Jrc104993 Pilot Phase Final Report

Assessment of the monitoring
methodology for CO₂ emissions

from heavy duty vehicles
Pilot phase test-campaign
report and analysis of the
ex-post verification options
Theodoros Grigoratos, Georgios Fontaras,

Barouch Giechaskiel and Biagio Ciuffo
2017
EUR 28409 EN
This publication is a Science for Policy report by the Joint Research Centre (JRC), the European Commission’s
science and knowledge service. It aims to provide evidence-based scientific support to the European
policymaking process. The scientific output expressed does not imply a policy position of the European
Commission. Neither the European Commission nor any person acting on behalf of the Commission is
responsible for the use that might be made of this publication.
Contact information
Name: Georgios Fontaras & Theodoros Grigoratos
Address: European Commission Joint Research Center, Via E. Fermi 2749, 21027, Ispra (VA), Italy
Email: [email protected] & [email protected]
JRC Science Hub

https://ec.europa.eu/jrc
JRC104993
EUR 28409 EN
PDF ISBN 978-92-79-64966-0 ISSN 1831-9424 doi:10.2760/081827
Luxembourg: Publications Office of the European Union, 2017
© European Union, 2017
The reuse of the document is authorised, provided the source is acknowledged and the original meaning or
message of the texts are not distorted. The European Commission shall not be held liable for any consequences
stemming from the reuse.
How to cite this report: Grigoratos T., Fontaras G., Giechaskiel B., Ciuffo B., Assessment of the monitoring
methodology for CO₂ emissions from heavy duty vehicles - Pilot phase test-campaign report and analysis of the
ex-post verification options, EUR 28409 EN, doi:10.2760/081827
All images © European Union 2017
Title: Assessment of the Monitoring Methodology for CO2 Emissions from Heavy Duty Vehicles
Abstract
Following a request from DG-Clima and DG-GROW, JRC launched a test-campaign in order to investigate the
validity, accuracy and plausibility of the methodology proposed for the verification of the certified CO2 emissions
from Heavy Duty Vehicles (aka ex-post verification methodology). In addition scope of the test campaign was
to demonstrate the representativeness of the CO2 emissions calculations made by the official simulator (VECTO)
by comparing against the actual performance of vehicles. Experiments were conducted on four Euro VI trucks,
both on the chassis dyno and on the road with the aim of understanding the advantages and disadvantages of
different approaches proposed. Two main verification approaches were investigated, steady state
measurements in chassis-dyno / controlled conditions, and measurements under transient conditions on
chassis-dyno and actual on-road operating conditions. The official simulation software (VECTO) was used for
simulating the operation of vehicles under the different test conditions. The key conclusion of the test campaign
is that an ex-post verification method which is based on transient, on-road tests is possible for trucks and
comes with the advantage that it could potentially cover also other vehicle types which are difficult to be
validated under steady state conditions in a laboratory or on a test track under controlled conditions. However,
there is a clear need to work on the details of the test protocol to be finally implemented, define boundary
conditions for transient tests on road, and establish the necessary acceptance and rejection margins for any
such validation. Finally, additional testing is necessary in order to calculate accurately any systematic deviation
between the officially reported, simulated, CO2 values and those actually occurring in reality. VECTO results
should be periodically controlled and assessed in order to make sure that its CO2 estimates remain
representative and minimize the possibility that discrepancies will occur in the future between the officially
reported and actually experienced fuel consumption.
Contents
Acknowledgements .............................................................................................. 1
Executive Summary ............................................................................................. 2
1 Introduction ................................................................................................... 7
2 Experimental methods ..................................................................................... 9
2.1 VELA 7 facilities and setup .......................................................................... 9
2.2 Test vehicles ............................................................................................11
2.3 Daily test protocol and test cycles ...............................................................12
2.4 Vehicle simulator ......................................................................................16
3 Results and discussion ....................................................................................19
3.1 Chassis dynamometer measurements ..........................................................19
3.1.1 Vehicle #1 .......................................................................................19
3.1.1.1 Steady State Tests ...........................................................................19
3.1.1.2 Transient Tests ................................................................................23
3.1.2 Vehicle #2 .......................................................................................24
3.1.2.2 Transient Tests ................................................................................28
3.1.3 Vehicle #3 .......................................................................................30
3.1.3.2 Transient Tests ................................................................................34
3.1.4 Vehicle #4 .......................................................................................35
3.1.4.2 Transient Tests ................................................................................39
3.2 On-road measurements .............................................................................41
3.2.1 Vehicle #1 .......................................................................................41
3.2.2 Vehicle #2 .......................................................................................44
3.2.3 Vehicle #3 .......................................................................................47
3.2.4 Vehicle #4 .......................................................................................50
4 Conclusions ...................................................................................................54
References ........................................................................................................57
List of abbreviations and definitions ......................................................................58
List of figures ....................................................................................................59
List of tables......................................................................................................60
i
Acknowledgements
The authors would like to acknowledge ACEA for providing the vehicles and in particular
the following companies that supported the tests also by running VECTO simulations (in
alphabetical order):
 DAF (Henk Voets, Bart Lipsch)
 Daimler (Michael Berner, Ralf Krukenberg)
 Scania (Henrik Wentzel)
 Volvo (Jan Melin)
We are gratefully thankful to the technical staff of VELA laboratories for the execution of
the tests in such a short period of time and in particular to: Mauro CADARIO, Rinaldo
COLOMBO, Alessandro ZAPPIA and Andrea BONAMIN.
1
Executive Summary
Policy context
The European Commission has been working on the preparation of a new regulatory
initiative for monitoring CO2 emissions and fuel consumption from Heavy Duty Vehicles
(HDV) in Europe. The new methodology is based on a combination of component testing
and computer simulation of the vehicles' fuel consumption. Dedicated software simulator
has been developed for this purpose (Vehicle Energy Consumption calculation Tool –
VECTO). In parallel a series of new component testing protocols and methods were also
developed in order to measure vehicle components and provide the necessary input data
for running VECTO. As a result the final vehicle CO2 emissions are calculated based on
data received from components testing and computer simulations without the physical
need of the complete vehicle. This method allows specific CO2 emission values to be
attributed to each vehicle, provides the necessary flexibility to the vehicle manufacturers
as the HDV market is highly differentiated with limited common features between
different vehicle models and reduces the costs of vehicle certification. However, some
form of verification of the final CO2 result and the quality of input data used in the
simulations was deemed necessary by various stakeholders and European Member States
who requested the development of an appropriate verification procedure to be applied
randomly on complete vehicles after the certification processes has taken place (ex-post
verification).
Initially two possible verification approaches were proposed, one foreseeing steady state
(SS) tests under controlled conditions (chassis dyno or test-track testing) and a second
foreseeing transient testing under on-road conditions similar to – but not the same as –
the in-service-conformity testing. Following a request from DG-CLIMA and DG-GROW,
JRC launched a test-campaign in order to investigate the validity, accuracy and
plausibility of each one of the two methodologies. In addition JRC was ask to produce
data that demonstrated the representativeness of VECTO’s fuel consumption calculations
by comparing simulation results against the measured fuel consumption of the vehicles.
Experiments were conducted on four Euro VI trucks, both on the chassis dyno and on the
road with the aim of understanding the advantages and disadvantages of the different
approaches proposed.
Key conclusions
The key conclusion of the test campaign is that an ex-post verification method that is
based on transient, on-road tests is plausible for trucks and comes with the advantage
that it could potentially cover also other vehicle types that are difficult, if not impossible,
to be measured under steady state conditions in a laboratory or on a test track. The
steady state option presented other disadvantages such as higher costs, difficulty to be
reproduced without the involvement of the vehicle manufacturer and in certain situation
lower stability and repeatability. In order to introduce a transient ex-post verification
method in the HDV certification scheme there is a clear need to work on the details of the
respective test protocol, define clear boundary conditions for the tests and establish the
necessary acceptance and rejection margins for any such validation. With regards to
VECTO’s stability and capacity to produce realistic results, no new major issues were
identified and VECTO performed within the expected, previously reported margins. A
beta-version of the tool was used in the study so an additional assessment is advisable
2
once the official release of the tool is made. Finally, additional testing is necessary in
order to calculate accurately any systematic deviation between the officially reported,
simulated, CO2 values and those actually occurring in reality. VECTO results should be
periodically controlled and assessed in order to make sure that its CO2 estimates remain
representative and minimize the possibility that discrepancies will occur in the future
between the officially reported and actually experienced fuel consumption.
Quick guide / Experimental

Tests were conducted at the facilities of JRC (VELA 7). Four vehicles were tested (all of
them N3 category) equipped with state of the art exhaust after treatment systems like
Diesel Particulate filters (DPF) and Selective Catalyst Reduction of NOx (SCR). All tested
vehicles were Euro VI certified. The vehicles were tested in the laboratory over the ACEA
Regional Delivery cycle (RD) and the World Harmonized Vehicle Cycle (WHVC).
Furthermore, steady state points at various load and rpm pairs were measured according
to the provisional SS Cycle (SiCO test). Finally, on-road tests were performed over a 200
km route which included distinct urban, rural, and motorway parts.
An AVL i60 AMA 4000 system was used for the analysis of pollutants emissions. A Heated
NDIR (Non-Dispersive Infrared sensor) was used for CO2 emissions measurement and
the subsequent calculation of the fuel consumption. For the chassis dyno tests the
climatic room was adjusted at the temperature of 20°C. CO2 was measured downstream
of the exhaust after-treatment system of the truck. The calculation of the engine work
output over each sub-cycle was based on the instantaneous engine torque and rpm
values which were recorded via the vehicle's OBD (On-Board Diagnostics) system. The
calculation of the wheel work output (or cardan shaft work output) over each sub-cycle
was based on the instantaneous wheel torque values (or cardan shaft torque values)
measured by a Kistler wheel-rim torque measurement system, or in the case of one of
the vehicles, with a cardan-shaft mounted torque meter. Further to the standard
instantaneous CO2 measurement, instantaneous fuel consumption was measured also
with an AVL KMA Mobile fuel flow meter for crosschecking purposes. However, this
method was not always applied due to certain limitations imposed by the fueling system
of two of the tested trucks.
At least five repetitions of each cycle were conducted, of which at least three were
considered for a robust statistical analysis of the results. Tests were performed during
the same day but also over different days. Tests were performed always under warm
start conditions under the following sequence: WHVC, RD, and SiCO points. WHVC and
RD were tested in order to evaluate VECTO’s capability to reproduce also transient cycles
in the lab. At least three on-road tests were performed for all vehicles.
All FC values provided in the report are normalized to the average FC of each
vehicle separately. Thus, normalized FC values of different vehicles cannot be
compared to each other by any means.
Results Analysis - Main findings

The ex-post verification exercise was conducted in two phases:
 The experimental phase which took place between February and July 2016 in JRC
and involved testing of four Euro VI trucks in the laboratory and on-road.
3
 The simulations phase, which took place between May and November 2016,
during which simulations were performed by each individual OEM following the
guidelines of the JRC.
Four major OEMs participated at the exercise (in alphabetical order DAF, Daimler, Scania,
and Volvo) by providing a truck along with the necessary technical support. The
manufacturers performed the simulations after tests had been finalized by the JRC
without knowing the fuel consumption / CO2 emissions results. Simulation results were
then communicated to the JRC who performed an independent comparison between the
results of the simulations and those of the measurements. At a final step conclusions
were communicated to the respective OEM. The findings of the main evaluation phase
can be summarized to the following:
Steady State testing in the laboratory

 Steady state testing (SiCO test) proved to be satisfactorily repeatable in the
laboratory in terms of fuel consumption, particularly when testing medium and
high load points (i.e. power at the wheel or shaft >100 kW). Low load points (i.e.
power at wheel or shaft <50 kW) proved to be less repeatable and thus more
unstable compared to medium and high load points. This observation was
confirmed by all OEMs and it needs to be considered seriously as an important
share of the fuel consumption occurs over low load points, particularly in the
cases of smaller trucks and other types of HDV operating in urban and rural
conditions. During on-road low load operation occurred during a substantial part
of the trips, therefore they should not be completely neglected when performing
CO2 emissions validations.
 A good agreement between measured and simulated fuel consumption values
over medium and high loads was observed (deviations between measured and
simulated specific fuel consumption (g Fuel/kWh) was always lower than 2%) with
two of the trucks tested. This was not the case for low load points (i.e. power at
wheel or shaft <50 kW) where in some cases the deviation between measured
and simulated specific fuel consumption reached 4%. Vehicles #3 and #4
demonstrated generally higher deviations between measured and simulated fuel
consumption values, regardless the tested load. Further investigation is required
to understand the higher deviations found for these two vehicles.
 Apart from the issue with low load points this test campaign revealed other
drawbacks related to the SiCO test methodology. First of all, it is very difficult to
cover the full engine and gearbox map with only 12 (or even 18) steady state
points. Furthermore, it is not possible with one single test protocol to cover the
full range of HD vehicles (trucks, buses, coaches, etc.). Finally, this type of testing
requires expensive and difficult to maintain equipment, both when tests are
performed in the lab (expensive chassis dyno installations) and when they have to
be performed on road (special braking trailers, dedicated testing facilities, possibly
longer duration of testing due to varying weather conditions).
Transient testing in the laboratory

