Energy Conversion and Management 270 (2022) 116161

Contents lists available at ScienceDirect

Energy Conversion and Management

journal homepage: www.elsevier.com/locate/enconman

Analysing the real-world fuel and energy consumption of conventional and

electric cars in Europe☆
D. Komnos a, S. Tsiakmakis b, J. Pavlovic c, L. Ntziachristos a, G. Fontaras c, *
a
Aristotle University, Thessaloniki, Greece
b
BioCHP, Thessaloniki, Greece
c
European Commission, Joint Research Centre, Ispra, Italy

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.

1. Introduction The effectiveness of the measures and their contribution to curbing

The European Union (EU) Climate law sets the target of climate
neutrality by 2050 [1]. Passenger cars and vans are responsible for 15 %
of the total European Union (EU) CO2 emissions [2], and the emissions
reduction pace in transport is the slowest among all the sectors
contributing [3]. According to current regulations, the EU fleet-wide
CO2 emissions of new car sales will have to be reduced by 15 % by
2025 and 37.5 % by 2030 compared to the 2021 reference[4]. The path
to achieving these objectives is to create incentives for investments in
new technologies and increase the diffusion of zero- and low-emissions
vehicles in the EU market.
climate change will also be assessed considering the greenhouse gas
emission reductions achieved in real-world vehicle operation. During
the past decade, studies using different approaches demonstrated that
the official fuel consumption and that experienced during real-world
(RW) driving conditions were diverging. Thus, counting solely on the
official certification fuel/energy consumption or CO2 emissions and not
real-world performance values when analysing different policy mea­
sures could undermine their effectiveness. This divergence from the
officially declared CO2 values, the so-called gap, has been increasing in
the previous decades. When assessing databases with driver self-
reported consumption figures, the gap was up to 40 % higher than the

☆
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.
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
an one-year on-road campaign where several drivers used the same car
for their day-to-day needs, that the gap reached values up to 100 % for
certain drivers. Battery electric vehicles (BEVs) have undergone rapid
global growth during the last years. Their share has tripled due to higher
public environmental awareness [7], and the efforts by manufacturers to
meet the new CO2 targets [8]. Although tailpipe emissions are not
relevant in BEVs, there is interest in their actual energy consumption
which eventually translates to upstream emissions, their driving range
that is of high consumer concern, and the relevant influencing factors.
Early investigations using simulation approaches [9], or assessing fleet
data [10], placed the RW values close to the officially declared ones.
Nonetheless, Paffumi et al. [11] experimentally found that the energy
consumption due to heating, ventilation, and air-conditioning (A/C),
reffered to as HVAC, increased consumption by up to 70–80 % in lab­
oratory tests in − 10 ◦ C ambient temperature.
Recent provisions were added to monitor the RW fuel and electric
energy consumption of passenger cars and vans to prevent the gap from
growing again [12]. To achieve this and to have the possibility to assess
RW representativeness of the consumption determined at the certifica­
tion procedure, car manufacturers were required to install onboard fuel
consumption monitoring (OBFCM) devices that store information about
energy consumption by January 2021 [13]. This information will
eventually be reported to the European Commission [14]. However,
understanding the gap and its variability due to different factors will be
crucial to monitor the effectiveness of the CO2 standards [15]. The
influencing factors can be grouped in 4 main categories: the vehicle
characteristics and systems; the environmental and traffic conditions;
driver and user-related factors; and vehicle certification test [16]. While
the latter factor has been addressed in 2017 with the introduction of the
new test protocol, the Worldwide harmonised Light-vehicles Test Cycle
and Procedure (WLTC and WLTP, respectively) [17], the rest still
construct an essential research topic.
Experimental campaigns can accurately contribute to this research
topic [18,19], but they are not cost-effective. Additionally, the outcome
can not always show the broader picture, and the findings can not safely
apply in another case study. Several other practices exist for assessing
the factors influencing energy performance and the gap in an organised
way. Fontaras et al. [16], in an extensive literature review, presented the
parameters to consider during driving, and set indications for their RW
2
Energy Conversion and Management 270 (2022) 116161
impact. Zacharof et al. [20] combined literature with detailed vehicle
simulations to analyse the sensitivity of factors such as road gradient,
aerodynamics, and auxiliaries usage in CO2 emissions in indicative
conventional vehicles. Moreover, methodologies are proposed to assess
the effect of such non-regulatory parameters (e.g. wind [21], surface
wetness, auxiliary systems [22], and road gradient [23]) for a conven­
tional vehicle. Ktistakis et al. [14], using statistical approaches, analysed
databases with driver-reported CO2 values and characterised the vari­
ability of the gap due to vehicle characteristics. More recently, studies
have compared conventional powertrains with electrified ones. Fiori
et al. [24], and Mamarikas et al. [25] assessed the impact of traffic
conditions for BEVs and found that the consumption patterns were
different between BEVs and conventional vehicles. HVAC holds a
considerable share in the increase of the energy consumption of a
vehicle, and detailed models quantify the impact in RW driving cycles
[26]. The above studies substantially contribute to the research on RW
CO2, energy consumption gap, and variability. However, they focus on
determining the sensitivity of different parameters, and to the authors’
knowledge, there is no study assessing their combined effect (including
the influence of the certification test) in the current European car fleet
composition under specific, realistic scenarios applicable to Europe for a
one-year period.
In this context, the present study proposes a simulation-based
approach to study Europe’s CO2 and energy consumption gap. PyCSIS,
a vehicle simulator suitable for fleet analysis [27], has been updated to
host all the powertrain configurations present in the European fleet and
extensively validated versus real test cases with characteristics that
represent a wide range of the fleet. The simulator will construct a basis
for applying scenarios for the gap, as calculated over an one-year period;
for this reason, it was combined with a set of representative RW trips and
literature on environmental factors and their effect on the consumption
of the vehicle. The study consists of 3 steps: model validation to simulate
modern European passenger cars and quantification of the certification
procedure impact; review and quantification of the impact of different
factors in the gap; and finally, fleet data collection and scenarios ap­
plications to understand the variability reported by previous studies for
the European fleet. Insights on the simulation results intend to
contribute to the research and characterisation of the gap and its vari­
ability and provide realistic indicators of vehicle consumption and
driving range.
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.

