

Comparing the results from the three methods in Table 2, for the saloon cars the method
range is within 0.005 and an average of 0.002, and it is similar for the SUV and 2-box cars
with a maximum range of 0.006 and an average of 0.003. The maximum and minimum
∆CDW for the sports and saloon cars is +0.018 and -0.002 respectively with an average of
∆CDW= 0.009. For the SUV vehicles the maximum and minimum ∆CDW is +0.017 and -0.004
respectively with an average of ∆CDW= 0.010 and no obvious anomalies.
From the full scale wind tunnel tests some vehicles have pronounced CD minima at 0° yaw
and it was thought that this type of yaw response would have a large effect on CDW. Before
the CDW calculations were made it was considered that vehicles with smaller drag rise at
yaw would have lower “real world” drag. This hypothesis is not particularly obvious in Table
2, but if the sensitivity of CDW is extracted from the wind tunnel data in Figure 5, the
evidence becomes more apparent.

For saloons and sports cars the sensitivity of wind averaged drag, taken as ∆(CDW - CD),
becomes more pronounced as CD decreases. From Figure 5, for example those vehicles
with a 0° yaw drag coefficient of CD= 0.290 fall into two distinct groups depending on their
CD yaw response. Vehicles with a parabolic response show a CDW increase between 0.004
and 0.006, whereas those with a pronounced minima show a CDW increase between 0.010
and 0.015. It is clear that CDW should be calculated on an individual basis.

Figure 5 Sensitivity of Wind Averaged Drag (CDW) for Saloons and Sports Cars

It is the value of CDW that, potentially, most accurately reflects the performance of the
vehicle in the real world. The truck industry in North America has been aware of the need to
characterize a vehicle’s behaviour in the natural environment for many years [4] [5], as the
rate of change of drag with yaw is approximately ten times that of saloon cars. More
recently Cooper [6] and Leuschen [7] describe how some aerodynamic devices are sub-
optimal because the effect of turbulence and wind direction has been ignored.

SAE recommends that CDW at vehicle highway speed is used for vehicle performance and
economy predictions.
From Table 2 it can be seen that CDW is between 3-5% higher than CD measured at straight
ahead. A 5% increase in CD is equivalent to approximately only a 0.5% increase in fuel
consumption on the New European Driving Cycle (NEDC); but the NEDC only considers
that the vehicle will be travelling above 70 kph (19 m/s) for a short period of the cycle.

The NEDC is supposed to represent typical car usage in Europe and is used to assess the
emissions levels of cars. It consists of four urban driving cycles and an extra urban driving
cycle as shown in Figure 6. The mean speeds in the urban cycle are low and therefore the
aerodynamic forces are not overly significant.

Figure 6 NEDC Drive Cycle

It is not until the last third of the NEDC, the extra urban cycle, where aerodynamic drag
becomes an important consideration as speed rises above 70kph. But even in this part of
the cycle the vehicle only travels at speeds >70kph for 215s, which is only 18% of the total
NEDC test.

Real World fuel economy should ideally be calculated from a drive cycle that represents the
full range of driving speeds and nearer to real world scenarios. In 1994 BMW published
data [8] for their vehicles based on annual distances driven (Figure 7), and this showed that
on average their cars spent 70% of their annual mileage on motorways or national
highways and only 30% in the city. On this data, aerodynamic drag is very significant in
affecting real world fuel economy and even more so in Germany where the motorway
speeds are higher than in the UK. In the case of the BMW 7 Series it spends nearly 50% of
its annual mileage on motorways and one could safely assume at speeds regularly in
excess of 160kph.
Figure 7 BMW Press Information 1994 [8]

The United Nations is currently developing a proposal for a new global technical regulation
on the Worldwide harmonized Light vehicles Test Procedure (WLTP) which aims to provide
a harmonized method to determine pollutants and CO2 emissions.

This new procedure takes into account higher vehicle speeds in the drive cycle (see Figure
8), but still only over a short period, and stipulates the need for wind tunnels that better
replicate the real world with moving ground and rotating wheels. However the vehicles
continue to be tested in turbulence levels of <1% and more importantly the effect of CD
increase with yaw angle is overlooked.

0 200 400 600 800 1000 1200 1400 1600 1800


Time (s)

Figure 8 Proposed WLTP Drive Cycle


7. CONCLUSIONS

In the “real world”, ambient conditions and vehicle set-up must be taken into account when
calculating the vehicle’s CD. The measured CD, either from wind tunnel tests or from CFD, is
certainly not the CD experienced by the customer. The real world drag coefficient needs to
allow for ambient wind strength and direction (wind averaged drag), vehicle loading (or
attitude), trim level differences and wind tunnel set up.

The suggested method to calculate wind averaged drag coefficient (CDW) is the MIRA
method as it is more appropriate than the SAE or TRRL methods. It uses a wider range of
ambient wind speeds and yaw angles rather than a single wind speed at 6 yaw angles. For
the computation, yaw data is required over the yaw range ±5° in 1° increments and at
±10°and ±15°.

