Professional Documents
Culture Documents
6Analysing the real-world fuel and energy consumption of conventional and
6Analysing the real-world fuel and energy consumption of conventional and
A R T I C L E I N F O A B S T R A C T
Keywords: The transport sector constitutes one of the main sources of greenhouse gas emissions in the European Union.
CO2 emissions Although the literature is rich in studies assessing the factors influencing energy consumption in real world
Real-world driving, few analyse the divergence between certified and real-world consumption on a fleet level and annual
Passenger cars
temporal resolution. The present work builds on the findings of previous studies to provide a complete
Energy consumption
Fuel consumption gap
simulation-based approach for fleet-level projections in real-world scenarios. A tailored simulation framework
for fleet-wide analysis has been updated and validated extensively for current vehicle configurations. In addition
to modelling, the analysis used data describing real-world conditions and vehicle characteristics representative of
the European passenger car fleet and operation profile. Validation demonstrates good accuracy in simulating the
vehicle measured consumption values with a near-zero mean error when simulating laboratory and on-road
measurements. The impact of the certification procedure was quantified at 6% for both the combustion en
gine and battery electric vehicles. The influence of real-world factors such as traffic, ambient temperature and
cabin air conditioning was quantified, and the results show a greater energy impact in the cold than in warm
conditions. Different operating conditions were simulated to investigate the model’s capacity to estimate the
distribution of the average sales-weighted gap in energy consumption between officially reported CO2 values and
the real-world ones. The study focuses on three scenarios: average European conditions, the best and worst case
for assessing the extremes, and a Monte Carlo-based scenario to determine the impact of warm and cold periods.
Results show average gap values of 13.5% (best case) and 34.5% (worst case) for conventional vehicles and
− 4.5% (best case) and 23.9% (worst case) for battery electric ones. Warm and cold weather conditions result in
5.5 and 7.5% increases for conventional and hybrids and 7.5 and 15% for electrics, respectively.
☆
The views expressed in the paper are purely those of the authors and shall not be considered as an official European Commission position under any circumstance.
* Corresponding author.
E-mail address: georgios.fontaras@ec.europa.eu (G. Fontaras).
https://doi.org/10.1016/j.enconman.2022.116161
Received 19 April 2022; Received in revised form 18 July 2022; Accepted 18 August 2022
Available online 13 September 2022
0196-8904/© 2022 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY license (http://creativecommons.org/licenses/by/4.0/).
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 1. Flowchart of the methodology followed. The different colours of the objects correspond to the different processes (validation, impact simulations, projection).
The different shapes correspond to the data (vehicle databases, trip databases, factors for assessment).
figures reported officially [5]. Pavlovic et al. [6] demonstrated, through impact. Zacharof et al. [20] combined literature with detailed vehicle
an one-year on-road campaign where several drivers used the same car simulations to analyse the sensitivity of factors such as road gradient,
for their day-to-day needs, that the gap reached values up to 100 % for aerodynamics, and auxiliaries usage in CO2 emissions in indicative
certain drivers. Battery electric vehicles (BEVs) have undergone rapid conventional vehicles. Moreover, methodologies are proposed to assess
global growth during the last years. Their share has tripled due to higher the effect of such non-regulatory parameters (e.g. wind [21], surface
public environmental awareness [7], and the efforts by manufacturers to wetness, auxiliary systems [22], and road gradient [23]) for a conven
meet the new CO2 targets [8]. Although tailpipe emissions are not tional vehicle. Ktistakis et al. [14], using statistical approaches, analysed
relevant in BEVs, there is interest in their actual energy consumption databases with driver-reported CO2 values and characterised the vari
which eventually translates to upstream emissions, their driving range ability of the gap due to vehicle characteristics. More recently, studies
that is of high consumer concern, and the relevant influencing factors. have compared conventional powertrains with electrified ones. Fiori
Early investigations using simulation approaches [9], or assessing fleet et al. [24], and Mamarikas et al. [25] assessed the impact of traffic
data [10], placed the RW values close to the officially declared ones. conditions for BEVs and found that the consumption patterns were
Nonetheless, Paffumi et al. [11] experimentally found that the energy different between BEVs and conventional vehicles. HVAC holds a
consumption due to heating, ventilation, and air-conditioning (A/C), considerable share in the increase of the energy consumption of a
reffered to as HVAC, increased consumption by up to 70–80 % in lab vehicle, and detailed models quantify the impact in RW driving cycles
oratory tests in − 10 ◦ C ambient temperature. [26]. The above studies substantially contribute to the research on RW
Recent provisions were added to monitor the RW fuel and electric CO2, energy consumption gap, and variability. However, they focus on
energy consumption of passenger cars and vans to prevent the gap from determining the sensitivity of different parameters, and to the authors’
growing again [12]. To achieve this and to have the possibility to assess knowledge, there is no study assessing their combined effect (including
RW representativeness of the consumption determined at the certifica the influence of the certification test) in the current European car fleet
tion procedure, car manufacturers were required to install onboard fuel composition under specific, realistic scenarios applicable to Europe for a
consumption monitoring (OBFCM) devices that store information about one-year period.
energy consumption by January 2021 [13]. This information will In this context, the present study proposes a simulation-based
eventually be reported to the European Commission [14]. However, approach to study Europe’s CO2 and energy consumption gap. PyCSIS,
understanding the gap and its variability due to different factors will be a vehicle simulator suitable for fleet analysis [27], has been updated to
crucial to monitor the effectiveness of the CO2 standards [15]. The host all the powertrain configurations present in the European fleet and
influencing factors can be grouped in 4 main categories: the vehicle extensively validated versus real test cases with characteristics that
characteristics and systems; the environmental and traffic conditions; represent a wide range of the fleet. The simulator will construct a basis
driver and user-related factors; and vehicle certification test [16]. While for applying scenarios for the gap, as calculated over an one-year period;
the latter factor has been addressed in 2017 with the introduction of the for this reason, it was combined with a set of representative RW trips and
new test protocol, the Worldwide harmonised Light-vehicles Test Cycle literature on environmental factors and their effect on the consumption
and Procedure (WLTC and WLTP, respectively) [17], the rest still of the vehicle. The study consists of 3 steps: model validation to simulate
construct an essential research topic. modern European passenger cars and quantification of the certification
Experimental campaigns can accurately contribute to this research procedure impact; review and quantification of the impact of different
topic [18,19], but they are not cost-effective. Additionally, the outcome factors in the gap; and finally, fleet data collection and scenarios ap
can not always show the broader picture, and the findings can not safely plications to understand the variability reported by previous studies for
apply in another case study. Several other practices exist for assessing the European fleet. Insights on the simulation results intend to
the factors influencing energy performance and the gap in an organised contribute to the research and characterisation of the gap and its vari
way. Fontaras et al. [16], in an extensive literature review, presented the ability and provide realistic indicators of vehicle consumption and
parameters to consider during driving, and set indications for their RW driving range.
2
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Table 1 activities; they differ from the on-road trips used in the validation step
Technical characteristics of the ICEVs (1 to 6) and BEVs (7 to 10) tested at JRC. because they do not specifically fulfil legislative criteria. The results
Nomenclature for the engine types: CT: Compression turbo; PT: positive turbo; from the previous steps were afterwards used in the Fleet Projection
PNA: Positive natural aspiration. (Fig. 1, coloured in blue), in a representative share of vehicle models
Vehicle Vehicle Engine type [-], Battery Mass in Engine/ that recently entered the EU market. The vehicles were simulated ac
No. body [-] Engine capacity capacity running motors cording to the certification procedure and RW driving conditions. The
[lt], Gear box [Ah] order [kg] max power
in-use factors relations derived in the RW impact analysis were applied to
type – Gears [kW]
the fleet data to create different RW scenarios.