 In lab tests presented very good measurement repeatability (i.e. coefficient of
variation for three fuel consumption measurements <2%) over the Regional
4
Delivery cycle. Higher deviations were observed with the WHVC cycle for almost
all vehicles (coefficient of variation close to 5%). Compared to Regional Delivery,
WHVC speed profile includes many more braking events and is characterized from
generally lower speeds and lower load points. This could explain the lower
repeatability of WHVC. Finally, specific fuel consumption was found to be higher
over the WHVC than the Regional Delivery probably due to the more transient
nature of WHVC.
 A satisfactory agreement was observed between measured and simulated fuel
consumption over the Regional Delivery tests with the deviation between tests
and simulations being generally lower than 3% and only once reaching up to 5%.
VECTO PWheel mode (hereafter mentioned as VECTO SiCO mode) provided more
precise and in several cases also more accurate results compared to the VECTO
Engineering mode (whenever both modes were examined over the RD cycle). This
was an expected finding as the SiCO mode generally exhibits lower uncertainties,
the origin of which relates to the uncertainties in the estimation of the vehicle’s
road loads (air drag and rolling resistance) under different operating conditions, a
factor which is more pronounced in real world driving but is also present during
chassis dyno testing. WHVC simulations were less accurate compared to the RD
ones, regardless the vehicle tested. In general, it could be concluded that VECTO
is capable of providing reliable fuel consumption estimates for the in-lab tests,
over different transient cycles, exhibiting, however, a slightly higher uncertainty
compared to the SiCO results.
 There are various drawbacks related to transient testing method in the laboratory.
There are several difficulties for the driver to reproduce braking events over
transient cycles and specifically over a highly transient cycle such as the WHVC
(or potentially any of the lower speed cycles included in VECTO). Despite that the
operating points for the engine and the gearbox are closer to those experienced
over real world driving conditions, they do not cover the full range of the
engine/gearbox maps. In addition, some vehicles currently, and more vehicles in
the future, are equipped with sensors or GPS systems that define the operation of
certain components (e.g. gearbox) according certain external parameters (under
which conditions the vehicle operates or is expected to be operating). The effect
of such systems is totally excluded when testing on a dyno (i.e. vehicle at stand
still). Furthermore, it is not possible with one single test protocol to cover the full
range of HD vehicles (trucks, buses, coaches, etc.). There is a need for expensive
and difficult to maintain equipment (chassis dyno, special braking trailers, etc.).
Overall, it seems to be an unfavorable compromise between steady state and on-
road tests as it doesn’t solve the issues related to the other two methods.
On-road testing
 Very good repeatability was observed in the measurements of specific fuel
consumption over on-road tests, regardless the truck tested. The coefficient of
variation of specific fuel consumption measurements over three repetitions was
always lower than 1.5%. This result is somewhat surprising since these tests were
known to be more uncertain than the ones on the chassis dyno and difficult to
repeat with high precision.
5
 Overall, a good agreement between measured and simulated specific fuel
consumption values was observed over on-road tests with the deviation never
exceeding <5%. When the wheel rims were used for the measurement of the
torque at wheel the deviation between measured and simulated specific fuel
consumption did not exceed 3% for vehicles #2, #3 and #4, while in the case of
vehicle #1 it was found to be close to 5% due to a drift of the torquemeter
installed at the cardan shaft. VECTO SiCO mode proved to be more precise in
simulating measured fuel consumption values compared to Engineering mode.
Overall, it seems that VECTO is capable of providing reliable results over on-road
tests. However, differences among different VECTO modes should be further
investigated.
 On-road tests seem to be a good solution for the ex-post verification procedure as
they overcome most of the drawbacks related to the laboratory-based testing
methodologies. First of all, a wider area of the engine and gearbox maps is
investigated as the truck operates under real world conditions. A final testing
methodology could be adopted to cover even a wider range of HD Vehicle maps,
by for example introducing testing with different loadings (e.g. different vehicle
payloads). Finally, on road tests overcome the need for very expensive to
purchase and maintain equipment (chassis dyno, special braking trailers, etc.).
 Still, as in the previous options (SiCO test, In-lab transient cycle tests) it is
necessary to include in an on-road verification test, torque measurement systems
which have a certain cost. However, such systems are also used for the
measurement and definition of the vehicle air drag value (CdxA) according to the
respective test protocol.
 Regardless the testing methodology, a better agreement between measured and
simulated specific fuel consumption values was observed when using the wheel
rim torque measurement systems as opposed to the cardan shaft torque
measurement system. This observation became more obvious over on-road tests
due to their longer duration compared to laboratory tests. Whether this behavior
occurred due to the characteristics of the specific cardan shaft sensor, or could it
be a generalized behavior, remains an open point that reaches beyond the scope
of the study.
Related and future JRC work

Based on the results of the Ex-Post validation phase, the Graz University of Technology
(TUG) will conduct an error propagation analysis in order to confirm the findings and
conclusions of JRC. Afterwards, TUG may jointly proceed into drafting a first version of
the ex-post verification test protocol. JRC will support/participate in the further steps.
6
1 Introduction
The European Commission published in 2011 its “White Paper on transport” (2011) with
the aim of providing a pathway to increase the sustainability of the transport system. In
this document the European Commission suggests that reducing vehicle fuel consumption
will have a positive effect on overall CO2 emissions. However, without a robust CO2 and
fuel consumption monitoring methodology it is not feasible to achieve neither short-term
policy planning nor any additional relative initiative. A robust CO2 and fuel consumption
monitoring method should reflect the real world performance of the vehicles as well as
the comparative advantages of different vehicle models and technology packages
available in the market. This way necessary and useful information will arrive to the end
user and will allow the introduction into the market of vehicles with lower fuel
consumption. The Commission’s European Strategy for Low Emission Mobility, published
in July 2016, reiterates the importance of a low carbon transport sector and sets out an
overall vision built on three pillars: moving towards zero-emission vehicles; low emission
alternative energy for transport and efficiency of the transport system. Robust emissions
monitoring is necessary for the successful deployment of initiatives across all three
pillars. The aim of the Commission according to the European Strategy for Low Emission
Mobility is to speed up analytical work on design options for CO2 emission standards for
HDVs such as lorries, buses and coaches and is planning to launch a public consultation
to prepare the ground for a proposal during this mandate (EC, 2016).
Heavy-duty vehicle emissions are not yet monitored in a commonly agreed way in
Europe, while at the same time until recently there was no standardized and consistent
method for quantifying such emissions. The acceptance of the draft HDV CO2 certification
legislation in May 2017, initially covering Heavy Duty Trucks, and the respective
simulation-based CO2 quantification methodology, is an important first step in addressing
this issue and is expected to contribute towards lowering CO2 emissions. Still there is an
absence of consistent CO2 emissions monitoring; the EC has initiated a series of projects
with the aim of establishing a comprehensive, standardized and accurate method to
quantify and report CO2 emissions from HDVs. The issue of energy efficiency of HDVs is
important also for other policy instruments. For instance, the public procurement
legislation in Europe requires that within the criteria set for the procurement of vehicles,
energy efficiency and environmental performance specifications have to be taken into
consideration.
Initial studies and feedback received from involved stakeholders suggested that the
approach that best fits the characteristics and particularities of the HDV sector is founded
on a combination of component testing and computer simulation (AEA-Ricardo, 2013).
Similar approaches have already been adopted by the US and Japan. Measurement of
vehicles or their components is fundamental for building accurate and reliable models
and it is foreseen in all certification approaches already established. At the time of writing
of this report, vehicle simulation software (Vehicle Energy consumption Calculation Tool,
or VECTO) is being developed to be used for the purpose (Fontaras et al. 2013), while its
beta version has been tested both by the EC and individual OEMs regarding its capacity
to calculate representative CO2 emissions. In this model total fuel consumption is
simulated based on vehicle longitudinal dynamics from the input data on the vehicle and
engine characteristics. Equally important are the established test protocols for measuring
individual vehicle components and producing the required input data for running the
simulations (EC, 2017). The plausibility of such a simulation-based approach was
assessed in an extensive experimental campaign conducted by the EC's Joint Research
7
Centre (JRC). This study provided detailed experimental results for supporting the
plausibility of the simulation-based approach and its results have been described
elsewhere (Fontaras et al. 2013).
It should be noted that in the adopted simulation-based procedure, the final vehicle CO2
emissions are calculated based on data received from components testing and computer
simulations without the physical need of the complete vehicle. On one hand the method
allows specific CO2 emission values to be attributed to each vehicle, providing the
necessary flexibility to the vehicle manufacturers as the HDV market is highly
differentiated with limited common features between different vehicle models and
reducing the costs of vehicle certification. However, some form of verification of the final
CO2 result and the quality of input data used in the simulations was deemed necessary
by various stakeholders and European Member States for transparency, quality control
and trust-building reasons. As a result it was decided to develop an appropriate
verification procedure to be applied on complete vehicles after the certification process
has taken place (ex-post verification).
Initially two possible verification approaches were proposed: one foreseeing steady state
tests at controlled conditions (chassis dyno or test-track testing); the second foreseeing
transient testing under on-road conditions similar to but not the same as the in-service-
conformity testing. Following a request from DG-CLIMA and DG-GROW, JRC launched a
test-campaign in order to investigate the validity, accuracy and plausibility of each one of
the two methodologies. In addition JRC was ask to produce data that demonstrated the
representativeness of VECTO’s fuel consumption calculations by comparing simulation
results against the measured fuel consumption of the vehicles. Experiments were
conducted on four Euro VI trucks, both on the chassis dyno and on the road with the aim
of understanding the advantages and disadvantages of the different approaches
proposed.
This report summarizes the outcome of the abovementioned experimental test campaign
and attempts to provide insight with regards to the advantages and disadvantages of the
two different verification methods. In addition the data retrieved from the measurements
come to supplement those of the previous test campaigns regarding the capacity of
VECTO and the proposed approach to capture the CO2 emissions of vehicles. Experiments
were conducted on four Euro VI long haul trucks both on the chassis dyno and on the
road. The VECTO simulation tool was used for simulating the tests.
8
2 Experimental methods
Measurements took place between January and September 2016. Parts of the tests were
combined with other projects concurrently running at the JRC after prior coordination and
agreement with the respective OEMs1. In this report only the Ex-Post validation test
campaign results will be presented and discussed.
All tested HD vehicles were Euro VI. Measurements included tests on the chassis dyno at
steady state conditions and at dynamic conditions and on the road following real-world
driving patterns. Detailed descriptions of the vehicles, protocols and test conditions are
provided in this chapter.
2.1 VELA 7 facilities and setup

Chassis dyno measurements were performed at the Heavy Duty Chassis dynamometer of
the Vehicle Emissions Laboratory (VELA 7) of the European Commission’s Joint Research
Centre (JRC).
The two-roller chassis dynamometer (Zoellner GmbH, Germany) has been designed to
host even 4-wheel drive HDVs of up to 30 t in weight, 12 m in length, and 5 m in height.
HDVs of 2 axles can also be accommodated. Maximal test speed is set at 150 km/h. The
application of pull-down up to 20 kN in order to avoid slipping of the tyres is also
feasible. The test cell can be conditioned in temperatures between -30°C and +50°C and
relative humidity between 15% and 95%, providing thus the ability to test vehicles under
extreme conditions. All laboratory tests were conducted at 20°C and 50%. The constant-
volume sampler (CVS) for full exhaust dilution (AVL, Graz, Austria) is equipped with 4
Venturis of 10, 20, 40, and 80 m3/min in order to achieve a maximum air flow of 150
m3/min. Tests were usually performed with an air flow of 100 m3/min, except for some
demanding high load steady state points which were performed with maximal air flow.
Dilution air is taken from the test cell, conditioned to 22°C, and filtered through high
efficiency particulate air (HEPA) and activated charcoal filters. The climatic test cell of
VELA 7 has an air circulation system that provides enough number of cell air changes
(≥15) in order to allow the testing of vehicles regardless the fuel used. Figure 1 provides
an overview of the VELA 7 facilities.
In certain cases in addition to CO2 other gas pollutants were measured (i.e. THC, CH4,
CO, NOx). An AVL i60 AMA 4000 system was used for the analysis of gaseous emissions
in the laboratory. A Heated Flame Ionization Detector (HFID) was employed for
measuring exhaust gas concentrations of THC and CH4. A Heated Non-Dispersive Infrared
sensor (NDIR) was used for CO2 and CO emissions. Finally, a Heated ChemiLuminescence
Detector (CLD) was employed for the measurement of exhaust NOx. Pollutants were
measured downstream of the exhaust aftertreatment system of each truck. All fuel
consumption calculations in the laboratory were performed based on the CO2
measurements from the AMA analyzer unless mentioned otherwise. Figure 2 shows the
instrumentation used for the purposes of the study.
1
During the period from January to September 2016 different HD related projects were running concurrently at
VELA 7 facilities. In addition to the Ex-Post validation exercise, certain tests for the PN PEMS and the Cold
Start projects were conducted. Results from these two test campaigns have been published in another JRC
Science for Policy Report (Giechaskiel et al. 2016). Furthermore, some tests for the needs of the PMP sub23
project were performed.
9
Figure 1: VELA 7 facilities
A Gas-PEMS system was used both in the lab and on-road. The reason for using gas-
PEMS in all tests was to maintain a common reference instrument for all vehicles and all
test conditions. All other CO2 / future consumption signals were compared against this
common reference. A Semtech-DS PEMS system was used, manufactured by Sensors
Inc., and it consisted of tailpipe attachment, heated exhaust lines, an exhaust flow meter
(EFM) (4’’ or 5’’ depending on the vehicle tested), exhaust gas analysers, data logger to
vehicle network, a global positioning system (GPS), and a weather station for ambient
temperature and humidity. All data were recorded at a frequency of 1 Hz and the whole
system added further ~100 kg of instrumentation to the vehicle. An independent power
generator was used to produce current for the needs of the PEMS. The Semtech DS
measured exhaust gas concentrations of unburned hydrocarbons (THC) by HFID, carbon
monoxide (CO) and carbon dioxide (CO2) by a NDIR, and nitrogen monoxide (NO) and
nitrogen dioxide (NO2) by a non-dispersive ultraviolet sensor (NDUV). The oxides of
nitrogen (NOx) could be calculated by the sum of the concentrations of NO and NO2. The
measurement principles and accuracy from the Semtech DS were in-line to those
described by current legislation for this type of testing. As a standard procedure, test
runs preparation included routine calibration of pollutant analysers (zero and span of
gases). All fuel consumption calculations on-road were performed based on the CO2
measurements from the PEMS system while is some cases on board fuel measurement
systems were used for comparison purposes. No gaseous pollutants results will be
presented in this report as it is out of the scope of the current exercise.
Mobile fuel flow meter was not employed for measuring instantaneous fuel flow in all
tests due to functional problems. In all cases the on-board fuel flow indication provided
by the vehicles was used for recording instantaneous fuel consumption. However, fuel
consumption results were reported based on the gas PEMS system which was used with
the 5’’ Exhaust Flow Meter (EFM) for vehicles #1, #2 and #4 and with the 4’’ EFM for
vehicle #3. A validation against instantaneous CO2 data recorded during the lab tests
was performed by installing the gas PEMS system also in some laboratory
measurements. The gas PEMS system in the lab provided highly comparable readings to
the AMA system (<2% difference) especially when transient cycles were compared.
10
Figure 2: Setup of the instrumentation in the controlled temperature dilution tunnel room
2.2 Test vehicles

