ICCT Coastdowns-EU 201605

WHITE PAPER MAY 2016
THE IMPACT OF OFFICIAL VERSUS

REAL-WORLD ROAD LOADS ON CO2
EMISSIONS AND FUEL CONSUMPTION
OF EUROPEAN PASSENGER CARS
Jörg Kühlwein
www.theicct.org
communications@theicct.org
BE I J I N G | BERLIN | B R USS E LS | SAN FRANCIS CO | WAS H INGTO N

ACKNOWLEDGEMENTS
The author would like to thank all reviewers of this report for their constructive
comments. This study could not have been conducted without the valuable support
and data delivery of independent European vehicle laboratories. The author would like
to thank TNO in the Netherlands, the Technical University of Graz in Austria, the VTT
Technical Research Centre of Finland, and the Laboratory of Applied Thermodynamics
(LAT) of the Aristotle University in Thessaloniki, Greece. Germany and France
graciously provided the requested official road load data sets, particularly the German
Kraftfahrtbundesamt (KBA) and the French L’Union Technique de l’Automobile du
Motocycle et du Cycle (UTAC).
International Council on Clean Transportation Europe

Neue Promenade 6, 10178 Berlin
+49 (30) 847129-102
communications@theicct.org | www.theicct.org | @TheICCT
© 2016 International Council on Clean Transportation

EXECUTIVE SUMMARY
Driving resistances of a moving vehicle, such as rolling resistance and aerodynamic
drag, have a strong impact on its CO2 emissions and fuel consumption. To properly
simulate these resistance forces when testing stationary vehicles tied down on a chassis
dynamometer, the vehicle’s road loads have to be determined beforehand. European
legislation includes provisions on how to determine the road loads using coastdown
runs. The current rules include high tolerances and systematic errors and do not cover all
technical aspects. This leads in practice to a general underestimation of vehicles’ official
road loads and the corresponding CO2 emissions.
In this report, realistic road load data were determined on 29 light-duty vehicles from
four independent (non-industrial) labs in Europe. The results were compared to the
official road loads determined by manufacturers for certification emission tests. Nine of
the vehicles are also sold in the United States; for these vehicles the lab road load data
was also compared with official EPA road load values. Directly comparing the road load
coefficients is not meaningful, due to different influences for each coefficient and the
effect of mass. Thus, the observed road load differences were translated into total NEDC
cycle energy and CO2 emissions by applying a numeric vehicle emission model.
With the introduction of Euro 5 emission standards for passenger cars in September
2009, manufacturers are required to submit their official coastdown parameters as
part of the type-approval documentation. However, there is no EU-wide database
collecting all official emission test results or the related road load data of the certified
vehicles. Hence, when requesting coastdown data for selected vehicles each individual
national agency has to be contacted separately. For this study, only two member states
(Germany and France) provided the relevant data - four member states (Italy, Great
Britain, Luxembourg and Spain) refused to provide the road load data upon request.
Official road load data for 15 of the 29 vehicles were obtained from Germany and France
and data for four additional vehicles were obtained from the test labs. The other 10
vehicles had to be discarded because of missing official road load data.
The application of realistic road loads instead of the official EU road loads increased
total NEDC cycle energy by 15.0% and CO2 emissions under the NEDC driving regime
by 7.2%, on average. Car road loads under realistic conditions were higher than the EU
data for all 19 vehicles, with corresponding CO2 increases of 0.7% to 14.5%. Regarding
the observed load and CO2 gaps, no significant differences between manufacturers
were observable. A biased reference ambient temperature during the coastdown
test (20 °C instead of 14 °C) and additional coastdown tolerances on the chassis
dynamometer further contribute to the overall CO2 gap by 2.3%. Altogether, assuming
an overall divergence between official and real-world CO2 emissions of 25% in 2010,
more than one third of this gap can be explained by exploited tolerances and errors of
the road load procedures.
i
ICCT WHITE PAPER
Average: 7.2%
Audi A5
Gasoline
BMW 116i
Škoda Octavia Diesel
Dacia Sandero
Fiat 500
Peugeot 407
BMW X1
VW Polo
Renault Scenic
Audi A5
Mitsubishi ASX
Opel Astra
Mercedes E350
Audi A3
Peugeot 508
Opel Insignia
Opel Astra
VW Passat
Peugot 207
0% 2% 4% 6% 8% 10% 12% 14%
CO2 increase with real-world road loads
Figure ES-1. Increase of CO2 emissions with real-world road loads compared to official loads.
The reasons for the observed deviations are manifold. The actual European
legislation (NEDC) is based on a directive from 1970 and was last updated in 1998.
The current rules do not adequately consider improvement in dynamometer and
vehicle technology. They include a large variety of technical tolerances and imprecise
definitions, which still reflect the poor technical standards and imprecision of
technologies in the 1970s. Manufacturers take advantage of the outdated tolerances
to adjust the relevant parameters close to the lower boundary of the tolerance
bandwidth and to create artificially modified vehicles for the official coastdown runs.
Furthermore, the legislation includes systematic errors and outdated assumptions that
lead to a biased calculation of the vehicles’ road loads, which alone can result in CO2
underestimations of more than 7%.
Compared to the situation in Europe, road load data determined under the U.S.
certification requirements (for the same vehicles) are higher and closer to reality.
The average NEDC cycle energy increase was only 4.2% and the CO2 increase
was only 1.8% compared to the official U.S. road load data. That is because of a
better enforcement system and a higher risk of manufacturers getting caught,
as EPA periodically conducts its own road load tests. EPA also releases all of the
manufacturer road load data to the public, allowing anyone to verify the accuracy of
the manufacturer data. The danger of vehicle recalls and the financial consequences
in the U.S. are much greater.
The replacement of the current emission legislation in the EU by the Worldwide

Harmonized Light Vehicles Test Procedure (WLTP), planned for 2017, will
entail improvements and eliminate some of the existing methodological errors.
Manufacturers will be directly responsible for the officially declared road loads. On
the other hand, the WLTP offers completely new options on road load determination,
ii
IMPACT OF OFFICIAL VS. REAL-WORLD ROAD LOADS IN EUROPE
increasing complexity, and the corresponding descriptions of these new methods are
rather vague.
Transparency and independent control measures will become even more important
than today. The vehicles’ driving resistance data relevant for CO2 type-approval tests
should be included in the certificate of conformity (CoC) and summarized in a public
database, together with all certified fuel consumption and emission data. Free access
to the official road load forces is a vital precondition for any independent verification
measures. Furthermore, official in-use compliance measurements must be extended,
and distorted official CO2 emissions based on false assumptions on road load forces
must be discovered and corrected. This requires the establishment of a completely
new road load validation procedure under the WLTP regime.
iii
ICCT WHITE PAPER
TABLE OF CONTENTS
Executive Summary..................................................................................................................... i
Abbreviations.............................................................................................................................. v
1 Introduction........................................................................................................................... 1
1.1 Physical principles of driving resistances................................................................................ 1
1.2 Terminology........................................................................................................................................ 4
1.3 Coastdown runs and road load determination – legal requirements........................... 5
2 Evaluated coastdown runs from independent labs.......................................................... 8

2.1 TNO........................................................................................................................................................ 8
2.2 TUG........................................................................................................................................................ 8
2.3 VTT......................................................................................................................................................... 9
2.4 LAT.......................................................................................................................................................10
2.5 Lab comparisons.............................................................................................................................10
3 Request of official coastdown data.................................................................................. 12

3.1 The data situation in Europe......................................................................................................12
3.2 Responses from national type-approval authorities.........................................................12
4 Comparisons between official and real-world road loads............................................ 15

4.1 Differences of road load forces.................................................................................................15
4.2 Comparison with official U.S. road loads..............................................................................16
4.3 Vehicle simulation and impact on CO2 emissions..............................................................17
4.4 Reasons for road load deviations.............................................................................................22
4.5 NEDC deficiencies not contributing to the road load deviations................................26
5 Conclusions.......................................................................................................................... 28
6 References............................................................................................................................ 30
Appendix A Administrative and technical vehicle data..................................................... 32
Appendix B Coastdown data as part of the EU type-approval......................................... 36
Appendix C Road load determination in NEDC/UNECE legislation................................ 37
Appendix D Road load determination in WLTP regulation............................................... 41
Appendix E Average gasoline and diesel car specifications............................................. 46
iv
ABBREVIATIONS
a Vehicle acceleration
A, B, C Road load coefficients (U.S. labeling)
AF Frontal area
Acc Acceleration
A D Aerodynamic Drag (= Cd * A), with m² as the unit
AT Automatic transmission
CADC Common Artemis Driving Cycle
Cd Aerodynamic drag coefficient
CoC Certificate of Conformity
CO2 Carbon dioxide
DRIRE Centre National de Reception des Vehicules (France)
DVT Data Visualization Tool
EEA European Environmental Agency
EPA United States Environmental Protection Agency
EU European Union
f Front axle driven
F Force
FC Fuel Consumption
fRR Rolling resistance coefficient
f0, f1, f2 Road load coefficients (European labeling)
g Gravity constant
GTR Global technical regulation
HBEFA Handbook Emission Factors for Road Transport
IRP Inertia of rotating parts (expressed as mass equivalent)
KBA Kraftfahrtbundesamt (Germany)
lbf Pound-force
LDV Light-duty vehicle
mass iro Mass in running order (EU definition)
mph Miles per hour
MT Manual transmission
mV Vehicle mass
MY Model Year
N Newton
N1 Light Commercial Vehicles with a maximum mass not exceeding 3.5 tonnes
NEDC New European Driving Cycle
v
ICCT WHITE PAPER
PHEM Passenger Car and Heavy Duty Emission Model

q Two-axle driven
r Rear axle driven
RG Road Gradient
RM Reference mass (mass iro + 25 kg)
RR Rolling Resistance
RRC Rolling Resistance Coefficient
RW Reference weight (= reference mass)
SNCH Société Nationale de Certification et d’Homologation (Luxembourg)
t Time
TA Type-approval
TAN Type-approval number
TMH Test Mass High (WLTP)
TML Test Mass Low (WLTP)
TU Technical University
UNECE United Nations Economic Commission for Europe
UTAC L’Union Technique de l’Automobile du Motocycle et du Cycle (France)
v Vehicle velocity
VCA Vehicle Certification Agency (United Kingdom)
WLTC Worldwide harmonized Light-duty Test Cycle
WLTP Worldwide harmonized Light vehicles Test Procedure
α Road gradient
ρAir Air density
vi
1 INTRODUCTION
Driving resistances of a moving vehicle, such as rolling resistance and aerodynamic
drag, have a strong impact on its CO2 emissions and fuel consumption. To properly
simulate these resistance forces when testing stationary vehicles tied down on a chassis
dynamometer, the vehicle’s road loads have to be determined beforehand. European
legislation includes provisions on how to determine the road loads using experimental
coastdown runs. It was previously reported that the current rules include tolerances and
systematic errors and do not cover all technical aspects, thereby in practice resulting in
a general underestimation of official road load data and the corresponding official CO2
emissions (Kadijk & Ligterink, 2012; Stewart et al., 2015).
This report aims to better quantify the discrepancy between official type-approval and
real-world CO2 emissions caused by inaccurate road load data, by:
»» Comparing real road loads measured by independent labs with official road loads
used for EU type approval for selected light-duty vehicles, and
»» Quantifying the impact of systematic errors in the road load determination
procedure and the biased use of tolerances on CO2 emissions by applying a vehicle
emissions simulation tool.
1.1 PHYSICAL PRINCIPLES OF DRIVING RESISTANCES

The actual fuel consumption and CO2 emissions of a vehicle depend on the vehicle’s
driving resistances, the powertrain efficiency, and the energy demand of potentially
activated auxiliary consumers. The efficiency of the powertrain describes those parts
of the total fuel’s energy content that can be used for the mechanical propulsion of the
vehicle. The majority of the employed chemical energy is lost by heat dissipation and
friction of the powertrain. Engine efficiencies vary among different types of engines
and also among different loads within the engine maps, described by engine speed and
engine power (or torque). Accurate engine efficiency maps are essential for accurate
numerical simulations of vehicles’ fuel consumption and CO2 emissions.
AERODYNAMIC DRAG
ACCELERATION
α
ROAD GRADE
ROLLING
RESISTANCE
Figure 1. Forces affecting a moving vehicle.
1
ICCT WHITE PAPER
The driving resistances of a vehicle are directly linked to the vehicle’s body
characteristics and follow basic physical principles. The total force occurring at the
contact area between tires and road surface consists of four parts: aerodynamic drag,
rolling resistance, acceleration, and a gravity component, which varies based on the road
grade (Figure 1). These forces can be calculated by the following formula:
Total force:
FTOTAL = FAD + FRR + FAcc + FRG
The aerodynamic drag (FAD) of a vehicle consists of the aerodynamic shape of the body,
described by the drag coefficient (Cd), and by the projected frontal area of the vehicle
(AF). The aerodynamic force increases with the square of the vehicle’s velocity:
FAD = Cd * AF * ρAir /2 * v²
The rolling resistance forces (FRR) are mainly determined by the tires, but also by parts
of the driveline. They are characterized by the rolling resistance coefficient, fRR, which is
dependent on the vehicle’s velocity. The mass of the vehicle (perpendicular to the road)
also has a linear influence:
FRR = mV * g * fRR * cos(α)
The acceleration forces (FAcc) increase proportionally with the vehicle’s mass. The
inertias of the rotating parts (IRP, in particular of the wheels) also must be considered:
FAcc = (mV + IRP) * a
The slope forces (FRG) can be calculated directly based on the road gradient and the
vehicle mass:
FRG = mV * g * sin(α)
With:
Cd Aerodynamic drag coefficient
AF Frontal area
ρAir Air density
v Vehicle velocity
mv Vehicle mass
g Gravity constant
fRR Rolling resistance coefficient
α Road gradient
IRP Inertia of rotating parts (mass equivalent)
a Vehicle acceleration
Fuel consumption and CO2 emissions of a light-duty vehicle normally are measured
on a chassis dynamometer under standardized conditions, such as defined driving
patterns and constant ambient conditions. The resistances of the rollers of a chassis
dynamometer have to be adjusted to the vehicle’s driving resistances in the real world
and its mass. For this adjustment, measured rolling resistance and aerodynamic drag
are used. The acceleration forces occurring during the driving cycle are adjusted by
applying the matching inertia. Road gradients are currently not considered on chassis
dynamometers under statutory conditions, but can be simulated by adjusting the inertia.
2
OUTSIDE TEST TRACK: INSIDE LABORATORY:
Vehicle coastdown
140
120
Chassis dyno calibration
Road Load
100 >>>> >
Velocity [km/h]
parameters
80 f0, f1, f2
60
40
Forces
20
0
0 50 100 150 200
t120
t100
Coastdown Test Emissions and