1 Hatch- CT, 1.5, Manual 1,255 96
back –6
2 Sedan CT, 1.6, 1,565 143 2.2. Vehicle data
Automatic – 9
3 Hatch- PNA, 1.4, 1,120 64
back Manual – 5
The following sections discuss details on the vehicle data used for the
4 SUV CT, 1.6, Manual 1,449 100 validation and the fleet projection. For the present study, although
–6 conventional and hybrid electric vehicles (HEVs) are handled differently
5 Van CT, 2.0, Manual 2,051 107 by the model, they are grouped under internal combustion engine ve
–6
hicles (ICEVs) for simplicity.
6 Sedan PT, 2.0, 1,615 135
Automatic – 8
7 Hatch- 125 1,700 96 2.2.1. Tested vehicles
back Six ICEVs, representative of specific segments of 2018 to 2020 Eu
8 Van 278 3,049 143 ropean fleet, were selected from a pool of vehicles tested on JRC pre
9 SUV 216 2,269 64
10 Sedan 278 1,999 100
mises. The cars were tested under a general JRC measurement campaign
[28–30], and the selection criteria for the vehicles include sales
numbers, manufacturer diversity, segments and technology market
2. Materials and methods coverage. Four BEVs from 2021, tested in JRC for their energy effi
ciency, were also used for the study. Although fewer BEVs were tested
The following chapter provides a detailed description of the mate compared to the ICEVs, the authors consider their number sufficient to
rials collected and the methodological steps proposed to study the CO2 cover their current fleet penetration. The main characteristics of the ten
and energy consumption gap and its variability. vehicles are summarised in Table 1.
Fig. 2. Type Approval vehicles characteristics of the subset used: (a) 74 ICEVs; (b) 14 BEVs.
3
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 3. Sales expanded vehicle characteristics of the selected fleet sample for the simulations: (a) 2.5 million ICEVs; (b) 0.3 million BEVs.
Table 2
On-road trips characteristics.
Trip 1 Trip 2 Trip 3 Trip 4 Trip 5 Trip 6 Trip 7
new passenger cars for the years 2018 and 2019 for ICEVs and 2019 for ICEVs, and the majority are small vehicles with engine capacity below
BEVs, developed and maintained by the European Environment Agency 1.4 lt. The European Automobile Manufacturers’ Association confirms
[33], were used as a reference to study the RW impact on the fleet. These these facts for 2009 to 2020 [35].
databases collect information needed to assess vehicle manufacturers’
compliance with their CO2 targets. In each studied year, around 15.5
million new cars were registered in the 27 Member states, which is 23.7 2.3. Trip data
% higher than the registrations during 2020 and onwards due to the
unprecedented consequences of the COVID-19 pandemic [34]. For each This section provides details of the trips performed during the
entry, official CO2 values (for ICEVs), battery energy consumption (for physical tests and used for the validation, and the trips adopted for the
BEVs), along with vehicle-specific information such as mass in running simulations of the RW scenarios. The main factors used for the trip
order, technical characteristics, and the number of registrations, are characterisation are the mean trip speed, accumulated distance, and
provided. maximum and minimum elevation.
The information included in the fleet database is not sufficient to
perform vehicle-specific modelling. Information regarding engine 2.3.1. Type approval test
characteristics, vehicle body shape, gearbox type, and battery capacity The gap experienced during the RW driving is studied against the
are critical and profoundly affect the selected model’s performance. In current EU certification protocol, the WLTP, performed on a test bench.
that aspect, data from the official database was combined with data The driving cycle lasts 30 min, and the distance driven is 23.25 km, with
retrieved from publicly available sources (i.e., online databases like car a mean speed of 46.5 km/h. The cell temperature is set to 23 ◦ C, and the
folio.com, cars-data.com, carspector.com, etc., and vehicle manufac vehicle is tested on cold start (its thermal state is at the same tempera
turers’ websites) used to formulate a more detailed database. More de ture as the cell). Resistance coefficients, a function of vehicle speed,
tails on the European Environment Agency database, the publicly need to be set as a target in the dynamometer for simulating realistic on-
provided vehicles’ specifications database, the combined database, and road resistances; these coefficients need to be derived experimentally in
all data treatment steps are provided in a previous study by Tsiakmakis test tracks, with the most frequent procedure being the coast-down test
et al. [17]. The combined database created consists of 993 ICEVs from [17]. During the coast down the vehicle is left to decelerate from a high
2018 and 574 from 2019, representing 12 % and 5 % of the 2018 and speed, and the deceleration intervals are transformed to forces resisting
2019 vehicles registered in the EU in the respective years, and 63 BEVs the motion.
representing 91 % of the electric vehicles that entered the EU roads on
2019. 2.3.2. On-road tests
Fig. 3 gives a visual representation of the vehicles’ technical con On-road trips were performed on vehicles tested at JRC, and they
figurations expanded for their EU sales. The study team simulated ve were used to validate the simulation tool under real driving conditions
hicles representing more than 2.8 million registrations in the European (extra driving resistances, HVAC operation). Table 2 presents some
market. The representation of both ICEVs and BEVs is considered indicative information of the trips used. The distance shares travelled in
satisfactory: Sport utility vehicles (SUVs) have the highest share for urban, rural, and highway driving, as defined in regulation [17], are
considered representative of real driving, and all vehicles were tested in
4
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 4. RW trips selection. Blue dots present the distance and average speed pairs of each recorded RWD trip in the campaign described in [6]. Pentagons present the
30 selected trips. Their colours characterise the trip speed categories they fall into: low, mean and high are coloured with green, black, and red, respectively. The
yellow petagon represents WLTC. The light blue vertical dashed lines separate the distance clusters.
5
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Table 4
Simulated versus measured consumption percentage error for ICEVs (vehicles 1 to 6), and BEVs (vehicle 7 to 10). Values with an asterisk (*) indicate the average error
from all the trips performed with a specific vehicle.
Fig. 5. Simulation error of the tested vehicles under the on-road trips versus average ambient temperature.
6
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 6. Simulated average gap from the vehicles tested by the JRC: (a) CO2 gap for ICEVs; (b) battery-electric consumption gap for BEVs. Red dots depict the 30
actual trips simulated. The background colouring represents the map produced by fitting in the simulated CO2 gap on the trips. Red shades represent a positive gap,
blue shades represent a negative one.
2.6.1. Validation
Table 6 Dyno test data following the WLTC velocity profile were collected
In-Use & ambient factors effect on CO2 emissions (ICEVs) and battery electric only for the JRC tested ICEVs, but on-road tests were obtained for both
consumption (BEVs). configurations: CO2 emissions were measured for the ICEVs (vehicles 1
Difference from Base-Case [%] to 6) with standard Portable Emissions Measurement Systems (PEMS),
In-use factor and battery and electric motor signals were collected through Controller
Values ICEVs BEVs
Area Network (CAN) protocol for BEVs (vehicles 7 to 10), supported by
No. of passengers [-] 1 − 0.94 − 0.8
power analysers for validation. Other signals were retrieved via the
1.45 (Base-case) 0 0
3 3.37 2.6 Onboard diagnostics (OBD) port for both ICEVs and BEVs, together with
Vehicle auxiliaries [kW] 0.3 − 2.89 − 5.0 instant position information, elevation, and ambient temperature
0.75 (Base-case) 0 0 through GPS. The vehicle speed was measured via OBD and GPS. The
1.5 4.91 8.5 GPS speed signal was overlapped with the OBD one for correcting the
Ambient temperature [oC] 5 2.39 1.2
14 (Base-case) 0 0
absolute speed values of the OBD signal with a quadratic function, and
23 − 2.7 − 1.1 the sampling frequency of all data was reduced to 1 Hz. The simulations
were performed using as inputs the specifications of each test, i.e., the
heater is used to ensure cabin thermal comfort in cold conditions. vehicle speed, road gradient, ambient temperature, and vehicle mass,
Europe’s largest motoring association, ADAC, reported a non-negligible including the PEMS instrumentation. Then, the simulated signals for the
average electric consumption of 2 to 3 kW for heating in extreme tem engine speed, engine coolant temperature, etc., were compared to the
peratures ranging from − 9 to − 14 ◦ C for two city-BEVs [48]. To this end, recorded.