Four vehicles were employed for the purposes of the present study. Three long haul
vehicles and one regional vehicle were selected. Some generalized properties of the
vehicles are provided in Table 1. All vehicles were provided by the respective OEMs in
their standard operating. Figure 3 demonstrates vehicles tested in VELA 7 and on-road.
The numbering of the vehicles (i.e. #1 to #4) does not correspond in any way to the
presentation of the vehicles in Figure 3.
Table 1: Main vehicle characteristics and main input data origin
Characteristic Range of Values
Engine Displacement [cm3] 7700 – 12900
Rated Power [kW] 240 – 355
Rated Torque [Nm] 1224 – 2500
Gearbox Manual or AMT
Max load [kg] 24,000 – 40,000
Test Mass [kg] 17,000 – 27,000
Emissions Category EURO VI
Torque measurement Cardan shaft transducer or Wheel rim torquemeter
Exhaust emissions control EGR, DPF, SCR
11
During all tests signals from the vehicles On Board Diagnostic (OBD) port were recorded.
The calculation of the engine work output over each cycle was based on the
instantaneous engine torque and rpm values which were recorded via the vehicle’s ECU
(Engine Control Unit). However, this value was not used for the calculation of the specific
fuel consumption but only for cross validation purposes.
Vehicle specific fuel consumption was calculated using the loads imposed on the vehicles
during the tests. For these calculations the total work output of the driveline system
(positive or absolute) was calculated from torque measurement devices installed either at
the shaft or at the wheels. This allowed on a second step a better validation of the
resistances simulated by VECTO and an assessment of the origin of the inaccuracies in
the calculations.
Figure 3: Trucks tested in the climatic room and on-road
2.3 Daily test protocol and test cycles

The daily test protocol in the laboratory consisted of three different types of test cycles
that covered from commonly used test cycles to highly dynamic speed-versus-distance
cycles. The WHVC (World Harmonized Vehicle Cycle) has been commonly used for
emissions study of different types of HDVs, and consist of three very distinct phases, the
urban, rural and motorway. The Regional Delivery Cycle (RD) is a distance-based cycle
which is considered to provide a more realistic approach for evaluating the fuel
consumption performance of a multitude of different HDV without overlooking their
12
optimal performance conditions. Finally, constant speed tests (SiCO tests) with
predefined combinations of engine rpm and real time measured torque were executed
with the aim of investigating the CO2 emissions at various engine points.
WHVC is the equivalent of the engine World Harmonized Test Cycle (WHTC) to a vehicle
cycle for the chassis dynamometer. The duration of the WHVC test is 1800 s. The first
900 s represent urban driving with an average speed of 21.3 km/h and a maximum
speed of 66.2 km/h. This segment includes frequent starts, stops and idling. The
following 481 s represent rural driving with an average speed of 43.6 km/h and a
maximum speed of 75.9 km/h. The last 419 s are defined as highway driving with
average speed of 76.7 km/h and a maximum speed of 87.8 km/h. No slopes were
applied in all cases, thus the work of this cycle was different to the type approval work of
the WHTC. The speed versus time profile of WHVC is given in Figure 4.
Figure 4: Speed versus time profile of WHVC
Regional Delivery2 belongs to a group of distance based cycles where the actual
speed of the vehicle at any moment in time is produced by the simulator as a function
of the vehicle characteristics (weight, resistances, and available power) as well as the
modelled behaviour of the driver. Such an approach is considered much more realistic
for evaluating the performance of a multitude of different HDV without overlooking
their optimal performance conditions. The benefits of various driver aids, which are
very common in HDV applications, can also be demonstrated in the distance-based
approach. Finally, in order to match realistic conditions more closely, the RDC
features additionally the slope as a function of the traveled distance. The RDC has
been proposed to be included in the forthcoming CO2 monitoring and reporting
legislation as a representative cycle for rural delivery conditions in Europe. The speed
versus distance profile of the RDC is depicted in Figure 5.
2
The 2016 version of the regional delivery cycle was used in the study
13
Figure 5: Speed versus distance profile of Regional Delivery
Constant Speed test (SiCO test) with predefined combinations of engine rpm and real
time measured wheel (or shaft) torque (predefined test cycle in VECTO consisting of
constant speeds for optional later validation of the CO2 result produced for a HDV) were
also performed with the aim of investigating the CO2 specific emissions at various engine
load points. For each vehicle a different sequence of test points was applied depending
on its characteristics (i.e. rated power, engine displacement, etc.). At least 15 load points
were examined for each vehicle. Warm-up of the vehicle for at least 1h at high load
conditions in order to achieve a minimum of 60°C for the axle and 75°C for the
transmission oil was required. A repetition of each point is defined as a minimum of 60s
after stabilization phase. Tests are performed with all auxiliaries being switched off (i.e.
A/C switched OFF, data with Air Compressor Status ON are disregarded). A minimum of 3
repetitions of the points sequence is required for a robust statistical analysis. Figure 6
shows the applied constant speed profile of engine rpm versus torque for vehicle #2.
Figure 6: Example of a constant speed profile of engine rpm vs. torque for vehicle #2
14
Figure 7: Route for on-road tests performed. Red line - Motorway; Blue line - Urban & Rural
On-road tests around the JRC site were performed to simulate real-world emissions. A
mixed route of total distance of approximately 200 km which consists of urban, rural and
highway parts was driven (Figure 7). The highway part accounts for approximately 150
km allowing thus to simulate the real world operating conditions of long-haul trucks. In
Figure 7 the highway part has been spotted with the red line, whereas urban and rural
parts are shown with the blue line. Figure 8 shows the speed profile of vehicle #2 over
the described route.
The scope of on-road tests was to obtain a mix of operating conditions similar to those of
the chassis dynamometer tests. Also there was a need for investigating parameters such
as the repeatability of the tests and the agreement between measured and simulated
values since on-road tests were considered to be less repeatable compared to laboratory
tests.
The route's statistics with the four vehicles are summarized in Table 2. A PEMS compliant
trip would stop at around time 6000s. However, it was decided to extend the
measurement period throughout the whole trip for the purposes of this project. The
highest deviations were observed at the rural phase where in some cases vehicles
15
exceeded 70 km/h, thus attributing the part of the trip to the motorway phase. All
vehicles were tested at least three times except for vehicle #4 which was tested only
twice due to time restrictions.
Table 2: Driving phase distributions of on-road trips
Speed Classification Share in total trip duration
Low Speed <50 km/h [~Urban] 23‐27%
Medium Speed 50‐70 km/h [~Rural] 4‐15%
High Speed >70 km/h [~Motorway] 60‐69%
Figure 8: Speed profile of vehicle #2 during one typical on-road test
2.4 Vehicle simulator

The VECTO-simulator is the core component of the proposed methodology. The software
simulates CO2 emissions and fuel consumption based on vehicle longitudinal dynamics
using a driver model for simulation of target speed cycles. The load required by the
internal combustion engine is calculated internally in 1Hz steps based on the driving
resistances, the power losses in the drive train system, and the power consumption of
the vehicle auxiliary units. Engine speed is determined based on a gear-shifting model,
the gear ratios, and the wheel diameter. Fuel consumption and CO2 emissions are then
interpolated from an engine fuel/CO2 map.
Currently, in each timestamp VECTO interpolates the engine fuel consumption based on
the simulated engine speed and torque from an engine fuel map measured in steady
state conditions at the engine test bed. To overcome the shortcomings introduced by the
use of steady state fuel map in for the simulation of transient operating conditions, it is
foreseen in VECTO that a correction factor is applied. This correction factor shall be
16
determined based on the quotient of measured fuel consumption in a transient real world
cycle (most probably the WHTC) and the simulated fuel consumption for this cycle based
on the steady state engine fuel map. More details on the procedure for obtaining the map
and the correction function are provided by Luz et al. [2014].
The main characteristics of the current VECTO version can be summarized in the
following list:
 Backwards-calculating, quasi-stationary longitudinal dynamics model with pre-
and post-processing loops (e.g. for time to distance conversions, driving aids and
WHVC corrections);
 Time-based or Distance-based cycles (time-steps may have varying duration,
distance-steps must be at most 1min length);
 1 s (1 Hz) Internal and Output time-steps;
 Driving model considers real life driving behavior (e.g. acceleration and breaking
curves, gear shifting, coasting);
 Input and output via text-files;
 Implemented as Visual Basic. NET application (Windows);
 Graphical user interface for calculation control and editing of the main input files;
 Declaration mode with locked-values and cryptographic signing of results for
certification purposes.
Figure 9: VECTO’s simulation core
The simulation-core is summarized schematically in Figure 9. Additional information

about the software and its functionality can be found in Fontaras et al. (2013) and Luz et
al. (2014). A series of studies have shown that VECTO performs adequately and in a
17
similar way as other established commercial or regulation oriented simulators Franco et
al. (2015), ACEA (2013).
Table 3 briefly describes the available VECTO modes. The official (declaration) mode uses
official values and input (as in certification) for all parameters (i.e. mass, road loads,
gearbox, axle, engine) as well as constant predefined values for the auxiliaries.
Declaration mode was not examined in the current study. The engineering mode uses as
input values for mass, road loads and drive cycle those measured by the JRC.
Additionally, VECTO considers power losses for gearbox, axle, engine and some of the
auxiliaries. Engineering mode was applied in several tests and in particular to those
conducted on-road. Finally, the SiCO mode requires the measured values of the torque at
wheel or at shaft along with the measured engine RPM. All other input comes from the
official gearbox, axle and engine maps. The SiCO mode was applied in all executed tests.
VECTO simulations were all run by the respective OEM with data provided by the JRC.
Different versions of the tool were used depending on each OEM. The versions used by
each OEM are given at the respective results section.
Table 3: VECTO modes description
Mass & Gearbox &

Test Engine Auxiliaries Drive Cycle
Road loads Axle
Official values Official values Official values Official values Official values
Declaration
and input and input and input and input and input
mode
(certification) (certification) (certification) (certification) (certification)
Speed profile
Engineering JRC tests Official Official Official & JRC
& slope
mode (F0, F1, F2) Values Values test values
JRC tests
Wheel Torque,
Not Official Official Official & JRC
SiCO Mode Engine RPM
relevant Values Values test values
JRC tests
18
3 Results and discussion
3.1 Chassis dynamometer measurements

The scope of the laboratory tests was twofold, to check the repeatability of the
methodologies in the lab, under highly controlled conditions, with particular view on the
transient testing cycles (i.e. Regional Delivery) and to investigate the quality of the
simulations under different operating conditions. According to feedback received from
involved stakeholders and indications from previous experimental campaigns the steady
state tests (SiCO test) were expected to be highly repeatable in the laboratory. However,
at the moment there were few data to support the repeatability of transient cycles. On
the contrary feedback from stakeholders involved in testing suggested that on road fuel
consumption testing exhibits poor repeatability. Further to the test’s robustness the data
would be used to compare the uncertainty of the simulation runs under transient
conditions, and obtain a broader picture of VECTO’s accuracy. These aspects were
partially covered in the previous JRC study (Fontaras et al. 2016).
3.1.1 Vehicle #1
3.1.1.1 Steady State Tests

Table 4 describes the steady state points tested with vehicle #1 on the dyno. In this case
torque measurements were conducted with a torque measurement device on the cardan
shaft (Figure 10). All tests were preceded by at least 1h of vehicle warm up in order to
achieve a minimum of 60°C for the axle and 75°C for the transmission lubricant and
stabilize thermally the vehicle. Table 4 provides information regarding the vehicle’s
speed, the actual power measured at the cardan shaft, the gear engaged as well as the
engine’s and shaft’s speed for each one of the 15 tested points. A repetition of each point
is defined as a minimum of 60 s after stabilization phase. Measured shaft power and
engine speed for each point shall not deviate more than ±10 kW and ±25 rpm between
different measurements, otherwise the measurement is considered invalid.
Figure 10: Torque measurement device on the cardan shaft of vehicle #1
19
Table 4: SiCO points for vehicle #1
Vehicle’s Speed Power at Shaft Engine Speed Shaft Speed Gear

