You are on page 1of 11

journal of environmental sciences 125 (2023) 266–276

Available online at www.sciencedirect.com

www.elsevier.com/locate/jes

Variability of fuel consumption and CO2 emissions


of a gasoline passenger car under multiple
in-laboratory and on-road testing conditions

Boya Zhou 1, Liqiang He 2,∗, Shijian Zhang 1, Rui Wang 1, Luowei Zhang 1,
Mengliang Li 1, Yu Liu 1, Shaojun Zhang 2,3,4, Ye Wu 2,3,4, Jiming Hao 2,3,4
1 China Automotive Technology and Research Center Co., Ltd. Tianjin 300300, China
2 School of Environment, State Key Joint Laboratory of Environment Simulation and Pollution Control, Tsinghua
University, Beijing 100084, China
3 State Environmental Protection Key Laboratory of Sources and Control of Air Pollution Complex, Beijing 100084,

China
4 Beijing Laboratory of Environmental Frontier Technologies, School of Environment, Tsinghua University, Beijing

100084, China

a r t i c l e i n f o a b s t r a c t

Article history: An increasing divergence regarding fuel consumption (and/or CO2 emissions) between real-
Received 1 November 2021 world and type-approval values for light-duty gasoline vehicles (LDGVs) has posed severe
Revised 27 December 2021 challenges to mitigating greenhouse gases (GHGs) and achieving carbon emissions peak and
Accepted 28 December 2021 neutrality. To address this divergence issue, laboratory test cycles with more real-featured
Available online 4 January 2022 and transient traffic patterns have been developed recently, for example, the China Light-
duty Vehicle Test Cycle for Passenger cars (CLTC-P). We collected fuel consumption and
Keywords: CO2 emissions data of a LDGV under various conditions based on laboratory chassis dy-
Fuel consumption namometer and on-road tests. Laboratory results showed that both standard test cycles
CO2 emissions and setting methods of road load affected fuel consumption slightly, with variations of less
Light-duty gasoline vehicle (LDGV) than 4%. Compared to the type-approval value, laboratory and on-road fuel consumption of
Real driving the tested LDGV over the CLTC-P increased by 9% and 34% under the reference condition
(i.e., air conditioning off, automatic stop and start (STT) on and two passengers). On-road
measurement results indicated that fuel consumption under the low-speed phase of the
CLTC-P increased by 12% due to the STT off, although only a 4% increase on average over
the entire cycle. More fuel consumption increases (52%) were attributed to air conditioning
usage and full passenger capacity. Strong correlations (R2 > 0.9) between relative fuel con-
sumption and average speed were also identified. Under traffic congestion (average speed
below 25 km/hr), fuel consumption was highly sensitive to changes in vehicle speed. Thus,
we suggest that real-world driving conditions cannot be ignored when evaluating the fuel
economy and GHGs reduction of LDGVs.
© 2022 The Research Center for Eco-Environmental Sciences, Chinese Academy of
Sciences. Published by Elsevier B.V.


Corresponding author.
E-mail: helq@tsinghua.edu.cn (L. He).

https://doi.org/10.1016/j.jes.2021.12.042
1001-0742/© 2022 The Research Center for Eco-Environmental Sciences, Chinese Academy of Sciences. Published by Elsevier B.V.
journal of environmental sciences 125 (2023) 266–276 267