the extensive evaluation of different HVAC systems on the energy con For the set of vehicle models collected from Type Approval data, the
sumption of BEVs, performed by Lajunen et al. [26], was used as a basis simulated certified values (CO2 emissions [g/km] or electric energy
for the implementation of a linear model for heating as a function of the consumption [kWh/km]), which include the regulation corrections to
ambient temperature. The consumption of the implemented model falls eliminate cycle flexibilities and battery energy balance, were compared
between a conventional HVAC system (A/C and Positive Temperature to the manufacturer declared values. This step defined the average dif
Coefficient (PTC) electric heater) and a more efficient system that ference between measured and declared values due to the procedural
combines Heat pump and PTC heater. Higher efficiency systems, as the change in WLTP: the manufacturer shall declare a value at least 1 %
Joule heating, were not considered [43]. More details on the HVAC higher than the measured [49]. Equation (1) presents the formula used
implementation and the vehicle information needed for the simulations for the calculation of error or divergence.
are presented in Appendix A. (
Simulated
)
Error = − 1 • 100 (1)
Target
7
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 7. Vehicle losses due to ambient conditions, HVAC auxiliaries, and their combined effect on the gap, as a function of ambient temperature. Blue lines indicate
the effect on ICEVs, and orange lines on BEVs.
Table 7
In-use factors percentage effect in consumption comparison to literature. T: Temperature.
Configuration Scenario Simulation Literature
Table 8
RW gap versus official values for the simulated fleet. Diesel: Diesel engine; Gas: Gasoline engine; Auto: Automatic gearbox; Manual: Manual gearbox.
Gap
Configuration Counts (sales in thousands) Mean (Sales weighted) STD Interquartile range
[-] [%] [%] [%]
km, which, according to the literature [50], falls inside the average is provided in Appendix B for better understanding. Then, the individual
distance a vehicle runs in a year period. As mentioned earlier, only the gap of each car model simulated was calculated according to the equa
30 selected RW trips and the WLTP profile were simulated for each tion (3). The certified consumption of each vehicle is the manufacturer’s
vehicle from the fleet. The influence of the remaining trips is assigned to declared value according to the WLTP.
each individual simulated RW trip as a weight for producing a weighted Gapind.vehicle = (RWconsumptionind.vehicle /Certifiedconsumptionind.vehicle − 1)
average RW consumption value per vehicle (equation (2)).
• 100
(3)
The average gap value for the 2018 and 2019 fleet datasets was
8
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
conditions.
In the final step, Monte-Carlo simulations were performed to calcu
late the gap and its dispersion separately for warm and cold weather
conditions. Replicating the whole year, the analysis in European coun
tries revealed average values for the warm (spring and summer) and the
cold (autumn and winter) months, but also their dispersion. Subse
quently, randomly generated temperatures were produced following
normal distributions for both periods. According to the randomly
generated temperatures, the HVAC consumption was calculated sepa
rately for ICEVs and BEVs. Regarding the number of passengers,
randomly generated weight for 1 (i.e. only the driver) to 5 (i.e. driver
plus four passengers) people was produced. The trip characteristics were
produced by generating a random pair of trip distance and average trip
velocity following the distributions of the distance and speed of the full
RW dataset, and the gap from the official values was produced by
feeding the pair of values to a gap map produced for each individual
vehicle.
3.1. Validation
Fig. 8. Range driven with one full battery charge (left y-axis) for different RW The following sections compare the simulation results to experi
trip distances (x-axis). Each vehicle is represented by multiple dots across a mental data and the certified values, as provided by the car
range of trip distances. The trip distances shown do not surpass the manufac manufacturers.
turer declared ranges. The dot colour indicates the manufacturers’ declared
WLTP range. The bars show the percentage of vehicles failing to complete the 3.1.1. Validation against test data
route with 1 full battery (right y-axis).
Table 4 presents the error between the simulated CO2 emissions for
ICEVs, and battery energy consumption for BEVs, versus the recorded
Table 9 values. For ICEVs, only vehicle 3 presented an error higher than 4 %
Details for the worst and best-case scenarios. during the WLTC. Οverall, the average error from simulation relative to
Passengers [-] Ambient Temperature [oC] Auxiliaries certification test results for the six ICEVs was equal to 1.54 %. Good
[kW] results also appeared for the on-road tests, where the individual average
ICEVs BEVs errors for all ICEVs did not exceed 4 %. The average error for the six
vehicles was equal to − 0.24 %. The average error for all BEVs was low
Worst case 5 − 1 1 2.2
Best case 1 23 0.75 0.75
(-1.1 % on average). Some high divergences (above ± 9 %) found in
specific tests were mainly attributed to low-quality recordings.
Fig. 5 presents the individual trip error for the on-road simulations as
calculated in two ways: once considering each individual vehicle
a function of ambient temperature due to its critical role in HVAC
simulated as one entry, and secondly, as a fleet average, where the sales
consumption and the cold-start (for ICEVs). 92.5 % (37 from the 40 trips
each models were considered, as provided in equation (4).
in total) of the simulations fell inside the ± 10 % error. Several pa
∑
SalesweightedaverageGap = (Gapind.vehicle(i) rameters during on-road tests were neglected or not considered in detail
∑ in the simulation process (e.g., exact road load coefficients, wind impact,
• Salesind.vehicle(i) )/ (Salesind.vehicle(i) )∀i
interaction with traffic). In addition, instruments providing vital signals
∈ 2018, 2019fleetdata (4) for the tool, like the GPS (e.g., vehicle speed, elevation) have limited
accuracy. The aforementioned can explain the error dispersion shown.
The RW impact outcome was initially used to find the average, or
The error was intensified in vehicle 7, and the authors attribute this
base-case gap in the fleet at 14 ◦ C ambient temperature and the weight of
error to the battery state of health since parallel physical consumption
1.45 passengers (i.e., 123 kg), as presented in Table 3. Since the vehicle
measurements with power analysers showed better agreement. No trend
efficiency and the auxiliaries’ consumption are dependent on the
as a function of the ambient temperature was observed, indicating the
ambient temperatures, European monthly ambient temperatures were
realistic assumptions done for the HVAC consumption and the efficient
collected from online sources [51], and combined with population in
capture of the cold-start effect.
formation to produce population-weighted values for a better repre
Regarding the cold-start effect and the performance in short trips for
sentation of the European environmental conditions.
the ICEVs, a separate evaluation was performed using data from a test
For the worst-case scenario, the weight of 5 passengers was consid
campaign in extreme laboratory conditions (more details in [44]). The
ered, while for the best-case the weight of one, the driver. The interac
simulated CO2 emissions showed good agreement with the experimental
tion between the in-use factors auxiliaries and ambient temperature
ones. The tool closely followed the emitted CO2 for a test repeated in
through the HVAC consumption does not allow for a straightforward
ambient temperatures of − 10 ◦ C, 5 ◦ C and 23 ◦ C, and the same initial
assignment of exact values to the worst and best cases. For example, the
temperature for the vehicle state (the reader will find further details in
impact of the cold-start of ICEVs in cold conditions needs to be compared
Appendix C). Furthermore, the tool’s sensitivity to capture the labora
to the impact of the A/C in hot conditions to see by which factor the
tory versus on-road variability was analysed for all ICEVs for which both
consumption is influenced the most in different ambient temperatures.
WLTP and on-road tests were performed. The average error calculated
Thus, the details of the two scenarios will be presented in the results
was 2.2 %. More details for the above assessment are provided in Ap
section, after the quantification of the importance of the auxiliaries and
pendix D. Overall, considering the limited input data for each vehicle,
the ambient temperature. This information will be later combined with
and the fact that there was no artificial model calibration to minimise
indicative ambient temperatures for average cold and hot European
the error of specific vehicles, the tool corresponds with satisfactory
9
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 9. Fleet models gap distributions for the base- (light blue), worst- (red), and best- (green) case scenarios; (a) ICEVs; (b) BEVs.