#
[km/h] [kW] [RPM] [RPM] [‐]
1 86.1 110.0 1252 1252 12
2 87.0 299.1 1249 1248 12
3 83.3 192.5 1220 1220 12
4 85.1 291.6 1222 1222 12
5 80.0 280.0 1179 1178 12
6 80.1 258.3 1178 1178 12
7 83.6 306.8 1481 1198 11
8 78.0 277.2 1150 1149 12
9 74.0 312.3 1311 1061 11
10 74.2 209.6 1315 1064 11
11 69.0 289.4 1223 990 11
12 69.8 141.8 1237 1001 11
13 67.6 103.3 1199 970 11
14 55.2 192.3 1230 792 10
15 55.2 96.3 1230 792 10
Table 5 briefly describes SiCO test results including the application of specific filters. In
all cases measurements were conducted with A/C being switched off, while data with air
compressor status “ON” were disregarded. All values given in Table 5 are averaged over
3 measurements, while ± values correspond to the standard deviation of the 3
measurements. Specific fuel consumption is normalized to the average value in g
Fuel/kWh of all 15 SiCO points. The last column shows the relative standard deviation
[%] of the three measurements of the fuel consumption (g Fuel/kWh) at the shaft
(coefficient of variation).
Despite that high load points (i.e. shaft power >200 kW) come with high values of fuel
consumption per hour (i.e. g Fuel/h), relatively low specific fuel consumption is observed.
This is seen in points 2, 4, 5, 6, 7, 8 and 11 which all demonstrate specific fuel
consumption lower than the average value of the 15 points (i.e. <1.0). On the other
hand, medium load points (i.e. shaft power of 100-200 kW) showed specific fuel
consumption higher than the average value of the 15 points (see points 1, 12, 13, 15). It
could be concluded that lower load points are linked to higher specific fuel consumption.
Unfortunately, this could not be confirmed for low load points (i.e. <50 kW) as they were
not evaluated with vehicle #1 due to stability issues. More specifically it seems to be
difficult for the dyno (and/or the analyzers) to perform repeatable runs of low load points
with high accuracy, maybe due to the fact that relatively low forces are applied and
therefore the relative error is maximized.
Coming to the repeatability of the SiCO test in the laboratory, it is seen from Table 5 that
almost all points exhibited very low coefficient of variation (<1.5%). Only in the case of
point 13 the relative standard deviation of the three measurements was higher than 2%.
Of course, as mentioned previously this does not include low load points (i.e. <50 kW)
20
which proved to be quite unstable and the measurements could not be repeated
accurately.
Table 5: SiCO dyno test results for vehicle #1
Normalized Relative
Point Engine Speed Shaft Power
Specific FC at Shaft SD
[#] [rpm] [kW]
[g/kWh] [%]
1 1251±2 109.5±0.4 1.096 1.1
2 1250±1 299.2±0.8 0.968 1.1
3 1221±1 194.5±1.7 0.989 1.3
4 1221±1 291.9±0.7 0.975 0.1
5 1180±4 281.8±1.7 0.972 1.0
6 1181±5 259.8±1.8 0.967 0.7
7 1478±5 307.0±0.3 0.989 0.4
8 1152±5 276.4±0.8 0.973 0.6
9 1316±6 312.8±0.8 0.980 0.2
10 1316±5 210.1±0.8 0.990 0.9
11 1229±7 291.3±2.0 0.974 0.6
12 1229±7 141.5±0.3 1.014 0.8
13 1194±5 101.6±2.3 1.061 2.5
14 1228±2 191.9±0.6 0.990 0.0
15 1228±2 95.8±0.4 1.061 0.2
Table 6 briefly presents the comparison between measured and simulated fuel
consumption values over the 15 SiCO points. Comparison is performed over normalized
to the average of all measurements g Fuel/h value in order to better reflect the different
FC over the whole range of points tested. Only VECTO SiCO mode was considered in this
case. VECTO simulations were performed for one of the three available sequences of
points. This practically means that Table 7 shows the direct comparison of the measured
fuel consumption over this particular series and the respective simulated values and that
no average values were considered for this comparison. The last column presents the
deviation between the measured and the simulated fuel consumption values (g
Fuel/kWh) for each point. The deviation between measured and simulated values is
better depicted in Figure 11.
Table 6 shows that there is a good agreement between measured and simulated values
for almost all examined points. High load points (i.e. shaft power >200 kW) showed
deviations between measured and simulated values lower than 2%. Some medium load
points (i.e. shaft power of 100-200 kW) showed slightly high deviations close to 3% (see
points 13, 15). Again it seems that lower load points are linked to higher uncertainties
compared to high load points. This could not be confirmed for low load points (i.e. <50
kW) as they were not evaluated with vehicle #1 due to stability issues.
Overall, it is concluded that there are no big difficulties in repeating steady state points in
the lab. There is a question regarding low load points as they seem to be more difficult to
21
be stabilized. Also, it seems from vehicle #1 that VECTO can simulate most points with a
deviation from the measured value lower than 2%. Slightly higher deviations are found
over lower load points. This shall not be neglected as important part of the fuel
consumption occurs over low load points (Grigoratos et al. 2016).
Table 6: Measured vs. simulated fuel consumption over SiCO dyno test for vehicle #1
Normalized Normalized
Point Engine Shaft Power Deviation
FC VECTO FC Measured
[#] Speed [rpm] [kW] [%]
[g Fuel/h] [g Fuel/h]
1 1252 110.0 0.634 0.644 ‐1.5
2 1249 299.1 1.531 1.556 ‐1.6
3 1220 192.5 1.013 1.034 ‐2.1
4 1222 291.6 1.501 1.525 ‐1.6
5 1179 280.0 1.447 1.476 ‐1.9
6 1178 258.3 1.330 1.349 ‐1.4
7 1481 306.8 1.629 1.622 0.5
8 1150 277.2 1.423 1.453 ‐2.0
9 1311 312.3 1.633 1.641 ‐0.5
10 1315 209.6 1.114 1.104 0.9
11 1223 289.4 1.516 1.522 ‐0.4
12 1237 141.8 0.782 0.777 0.7
13 1199 103.3 0.594 0.579 2.7
14 1230 192.3 1.024 1.019 0.5
15 1230 96.3 0.566 0.548 3.3
Figure 11: Deviation of the measured vs. the simulated FC values for all examined SS points
22
3.1.1.2 Transient Tests
Table 7 briefly describes WHVC and RD test results without the application of any filter.
In all cases measurements were conducted with A/C being switched off. All transient
tests were performed with at least 1h warming up of the vehicle, therefore all WHVCs are
hot-start. Pull down of 20 kN was applied in order to avoid slipping of tyres. All values
given in Table 8 are averaged over 3 measurements, while ± values correspond to the
standard deviation of the 3 measurements. Specific fuel consumption has been
normalized to the average value of all tested SiCO points in g Fuel/kWh. The last column
represents the relative standard deviation [%] of the three measurements of fuel
consumption (g Fuel/kWh) at the shaft (coefficient of variation).
Table 7: WHVC and Regional Delivery dyno test results for vehicle #1
Average Shaft Energy Normalized

Relative SD
Test Speed Measured Specific FC at Shaft
[%]
[km/h] [kWh] [g/kWh]
Regional
69.2±0.7 28.9±0.4 1.095 1.3
Delivery
Hot Start
40.8±0.1 22.1±0.9 1.437 3.7
WHVC
Table 7 shows that Regional Delivery cycle is highly repeatable in the laboratory. On the
other hand, WHVC exhibited a coefficient of variation for the 3 different tests of about
4%. It has to be noted that there were several difficulties for the driver to reproduce
braking events over transient cycles and mainly WHVC. Compared to RD, WHVC speed
profile includes many more braking events and is characterized from generally lower
speeds (average speed of 40.8 km/h vs. 69.2 km/h). This could explain the lower
repeatability of WHVC compared to RD. Finally, normalized specific fuel consumption is
found to be higher over the WHVC than the Regional Delivery due to the more transient
nature of WHVC.
Table 8: Measured vs. simulated FC over WHVC and Regional Delivery for vehicle #1
Average Shaft Energy Normalized Normalized

Deviation
Test Speed Measured FC VECTO FC Measured
[%]
[km/h] [kWh] [g/h] [g/h]
Regional
69.8 28.6 0.464 0.443 4.7
Delivery
Hot Start
40.9 22.0 0.311 0.331 ‐6.0
WHVC
consumption values over both transient cycles. Comparison is performed over normalized
to the average of all measurements g Fuel/h value. Only VECTO in SiCO mode was tested
for both transient cycles. Also in this case, VECTO simulations were performed for one of
the 3 available tests for each cycle. This practically means that Table 8 shows the direct
23
comparison of the measured fuel consumption over one WHVC and one RD with the
simulated values of exactly the same run of WHVC and RD. Also, since VECTO SiCO mode
uses as input the measured torque at the shaft, in this case the deviation between
measured and simulated fuel consumption could be expressed both in g Fuel/h and g
Fuel/kWh without being different. The deviation between measured and simulated values
is also given in Figure 12.
As shown in Table 8 the total fuel consumption simulated with VECTO did not match very
closely the experimentally measured for WHVC and RD cycles. The differences between
calculated and measured results were up to 6%. Over the regional delivery the difference
between measured and simulated fuel consumption was found to be at a 5% margin. The
increased fuel consumption over the simulation suggests a possible overestimation of a
particular vehicle load, possibly the consumption of some of the auxiliary systems.
Furthermore, there is an offset due to torque drift which has not been corrected and
overestimates the measured torque at the shaft. Similar deviation for the RD was
reported previously for another truck tested by Fontaras et al. (2016). In any case we
see that the accuracy of the simulations is lower compared to that of the steady state
tests.
Figure 12: Normalized simulated vs. measured (=1.0) FC of vehicle #1 for WHVC and RD
Overall, it is concluded that while regional delivery is highly repeatable in the lab, this is
not the case for WHVC. Increased number of braking events and generally lower speeds
over WHVC seem to affect the repeatability of this cycle in the dyno. VECTO SiCO mode
is able to simulate both cycles with, however, a relatively high deviation (approximately
5%) from the measured value. Slightly higher deviation is observed over WHVC.
3.1.2 Vehicle #2

Table 9 briefly presents the points tested with vehicle #2 in the dyno for the purposes of
the SiCO test. Information regarding the vehicle’s speed, the actual power measured at
24
the wheel, the gear engaged as well as the engine’s speed for each one of the 17 tested
points is given. Torque measurements were conducted with the Kistler torquemeter
device placed at the wheel rims (Figure 13). All tests were performed with at least 1h
warming up of the vehicle. A repetition of each point is defined as a minimum of 60 s
after stabilization phase. Measured wheel power and engine speed for each point shall
not deviate more than ±10 kW and ±25 rpm between different measurements.
Vehicle’s Speed Power at Wheel Engine Speed Shaft Speed Gear

#
1 86.4 163.8 1178 1177 12
2 86.5 127.9 1178 1178 12
3 86.8 75.3 1178 1177 12
4 86.9 39.9 1178 1177 12
5 83.9 281.5 1472 1151 11
6 82.1 254.8 1124 1124 12
7 84.7 143.0 1475 1153 11
8 82.2 228.6 1124 1124 12
9 82.8 90.1 1124 1124 12
10 75.4 124.0 1026 1026 12
11 75.6 80.4 1026 1026 12
12 75.8 32.2 1026 1026 12
13 67.0 226.1 1176 919 11
14 58.8 152.8 1028 804 11
15 47.3 48.9 1318 642 9
16 43.3 169.9 1222 596 9
17 75.6 260.7 1327 1037 11
Table 10 presents SiCO test results filtered with regard to data with air compressor
status “ON”. Measurements were conducted with A/C being switched off. Values in Table
10 are averaged over 3 measurements and ± values correspond to the standard
deviation of the 3 measurements. Vehicle specific fuel consumption is normalized to the
average measured value (in g Fuel/kWh) of all examined SiCO points. The last column
represents the relative standard deviation of the three measurements (g Fuel/kWh)
elsewhere referred as coefficient of variation.
Like in case of vehicle #1, high load points (i.e. wheel power >200 kW) demonstrate the
higher fuel consumption per hour but at the same time their vehicle specific fuel
consumption is relatively low. Points 5, 6, 8, 13 and 17 all demonstrate VSFC lower than
the average value of the 17 points (i.e. <1.0). On the other hand, medium to low load
points (i.e. wheel power <100 kW) exhibited fuel consumption per kWh higher or very
close to the average value of the 17 points (see points 3, 4, 9, 11, 12, 15) with low load
points (i.e. wheel power <50 kW) being the most energy consuming (see points 4, 12,
15). Once more it is demonstrated that load points are linked to higher vehicle specific
fuel consumption.
25
Figure 13: Torque measurement device at the wheels of vehicle #2
Normalized
Point Engine Speed Wheel Power Relative SD
VSFC
[#] [rpm] [kW] [%]
[g/kWh]
1 1178±0 163.8±1.2 0.933 1.2
2 1178±0 127.9±1.0 0.952 0.4
3 1178±0 75.3±0.9 1.025 2.0
4 1178±0 39.9±0.2 1.248 1.9
5 1472±0 281.5±0.2 0.935 0.1
6 1124±0 254.8±0.7 0.911 0.1
7 1475±5 143.0±0.9 1.007 0.2
8 1124±0 228.6±1.2 0.914 0.7
9 1124±0 90.1±1.3 0.971 0.3
10 1026±4 124.0±1.0 0.935 0.5
11 1026±4 80.4±1.6 0.980 0.9
12 1026±4 32.2±1.9 1.274 4.4
13 1176±5 226.1±1.5 0.921 0.7
14 1028±0 152.8±2.3 0.938 1.1
15 1318±0 48.9±1.9 1.190 2.5
16 1222±0 169.9±2.0 0.933 0.7
17 1327±0 260.7±2.9 0.935 1.1
When the repeatability of the SiCO test in the laboratory is examined, it is seen that
almost all steady state points exhibited low coefficient of variation (<2.0%). Most points
26
proved to be highly repeatable (coefficient of variation <1%) and only in the case of
points 12 and 15 the relative standard deviation of the 3 measurements was higher than
2%. Unstable points are low load points, therefore in conjunction with the observations
from vehicle #1 it can be concluded that low load points are quite unstable and not easy
to be repeated accurately.
consumption over the 17 steady state points. Comparison is performed over normalized
to the average of all measurements g Fuel/h value in order to better reflect the different
FC over the whole range of points tested. The simulations were run with VECTO SiCO
mode (3.0.3.495 version) for one of the three available sequences of points. The last
column presents the deviation between the measured and the simulated fuel
consumption values (g Fuel/h) for each point. The deviation between measured and
simulated values is also graphically given in Figure 14.
Point Engine Wheel Power Deviation
[#] Speed [rpm] [kW] [%]
1 1178 165.1 1.215 1.218 ‐0.2
2 1178 129.0 0.973 0.973 0.0
3 1178 76.4 0.628 0.609 3.1
4 1178 40.2 0.402 0.392 2.5
5 1472 281.7 2.070 2.095 ‐1.2
6 1124 254.1 1.120 1.843 ‐0.7
7 1472 142.4 1.830 1.140 ‐1.8
8 1124 227.2 1.650 1.665 ‐0.9
9 1124 88.7 0.718 0.685 4.8
10 1031 123.0 0.940 0.920 2.1
11 1031 78.7 0.646 0.618 4.5
12 1031 30.2 0.330 0.321 2.9
13 1181 224.6 1.652 1.659 ‐0.4
14 1028 150.3 1.148 1.135 1.1
15 1318 46.9 0.472 0.456 3.6
16 1222 167.6 1.266 1.254 1.0
17 1327 257.3 1.900 1.938 ‐2.0
Table 11 demonstrates that there is a good agreement between measured and simulated
values for all high load points (i.e. wheel power >200 kW). Deviations between measured
and simulated values did not exceed 2%. Also medium to high load points (i.e. wheel
power of 100-200 kW) exhibited very good repeatability close to 1% (see points 1, 8, 14,
16). Medium to low load points (i.e. wheel power of 50-100 kW) and low load points (i.e.
wheel power <50 kW) are linked to higher uncertainties compared to high load points.
The most difficult points to reproduce and therefore with the highest values of deviation
27
(>3%) proved to be these of approximately 80 kW wheel power (see points 3, 9, 11).
Similar conclusions were drawn for vehicle #1.
Overall, it is concluded that steady state points are satisfactorily repeatable in the lab
with a bigger deviation when it comes to low load points. Even in the very advanced
chassis dyno of VELA 7 low load points seem to be more difficult to be stabilized. VECTO
can simulate medium to high and high load points with a deviation from the measured
value lower than 2%. Slightly higher deviations are found over medium to low and low
load points making the method questionable at least at this range of loads. This is an
important issue related to the SiCO method as important part of fuel consumption occurs
over low load points.