can save fuel and reduce CO2 emissions, such as gasoline di-
Introduction rect injection (GDI) (He et al., 2018; Wu et al., 2019), turbocharg-
ing (Edwards et al., 2008), vehicle lightweighting and down-
The rapid socio-economic development and urbanization sizing (He et al., 2020), powertrain electrification (Wu et al.,
over the past two decades have spurred substantial trans- 2012), alternative or blended fuels (Liu et al., 2019; Wu et al.,
portation demands demonstrated by skyrocketing vehicle 2019) and better fuel properties (Zhu et al., 2017; Wen et al.,
ownership (National Bureau of Statistic of China NBSC 2020).
2021; European Environment Agency EEA 2021; United States Increasingly stringent fuel economy standards and up-
Department of Transportation, Bureau of Transportation dated technologies have succeeded in reducing declared fuel
Statistics U.S. BTS 2021), which poses severe challenges consumption and CO2 emission values measured in the lab-
to ensuring energy security and alleviating climate change oratory under standard testing protocols. However, there is
(Ntziachristos et al., 2014; Wu et al., 2017). Transportation is a still widespread suspicion about these values, given that they
major consumer of fossil energy and consequently an impor- rarely reflect real-world fuel consumption and CO2 emissions.
tant contributor to greenhouse gases (GHGs), such as 28.6% of Previous studies have pointed out the divergence between
total GHG emissions in 2019 for the US (United States Envi- the real-world performance and official type-approval values,
ronmental Protection Agency U.S. EPA 2021) and 26% in 2018 meanwhile noting that its range becomes wider over time
for the European Union (EU) (European Environment Agency (Fontaras et al., 2017; Zheng et al., 2020). Based on real-world
EEA 2020). Among all transportation sectors, road vehicles fuel consumption data of more than 1.5 million passenger
are the largest source in China, accounting for up to 84.1% cars from China, the EU, the US, and Japan, the international
of total transportation GHG emissions, 50.9% of which origi- council on clean transportation (ICCT) reported that the di-
nate from light-duty vehicles (Ministry of Ecology and Envi- vergence increased from an 8% to 14% range in 2001 to a
ronment of the People’s Republic of China MEE 2020). Mitigat- 25% to 40% range in 2014 and that the growth in the diver-
ing GHG emissions (mainly carbon dioxide (CO2 )) from road gence was much more pronounced in China and the EU than
vehicles has been identified as an essential task in China, in the US (Tietge et al., 2017). The reasons for the growth of
responding to the Paris Climate Agreement (i.e., limit global the divergence in the above regions are probably a complex
warming to below 2°C preferably to 1.5°C, compared to pre- combination of various factors, however considerable stud-
industrial levels) and national climate goals (e.g., to achieve ies in China and the EU reveal that the design weaknesses of
carbon emissions peak before 2030 and carbon neutrality the New European Driving Cycle (NEDC) is the leading reason
by 2060). (Marotta et al., 2015; Zheng et al., 2020; Xing et al., 2021). In con-
To reduce GHG emissions and fuel consumption, fuel econ- trast, no unambiguous evidence of systematic manipulation is
omy or CO2 emission standards have been implemented by available in the US and Japan (Tietge et al., 2017). To address
most of the world’s major vehicle markets (e.g., China, Eu- the laboratory-to-road divergence issues of exhaust pollutant
rope, the US, and Japan). Taking light-duty gasoline vehicles emissions as well as fuel consumption, the Worldwide Harmo-
(LDGVs) as an example, China has issued the first fuel con- nized Light Vehicles Test Cycle (WLTC) has been developed by
sumption standards for individual vehicle models in 2004, the United Nations Economic Commission for Europe (UNECE)
which took effect in two phases (i.e., Phase I in 2005 and as the type-approval cycle, instead of the NEDC. The latest reg-
Phase II in 2008). In contrast, Phase III adopted in 2012 and ulations in China, the EU, and Japan have already adopted the
Phase IV in 2016 not only specify per-vehicle fuel consump- WLTC (MEP and AQSIQ, 2016; SAMR and SA, 2021; EU, 2017;
tion limits but also set corporate-average fuel consumption MOE, 2015). China also developed the China light-duty vehi-
(CAFC) targets for each manufacturer, targeting a fleet-average cle test cycle for passenger cars (CLTC–P) based on more lo-
fuel consumption of 6.9 L/100 km in 2015 and 5.0 L/100 km cal features of real-world road traffic (SAMR and SA, 2019).
in 2020 (MIIT, 2015). China now is implementing the Phase V Both the WLTC and CLTC-P exhibit more real-featured and
fuel consumption standards that took effect on July 1, 2021, transient traffic patterns, compared to the NEDC. In addition,
with a target of 4.0 L/100 km (95 g CO2 /km) by 2025 (SAMR and other factors have been identified that cause the increasing
SA, 2021). It is worth noting that worldwide progress on fuel divergence between real-world performance and official type-
economy standards is also remarkable in other major vehi- approval values. The boundary conditions specified in the lab-
cle markets. For example, the first US fuel economy stan- oratory test procedures of current fuel economy regulations
dard for light-duty vehicles can date back to 1975, aiming are generally loose and unrealistic (e.g., restricted use of ve-
to double the fleet average fuel economy of passenger cars hicle auxiliary systems and lower augmented vehicle mass)
from 13.6 miles per gallon (mpg) (1974) to 27.5 mpg (1985), (Zhang et al., 2014; Marotta et al., 2015; Fontaras et al., 2017).
while the current standards (i.e., 2012 rule) set a target of Hence, to achieve real-world benefits, it is very necessary to
an estimated 52.5 mpg (89 g CO2 /km) for passenger cars convert the type-approval values measured in the laboratory
by 2025 (Tietge et al., 2017). Since 2009 the EU has also set into on-road fuel savings and emission reductions.
binding CO2 targets mandating each passenger car manu- This study aims to clarify the divergence regarding fuel
facturer shall comply with an average CO2 emission at 130 consumption and CO2 emissions between laboratory and on-
g/km (2015) and 95 g/km (2021) under type-approval proce- road measurements and evaluate the impacts from on-road
dures (European Commission EC, 2009). These progressively driving conditions (e.g., air conditioning on, full loads, aver-
stringent fuel economy standards have posed substantial im- age vehicle speed). We recruited a typical GDI passenger car
pacts on the global automotive industry, motivating the appli- in China and collected its fuel consumption and CO2 emis-
cation of numerous technologies and fuel improvements that sions data. Laboratory tests over the standard test cycles (i.e.,
268 journal of environmental sciences 125 (2023) 266–276

CLTC-P, NEDC, and WLTC) were first conducted on a chassis


dynamometer. Then on-road measurements were performed
over the CLTC-P cycle on the high-speed loop test road by us-
ing a fuel consumption flow meter. Our findings can provide
meaningful first-hand data to policymakers and researchers,
and enhance their understanding regarding the impact of real
driving conditions on fuel consumption and CO2 emissions.

1. Materials and methods

1.1. Vehicle information

The comparative experiment of fuel consumption based on


chassis dynamometer tests and on-road measurements was
conducted in 2017. One LDGV in compliance with the China
5 (Euro 5 equivalent) emission standard was recruited from
a well-known Chinese automobile manufacturer. This tested
car is a sport utility vehicle (SUV) equipped with three-way
catalytic converters to control exhaust emissions. Market
gasoline from a certified gas station was used throughout this
study and the fuel quality complied with Beijing VI standards
with a sulfur content below 10 mg/kg. The detailed vehicle and
fuel specifications are: production year 2017, odometer ∼9000
km, curb mass 1709 kg, automatic transmission, inline four-
cylinder engine, direct-injection fuel supply, turbocharged air
intake, displacement 2.0 L, power rating 140 kW, and gasoline
with the research octane number of 92 as fuel. Fig. 1 – Vehicle speed profiles of (a) the China light-duty
vehicle test cycle for passenger cars (CLTC-P), (b) the New
1.2. Test cycles and experimental procedures European Driving Cycle (NEDC) and (c) the Worldwide
Harmonized Light Vehicles Test Cycle (WLTC) Class 3b. ECE
Three standard test cycles (i.e., the CLTC-P, NEDC and WLTC, 15: Econonic Comminssion for Europe 15; EUDC:
see Fig. 1) were selected to evaluate the fuel consumption extra-urban driving cycle.
and CO2 emission characteristics of the tested vehicle. The
CLTC-P comprises three phases with different allocation of
driving conditions, namely, low-speed phase (average speed consumption of potentially activated auxiliary devices (e.g.,
of 13.1 km/hr), medium-speed phase (30.7 km/hr) and high- air conditioning and entertainment systems, lights, and horn)
speed phase (50.9 km/hr), respectively. The total distance of (Zhang et al., 2021; Kühlwein, 2016; EU, 2017). In this study, we
the entire CLTC-P cycle is 14.5 km, and the duration is 1800 focused on the impact of driving conditions (including road
sec (SAMR and SA, 2019). The NEDC consists of four repe- load coefficients, test cycles, vehicle load mass), automatic
titions of urban driving segments (i.e., Econonic Commins- stop and start (STT) system, and major auxiliary equipment
sion for Europe (ECE) 15) featured with low engine load and (i.e., air conditioning) on fuel consumption and CO2 emissions.
low vehicle speed, followed by the extra-urban driving cy- Laboratory tests were first performed on a chassis dy-
cle (EUDC) which represents more aggressive and high-speed namometer, which was employed to simulate the driving re-
driving modes with a maximum vehicle speed of 120 km/hr sistances (i.e., road loads) imposed on the vehicle. In this
(MEP and AQSIQ, 2013; Marotta et al., 2015). The Class 3b WLTC study, three CLTC-P tests, two NEDC tests and one WLTC test
applicable to the highest power-to-mass vehicle categories with different road loads were conducted (Table 1). Road load
(81.9 W/kg for the tested vehicle), includes low-speed (18.9 refers to the forces resisting the vehicle forward motion and
km/hr), medium-speed (39.5 km/hr), high-speed (56.6 km/hr), is approximated by the three coefficients of a second-order
and extra high-speed (92.0 km/hr) phases, characterized by polynomial (EU, 2017; MEP and AQSIQ, 2016), as Eq. (1):
more dynamic vehicle speed profile, lower stop duration and
higher travel mileage (EU, 2017). The selected parameters of F = f 0 + f 1 × v + f 2 × v2 (1)
the three cycles are summarized in Table S1.
The vehicle was tested as received. Before each test, we where F (N) is the road load force as a function of vehicle
carefully checked and demonstrated the normal status of the speed, f0 (N), f1 (N•hr/km), f2 (N•hr2 /km2 ) are the applicable
vehicle including engine and after-treatment systems. Fuel road load coefficients, respectively, and v (km/hr) is the vehicle
consumption and CO2 emissions are determined by the vehi- speed.
cle’s driving resistances (e.g., aerodynamic drag, rolling resis- To accurately measure the fuel consumption and pollutant
tance, inertia resistances) affected by driving conditions and emissions, the vehicle on the chassis dynamometer needs to
vehicle load mass, the powertrain’s efficiency, and the energy overcome running resistances comparable to normal on-road
journal of environmental sciences 125 (2023) 266–276 269