Table 10
Range of the gap, defined by the 1st quartile of the best- case (lowest value) and
Table 11
the 3rd quartile of the worst-case (highest value).
Simulations on the gap using various numbers of trips with a combination of
Conventional ICEVs Electrified ICEVs (Hybrids) BEVs randomly produced trip characteristics.
Gasoline Diesel Cold period gap [%] Warm period gap [%]
Range [%] 12.5 – 41 7 – 31 27 – 55 − 6.5 – 26.5 No of trips mean STD mean STD
Interval 27.5 24 28 33
ICEV 250 ~25.5 7.9 ~23.5 8
50 8.7 8.5
accuracy in all conditions for a wide range of vehicle models. 25 8.8 9.2
5 13.8 13.8
BEV 250 ~12.5 2.8 ~5 2.8
3.1.2. Validation against official data
50 3.8 3.9
Pavlovic et al. [11] estimated a 4–5 % over-declaration in official CO2 25 4 4.3
values, which was also proven by an analysis performed in 221 Type 5 9.8 10
Approval documents by Chatzipanagi et al. [32]. Thus, an underesti
mation (according to equation (1)) in the order of 5 % was expected in
the simulated values based on the manufacturer’s declared ones.
Table 12
Table 5, which presents the comparison statistics, shows a clear shift
Variability of annual distance driven depending to the number of different route
of the results towards negative values (i.e. the simulated results are
profiles.
lower than the official values) that were not observed in the tool vali
dation versus the measured consumption during WLTC. The difference’s No of trips 250 50 25 5
standard deviation (STD) is 6.3 % for ICEVs which is considered satis Annual distances min–max (x 1000 11 – 18 10 – 20 7.5 – 25 2.5 – 35
factory, while BEVs show higher scattering (10.9 %). Although this step km)
10
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
analysis of the official versus RW CO2 and energy consumption gap. 3.3.1. Base-case scenario
The overall gap for all simulated ICEV models in the average base-
3.2. Real-world influence case conditions was 18 %. The overall sales-weighted mean gap was
16.5 %. The gap calculated is in agreement with the fuel consumption
Having proven the accuracy of the simulation tool, the influence of gap of 20 % (adjusted to the WLTP), reported in other studies [6,54]. In a
on-road parameters are presented in the next two sections. study by Ktistakis et al. [55] assessing three different databases, the CO2
gap ranged from 25 % to 39 %. The divergence was due to the share of
3.2.1. Trip effect PHEVs that increases the CO2 gap.
Fig. 6 presents the average gap due to the trip characteristics. It is BEVs average gap was − 2.6 %, ranging from − 4.6 % to − 0.3 %.
separated in ICEVs and BEVs because the two powertrain configurations Considering the EU 2019 sales, the average gap was − 0.8 %. Fiori et al.
show different trends. The increased stop & go regimes in the short [56], in measurements in six different legislative laboratory cycles,
distance trips, combined with the engine idling and the high shares in showed that the gap compared to WLTP consumption ranged from − 30
the trip duration, where the cold-start period occurs, resulted to the % to 12 %, identifying a strong correlation with the cycle energy de
highest positive gaps for the six ICEVs. In average speeds between 40 mand. This confirms the slightly negative gap in some of the simulations:
and 80 km/h, the lowest gap was observed: there were trips where the the majority of the trips used in the analysis had lower trip speed and
gap turned negative, i.e., the CO2 emissions were lower than what was distance than WLTP, and according to the assessment of the cycle energy
simulated for WLTC velocity profile. In higher speeds, CO2 emissions demand in different RW profiles performed by Komnos et al. [36], lower
increased again due to excessive aerodynamic losses. This overall trend trip speed equals lower overall cycle energy demand.
is consistently mentioned in the literature as the ICEV U-shape [24]. The More details for both ICEVs and BEVs are presented in Table 8. Diesel
average gap calculated for the trips with average speed below the vehicles appeared to have a lower average gap than gasoline ones. The
average WLTP (46.5 km/h) was 13 %, while for higher speeds it was 7 reason is the lower impact of cold-start calculated for diesel in short-
%. distance trips. This outcome is supported by Pielecha et al. [57] who
The four BEVs completed all the RW trips with only one full battery showed that the gasoline configuration emitted two times more than the
charge in the simulation. The gap was monotonic as a function of the diesel during the cold-start for the same brand model. In trips with
average speed; thus, trips with average speeds below the WLTP had a higher duration, the gap found in diesel vehicles was similar to the
negative average gap (-7%), while for higher trip speeds, it was 20 %. gasoline ones. For conventional ICEVs, the range of the gap (1st to 3rd
The overall trend agrees with observations of Mamikoglu et al. [26], quartiles) was 14 %–22 %, while for electrified ICEVs (Hybrids), the
which reported better efficiency in low-speed conditions like city cycles range was 31.5 %-36 %. This 10 % higher gap between hybrids and
compared to cycles with higher speeds. conventional vehicles is supported by an International Council of Clean
Transportation study [58] that assessed user-communicated consump
3.2.2. Effect of In-Use factors tion. A detailed assessment of the hybrid simulation outputs revealed
Table 6 presents the average percentage gap produced in the 30 RW that this extra gap resulted from the hybrid controller that promoted the
trips. The impact of passengers is limited for both configurations, but drive battery charging during trips with high dynamicity, resulting in
especially for BEVs that are generally heavy vehicles due to the fitted higher fuel consumption. This phenomenon was observed in a few short-
batteries. The highest impact was observed when tuning the value of distance trips; their higher influence in the RW weighted average
auxiliaries consumption, and BEVs had almost double the impact intensified the impact in the gap. More details are presented in Appendix
compared to ICEVs. This is because the auxiliary demand in low-speed F.
trips, that are more common in RW driving, is comparable to the Regarding the influence of the trip characteristics in the driving
battery-electric energy consumption for the vehicle propulsion. In range of a BEV, Fig. 8 shows that for trips longer than 100 km, several
regards to the ambient temperature, ICEVs had a greater impact models from the market failed to complete the route with one full
attributed mainly to the cold-start operation. Another reason was the charge, even if the manufacturers’ declared range was higher than the
better aerodynamics of BEVs [52], resulting to a lower impact of the trip distance. This was intensified in trips with higher type-approved
reduced air density compared to ICEVs. ranges: one charge was insufficient for 36 % of the models with a
Fig. 7 shows the individual and the combined effect of the HVAC- manufacturer- stated range between 230 and 320 km.
related auxiliaries, and the vehicle losses, due to the ambient tempera
ture. The gap was higher during cold conditions (ambient temperatures 3.3.2. Best and worst-case scenario
below 14 ◦ C) than warmer conditions (temperatures above 14 ◦ C) for The analysis of ambient European temperatures indicated the
both ICEVs and BEVs, and the best efficiency appeared in an ambient extreme value for the average cold conditions was − 1 ◦ C and for hot
temperature of 21 ◦ C. The impact of the ambient temperature was higher conditions, 23 ◦ C. According to the assessment of the in-use factors, the
in the case of BEVs because the energy needed for the HVAC was com cold conditions were used as the worst case and the hot conditions as
parable to the energy required for the propulsion, especially for the low- best-case scenario. Table 9 presents further details of the values
speed trips. considered. For the worst case, the auxiliaries from ICEV base-case were
Although there are methodological differences, and direct compari increased by 0.25 kW, and for BEVs by 0.2 kW to account for extra
sons with the literature cannot be made, Table 7 shows that the influ infotainment.