2.1. Study structure 2.2.2. Type Approval data

As a first item, Fig. 1 provides a graphical representation of the study
structure, split in the Validation, the RW impact simulations, and Fleet
projection processes.
Laboratory and on-road experimental data from tests performed by
the EU Joint Research Centre (JRC) in the period between 2018 and
2021 were used to validate (Fig. 1, coloured in red) the simulation tool.
The 2018 vehicle certification data were used to further validate and
quantify the Type Approval procedure effect. The vehicle weight, aux­
iliaries consumption, and the ambient temperature - parameters
responsible for the gap- were subsequently assessed with simulations in
the RW impact section (Fig. 1, coloured in green), and the importance of
each factor was quantified in a set of RW driving profiles. The term RW
implies that the trips performed on-road and their characteristics (e.g.
velocity profile, distance travelled, stop share) cover the day-to-day
EU Member States use a standard secure electronic exchange system
for sharing information about motor vehicles, trailers and systems,
certified in the EU [31]. The JRC has collected technical characteristics,
the manufacturers’ declared CO2 emission values (for the ICEVs), and
the battery-electric energy consumption (for BEVs), for 1,500 vehicle
variants type-approved in EU during the first year of the WLTP intro­
duction (August 2017 to December 2018). A database was built that
represents the European market as a whole [32]. A subset consisting of
74 ICEVs and 14 BEVs, for which the WLTP declared values were
available, was selected for the needs of the study. The representation of
all main configurations of ICEVs and BEVs is well captured, as seen in
Fig. 2.
2.2.3. Fleet data
The official European Monitoring databases of CO2 emissions from

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

Average Speed [km/h] 49 53 63 60 60 36 65

Trip distance [km] 83 96 139 120 130 62 65
Urban/ Rural/ Highway distance shares [%] 40/32/28 39/28/33 29/13/58 41/23/36 42/13/45 91/9/0 17/22/61
Altitude max – min [m] 98 223 188 235 178 950 70
Vehicles 1–6 1–6 1–6 7, 9 8 7, 8, 9 10