CDW was calculated for a range of vehicles using a sample set of 51 sports, saloons, 2-box
and SUVs, and the average increase was ∆CD= +0.007 to +0.010. Two saloon cars had an
increase of more than ∆CD= +0.015 over their CD at 0° yaw. Therefore it is important to
either use this delta, or the actual CDW for the vehicle in question, when calculating real
world fuel economy.

It should be remembered that drag coefficients obtained from wind tunnel tests and CFD
are highly idealized. The vehicles are set at a predetermined trim height, with a particular
wheel and tyre combination, with the wind at 0° yaw and tested in low turbulence all of
which vary in the real world.

ACKNOWLEDGEMENTS
I thank Jaguar Land Rover for permission to publish this paper. I am also grateful to
colleagues past and present for their encouragement and discussion over its contents. The
views expressed are the author’s own and do not necessarily represent company policy.

REFERENCES

1. Hucho, W-H., Aerodynamics Of Road Vehicles, Fourth Edition


2. MIRA Aerodynamic Wind Tunnel Facilities Users’ Handbook
3. SAE J1252 JUL2012, SAE Wind Tunnel Test Procedure for Trucks and Buses
4. Ingram, K.C. The Wind Averaged Drag Coefficient Applied to Heavy Goods Vehicles,
Transport and Road Research Laboratory (TRRL) Supplementary Report 392, 1978
5. Buckley, F. and Sekscienski, W., Comparisons of Effectiveness of Commercially
Available Devices for the Reduction of Aerodynamic Drag on Tractor-Trailers, SAE
Technical Paper 750704, 1975
6. Cooper, K., Truck Aerodynamics Reborn - Lessons from the Past, SAE Technical
Paper 2003-01-3376, 2003
7. Leuschen, J. and Cooper, K., Full-Scale Wind Tunnel Tests of Production and
Prototype, Second-Generation Aerodynamic Drag-Reducing Devices for Tractor-
Trailers, SAE Technical Paper 2006-01-3456, 2006
8. BMW Techniktag ’94 (annual in-house Engineering Conference)
APPENDICES

Calculation of Wind Averaged Drag Coefficients

Appendix 1 MIRA Method

Notation

VV = vehicle velocity (kph)


VW = wind velocity (kph)
VRES = relative wind velocity (kph)
Ѳ = wind direction relative to the vehicle
Ψ = relative yaw angle
CDΨ = drag coefficient at yaw angle Ψ
CDW = wind averaged drag coefficient for wind velocity VW

Calculations

2 2
VRES = VV + VW + 2VVVW Cosθ (1)

 VW sin θ 
ψ = tan −1   (2)
 VV + VW cos θ 

2
1 π V 
CDw = ∫ CDψ  RES  dθ (3)
π 0 V 
 Vψ 

Where: VV= 110kph

The following values of VW were used with the weighting factor as indicated

VW Weighting
(kph) Proportion
2 0.210
6 0.330
10 0.250
14 0.130
18 0.055
22 0.020
26 0.005

Yaw data is required at 0, 1, 2, 3, 4, 5, 10, 15° in both positive and negative directions
Appendix 2 SAE J1252 Method

Notation

VV = vehicle velocity (kph)


VW = wind velocity (kph)
VRES = relative wind velocity (kph)
Ѳ = wind direction relative to the vehicle
Ψ = relative yaw angle
CDΨ = drag coefficient at yaw angle Ψ
CDW = wind averaged drag coefficient for wind velocity VW

Calculations

 VW sin θ 
ψ = tan −1   (1)
 VV + VW cos θ 

2 2
VRES = VV + VW + 2VVVW Cosθ (2)

2
V  V 
M i = 1 +  W  + 2. W .Cos θ (3)
 VV   VV 

θi = (i × 30°) − 15° (4)

1 6
C DW = ∑ [CDψ M i ]
6 i =6
(5)

Where: VV= 110kph


VW= 11.3kph (7mph)
i = 1 to 6 in increments of 1

Yaw data is required at 0, 1, 2, 3, 4, 5, 10, 15° in both positive and negative directions

It is assumed that the wind approaches the vehicle with equal probability from any direction
Appendix 3 TRRL Method

Notation

VV = vehicle velocity (kph)


VW = wind velocity (kph)
VRES = relative wind velocity (kph)
Ѳ = wind direction relative to the vehicle
Ψ = relative yaw angle
CDΨ = drag coefficient at yaw angle Ψ
CDW = wind averaged drag coefficient for wind velocity VW

Calculations

2 2
VRES = VV + VW + 2VVVW Cosθ (1)

 VW sin θ 
ψ = tan −1   (2)
 VV + VW cos θ 

2
1 π V 
CDw = ∫ CDψ  RES  dθ (3)
π 0 V 
 Vψ 

Where VV= 110kph and VW= 11.3kph (7mph)

The following values of Ѳ were used with the probability / weighting factor as indicated

Wind Direction Probability


(Ѳ°) p(Ѳ)
166-15 0.155
16-45 0.168
46-75 0.170
76-105 0.169
106-135 0.171
136-165 0.167

Yaw data is required at 0, 1, 2, 3, 4, 5, 10, 15° in both positive and negative directions
Appendix 4 MIRA Method Worked Example

Measured data from MIRA:

Wind Averaged Drag Calculation:


View publication stats

You might also like