ence calculated for different scenarios is similar to studies by Zacharof Fig. 9 presents the updated gap results produced by the adjusted RW
et al. [20], Fiori et al. [53] and Lajunen et al. [26]. This gives confidence weighted-average consumption for each model’s worst- and best-case
both in the accuracy of the tool and in the methodology used to apply scenario versus its simulated certified value. The best- and worst-case
realistic projections of the influence of the in-use factors in RW average gap for ICEVs was 13.8 % and 34.5 %, respectively. These
conditions. values, translated to the 2018 and 2019 model sales, changed to 12.5 %
and 33 %, respectively. For BEVs the average values for the models
3.3. Full fleet scenario results simulated were − 4.5 % and 23.9 % for best and worst-case scenario,
which translates to − 2.7 % and 26.1 % average gaps considering the
This chapter presents the assessment of the gap based on the fleet sales numbers. The difference between the best-case and the base-case
data (equation (4)). The simulated Type Approval figures were increased scenario was more profound for ICEVs (4.2 %) than BEVs (1.9 %). The
by a constant value of 6 % to account for the inflation due to the certi difference can be attributed to the engine cold-start period, which was
fication procedure, as calculated by the Type Approval assessment. not included in the case of the BEVs. In contrast, the worst-case scenario
11
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
is more evident for BEVs, calculated to be 26.5 %, while for ICEVs it was influence. When combining factors that are influenced by ambient
16.5 %. The reason for the difference was the extra consumption for temperature, the cold conditions appear to have a greater impact on the
cabin heating needed in BEVs that comes for free in ICEVs. gap than warm conditions, for all powertrain configurations tested. For
Τhe range of the gap, defined as the 1st quartile of the best-case up to the average European conditions, the gap of RW driving versus certified
the 3rd quartile of the worst-case, for different vehicle configurations is figures was found to be 16.5 % for ICEVs, and − 0.8 % for BEVs.
presented in Table 10. Diesel vehicles showed a lower range but also gap Considering the best- and worst- case scenarios for Europe, the gap from
interim compared to gasoline. BEVs showed the highest range and cor the certified values range was 12.5 – 41 % for gasoline vehicles, 7 – 31 %
responded well with the suggested range by De Cauwer et al. [27], for diesel conventionals, 27 – 55 % for HEVs, and − 6.5 – 26.5 % for
particularly 10 to 40 % (adjusted to WLTP). BEVs. Regarding the values gap distribution of the fleet, a Monte-Carlo
simulation approach revealed that driving in a limited range of trip
3.3.3. Monte-Carlo simulations characteristics –mainly city routes– resulted in higher gap dispersity, but
The analysis of the temperatures of European countries showed that the average fleet values change was limited.
the average temperature was 14 ◦ C (STD ± 4 ◦ C) and 7 ◦ C (STD ± 5 ◦ C) The results of the present study represent indicative average Euro
in warm and cold conditions, respectively. The cold temperatures pean scenarios. The same approach and simulation tools could be
reached values down to − 5◦ C, and the warm ones up to 30 ◦ C. The adopted to assess different scenarios, more focused on specific fleet
validation of the online temperature database, examples of the method compositions, conditions, and driving patterns, to support future pol
for the random in-use factors values generation and trip characteristics icies scientifically. Future work should focus on further analysing the
that influence the gap per vehicle are presented in Appendix G. parameters affecting the gap at a vehicle level, improving the model to
To replicate a full year, similar to the 483 trips that corresponded to capture new technologies and their effectiveness on-road; and finally, on
11 months, 500 random values were produced per vehicle model for providing insights on the range and potentials of BEVs that are becoming
different combinations of number of passengers and ambient tempera one of the main alternatives for meeting the CO2 targets set.
tures, split in 250 warm and 250 cold conditions. The HVAC auxiliaries
were calculated according to the ambient temperatures. Four scenarios Authors contributions
(for 250, 50, 25, and 5 trips) were simulated for each type of powertrain
to understand the gap variability. For each scenario, a random combi The authors, Dimitrios Komnos (DK), Stefanos Tsiakmakis (ST) Jelica
nation of trip characteristics that could be driven throughout a year was Pavlovic (JP), Leonidas Ntziachristos (LN) and Georgios Fontaras (GF)
combined with the random in-use factors for the warm and the cold contributed to the paper as follows:
period. When moving from 250 random trip characteristics to 5, the Study conceptualisation: ST, DK, GF, LN; data collection, curation,
mean gap remains about the same but the STD significantly increases and analysis: DK, ST, JP; model simulations: DK, ST; analysis and
(Table 11). interpretation of results: DK, ST, GF; manuscript preparation: DK, ST;
This showed that a low variability in the driving patterns spread the revision and comments: JP, LN, GF; research coordination: GF. All au
fleet gap dispersion. For example, a specific vehicle model driven mainly thors reviewed the results and approved the final version of the
on a motorway would have a different gap than when driven on city manuscript.
roads. This can be depicted in the kilometres driven annually according
to the four scenarios (Table 12). Declaration of Competing Interest
The variability in the mean gap values in the four scenarios was very
low because the high number of vehicle simulations converged to The authors declare that they have no known competing financial
similar values. However, compared to the base-case, it is interesting that interests or personal relationships that could have appeared to influence
both cold and warm periods showed higher average gaps: 7.5 and 5.5 % the work reported in this paper.
higher for ICEVs, and 15 and 7.5 % for BEVs. For ICEVs, the difference
can be attributed to the sort distance trips and the cold-start period, Data availability
while for BEVs, it can be assigned to HVAC.
Data will be made available on request.
4. Conclusions
Acknowledgements
Understanding the effect of the factors that influence real-world
consumption is essential for proper policy implementation. The The authors would like to acknowledge the JRC Vehicle Emissions
distinct impact of different factors influencing the gap on CO2 emissions Laboratory (VELA) team, and Dr Alessandro Tansini for carrying out the
and energy consumption has been extensively studied. However, under experimental activities, and providing the experimental data. The au
RW conditions, little is known on the combined effect of these factors on thors would also like to thank Mr Francois Filip, Mr Serra Carlos and Ms
the gap in a representative sample of the European fleet. In this context, Lakatos Katalin for their constructive review.
the present study proposed an approach based on a detailed simulation
tool, PyCSIS, for estimating fleet vehicle consumption. Τhe accuracy of Appendix A. . Simulation tool inputs and HVAC model
the tool to reproduce the recorded consumption during RW driving was
satisfactory, with 92.5 % of the simulated values falling between ± 10 % Table 13 lists the main inputs required from the user and Fig. 10 the
of the recorded. PyCSIS was subsequently combined with literature and HVAC model consumption as function of external ambient temperature.
other data sources to provide a platform for quantifying the gap between
the RW consumption and the manufacturer-declared values for vehicle
configurations added in the EU market during 2018 and 2019. Results
are in line with prior studies, suggesting a 6 % certification procedure
12
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Table 13
Inputs for simulation tool.
Input Unit Format
General inputs
Mass in running order [kg] Float
Tyre code [-] String
Vehicle body [-] String
Frontal area [m2] Float
Auxiliary losses [kW] Float
Ambient air density [kg/m3] Float
Elevations [m] [Float, Float, …]
Velocity profile (RW or official cycle) [km/h] [Float, Float, …]
Full HEV configuration [-] Boolean
PHEV configuration [-] Boolean
Electric configuration [-] Boolean
ICEV inputs
Engine capacity [cc] Float
Fuel type [-] String
Gear box type [-] String
Engine max power [kW] Float
Engine speed at max power [RPM] Float
Gear box ratios [-] [Float, Float, …]
Final drive ratio [-] Float
BEV inputs
Drive battery voltage [V] Float
Drive battery capacity [Ah] Float
Drive battery initial SOC [%] Float
Electric motor max torque [Nm] Float
Electric motor max power [kW] Float
Fig. 10. HVAC consumption as function of the average ambient temperature for ICEVs (blue line), and BEVs (orange line). For comparison, dots are added to present
literature findings. Nomenclature: HP: Heat pump; PTC: Positive Temperature Coefficient heater.