new passenger cars for the years 2018 and 2019 for ICEVs and 2019 for
BEVs, developed and maintained by the European Environment Agency
[33], were used as a reference to study the RW impact on the fleet. These
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
% higher than the registrations during 2020 and onwards due to the
unprecedented consequences of the COVID-19 pandemic [34]. For each
entry, official CO2 values (for ICEVs), battery energy consumption (for
BEVs), along with vehicle-specific information such as mass in running
order, technical characteristics, and the number of registrations, are
provided.
The information included in the fleet database is not sufficient to
perform vehicle-specific modelling. Information regarding engine
characteristics, vehicle body shape, gearbox type, and battery capacity
are critical and profoundly affect the selected model’s performance. In
that aspect, data from the official database was combined with data
retrieved from publicly available sources (i.e., online databases like car
folio.com, cars-data.com, carspector.com, etc., and vehicle manufac­
turers’ websites) used to formulate a more detailed database. More de­
tails on the European Environment Agency database, the publicly
provided vehicles’ specifications database, the combined database, and
all data treatment steps are provided in a previous study by Tsiakmakis
et al. [17]. The combined database created consists of 993 ICEVs from
2018 and 574 from 2019, representing 12 % and 5 % of the 2018 and
2019 vehicles registered in the EU in the respective years, and 63 BEVs
representing 91 % of the electric vehicles that entered the EU roads on
2019.
Fig. 3 gives a visual representation of the vehicles’ technical con­
figurations expanded for their EU sales. The study team simulated ve­
hicles representing more than 2.8 million registrations in the European
market. The representation of both ICEVs and BEVs is considered
satisfactory: Sport utility vehicles (SUVs) have the highest share for
ICEVs, and the majority are small vehicles with engine capacity below
1.4 lt. The European Automobile Manufacturers’ Association confirms
these facts for 2009 to 2020 [35].
2.3. Trip data
This section provides details of the trips performed during the
physical tests and used for the validation, and the trips adopted for the
simulations of the RW scenarios. The main factors used for the trip
characterisation are the mean trip speed, accumulated distance, and
maximum and minimum elevation.
2.3.1. Type approval test
The gap experienced during the RW driving is studied against the
current EU certification protocol, the WLTP, performed on a test bench.
The driving cycle lasts 30 min, and the distance driven is 23.25 km, with
a mean speed of 46.5 km/h. The cell temperature is set to 23 ◦ C, and the
vehicle is tested on cold start (its thermal state is at the same tempera­
ture as the cell). Resistance coefficients, a function of vehicle speed,
need to be set as a target in the dynamometer for simulating realistic on-
road resistances; these coefficients need to be derived experimentally in
test tracks, with the most frequent procedure being the coast-down test
[17]. During the coast down the vehicle is left to decelerate from a high
speed, and the deceleration intervals are transformed to forces resisting
the motion.
2.3.2. On-road tests
On-road trips were performed on vehicles tested at JRC, and they
were used to validate the simulation tool under real driving conditions
(extra driving resistances, HVAC operation). Table 2 presents some
indicative information of the trips used. The distance shares travelled in
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.

2.4. In-Use factors

Table 3
Information for the in-use parameters used for each simulation batch run for the
Three parameters were evaluated in the present study for their effect
10 JRC vehicles over the 30 RW mission profiles. The first column includes the
base-case values. *: Value same as Base-case. on the RW driving emissions and the respective gap between RW and the
certified figures: the mass, the auxiliary system consumption, and the
Simulation batch Number 1 (Base-Case) 2 3 4 5 6 7
ambient temperature. These parameters have been selected based on the
Passengers [-] 1.45 1 3 * * * * findings of Fontaras et al. [16]. Combined with a wide range of vehicle
Vehicle auxiliaries [kW] 0.75 * * 0.3 1.5 * *
and RW trip characteristics they effectively cover all the factors affecting
Ambient temperature [oC] 14 * * * * 5 23
RW emissions: vehicle, environmental and traffic, and driver [6]. More
details for each factor are added below:
all three road types. Trip 6 does not contain a highway part because the
route was in the mountain with high altitude gain to test the regenera­ • Extra mass, related to additional mass coming from passengers and
tion of battery of electrified vehicles. It should be noted that there was luggage. Three values are considered based on the vehicle passen­
no need for battery charging during the tests for BEVs. More details gers. According to the European Environment Agency [37], the
regarding the instrumentation are presented later in the Methodological average passenger occupancy is 1.45. The passenger mass is 75 kg
steps section. (taken from certification regulation [17]), and 10 kg extra weight is
added as luggage per passenger. The scenarios include 1/ 1.45/ 3
2.3.3. Real-world driving trips passengers, using as base-case scenario 1.45 passengers.
Real trips were introduced in the simulation plan to analyse the in­ • Auxiliaries, related to electrical and mechanical loads coming from
fluence of RW driving on an average vehicle over one year. The data was vehicle’s auxiliaries, when used under different conditions (lights,
collected in a campaign that aimed to examine the variability of RW CO2 A/C, radio, wipers, pumps, fans, compressors, etc.). Three values are
from multiple drivers [6]. Due to the computational cost, a subset from considered for this parameter: 0.3/ 0.75/ 1.5 kW. The low and the
the 483 in total RW trips had to be selected. There is a logarithmic base-case scenario consumptions are mentioned as indicative con­
pattern between distance and average speed. In a previous study, one sumption values during the Type Approval test and on-road driving
trip per distance segment was selected that was falling close to the [38]. On the other hand, the value used as maximum auxiliary
logarithmic line fitted, for 10 different distance segments [36]. In order consumption is indicative for including HVAC and infotainment
to maximise the representativeness of all the trips, the authors adopted a electric consumption.
method to select 3 trips per segment. The selection criteria to identify • Ambient temperature, related to different ambient temperatures that
the subset are the following: affect the heat losses on the engine’s body, cold-start operation (for
ICEVs), and vehicle’s resistances. Three values are considered: 5/
1. The data were grouped in 10 distance clusters (vertical blue dashed 14/ 23 ◦ C, using as base the 14 ◦ C, which is considered as an average
lines in Fig. 4). global ambient temperature across the year [17]. For the present
2. Trip distance and mean vehicle speed were used to characterise the study, also the vehicle thermal state of ICEVs starts from the same
complete set of recordings (light blue points in Fig. 4). temperatures (e.g. engine coolant and gearbox temperatures).
3. The K-means algorithm was applied in each distance cluster to form 3
groups: low, mean, and high speed (the reader will find an example Table 3 summarises the simulation setup. The first batch was simu­
in Appendix B). lated in a base-case scenario. Then, six more simulation batches fol­
4. The trips falling closer to the centre of each K-means cluster were lowed. Each time one of the in-use factors was replaced with the best or
selected, resulting in 3 trips per segment (green, black and red worst assumption. The influence of each of the factors compared to the
hexagons in Fig. 4), thus 30 trips in total. base-case was assessed in the ten vehicles (6 ICEVs and 4 BEVs) tested in
the JRC. А total of 2,100 simulations were carried out: 10 vehicles in 30
RW trips, having 7 different in-use factor combinations.