Table 1 – Summary of testing conditions.

Chassis dynamometer tests

Driving condition Reference condition Reference condition Reference condition


Driving cycle CLTC-P NEDC WLTC
Determined method of road load Look-up table (1) Look-up table (1) Calculation (1)
coefficients (Test number) Coast-down (1) Coast-down (1)
Calculation (1)
Passengers 2 2 2
Windows Closed Closed Closed
Air conditioning Off Off Off
Automatic stop and start (STT) On On On
Ambient temperature 30°C
Relative humidity 45%

On-road measurements

Driving condition Reference condition STT off STT off and full load
Driving cycle CLTC-P CLTC-P CLTC-P
Test number 1 2 1
Passengers 2 2 5
Windows Closed Closed Closed
Air conditioning Off Off On
STT On Off Off
Ambient temperature 30-32°C
Relative humidity 72%-78%

driving. There are several methods to determine the road load the temperature of 22°C and the automatic air volume setting
coefficients of a test vehicle in current regulations (EU, 2017). scheme.
During our experiments, the road load coefficients of each test
were determined by different methods to evaluate the impact 1.3. Measurement systems and data processing
of driving resistances, including look-up table method based
on the vehicle reference mass (i.e., curb mass augmented All the chassis dynamometer tests were conducted in
by 100 kg), on-road coast-down test, and calculation method the China Automotive Technology and Research Center
based on vehicle parameters (i.e., test mass, vehicle width, and (CATARC)’s laboratory in Tianjin, China, where the fuel con-
height). In this study, the look-up table method and on-road sumption testing capability has been qualified by the Ministry
coast-down method comply with China 5 emission standard of Industry and Information Technology (MIIT) of China. A 48-
(MEP and AQSIQ, 2013), and the calculation method for default inch, 2-wheel-drive light-duty chassis dynamometer (Model
road load meets with China 6 emission standards (MEP and 4000 2WD MIM, BURKE E. PORTER Machinery Company, USA)
AQSIQ, 2016). Three road load curves for the tested vehicle are was used to operate the standard test cycles (i.e., CLTC-P,
illustrated in Fig. S1a. NEDC, and WLTC in this study). A constant volume sampler
Then on-road measurements were carried out over the (Model CVS 7200, HORIBA, Japan) integrated with an analysis
CLTC-P cycle on a high-speed loop test road to compare the system (Model MEXA-7200, HORIBA, Japan) was employed to
difference in fuel consumption between laboratory and on- measure gaseous emissions (e.g., CO2 , carbon monoxide (CO),
road tests and evaluate the impacts from the typical real- total hydrocarbon (THC) and nitrogen oxide (NOx )). According
world driving conditions (Table 1), for example, STT off, high to the recommendatory standard (i.e., GB/T 19233-2020), the
load mass and air conditioning usage. Throughout our tests, distance-specific fuel consumption is calculated by the carbon
auxiliary devices of the tested vehicle were switched off or de- balance method, as Eq. (2).
activated, except as indicated in Table 1. It should be worth
noting that the same passengers (including driver) were se- 100  
lected for multiple tests without personnel replacement dur- FC = × 0.273 × EFCO2 + 0.429 × EFCO + 0.866 × EFTHC
D × WC
ing the entire test. (2)
Reference condition: a test condition at an ambient tem-
perature of about 30°C with the air conditioning off, window
where FC (L/100 km) is the fuel consumption, D (g/L) is the
closed, automatic stop and start (STT) on and two passengers
density of gasoline, WC is the ratio of carbon mass to the total
seated in the tested LDGV; STT off: switching off the STT sys-
fuel mass, 0.866 for gasoline, EFCO2 (g/km), EFCO (g/km), and
tem in the case of reference condtion; STT off and full load:
EFTHC (g/km) are the distance-specific emission factors of CO2 ,
compared to the reference condition, the STT system was
CO, and THC, respectively.
turned off, passengers were increased from 2 to a full occu-
On-road fuel consumption measurements over the CLTC-
pancy (i.e., 5 people), and the air conditioning was adjusted
P were performed on a high-speed loop test road with a to-
and set to the inner circulation blowing mode according to
tal length of 7.85 km in Yancheng, Jiangsu Province, China.
270 journal of environmental sciences 125 (2023) 266–276