Table 12 shows WHVC and Regional Delivery test results without the application of any
filter. Measurements were conducted with A/C being switched off. All transient tests were
performed with at least 1h warming up of the vehicle with an application of 20 kN pull
down to avoid slipping of tyres. Values given in Table 12 are averaged over 3
measurements performed over different days. Vehicle specific fuel consumption is
normalized to the average value in g Fuel/kWh of all SiCO points tested. The last column
represents the coefficient of variation of 3 measurements of the VSFC (g Fuel/kWh).
Table 12 shows that both transient cycles tested are quite repeatable in the laboratory.
In this case and despite the difficulty of the driver to reproduce some braking events
over WHVC, this cycle proved to be highly repeatable. WHVC was selected to perform a
cross check of the AMA performance. For that reason the AVL KMA fuel flowmeter was
connected and the values of the two instruments were compared. WHVC was selected
because it is a highly transient cycle and therefore representative of many different
operating conditions and engine loads. The difference among the two instruments was
found to range from 0.5-1.0% depending on the test. This range of deviation is
28
considered low taking into account the analyser inaccuracies. Once more, normalized fuel
consumption is higher over the WHVC than the Regional Delivery due to the more
transient nature of WHVC.
Average Wheel Energy Normalized

Relative SD
Test Speed Measured VSFC
[%]
Regional
67.8±0.3 31.6±0.3 0.979 1.4
Delivery
Hot Start
39.2±1.2 25.3±0.2 1.018 0.1
WHVC
Table 13 presents the comparison between measured and simulated fuel consumption
values over Regional Delivery. Comparison is performed over normalized to the average
of all measurements g Fuel/h value. VECTO SiCO mode in 2 different versions (with 4.51
kW additional auxiliaries consumption and without any Padd) was considered in this case.
Unfortunately, no comparison is available for the WHVC cycle. VECTO simulations for
Regional Delivery were performed for two out of the three available tests and all results
are given in Table 13. Since VECTO SiCO mode uses as input the measured torque at the
wheel also in this case the deviation between measured and simulated fuel consumption
could be expressed both in g Fuel/h and g Fuel/kWh without any difference. The
deviation between measured and simulated values is also depicted in Figure 15.
Table 13: Measured vs. simulated FC over Regional Delivery tests for vehicle #2
Wheel Normalized Normalized Deviation Deviation

Average Normalized
Energy FC VECTO FC VECTO Measured Measured vs.
Test Speed FC Measured
Measured w/o Padd with Padd vs. No Padd 4.51kW Padd
[km/h] [g/h]
[kWh] [g/h] [g/h] [%] [%]
Regional
68.1 31.2 0.641 0.668 0.650 ‐1.5 2.7
Delivery 1
Regional
68.0 31.7 0.647 0.674 0.647 0.0 4.2
Delivery 2
Regional
Delivery 68.1 31.5 0.644 0.671 0.649 ‐0.8 3.4
Averaged
As shown in Table 13 the total fuel consumption calculated by VECTO was very close to
the measured value when no Padd was applied. Actually, the deviation between measured
and simulated FC was at the same level as stable state points. On the other hand, the
total fuel consumption simulated by VECTO was found to be less precise with the 4.51kW
Padd version. The difference with the latest reached 4% and was similar to that found with
the vehicle #1 over the same cycle. Increased fuel consumption over the simulation
suggests a possible overestimation of the consumption of some of the auxiliary systems.
In any case we see that the accuracy of the simulations with vehicle #2 is similar to that
of the steady state tests.
29
Figure 15: Normalized simulated (w/o Padd blue bars – 4.51kW Padd red bars) vs. measured (=1.0 –
green bars) FC of vehicle #2 for Regional Delivery tests
Overall, it is concluded that both transient cycles are highly repeatable in the laboratory
in contrast to what was seen with vehicle #1. VECTO SiCO mode seems to be able to
simulate Regional Delivery cycle quite accurately with both modes and slightly better
when no Padd is applied. No test demonstrated deviation higher than 5%.
3.1.3 Vehicle #3

Table 14 describes the 16 steady state points tested with vehicle #3 in the dyno for the
SiCO test. Information regarding the vehicle’s speed, the actual power measured at the
wheel, the gear engaged as well as the engine’s speed for each point is given. Also in this
case torque measurements were conducted with the Kistler torquemeter device placed at
the wheel rims. Shaft speed was not recorded in this case as the respective signal from
the CAN was missing. All tests were performed with at least 1h warming up of the
vehicle. A repetition of each point is defined as a minimum of 60 s after stabilization
phase. Wheel power and engine speed for each one of the 16 points shall not deviate
more than ±10 kW and ±25 rpm between different measurements otherwise the
measurement is considered invalid.
Table 15 shows SiCO test results of vehicle #3. In all cases measurements were
conducted with A/C being switched off, while data with air compressor status “ON” were
disregarded. Values are averaged over 3 measurements and ± values correspond to the
SD of the 3 measurements. Vehicle specific fuel consumption is normalized to the
average value of all measured SiCO points in g Fuel/kWh. The last column represents the
relative standard deviation of the three measurements (g Fuel/kWh).
30
Despite the fact that vehicle #3 is different from the other vehicles with respect to its
maximum load and engine characteristics, there were no significant differences in its
behavior on the chassis dyno at least with regard to the SiCO test. High load points (i.e.
wheel power >150 kW – different for this truck compared to the other three)
demonstrated the higher fuel consumption in g Fuel/h, while at the same time
normalized vehicle specific fuel consumption is relatively low. For instance points 1, 5, 7,
10, 12, 13 and 14 all exhibited fuel consumption per kWh lower than the average value
of the 16 points (i.e. <1.0). On the other hand, low load points (i.e. wheel power close to
or <50 kW) exhibited normalized fuel consumption higher or very close to the average
value of the 16 points (see points 6, 11, 15).
Vehicle’s Speed Power at Wheel Engine Speed Shaft Speed Gear

#
1 84.7 161.1 1420 * 9
2 83.8 138.7 1403 * 9
3 84.0 103.1 1403 * 9
4 84.2 76.0 1403 * 9
5 81.8 157.1 1370 * 9
6 82.5 30.3 1371 * 9
7 77.9 150.0 1306 * 9
8 74.0 156.5 1648 * 8
9 64.5 75.0 1431 * 8
10 63.1 151.2 1409 * 8
11 63.7 55.1 1409 * 8
12 60.2 149.3 1344 * 8
13 54.1 173.0 1633 * 7
14 49.6 167.5 1497 * 7
15 48.0 36.8 1429 * 7
16 46.6 118.4 1400 * 7
From Table 15 it is seen that most steady state points proved to be highly repeatable as
almost all points exhibited low coefficient of variation (<2.0%). Only in the case of points
11 and 15 the relative standard deviation of the 3 measurements was higher than 2%.
Once more, higher instability is observed for low load points.
over the 16 steady state points. Comparison is performed over normalized to the average
of all measurements g Fuel/h value in order to better reflect the different FC over the
whole range of points tested. The simulations were run with VECTO SiCO mode
(3.0.2.466 version) for one of the three available sequences of points. The last column
presents the deviation [%] between the measured and the simulated fuel consumption
31
values (g Fuel/kWh) for each point. The deviation between measured and simulated
values is also graphically given in Figure 16.
Normalized
VSFC
[#] [rpm] [kW] [%]
[g/kWh]
1 1420±0.1 161.1±1.2 0.956 1.5
2 1403±0.0 138.7±0.7 0.964 2.1
3 1403±0.0 103.1±0.4 0.978 1.9
4 1403±0.1 76.0±0.2 1.018 0.4
5 1370±0.5 157.1±0.5 0.947 0.7
6 1371±0.2 30.3±0.6 1.272 0.2
7 1306±0.1 150.0±1.1 0.937 1.3
8 1648±0.1 156.5±1.4 0.947 0.3
9 1431±0.0 75.0±1.3 0.995 0.1
10 1409±0.2 151.2±1.2 0.921 0.7
11 1409±0.1 55.1±1.8 1.035 3.2
12 1344±0.3 149.3±1.5 0.922 1.4
13 1633±0.1 173.0±2.2 0.975 0.7
14 1497±16.7 167.5±1.6 0.962 1.2
15 1429±0.1 36.8±1.7 1.218 3.3
16 1400±0.1 118.4±1.6 0.953 1.7
Table 16 shows a lack of agreement between measured and simulated values regardless
the examined load. Deviations up to 8% were observed both for high (i.e. wheel power
>150 kW) and low (i.e. wheel power <50 kW) load points. Only some medium load
points (i.e. wheel power of 80-120 kW) demonstrated a fair repeatability of 1.5-2.5%
(see points 3, 4, 9, 16), which can be compared to this of the other examined vehicles.
VECTO seems to underestimate most of the measured points. One source of the error is
probably the false Air-Compressor status signal received by the CAN. It seems that there
was a constant signal showing that the Air-Compressor was always switched off even if it
was functioning. This way there were no data to disregard leading to an increased
measured fuel consumption with respect to the one simulated by the tool.
Overall, it is once more confirmed that steady state points are satisfactorily repeatable in
the lab with however a higher uncertainty when it comes to low load points. Difficulties
trying to repeat low load steady state points have been also reported by the respective
OEMs. In this case, VECTO failed to accurately simulate steady state points. This is
32
partially attributed to a false signal of the Air Compressor status received by the ECU.
Slightly lower deviations are found over medium load points.
Point Engine Speed Wheel Power Deviation
[#] [rpm] [kW] [%]
1 1420 159.8 1.477 1.584 ‐6.7
2 1403 138.1 1.279 1.385 ‐7.7
3 1403 102.8 1.022 1.037 ‐1.4
4 1403 75.9 0.774 0.795 ‐2.6
5 1371 157.0 1.448 1.515 ‐4.4
6 1371 30.7 0.374 0.396 ‐5.7
7 1306 151.2 1.389 1.435 ‐3.2
8 1648 156.9 1.516 1.523 ‐0.5
9 1431 76.5 0.799 0.778 2.6
10 1409 152.3 1.439 1.423 1.2
11 1409 57.2 0.617 0.583 5.9
12 1344 151.0 1.422 1.403 1.3
13 1633 175.5 1.644 1.752 ‐6.2
14 1487 168.4 1.563 1.665 ‐6.3
15 1429 38.7 0.439 0.466 ‐5.8
16 1400 120.2 1.127 1.144 ‐1.6
33
Table 17 shows WHVC and Regional Delivery test results without the application of any
filter. Measurements were conducted with A/C being switched off and after 1h warming
up of the vehicle. Values are averaged over 3 measurements which were performed over
different days. Vehicle specific fuel consumption is normalized to the average value of the
16 steady state points tested. The last column represents the coefficient of variation of 3
measurements of the fuel consumption at wheel (g Fuel/kWh).

Relative SD
[%]
Regional
69.1±0.4 23.7±0.1 1.038 0.9
Delivery
Hot Start
39.9±0.4 16.8±0.1 1.184 2.7
WHVC
Table 17 shows that both transient cycles are satisfactorily repeatable in the laboratory.
WHVC exhibited a coefficient of variation for the three different tests of about 3%. This
observation is similar to vehicle #1 and can be attributed to the fact that WHVC cycle is
too dynamic to stay within the speed band and in any case it is more transient than the
RD. On the other hand, Regional Delivery proved to be highly repeatable with the
coefficient of variation for the 3 different tests not exceeding 1% in accordance with the
results of the other trucks. Finally, normalized fuel consumption is higher over the WHVC
than the Regional Delivery due to the more transient nature of WHVC.
Table 18: Measured vs. simulated FC over Regional Delivery tests for vehicle #3
Normalized Normalized Deviation Deviation

Average Normalized
FC VECTO FC VECTO Measured Measured
SiCO Mode Eng. Mode vs. SiCO vs. Eng.
[km/h] [g/h]
[g/h] [g/h] [%] [%]
Regional
69.1 0.668 0.832 0.676 ‐1.1 ‐7.9
Delivery 1
Regional
68.7 0.658 0.825 0.668 ‐1.4 ‐8.4
Delivery 2
Regional
69.4 0.663 0.837 0.675 ‐1.7 ‐8.9
Delivery 3
Regional
Delivery 69.1±0.4 0.663 0.831 0.673 ‐1.4 ‐8.4
Averaged
WHVC 39.9 0.393 0.545 0.405 ‐2.8 ‐13.6
Table 18 shows the details of the comparison between measured and simulated fuel
consumption values over all Regional Delivery tests as well as over the one WHVC test
that was selected to be simulated. Comparison is performed over normalized to the
34
average of all measurements g Fuel/h value. Both VECTO SiCO and Engineering modes
were considered in this case for the simulations. The deviation between measured and
simulated fuel consumption is calculated using the fuel consumption at the wheel in g
Fuel/h. The deviation between measured and simulated values is graphically given in
Figure 17.
As shown in Table 18 the total fuel consumption calculated by VECTO SiCO mode was
very close to the measured value both for Regional Delivery and WHVC tests. Deviations
lower than 2% were found for the RD cycle, while it was found to be somewhat higher for
the WHVC (3%). On the other hand, the total fuel consumption simulated by VECTO
Engineering mode was found to be very far from the measured value with deviations
being as high as 9%. The difference was even higher for the unique WHVC test
examined. In all cases the VECTO Engineering mode underestimates the fuel
consumption suggesting a possible underestimation of the consumption of some of the
auxiliary systems.
Figure 17: Normalized simulated (SiCO mode blue bars – Engineering Mode red bars) vs. measured
(=1.0 – green bars) FC of vehicle #3 for Regional Delivery tests
Overall, it is once more demonstrated that both transient cycles are satisfactorily
repeatable in the laboratory with Regional Delivery being much easier to replicate
compared to WHVC. VECTO SiCO mode seems to be able to simulate both transient
cycles quite accurately, whereas the VECTO Engineering mode failed to do so. There is no
obvious reason for this big difference among the two modes and clearly there is a need
for further investigation.
3.1.4 Vehicle #4

Table 19 describes the stable state points tested with vehicle #4 in the dyno. Also in this
case measurements were conducted with A/C being switched off. Torque measurements
were conducted with the Kistler torquemeter device placed at the wheel rims. All tests
were performed with at least 1h warming up of the vehicle. A repetition of each point is
35
defined as a minimum of 60 s after stabilization phase. Measured wheel power and
engine speed for each point shall not deviate more than ±10 kW and ±25 rpm between
different measurements otherwise the measurement is considered invalid.
Vehicle’s Speed Wheel Power Engine Speed Shaft Speed Gear