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.

• Derivation of indicative values for the hybrid control strategy, hybrid

Table 5
architecture, and technical characteristics of HEVs and plug-in HEVs
Official (manufacturers’ declared) values versus simulation error. STD: standard
from CO2MPAS calibrated models.
deviation.
• Development of a power-based model for BEVs for the battery energy
ICEVs BEVs Total consumption as a function of the motive power that is calculated
Count [-] 74 14 88 from the resistive forces. The study did not focus on different possible
Mean Difference [%] − 6.6 − 5.6 − 6.4 BEV charging practices, even though, the percentage of the trip
STD [%] 6.3 10.9 7.1
driven with one full battery charge is recorded for post-processing.
• Reassessment of the road load coefficients, i.e., the coefficients for
2.5. Simulation tool calculating the resistive forces as a function of speed. The coefficients
were evaluated versus the certified ones which were produced
The simulation tool incorporated, PyCSIS, uses the same underlying following the official coast-down procedure. The impact of the
simulation methodology as CO2MPAS [39]. The suitability and accuracy simulated road loads compared to the official ones is analysed in
for simulating fleet-average CO2 emissions have been thoroughly tested detail on the energy demand and CO2 emissions during RW, showing
and validated for the laboratory [27] and on-road [40] conditions. that the error introduced is limited and averages to 1.5 % [36].
Compared to other models such as the widely known Guidebook method
[41], which is macro-perspective, or the implementation of Duarte et al. An important factor affecting consumption that is gaining attention
[42] which provides accurate results when averaging over a number of is the HVAC, especially in the case of BEVs that can significantly reduce
vehicles, PyCSIS detailed implementation supports the investigation of their driving range by 25 % [43]. Regarding the energy consumption for
specific vehicle configurations in specific conditions [40]. the cabin cooling, the model used in a study by Giechaskiel et al. [44] to
The sub-models have been calibrated with tests performed in the JRC assess the vehicle performance under extreme ambient temperatures
premises, as previously described by Tsiakmakis et al. [27]. The vehicle was adopted. A linear model was introduced using the results from
calibration database consisted of 100 conventional and 6 electrified Fayazbakhsh et al. [45], and Lee et al. [46] that assessed the A/C con­
ICEVs, that entered the EU market from 2014 to 2020. In detail, the tool sumption during ambient temperatures of 35 ◦ C and 45 ◦ C, respectively.
update included: Linear models are also found in a study by Mebarki et al. [47] for BEV A/
C simulations. ICEVs employ the engine heat for the cabin heating; thus,
the energy consumed is little and assigned to the electric fan propulsion.
For model simplicity, no extra loss is assigned. In contrast, the BEVs
powertrain produces minimal waste heat, and consequently, a separate

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

2.6. Methodological steps

2.6.2. Real-world impact
The 10 tested vehicles were simulated for the 30 RW trips (the inputs
In the following sections, the Validation, the RW impact simulations,
were the velocity profiles and road gradient) selected during this pro­
and the Fleet projection steps mentioned at the beginning of the chapter
cess. The simulation step was performed multiple times, assessing each
are described in more detail.