To achieve CLTC-P curve following, an advanced driving assis- distance-specific fuel consumption and CO2 emissions of each
tance system developed by the School of Vehicle and Mobil- laboratory test under the reference condition. Our measure-
ity, Tsinghua University was applied during the on-road tests. ment results indicate that the fuel consumption and CO2
Fig. S2 presents that the on-road vehicle speed profiles are al- emissions show a slight upward as the driving resistances in-
most within the error criteria as time changes. The error du- crease. For example, when the road load coefficients are ob-
ration of each on-road CLTC-P test was kept within 60 sec, tained by the look-up table method, fuel consumption and
accounting for less than 3.4% of the total test time (i.e., 1800 CO2 emission factor of the test vehicle over the CLTC-P are 8.35
sec), and the speed readjustment could be performed within L/100 km and 197 g/km. Since driving resistances determined
0.7 sec (Table S2), implying that the CLTC-P cycle can be accu- by look-up table method, on-road coast-down test, and calcu-
rately simulated during the on-road tests. Real-time fuel con- lation method increases in order (Fig. S1a), the fuel consump-
sumption was collected by a fuel consumption flow meter in- tion values based on coast-down test and calculation method
cluding a gasoline flow meter (Model YT-985/826, Aite, China) are higher 2.1% and 3.4% than the look-up table method, re-
and a speedometer (Model YT-1000, Aite, China). Based on spectively. Further, the fuel consumption under three different
the instantaneous fuel rate and vehicle speed, fuel consump- CLTC-P phases (i.e., low-, medium- and high-speed phases)
tion (L/100 km) for each operating trip were estimated with is calculated. The discrepancy of fuel consumption and CO2
Eq. (3). emissions mainly occurred under the low-speed and medium-
speed phases. Compared to the look-up table method, the fuel
T
100 t=1 FRt consumption increases by 0.8%-6.0% for the coast-down test
FC = T (3)
t=1 vt and 3.4%-6.5% for the calculation method (Fig. S2). However,
there is a very small distinction under the high-speed phase
where FRt (L/hr) is the instantaneous fuel rate at second t, vt (-0.6% and 1.1%). A possible reason is that the differences re-
(km/hr) is the instantaneous vehicle speed at second t, and T garding road loads gradually decrease as the vehicle speed in-
(sec) is the number of seconds for an operating trip. creases (Fig. S1b).
Furthermore, on-road CO2 emission factors were estimated Obviously, the results over different standard cycles also
by approximate back-calculation based on the carbon balance confirmed that the fuel consumption and CO2 emissions in-
method, as given in Eq. (4). It should be pointed out that the crease with the increase of road load coefficients (Fig. 2). We
CO and THC were ignored due to their low contribution to clearly observe that various aspects of inter-cycle distinctions
fuel consumption (Zhang et al., 2014). For this tested vehicle, lead to different impacts on fuel consumption and CO2 emis-
the results of our laboratory tests show that CO and THC are sions. Average fuel consumption values of the tested vehi-
responsible for an average share of less than 1% of fuel con- cle over the CLTC-P, NEDC, and WLTC are 8.50±0.14 L/100
sumption. km (N=3, the test number and hereinafter), 8.46±0.03 L/100
km (N=2), and 8.82 L/100 km (N=1), respectively. Note that
FC × D × WC the fuel consumption over the NEDC is on average 8.4%
EFCO2 = (4)
27.3
higher than the corresponding type-approval value of the
The technical specifications of the main laboratory and tested vehicle (7.8 L/100 km) with a standard deviation of
on-road analyzers are illustrated in Table S3. For each test, 0.4%, which is consistent with the finding (9%±8%) reported
the installation, calibration, measurement, and verification by Marotta et al. (2015). This is probably because, unlike the
of the above analyzers followed the instructions issued by test conditions in this study (Table S4), manufactures tend
the instrument manufacturers and the relevant official stan- to prefer more fuel-saving conditions (e.g., calm wind, low
dards (e.g., GB/T 19233-2008 and GB 18352.6-2016) to en- load mass, and low road slope) for coast-down parameters set
sure measurement accuracy. In addition, the same pro- during the type approval, although fuel consumption mea-
fessional driver who is skilled and trained operated the surments are also conducted by using the NEDC. The corre-
tested vehicle throughout the laboratory and on-road driving sponding CO2 emissions over the above three test cycles are
tests. 199±3 g/km (N=3), 200±1 g/km (N=2), and 206 g/km (N=1), re-
spectively. We further estimated the ratio of WLTC to CLTC-P
fuel consumption using the same road load coefficients (i.e.,
2. Results and discussion determined by the calculation method) with a value of 1.02.
The WLTC cycle is expected to increase fuel consumptions
2.1. Fuel consumption and CO2 emissions over the and CO2 emissions might be attributed to the fact that the
standard cycles with various road load coefficients higher proportions of transient and aggressive driving condi-
tions (e.g., higher acceleration and speed) in the WLTC than
According to the current certification requirements, the fuel the CLTC-P (Fig. 1 and Table S1) resulting in stronger accel-
consumption and CO2 emissions of LDGVs are normally mea- eration forces and more fuel usage during vehicle operating
sured on a chassis dynamometer under defined driving pa- (Mock et al., 2014; He et al., 2018). Moreover, a lower idling
rameters (e.g., standard test cycles) and specific external con- share in the WLTC (12.6%), roughly about half that of the CLTP-
ditions (e.g., road load coefficients and ambient conditions) P (22.1%), allows for a smaller fuel-saving benefit from acti-
(Tietge et al., 2017). In this study, a direct-injection gasoline vated STT systems, which might cause a slightly higher fuel
SUV was tested to address the different impacts of the stan- consumption growth than the CLTC-P. For the ratio NEDC to
dard test cycles and road load coefficients on vehicle fuel CLTC-P, a similar value of 1.00±0.01 is found, although the
consumption and CO2 emissions. Fig. 2 presents the detailed CLTC-P possesses more transient driving conditions. It is evi-
journal of environmental sciences 125 (2023) 266–276 271

Fig. 2 – (a) Fuel consumption and (b) CO2 emissions with different road load coefficients over various test cycles under
reference condition.

dent from our results that the difference in fuel consumption


among these three test cycles is not significant, which agrees
with several previous studies of the European Commission’s
Joint Research Centre (JRC) (Marotta et al., 2015; Pavlovic et al.,
2016). More descriptions of test cycle conversions among
worldwide light-duty vehicle fuel consumption (or CO2 emis-
sions) have been reported by Kühlwein et al. (2014).