t80
t60
t40
t20
Time [s]
times mass fuel consumption
Recorded in Type Recorded in the

approval documentation, Certificate of Conformity
not accessible to public (CoC), public data
Figure 2. Schematic illustration of the road load determination procedure in the EU.
For the experimental determination of rolling resistance and aerodynamic drag a

coastdown run with the test vehicle normally is performed beforehand. The vehicle is
accelerated on a flat and straight road to a certain velocity (e.g., 130 km/h). The engine
then is decoupled from the drivetrain by switching the gearbox into neutral position,
and the vehicle coasts to a standstill. The velocities and times during this coastdown run
are monitored continuously. A typical velocity-time course and the principle road load
determination procedure are depicted in Figure 2.
The balance of forces during the coastdown deceleration of the vehicle is described by
the following formulae:
- FAcc = FRR + FAD
- (mV+IRP) * a = mV * g * fRR + Cd * AF * ρAir / 2 * v²
Because of the velocity dependency of the rolling resistance coefficient (fRR), it is

difficult to derive the relevant resistance coefficients (fRR and Cd*AF) directly from the
experimental data. Aerodynamic drag increases with the square of the velocity, whereas
rolling resistance is rather constant at low velocities, but increases strongly at high
speeds. The distribution of the forces and their absolute values strongly depend on the
characteristics of the vehicle’s body and tires.
Instead, the deceleration forces are calculated at certain velocity points, and a quadratic
correlation with the velocity as independent variable following the principle of the least
squares deviation is applied. European regulations prescribe the use of six fixed-velocity
intervals for this correlation (see Appendix C). The basic formula for this approach is:
FRR + FAD = f0 + f1 * v + f2 * v²
The derived correlation factors f0, f1 and f2 are called the “road load coefficients.” In
the U.S., these are labeled as A, B and C “target” coefficients. In practice, these three
factors are used together with the vehicle test mass to calibrate the dynamometer
3
ICCT WHITE PAPER
rollers’ resistances. Finally the vehicle on the chassis dynamometer has to overcome
the same forces as on the road during normal driving. This is controlled by additional
dynamometer coastdown runs where the deceleration forces and the times needed for
the predefined velocity intervals have to be identical to the coastdown behavior on the
road. Some tolerances are permitted in the EU, but not in the U.S.
1.2 TERMINOLOGY
Coastdown describes the practical test where the vehicle accelerates to a maximum
velocity and afterward decelerates with a decoupled gearbox. The test can be
performed on an outdoor test track, a chassis roller or a flat belt dynamometer. The
velocity-time course is recorded, and time intervals for certain velocity ranges are
calculated. These time intervals are seen as “coastdown parameters.” In addition, the
exact mass of the vehicle as driven during the experiment (including driver) is essential
for the subsequent data evaluation.
Road load refers to the deceleration forces of the vehicle during coastdown testing
on a test track. The road load is equivalent to the sum of the rolling resistance and
aerodynamic forces and is calculated at certain velocity points by the measured time
intervals and vehicle mass, including the additional equivalent mass of the rotational
inertia, which is caused primarily by the tires. The force-time curve is approximated by a
quadratic correlation approach, following the least squares method. The three resulting
correlation coefficients are referred to as “road load parameters.” Note that the road
load parameters depend on vehicle mass. The usual symbols and units in EU and the U.S.
are summarized in Table 1.
Table 1. Road load parameters in EU and U.S.
EU U.S.
Symbols Units Symbols Units
f0 N N A lbf
f1 N/(m/s) N/(km/h) B lbf/mph
f2 N/(m/s)² N/(km/h)² C lbf/mph²
The road load parameters reflect the driving forces of the vehicle on the test track and
are not to be confused with the dynamometer brake loads (U.S.: Dynamometer set
coefficients). The brake loads of the rollers or the flat belt are important in simulating
realistic driving forces on the chassis dynamometer. They compensate for the missing
aerodynamic drag and the different rolling resistance of the vehicle on the test bench
compared to driving on the road. The brake loads are determined by the difference of the
road loads of the tested vehicles driven on the road and on the chassis dynamometer.
Driving resistances in this report refer to the physical parameters acting in opposition to
the vehicle’s driving forces. These are:
»» Rolling resistance coefficient, in kg/t

»» Aerodynamic drag (as the product of aerodynamic drag coefficient and frontal
area), in m²
»» Vehicle mass, in kg (responsible for acceleration forces)
»» Road gradient (currently not considered in EU legislation)
4
1.3 COASTDOWN RUNS AND ROAD LOAD DETERMINATION – LEGAL

REQUIREMENTS
1.3.1 New European Driving Cycle (NEDC) / United Nations Economic

Commission for Europe (UNECE)
The legal requirements for the official coastdown runs and the mathematical derivation
of road load parameters are laid down in the EU Regulation EC/692/2008. This
regulation corresponds to the UNECE Regulation R83 (United Nations Economic
Commission for Europe [UNECE], 2015b). A summary of the main legal features is given
in Appendix C.
The European legislation is based on a 1970 directive last updated in 1998 (Council
Directive of 20 March 1970). The coastdown legislation’s original intent was to
determine the road loads of a vehicle in the condition that it is driven on the road.
However, the current rules do not adequately consider advances in technology. They
include a large variety of technical tolerances and imprecise definitions that still reflect
the poor technical standards and precision of technologies in the 1970s, for example
when adjusting the rollers’ braking forces of the chassis dynamometer. However,
modern technologies and control devices are much more precise, and today vehicle
manufacturers take advantage of the outdated tolerances by adjusting the relevant
parameters close to the lower boundary of the tolerance bandwidth, disregarding the
statistical nature of measurement tolerances.
Furthermore, the legislation includes systematic errors and outdated assumptions

that lead to a biased calculation of vehicles road loads. Some of these distortions
are addressed in this report, and their impact on NEDC total cycle energies and CO2
emissions are quantified (see section 4.4.2). These include:
»» Road inclination of the test track

»» Inertia of rotating parts
»» Tire tread depth
»» Missing humidity in the air density calculation
»» Shifted share of rolling resistance and aerodynamic drag
As an alternative to the velocity–time measurements made during the coastdown run,
the NEDC legislation also allows for direct torque measurements at the wheels. The
measured torques have to be registered at the predefined vehicle velocities and must be
reproduced during the control coastdown on the chassis dynamometer. However, under
practical aspects, measuring torque of a moving vehicle is still a rather sophisticated
methodology and normally not preferred by manufacturers.
1.3.2 World Harmonized Light Vehicles Test Procedure (WLTP)

A new type-approval procedure, the Worldwide Harmonized Light Vehicles Test
Procedure (WLTP), is currently under development. Its EU introduction is planned
for 2017 (Mock et al., 2014). The current version of the Global Technical Regulation
(GTR) also includes a section on road load determination procedures. In general, the
WLTP rules about the road load issue are much more extensive than the current NEDC
regulation. The WLTP is more precise in the classic road coastdown methodology and
eliminates most of the systematic NEDC errors, with the exception of the air humidity
5
ICCT WHITE PAPER
issue. On the other hand, the WLTP offers completely new options on road load
determination, although the corresponding descriptions of these new methods are
rather vague. Appendix D summarizes the main WLTP road load topics.
It is remarkable that the WLTP regulation clearly states that the manufacturer shall be
responsible for the accuracy of the road load coefficients. Compared to the NEDC, this
is a completely new approach that is meant to prevent large discrepancies between
a coastdown test vehicle and a regular series vehicle, as they are currently observed.
Another new provision in the WLTP is that tolerances within the procedure shall not be
used to underestimate the derived road load coefficients.
In addition to coastdown runs that measure velocities or wheel torques, there are three
new options in the WLTP that can be applied by vehicle manufacturers to derive vehicle
road loads:
»» Road coastdown with on-board anemometry

»» Aerodynamic drag measured in a wind tunnel, with rolling resistance measured on a
chassis roller or flat belt dynamometer
»» By calculation (interpolation approach) from vehicle H (high load) and L (low load)
The procedures for these new determination options are either highly complex (e.g., a
complicated correlation process included in the on-board anemometry methodology)
or in the current form are imprecisely defined (e.g., the measurement and calculation
procedure for determining the aerodynamic drag in a wind tunnel is not described).
Other examples for new “flexibilities” in the WLTP include the following:
»» The criteria of vehicle selection for the interpolation method are not clearly defined.
The introduction of “road load families” makes the situation even more opaque.
»» There is no clear definition of a “vehicle coastdown mode” and no clear rationale for
the use of such a software modification. The full functionality of such a device is not
scrutinized. Hence, a high risk of misuse during the exhaust measurements, which is
to say hidden software modifications, must be expected.
»» The tolerances of the torque meter method are much more lax under WLTP than
under NEDC (6 Nm measurement accuracy compared to 1 Nm with NEDC). The higher
tolerances might increase the attractiveness of this method for vehicle manufacturers.
»» The method to measure rolling resistance on a chassis roller dynamometer includes
a correction formula to consider the higher resistances of the tires on a curved
roller than on a flat road. The suggested correction is rather general, and its source
is unclear.
»» The procedure of load adaptation on the chassis dynamometer is complex, still
erroneous and difficult to figure out. The point of time and number of load setting
adjustments are not prescribed.
»» A description of a control coastdown after the emission test is missing.
Differentiated tolerances for the vehicle forces are no longer included.
Summarizing, these new features of the WLTP are expected to make the process of road
load determination even more complex and difficult to control. Compared to the limited
options in the NEDC, this also could entail new flexibilities and distortions of reality.
6
On the other hand, the WLTP will eliminate some of the systematic errors included in
NEDC/UNECE legislation, and it makes vehicle manufacturers directly responsible for
the accuracy of their declared road load data. At this moment in time, these pros and
cons make it difficult to assess if the introduction of the WLTP will lead to more realistic
official road load data overall.
7
ICCT WHITE PAPER
2 EVALUATED COASTDOWN RUNS FROM

INDEPENDENT LABS
Realistic coastdown data from 29 light-duty vehicles were collected from independent
research laboratories across Europe. In total, four labs contributed data for this study.
The criteria of the considered vehicles were:
»» Light-duty vehicle (car or van)

»» Certified under Euro 5 or Euro 6 emission standard
»» Minimum standards of coastdown procedure fulfilled following the NEDC
regulations
2.1 TNO
TNO, Netherlands, performed an experimental study on cars’ road loads on behalf of
the European Climate Foundation and the Dutch Ministry of Infrastructure and the
Environment (Kadijk & Ligterink, 2012). Besides some Euro 4 cars, five Euro 5 and one
Euro 6 cars were involved in these investigations (Table 2). TNO already evaluated the
achieved coastdown data, compared them with the official manufacturers’ road load
parameters and draw initial conclusions on the reasons of occurring deviations.
Table 2. Available real-world road load data sets from TNO
Engine Engine Mass in

Capacity Power Build running order
ID Model Euro Fuel Transmission ccm kW year kg
TNO01 VW Passat 5 diesel M6f 1598 77 2012 1543
TNO02 Peugeot 207 5 diesel M5f 1560 68 2012 1275
TNO03 Fiat 500 5 gasoline A5f 1242 51 2009 975
Mercedes E350
TNO04 6 diesel A7r 2987 155 2009 1845
Bluetec
TNO05 Renault Scenic 5 diesel M6f 1461 81 2010 1485
TNO06 Peugeot 508 5 gasoline M6f 1598 115 2012 1475
2.2 TUG
The Technical University of Graz, Austria, Institute of Internal Combustion Engines and
Thermodynamics, performs coastdown experiments of light-duty vehicles in irregular
intervals for different publically funded projects. The coastdown data are used for
the calibration of the chassis dynamometer rollers’ brake forces to achieve accurate
emission test results. TU Graz developed its own methodology for the correction of
gradient forces, being applied at a small scale instead of using averaged gradients over
the whole test track. Coastdown data for 10 Euro 5 vehicles were provided (Table 3).
8
Table 3: Available real-world road load data sets from TUG

TUG01 Peugeot 407 SW 5 diesel M6f 1997 103 2010 1612
TUG02 Opel Astra EU5 1.7CDTI 5 diesel M6f 1686 92 2010 1503
TUG03 Fiat Doblo 1.6l 5 diesel M6f 1598 77 2010 1485
TUG04 Fiat Punto EVO 5 gasoline M6f 1368 77 2010 1150
TUG05 Audi A3 Sportback 5 diesel M5f 1598 66 2010 1395
TUG06 Honda Civic 1.4 Comfort 5 gasoline M6f 1339 73 2010 1257
TUG07 Mitsubishi ASX 4WD 5 diesel M6q 1798 110 2011 1600
TUG08 Audi A5 2.0l TFSI 5 gasoline M6f 1984 155 2011 1580
TUG09 Mazda CX-5 5 diesel M6q 2191 110 2012 1663
TUG10 Peugeot Boxer 2.2 HDI 5 diesel M6f 2198 81 2012 2164
2.3 VTT
VTT Technical Research Centre of Finland Ltd is the largest multidisciplinary not-for-
profit research organization in Northern Europe. In 2012, VTT tested the CO2 emission
performance of 10 Euro 5 cars (Table 4). The focus of this study was to determine
real-world CO2 emissions and fuel consumption. In addition to outdoors measurements
at constant speeds and during normal traffic situations, VTT also aimed at reproducing
the type-approval NEDC results, but had no access to the manufacturers’ driving
resistances and “did not want to use them, since we suspected that they were not
entirely reliable” (Ahonen et al., 2012). For this study, VTT provided the measured and
corrected coastdown times and the corresponding vehicle masses.
Table 4. Available real-world road load data sets from VTT