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

Mass ICEV +130 kg 3.4 % 3 % [20]

Auxiliaries ICEV +0.5 kW 3.3 % 4 % [20]
Cold-start ICEV − 2◦ C (amb. T) 8 % (RW) 18 % (Urban) [53]
10 % (NEDC) [53]
Heating BEV − 10 ◦ C (amb. T) 21.5 % (RW) 33 % (WLTP) [26]

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
[-] [%] [%] [%]

ICEVs All ICEVs 1567 (2539.3) 18 (16.5) 6.2 13 – 22

Diesel Auto 378 (222.6) 12.9 (11.9) 3.2 11 – 15
Diesel Man 168 (164.5) 12.4 (11.5) 2.5 11 – 14
Gas Auto 483 (595.0) 20.3 (18.5) 4.6 18 – 24
Gas Auto Hybrid 30 (98.0) 31.8 (26.5) 7.1 31.5 – 36
Gas Manual 502 (1,459.0) 19.7 (16.5) 4 17 – 23
Gas Manual Hybrid 6 (0.23) 20.6 (18) 19.1 3 – 38
BEVs Electric 63 (313.5) − 2.6 (-0.8) 3.5 − 4.6 – − 0.3

time the influence of an in-use factor related to RW driving. It should be ∑

RWconsumptionvalueind.vehicle,trip(i) • wtrip(i)
noted that the assumption of a fully charged battery before each indi­ RWconsumptionind.vehicle = ∑ , ∀i
vidual trip was made for all the 4 BEVs simulated. wtrip(i)
∈ [1, 30]
2.6.3. Fleet projection (2)
The 483 trips recorded over 11 months support the study’s goal for
The weight of each trip, wtrip, is the number of the trips falling in the
RW representativeness because they describe day-to-day needs and a ∑
depiction of one-year vehicle usage. The total distance driven is 13,167 same distance, speed group, thus 30 i wtrip (i) = 483. A graphical example

km, which, according to the literature [50], falls inside the average
distance a vehicle runs in a year period. As mentioned earlier, only the
30 selected RW trips and the WLTP profile were simulated for each
vehicle from the fleet. The influence of the remaining trips is assigned to
each individual simulated RW trip as a weight for producing a weighted
average RW consumption value per vehicle (equation (2)).
is provided in Appendix B for better understanding. Then, the individual
gap of each car model simulated was calculated according to the equa­
tion (3). The certified consumption of each vehicle is the manufacturer’s
declared value according to the WLTP.
Gapind.vehicle = (RWconsumptionind.vehicle /Certifiedconsumptionind.vehicle − 1)
• 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. Results & discussion

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)

is mainly performed to investigate the certification procedural influ­

ence, it provides further assurance of the tool’s accuracy. More insight
into the distribution of values is provided in Appendix E. From this step,
the determined average certification influence was calculated to 6 %,
which is close to what was expected. This value is later used on the