2.2. Comparisons of laboratory and on-road fuel


consumption and CO2 emissions over the CLTC-P

To minimize the effects of vehicle configuration, operating


conditions, driving behaviors that could cause significant dis-
crepancies in fuel economy between the laboratory test and
real-world driving, we conducted a comparative experiment
using the same vehicle, with the same professional driver,
over the same standard test cycle. As shown in Fig. 3, on-road
fuel consumption and CO2 emission of the tested vehicle over
the CLTC-P are 10.44 L/100 km and 248 g/km under reference
condition, approximately 23% higher than those measured on
the chassis dynamometer in the laboratory and 34% higher
than the type-approval figure (7.8 L/100 km). An on-road test
study based on 60 passenger cars indicated real-world fuel
consumption is 30%±12% higher than type-approval figures
(Zhang et al., 2014). This great divergence, far higher than
that between the three standard test cycles, might be at-
tributed to a series of easily neglect factors, such as weather
conditions (e.g., wind, rain, and snow), road characteristics
(e.g., morphology, grade, and road surface) and vehicle turning
(Fontaras et al., 2017). These are prone to affect real-world fuel
Fig. 3 – Comparisons of laboratory and (a, b) on-road fuel
consumption and CO2 emissions but are currently difficult to
consumption and (c, d) CO2 emissions under the reference
reflect in chassis dynamometer tests. For example, a wind ve-
condition.
locity of 3 m/sec could cause changes in aerodynamic resis-
tance, either positive or negative, by up to 10% (Ligterink et al.,
2016), with respect to change in fuel consumption of about
272 journal of environmental sciences 125 (2023) 266–276

2% on average (Fontaras and Samaras, 2010). Additionally, ve- and load mass on real-world fuel consumption and CO2 emis-
hicles require more power output to overcome their gravity sions. Fig. 4 illustrates the results obtained from on-road mea-
when driving uphill than on flat roads. Park and Rakha (2005) surements over the CLTC-P that the fuel consumption and
reported that a 1.5% increase in road grade may trigger a 9% in- CO2 emission changes are strongly associated with on-road
crease in fuel consumption. It is noted that the selected high- driving conditions. We confirmed that the use of STT systems
speed loop road has very slight uphill and downhill slopes, in vehicles could provide benefits to fuel-saving (Sales et al.,
rather than using a fixed road grade as in the laboratory (zero 2017; Prati et al., 2018). With the STT system off, it is obviously
degrees set in this study). found that average fuel consumption and CO2 emissions of
Besides the vehicle driving tests, we collected large the tested vehicle over the CLTC-P are 10.89 L/100 km and 259
amounts of real-world fuel consumption data from 3239 g/km, 4% higher than those under the reference condition. In
owners of the same model vehicle with a total mileage of the case of vehicles deactivated with the STT systems, fuel
27.06 million kilometers by using XiaoXiongYouHao, a Chinese consumption and CO2 emissions may increase to some ex-
mobile application that allows users to track and compare tent, especially under driving conditions that include long pe-
their fuel consumption. It is necessary to highlight that the riods of idling (e.g., urban road and congestion conditions). As
driving behaviors of these vehicle owners were completely shown in Fig. 4, fuel consumption and CO2 emissions under
autonomous without any interference. This means that the low- and medium-speed phases of the CLTC-P increase by 12%
impact on fuel consumption from real driving factors, includ- and 4% on average, due to the STT system being turned off.
ing air conditioning, loads, road characteristics, weather con- However, for the CLTC-P high-speed phase with a low propor-
ditions, fuel variations, vehicle maintenance, driving behav- tion of idling (∼5.5%), the effect of the STT system is extremely
iors, etc., were not intentionally removed. Coincidentally, the slight (lower than 1%). This is attributed to the fact that the
average fuel consumption of 10.42 L/100 km from XiaoXiongY- STT system could shut down the engine during vehicle stop
ouHao, is not significantly different from that of on-road CLTC- phases and ideally reduce idling emissions to zero.
P test under the reference condition. Likewise, the result is 23% The impact of activated auxiliary devices and increased
and 34% higher than in-lab CLTC-P and official certification payloads on fuel consumption and CO2 emissions are also
value, respectively. This finding is new evidence implying that concerned in our study. We focused on evaluating on-road
chassis dynamometer testing is overestimating the fuel effi- fuel consumption and CO2 emissions under the scenario of
ciency of vehicles. The innovation center for energy and trans- air conditioning usage (set 22°C with internal air circula-
portation (iCET) in the past few years has also identified the tion) and full passenger capacity with five passengers (here-
great gap between XiaoXiongYouHao and type-approval fuel inafter referred to as full load). The fuel consumption and
consumption in the Chinese vehicle market (Qin et al., 2016; CO2 emission factor over the CLTC-P under full load increase
Ding et al., 2015). Qin et al. (2016) demonstrated that the aver- by 52%, compared to those under the reference condition. In
age gap increase was up to 31% in 2015 for the LDGVs equipped addition, we observe that full load leads to a significant in-
with automatic transmission based on a set of fuel consump- crease in fuel consumption and CO2 emissions under each
tion data from nearly 600,000 users of XiaoXiongYouHao. phase of the CLTC-P, 81% for the low-speed phase, 48% for
Average fuel consumption values under low-, medium- the medium-speed phase, and 43% for the high-speed phase
and high-speed phase of the CLTC-P in the laboratory are (Fig. 4). Several previous studies also demonstrated the im-
12.37±0.21, 7.58±0.25, and 7.84±0.07 L/100 km, with corre- pact of air conditioning and load mass on fuel consumption
sponding CO2 emission factors of 290±4, 178±5 and 184±2 or CO2 emissions. Air conditioning, one of single largest aux-
g/km, respectively (Fig. 3b and d). We can observe that on-road iliary systems for LDGVs, is a key device deteriorating the fuel
fuel consumption values under CLTC-P medium- and high- economy (Lee et al., 2013), due to the extra increase of engine
speed phases are higher by 24% and 38% than those in the loads (He et al., 2018). Lee et al., (2013) claimed that increase
laboratory, and the CO2 emission results are comparable. In rates of LDGV fuel consumption with air conditioning on ver-
contrast, there is no significant divergence between labora- sus off were ∼25% at a vehicle speed of 50 km/hr, ∼35% at 75
tory and on-road fuel consumption and CO2 emissions un- km/hr, and even up to 90% during idling operation. However,
der the CLTC-P low-speed phase. This finding may be under- Zacharof and Fontaras (2016) estimated that the average in-
standable because, unlike the laboratory chassis dynamome- crease in fuel consumption was only ∼9% due to air condition-
ter, the road selected in this study is a high-speed loop runway ing usage. The large variation in these studies may be owing
where not only the differences in headwinds and road char- to the lack of consensus on the measurement conditions (e.g.,
acteristics between the laboratory and actual road, but also air conditioning operation setting, ambient temperature, and
crosswinds, vehicle turning or changing lanes, etc., could ad- vehicle speed profile). The increase of vehicle load mass can
ditionally increase the vehicle driving resistance coefficients, increase fuel consumption because it demands more power to
especially the quadratic one, thus resulting in much greater accelerate the vehicle during acceleration phases and rolling
impact on the vehicle driving performances at high speeds resistance is increased proportionally (Fontaras et al., 2017). It
(Fontaras et al., 2017). has generally been reported a 100 kg increase account for a 5%
to 10% increase in fuel consumption (Mickūnaitis et al., 2007;
2.3. Impacts from STT and vehicle loads on the on-road Fontaras et al., 2010). In this study, the full passenger capacity
fuel consumption and CO2 emissions was increased by three passengers compared to the reference
condition, roughly 200 kg.
Besides the above on-road test under the reference condition, Traffic conditions have been recognized as a substantial
we explored the impact of the STT system, air conditioning, impact on the fuel economy of LDGVs (Zhang et al., 2014;
journal of environmental sciences 125 (2023) 266–276 273