#
1 83.4 265.8 1153 1153 12
2 84.0 130.5 1153 1152 12
3 84.5 24.3 1153 1152 12
4 81.8 200.5 1125 1125 12
5 82.0 158.0 1126 1125 12
6 82.1 121.7 1126 1125 12
7 82.3 72.4 1126 1125 12
8 80.6 234.1 1112 1111 12
9 76.5 244.7 1056 1055 12
10 73.5 266.2 1308 1016 11
11 64.5 252.1 1150 893 11
12 65.2 109.5 1150 893 11
13 65.4 65.6 1150 893 11
14 61.6 250.5 1099 854 11
15 70.4 291.0 1590 975 10
16 66.4 290.9 1501 921 10
17 50.9 175.2 1146 703 10
18 51.4 48.6 1146 703 10
Table 20 briefly describes SiCO test results including the application of specific filters (i.e.
A/C switched off, data with air compressor status “ON” disregarded). All values are
averaged over 3 measurements, while ± values correspond to the standard deviation of
the 3 measurements. Fuel consumption was recorded from the PEMS instrument as there
were some issues with the AMA device. Vehicle specific fuel consumption is normalized to
the average value of all 18 points measured in g Fuel/kWh. The last column represents
the coefficient of variation of the 3 fuel consumption measurements (g Fuel/kWh).
High load points (i.e. wheel power >200 kW) showed higher fuel consumption in g Fuel/h
(Table 21) but relatively low fuel consumption per kWh. This is seen in points 1, 10, 11,
14, 15 and 16 which all showed fuel consumption per kWh lower than the average value
of the 18 points (i.e. <1.0). Medium to low load points (i.e. wheel power of 100-200 kW)
showed vehicle specific fuel consumption close to the average value of the 18 points (see
points 2, 6, 12). On the other hand lower load points (i.e. wheel power <70 kW)
demonstrated higher normalized fuel consumption than the average of the 18 points (see
points 3, 13, 18). This observation is confirmed for all tested vehicles.
36
Normalized
VSFC
[#] [rpm] [kW] [%]
[g/kWh]
1 1153±2.6 265.8±2.0 0.931 0.4
2 1153±2.5 130.5±1.4 0.971 0.2
3 1153±2.5 24.3±0.9 1.537 0.8
4 1125±2.7 200.5±1.0 0.926 1.0
5 1126±2.2 158.0±0.9 0.941 0.7
6 1126±2.7 121.7±1.3 0.963 1.3
7 1126±2.6 72.4±0.8 1.051 0.5
8 1112±2.9 234.1±0.9 0.924 1.0
9 1056±2.6 244.7±1.2 0.935 1.0
10 1308±2.9 266.2±0.7 0.928 0.8
11 1150±2.4 252.1±0.9 0.946 0.7
12 1150±2.5 109.5±0.6 0.981 0.3
13 1150±2.3 65.6±0.5 1.061 0.0
14 1099±3.1 250.5±0.3 0.951 1.0
15 1590±3.3 291.0±0.7 0.937 0.4
16 1501±3.2 290.9±0.9 0.936 0.5
17 1146±4.1 175.2±0.9 0.947 0.5
18 1146±3.7 48.6±1.4 1.133 1.3
Regarding the repeatability of the SiCO test in the laboratory once more it is
demonstrated that most points exhibit very low coefficient of variation (<1.5%). In this
case no point exhibited relative standard deviation of the three measurements higher
than 2%. Even low load points (i.e. <50 kW) proved to be quite stable with relative
standard deviation of 0.8 and 1.3%.
Table 21 briefly presents the comparison between measured and simulated (VECTO SiCO
mode - 3.0.3 version) fuel consumption values over the 18 SiCO points. Comparison is
performed over normalized to the average of all measurements g Fuel/h value in order to
better reflect the different FC over the whole range of points tested. VECTO simulations
were performed for one of the three available sequences of points. This practically means
that Table 21 presents the direct comparison of the measured fuel consumption over this
particular series and the respective simulated values. The last column presents the
deviation [%] between the measured and the simulated fuel consumption values (g
Fuel/kWh) for each point. The deviation between measured and simulated values is also
graphically presented in Figure 18.
Also in this case there is not a good agreement between measured and simulated values,
regardless the point examined. All points demonstrated a deviation higher than 3%,
while in some cases it exceeded 10%. In all cases VECTO SiCO mode underestimates the
fuel consumption suggesting a possible underestimation of the load or/and the
37
consumption of some of the auxiliary systems. This behavior is similar to that of vehicle
#3.
Point Engine Speed Wheel Power Deviation
[#] [rpm] [kW] [%]
1 1155 264.1 1.532 1.668 ‐4.8
2 1154 129.3 0.765 0.829 ‐7.0
3 1155 23.9 0.209 0.244 ‐12.8
4 1128 200.2 1.156 1.266 ‐4.2
5 1127 157.5 0.918 0.976 ‐5.4
6 1128 121.4 0.719 0.766 ‐5.4
7 1128 72.0 0.453 0.488 ‐8.3
8 1114 234.0 1.358 1.462 ‐3.5
9 1054 243.6 1.432 1.539 ‐3.3
10 1310 265.6 1.499 1.679 ‐6.8
11 1150 251.1 1.422 1.608 ‐8.4
12 1150 108.9 0.643 0.701 ‐8.0
13 1150 65.2 0.410 0.453 ‐9.7
14 1102 250.8 1.368 1.593 ‐11.7
15 1592 290.6 1.656 1.859 ‐6.7
16 1503 290.5 1.647 1.853 ‐7.2
17 1142 174.4 0.997 1.091 ‐7.5
18 1143 48.3 0.321 0.356 ‐11.1
38
Overall, it is concluded that there are no big difficulties in repeating steady state points in
the lab. There is a question regarding low load points as they seem to be more difficult to
be stabilized. However, VECTO doesn’t seem capable of simulating steady state points
with a deviation from the measured value lower than 3%.

Table 22 briefly describes WHVC and RD test results without the application of any filter.
Measurements were conducted with at least 1h warming up of the vehicle and A/C being
switched off. Pull down of 20 kN was applied in order to avoid slipping of tyres. All values
are averaged over 3 measurements and ± correspond to the standard deviation of the
measurements. Fuel consumption is given in normalized to the average of SiCO points
value as g Fuel/kWh. The last column represents the coefficient of variation of the 3
different fuel consumption measurements (g Fuel/kWh).
Wheel Energy Normalized

Average Speed Relative SD
Test Measured VSFC
[km/h] [%]
[kWh] [g/kWh]
Regional
68.0±0.4 31.1±0.2 1.001 0.6
Delivery
Hot Start
39.9±0.7 23.9±0.9 1.132 3.1
WHVC
Once more it is demonstrated that Regional Delivery cycle is highly repeatable in the
chassis dyno of the VELA 7. On the other hand, WHVC exhibited a coefficient of variation
for the three different tests that were considered of about 3%. This deviation is similar to
that of vehicles #1 and #3. Again there were several difficulties for the driver to
reproduce braking events particularly over WHVC. Also it is seen that normalized fuel
consumption is higher over the WHVC than the Regional Delivery due to the more
Table 23 briefly presents the measured vs. simulated fuel consumption values over the
Regional Delivery cycle. Comparison is performed over normalized to the average of all
measurements g Fuel/h value. Only VECTO SiCO mode was considered for the
simulations. Unfortunately, no comparison is available for the WHVC cycle. VECTO
simulations for Regional Delivery were performed for all three available tests. Also, since
VECTO SiCO mode uses as input the measured torque at the wheel, in this case the
deviation between measured and simulated fuel consumption could be expressed both in
g Fuel/h and g Fuel/kWh. The deviation between measured and simulated values is also
given in Figure 19.
As shown in Table 23 the total fuel consumption simulated with VECTO matched
satisfactorily the experimentally measured for all Regional Delivery cycles. The difference
between calculated and measured fuel consumption values did not exceed 4% for
individual tests, while the averaged deviation was found to be lower than 3%. Again, the
VECTO underestimates actual fuel consumption but in this case there is no significant
39
difference like in case of steady state points. Similar deviation for the Regional Delivery
was also found for the other trucks tested.
Table 23: Measured vs. simulated FC over Regional Delivery cycle for vehicle #4
Average Wheel Energy Normalized Normalized

Deviation
Test Speed Measured FC VECTO FC Measured
[%]
[km/h] [kWh] [g/h] [g/h]
Regional
67.6 31.0 0.514 0.528 ‐4.0
Delivery #1
Regional
68.0 31.0 0.513 0.533 ‐3.9
Delivery #2
Regional
68.4 31.4 0.535 0.537 ‐0.5
Delivery #3
Regional
Delivery 68.0±0.4 31.1±0.2 0.521 0.533 ‐2.8
Average
Overall, it is concluded that while regional delivery is highly repeatable in the lab, this is
not the case for WHVC. Increased number of braking events and generally lower speeds
over WHVC seem to affect the repeatability of this cycle in the dyno. VECTO SiCO mode
is able to simulate Regional Delivery satisfactorily with a relatively low deviation
(approximately 4%) from the measured value. This conclusion can be generalized as it
was observed for all four vehicles tested.
Figure 19: Normalized simulated (SiCO mode - blue bars) vs. measured (=1.0 – red bars) FC of
vehicle #4 for Regional Delivery tests
40
3.2 On-road measurements
The scope of on-road tests was to check the repeatability of the applied methodology due
to the high level of uncertainty compared to laboratory tests and to investigate the
quality of the simulations under not controlled environment and operating conditions in
order to obtain a broader picture of the simulator's accuracy. Overall, the aim was to test
if ex-post verification based on on-road tests is possible and under what conditions.
3.2.1 Vehicle #1
Table 24 gives an overview of on-road test results without the application of any filter. In
all cases measurements were conducted with A/C being switched off. All tests were
performed with 30 min warming up of the vehicle and the first 15 min of the route were
disregarded. This way a minimum of conditioning for the engine as well as for the
gearbox and the axle was achieved as VECTO does not consider the cold behavior of the
components. Tests were performed at an average temperature of 7.8±2.9°C and 45±9%
RH without any form of precipitation. Values given in the last raw of Table 24 are
averaged over the 3 selected measurements and ± values correspond to the SD of the 3
measurements. Specific fuel consumption is given normalized to the average of the 15
SiCO points value as g Fuel/kWh. The last column represents the relative standard
deviation of the three measurements of fuel consumption (g Fuel/kWh) as measured at
the shaft (coefficient of variation).
Table 24 clearly demonstrates that on-road tests proved to be highly repeatable. The
coefficient of variation over the three different tests was lower than 1%. Unlike transient
tests in the laboratory there are no specific difficulties for the driver to reproduce braking
events over on-road tests due to the fact that the driving behaviour is more normal. All
tests exhibited similar speed profile with average speed being close to 67 km/h. A
difference in the energy consumed was observed in trip #3 compared to the other two
trips but it did not affect the overall fuel consumption which was found to be similar for
all trips.
Table 24: On-road test results for vehicle #1
Energy at Shaft Normalized

Test Measured Specific FC at Shaft
[km/h] [%]
[kWh] [g/kWh]
On‐Road #1 66.9 204.8 1.083 ‐
On‐Road #2 66.4 205.9 1.066 ‐
On‐Road #3 66.6 197.2 1.081 ‐
On‐road
66.6±0.3 202.6±4.7 1.077 0 .9
Average
values over all on-road tests. Comparison is performed over normalized to the average of
all measurements g Fuel/h value in order to better reflect the different FC over the whole
range of points tested. Both VECTO SiCO and Engineering mode were tested and
41
compared to the measured fuel consumption values. VECTO simulations were performed
for all three selected tests. Values given in the last raw are averaged over the 3
measurements and ± values correspond to the SD of the 3 measurements. The deviation
between measured and simulated values is also given in Figure 20.
Table 25: Measured vs. simulated FC over on-road tests for vehicle #1
Normalized Normalized Normalized Deviation Deviation

Average
FC VECTO FC VECTO FC Measured vs. Measured vs.
Test Speed
SiCO Mode Eng. Mode Measured SiCO Mode Eng. Mode
[km/h]
[g/h] [g/h] [g/h] [%] [%]
On‐Road #1 66.9 0.481 0.451 0.467 3.1 ‐3.3
On‐Road #2 66.4 0.486 0.446 0.459 5.8 ‐2.8
On‐Road #3 66.6 0.475 0.451 0.452 5.1 ‐0.2
On‐road
66.6±0.3 0.481 0.449 0.459 4.7 ‐2.1
Average
As shown in Table 25 the average deviation between the measured on-road fuel
consumption and the simulated one was calculated to be 4.7% with the VECTO SiCO
mode and 2.1% with the VECTO Engineering mode. This is a very satisfactory figure and
similar to that reported previously from Fontaras et al. (2016) with another Euro VI
truck. The most accurate results were achieved with the Engineering mode which takes
into account the road gradient but not the measured torque. On the other hand, SiCO
mode exhibited a slightly higher deviation which is partly attributed to the offset due to
torque drift as no correction was applied. This is clearly demonstrated in Figure 21 where
second by second measured and simulated (SiCO mode) fuel consumptions are
displayed. It is seen that the simulated fuel consumption is constantly increasing
compared to the measured value with the time (green spotted area). A 2-3%
overestimation of the measured torque at the shaft takes place and therefore is
incorrectly fed to the simulator resulting in an overestimation of the simulated fuel
consumption. Thus a correction should be applied for this reason making both modes
equally capable of predicting the vehicle’s fuel consumption.
Overall, it is seen that on-road tests proved to highly repeatable. There was a good
agreement between measured and simulated values for both VECTO Engineering and
SiCO mode. Furthermore it is seen that the VECTO tool seems to perform in a robust way
as the coefficient of variation for the 3 measurements was found to be 1.1% for the SiCO
mode and 0.6% for the Engineering mode. It is seen that ex-post verification based on
on-road tests could be possible with the advantage of being able to cover all vehicle
types. However, there is still a need to work on the details of the test protocol, define
test boundary conditions and understand the differences found among the two different
VECTO modes.
42
Figure 20: Normalized simulated vs. measured (=1.0) FC of vehicle #1 for on-road tests
Figure 22 shows the measured shaft power map for all different types of tests (i.e. on-
road, regional delivery and steady state points) conducted with vehicle #1. It can be
seen that SiCO test covers only a small part of the actual map and is more representative
of medium and higher loads. Also it is seen that Regional Delivery is not highly
representative of real-world operation as there are 2 main blocks of points which are not
found in the RD cycle (one of 1080 RPM and another with 1280 RPM). Finally, it is
apparent that under real-world conditions there are many low load points (i.e. <50 kW)
making clear that they should also be tested during the SiCO test.
Figure 21: Real time normalized simulated vs. measured FC for on-road test 1 (vehicle 1)
43
Figure 22: Shaft power map over on-road, Regional Delivery and SiCO tests
3.2.2 Vehicle #2
Table 26 gives an overview of on-road test results with vehicle #2. Among the four
executed tests the two closest to the average fuel consumption value were selected for
the processing of the results. No filters were applied and measurements were conducted
with A/C being switched off. A minimum of conditioning for the engine, gearbox and the
axle was achieved by a 30 min warming up of the vehicle. Also the 15 min of the route
were disregarded. Tests were performed at an average temperature of 13.9±2.3°C and
42±16% RH without any form of precipitation. Values in the last raw are averaged over
the 3 selected measurements with ± values corresponding to the SD of the 3
measurements. Vehicle specific fuel consumption is normalized to the average of the 17
SiCO points value as g Fuel/kWh. The last column represents the coefficient of variation
of the 3 fuel consumption (g Fuel/kWh) measurements.