10
D. Komnos et al.
analysis of the official versus RW CO2 and energy consumption gap.
3.2. Real-world influence
Having proven the accuracy of the simulation tool, the influence of
on-road parameters are presented in the next two sections.
3.2.1. Trip effect
Fig. 6 presents the average gap due to the trip characteristics. It is
separated in ICEVs and BEVs because the two powertrain configurations
show different trends. The increased stop & go regimes in the short
distance trips, combined with the engine idling and the high shares in
the trip duration, where the cold-start period occurs, resulted to the
highest positive gaps for the six ICEVs. In average speeds between 40
and 80 km/h, the lowest gap was observed: there were trips where the
gap turned negative, i.e., the CO2 emissions were lower than what was
simulated for WLTC velocity profile. In higher speeds, CO2 emissions
increased again due to excessive aerodynamic losses. This overall trend
is consistently mentioned in the literature as the ICEV U-shape [24]. The
average gap calculated for the trips with average speed below the
average WLTP (46.5 km/h) was 13 %, while for higher speeds it was 7
%.
The four BEVs completed all the RW trips with only one full battery
charge in the simulation. The gap was monotonic as a function of the
average speed; thus, trips with average speeds below the WLTP had a
negative average gap (-7%), while for higher trip speeds, it was 20 %.
The overall trend agrees with observations of Mamikoglu et al. [26],
which reported better efficiency in low-speed conditions like city cycles
compared to cycles with higher speeds.
3.2.2. Effect of In-Use factors
Table 6 presents the average percentage gap produced in the 30 RW
trips. The impact of passengers is limited for both configurations, but
especially for BEVs that are generally heavy vehicles due to the fitted
batteries. The highest impact was observed when tuning the value of
auxiliaries consumption, and BEVs had almost double the impact
compared to ICEVs. This is because the auxiliary demand in low-speed
trips, that are more common in RW driving, is comparable to the
battery-electric energy consumption for the vehicle propulsion. In
regards to the ambient temperature, ICEVs had a greater impact
attributed mainly to the cold-start operation. Another reason was the
better aerodynamics of BEVs [52], resulting to a lower impact of the
reduced air density compared to ICEVs.
Fig. 7 shows the individual and the combined effect of the HVAC-
related auxiliaries, and the vehicle losses, due to the ambient tempera­
ture. The gap was higher during cold conditions (ambient temperatures
below 14 ◦ C) than warmer conditions (temperatures above 14 ◦ C) for
both ICEVs and BEVs, and the best efficiency appeared in an ambient
temperature of 21 ◦ C. The impact of the ambient temperature was higher
in the case of BEVs because the energy needed for the HVAC was com­
parable to the energy required for the propulsion, especially for the low-
speed trips.
Although there are methodological differences, and direct compari­
sons with the literature cannot be made, Table 7 shows that the influ­
ence calculated for different scenarios is similar to studies by Zacharof
et al. [20], Fiori et al. [53] and Lajunen et al. [26]. This gives confidence
both in the accuracy of the tool and in the methodology used to apply
realistic projections of the influence of the in-use factors in RW
conditions.
3.3. Full fleet scenario results
This chapter presents the assessment of the gap based on the fleet
data (equation (4)). The simulated Type Approval figures were increased
by a constant value of 6 % to account for the inflation due to the certi­
fication procedure, as calculated by the Type Approval assessment.
11
Energy Conversion and Management 270 (2022) 116161
3.3.1. Base-case scenario
The overall gap for all simulated ICEV models in the average base-
case conditions was 18 %. The overall sales-weighted mean gap was
16.5 %. The gap calculated is in agreement with the fuel consumption
gap of 20 % (adjusted to the WLTP), reported in other studies [6,54]. In a
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
PHEVs that increases the CO2 gap.
BEVs average gap was − 2.6 %, ranging from − 4.6 % to − 0.3 %.
Considering the EU 2019 sales, the average gap was − 0.8 %. Fiori et al.
[56], in measurements in six different legislative laboratory cycles,
showed that the gap compared to WLTP consumption ranged from − 30
% to 12 %, identifying a strong correlation with the cycle energy de­
mand. This confirms the slightly negative gap in some of the simulations:
the majority of the trips used in the analysis had lower trip speed and
distance than WLTP, and according to the assessment of the cycle energy
demand in different RW profiles performed by Komnos et al. [36], lower
trip speed equals lower overall cycle energy demand.
More details for both ICEVs and BEVs are presented in Table 8. Diesel
vehicles appeared to have a lower average gap than gasoline ones. The
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
showed that the gasoline configuration emitted two times more than the
diesel during the cold-start for the same brand model. In trips with
higher duration, the gap found in diesel vehicles was similar to the
gasoline ones. For conventional ICEVs, the range of the gap (1st to 3rd
quartiles) was 14 %–22 %, while for electrified ICEVs (Hybrids), the
range was 31.5 %-36 %. This 10 % higher gap between hybrids and
conventional vehicles is supported by an International Council of Clean
Transportation study [58] that assessed user-communicated consump­
tion. A detailed assessment of the hybrid simulation outputs revealed
that this extra gap resulted from the hybrid controller that promoted the
drive battery charging during trips with high dynamicity, resulting in
higher fuel consumption. This phenomenon was observed in a few short-
distance trips; their higher influence in the RW weighted average
intensified the impact in the gap. More details are presented in Appendix
F.
Regarding the influence of the trip characteristics in the driving
range of a BEV, Fig. 8 shows that for trips longer than 100 km, several
models from the market failed to complete the route with one full