VTT01 Audi A5 1.8 TFSI 5 gasoline M6f 1798 125 2012 1500
VTT02 BMW 116i 5 gasoline M6r 1598 100 2011 1365
VTT03 Citroen C1 1.0i 5 gasoline M5f 998 50 2012 905
VTT04 Dacia Sandero 1.6 Hi-Flex 5 gasoline M5f 1598 77 2011 1157
VTT05 Toyota Yaris Verso-S 5 gasoline M6f 1329 73 2012 1187
VTT06 Fiat Punto 1.3 M-Jet 5 diesel M5f 1248 62 2011 1220
Ford Mondeo 1.6 TDCi
VTT07 5 diesel M6f 1560 85 2012 1560
ECOnetic
Opel Insignia Sports
VTT08 5 diesel M6f 1956 118 2011 1788
Tourer 2.0 CDTI EcoFlex
Škoda Octavia HB 1.6 TDI
VTT09 5 diesel M5f 1598 77 2011 1390
GreenLine
VTT10 Volvo V70 1.6D DRIVe 5 diesel M6f 1560 84 2011 1714
9
ICCT WHITE PAPER
2.4 LAT
The Laboratory of Applied Thermodynamics (LAT) of the Aristotle University in
Thessaloniki, Greece, is operating a single-axle chassis dynamometer for light-duty
vehicle emission measurements. Coastdown experiments have been performed on
behalf of the European Commission. Available data comprise three Euro 5 cars, as
described in Table 5.
Table 5. Available real-world road load data sets from LAT

Trans- Capacity Power Build running order
ID Model Euro Fuel mission ccm kW year kg
LAT01 Opel Astra EU5 1.3 D 5 diesel M5f 1248 70 2012 1393
LAT02 VW Polo 1.2 TSI 5 gasoline M5f 1197 66 2012 1102
LAT03 BMW X1 sDrive20d 5 diesel M6r 1995 120 2012 1565
2.5 LAB COMPARISONS

The applied procedure of deriving the vehicle forces from the measured coastdown
times differs among the four labs under consideration. Hence, a harmonization and
recalculation of the provided data sets was necessary to ensure that the resulting
real-world road loads are based on equal boundary conditions. Table 6 summarizes the
methodologies being used by the different labs and the corrections being applied for
this study.
Table 6. Lab methodologies and data corrections
Averaging of coastdown runs Rotating inertias taken into

in opposite directions? Correction to ambient account for force calculation?
standard conditions? Measured
corrected (20 °C, 1 bar) corrected to vehicle mass?
TNO YES, times Forces YES NO 3% YES
YES, originally
TUG YES, forces (small-scale) - YES 3% YES
4%
VTT YES, forces - YES NO 3% YES
LAT NO, only one direction available - NO NO 3% NO
»» Averaging of coastdown runs: TUG and VTT delivered data with a correct force
averaging (TUG even with small-scale correction approach). TNO data originally
were calculated by applying the erroneous NEDC time averaging procedure (see
Chapter 4.4.2) and were translated into force averaged road load coefficients.
LAT provided data in only one driving direction, assuming no inclination of the
test track.
»» Ambient conditions: TNO, TUG and VTT adapted the directly derived road loads
to ambient standard conditions (20 °C, 1 bar). LAT could not provide information
about the ambient conditions during the coastdown runs. Hence, data could not
be standardized.
»» Rotating inertias: TNO, VTT and LAT did not take into account the equivalent
masses of the rotating inertias. TUG normally assumes an extra charge of 4%
related to the total vehicle mass. All data have been recalculated to 3% extra mass
10
based on the suggestions of the current version of the WLTP GTR (United Nations
Economic Commission for Europe [UNECE], 2015a).
»» Vehicle mass: TNO, TUG and VTT weighed the coastdown vehicles and provided
exact masses. LAT data did not include the real masses of the tested vehicles.
Instead the “mass in running order” was used as stated in the vehicles’ registration.
11
ICCT WHITE PAPER
3 REQUEST OF OFFICIAL COASTDOWN DATA
3.1 THE DATA SITUATION IN EUROPE

With the introduction of Euro 5 emission standards for passenger cars, mandatory
for type-approval since September 2009, the official coastdown parameters are part
of the technical description of the type-approval documentation as specified in EU
Regulation EC/692/2008. The entries in the relevant part of the information document
(see Appendix B for the details) are therefore public data and, in theory, accessible to all
interested parties.
In practice, there is no central agency of the European Union being authorized for
the certification of road vehicles. Instead, the national type-approval authorities are
responsible. Each manufacturer chooses any technical service company in Europe that
is registered in a certain member state and associated with a certain national authority
(Mock & German, 2015). So, theoretically each of the 28 member states can be chosen to
certify a vehicle on behalf of all other EU member states. Practically, at least six different
member states currently are involved in certifications of the major manufacturers’ cars
(see Table 7).
The European type-approval scheme differentiates between a “whole vehicle”

certification and sub-certifications for specific parts of the vehicle. The emission
certification is one of these subset procedures, including all issues on the emission
testing on chassis dynamometer and engine test benches and on the underlying
coastdown experiments. Manufacturers may decide to commission different technical
services for the specific sub-certifications. Hence, the “whole vehicle” certification can
be under different responsibility than the emission certification.
The national type-approval authorities or the European Commission do not publish any
type-approval data which are not already part of the Certificate of Conformity (CoC).
So far, coastdown or road load data are not included in the CoC. And, in contrast to
the U.S., there is no EU-wide database collecting all official emission test results or the
related road load data of the certified vehicles. Hence, up to now, when requesting
coastdown data for selected vehicles, researchers have had to contact each individual
national agency separately.
3.2 RESPONSES FROM NATIONAL TYPE-APPROVAL AUTHORITIES

In the framework of this study, the responsible national type-approval authorities
for 29 individual vehicles were identified. Their type-approval numbers (TAN for
“whole vehicle” certification) and their acronyms for type, variant and version were
extracted from the CO2 monitoring database (European Environment Agency, 2015).
The vehicles were clustered, and the resulting lists were sent to the authorities. Table
7 summarizes the responsible type-approval authorities in Europe and their reactions
to ICCT’s data request.
12
Table 7. Requested type-approval authorities and their reactions
Number of evaluated vehicles

falling within the authority’s remit
Provided official
Country Authority Whole vehicle TA Emissions TA coastdown data?
Germany Kraftfahrtbundesamt (KBA), Flensburg 12 7 YES
Centre National de Reception des YES (data provided
France 5 5
Vehicules (DRIRE), Montlhery by UTAC)
Great Britain Vehicle Certification Agency, Bristol 4 5 NO, declined
Ministero della Infrastrutture e dei
Italy 5 5 NO, declined
Trasporti, Roma
Société Nationale de Certification et
Luxembourg 2 6 NO, no response
d’Homologation (SNCH), Sandweiler
Ministerio de industria, turismo y
Spain 1 1 NO, no response
comercio, Madrid
The German Kraftfahrtbundesamt (KBA) approved a formal application, following

the procedure of the German “Umweltinformationsgesetz” based on the European
Environmental Information Directive (Directive 2003/4/EC). The coastdown data under
French responsibility was provided directly by “L’Union Technique de l’Automobile du
Motocycle et du Cycle (UTAC)” upon request by ICCT. VCA in Great Britain and the
respective ministry in Italy declined to provide coastdown data, pointing to the alleged
confidentiality of the data. Luxembourg and Spain did not respond to ICCT’s request.
The refusal of four out of the six requested type-approval authorities led to an
unintentional bias of the vehicle database, as some manufacturers could not be
considered despite the availability of realistic coastdown data. Vehicles from Toyota,
Honda, Ford and Škoda are usually type-approved in Great Britain; Fiat Group (and
associated manufacturers) vehicles are type-approved in Italy; and the Volvo car in
the database was emission type-approved in Spain. Luxembourg did not provide the
official data for the Mazda car. Audi and BMW vehicles are also emission type-approved
in Luxembourg, but the “whole vehicle” certification for these manufacturers is the
responsibility of the German KBA. As a result, KBA was able to provide the official
coastdown data for these vehicles, with the exception of one BMW X1.
In addition to the official coastdown extracts from German and French type-approval
documentation, manufacturers’ road load coefficients for four vehicles were also
available directly from the labs that provided the realistic coastdown data for this study
(two from VTT, one from LAT, and one from TNO).
Table 8 provides an overview of the sources of the official coastdown and derived road
load data. In summary, 19 vehicles could be used for the road load comparisons. Ten
vehicles had to be excluded because official road load data was missing.
13
ICCT WHITE PAPER
Table 8. Availability of EU official coastdown/road load data
Official coastdown /
ID Model road load available? Source
TNO01 VW Passat YES KBA
TNO02 Peugeot 207 YES UTAC
TNO03 Fiat 500 YES TNO
TNO04 Mercedes E350 Bluetec YES KBA
TNO05 Renault Scenic YES UTAC
TNO06 Peugeot 508 YES UTAC
TUG01 Peugeot 407 SW YES UTAC
TUG02 Opel Astra EU5 1.7CDTI YES KBA
TUG03 Fiat Doblo 1.6l NO
TUG04 Fiat Punto EVO NO
TUG05 Audi A3 Sportback YES KBA
TUG06 Honda Civic 1.4 Comfort NO
TUG07 Mitsubishi ASX 4WD YES KBA
TUG08 Audi A5 2.0l TFSI YES KBA
TUG09 Mazda CX-5 NO
TUG10 Peugeot Boxer 2.2 HDI NO 1)
VTT01 Audi A5 1.8 TFSI YES KBA

VTT02 BMW 116i YES KBA
VTT03 Citroen C1 1.0i NO 2)
VTT04 Dacia Sandero 1.6 Hi-Flex YES VTT

VTT05 Toyota Yaris Verso-S NO
VTT06 Fiat Punto 1.3 M-Jet NO
VTT07 Ford Mondeo 1.6 TDCi ECOnetic NO
VTT08 Opel Insignia Sports Tourer 2.0 CDTI EcoFlex YES KBA
VTT09 Škoda Octavia HB 1.6 TDI GreenLine YES VTT
VTT10 Volvo V70 1.6D DRIVe NO
LAT01 Opel Astra EU5 1.3 D YES KBA
LAT02 VW Polo 1.2 TSI YES KBA
LAT03 BMW X1 sDrive20d YES LAT
1) Italy’s responsibility
2) Great Britain’s responsibility
14
4 COMPARISONS BETWEEN OFFICIAL AND

REAL-WORLD ROAD LOADS
Both sets of road load coefficients, EU official and real-world, were collected for direct
comparison of 19 vehicles. Furthermore, regarding the vehicles chassis and the related
aerodynamic drag, nine cars matched equivalent U.S. models for which official road
load data was available from the U.S. EPA’s public database. Comparing the three road
load coefficients f0, f1, f2 directly to each other is not helpful in most cases, especially
with high values of the f1 parameter that cannot be clearly assigned to one of the two
types of coastdown resistances, specifically rolling resistance and aerodynamic drag.
Moreover, the road load coefficients are always in relation to a specific vehicle mass,
affecting rolling resistance and acceleration forces, which is different in most cases for
the two related road load sets and needs to be harmonized first.
Instead, the vehicle simulation tool PHEM (discussed in Chapter 4.3 ) was applied
to perform NEDC cycle runs under the different vehicle load conditions. The model
runs result in total cycle energies (including the relevant driving forces, i.e., rolling,
aerodynamic and acceleration resistances) and in total CO2 emissions, which can be
directly compared to each other. Average Euro 5 gasoline and diesel car specifications
were used as technical input parameters for the model runs (see Appendix E for
technical details). These specifications correspond to the average definitions for Euro 5
cars from the Handbook Emission Factors for Road Transport (INFRAS, 2014).
4.1 DIFFERENCES OF ROAD LOAD FORCES

In Table 9 all realistic road load data for the 19 selected vehicles and their equivalent
official EU data sets are summarized. The corresponding reference masses for each
set of road load coefficients are also included. To make them comparable, the road
load data have to be converted into sets of a unique vehicle mass (see Chapter 4.3),
identified in the table as “Test Mass.” This mass was used for the simulation runs to
achieve the vehicle’s energy demand over the NEDC cycle. The calculated cycle energies
clearly show that the official EU road load sets are less demanding than the realistic
road loads for all 19 of the vehicles examined.
15
ICCT WHITE PAPER
Table 9. Road load parameters applied for type-approval measurements and derived from real-world coastdown experiments
Road loads EU type approval Road loads real-world
Test Mass f0 f1 f2 Reference NEDC cycle f0 f1 f2 Reference NEDC cycle

Vehicle ID kg N N/(m/s) N/(m/s)² Mass kg energy Wh N N/(m/s) N/(m/s)² Mass kg energy Wh
TNO01 1810 115.0 1.33 0.366 1360 1269 218.2 -0.97 0.462 1470 1483
TNO02 1250 78.4 1.29 0.383 1250 1246 149.0 2.08 0.488 1130 1423
TNO03 1020 86.0 0.61 0.415 1590 1590 126.4 1.07 0.435 1590 1852
TNO04 1930 157.0 2.23 0.382 1590 1532 305.0 -0.42 0.478 1360 1560
TNO05 1470 74.4 1.92 0.487 1360 1256 213.9 -0.48 0.519 1360 1389
TNO06 1470 78.6 3.01 0.312 1810 1507 159.6 6.56 0.298 1566 1856
TUG01 1590 111.9 2.81 0.348 1250 1217 170.2 1.78 0.461 1277 1538
TUG02 1470 56.8 4.34 0.341 1020 1149 166.1 6.91 0.353 1124 1266
TUG05 1590 109.0 1.33 0.377 1930 1759 73.4 9.84 0.215 1962 2134
TUG07 1700 147.5 3.12 0.420 1470 1463 173.1 6.24 0.463 1652 1697
TUG08 1700 123.1 1.60 0.344 1470 1335 100.5 4.73 0.284 1523 1679
VTT01 1700 110.0 1.62 0.340 1590 1501 170.9 0.00 0.499 1700 1688
VTT02 1360 125.1 1.19 0.410 1470 1361 154.4 0.00 0.478 1640 1706
VTT04 1130 69.7 1.84 0.440 1590 1414 120.7 0.00 0.562 1460 1617
VTT08 1700 97.7 1.29 0.365 1700 1752 186.2 0.00 0.507 1620 2048
VTT09 1360 84.0 1.33 0.353 1700 1487 112.6 0.00 0.408 1660 1529
LAT01 1360 69.5 1.04 0.407 1700 1431 161.8 -3.68 0.614 1605 1705
LAT02 1250 84.9 1.34 0.390 1360 1395 123.2 1.23 0.426 1595 1428
LAT03 1590 140.4 1.11 0.445 1130 1266 317.2 -8.47 0.681 1240 1413
1622
Average cycle energies [Wh]: 1410
(+15.0%)
4.2 COMPARISON WITH OFFICIAL U.S. ROAD LOADS