Test Measured VSFC
[km/h] [%]
[kWh] [g/kWh]
On‐Road #1 62.5 231.5 1.016 ‐
On‐Road #2 62.9 231.3 1.002 ‐
On‐road
62.7±0.3 231.4±0.2 1.009 1 .0
Average
44
Like in case of vehicle #1, it is seen from Table 26 that on-road tests can be highly
repeatable. The coefficient of variation over the two different tests again did not exceed
1%. Both tests exhibited similar speed profile with an average speed of approximately 63
km/h while no significant difference in the consumed energy was observed. This is a very
important finding and is confirmed for two different vehicles. It is demonstrated that on-
road tests can be equally repeatable to chassis dyno tests allowing thus the ex-post
verification procedure to be based on on-road potentially overcoming all drawbacks
related to the laboratory testing.
Normalized Normalized Deviation Deviation

Average Normalized
FC VECTO FC VECTO Measured Measured
w/o Padd 4.51kW Padd vs. w/o Padd vs. Padd
[km/h] [g/h]
[g/h] [g/h] [%] [%]
On‐Road
62.5 0.549 0.577 0.569 ‐3.6 1.4
#1
On‐Road
62.9 0.558 0.586 0.565 ‐1.4 3.6
#2
On‐road
62.7±0.3 0.553 0.581 0.567 ‐2.5 2.5
Average
values over on-road tests with vehicle #2. Comparison is performed over normalized to
the average of all measurements g Fuel/h value. In this case only VECTO SiCO mode was
tested. However, two different SiCO mode scenarios were applied (without Padd and with
4.51kW). Values given in the last raw are averaged over two measurements and ±
values correspond to the SD of the two measurements. The deviation between measured
and simulated fuel consumption is expressed both in g Fuel/h and g Fuel/kWh without
any difference. The deviation between measured and simulated values is also depicted in
Figure 23.
Table 27 demonstrates that the average deviation between the measured and the
simulated on-road fuel consumption was found to be 2.5% with both VECTO SiCO modes.
Once more this is a quite satisfactory figure with respect to these reported previously
(Fontaras et al. 2016). In this case SiCO mode proved to be more accurate compared to
vehicle #1 probably due to the use of the wheel torquemeter device. A zeroing and a
correction based on the measured drift is applied before and after testing, therefore the
simulated and measured fuel consumption do not deviate with the time like in case of
vehicle #1. This is clear from Figure 24 where the real time normalized simulated and
measured fuel consumption for on-road test 1 is shown. The difference in the
performance of the two devices could satisfactorily explain the 2-3% more accurate
simulation achieved with vehicle #2.
Once more on-road tests proved to be highly repeatable. A very good agreement
between measured and simulated values for both VECTO SiCO modes was observed
(deviation <3%). VECTO tool seems to be able to simulate the actual fuel consumption in
a robust way as the coefficient of variation for the two simulations was found to be lower
45
than 2%. It is seen that ex-post verification based on on-road tests could be possible
with the advantage of being able to cover all vehicle types.
Figure 25 depicts the measured wheel power map for all different types of tests
conducted with vehicle #2. Like in case of vehicle #1, SiCO test covers only a small part
of the actual map and is more representative of medium and higher loads. Regional
Delivery cannot represent real-world operation very accurately as it is focused mainly in
one block of points (at approximately 1170 RPM), thus leaving out several other
important blocks where the engine actually operates. Once more it is demonstrated that
under real-world conditions there are many low load points (i.e. <50 kW) making clear
that they should also be taken into account for the SiCO test.
46
Figure 25: Wheel power map over on-road, Regional Delivery and SiCO tests
3.2.3 Vehicle #3
Table 28 gives an overview of on-road test results with vehicle #3. Values in the last raw
represent the average over the 3 selected measurements while ± values correspond to
the SD of the 3 measurements. No filters were applied and measurements were
conducted with A/C being switched off. A minimum of conditioning for the engine,
gearbox and the axle was achieved by a 30 min warm up of the vehicle. The first 15 min
of the route were disregarded. Tests were performed at an average temperature of
25.5±0.7°C and 43±6% RH without any form of precipitation. Fuel consumption is
normalized to the average value measured from the 16 SiCO points in g Fuel/kWh. The
last column represents the coefficient of variation of the 3 fuel consumption (g Fuel/kWh)
measurements.

Relative SD
[%]
On‐Road #1 69.3 145.5 1.056 ‐
On‐Road #2 67.6 153.0 1.062 ‐
On‐Road #3 69.5 149.6 1.071 ‐
On‐road
68.8±1.1 149.3±3.8 1.063 1 .4
Average
47
It is one more confirmed that on-road tests can be highly repeatable. The coefficient of
variation over the three different tests slightly exceeded 1% and was very close to the
values found for vehicles #1 and #2. All tests exhibited similar speed profile with an
average speed of approximately 69 km/h, while there was no significant difference in the
measured wheel energy. It is clearly demonstrated with three different vehicles that on-
road tests can be equally repeatable to chassis dyno tests, allowing thus the ex-post
verification procedure to be based on real world testing.
Normalized Normalized Normalized Deviation Deviation

Average
FC VECTO FC VECTO FC Measured vs. Measured vs.
Test Speed
SiCO Mode Eng. Mode Measured SiCO Mode Eng. Mode
[km/h]
[g/h] [g/h] [g/h] [%] [%]
On‐Road
69.3 0.552 0.793 0.560 ‐1.4 ‐9.2
#1
On‐Road
67.6 0.566 0.773 0.563 0.6 ‐6.3
#2
On‐Road
69.5 0.570 0.797 0.568 0.4 ‐7.2
#3
On‐road
68.8±1.1 0.563 0.788 0.564 ‐0.2 ‐7.6
Average
values over the individual on-road tests with vehicle #3. Comparison is performed over
normalized to the average of all measurements g Fuel/h value in order to better reflect
the different FC over the whole range of points tested. Also the averaged values are
provided in the last raw and ± values correspond to the SD of the three measurements.
Both VECTO SiCO and Engineering mode were examined. Since also VECTO Engineering
mode was used the deviation between measured and simulated fuel consumption is
calculated using the fuel consumption at the wheel in g Fuel/kWh. The deviation between
measured and simulated values is also depicted in Figure 26.
Table 29 shows that the average deviation between the measured and the simulated on-
road fuel consumption was found to be 0.2% with VECTO SiCO mode. This practically
means that VECTO can precisely simulate the fuel consumption for this particular truck.
Again VECTO SiCO mode proved to be more accurate compared to vehicle #1, probably
due to the use of the wheel torquemeter device instead of the torquemeter at the cardan
shaft. It is shown in Figure 27 that simulated (VECTO SiCO mode) and measured fuel
consumption do not deviate with the time like in case of vehicle #1. On the other hand,
there is a high deviation between the measured and the simulated on-road fuel
consumption with VECTO Engineering mode. This was also observed with the transient
cycles in the lab and shall be further investigated.
Overall, on-road tests proved to be highly repeatable. A very good agreement between
measured and simulated values for VECTO SiCO mode was observed. VECTO tool seems
to be able to simulate the actual fuel consumption in a robust way as the coefficient of
variation for the three simulations was found to be lower than 1%.
48
conducted with vehicle #3. Like in case of the other tested vehicles, SiCO test covers
only a small part of the actual map and is more representative of medium and higher
loads. Regional Delivery cannot represent real-world operation very accurately as it is
focused mainly in one block of points (at approximately 1400 RPM), thus leaving out
several other important blocks where the engine actually operates (close to 1450 RPM).
Once more it is demonstrated that under real-world conditions there are many low load
points (i.e. <40 kW).
Figure 27: Real time normalized simulated vs. measured FC for on-road test 1 (vehicle #3)
49
3.2.4 Vehicle #4
Table 30 gives an overview of on-road test results with vehicle #4. Unfortunately due to
time restrictions only two tests were performed. No filters were applied and
measurements were conducted with A/C being switched off. A minimum of conditioning
for the engine, gearbox and the axle was achieved by a 30 min warming up of the
vehicle. The first 15 min of the route were disregarded. Values in the last raw are
averaged over the 2 measurements. Tests were performed at an average temperature of
28.0±1.7°C and 55±7% RH without any form of precipitation. Fuel consumption at the
shaft is expressed in g Fuel/h and normalized to the average of the 18 SiCO points value
as g Fuel/kWh. The last column represents the coefficient of variation of the two fuel
consumption (g Fuel/kWh) measurements.

Test Measured VSFC
[km/h] [%]
[kWh] [g/kWh]
On‐Road #1 67.9 216.9 0.996 ‐
On‐Road #2 69.6 213.1 1.001 ‐
On‐road
68.8±1.2 215.0±2.6 0.998 0 .3
Average
50
Table 30 demonstrates that on-road tests with vehicle #4 showed an excellent
repeatability. The coefficient of variation over the two different tests was lower than 1%.
Individual tests exhibited similar speed profile with an average speed of approximately
69 km/h while no significant difference in the consumed energy was observed. Overall,
the high repeatability of on-road tests is a very important finding and is confirmed for all
tested vehicles. It is demonstrated that on-road tests can be equally repeatable to
chassis dyno tests allowing thus the ex-post verification procedure to be based on on-
road potentially overcoming all drawbacks related to the laboratory testing.
Deviation
Average Wheel Energy Normalized Normalized Measured
Test Speed Measured FC VECTO FC Measured vs. SiCO
[km/h] [kWh] [g/h] [g/h] Mode
[%]
On‐Road
67.9 216.9 0.466 0.483 ‐3.5
#1
On‐Road
69.6 213.1 0.477 0.489 ‐2.6
#2
On‐road
68.8±1.2 215.0±2.6 0.471 0.486 ‐3.1
Average
values over on-road tests with vehicle #4. Comparison is performed over normalized to
the average of all measurements g Fuel/h value. In this case only VECTO SiCO mode was
tested. Values given in the last raw are averaged over the two measurements and ±
values correspond to the SD of the two measurements. The deviation between measured
and simulated fuel consumption is expressed in g Fuel/h. The deviation between
measured and simulated values is also depicted in Figure 29.
Table 31 demonstrates that the average deviation between the measured and the
simulated on-road fuel consumption was found to be approximately 3.0%. Once more
this is a very satisfactory figure and comparable to that found for the other vehicles. Both
individual trips exhibited similar difference. Also in this case SiCO mode proved to be
more accurate compared to vehicle #1 probably due to the use of the wheel torquemeter
device. Figure 30 presents the real time normalized simulated and measured fuel
consumption.
Once more on-road tests proved to be highly repeatable. A very good agreement
between measured and simulated values for VECTO SiCO mode was observed (deviation
close to 3%). VECTO tool seems to be able to simulate the actual fuel consumption in a
robust way as the coefficient of variation for the two simulations was found to be lower
than 1%.
conducted with vehicle #4. Steady state test covers only a small part of the actual map
while Regional Delivery cannot represent real-world operation very accurately as it is
focused mainly in one block of points (at approximately 1170 RPM), thus leaving out
51
several other important blocks where the engine actually operates (close to 1200 RPM).
Also it is seen that under real-world conditions there is a big portion of low load points
(i.e. <50 kW) making clear that they should also be taken into account when the SiCO
test is executed.
52
53
4 Conclusions
The study was conducted in two phases: the experimental phase which took place in JRC
and involved testing of four Euro VI trucks in the laboratory and on-road, and the
simulations phase which was performed within each individual OEM after the completion
of the respective experimental campaign. Afterwards, JRC performed an independent
comparison between the results of the simulations and those of the measurements and
the conclusions were communicated to the respective OEM and to Member States at the
8th meeting of the CO2 Heavy-Duty Editing Board. The findings of the main evaluation
phase can be summarized to the following.
Steady State Transient Cycle

Option On-road operation
Chassis Dyno Chassis Dyno
Very good except

Repeatability Very good Very good
for low-load points
Representativeness High with some

of actual vehicle Lowest restrictions in Highest
operation brake applications
Applicability to all Restrictions for Restrictions for

Possible
HDVs specific categories specific categories
Cost High High Lowest
Low provided all Low but requires

High due to the
Complexity equipment specific test
nature of the test
available protocol
Lowest directly
High needs specific
comparable to
Test Data analysis Low boundary
specific VECTO
conditions
output
Fair - Elements
Maturity Good - A first draft
Poor - New from PEMS
(how close to actual of the protocol
protocol is required protocol can be
implementation) described
adopted
Steady State testing in the laboratory (SiCO)

 Steady state testing (SiCO test) proved to be satisfactorily repeatable in the
laboratory in terms of fuel consumption, particularly when medium and high load
points were examined (i.e. power at the wheel or shaft >100 kW). Low load points
(i.e. power at wheel or shaft <50 kW) proved to be more difficult to repeat and
thus more unstable compared to medium and high load points. This observation
54
was confirmed by all OEMs and it needs to be considered seriously as important
part of fuel consumption occurs over low load points. Furthermore, on-road tests
demonstrated that low load points hold a significant share of the engine map
under real world conditions; therefore they should not be neglected when
investigating CO2 emissions particularly of vehicle classes operating mostly in low
vehicle speed conditions.
 A good agreement between measured and simulated fuel consumption values
over medium and high loads was observed (deviation between measured and
simulated fuel consumption in g/kWh was always lower than 2%) with two of the
trucks tested. This was not the case for low load points (i.e. power at wheel or
shaft <50 kW) where in some cases the deviation between measured and
simulated fuel consumption reached 4%. On the other hand, vehicles #3 and #4
demonstrated generally higher deviations between measured and simulated fuel
consumption values, regardless the tested load. Further investigation is required
to understand the high deviations found for the two trucks.
 Apart from the issue with low load points there are several other drawbacks
related to the SiCO test methodology. First of all, it is very difficult to cover the
full engine and gearbox map with only 12 (or even 18) steady state points.
Furthermore, it is not possible with one test to cover the full range of HD vehicles
(trucks, buses, coaches, etc.). Finally, it needs expensive and difficult to maintain
equipment, specifically when it needs to be performed on road (special braking
trailers, dedicated testing facilities).
Transient testing in the laboratory