charge, even if the manufacturers’ declared range was higher than the
trip distance. This was intensified in trips with higher type-approved
ranges: one charge was insufficient for 36 % of the models with a
manufacturer- stated range between 230 and 320 km.
3.3.2. Best and worst-case scenario
The analysis of ambient European temperatures indicated the
extreme value for the average cold conditions was − 1 ◦ C and for hot
conditions, 23 ◦ C. According to the assessment of the in-use factors, the
cold conditions were used as the worst case and the hot conditions as
best-case scenario. Table 9 presents further details of the values
considered. For the worst case, the auxiliaries from ICEV base-case were
increased by 0.25 kW, and for BEVs by 0.2 kW to account for extra
infotainment.
Fig. 9 presents the updated gap results produced by the adjusted RW
weighted-average consumption for each model’s worst- and best-case
scenario versus its simulated certified value. The best- and worst-case
average gap for ICEVs was 13.8 % and 34.5 %, respectively. These
values, translated to the 2018 and 2019 model sales, changed to 12.5 %
and 33 %, respectively. For BEVs the average values for the models
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
sales numbers. The difference between the best-case and the base-case
scenario was more profound for ICEVs (4.2 %) than BEVs (1.9 %). The
difference can be attributed to the engine cold-start period, which was
not included in the case of the BEVs. In contrast, the worst-case scenario
D. Komnos et al.
is more evident for BEVs, calculated to be 26.5 %, while for ICEVs it was
16.5 %. The reason for the difference was the extra consumption for
cabin heating needed in BEVs that comes for free in ICEVs.
Τhe range of the gap, defined as the 1st quartile of the best-case up to
the 3rd quartile of the worst-case, for different vehicle configurations is
presented in Table 10. Diesel vehicles showed a lower range but also gap
interim compared to gasoline. BEVs showed the highest range and cor­
responded well with the suggested range by De Cauwer et al. [27],
particularly 10 to 40 % (adjusted to WLTP).
3.3.3. Monte-Carlo simulations
The analysis of the temperatures of European countries showed that
the average temperature was 14 ◦ C (STD ± 4 ◦ C) and 7 ◦ C (STD ± 5 ◦ C)
in warm and cold conditions, respectively. The cold temperatures
reached values down to − 5◦ C, and the warm ones up to 30 ◦ C. The
validation of the online temperature database, examples of the method
for the random in-use factors values generation and trip characteristics
that influence the gap per vehicle are presented in Appendix G.
To replicate a full year, similar to the 483 trips that corresponded to
11 months, 500 random values were produced per vehicle model for
different combinations of number of passengers and ambient tempera­
tures, split in 250 warm and 250 cold conditions. The HVAC auxiliaries
were calculated according to the ambient temperatures. Four scenarios
(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
nation of trip characteristics that could be driven throughout a year was
combined with the random in-use factors for the warm and the cold
period. When moving from 250 random trip characteristics to 5, the
mean gap remains about the same but the STD significantly increases
(Table 11).
This showed that a low variability in the driving patterns spread the
fleet gap dispersion. For example, a specific vehicle model driven mainly
on a motorway would have a different gap than when driven on city
roads. This can be depicted in the kilometres driven annually according
to the four scenarios (Table 12).
The variability in the mean gap values in the four scenarios was very
low because the high number of vehicle simulations converged to
similar values. However, compared to the base-case, it is interesting that
both cold and warm periods showed higher average gaps: 7.5 and 5.5 %
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,
while for BEVs, it can be assigned to HVAC.
4. Conclusions
Understanding the effect of the factors that influence real-world
consumption is essential for proper policy implementation. The
distinct impact of different factors influencing the gap on CO2 emissions
and energy consumption has been extensively studied. However, under
RW conditions, little is known on the combined effect of these factors on
the gap in a representative sample of the European fleet. In this context,
the present study proposed an approach based on a detailed simulation
tool, PyCSIS, for estimating fleet vehicle consumption. Τhe accuracy of
the tool to reproduce the recorded consumption during RW driving was
satisfactory, with 92.5 % of the simulated values falling between ± 10 %
of the recorded. PyCSIS was subsequently combined with literature and
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
Energy Conversion and Management 270 (2022) 116161
influence. When combining factors that are influenced by ambient
temperature, the cold conditions appear to have a greater impact on the
gap than warm conditions, for all powertrain configurations tested. For
the average European conditions, the gap of RW driving versus certified
figures was found to be 16.5 % for ICEVs, and − 0.8 % for BEVs.
Considering the best- and worst- case scenarios for Europe, the gap from
the certified values range was 12.5 – 41 % for gasoline vehicles, 7 – 31 %
for diesel conventionals, 27 – 55 % for HEVs, and − 6.5 – 26.5 % for
BEVs. Regarding the values gap distribution of the fleet, a Monte-Carlo
simulation approach revealed that driving in a limited range of trip
characteristics –mainly city routes– resulted in higher gap dispersity, but
the average fleet values change was limited.
The results of the present study represent indicative average Euro­
pean scenarios. The same approach and simulation tools could be
adopted to assess different scenarios, more focused on specific fleet
compositions, conditions, and driving patterns, to support future pol­
icies scientifically. Future work should focus on further analysing the
parameters affecting the gap at a vehicle level, improving the model to
capture new technologies and their effectiveness on-road; and finally, on
providing insights on the range and potentials of BEVs that are becoming