In contrast to Europe, in the U.S. road load coefficients applied at official emission tests
are published in large databases and freely accessible on the EPA website (epa.gov).
For the purpose of this study, where appropriate, equivalent U.S. models were identified
and their U.S. road loads were extracted from EPA’s test database (U.S. Environmental
Protection Agency, 2014b). The criteria for the U.S. road load selection were that
the vehicle model should belong to the same vehicle generation and have the same
bodywork as its European counterpart. This ensures equivalent aerodynamics between
U.S. and EU models. In most cases vehicle models and their names were identical.
Sometimes one will find different model names for similar vehicles in the different
markets (e.g., Mitsubishi ASX and Outlander; Opel Insignia and Buick Regal). It should be
noted that the test masses (equivalent inertias) for the same vehicle model may differ
between official EU and U.S. emission tests – they are mostly higher in the U.S. This has
been taken into account when deriving the rolling resistance coefficients.
Table 10 includes the selected U.S. models, their model year, engine size and type of
transmission, and their related European matches. Altogether matching criteria were
found for nine vehicles, including two matches for the Audi A5, for a net of eight
different model matches.
16
Table 10: European vehicle models with real-world coastdown data and U.S. equivalents
Build Model
Vehicle ID EU vehicle model year Engine Transmission US vehicle model Year Engine Transmission
LAT03 BMW X1 sDrive20d 2012 2.0 D M6q BMW X1 xDrive28i 2013 2.0 G A8q
Mercedes E350
TNO04 Mercedes E350 Bluetec 2009 3.0 D A7r 2011 3.0 D A7r
Bluetec
TUG05 Audi A3 Sportback 2010 1.6 D M5f Audi A3 Sportback 2009 2.0 G M6f
TUG06 Honda Civic Comfort 2010 1.4 G M6f Honda Civic 2008 1.8 G M5f
Mitsubishi
TUG07 Mitsubishi ASX 2011 1.8 D M6q 2012 2.0 G A6q
Outlander
TUG08 Audi A5 2011 2.0 G M6f Audi A5 2013 2.0 G A8f
TUG09 Mazda CX-5 2012 2.2 D M6q Mazda CX-5 2013 2.0 G M6f
VTT01 Audi A5 2012 1.8 G M6f Audi A5 2013 2.0 G A8f
VTT08 2011 2.0 D M6f Buick Regal 2012 2.0 G M6f
Tourer EcoFlex
Table 11 directly compares the official U.S. road load coefficients of the selected
U.S. vehicle models and the related reference masses (as tested on the chassis
dynamometer) with the equivalent road loads derived from the real-world coastdown
experiments. In addition, as a result from the simulation runs, the total cycle energies
needed for a NEDC test run are given. It can be seen that the energy demands when
applying the U.S. road loads are much closer to the real-world energies (sometimes even
higher) than is the case for the EU official road loads from Table 9.
Table 11. Road load parameters applied for U.S. certification and derived from real-world coastdown experiments
Road loads, U.S. certification Road loads, real-world

Test
Vehicle Mass f0 f1 f2 Reference NEDC cycle f0 f1 f2 Reference NEDC cycle
ID kg N N/(m/s) N/(m/s)² Mass kg energy Wh N N/(m/s) N/(m/s)² Mass kg energy Wh
LAT03 1590 220.2 -3.84 0.566 1758 1657 317.2 -8.47 0.681 1590 1852
TNO04 1930 181.7 3.35 0.347 2041 1836 305.0 -0.42 0.478 1962 2134
TUG05 1590 137.9 1.49 0.445 1644 1555 73.4 9.84 0.215 1460 1617
TUG06 1250 104.1 3.95 0.326 1361 1358 156.3 0.00 0.424 1300 1432
TUG07 1700 194.9 -2.00 0.616 1644 1851 173.1 6.24 0.463 1620 2048
TUG08 1700 164.0 4.43 0.387 1814 1733 100.5 4.73 0.284 1660 1529
TUG09 1700 98.0 2.89 0.513 1588 1722 81.5 9.82 0.313 1720 1767
VTT01 1700 164.0 4.43 0.387 1814 1715 170.9 0.00 0.499 1605 1705
VTT08 1700 178.4 1.88 0.459 1814 1735 186.2 0.00 0.507 1800 1724
1756
Average cycle energies [Wh]: 1685
(+4.2%)
4.3 VEHICLE SIMULATION AND IMPACT ON CO2 EMISSIONS

To make the different road loads more comparable to each other, their direct impact
on CO2 emissions for passenger cars was determined by the application of an emission
simulation tool. The applied “Passenger car and heavy duty emission model” (PHEM)
is an emission map-based instantaneous tool, which has been developed by TU Graz
17
ICCT WHITE PAPER
since the late 1990s (Luz & Hausberger, 2014). It calculates the fuel consumption
and emissions of road vehicles in 1 Hz temporal resolution for a given driving cycle
based on the vehicle longitudinal dynamics and emission maps (Figure 3). The engine
emission maps are generated based on emission measurements on engine test stands
or more frequently by chassis dynamometer tests or portable emissions measurement
systems (PEMS).
In the frame of this study, the specific CO2 emission maps for each individual vehicle
with available realistic coastdown data were not available. Instead, the latest average
Euro 5 emission maps for gasoline and diesel passenger cars included in the PHEM
database were applied. These emission maps were derived from instantaneous emission
measurements of 18 gasoline and 27 diesel vehicles, driven at different driving cycles
also covering high load conditions, for example under the regime of the Common
Artemis Driving Cycle (CADC). Applying the same CO2 emission maps for all examined
gasoline respectively diesel cars when simulating NEDC results might lead to some over-
or under-estimations of the absolute CO2 test result. However, the focus of this exercise
is to quantify the relative difference when applying two different sets of road loads. For
this purpose, the absolute emission levels are immaterial.
DRIVING CYCLE IN 1 HZ VEHICLE DATA ENGINE EMISSION MAPS
Driving Drivetrain
Velocity resistances losses
Engine Emissions and

power fuel consumption
Gear schedule (MT) Transmission ratios
Engine
speed
Gear shift model (AT)
Figure 3. Scheme of the applied simulation tool PHEM.
Specific technical data for each vehicle model was used as input for the PHEM
simulation runs (see Appendix E):
»» Mass
»» Rated engine power
»» Transmission ratios
»» Tire dimensions
Idle and rated engine speeds were not known in detail for every vehicle model. Instead,
the Euro 5 averages of the PHEM database were applied. Also standard values were
used for required energy losses of the drivetrain.
18
The PHEM model requires the driving resistance parameters as input to describe the
vehicles’ resistance forces. Hence, before starting the model runs, both sets of calculated
road load parameters (official and realistic) were translated into rolling resistance
coefficients (from f0 and f1 parameters) and aerodynamic drag (from f2 parameter). This
procedure enables a fully accurate usage of the road load data, although the derived
driving resistance parameters themselves include some uncertainties, especially with
high f1 values that cannot be clearly assigned to rolling resistance or aerodynamic drag.
An advantage of this transformation of road load coefficients into driving resistance
parameters is the mass independency of the latter, allowing for direct comparisons
between official and realistic driving resistance-based simulated cycle energy and
CO2 emissions. For each vehicle model, a unique vehicle test mass was applied in the
simulation runs for both compared load variants.
CO2 emissions over the complete NEDC driving cycle were simulated. Starts with a
hot engine were considered, assuming that the relative cold start effect on CO2 is not
affected by varying vehicle road loads. Note that the only target of these simulation
studies was to identify the relative effect of altered road load parameters. The resulting
absolute CO2 levels are not necessarily realistic, are not relevant here, and must not
be compared with CO2 emission data from other sources, like type-approval or other
measurement values.
Table 12 includes the NEDC CO2 results as achieved from the PHEM model runs. Only
those vehicle models with available realistic road loads and at least one comparable
set of official road loads (EU or U.S.) were considered. The relative deviations between
CO2 emissions based on realistic road loads compared to those based on official load
data are given in the last two columns. The vehicle models and their results in this table
are sorted by decreasing relative differences based on the official EU load data, that is,
vehicles with the largest gaps are listed at the top. Figure 4 gives an overview on the
resulting CO2 increases.
19
ICCT WHITE PAPER
Table 12. Differences between simulated NEDC (hot start) CO2 emissions applying EU type-approval road loads, U.S. certification road
loads and real-world road loads, sorted by relative deviations real-world against EU
Test Mass CO2 EU road CO2 US road CO2 real road delta CO2 EU delta CO2 US
Vehicle ID Vehicle Model kg loads g/km loads g/km loads g/km real/EU-1 real/US-1
TNO02 Peugeot 207 1250 104.3 119.4 14.5%
TNO01 VW Passat 1810 126.0 142.1 12.8%
TUG02 Opel Astra EU5 1.7CDTI 1470 132.3 147.3 11.3%
VTT08 1700 142.8 157.0 156.5 9.6% -0.3%
TNO06 Peugeot 508 1470 178.6 194.4 8.8%
TUG05 Audi A3 Sportback 1590 111.1 118.1 120.9 8.8% 2.4%
TNO04 Mercedes E350 Bluetec 1930 192.0 197.3 209.0 8.8% 5.9%
LAT01 Opel Astra EU5 1.3 D 1360 117.1 126.8 8.2%
TUG07 Mitsubishi ASX 4WD 1700 183.5 187.7 197.1 7.4% 5.0%
VTT01 Audi A5 1.8 TFSI 1700 172.7 185.8 185.5 7.4% -0.2%
TNO05 Renault Scenic 1470 142.7 153.1 7.3%
LAT02 VW Polo 1.2 TSI 1250 126.5 135.0 6.7%
LAT03 BMW X1 sDrive20d 1590 165.9 168.5 175.8 6.0% 4.4%
TUG01 Peugeot 407 SW 1590 146.5 154.0 5.1%
TNO03 Fiat 500 1020 109.9 115.4 5.0%
VTT04 Dacia Sandero 1.6 Hi-Flex 1130 151.0 157.9 4.6%
VTT09 1360 112.9 115.6 2.3%
GreenLine
VTT02 BMW 116i 1360 157.2 158.7 0.9%
TUG08 Audi A5 2.0l TFSI 1700 203.4 214.7 204.9 0.7% -4.5%
TUG06 Honda Civic 1.4 Comfort 1250 152.3 155.7 2.3%
TUG09 Mazda CX-5 1700 180.3 181.9 0.9%
Average EU (19 vehicles): 7.2%
Average US (9 vehicles): 1.8%
Averages EU and US (7 vehicles): 7.0% 1.8%
20
Average: 7.2%
Audi A5
Gasoline
BMW 116i
Škoda Octavia Diesel
Dacia Sandero
Fiat 500
Peugeot 407
BMW X1
VW Polo
Renault Scenic
Audi A5
Mitsubishi ASX
Opel Astra
Mercedes E350
Audi A3
Peugeot 508
Opel Insignia
Opel Astra
VW Passat
Peugot 207
0% 2% 4% 6% 8% 10% 12% 14%
CO2 increase with real-world road loads
Figure 4. Increase of CO2 emissions with real-world road loads compared to official loads.
The main results of this study are:
»» The application of realistic road loads instead of official EU road loads on average
leads to 15.0% higher total NEDC cycle energy and 7.2 % higher CO2 emissions
under the NEDC driving regime.
»» Realistic road loads are higher than official EU road loads for all examined cars.
CO2 increases ranged from 0.7% to 14.5%. The highest gaps can be observed for a
Peugeot 207, a VW Passat and an Opel Astra.
»» Concerning the quantified CO2 gaps, no significant differences are observable
between manufacturers.
»» Comparing realistic to official U.S. road loads leads to just 1.8% higher CO2
emissions on average. This value should be compared to the EU average of 7.0%,
where only those vehicles with all three available road load data sets (real, EU and
U.S.) are considered.
»» Some vehicle models offered in the U.S. show even higher road loads under U.S.
testing conditions than derived from the real-world coastdowns, with CO2 impacts
ranging from a decrease of 4.5% to an increase of 5.9% overall.
»» Regarding the calculated gaps, no lab specific systematic differences are
observable after applying data corrections and harmonization of the data
evaluation methods.
21
ICCT WHITE PAPER
4.4 REASONS FOR ROAD LOAD DEVIATIONS

The results clearly indicate that the current NEDC regulation is not adequate to establish
realistic road load parameters to be applied at the official type-approval measurements.
The potential reasons for the observed gaps between realistic and official road loads are
diverse. The NEDC procedure is rather imprecise and the flexibilities and shortcomings
of unregulated issues it contains are exploited, and in some cases misinterpreted,
by vehicle manufacturers. This results in modifications of the coastdown test vehicle
deviating from a regular series vehicle and in the use of special coastdown testing
conditions. Furthermore, some systematic errors implemented in the NEDC regulation
since its introduction in the 1970s have never been corrected.
Chapter 5 discusses additional testing deficiencies that do not contribute to the

observed road load gap, but additionally contribute to the increasing gap between
realistic and official CO2 emissions.
4.4.1 Legal flexibilities and shortcomings of the NEDC procedure

The intention of the NEDC regulation as created in the 1970s was to test a vehicle in
the same condition as it is driven on the road. However, technical specifications and
details have not been addressed comprehensively. Over time, these missing definitions
were increasingly exploited by vehicle manufacturers, establishing artificial test cars and
applying unusual testing conditions to lower the official road load values. Some of these
measures violate the actual regulation. For example, tire pretreatment is in clear contrast
to the legislation stating that “the rolling resistance characteristics of the tires fitted to
production vehicles shall reflect those of the tires used for type approval.”
Some of the possible measures used by manufacturers are:
»» Pretreatment of tires, baking them in an oven such that they are thermally hardened
and/or shaving them so they are almost bald.
»» Optimizing aerodynamics by modifying the vehicle’s chassis, such as closing or
making new openings, removing exterior mirrors, and adapting the suspension
system to reduce vehicle ride height.
»» Optimizing the warm-up procedure of the test vehicle and relevant parts: Tires,
bearings, gearbox and differential oil can be preheated, which reduces rolling
resistance and friction losses of mechanical parts and lubricants.
»» Elongated running-in period of the vehicle: Additional driving before testing results
in lower friction of the relevant rotating parts (gearbox, driveshaft, wheels).
»» Optimizing wheel alignments: Toe and camber can be adjusted to minimize tires’
rolling resistance, deviating from the regular adjustment that incorporates safety
and comfort aspects.
»» Manually opening brake calipers: The NEDC test procedure explicitly allows manual
brake opening before the coastdown test to eliminate parasitic losses caused by
grinding friction pads.
»» Optimizing resistance of the road surface of the test track: A smooth, hard and
clean road surface reduces tires’ rolling resistance.
All of these listed items potentially contribute to the observed road load gap. However,
their use is not documented in the official type-approval records and is not reported by
22
vehicle manufacturers elsewhere. Hence, it is unknown which of these measures, and

perhaps others, are applied by which manufacturers and to what extent. This also makes
it difficult to quantify the effect of each of these individual measures in terms of reduced
CO2 emissions, as measurement data is not published by manufacturers and data from
independent sources are rare.
4.4.2 Systematic errors of the NEDC coastdown procedure