 The results showed good measurement repeatability (i.e. coefficient of variation
for three fuel consumption measurements <2%) was observed in the laboratory
when the Regional Delivery cycle was performed. Higher deviation was observed
with the WHVC cycle for almost all vehicles (coefficient of variation close to 5%).
Compared to Regional Delivery, WHVC speed profile includes many more braking
events and is characterized from generally lower speeds. This could explain the
lower repeatability of WHVC. Finally, normalized fuel consumption is found to be
higher over the WHVC than the Regional Delivery probably due to the more
 A satisfactorily agreement of measured and simulated fuel consumption was
observed over Regional Delivery tests with the deviation being generally lower
than 3% and only once reaching 5%. VECTO PWheel mode proved to be more
reliable in simulating measured fuel consumption values compared to Engineering
mode whenever both modes were examined. On the other hand, WHVC proved to
be more difficult to simulate accurately regardless the vehicle tested. In general,
it could be concluded that VECTO is capable of providing reliable fuel consumption
values in the laboratory over different transient tests within, however, a slightly
higher uncertainty compared to steady state points.
 There are several drawbacks related to transient testing method in the laboratory.
There are several difficulties for the driver to reproduce braking events over
transient cycles and specifically over WHVC. Despite that the engine and gearbox
map is more realistic and closer to the real world operation of the trucks it doesn’t
yet cover the full range. Furthermore, it is not possible with one test to cover the
55
full range of HD vehicles (trucks, buses, coaches, etc.). Again there is a need for
expensive and difficult to maintain equipment (chassis dyno, special braking
trailers, etc.). Overall, it seems to be a poor compromise between steady state
and on-road tests as it doesn’t solve any issues related to neither of the two
methods.
On-road testing
 Excellent fuel consumption repeatability was observed over on-road tests,
regardless the truck tested. The coefficient of variation of fuel consumption
measurements over three repetitions was always lower than 1.5%. This result is
somewhat surprising since these tests were known to be quite uncertain and
difficult to repeat accurately.
 Overall, a good agreement of measured and simulated fuel consumption values
was observed over on-road tests with the deviation never exceeding <5%. When
the wheel rims were used for the measurement of the torque at wheel the
deviation between measured and simulated fuel consumption did not exceed 3%,
while in the case of vehicle #1 it was found to be close to 5% due to the drift of
the torquemeter installed at the cardan shaft. VECTO SiCO mode proved to be
more reliable in simulating measured fuel consumption values compared to
Engineering mode. Overall, it seems that VECTO is capable of providing reliable
results over on-road tests. However, differences among different VECTO modes
should be further investigated.
 On-road tests seem to be a good solution as they overcome most of the
drawbacks related to laboratory testing methodologies. First of all, the full engine
and gearbox map is investigated as the truck operates only under real world
conditions. In this case one testing methodology with very limited modifications
could be adopted to cover the full range of HD Vehicles. Finally, it overcomes the
need for expensive equipment (chassis dyno, special braking trailers, etc.).
Overall, the ex-post verification based on on-road tests is possible for trucks.
However, there is a clear need to work on the details of the test protocol, define
boundary conditions, and further understand the differences found among the
different VECTO modes. Finally, it is necessary to define what would be considered
as acceptable maximum deviation between measured and simulated values for
CO2 emissions.
 Regardless the testing methodology, a better agreement between measured and
simulated fuel consumption values was observed when the wheel rims were used
for the measurement of the wheel torque compared to the application of the
cardan shaft torquemeter. This became more obvious over on-road tests due to
their longer duration compared to laboratory tests.
56
References
ACEA; Evaluation of VECTO tool. In: Clima JD, editor. ACEA report communicated to the
European Commission; 2013
AEA-Ricardo; Reduction and testing of GHG emissions from HDVs - Final Report. 2011.
http://ec.europa.eu/clima/policies/transport/vehicles/docs/ec_hdv_ghg_strategy_en.pdf
European Commission WHITE PAPER - Roadmap to a Single European Transport Area e

towards a competitive and resource efficient transport system. 2011. COM(2011) 144
European Commission 2016; Communication from the Commission to the European

Parliament, the Council, the European Economic and Social Committee & the Committee
of the regions. Available at ec.europa.eu/transparency/regdoc/rep/1/2016/EN/1-2016-
501-EN-F1-1.PDF
European Commission 2017; Implementing Regulation (EU) No 595/2009 of the

European Parliament and of the Council as regards the determination of the CO2
emissions and fuel consumption of Heavy-Duty vehicles and amending Directive
2007/46/EC of the European Parliament and of the Council and Commission Regulation
(EU) No 582/2011. Annexes 1-11. Available at
https://circabc.europa.eu/faces/jsp/extension/wai/navigation/container.jsp?FormPrincipal
:_idcl=FormPrincipal:_id1&FormPrincipal_SUBMIT=1&id=419045db-b7c5-4935-a7ab-
91332f168b0a&javax.faces.ViewState=AiF8vIrLRDhX8GPwNN49KQLs2HhisuQb6lzN8Cmw
zyIBjWRolCjqDNWCt5MZp%2FpnPwzC%2FfQVLy8VC3lVNILLNypIybTNR%2BYMqMp8fnKf
oUW9%2Bei91P8oQ9CO%2Fw9aU6vQott9kxlFKg8B%2B27guRIpLmRO8A8%3D
Fontaras G., Grigoratos T., Savvidis D., Anagnostopoulos K., Luz R., Rexeis M.,
Hausberger S.; An experimental evaluation of the methodology proposed for the
monitoring and certification of CO2 emissions from heavy-duty vehicles in Europe; Energy
102 (2016) 354-364
Fontaras G., Rexeis M., Dilara P., Hausberger S., Anagnostopoulos K.; The development
of a simulation Tool for monitoring heavy-duty vehicle CO2 emissions and fuel
consumption in Europe. 2013. SAE Technical Paper 2013-24-0150
Franco V., Delgado O., Muncrief R.; Heavy-duty vehicle fuel-efficiency simulation: A
comparison of US and EU tools. April 2015. ICCT report – White Paper, Available at:
http://www.theicct.org/sites/default/files/publications/ICCT_GEM-VECTO-comparison_20150511.pdf
Giechaskiel B., Riccobono F., Villafuerte PM., Grigoratos T.; Particle Number (PN) -
Portable Emissions Measurement Systems (PEMS). Heavy Duty Vehicles Evaluation phase
at the Joint Research Centre (JRC); JRC Science for Policy Report. 2016
Grigoratos T. and Fontaras G.; Ex-post Verification – SiCO and VECTO validation exercise
Preliminary Results; September 2016. Editing Board
Luz R., Rexeis M., Hausberger S., Jajcevic D., Lang W., Schulte L.E.; Development and
validation of a methodology for monitoring and certification of greenhouse gas emissions
from heavy duty vehicles through vehicle simulation. 2014. Final Report. Available at:
http://ec.europa.eu/clima/policies/transport/vehicles/heavy/docs/final_report_co2_hdv_e
n.pdf
57
List of abbreviations and definitions
ACEA European Automobile Manufacturers' Association

CLD ChemiLuminescence Detector
CVS Constant Volume Sampler
DPF Diesel Particulate Filter
EC European Commission
EFM Exhaust Flow Meter
FC Fuel Consumption
FTIR Fourier Transform Infra-Red spectrometer
GPS Global Positioning System
HDV Heavy Duty Vehicle
HEPA High Efficiency Particulate Air filter
HFID Heated Flame Ionization Detector
JRC Joint Research Centre
NDIR Non-Dispersive Infra-Red sensor
NDUV Non-Dispersive Ultra-Violet sensor
OBD On-Board Diagnostics
PEMS Portable Emissions Measurement System
PN Particle Number
RD Regional Delivery cycle
SCR Selective Catalyst Reduction of NOx
SiCO Simplified Constant Speed Test
SS Steady State
THC Total Hydrocarbons
VECTO Vehicle Energy consumption Calculation Tool
VSFC Vehicle Specific Fuel Consumption
VELA Vehicle Emissions Laboratories
WHTC World Harmonized Transient Cycle
WHVC World Harmonized Vehicle Cycle
58
List of figures
Figure 1: VELA 7 facilities .................................................................................................. 10
Figure 2: Setup of the instrumentation in the controlled temperature dilution tunnel room .......... 11
Figure 3: Trucks tested in the climatic room and on-road ....................................................... 12
Figure 4: Speed versus time profile of WHVC ........................................................................ 13
Figure 6: Example of a constant speed profile of engine rpm vs. torque for vehicle #2 ............... 14
Figure 7: Route for on-road tests performed. Red line - Motorway; Blue line - Urban & Rural ....... 15
Figure 8: Speed profile of vehicle #2 during one typical on-road test ....................................... 16
Figure 9: VECTO’s simulation core ...................................................................................... 17
Figure 10: Torque measurement device on the cardan shaft of vehicle #1 ................................ 19
Figure 11: Deviation of the measured vs. the simulated FC values for all examined SS points ...... 22
Figure 12: Normalized simulated vs. measured (=1.0) FC of vehicle #1 for WHVC and RD .......... 24
Figure 13: Torque measurement device at the wheels of vehicle #2 ........................................ 26
Figure 15: Normalized simulated (w/o Padd blue bars – 4.51kW Padd red bars) vs. measured (=1.0 –
green bars) FC of vehicle #2 for Regional Delivery tests ........................................................ 30
Figure 17: Normalized simulated (SiCO mode blue bars – Engineering Mode red bars) vs. measured
(=1.0 – green bars) FC of vehicle #3 for Regional Delivery tests ............................................. 35
Figure 19: Normalized simulated (SiCO mode - blue bars) vs. measured (=1.0 – red bars) FC of
vehicle #4 for Regional Delivery tests ................................................................................. 40
Figure 20: Normalized simulated vs. measured (=1.0) FC of vehicle #1 for on-road tests ........... 43
Figure 21: Real time normalized simulated vs. measured FC for on-road test 1 (vehicle 1) .......... 43
Figure 22: Shaft power map over on-road, Regional Delivery and SiCO tests ............................. 44
Figure 25: Wheel power map over on-road, Regional Delivery and SiCO tests............................ 47
Figure 27: Real time normalized simulated vs. measured FC for on-road test 1 (vehicle #3) ........ 49
59
List of tables
Table 1: Main vehicle characteristics and main input data origin .............................................. 11
Table 2: Driving phase distributions of on-road trips .............................................................. 16
Table 3: VECTO modes description ...................................................................................... 18
Table 4: SiCO points for vehicle #1 ..................................................................................... 20
Table 5: SiCO dyno test results for vehicle #1 ...................................................................... 21
Table 6: Measured vs. simulated fuel consumption over SiCO dyno test for vehicle #1 ............... 22
Table 7: WHVC and Regional Delivery dyno test results for vehicle #1 ..................................... 23
Table 8: Measured vs. simulated FC over WHVC and Regional Delivery for vehicle #1................. 23
Table 9: SiCO points for vehicle #2 ..................................................................................... 25
Table 10: SiCO dyno test results for vehicle #2 .................................................................... 26
Table 11: Measured vs. simulated fuel consumption over SiCO dyno test for vehicle #2 .............. 27
Table 12: WHVC and Regional Delivery dyno test results for vehicle #2 .................................... 29
Table 13: Measured vs. simulated FC over Regional Delivery tests for vehicle #2 ....................... 29
Table 14: SiCO points for vehicle #3 ................................................................................... 31
Table 18: Measured vs. simulated FC over Regional Delivery tests for vehicle #3 ....................... 34
Table 19: SiCO points for vehicle #4 ................................................................................... 36
Table 23: Measured vs. simulated FC over Regional Delivery cycle for vehicle #4 ...................... 40
Table 24: On-road test results for vehicle #1........................................................................ 41
Table 25: Measured vs. simulated FC over on-road tests for vehicle #1 .................................... 42
60
Europe Direct is a service to help you find answers
to your questions about the European Union.
Freephone number (*):
00 800 6 7 8 9 10 11
(*) The information given is free, as are most calls (though some operators, phone boxes or hotels may
charge you).
More information on the European Union is available on the internet (http://europa.eu).
HOW TO OBTAIN EU PUBLICATIONS
Free publications:
• one copy:
via EU Bookshop (http://bookshop.europa.eu);
• more than one copy or posters/maps:

from the European Union’s representations (http://ec.europa.eu/represent_en.htm);
from the delegations in non-EU countries (http://eeas.europa.eu/delegations/index_en.htm);
by contacting the Europe Direct service (http://europa.eu/europedirect/index_en.htm) or
calling 00 800 6 7 8 9 10 11 (freephone number from anywhere in the EU) (*).
(*) The information given is free, as are most calls (though some operators, phone boxes or hotels may charge you).
Priced publications:
• via EU Bookshop (http://bookshop.europa.eu).
KJ-NA-28409-EN-N
doi:10.2760/081827
ISBN 978-92-79-64966-0

Jrc104993 Pilot Phase Final Report

Uploaded by

Copyright:

Available Formats

Jrc104993 Pilot Phase Final Report

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

Jrc104993 Pilot Phase Final Report

Uploaded by

Copyright:

Available Formats

Assessment of the monitoring

methodology for CO₂ emissions

Theodoros Grigoratos, Georgios Fontaras,

JRC Science Hub

PDF ISBN 978-92-79-64966-0 ISSN 1831-9424 doi:10.2760/081827

Luxembourg: Publications Office of the European Union, 2017

© European Union, 2017

All images © European Union 2017

Quick guide / Experimental

Results Analysis - Main findings

Steady State testing in the laboratory

Transient testing in the laboratory

Related and future JRC work

2.1 VELA 7 facilities and setup

2.2 Test vehicles

Table 1: Main vehicle characteristics and main input data origin

Characteristic Range of Values

Engine Displacement [cm3] 7700 – 12900

Rated Power [kW] 240 – 355

Rated Torque [Nm] 1224 – 2500

Gearbox Manual or AMT

Max load [kg] 24,000 – 40,000

Test Mass [kg] 17,000 – 27,000

Emissions Category EURO VI

Torque measurement Cardan shaft transducer or Wheel rim torquemeter

Exhaust emissions control EGR, DPF, SCR

Figure 3: Trucks tested in the climatic room and on-road

2.3 Daily test protocol and test cycles

Figure 4: Speed versus time profile of WHVC

Table 2: Driving phase distributions of on-road trips

Speed Classification Share in total trip duration

Low Speed <50 km/h [~Urban] 23‐27%

Medium Speed 50‐70 km/h [~Rural] 4‐15%

High Speed >70 km/h [~Motorway] 60‐69%

Figure 8: Speed profile of vehicle #2 during one typical on-road test

2.4 Vehicle simulator

Figure 9: VECTO’s simulation core

The simulation-core is summarized schematically in Figure 9. Additional information

Table 3: VECTO modes description

Mass & Gearbox &

3.1 Chassis dynamometer measurements

3.1.1.1 Steady State Tests

Figure 10: Torque measurement device on the cardan shaft of vehicle #1

Vehicle’s Speed Power at Shaft Engine Speed Shaft Speed Gear

Table 5: SiCO dyno test results for vehicle #1

Average Shaft Energy Normalized

Average Shaft Energy Normalized Normalized

3.1.2.1 Steady State Tests

Table 9: SiCO points for vehicle #2

Vehicle’s Speed Power at Wheel Engine Speed Shaft Speed Gear

Table 10: SiCO dyno test results for vehicle #2

3.1.2.2 Transient Tests

Average Wheel Energy Normalized

Wheel Normalized Normalized Deviation Deviation

3.1.3.1 Steady State Tests

Table 14: SiCO points for vehicle #3

Vehicle’s Speed Power at Wheel Engine Speed Shaft Speed Gear

Table 15: SiCO dyno test results for vehicle #3

2 1403±0.0 138.7±0.7 0.964 2.1

3 1403±0.0 103.1±0.4 0.978 1.9