Fig. 4 – Estimated increase in (a) fuel consumption and (b) CO2 emission over the CLTC-P phases under different on-road
driving conditions.

sitive to average vehicle speed exceeding 25 km/hr. Based on


the fitting functions, we conducted the sensitivity analysis to
quantitatively evaluate the impact of average speed on fuel
consumption. When average vehicle speed decreased from 25
to 10 km/hr, representing typical traffic congestions, fuel con-
sumption was estimated to increase by 63%, while the varia-
tion in fuel consumption from 25 to 45 km/hr ranged from -5%
to 8%.

3. Conclusions

Fuel consumption and CO2 emissions data of a GDI passenger


car were collected over the standard test cycles (e.g., CLTC-
P, NEDC, and WLTC) based on chassis dynamometer tests in
the laboratory and on-road measurements on a high-speed
loop test road. Laboratory testing results indicate that there
are slight variations, less than 4%, in fuel consumption and
CO2 emissions between the three sets of road load coefficients
Fig. 5 – Correlations of relative fuel consumption and specified according to different methods (i.e., look-up table,
average vehicle speed under 11 short trips of the CLTC-P. coast-down test and calculation method), although they tend
to increase with the increase of road loads. Likewise, the dif-
ferences in the in-lab fuel consumption and CO2 emissions
between the three standard cycles mentioned above are also
Zheng et al., 2020). To better illustrate the impact of real-world not significant. Average fuel consumption values are 8.50±0.14
driving mode on fuel consumption, we divided the CLTC-P into L/100 km over the CLTC-P, 8.46±0.03 L/100 km over the NEDC,
eleven short trips determined by the idling times and calcu- and 8.82 L/100 km over the WLTC, with the corresponding CO2
lated the fuel consumption of each short trip (Fig. S4). Our emissions of 199±3, 200±1, and 206 g/km, respectively. We ob-
results indicate that average speed may be a dominant fac- serve that the fuel consumption over the NEDC was on av-
tor affecting fuel consumption. A strong relationship (R2 ≥ erage 8.4%±0.4% higher than the corresponding NEDC type-
0.93) is identified between relative fuel consumption and the approval value of the tested vehicle (7.8 L/100 km).
average speed of the short trips (Fig. 5). We find that a dra- According to a comparative experiment over the CLTC-P,
matic increase in fuel consumption under traffic congestion we find that on-road fuel consumption and CO2 emissions are
due to its lower average speed and frequent vehicle speed 10.44 L/100 km and 248 g/km, approximately 23% higher than
variability (i.e., accelerations and decelerations) (Bigazzi and those measured in the laboratory and 34% higher than the
Clifton, 2015). In contrast, the fuel consumption is less sen- type-approval value. Coincidentally, average fuel consump-
274 journal of environmental sciences 125 (2023) 266–276