The current NEDC regulation includes systematic errors, which lead to
underestimations of the derived vehicle road loads. This report quantifies the impact
on CO2 emissions of average gasoline and diesel cars for five of these errors: faulty
averaging method of test runs in opposite directions, missing rotating inertias,
faulty wording regarding tire preconditioning, biased shares of rolling resistance and
aerodynamic drag, and missing humidity in air density calculations. These systematic
deficiencies of the NEDC have existed since its introduction in the 1970s and have
never been officially addressed and corrected.
Road inclination of the test track

The NEDC regulation includes a faulty calculation procedure regarding the method
of averaging the coastdown runs in opposite directions. It prescribes an average
of the measured coastdown times from the opposite runs instead of averaging
the calculated forces. This error is especially relevant at sloping test tracks, as it
increases overproportionally with the track’s gradient. Table 13 shows the effect of the
miscalculations for stepwise road inclinations. Already a small gradient of 1% leads to
a dramatic distortion of the resulting forces and the vehicle’s energy demand at the
dynamometer test, resulting in CO2 underestimations of approximately 5%. The NEDC
allows gradients up to 1.5%, but gradients of more than 1% are not applicable for vehicles
with low rolling resistance, as explained in the footnote to Table 13.
Table 13. The effect of a faulty averaging procedure in NEDC on cycle energy and CO2
Average gasoline car Average diesel car
Gradient of NEDC cycle NEDC CO2 NEDC cycle NEDC CO2

test track energy Wh emissions g/km Systematic error energy Wh emissions g/km Systematic error
0.0% 1477 133.2 1620 132.1
0.1% 1475 133.1 0.0% 1618 132.1 -0.1%
0.2% 1470 132.9 -0.2% 1612 131.9 -0.2%
0.3% 1460 132.6 -0.4% 1601 131.5 -0.5%
0.4% 1447 132.2 -0.8% 1587 131.0 -0.8%
0.5% 1431 131.6 -1.2% 1568 130.4 -1.3%
0.7% 1386 130.0 -2.4% 1518 128.6 -2.6%
1.0% 1291 126.7 -4.9% 1412 125.0 -5.4%
1.5%1
1072 116.3 -12.7% 1
1164 111.5 -15.6%1
1. On

test tracks with a gradient of more than 1.0%, during downhill coasting the final stationary velocity of the average cars is above the
required minimum velocity of 15 km/h. Hence, the road load procedure cannot be applied, and the tabulated values are only theoretical.
Inertia of rotating parts

In the NEDC regulation, the equivalent masses of the inertias of the vehicle’s rotating
parts (mainly wheel assemblies, consisting of tires, rims, and mounted brake discs or
drums) are not added to the vehicle mass for the calculation of the driving forces. This
produces an underestimation of the total effective mass of the tested coastdown vehicle
23
ICCT WHITE PAPER
and consequently results in underestimating resistance forces and cycle energies. Table
14 summarizes the effect on the NEDC cycle energy and CO2 emissions, assuming a
share of the rotational equivalent mass in the total vehicle mass of 3%.1 For averaged
gasoline and diesel cars, the CO2 emissions are underestimated by 0.8%.
Table 14. The effect of missing equivalent masses for inertias of rotating parts on cycle energy and CO2

Inertias NEDC cycle NEDC CO2 Systematic NEDC cycle NEDC CO2 Systematic
considered? energy Wh emissions g/km error energy Wh emissions g/km error
YES 1477 133.2 1620 132.1
NO 1452 132.1 -0.8% 1593 131.0 -0.8%
Tire tread depth

The NEDC regulation wording is deficient regarding the accepted run-in procedure of
the tires used for the coastdown experiment. It reads, “Run-in at the same time as the
vehicle or tread depth 50-90%.” This wording allows manufacturers the option to use
tires with low tread or even completely treadless tires, not even complying with security
regulations for driving on public roads, as long as they are mounted during the run-in of
the test vehicle. Assuming an average tire tread depth for new summer tires of 8 mm, a
regular baseline tire with 70% tire tread and a reduction of the rolling resistance of 1%
with a reduced tread depth of 1 mm, the “benefit” of a treadless tire in terms of CO2 is
0.6% (European Commission, 2015b). Table 15 summarizes the model results.
Table 15. The affect of using treadless tires on cycle energy and CO2

Tire tread NEDC cycle NEDC CO2 Systematic NEDC cycle NEDC CO2 Systematic
depth energy Wh emissions g/km error energy Wh emissions g/km error
70% / 5.6 mm 1477 133.2 1620 132.1
0% / 0.0 mm 1457 132.4 -0.6% 1598 131.3 -0.6%
Biased share of rolling resistance and aerodynamic drag

The formula to correct total driving forces for changed ambient conditions (atmospheric
temperature and pressure) in the NEDC regulation includes an estimate of shares for rolling
resistance and aerodynamic drag, depending on vehicle mass and actual velocity. That is
because air pressure only affects air density and aerodynamic drag, while temperature also
has an influence on tires’ rolling resistance (see Appendix C). These tabulated shares are
outdated, as rolling resistances of tires have improved more in recent decades than vehicle
chassis aerodynamics. Actual vehicle resistance data show that NEDC overestimates the
share of rolling resistance by approximately 5%, especially at higher driving speeds. This bias
leads to underestimated correction factors if the coastdown is performed at low ambient
temperature and at low atmospheric pressure (high altitudes). NEDC includes a tolerance
margin of 7.5% for air density only under reference conditions (20 °C, 1 bar). It does not
specify tolerances for temperature and pressure.
1 For measurements on the chassis dynamometer, 1.5% extra mass is added to the reference mass on a one-
axle driven dynamometer, reflecting the rotating inertia of the non-rotating wheels, that is, half of the 3%
contribution of rotating inertia for all driven wheels. On a two-axle driven dynamometer no extra charge is
applied as all wheels have to be driven by the vehicle and the required energy is reflected directly by the
test results.
24
Table 16 summarizes the effects of optimized ambient conditions. Two scenarios

were evaluated against the reference condition: 5 °C at 950 mbar (an approximate
altitude of 500 meters altitude) and -10 °C at 900 mbar (an approximate altitude of 1000
meters). With the use of regular tires, that is, market products as driven on the road,
CO2 underestimations of 0.6% (gasoline) and 0.7% (diesel) were found. Note that the
application of optimized tires as described above will increase the negative impact
on derived CO2 emissions. Furthermore, the NEDC regulation also explicitly allows
manufacturers to specify their own values for the shares of driving resistances. These
values are not published and can easily contribute to further distortions.
Table 16: The effect of biased shares of rolling resistance and aerodynamic drag on cycle energy and CO2

Ambient NEDC cycle NEDC CO2 Systematic NEDC cycle NEDC CO2 Systematic
conditions energy Wh emissions g/km error energy Wh emissions g/km error
20 °C, 1000 mbar 1477 133.2 1620 132.1
5 °C, 950 mbar 1469 132.8 -0.3% 1610 131.7 -0.3%
-10 °C, 900 mbar 1461 132.4 -0.6% 1599 131.2 -0.7%
Missing humidity in air density calculation

The NEDC algorithm assumes dry air, that is, a relative humidity of 0%. In reality, water
molecules are lighter than nitrogen and oxygen. Therefore, water vapor decreases the air
density and the vehicle’s aerodynamic drag. As quantified in Table 17, missing humidity
has only a minor impact, with a maximal 0.1% CO2 underestimation in the case of 100%
relative humidity at an ambient temperature of 20 °C.
Table 17. The effect of not considering humidity in air density calculation on cycle energy and CO2

Relative
humidity NEDC cycle NEDC CO2 Systematic NEDC cycle NEDC CO2 Systematic
(at 20 °C) energy Wh emissions g/km error energy Wh emissions g/km error
0% 1477 133.2 1620 132.1
50% 1475 133.1 -0.05% 1618 132.0 -0.05%
100% 1477 133.0 -0.1% 1616 131.9 -0.1%
Sum of systematic errors

The full exploitation of the identified systematic NEDC errors by a manufacturer –
selection of a test track with 1.0% gradient, omitting rotational inertias, completely
bald tires, optimized ambient temperature and pressure, omitting humidity – leads
to total CO2 underestimations of 7.0% for the average gasoline car and for 7.6%
for the average diesel car. Table 18 summarizes the total effects under the most
“optimized” conditions.
25
ICCT WHITE PAPER
Table 18. Summary of systematic errors in the NEDC/UNECE coastdown procedure
CO2 underestimation
Factor Assumption Gasoline car Diesel car
Road inclination of the test track 1% gradient -4.9% -5.4%
Inertia of rotating parts omitted -0.8% -0.8%
Tire tread depth 0% / 0 mm -0.6% -0.6%
Biased share of rolling resistance and aerodynamic drag -10 °C, 900 mbar -0.6% -0.7%
Missing humidity in air density calculation 100% humidity -0.1% -0.1%
SUM: -7.0% -7.6%
4.4.3 Additional considerations – simulation approaches

From informal sources it is known that for the majority of all vehicle models, the official
road load data used are from computer simulations, not direct measurements from
physical coastdown trials. The effort required to perform coastdown experiments is
rather high, and the large variety of vehicle models and variants requires the generation
of large road load databases. Hence, manufacturers enable numerical data processing
based on existing data of similar vehicle models.
The technical process of simulating road load data for slightly modified vehicles from
already existing data sets should principally be under the control of the technical
services companies. But the procedure of applying numerical instead of experimental
approaches is not regulated at all. There are no criteria about the accuracy of such
simulation tools, nor are there any external validations of the models’ results foreseen.
The methodologies and achieved results are not included in the official type-approval
documentation and are therefore not reproducible by third parties. Hence, the total
process is vulnerable to additional distortions. The impact of these additional flexibilities
caused by simulation processes on CO2 emissions cannot be assessed here because of
missing information.
4.5 NEDC DEFICIENCIES NOT CONTRIBUTING TO THE ROAD

LOAD DEVIATIONS
Other deficiencies of the NEDC road load regulations do not contribute to the road load
gap, but further increase the overall gap between official and real-world CO2 emissions.
For the sake of completeness, these factors are discussed in this section.
Tolerances of control coastdown

In the course of the emission certification process, the experimentally derived road loads
are transferred to the chassis dynamometer test bench where they are reproduced by
the calibration of the roller brakes. A control coastdown on the chassis dynamometer
is performed regularly after the NEDC emission test to check if the driving forces of
the dynamometer test vehicle match the original coastdown forces. Regarding the
alignment of test track and dynamometer forces, the NEDC regulation allows tolerances,
which are exploited by vehicle manufacturers in systematically setting the vehicle road
loads as close as possible to the lower boundary of the tolerance bandwidth. In practice,
the given force tolerances (±5% at 120, 100, 80, 60 and 40 km/h; ±10% at 20 km/h)
cannot be exploited completely, as the brake forces for the six speed ranges cannot be
26
adjusted separately. However, an exploitation of -4% at 120, 100, 80, 60 and 40 km/h
and -8% at 20 km/h seems to be achievable. Table 19 summarizes the results of such a
deliberately biased calibration of dynamometer forces. The level of CO2 emissions can
be cut by approximately 1.2%.
Table 19. The impact of exploitation of dynamometer load tolerances on cycle energy and CO2

Exploitation
of lower NEDC cycle NEDC CO2 Systematic NEDC cycle NEDC CO2 Systematic
boundary energy Wh emissions g/km error energy Wh emissions g/km error
0% 1477 133.2 1620 133.2
- 4% / - 8% 1439 131.6 -1.2% 1580 130.5 -1.2%
Ambient reference temperature
The ambient temperature during the coastdown experiment affects both resistance
types. A higher air temperature decreases the air density and, hence, decreases the
vehicle’s aerodynamic drag. A higher tire temperature also decreases the tires’ rolling
resistance. Realistic road loads must be harmonized to be comparable to the official
road loads. Hence, in this report all road load data have been normalized to NEDC
ambient reference conditions (1000 mbar, 20 °C). However, the average European
ambient temperature weighted by driven mileage is lower than 20 °C. The European
Commission assumes 14 °C to be a more realistic European average (European
Commission, 2015a). This lower temperature is going to be established as the reference
for the emission tests under the WLTP regime, but not for the related coastdown tests,
which perpetuates the inconsistency.
In Table 20 the effects of applying a more realistic ambient temperature of 14 °C instead

of 20 °C are shown. The higher aerodynamic drag corresponds to an increase of CO2
by 0.35%, while the tires’ higher rolling resistances result in 0.75% higher CO2 emissions.
Altogether the 6 °C reduction of the ambient temperature increases road loads and CO2
increase, the latter by 1.1%.
Table 20: The effect of ambient temperature on cycle energy and CO2