13
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Appendix B. . Graphical example of the real-world trips accumulated CO2 of a test repeated in ambient temperatures of − 10 ◦ C,
clustering 5 ◦ C, and 23 ◦ C, as function of time for the first 10 min of operation. The
simulation tool show satisfying correspondence, and overall, it is able to
Fig. 11 presents an example of the K-Means clustering performed and differentiate the 3 different test conditions. The same results show evi
the weights, wtrip, derivation. The RW simulated value of each vehicle in dence that the tool is capable of accurately simulating not only long
one of the 30 RW trips is fed together with the wtrip to the equation (4). trips, where different errors can compasate, but also short trips.
Appendix C. . Internal combustion engine old-start model Appendix D. . On-road versus laboratory assessment
assessment
Fig. 13 shows the gap between laboratory (WLTP) and on-road
The on-road tests used for the ICEV were starting with the vehicle driving as derived by the simulations and the experimental data. The
thermal state always between 20 ◦ C and 23 ◦ C. That is why the cold-start average error inserted between lab and on-road is 2.2 %. Considering
operation was assessed separately in additional laboratory tests per that literature suggests a gap in the order of 20 % or above [55,59,60],
formed to assess the influence of extreme temperatures in the emissions the authors consider these results adequately accurate and sensitive.
of a Euro 6d-temp gasoline direct injection. Fig. 12 shows the
Fig. 12. Model cold-start assessment in laboratory conditions for the urban traffic TfL (Transport for London urban interpeak). The first 10 min are assessed.
14
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 13. Tested ICEVs – CO2 emissions gap: Mean Simulated vs Real CO2 gaps per vehicle.
Appendix E. . Type Approval vehicles error distributions manufacturer declared value. In the first 600 s (10 min) that the cold-
start is present, the gap is considerably higher, especially for the gaso
Fig. 14 presents the distribution of the percentage difference be line vehicles.
tween the simulated type approval values versus the manufacturers’ Fig. 16 presents a comparison between similar models of the same
declared ones. The result of around − 6% for both ICEVs and BEVs brand, one conventional and one electrified ICEV. The simulations of
confirms literature suggestions on the impact of the certification pro electrified version show that in general it consumed significantly less,
cedure. The scattering of the values for both ICEVs and BEVs resemble except of some cases where, the hybrid controller promoted the driver
normal distributions. battery charging using the ICE, leading to increased fuel consumption.
Fig. 14. Type Approval Data: simulation error distribution for 74 ICEVs (shades of blue) and 14 BEVs (shades of orange). The vertical lines demonstrate the average
errors The background colour a normal distribution produced by the error mean and STD values.
Appendix F. . Individual real-world trip gap The RW gap, that previously presented to be 20 % and 32 % for con
ventional and electrified gasoline ICEVs, respectively, not considering
Fig. 15 presents the gap for each one of the 30 RW simulated trips. As the RW weighted average is 13 % and 15 %.
shown, the lower the trip duration, the higher the gap versus the
15
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 15. (a) Fleet average ICEVs CO2 gap as a function of the trip duration for various subcategories. (b) Fleet average gap difference between gasolines and diesels as
function of trip duration.
Fig. 16. (a) Comparison of the CO2 emissions of one conventional and one electrified ICEV, as function of trip distance. The horizontal dashed lines are the mean CO2
emissions from all the 30 RW trips. (b) Drive battery delta SOC (final minus initial) for the electrified ICEV, as function of trip distance. The red area indicates the
trips for which intensive drive battery recharge occurred.
Appendix G. . In-use factors and trip characteristics random together with EU countries average temperatures. Distributions were
distributions constructed by providing the mean temperatures for the EU warm, and
cold periods (Table 14). A comparison between the RW ambient values
The available information regarding the ambient temperature of the in Northern Italy and the distributions created by average EU monthy
483 RW trips was not adopted because the majority of the trips were temperatures is also provided. The dispersion of the RW values are
performed in Northern Italy, so they could not fully resemble average EU higher than EU average, since the later ones correspond to month av
temperatures. Still, the RW trip average ambient recordings were split in erages. The validity of the online database [51] is confirmed on Lom
warm (spring and summer) and cold (autumn and winter) periods to bardia region (Northern Italy), by comparing to the recorded RW
understand the temperature distributions. The analysis revealed normal average values in 2017, and ASAP, a tool produced by European Com
distributions for both warm and cold periods. This information was used mission [61,62].
16
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
Fig. 18. Method for introducing randomness in the trip characteristics influence. The trip distance and the average speed are produced randomly following the
gamma distributions fitted in the full RW dataset containing 483 trips (top figures). The pair of distance and speed is fed in the individual vehicle gap map produced
by fitting in the 30 simulated gap values (bottom figure).
17
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
References [22] Zachiotis AT, Giakoumis EG. Non-regulatory parameters effect on consumption
and emissions from a diesel-powered van over the WLTC. Transportation Research
Part D: Transport and Environment 2019;74:104–23. https://doi.org/10.1016/j.
[1] Cabuzel T. EU climate action and the European Green Deal. Climate Action -
trd.2019.07.019.
European Commission 2019. https://ec.europa.eu/clima/policies/eu-climate-
[23] Zachiotis A, Giakoumis E. Methodology to Estimate Road Grade Effects on
action_en (accessed April 28, 2021).
Consumption and Emissions from a Light Commercial Vehicle Running on the
[2] Transport could burn up the EU’s entire carbon budget. International Council on
WLTC Cycle. J Energy Eng 2020;146. https://doi.org/10.1061/(ASCE)EY.1943-
Clean Transportation 2021. https://theicct.org/transport-could-burn-up-the-eus-
7897.0000694.
entire-carbon-budget/ (accessed July 5, 2022).
[24] Fiori C, Arcidiacono V, Fontaras G, Makridis M, Mattas K, Marzano V, et al. The
[3] Is Europe reducing its greenhouse gas emissions? — European Environment
effect of electrified mobility on the relationship between traffic conditions and
Agency n.d. https://www.eea.europa.eu/themes/climate/eu-greenhouse-gas-
energy consumption ☆. Transportation Research Part D Transport and
inventory (accessed June 29, 2022).
Environment 2018;69:275–90. https://doi.org/10.1016/j.trd.2018.11.018.
[4] European Commission. Regulation (EU) 2019/631 of the European Parliament and
[25] Mamarikas S, Doulgeris S, Samaras Z, Ntziachristos L. Traffic impacts on energy
of the Council of 17 April 2019 setting CO2 emission performance standards for
consumption of electric and conventional vehicles. Transportation Research Part D:
new passenger cars and for new light commercial vehicles, and repealing
Transport and Environment 2022;105:103231. https://doi.org/10.1016/j.
Regulations (EC) No 443/2009 and (EU) No 510/2011 (Text with EEA relevance.).
trd.2022.103231.
2019.
[26] Lajunen A. Energy Efficiency and Performance of Cabin Thermal Management in
[5] Tietge U. From laboratory to road: A 2018 update of official and “real-world” fuel
Electric Vehicles. Warrendale, PA: SAE. International 2017. https://doi.org/
consumption and CO2 values for passenger cars in Europe n.d.:56.
10.4271/2017-01-0192.
[6] Pavlovic J, Fontaras G, Ktistakis M, Anagnostopoulos K, Komnos D, Ciuffo B, et al.
[27] Tsiakmakis S, Fontaras G, Ciuffo B, Samaras Z. A simulation-based methodology for
Understanding the origins and variability of the fuel consumption gap: lessons
quantifying European passenger car fleet CO2 emissions. Appl Energy 2017;199:
learned from laboratory tests and a real-driving campaign. Environ Sci Eur 2020;
447–65. https://doi.org/10.1016/j.apenergy.2017.04.045.
32:53. https://doi.org/10.1186/s12302-020-00338-1.