tion estimated by a large amount of real-world data collected Ding, D., Ben Dror, M., Kang, L., An, F., 2015. Real-world and
from XiaoXiongYouHao is not significantly different from the certified fuel consumption gap anlaysis. The Innovation
on-road CLTC-P testing value (10.42 L/100 km), further demon- Center for Energy and Transportation.. Available: https:
//www.efchina.org/Attachments/Report/report- ctp- 20150810/
strating the overestimation of vehicle’s fuel economy by the
real- world- and- certified- fuel- consumption- gap- analysis .
laboratory tests. We also explore the impact of the STT system,
Accessed September 25, 2021.
air conditioning, and vehicle load mass on fuel consumption. Edwards, S., Müller, R., Feldhaus, G., Finkeldei, T., Neubauer, M.,
Our measurement results indicate that the fuel consumption 2008. The reduction of CO2 emissions from a turbocharged DI
and CO2 emissions over the entire CLTC-P increase by approx- gasoline engine through optimised cooling system control.
imately 4% due to the STT system off, and we observe that MTZ worldwide 69, 12–17.
the increase mostly occurs at low- and medium-speed phases European Environment Agency (EEA), 2020. Annual European
Union greenhouse gas inventory 1990–2018 and inventory
where the idling share is comparatively high. It is crucial not-
report 2020.. Available
ing that air conditioning usage and full load (i.e., full passen- https://www.eea.europa.eu/publications/
ger capacity for LDGVs) increase more fuel consumption and european- union- greenhouse- gas- inventory- 2020 . Accessed
CO2 emissions by up to 52% over the CLTC-P. Of these, the September 30, 2021.
low-, medium- and high-speed phases of the CLTC-P increase European Commission (EC), 2009. Regulation No. 443/2009 of the
by 81%, 48%, and 43%, respectively. Furthermore, we identi- European Parliament and of the Council of 23 setting
emission performance standards for new passenger cars as
fied strong correlations (R2 > 0.9) between relative fuel con-
part of the Community’s integrated approach to reduce CO2
sumption and average vehicle speed of short trips. Under traf-
from light-duty vehicles. Available:
fic congestion, fuel consumption increases dramatically. For https://eur-lex.europa.eu/LexUriServ/LexUriServ.do?uri=OJ:L:
example, fuel consumption was estimated to increase by 63% 2009:140:0001:0015:EN:PDF. Accessed September 30, 2021.
when vehicle speed decreased for 25 to 10 km/hr. Therefore, European Environment Agency (EEA), 2021. Size of the vehicle
real-world driving conditions cannot be ignored when evalu- fleet in Europe.. Available:
ating the fuel consumption and GHG emissions of LDGVs. We https://www.eea.europa.eu/data- and- maps/indicators/
size- of- the- vehicle- fleet/size- of- the- vehicle- fleet- 10 .
suggest that future regulation designs regarding fuel economy
Accessed September 25, 2021.
and GHGs reduction should take full consideration of these
European Union (EU), 2017. Commission Regulation (EU)
complicated affecting factors and move from controlled lab- 2017/1151 of 1 June 2017 supplementing Regulation (EC) No
oratory to real-world road as pollutant emission regulations 715/2007 of the European Parliament and of the Council on
have done (e.g., the real driving emission (RDE) standard). type-approval of motor vehicles with respect to emissions
from light passenger and commercial vehicles (Euro 5 and
Euro 6) and on access to vehicle repair and maintenance
information, amending Directive 2007/46/EC of the European
Declaration of Competing Interest Parliament and of the Council, Commission Regulation (EC)
No 692/2008 and Commission Regulation (EU) No 1230/2012
The authors declare that they have no known competing fi- and repealing Commission Regulation (EC) No 692/2008.
nancial interests or personal relationships that could have ap- Available: https://eur-lex.europa.eu/legal-content/EN/TXT/
PDF/?uri=CELEX:32017R1151&from=IT. Accessed September
peared to influence the work reported in this paper.
27, 2021.
Fontaras, G., Samaras, Z., 2010. On the way to 130 g
CO2 /km—Estimating the future characteristics of the average
Acknowledgments European passenger car. Energy Policy 38, 1826–1833.
Fontaras, G., Zacharof, N., Ciuffo, B., 2017. Fuel consumption and
CO2 emissions from passenger cars in Europe - Laboratory
This work was sponsored by the National Natural Science
versus real-world emissions. Prog. Energ. Combust. 60, 97–131.
Foundation of China (Nos. 52170111 and 41977180) and the
He, L., Hu, J., Zhang, S., Wu, Y., Zhu, R., Zu, L., et al., 2018. The
first China First Automobile Works (FAW)-Volkswagen China impact from the direct injection and multi-port fuel injection
Environmental Protection Foundation automobile environ- technologies for gasoline vehicles on solid particle number
mental protection innovation leading plan. The contents of and black carbon emissions. Appl. Energy 226, 819–826.
this paper are solely the responsibility of the authors and do He, X., Kim, H., Wallington, T., Zhang, S., Shen, W., Kleine, R.D.,
not necessarily represent the official views of the sponsors. et al., 2020. Cradle-to-gate greenhouse gas (GHG) burdens for
aluminum and steel production and cradle-to-grave GHG
benefits of vehicle lightweighting in China. Resour. Conserv.
Recy. 152, 104497.
Appendix A Supplementary data Kühlwein, J., German, J., Bandivadekar, A., 2014. Development of
test cycle conversion factors among worldwide light-duty
Supplementary material associated with this article can be vehicle CO2 emission standards. The International Council on
Clean Transportation, Berlin. Available.
found in the online version at doi:10.1016/j.jes.2021.12.042.
Kühlwein, J., 2016. Driving resistances of light-duty vehicles in
Europe: present situation, trends, and scenarios for 2025
references (White paper). The International Council on Clean
Transportation. Available.
Lee, J., Kim, J., Park, J., Bae, C., 2013. Effect of the air-conditioning
system on the fuel economy in a gasoline engine vehicle. P. I.
Bigazzi, A.Y., Clifton, K.J., 2015. Modeling the effects of congestion Mech. Eng. D J. Aut. 227, 66–77.
on fuel economy for advanced power train vehicles.
Transport. Plan. Techn. 38 (2), 149–161.
journal of environmental sciences 125 (2023) 266–276 275