Ambient NEDC cycle NEDC CO2 Systematic NEDC cycle NEDC CO2 Systematic
Temperature Effect on energy Wh emissions g/km error energy Wh emissions g/km error
20 °C 1477 133.2 1620 132.1
14 °C Aerodynamic drag 1487 133.7 -0.35% 1630 132.6 -0.35%
14 °C Rolling resistance 1500 134.1 -0.75% 1645 133.1 -0.75%
14 °C Both 1510 134.6 -1.1% 1655 133.5 -1.1%
27
ICCT WHITE PAPER
5 CONCLUSIONS
The driving resistances of a vehicle (rolling resistance, aerodynamic drag, mass)
have a strong effect on total energy consumption. Hence, derived road loads used
for the official emission measurements on a chassis dynamometer strongly impact
the vehicles’ CO2 emissions. The official road loads of the European type-approval
emission procedure are clearly and systematically lower than for vehicles in the real
world. Consequently, the certified CO2 emission data based on the NEDC/UNECE rules
underestimate any real-world emission behavior. Even clear standards in the current
legislation do not prevent manufacturers from applying deceptive measures.
The current legislation in Europe and the associated control system by technical
services companies is not suitable to achieve realistic road loads. There are incentives
for manufacturers to exploit given tolerances and undefined issues in order to minimize
official road loads and CO2 emissions. Systematic errors in the rules further contribute
to this gap. The determination of road load coefficients is a largely opaque procedure.
The technical services cannot act as an independent institution as they are directly
commissioned by the manufacturers and do not have full access to the manufacturers’
internal processing.
The official coastdown data for light-duty vehicles are part of the emission type-
approval documentation and are available from the responsible national type-approval
authorities. In practice, these data sets are not published and must be requested by
interested third parties by following a time consuming process. Despite the public
character of these data, only Germany and France are willing to provide the data. All
other national agencies contacted (Italy, Great Britain, Luxembourg, and Spain) did not
provide the data upon request. For 29 vehicles with realistic road load sets in this study,
official coastdown data was provided on only 19 vehicles (66%). The refusal of the data
release and the irregular behavior among the different EU member states reflects a
major problem with the European type-approval process.
The application of realistic road loads instead of the official EU road loads increased
total NEDC cycle energy by 15.0% and CO2 emissions by 7.2 % on average under the
NEDC driving regime. Car road loads under realistic conditions were higher than the
EU data for all 19 vehicles, with corresponding CO2 individual increases between 0.7%
and 14.5%. A biased reference ambient temperature during the coastdown test (20 °C
instead of 14 °C) and additional coastdown tolerances on the chassis dynamometer
further contribute to the overall CO2 gap by 2.3%. Altogether, assuming an overall
divergence between official and real-world CO2 emissions of 25% in 2010 (Tietge et al.,
2015), more than one third of this gap can be explained by exploited tolerances and
errors of the road load procedures.
Compared to the situation in Europe, road load forces derived for the U.S. certification
for the same vehicles are higher and closer to reality. The average NEDC cycle energy
increase was only 4.2% and the CO2 increase was only 1.8% compared to the official
U.S. road load data. That is not because of a more precise and detailed methodology
to determine the road load coefficients, but because of a better enforcement system
and a higher risk of manufacturers getting caught. The danger of recalls and the
financial consequences are much more distinct in the U.S. (U.S. Environmental
Protection Agency, 2014b).
28
The replacement of the current emission legislation in the EU by the WLTP planned for
2017 will entail improvements and eliminate some of the existing methodological errors.
The manufacturers will be directly responsible for the officially declared road loads. This
could largely eliminate the current practice of using artificially modified vehicles for the
official coastdown runs. On the other hand, the WLTP will offer new methodologies for
road load determination with higher complexity. The standards foreseen for these new
methodologies are rather imprecise, which makes the whole regulation confusing.
Transparency and independent control measures will become even more important in
the future. Vehicle driving resistance data relevant for CO2 type-approval tests must
be made easily accessible to all interested parties. Coastdown times, test masses and
derived road load parameters of all type-approved light-duty vehicle versions should
be included in the CoC and summarized in a public database together with all certified
fuel consumption and emission data. Free access to the official road load forces is a
vital precondition for any independent verification measures. Furthermore, official
in-use compliance measurements must be extended by including coastdown testing,
and the results should be published in comprehensive reports. Distorted official CO2
emissions based on false assumptions regarding road load forces must be discovered
and corrected. This requires the establishment of a completely new road load validation
procedure under the WLTP regime.
29
ICCT WHITE PAPER
6 REFERENCES
Ahonen, H., Honkanen, V., Koisaari, T. (2012). An adjusted truth.Tekniikan Maailma, Issue 15.
Commission Regulation (EC) No 692/2008 of 18 July 2008 implementing and amending
Regulation (EC) No 715/2007 of the European Parliament and of the Council on
type-approval of motor vehicles with respect to emissions from light passenger
and commercial vehicles (Euro 5 and Euro 6) and on access to vehicle repair and
maintenance information. Consolidated version from 14 January 2015. Retrieved
from http://eur-lex.europa.eu/legal-content/EN/AUTO/?uri=CELEX:02008R0692-
20150204&qid=1448385832779
Council Directive 70/220/EEC of 20 March 1970 on the approximation of the laws of the
Member States on measures to be taken against air pollution by emissions from motor
vehicles. Retrieved from http://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=CONS
LEG:1970L0220:20070101:EN:PDF
Directive 2003/4/EC of the European Parliament and of the Council of 28 January 2003
on public access to environmental information and repealing Council Directive 90/313/
EEC. Retrieved from http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:
32003L0004&qid=1448387933020&from=EN
European Commission. (2015a). Draft Implementing Regulation setting out a
methodology for determining the correlation parameters necessary for reflecting the
change in the regulatory test cycle. Unpublished working paper, Version 20150702
European Commission. (2015b). 151119 - Annex XXI of Reg xxx-2015,Sub-Annex 6a:
Ambient Temperature Correction Test for the determination of CO2 emissions
under representative regional temperature conditions. Draft Regulation amending
EC/692/2008. Retrieved from https://circabc.europa.eu/w/browse/34525fd4-a226-
4371-a03a-76f3709c65d7
European Environment Agency. (2015). Monitoring of CO2 emissions from passenger
cars – Regulation 443/2009 [Provisional 2014 data]. Retrieved from http://www.eea.
europa.eu/data-and-maps/data/co2-cars-emission-8
INFRAS. (2014). Handbook Emission Factors for Road Transport (HBEFA),version 3.2.
[Database]. Retrieved from http://www.hbefa.net/e/index.html.
Kadijk, G., Ligterink, N. (2012). Road load determination of passenger cars. TNO report
2012 R10237, on behalf of the European Climate Foundation and the Dutch Ministry of
Infrastructure and the Environment.
Luz, R., Hausberger, S. (2014). PHEM – Passenger car and Heavy duty Emission Model.
User Guide for Version 11. Technical University of Graz.
Mock, P., German, J. (2015). The future of vehicle emissions testing and compliance. The
International Council on Clean Transportation. Retrieved from http://www.theicct.org/
future-of-vehicle-testing
Mock, P., Kühlwein, J., Franco, V., Bandivadekar, A., German, J. (2014). The WLTP:
How a new test procedure for cars will affect fuel consumption values in the EU.
Working paper 2014-9. The International Council on Clean Transportation. Retrieved
from http://www.theicct.org/wltp-how-new-test-procedure-cars-will-affect-fuel-
consumption-values-eu
30
Stewart, A., Hope-Morley, A., Mock, P., Tietge, U. (2015). Quantifying the impact of real-
world driving on total CO2 emissions from UK cars and vans. Element Energy Limited,
Cambridge, UK, and The International Council on Clean Transportation. Retrieved
from https://www.theccc.org.uk/publication/impact-of-real-world-driving-emissions/
Tietge, U., Zacharof, N., Mock, P., Franco, V., German, J., Bandivadekar, A., Ligterink,
N., Lambrecht, U. (2015). From laboratory to road – A 2015 update of official and
“real-world” fuel consumption and CO2 values for passenger cars in Europe. The
International Council on Clean Transportation. Retrieved from http://theicct.org/
laboratory-road-2015-update
United Nations Economic Commission for Europe (UNECE), Economic and Social
Council. (2015a). Proposal for a new global technical regulation on the Worldwide
harmonised Light vehicles Test Procedure (WLTP), ECE/TRANS/WP.29/2014/27.
Retrieved from https://www2.unece.org/wiki/display/trans/WLTP+12th+session
United Nations Economic Commission for Europe (UNECE), Economic and Social
Council. (2015b). Regulation No. 83 - Uniform provisions concerning the approval
of vehicles with regard to the emission of pollutants according to engine fuel
requirements, Revision 5, 22 January 2015. Retrieved from http://www.unece.org/
fileadmin/DAM/trans/main/wp29/wp29regs/R083r5e.pdf
U.S. Environmental Protection Agency. (2014a) Hyundai and Kia Clean Air Act
Settlement. Retrieved from http://www.epa.gov/enforcement/hyundai-and-kia-clean-
air-act-settlement
U.S. Environmental Protection Agency. (2014b). Test Car List Data Files, Model Years
2005 – 2015. Retrieved from http://www.epa.gov/otaq/tcldata.htm
31
ICCT WHITE PAPER
APPENDIX A ADMINISTRATIVE AND TECHNICAL VEHICLE DATA

Table 21. Type-approval CO2 emissions and certification numbers
TA CO2
Vehicle ID Vehicle Model g/km Type Variant Version TAN Emission certificate
TNO01 VW Passat 116 3C ACCAYCX0 FM6FM62S027STP17MQSNVR2O e1*2001/116*0307*32 e1*715/2007*00189*02
TNO02 Peugeot 207 110 W***** WE9HP* WE9HP0/1 e2*2001/116*0340*23 e2*715/2007*692/2008A*9090*04
TNO03 Fiat 500 110 312 AXA11 03C e3*2007/46*0064*00
TNO04 Mercedes E350 Bluetec 184 212 J0BAP0 NZAAB501 e1*2001/116*0501*11 e1*715/2007*692/2008N*0071*02
TNO05 Renault Scenic 135 JZ JZ1G JZ1GA6 e2*2001/116*0379*09 e2*715/2007*692/2008A*9096*04
TNO06 Peugeot 508 144 8 D5FV 8 e2*2007/46*0080*07 e2*715/2007*692/2008A*10218*02
TUG01 Peugeot 407 SW 150 6*RHF* 6ERHF* 6ERHF8 e2*2001/116*0369*08 e2*715/2007*692/2008A*8174*05
TUG02 Opel Astra EU5 1.7CDTI 125 P-J BF11 1A06AAAVFFE5 e1*2007/46*0141*00 e1*715/2007*692/2008A*0110*00
TUG05 Audi A3 Sportback 116 8P SCAYBF1 FM5A4051R8P607MGEM1 e1*2001/116*0217*30 e13*715/2007*692/2008A*0043*05
TUG07 Mitsubishi ASX 4WD 150 GA0 GA721 ABAAA6A5AAAC e1*2007/46*0368*02 e1*715/2007*692/2008A*0160*01
TUG08 Audi A5 2.0l TFSI 152 B8 SCDNCF1 FM6B1004R8S607MGEM1 e1*2001/116*0430*18 e13*715/2007*692/2008F*0011*08
VTT01 Audi A5 1.8 TFSI 134 B8 CCJEBF1 FM6B1004RB8T1S47MGEM1 e1*2001/116*0430*24 e13*715/2007*566/2011F*6353*01
VTT02 BMW 116i 132 1K4 1A11 H2 e1*2007/46*0283*04
VTT04 Dacia Sandero 1.6 Hi-Flex 153 SD BSDBS BSDBSP e2*01/116*0314*45 e2*715/2007*692/2008A*10187*00
Opel Insignia Sports Tourer 2.0
VTT08 119 0G-A FM11 4AALB2BVGKK5 e1*2007/46*0374*05 e1*715/2007*566/2011J*0617*00
CDTI EcoFlex
VTT09 99 1Z AACAYCX01 GFM5FM5A40540 e11*2001/116*0230*40
GreenLine
LAT01 Opel Astra EU5 1.3 D 104 P-J CACBC11 BA1J1EGTA5 e1*2007/46*0141*12 e1*715/2007*00189*02
LAT02 VW Polo 1.2 TSI 119 6R ABCBZC FM5FM52T106LLNVR07MQ e1*2001/116*0510*11 e1*715/2007*00189*02
LAT03 BMW X1 sDrive20d 119 X1 VZ91 5H0VC e1*2007/46*0275*06
32
Table 22. Technical specifications – engine and masses
Engine capacity Rated engine Mass in running Official test

Vehicle ID Vehicle Model Build year Euro class Fuel ccm power kW order kg inertia kg
TNO01 VW Passat 2012 5 diesel 1598 77 1543 1810
TNO02 Peugeot 207 2012 5 diesel 1560 68 1275 1250
TNO03 Fiat 500 2009 5 gasoline 1242 51 975
TNO04 Mercedes E350 Bluetec 2009 6 diesel 2987 155 1845 1930
TNO05 Renault Scenic 2010 5 diesel 1461 81 1485 1470
TNO06 Peugeot 508 2012 5 gasoline 1598 115 1475 1470
TUG01 Peugeot 407 SW 2010 5 diesel 1997 103 1612 1590
TUG02 Opel Astra EU5 1.7CDTI 2010 5 diesel 1686 92 1503 1470
TUG05 Audi A3 Sportback 2010 5 diesel 1598 66 1395 1590
TUG07 Mitsubishi ASX 4WD 2011 5 diesel 1798 110 1600 1700
TUG08 Audi A5 2.0l TFSI 2011 5 gasoline 1984 155 1580 1700
VTT01 Audi A5 1.8 TFSI 2012 5 gasoline 1798 125 1500 1700
VTT02 BMW 116i 2011 5 gasoline 1598 100 1365
VTT04 Dacia Sandero 1.6 Hi-Flex 2011 5 gasoline 1598 77 1157 1130
VTT08 Opel Insignia Sports Tourer 2.0 CDTI EcoFlex 2011 5 diesel 1956 118 1788 1700
VTT09 Škoda Octavia HB 1.6 TDI GreenLine 2011 5 diesel 1598 77 1390
LAT01 Opel Astra EU5 1.3 D 2012 5 diesel 1248 70 1393 1360
LAT02 VW Polo 1.2 TSI 2012 5 gasoline 1197 66 1102 1250
LAT03 BMW X1 sDrive20d 2012 5 diesel 1995 120 1565
33
ICCT WHITE PAPER
Table 23. Technical specifications – transmission
Transmission ratios
Gear
Vehicle ID Vehicle Model Type of gearbox 1 2 3 4 5 6 7 Differential