[28] Valverde V, Clairotte M, Bonnel P, Giechaskiel P, Carriero M, Otura M, et al. Joint
[7] Cihat Onat N, Aboushaqrah NNM, Kucukvar M, Tarlochan F, Magid HA. From
Research Centre 2018 light-duty vehicles emissions testing: contribution to the EU
sustainability assessment to sustainability management for policy development:
market surveillance: testing protocols and vehicle emissions performance. LU:
The case for electric vehicles. Energy Convers Manage 2020;216:112937. https://
Publications Office of the European Union; 2019.
doi.org/10.1016/j.enconman.2020.112937.
[29] Clairotte M, Valverde V, Bonnel P, Giechaskiel P, Carriero M, Otura M, et al. Joint
[8] Average CO2 emissions from new passenger cars registered in Europe decreased by
Research Centre 2019 light-duty vehicles emissions testing: contribution to the EU
12% in 2020 and the share of electric cars tripled as new targets started applying n.
market surveillance : testing protocols and vehicle emissions performance. LU:
d. https://ec.europa.eu/clima/news-your-voice/news/average-co2-emissions-new-
Publications Office of the European Union; 2020.
passenger-cars-registered-europe-decreased-12-2020-and-share-electric-cars-2021-
[30] Bonnel P, Clairotte M, Cotogno G, Gruening C, Loos R, Manara D, et al. European
06-29_en (accessed April 13, 2022).
market surveillance of motor vehicles Results of the 2020–2021. European
[9] Mamikoglu S, Andric J, Dahlander P. Impact of Conventional and Electrified
Commission Vehicle Emissions Testing Programme 2022. https://doi.org/
Powertrains on Fuel Economy in Various Driving Cycles. Warrendale, PA: SAE.
10.2760/59856.
International 2017. https://doi.org/10.4271/2017-01-0903.
[31] ETAES n.d. https://www.etaes.eu/ (accessed December 10, 2021).
[10] De Cauwer C, Maarten M, Heyvaert S, Coosemans T, Van Mierlo J. Electric Vehicle
[32] Chatzipanagi A, Pavlovic J, Fontaras G, Komnos D. Impact of WLTP introduction
Use and Energy Consumption Based on Realworld Electric Vehicle Fleet Trip and
on CO2 emissions from M1 and N1 vehicles: evidence from type approval and 2018
Charge Data and Its Impact on Existing EV Research Models. World Electric Vehicle
EEA data. LU: Publications Office; 2020.
Journal 2015;7:436–46. https://doi.org/10.3390/wevj7030436.
[33] Monitoring of CO2 emissions from passenger cars – Regulation (EU) 2019/631 —
[11] Paffumi E, Otura M, Centurelli M, Casellas R, Brenner A, Jahn S. Energy
European Environment Agency n.d. https://www.eea.europa.eu/data-and-maps/
Consumption, Driving Range and Cabin Temperature Performances at Different
data/co2-cars-emission-20 (accessed October 7, 2021).
Ambient Conditions in Support to the Design of a User-Centric Efficient Electric
[34] Passenger car registrations: -23.7% in 2020; -3.3% in December. ACEA - European
Vehicle: the QUIET Project n.d.:18.
Automobile Manufacturers’ Association 2021. https://www.acea.auto/pc-
[12] One goal, multiple pathways: A review of approaches for transferring on-board fuel
registrations/passenger-car-registrations-23-7-in-2020-3-3-in-december/ (accessed
consumption meter data to the European Commission | International Council on
April 11, 2022).
Clean Transportation n.d. https://theicct.org/publications/transferring_obfcm_
[35] New passenger cars by segment in the EU. ACEA - European Automobile
fuel_data_ec (accessed October 15, 2021).
Manufacturers’ Association 2021. https://www.acea.auto/figure/new-passenger-
[13] CABUZEL T. CO₂ emission performance standards for cars and vans. Climate Action
cars-by-segment-in-eu/ (accessed January 13, 2022).
- European Commission 2017. https://ec.europa.eu/clima/policies/transport/
[36] Komnos D, Tsiakmakis S, Pavlovic J, Laverde Marín A, Chatzipanagi A, Fontaras G.
vehicles/regulation_en (accessed June 14, 2021).
An Analysis of Modern Vehicle Road Loads for Fleetwide Energy Consumption
[14] Ktistakis MA, Pavlovic J, Fontaras G. Developing an optimal sampling design to
Modelling. Warrendale, PA: SAE. International 2021. https://doi.org/10.4271/
monitor the vehicle fuel consumption gap. Science of The Total Environment 2022:
2021-24-0080.
154943. 10.1016/j.scitotenv.2022.154943.
[37] Occupancy rates of passenger vehicles — European Environment Agency n.d.
[15] European Commission. Commission Implementing Regulation (EU) 2021/392 of
https://www.eea.europa.eu/data-and-maps/indicators/occupancy-rates-of-
4 March 2021 on the monitoring and reporting of data relating to CO2 emissions
passenger-vehicles/occupancy-rates-of-passenger-vehicles (accessed September 24,
from passenger cars and light commercial vehicles pursuant to Regulation (EU)
2021).
2019/631 of the European Parliament and of the Council and repealing
[38] Gil-Sayas S, Komnos D, Lodi C, Currò D, Serra S, Broatch A, et al. Analysing the
Commission Implementing Regulations (EU) No 1014/2010, (EU) No 293/2012,
potential of a simulation-based method for the assessment of CO2 savings from eco-
(EU) 2017/1152 and (EU) 2017/1153 (Text with EEA relevance). 2021.
innovative technologies in light-duty vehicles. Energy 2022;123238. https://doi.
[16] Fontaras G, Zacharof N-G, Ciuffo B. Fuel consumption and CO2 emissions from
org/10.1016/j.energy.2022.123238.
passenger cars in Europe – Laboratory versus real-world emissions. Prog Energy
[39] Fontaras G, Rexeis M, Hausberger S, Kies A, Hammer J, Schulte L-E, et al.
Combust Sci 2017;60:97–131. https://doi.org/10.1016/j.pecs.2016.12.004.
Development of a CO2 certification and monitoring methodology for Heavy Duty
[17] European Commission. Commission Regulation (EU) 2017/1151 of 1 June 2017
Vehicles - Proof of Concept report. 2014.
supplementing Regulation (EC) No 715/2007 of the European Parliament and of
[40] Tsiakmakis S, Fontaras G, Dornoff J, Valverde V, Komnos D, Ciuffo B, et al. From
the Council on type-approval of motor vehicles with respect to emissions from light
lab-to-road & vice-versa: Using a simulation-based approach for predicting real-
passenger and commercial vehicles (Euro 5 and Euro 6) and on access to vehicle
world CO2 emissions. Energy 2019;169:1153–65. https://doi.org/10.1016/j.
repair and maintenance information, amending Directive 2007/46/EC of the
energy.2018.12.063.
European Parliament and of the Council, Commission Regulation (EC) No 692/
[41] Ntziachristos L, Gkatzoflias D, Kouridis C, Samaras Z. COPERT: A European Road
2008 and Commission Regulation (EU) No 1230/2012 and repealing Commission
Transport Emission Inventory Model. In: Athanasiadis IN, Rizzoli AE, Mitkas PA,
Regulation (EC) No 692/2008 (Text with EEA relevance). 2017.
Gómez JM, editors. Information Technologies in Environmental Engineering,
[18] Suarez-Bertoa R, Valverde V, Clairotte M, Pavlovic J, Giechaskiel B, Vicente F,
Berlin, Heidelberg: Springer; 2009, p. 491–504. 10.1007/978-3-540-88351-7_37.
et al. On-road emissions of passenger cars beyond the boundary conditions of the
[42] Duarte GO, Gonçalves GA, Baptista PC, Farias TL. Establishing bonds between
real-driving emissions test. Environ Res 2019;176:108572. https://doi.org/
vehicle certification data and real-world vehicle fuel consumption – A Vehicle
10.1016/j.envres.2019.108572.
Specific Power approach. Energy Convers Manage 2015;92:251–65. https://doi.