Ligterink, N., Smokers, R., Spreen, J., Mock, P., Tietge, U., 2016. TNO Innovation Center for Energy and Transportation, Beijing.
report. Available.
Liu, H., Wang, X., Zhang, D., Dong, F., Liu, X., Yang, Y., et al., 2019. Sales, L.C., Souza, L.G., Monteiro, L.G., Rodrigues, M.S., Borges, F.R.,
Investigation on blending effects of gasoline fuel with 2017. Comparison of energy consumption and CO2 reduction
n-butanol, DMF, and ethanol on the fuel consumption and with combined application of the start and stop and
harmful emissions in a GDI vehicle. Energies 12, 1845. intelligent alternator in a flex fuel vehicle. SAE Technical
Marotta, A., Pavlovic, J., Ciuffo, B., Serra, S., Fontaras, G., 2015. Paper No. 2017-36-0115.
Gaseous emissions from light-duty vehicles: Moving from State Administration for Market Regulation of People’s Republic
NEDC to the new WLTP test procedure. Environ. Sci. Technol. of China (SAMR) and Standardization Administration of
49, 8315–8322. People’s Republic of China (SA), 2019. China automotive test
Mickūnaitis, V., Pikūnas, A., Mackoit, I., 2007. Reducing fuel cycle-Part 1: Light-duty vehicles. GB/T 38146.1-2019 (in
consumption and CO2 emission in motor cars. Transport 22, Chinese). Available: http://c.gb688.cn/bzgk/gb/showGb?type=
160–163. online&hcno=83CCA04B91B9B5FC6F19BD54E23320B7.
Ministry of Ecology and Environment of the People’s Republic of Accessed September 30, 2021.
China (MEE), 2020. China mobile source environmental State Administration for Market Regulation of the People’s
management annual report (in Chinese). Available Republic of China (SAMR), Standardization Administration of
http://www.mee.gov.cn/hjzl/sthjzk/ydyhjgl/202008/ the People’s Republic of China (SA), 2021. Fuel consumption
P020200811521365906550.pdf . Accessed limits for passenger cars. GB 19578-2021 (in Chinese).
September 21, 2021. Available: http://c.gb688.cn/bzgk/gb/showGb?type=
Ministry of Environmental Protection of the People’s Republic of online&hcno=57DA347FDF48E4743786873E5B2D670A .
China (MEP) and Administration of Quality Supervision and Accessed September 30, 2021.
Inspection Quarantine of the People’s Republic of China Tietge, U., Díaz, S., Yang, Z., Mock, P., 2017. From laboratory to road
(AQSIQ), 2013. Limits and measurement methods for international: A comparison of official and real-world fuel
emissions from light-duty vehicles (China 5), GB 18352.5-2013 consumption and CO2 values for passenger cars in Europe,
(in Chinese). Available: http://www.mee.gov.cn/ywgz/fgbz/bz/ the United States, China, and Japan. The International Council
bzwb/dqhjbh/dqydywrwpfbz/201309/t20130917_260352.shtml on Clean Transportation, Berlin. Available:.
. Accessed September 27, 2021. United States Department of Transportation, Bureau of
Ministry of Environmental Protection of the People’s Republic of Transportation Statistics (U.S. BTS), 2021. National
China (MEP) and Administration of Quality Supervision and Transportation Statistics of the United States. Available https:
Inspection Quarantine of the People’s Republic of China //www.bts.dot.gov/topics/national- transportation- statistics .
(AQSIQ), 2016. Limits and measurement methods for Accessed September 30, 2021.
emissions from light-duty vehicles (China 6), GB 18352.6-2016 United States Environmental Protection Agency (U.S. EPA), 2021.
(in Chinese). Available http://www.mee.gov.cn/gkml/hbb/bgg/ Inventory of U.S. Greenhouse Gas Emissions and Sinks
201612/t20161223_369497.htm . Accessed September 27, 2021. 1990-2019. Available https://www.epa.gov/ghgemissions/
Ministry of Industry and Information Technology of the People’s inventory- us- greenhouse- gas- emissions- and- sinks .
Republic of China (MIIT), 2015. Interpretation of the Phase IV Accessed September 30, 2021.
of fuel consumption limits for passenger cars (in Chinese). Wen, M., Zhang, C., Yue, Z., Liu, X., Yang, Y., Dong, F., et al., 2020.
Available https://www.miit.gov.cn/zwgk/zcjd/art/2020/ Effects of gasoline octane number on fuel consumption and
art_b3c3769150d041daac282a6c951b1524.html . Accessed emissions in two vehicles equipped with GDI and PFI
September 27, 2021. spark-ignition engine. J. Energ. Eng. 146, 04020069.
Ministry of the Environment of Japan (MOE), 2015. Light-duty: Wu, X., Zhang, S., Guo, X., Yang, Z., Liu, J., He, L., 2019. Assessment
New Post Long-Term Standards (In Japanese). Available of ethanol blended fuels for gasoline vehicles in China: Fuel
http://www.env.go.jp/air/car/gas_kisei/kisei.pdf . Accessed economy, regulated gaseous pollutants and particulate
August 18, 2021. matter. Environ. Pollut. 253, 731–740.
Mock, P., Tietge, U., Bandivadekar, A., German, J., 2014. The WLTP: Wu, Y., Yang, Z., Lin, B., Liu, H., Wang, R., Zhou, B., et al., 2012.
How a new test procedure for cars will affect fuel Energy consumption and CO2 emission impacts of vehicle
consumption values in the EU. The International Council on electrification in three developed regions of China. Energy
Clean Transportation, Berlin. Available:. Policy 48, 537–550.
National Bureau of Statistic of China (NBSC), 2021. China energy Wu, Y., Zhang, S., Hao, J., Liu, H., Wu, X., Hu, J., et al., 2017. On-road
statistical yearbook 2020 (in Chinese). Available vehicle emissions and their control in China: A review and
http://www.stats.gov.cn/tjsj/ndsj/2021/indexch.htm . outlook. Sci. Total Environ. 574, 332–349.
Accessed September 25, 2021. Xing, J., Liu, D., Jiang, S., Yu, H., Liu, Y., 2021. Research on energy
Ntziachristos, L., Mellios, G., Tsokolis, D., Keller, M., Hausberger, S., consumption test methods of light-duty pure electric vehicles
Ligterink, N., et al., 2014. In-use vs. type-approval fuel based on China automobile test driving cycle. In: IOP
consumption of current passenger cars in Europe. Energy Conference Series: Earth and Environmental Science, p. 835.
Policy 67, 403–411. Zacharof, N, Fontaras, G., 2016. Review of in use factors affecting
Pavlovic, J., Marotta, A., Ciuffo, B., 2016. CO2 emissions and energy the fuel consumption and CO2 emissions of passenger cars.
demands of vehicles tested under the NEDC and the new JRC science for policy report. European Commission Joint
WLTP type approval test procedures. Appl. Energy 177, Research Centre, Ispra (VA). Available.
661–670. Zhang, D., Gao, J., Tang, D., Wu, X., Shi, J., Chen, J., et al., 2021.
Prati, M., Costagliola, M.A., Pagliara, F., Mastantuono, E., 2018. Switching on auxiliary devices in vehicular fuel efficiency
Idling vehicle emissions and fuel consumption in urban use: tests can help cut CO2 emissions by millions of tons. One
Influence of the stop&start technology. In: 2018 IEEE Earth 4, 135–145.
International Conference on Environment and Electrical Zhang, S., Wu, Y., Liu, H., Huang, R., Un, P., Zhou, Y., et al., 2014.
Engineering and 2018 IEEE Industrial and Commercial Power Real-world fuel consumption and CO2 (carbon dioxide)
Systems Europe (EEEIC /I&CPS Europe), pp. 1–4. emissions by driving conditions for light-duty passenger
Qin, L., Ben Dror, M., Kang, L., Sun, H., An, F., 2016. Real-world and vehicles in China. Energy 69, 247–257.
certified fuel consumption gap analysis 2016 (in Chinese). The
276 journal of environmental sciences 125 (2023) 266–276

Zheng, X., Lu, S., Yang, L., Yan, M., Xu, G., Wu, X., et al., 2020. Zhu, R., Hu, J., Bao, X., He, L., Zu, L., 2017. Effects of aromatics,
Real-world fuel consumption of light-duty passenger vehicles olefins and distillation temperatures (T50 & T90) on particle
using on-board diagnostic (OBD) systems. Front. Environ. Sci. mass and number emissions from gasoline direct injection
Eng. 14, 1–10. (GDI) vehicles. Energy Policy 101, 185–193.

You might also like