TNO01 VW Passat M6f 4.111 2.118 1.360 0.971 0.733 0.592 3.389
TNO02 Peugeot 207 M5f 3.455 1.867 1.156 0.822 0.660 3.588
TNO03 Fiat 500 A5f 3.909 2.158 1.480 1.121 0.897 3.438
TNO04 Mercedes E350 Bluetec A7q 4.380 2.860 1.920 1.370 1.000 0.820 0.730 2.470
TNO05 Renault Scenic M6f 3.727 1.947 1.323 0.975 0.763 0.638 4.125
TNO06 Peugeot 508 M6f 3.538 1.920 1.323 0.975 0.761 0.646 4.176
TUG01 Peugeot 407 SW M6f 3.417 1.783 1.121 0.795 0.647 0.534 4.176
TUG02 Opel Astra EU5 1.7CDTI Mf M32-6/3,65 3.820 2.050 1.300 0.960 0.740 0.610 3.650
TUG05 Audi A3 Sportback FM5A4051(LUB) 3.778 1.944 1.185 0.816 0.625 3.647
TUG07 Mitsubishi ASX 4WD M6q 3.818 2.045 1.290 0.974 0.897 0.790 4.058
TUG08 Audi A5 2.0l TFSI FM6B1004 3.778 2.050 1.321 0.970 0.811 0.692 3.304
VTT01 Audi A5 1.8 TFSI FM6B1004(MVQ) 3.778 2.050 1.321 0.970 0.811 0.692 3.304
VTT02 BMW 116i M6r 4.552 2.548 1.659 1.230 1.000 0.830 2.813
VTT04 Dacia Sandero 1.6 Hi-Flex M5f 3.727 2.048 1.393 1.029 0.795 4.214
VTT08 Opel Insignia Sports Tourer 2.0 CDTI EcoFlex Mf F40-6/3,09 4.167 2.130 1.321 0.954 0.755 0.623 3.091
VTT09 Škoda Octavia HB 1.6 TDI GreenLine M5f 3.778 1.944 1.185 0.816 0.625 3.647
LAT01 Opel Astra EU5 1.3 D M5f F17-5ER/3,94 3.909 2.136 1.323 0.892 0.674 3.940
LAT02 VW Polo 1.2 TSI M5f 3.615 1.955 1.281 0.927 0.740 3.933
LAT03 BMW X1 sDrive20d M6r 5.140 2.830 1.804 1.257 1.000 0.831 2.643
34
Table 24. Technical specifications – tires
Type-approval coastdown Real-world coastdown

Pressure
Make and model Dimensions Diameter [m] Make and model Dimensions Pressure Diameter [m]
Vehicle ID Vehicle Model front / rear
Continental Contact
TNO01 VW Passat Michelin radial 205/55R16 220/220 0.6259 205/55R16 220/220 0.6259
Premium
Continental Premium
TNO02 Peugeot 207 Michelin Energy Saver S1 185/65R15 240/240 0.6155 195/55R16 230/220 0.6149
Contact 2E
TNO03 Fiat 500 Michelin Energy Saver 185/55R15 0.5785
Continental
TNO04 Mercedes E350 Bluetec Continental Premium Contact 2 225/55R16 260/280 0.6479 245/45R17 260/270 0.6463
Contisportcontact 3
TNO05 Renault Scenic Michelin Energy Saver 195/65R15 260/260 0.6285 205/60R16 0.6464
TNO06 Peugeot 508 Michelin Energy Saver 225/60R16 250/250 0.6704 Michelin Primacy HP 215/55R17 250/240 0.6623
TUG01 Peugeot 407 SW Michelin Pilot Primacy 205/60R16 260/260 0.6464 215/55R17 0.6623
TUG02 Opel Astra EU5 1.7CDTI Michelin Energy Saver 215/60R16 270/270 0.6584 235/40R19 0.6646
TUG05 Audi A3 Sportback Michelin Primercy 205/55R16 250/240 0.6259 205/65R16 0.6259
TUG07 Mitsubishi ASX 4WD Yokohama 215/60R17 220/220 0.6838 215/65R16 0.6799
TUG08 Audi A5 2.0l TFSI Dunlop SP Sport 225/50/R17 220/210 0.6508 225/50R17 0.6508
VTT01 Audi A5 1.8 TFSI 225/50/R17 210/200 0.6508 Pirelli P7 245/40R18 0.6472
VTT02 BMW 116i 225/45R17 0.6283
VTT04 Dacia Sandero 1.6 Hi-Flex Continental EcoContact 3 185/65R15 200/220 0.6155 Continental EcoContact3 185/65R15 0.6155
VTT08 Conti ECO Contact 5 225/50R17 290/320 0.6508 225/55R17 0.6733
VTT09 Michelin Energy Saver 195/65R15 0.6285
GreenLine
LAT01 Opel Astra EU5 1.3 D Conti ECO Contact 5 215/60R16 270/270 0.6584 215/60R16 0.6584
LAT02 VW Polo 1.2 TSI Bridgestone radial 185/60R15 220/200 0.5970 185/60R15 0.5970
LAT03 BMW X1 sDrive20d 225/50R17 0.6508
35
ICCT WHITE PAPER
APPENDIX B COASTDOWN DATA AS PART OF THE EU

TYPE-APPROVAL
This appendix includes an extract of the current EU type-approval documentation
regarding the data on coastdown and road load to be provided by the manufacturer.
Legal source: EU Regulation EC/692/2008, Annex I, Appendix 3, Appendix to
information document
Information on test conditions
1 …
2 …
3 …
4 Dynamometer load setting information (repeat information for each
dynamometer test)
4.1. Vehicle bodywork type (variant/version)
4.2. Gearbox type (manual/automatic/CVT)
4.3. Fixed load curve dynamometer setting information (if used)
4.3.1. Alternative dynamometer load setting method used (yes/no)
4.3.2. Inertia mass (kg):
4.3.3. Effective power absorbed at 80 km/h including running losses of the vehicle on
the dynamometer (kW)
4.3.4. Effective power absorbed at 50 km/h h including running losses of the vehicle
on the dynamometer (kW)
4.4. Adjustable load curve dynamometer setting information (if used)
4.4.1. Coast down information from the test track.
4.4.2. Tyres make and type:
4.4.3. Tyre dimensions (front/rear):
4.4.4. Tyre pressure (front/rear) (kPa):
4.4.5. Vehicle test mass including driver (kg):
4.4.6. Road coast down data (if used)
Mean corrected coast
V (km/h) V2 (km/h) V1 (km/h) down time (s)
120
100
80
60
40
20
4.4.7. Average corrected road power (if used)

V (km/h) CPcorrected (kW)
120
100
80
60
40
20
36
APPENDIX C ROAD LOAD DETERMINATION IN NEDC/

UNECE LEGISLATION
Sources: The EU regulation EC/692/2008 corresponds to the UNECE regulation R83
(Revision 5 from 22 January 2015)
UNECE R83 Annex 4a Par. 5 and Appendix 7 describe the “Measurement of vehicle
road load”
General requirements:
»» Test track: level, maximum slope 1.5%, constant within ±0.1%
»» Wind (measured 0.7 m above road surface): average <3 m/s, peak <5 m/s,
perpendicular component <2 m/s
»» Dry road
Selected vehicle:
The variant with the least aerodynamic body and tires with highest rolling resistance
(second worst if more than three tire types) have to be chosen. The rolling resistance
characteristics of the tires fitted to production vehicles shall reflect those of the tires
used for type approval. The variant with largest heat exchanger shall be chosen.
Separate tests are foreseen for different variants regarding AT/MT transmissions and
front/rear/all-wheel permanent and all-wheel switchable drive.
The testing mass shall be the reference mass (mass in running order + 25 kg) of the
vehicle with the highest “inertia range.” Additional on-board weights are foreseen if the
tested vehicle mass (including driver) does not reach the official test mass.
»» The vehicle shall be run-in for more than 3000 km. It shall be in normal running
order and adjustment.
»» The tires shall be run-in at the same time as the vehicle or shall have a tread depth
of 50%-90%.
»» The tire pressure shall be in accordance with the manufacturer’s specifications for
the use considered.
»» A manual brake adjustment before coastdown is explicitly allowed (elimination of
parasitic drag).
»» Windows shall be closed and any covers in non-operational position.
»» The vehicle shall be tested at a normal running temperature achieved “in an
appropriate manner.”
37
ICCT WHITE PAPER
Par. 5.1 Coastdown experiment:

Annex 4a, Par. 6.2.1.1 (Load determined with vehicle road test):
Table 25. EU velocity ranges for LDV coastdown
Europe: From km/h to km/h mean km/h

Step 1 125 115 120
Step 2 105 95 100
Step 3 85 75 80
Step 4 65 55 60
Step 5 45 35 40
Step 6 25 15 20
Minimum measurement requirements: time ±0.1 s, speed ±2 km/h
Averaging:
»» Average times (t) from both driving directions
»» Statistical accuracy: repetitions until statistical security of mean t for each velocity
range <2% (90% confidence) (equivalent to a variation coefficient of mean t of
appromixately <1%)
Load calculation:
Par. 6.2.1.2 (load determined by applying the vehicle reference mass without
rotational inertia):
»» Dynamometer with fixed load curve: power @ steady 80 km/h, according to Table 3
(alternative, choice by manufacturer)
»» Dynamometer with adjustable load curve: power @ steady 20, 40, 60, 80, 100, 120 km/h
Correction of power (driving forces by velocity) to ambient reference conditions: 20 °C,
1 bar, real air density shall not deviate by more than ±7.5% from reference conditions:
Pcorrected = K• Pmeasured
RR RA ρ0
K= • ( 1 + KR • (T – T0)) + –
RT RT ρ
with:
K: Correction factor
RR: Rolling resistance
RA: Aerodynamic drag
RT: RR + RA
KR: Correction factor for ambient temperature dependency of rolling resistance
(=0.00864 /K)
T: Ambient temperature (°C)
T0: Reference ambient temperature = 20 °C
ρ: Air density
ρ0: Air density at reference conditions (20 °C, 1 bar)
38
»» Aerodynamic forces: air density correction not explicitly prescribed, but physical
principles should be definite:
T 1 bar
ρ0 = ρ • •
293.2 K p
Humidity is not taken into account.

»» Rolling resistance: Regulation assumes decreasing tires’ rolling resistance with
increasing ambient temperature and suggests constant correction factor KR of
0.00864 /K. Manufacturer specific values can be applied if approved by the
authority.
»» Shares of RR and RA necessary (in the NEDC) to apply correction: manufacturer data
or alternatively tabulated values:
RR
=a•m+b
RT
m: Vehicle mass at coastdown [kg]
v a b
km/h 1/kg —
20 7.24E-05 0.82
40 1.59E-04 0.54
60 1.96E-04 0.33
80 1.85E-04 0.23
100 1.63E-04 0.18
120 1.57E-04 0.14
On the chassis dynamometer:

Corrected power (from road coastdown) shall be reproduced by adjusting the roller
brakes. This may be done by correcting the measured coastdown times by K and
mass differences (between reference mass on road “m” and equivalent inertia on
dynamometer “I”) and reproducing them by dynamometer coastdown:
tmeasured I
tcorrected = •
K m
Tolerances of control coastdown on the chassis dynamometer– Annex 4a,

Appendix 1, 1.2.4.:
The regulation allows tolerances for matching the vehicle’s road load on the chassis
dynamometer by adjusting the speed dependent braking forces of the rollers:
“The accuracy of matching dynamometer load to road load shall be 5% at

120, 100, 80, 60, and 40 km/h and 10% at 20 km/h. Below this, dynamometer
absorption shall be positive.” This means that braking load must not be negative,
that is, rollers must not be propelled by the dynamometer motor to balance high
rolling forces of the tested vehicle.
Tabulated brake loads

As an alternative to an experimental determination, a manufacturer may also decide
to apply fixed load values as given in Table 26. Note that the predefined “road load”
coefficients “a” and “b” in the table represent the roller brake loads, not the vehicle’s
39
ICCT WHITE PAPER
road loads. The tabulated load values are normally higher than those being adjusted
to experimental (“real”) vehicle loads, resulting in higher CO2 emission test results if a
manufacturer decides to circumvent the effort of performing experimental coastdowns.
Table 26. Alternative fixed road load parameters (A factor of 1.3 has to be applied at “a” and “b” for
vehicles with permanent all-wheel drive)
Reference mass Equivalent “Road Load” Coefficients

(RW: Reference weight) inertia (i.e., roller brake loads)
kg kg a (N) b (N/kph)²
RW < 480 455 3.8 0.0261
480 < RW 540 510 4.2 0.0282
540 < RW 595 570 4.4 0.0296
595 < RW 650 625 4.6 0.0309
650 < RW 710 680 4.8 0.0323
710 < RW 765 740 5 0.0337
765 < RW 850 800 5.2 0.0351
850 < RW 965 910 5.7 0.0385
965 < RW 1080 1020 6.1 0.0412
1080 < RW 1190 1130 6.4 0.0433
1190 < RW 1305 1250 6.8 0.046
1305 < RW 1420 1360 7.1 0.0481
1420 < RW 1530 1470 7.4 0.0502
1530 < RW 1640 1590 7.6 0.0515
1640 < RW 1760 1700 7.9 0.0536
1760 < RW 1870 1810 8.2 0.0557
1870 < RW 1980 1930 8.5 0.0577
1980 < RW 2100 2040 8.7 0.0591
2100 < RW 2210 2150 8.9 0.0605
2210 < RW 2380 2270 9.1 0.0619
2380 < RW 2610 2270 9.5 0.0646
2610 < RW 2270 9.9 0.0674
Par. 5.2 Torque measurements method:

As an alternative to the time and velocity measurement approach, direct torque
measurements can be taken at the wheels at constant velocity on road and on the
chassis dynamometer. The regulation includes defaults for measurement accuracy, which
are ±1 Nm for torque and ±0.2 km/h for velocity. Measured on-road torques must be
reproduced on the dynamometer with adjusted brake forces within certain tolerances.
40
APPENDIX D ROAD LOAD DETERMINATION IN WLTP

REGULATION
Source: UNECE Global Technical Regulation (GTR) – version 22 Sept 2015
General principle: Manufacturer shall be responsible for the accuracy of the road load
coefficients. Tolerances shall not be used to underestimate them.
Alternative methodologies for road load determination:

»» Road coastdown with external anemometer (standard procedure like NEDC)
»» Road coastdown with on-board anemometry (new)
»» Road coastdown with measured torque at wheels (like NEDC)
»» Measured aerodynamic drag in wind tunnel, measured rolling resistance on chassis
roller or flat belt dynamometer (new)
»» Predefined general load values (like NEDC, but new formula)
»» Calculation approach (interpolation) from vehicle H and L (new)
Reference measurement points at velocities: 20 km/h to 130 km/h, incremental steps of
10 km/h
General requirements:
»» Wind speed average < 5 m/s (7 m/s for on-board anemometry)
»» Peak wind speed < 8m/s (10 m/s for on-board anemometry)
»» Wind component across road < 2 m/s (4 m/s for on-board anemometry)
»» Correction term if wind speed > 3 m/s that cannot be cancelled out by alternate runs
»» Atmospheric temperature: 1 °C to 40 °C (35 to 45 °C on regional level)
»» Road conditions: flat, clean, dry, free of obstacles or wind barriers, representative
texture and composition, longitudinal average slope < 1%, local slope < 1.5%, sum of
slopes of the parallel test track segments 0 – 0.1 %, camber < 1.5 %.
Vehicle selection:
»» Standard: vehicle H producing the highest cycle energy demand
»» Interpolation method: vehicle H with “preferably” highest cycle energy demand,
vehicle L with “preferably” lowest cycle energy demand. Calculation of road
loads of an individual vehicle by interpolation of H and L road loads, applying tire
rolling resistance from tire labeling (class average) and Δ(Cd x Af) from wind tunnel
measurements or from simulation approaches
»» Road load families: Includes several “interpolation families,” vehicles HR and LR with
differences of cycle energy demand between 4% and 35%. Calculation of road loads
for HR and LR similar to the interpolation method
Test vehicle conditions:

»» Run-in: between 3t and 80t km
»» Vehicle must conform to manufacturer’s production vehicle specifications,
regarding tire pressure, wheel alignment, ground clearance, height, drivetrain, wheel
bearing lubricants, brake adjustment
41
ICCT WHITE PAPER
»» All manually-operated panels and openings shall be closed

Vehicle “coastdown mode” allowed in case of “non-reproducible forces.” Shall be
engaged during both the road load determination and on the chassis dynamometer
emission testing.
Condition of tires:
»» Not older than 2 years (after production date)
»» Not specially conditioned or treated
»» Run-in > 200 km on road
»» Tread depth between 100% and 80% (plus maximum 500 km driving distance after
tread depth measurement)
»» Pressure at lower limit as specified by manufacturer. Adjustment in case that soak
and ambient temperature differ by more than 5 °C: Δpt = 8 mbar/K x (Tsoak – Tamb)
Vehicle warm-up:
»» Before warm-up: moderate braking from 80 to 20 km/h within 5 to 10 seconds. No
further manual adjustment after this braking allowed.
»» Warm-up at 118 km/h for at least 20 minutes.
Measurement requirements during coastdown:
»» Minimum frequency of time and vehicle speed 5 Hz.
»» Measurement accuracy: time ±0.01 s, speed ±0.2 km/h
Coastdown:
»» Start: maximum 60 sec at 140-145 km/h
»» Coastdown with transmission in neutral
»» No steering, no braking during coastdown
»» Runs in opposite directions, minimum of 3 times (2 repetitions)
»» Split runs possible (not complete velocity range in one run)
Averaging of times:
»» Harmonized average of times in both directions: Δtj (equivalent to arithmetic
average of forces)
»» Relative statistical precision: ≤ 0.03 (95% confidence)
Force calculation:
1 2 • Δv
Fj = • (mav+ mr) •
3.6 Δtj
mr: Equivalent effective mass of all rotating components, measured or calculated or

estimated by: = 0.03 x RM (mass iro + 25 kg)
mav: Vehicle test mass
Δv: 5 km/h
Δtj: Coastdown time at velocity interval j
Road load coefficients f0, f1, f2 to be calculated from a least squares regression analysis.
42
On-board anemometry (alternative to on-board vehicle velocity plus external

wind measurements):
»» Wind speed measurements: ≥ 1 Hz resolution, ≤ 0.3 m/s accuracy, direction: ≤ 1°
resolution, ≤ 3° accuracy
»» Installation of on-board anemometer: a) boom 2 meters in front of the vehicle, b)
roof at vehicle’s centerline, 30 cm from windshield or c) on engine compartment
cover, midpoint position between front and windshield
»» Aerodynamic drag correction coefficient to be determined in a wind tunnel
(installed anemometer worsens vehicle’s aerodynamics)
»» Rolling resistance (mechanical drag) to be approximated by a three-term
polynomial as a function of velocity
»» Aerodynamic drag to be approximated by a five-term polynomial as a function of
“yaw angle”
»» Extra calibration coefficients for “vehicle blockage”
»» Requires a linear least squares regression analysis with five fixed (calculated)
and nine free parameters (constrained analysis possible if Cd and AF have been
previously determined.)
Torque meter method:

»» Measured torque on driven wheels for ≥ 5 seconds at each reference velocity point
»» Sampling frequency ≥ 10 Hz. Accuracy: ≤ ±6 Nm or ±0.5% of the maximum
measured total torque (whichever is greater)
»» Includes correction procedure for “drifting velocities”
Correction to reference conditions:
»» Correction factor for air resistance:
(Reference conditions: 1 bar, 20 °C)
T 100 kPa
K2 = •
293 K p
»» Correction for the temperature dependency of tires’ rolling resistance:
K0 = 8.6 • 10-3 / K
»» Correction of wind resistance (in case that wind speed alongside the test road
cannot be cancelled out by test runs in opposite directions):
w1 = 3.62 • f2 • vw2
(vw is the “lower average wind speed” of opposite directions alongside the test road)
»» Correction of test masses:

TM
K1 = f0 • (1 – )
mav
K1 = f0 x (1 – TM/mav)
TM: test mass on chassis dynamometer
mav: test mass during coastdown
(Mass correction for Rolling Resistance K1 only applied to f0 (not to f1))
43
ICCT WHITE PAPER
»» Total correction formula:

F* = ((f0 − w1 − K1) + f1 • v) • (1 + K0 • (T − 20 °C)) + K2 • f2 • v2
»» Correction of target coefficients:

At = (f0 − w1 − K1) • (1 + K0 • (T − 20 °C ))
Bt = f1 • (1 + K0 • (T − 20 °C ))
Ct = K2 • f2
Predefined general load values - Calculation method (alternative to measurements)

f0 = 0.14 x TM [N]
f1 = 0
f2 = 2.8 x 10-6 x TM + 0.017 x width [m] x height [m] [N*(h/km)²]
TM: Test mass [kg]
(Width and height of vehicle needed, included in CoC)
Wind tunnel method

Sum of road load forces measured
a) in a wind tunnel (aerodynamic drag), repeatability ≤ 0.015 m²
and
b) on a flat belt dynamometer or on a roller chassis dynamometer (rolling resistance and
drivetrain losses)
»» Eligibility of the facilities used for this method to be demonstrated by comparisons

with coastdown tests (maximal 2% average deviation).
»» Dynamometer measurements can be performed alternatively at constant speeds
(forces measured) or by coastdown (times measured), tolerance at each speed
reference point 10 N.
»» Chassis dynamometer method includes correction factor for higher rolling
resistance on a roller, depending on wheel diameter, which reduces measured forces
by app. 5%
Dynamometer load settings:

Three different methods for dynamometer coastdown and load adjustment are provided:
A) Vehicle is accelerated under its own power:
A1) Fixed run method:

»» Calculation of vehicle force (“measured road load”) at each reference velocity
»» Calculation of simulated road load coefficients (least squares approach)
»» Calculation of the tested vehicle resistance coefficients (simulated road load
coefficients minus dynamometer setting coefficients)
»» (Procedure applied 3x)
»» Calculation of final dynamometer setting coefficients (target road load
parameters minus averaged tested vehicle resistance coefficients)
44
A2) Iterative method
“Calculated forces” in the specified speed ranges shall be within a tolerance

of ±10 N after a least squares regression of the forces for two consecutive
coastdowns.
(Method currently unclear, missing definition of “calculated forces,” adaptation of

roller brake forces between the two consecutive coastdowns?)
B) V
ehicle is accelerated by the dynamometer:
(Method currently unclear, wrong reference in actual GTR version)
A similar approach is foreseen for the torque meter method.
Accuracy of dynamometer force transducer ±10 N or 2%, whichever is greater, measured

during unloaded coastdown (without vehicle).
45
ICCT WHITE PAPER
APPENDIX E AVERAGE GASOLINE AND DIESEL CAR

SPECIFICATIONS
The table includes the technical specifications of average Euro 5 gasoline and diesel
cars as applied for the PHEM model run CO2 calculations in this study. The specifications
correspond to Euro 5 car definitions of the current version of the Handbook Emission
Factors for Road Transport (INFRAS, 2014).
Average Euro 5 cars Parameter Unit Gasoline Diesel

Mass kg 1222 1547
Loading kg 50 50
Cd - 0.31 0.31
Frontal area m² 2.14 2.27
Inertia engine g*m² 0.4506 0.5458
Mass of wheels kg 40.06 42.4578
Inertia gearbox kg*m² 0.0576 0.06134
Engine rated power kW 80 97
Engine rated speed 1/min 5247 4014
Idle speed 1/min 706 798
Rolling resistance fR0 - 0.00900 0.00900
Rolling resistance fR1 s/m 0.00005 0.00005
Rolling resistance fR4 s /m
4 4
1.600E-09 1.600E-09
Transmission losses factor - 0.3 0.3
Axle ratio - 3.876 3.678
Wheel diameter m 0.607 0.6433
Transmission 1. gear - 3.672 3.798
Transmission 6. gear - - 0.61
46

ICCT Coastdowns-EU 201605

Uploaded by

Copyright:

Available Formats

ICCT Coastdowns-EU 201605

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

ICCT Coastdowns-EU 201605

Uploaded by

Copyright:

Available Formats

WHITE PAPER MAY 2016

THE IMPACT OF OFFICIAL VERSUS

BE I J I N G | BERLIN | B R USS E LS | SAN FRANCIS CO | WAS H INGTO N

International Council on Clean Transportation Europe

+49 (30) 847129-102

communications@theicct.org | www.theicct.org | @TheICCT

© 2016 International Council on Clean Transportation

0% 2% 4% 6% 8% 10% 12% 14%

CO2 increase with real-world road loads

The replacement of the current emission legislation in the EU by the Worldwide

2 Evaluated coastdown runs from independent labs.......................................................... 8

3 Request of official coastdown data.................................................................................. 12

4 Comparisons between official and real-world road loads............................................ 15

Appendix A Administrative and technical vehicle data..................................................... 32

Appendix B Coastdown data as part of the EU type-approval......................................... 36

Appendix C Road load determination in NEDC/UNECE legislation................................ 37

Appendix D Road load determination in WLTP regulation............................................... 41

Appendix E Average gasoline and diesel car specifications............................................. 46

PHEM Passenger Car and Heavy Duty Emission Model

1.1 PHYSICAL PRINCIPLES OF DRIVING RESISTANCES

Figure 1. Forces affecting a moving vehicle.

FRR = mV * g * fRR * cos(α)

FAcc = (mV + IRP) * a

OUTSIDE TEST TRACK: INSIDE LABORATORY:

Coastdown Test Emissions and

Recorded in Type Recorded in the

For the experimental determination of rolling resistance and aerodynamic drag a

- FAcc = FRR + FAD

- (mV+IRP) * a = mV * g * fRR + Cd * AF * ρAir / 2 * v²

Because of the velocity dependency of the rolling resistance coefficient (fRR), it is

Table 1. Road load parameters in EU and U.S.

»» Rolling resistance coefficient, in kg/t

1.3 COASTDOWN RUNS AND ROAD LOAD DETERMINATION – LEGAL

1.3.1 New European Driving Cycle (NEDC) / United Nations Economic

Furthermore, the legislation includes systematic errors and outdated assumptions

»» Road inclination of the test track

1.3.2 World Harmonized Light Vehicles Test Procedure (WLTP)

»» Road coastdown with on-board anemometry

2 EVALUATED COASTDOWN RUNS FROM

»» Light-duty vehicle (car or van)

Table 2. Available real-world road load data sets from TNO

Engine Engine Mass in

Table 3: Available real-world road load data sets from TUG

Engine Engine Mass in

Table 4. Available real-world road load data sets from VTT

Engine Engine Mass in

Table 5. Available real-world road load data sets from LAT

Engine Engine Mass in

2.5 LAB COMPARISONS

Table 6. Lab methodologies and data corrections

Averaging of coastdown runs Rotating inertias taken into

3 REQUEST OF OFFICIAL COASTDOWN DATA

3.1 THE DATA SITUATION IN EUROPE

The European type-approval scheme differentiates between a “whole vehicle”

3.2 RESPONSES FROM NATIONAL TYPE-APPROVAL AUTHORITIES

Table 7. Requested type-approval authorities and their reactions

Number of evaluated vehicles

The German Kraftfahrtbundesamt (KBA) approved a formal application, following

Table 8. Availability of EU official coastdown/road load data

VTT01 Audi A5 1.8 TFSI YES KBA

TNO02 Peugeot 207 110 W***** WE9HP* WE9HP0/1 e22001/116034023 e2715/2007692/2008A9090*04

TNO03 Fiat 500 110 312 AXA11 03C e32007/460064*00

TNO05 Renault Scenic 135 JZ JZ1G JZ1GA6 e22001/116037909 e2715/2007692/2008A9096*04

TNO06 Peugeot 508 144 8 D5FV 8 e22007/46008007 e2715/2007692/2008A10218*02

TUG01 Peugeot 407 SW 150 6RHF 6ERHF* 6ERHF8 e22001/116036908 e2715/2007692/2008A8174*05

TUG05 Audi A3 Sportback 116 8P SCAYBF1 FM5A4051R8P607MGEM1 e12001/116021730 e13715/2007692/2008A0043*05

TUG08 Audi A5 2.0l TFSI 152 B8 SCDNCF1 FM6B1004R8S607MGEM1 e12001/116043018 e13715/2007692/2008F0011*08

VTT01 Audi A5 1.8 TFSI 134 B8 CCJEBF1 FM6B1004RB8T1S47MGEM1 e12001/116043024 e13715/2007566/2011F6353*01

VTT02 BMW 116i 132 1K4 1A11 H2 e12007/460283*04

LAT02 VW Polo 1.2 TSI 119 6R ABCBZC FM5FM52T106LLNVR07MQ e12001/116051011 e1715/20070018902

LAT03 BMW X1 sDrive20d 119 X1 VZ91 5H0VC e12007/460275*06