[19] Mock P. Real-Driving Emissions test procedure for exhaust gas pollutant emissions
org/10.1016/j.enconman.2014.12.042.
of cars and light commercial vehicles in Europe | International Council on Clean
[43] Aragonés VG, Galiana BG, Diéguez CB, de Dios ÁM, Calabuig D, Cabanes A. Low
Transportation n.d. https://theicct.org/publications/real-driving-emissions-test-
energy heating system based on Joule effect: JOSPEL project. Twelfth International
procedure-exhaust-gas-pollutant-emissions-cars-and-light (accessed October 15,
Conference on Ecological Vehicles and Renewable Energies (EVER) 2017;2017:
2021).
1–3. https://doi.org/10.1109/EVER.2017.7935953.
[20] Zacharof NG, Fontaras G, Ciuffo B, Tsiakmakis S, Anagnostopoulos K, Marotta A,
[44] Giechaskiel B, Komnos D, Fontaras G. Impacts of Extreme Ambient Temperatures
et al. Review of in use factors affecting the fuel consumption and CO2 emissions of
and Road Gradient on Energy Consumption and CO2 Emissions of a Euro 6d-Temp
passenger cars. LU: Publications Office of the European Union; 2016.
Gasoline Vehicle. Energies 2021;14:6195. https://doi.org/10.3390/en14196195.
[21] Zachiotis AT, Giakoumis EG. Monte Carlo Simulation Methodology to Assess the
[45] Fayazbakhsh MA, Bahrami M. Comprehensive Modeling of Vehicle Air
Impact of Ambient Wind on Emissions from a Light-Commercial Vehicle Running
Conditioning Loads Using Heat Balance Method. Warrendale, PA: SAE.
on the Worldwide-Harmonised Light-Duty Vehicles Test Cycle (WLTC). Energies
International 2013. https://doi.org/10.4271/2013-01-1507.
2021;14:661. https://doi.org/10.3390/en14030661.
18
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161
[46] Lee J, Kim J, Park J, Bae C. Effect of the air-conditioning system on the fuel [60] Real-world vehicle fuel consumption gap in Europe is stabilising | International
economy in a gasoline engine vehicle. Proceedings of the Institution of Mechanical Council on Clean Transportation n.d. https://theicct.org/news/real-world-vehicle-
Engineers, Part D: Journal of Automobile Engineering 2013;227:66–77. https:// fuel-consumption-gap-europe-stabilizing (accessed October 4, 2021).
doi.org/10.1177/0954407012455973. [61] ASAP - EC n.d. https://mars.jrc.ec.europa.eu/asap/index.php (accessed January
[47] Mebarki B, Belkacem D, Allaoua B, Rahmani L, Benachour E. Impact of the Air- 10, 2022).
Conditioning System on the Power Consumption of an Electric Vehicle Powered by [62] Rembold F, Meroni M, Urbano F, Csak G, Kerdiles H, Perez-Hoyos A, et al. ASAP: A
Lithium-Ion Battery. Hindawi Publishing Corporation Modelling and Simulation in new global early warning system to detect anomaly hot spots of agricultural
Engineering, Article ID 935784, 2013;Volume 2013:6 pages. 10.1155/2013/ production for food security analysis. Agric Syst 2019;168:247–57. https://doi.
935784. org/10.1016/j.agsy.2018.07.002.
[48] Tipps für mehr Reichweite bei Kälte n.d. https://www.adac.de/rund-ums-
fahrzeug/elektromobilitaet/info/elektroauto-reichweite-winter/ (accessed
January 18, 2022). Glossary
[49] Pavlovic J, Ciuffo B, Fontaras G, Valverde V, Marotta A. How much difference in
type-approval CO2 emissions from passenger cars in Europe can be expected from
A/C: Air-conditioning for the vehicle cabin.
changing to the new test procedure (NEDC vs. WLTP)? Transportation Research
BEV: Battery-electric vehicle.
Part A: Policy and Practice 2018;111:136–47. https://doi.org/10.1016/j.
CAN: Controller Area Network, a message-based protocol designed to allow the Electronic
tra.2018.02.002.
Control Units of an automobile, as well as other devices, to communicate with each
[50] Dalla Chiara B, Deflorio F, Pellicelli M, Castello L, Eid M. Perspectives on
other.
Electrification for the Automotive Sector: A Critical Review of Average Daily
CO2: Carbon dioxide, a chemical compound.
Distances by Light-Duty Vehicles, Required Range, and Economic Outcomes.
Conventional vehicle: A vehicle powered by an internal combustion engine, with no elec
Sustainability 2019;11:5784. https://doi.org/10.3390/su11205784.
trification share.
[51] Travel Weather Averages (Weatherbase). Weatherbase n.d. http://www.
Driving range: The maximum distance driven.
weatherbase.com/ (accessed October 7, 2021).
EU: European Union, a political and economic union of 27 member states that are located
[52] Li Y, Zhu H. A Research on Electric Car Styling Design and Low Aerodynamic Drag.
primarily in Europe.
IOP Conference Series: Materials Science and Engineering 2019;573:012014.
Gap: Energy consumption or CO2 emissions divergence realised by driving on-road versus
https://doi.org/10.1088/1757-899X/573/1/012014.
the respective manufacturer certified figures.
[53] Roberts A, Brooks R, Shipway P. Internal combustion engine cold-start efficiency: A
HVAC: Heating, ventilation and air-conditioning of a vehicle cabin.
review of the problem, causes and potential solutions. Energy Convers Manage
JRC: Joint Research Centre of the European Commission
2014;82:327–50. https://doi.org/10.1016/j.enconman.2014.03.002.
Mass in running order: Actual vehicle mass including the mass of the driver, fuel and liquids,
[54] Tietge U, Mock P, Franco V, Zacharof N. From laboratory to road: Modeling the
with the standard equipment fitted in accordance with the manufacturer’s
divergence between official and real-world fuel consumption and CO2 emission
specifications.
values in the German passenger car market for the years 2001–2014. Energy Policy
OBD: On-board diagnostics, a standardised system for the interaction of external elec
2017;103:212–22.
tronics to with any automobile.
[55] Ktistakis M, Pavlovic J, Fontaras G. Sampling approaches for road vehicle fuel
OBFCM: On-board fuel consumption monitoring device
consumption monitoring. LU: Publications Office; 2021.
PEMS: Portable Emissions Measurement Systems used for on-road emissions testing.
[56] Fiori C, Ahn K, Rakha HA. Power-based electric vehicle energy consumption
RW: Real-world, refers to on-road driving for the day-to-day needs of a person.
model: Model development and validation. Appl Energy 2016;168:257–68.
STD: Standard deviation
https://doi.org/10.1016/j.apenergy.2016.01.097.
Weighted average: An average of a a data set that considers the varying degrees of impor
[57] Pielecha J, Merkisz J, Kurtyka K, Skobiej K. Cold start emissions of passenger cars
tance of each number in the data set.
with gasoline and diesel engines in Real Driving Emissions tests. Combustion
Type Approval proceduce: Procedure for granting a certificate of conformity for meeting the
Engines 2019;179:160–8. 10.19206/CE-2019-427.
minimum regulatory requirements.
[58] From laboratory to road: A 2018 update | International Council on Clean
WLTC: World harmonised Lightweight Test Cycle, from commission implemeting regula
Transportation n.d. https://theicct.org/publications/laboratory-road-2018-update
tion 2017/1151.
(accessed October 4, 2021).
WLTP: World harmonised Lightweight Test Procedure, from commission implementing
[59] Fontaras G, Ciuffo B, Zacharof N, Tsiakmakis S, Marotta A, Pavlovic J, et al. The
regulation 2017/1151.
difference between reported and real-world CO2 emissions: How much
improvement can be expected by WLTP introduction? Transp Res Procedia 2017;
25:3933–43. https://doi.org/10.1016/j.trpro.2017.05.333.
19