one of the main alternatives for meeting the CO2 targets set.
Authors contributions
The authors, Dimitrios Komnos (DK), Stefanos Tsiakmakis (ST) Jelica
Pavlovic (JP), Leonidas Ntziachristos (LN) and Georgios Fontaras (GF)
contributed to the paper as follows:
Study conceptualisation: ST, DK, GF, LN; data collection, curation,
and analysis: DK, ST, JP; model simulations: DK, ST; analysis and
interpretation of results: DK, ST, GF; manuscript preparation: DK, ST;
revision and comments: JP, LN, GF; research coordination: GF. All au­
thors reviewed the results and approved the final version of the
manuscript.
Declaration of Competing Interest
The authors declare that they have no known competing financial
interests or personal relationships that could have appeared to influence
the work reported in this paper.
Data availability
Data will be made available on request.
Acknowledgements
The authors would like to acknowledge the JRC Vehicle Emissions
Laboratory (VELA) team, and Dr Alessandro Tansini for carrying out the
experimental activities, and providing the experimental data. The au­
thors would also like to thank Mr Francois Filip, Mr Serra Carlos and Ms
Lakatos Katalin for their constructive review.
Appendix A. . Simulation tool inputs and HVAC model
Table 13 lists the main inputs required from the user and Fig. 10 the
HVAC model consumption as function of external ambient temperature.
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.
Appendix B. . Graphical example of the real-world trips
clustering
Fig. 11 presents an example of the K-Means clustering performed and
the weights, wtrip, derivation. The RW simulated value of each vehicle in
one of the 30 RW trips is fed together with the wtrip to the equation (4).
Appendix C. . Internal combustion engine old-start model
assessment
The on-road tests used for the ICEV were starting with the vehicle
thermal state always between 20 ◦ C and 23 ◦ C. That is why the cold-start
operation was assessed separately in additional laboratory tests per­
formed to assess the influence of extreme temperatures in the emissions
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
Energy Conversion and Management 270 (2022) 116161
accumulated CO2 of a test repeated in ambient temperatures of − 10 ◦ C,
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
differentiate the 3 different test conditions. The same results show evi­
dence that the tool is capable of accurately simulating not only long
trips, where different errors can compasate, but also short trips.
Fig. 11. Example of the clustering procedure. Round
dots correspond to all the recorded trips distance,
average speed pairs for the selected distance cluster
180–230 km. Pentagons correspond to the centers of
the clusters identified by the K-Means algorithm. X
shape points to the trip that falls closer to each
cluster center. The dot colour, i.e., red, black, and
green, is used to distinguish the high, mean, and low
speed group, respectively. The numbers correspond
to the number of trips falling in the same group.
Appendix D. . On-road versus laboratory assessment
Fig. 13 shows the gap between laboratory (WLTP) and on-road
driving as derived by the simulations and the experimental data. The
average error inserted between lab and on-road is 2.2 %. Considering
that literature suggests a gap in the order of 20 % or above [55,59,60],
the authors consider these results adequately accurate and sensitive.
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
Fig. 14 presents the distribution of the percentage difference be­
tween the simulated type approval values versus the manufacturers’
declared ones. The result of around − 6% for both ICEVs and BEVs
confirms literature suggestions on the impact of the certification pro­
cedure. The scattering of the values for both ICEVs and BEVs resemble
normal 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
line vehicles.
Fig. 16 presents a comparison between similar models of the same
brand, one conventional and one electrified ICEV. The simulations of
electrified version show that in general it consumed significantly less,
except of some cases where, the hybrid controller promoted the driver
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
distributions
The available information regarding the ambient temperature of the
483 RW trips was not adopted because the majority of the trips were
performed in Northern Italy, so they could not fully resemble average EU
temperatures. Still, the RW trip average ambient recordings were split in
warm (spring and summer) and cold (autumn and winter) periods to
understand the temperature distributions. The analysis revealed normal
distributions for both warm and cold periods. This information was used
together with EU countries average temperatures. Distributions were
constructed by providing the mean temperatures for the EU warm, and
cold periods (Table 14). A comparison between the RW ambient values
in Northern Italy and the distributions created by average EU monthy
temperatures is also provided. The dispersion of the RW values are
higher than EU average, since the later ones correspond to month av­
erages. The validity of the online database [51] is confirmed on Lom­
bardia region (Northern Italy), by comparing to the recorded RW
average values in 2017, and ASAP, a tool produced by European Com­
mission [61,62].

16
D. Komnos et al. Energy Conversion and Management 270 (2022) 116161

Fig. 17 shows examples of the in-use factors distributions produced

Table 14 randomly. Missing detailed information, the number of passengers
RW recorded ambient tempratures and EU monthly averages, split in warm and follow a continuous uniform distribution. According to the ambient
cold periods.
temperatures, randomly produced, an auxiliary consumption value for
EU average [51] RW trips (Northern Italy) the HVAC is derived from the model developed inside the tool.
period Mean [oC] STD [oC] Mean [oC] STD [oC]
An example of the gap value assigned randomly according to the trip
Warm 14 4 21 7.5 characteristics to each vehicle simulated is presented in Fig. 18.
Cold 7 5 8 7.5

Fig. 17. Example of the random in-use factors assignment.

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