Science of the Total Environment 367 (2006) 441 447

www.elsevier.com/locate/scitotenv

Short communication

A tank-to-wheel analysis tool for energy and emissions studies in

road vehicles
C.M. Silva , G.A. Gonalves, T.L. Farias, J.M.C. Mendes-Lopes
Mechanical Engineering Department, Instituto Superior Tcnico, Technical University of Lisbon, Avenida Rovisco Pais, 1049-001 Lisbon, Portugal
Received 22 June 2005; received in revised form 10 February 2006; accepted 14 February 2006
Available online 6 March 2006

Abstract
Currently, oil based fuels are the primary energy source of road transport. The growing need for oil independence and CO2
mitigation has lead to the increasing importance of alternative fuel usage. CO2 is produced not only as the fuel is used in the vehicle
(tank-to-wheel contribution), but also upstream, from the fuel extraction to the refueling station (well-to-tank contribution), and the
life cycle of the fuel production (well-to-wheel contribution) must be considered in order to analyse the global impact of the fuel
utilization. A road vehicle tank-to-wheel analysis tool that may be integrated with well-to-tank models was developed in the present
study. The integration in a demonstration case study allowed to perform a life cycle assessment concerning the utilization of diesel
and natural gas fuels in a specific network line of a bus transit company operating in the city of Porto, Portugal.
2006 Elsevier B.V. All rights reserved.
Keywords: Instantaneous fuel consumption and emissions model; Road vehicles; Internal combustion engines; Fuel life cycle analysis

1. Introduction
The emission of vehicle pollutants to the atmosphere
is an important health and environmental issue. For
example, hydrocarbons (HC), nitric oxides (NOx),
carbon monoxide (CO) and particles (PM) have direct
impacts on human health. Carbon dioxide (CO2) has no
direct health effect at the concentrations present in the
atmosphere but contributes to global warming. Currently
oil based fuels are the primary energy source of road
transport. The growing need for oil independence and
CO2 mitigation has lead to an increasingly importance of
Corresponding author. DTEA- Transportes, Energia e Ambiente,
Instituto Superior Tcnico, Av. Rovisco Pais, Pav. de Mecnica I, 2
andar, 1049-001 Lisboa, Portugal. Tel.: +351 218417632; fax: +351
218417640.
E-mail address: carlasilva@navier.ist.utl.pt (C.M. Silva).
0048-9697/$ - see front matter 2006 Elsevier B.V. All rights reserved.
doi:10.1016/j.scitotenv.2006.02.020

alternative fuel usage, including etanol, biodiesel and

hydrogen. Because CO2 is produced not only as the fuel
is used in the vehicle (tank-to-wheel contribution), but
also upstream, from the fuel extraction to the refueling
station (well-to-tank contribution), life cycle studies of
the fuel production (well-to-wheel contribution) must be
considered in order to analyse the global impact of the
fuel utilization.
The main contribution of this research work is the
creation of a model to simulate the use of conventional
and alternative fuels in road vehicles in terms of energy
consumption and emissions of HC, CO, NOx and CO2
(tank-to-wheel contribution). The main purpose of the
paper is to demonstrate the potentialities of this analysis
tool, namely the integration of its results with
commercial available life cycle software tool (well-totank) for fuel life cycle analysis assessments (well-towheel). The case study presented concerns the direct

442

C.M. Silva et al. / Science of the Total Environment 367 (2006) 441447

comparison of a Diesel bus and a CNG-Compressed

Natural Gas bus accounting also for the energy
consumption and CO2 emissions of diesel and natural
gas production pathways. The present paper is organized
as follows: in Section 2, the tank-to-wheel simulation
model is briefly described. Quantitative validity of the
model is presented in terms of an example, Fig. 2, in a
light-duty vehicle application. Section 3 presents the
well-to-tank model adopted in this study. Section 4

presents the case study along with model simulations.

Results for the buses are shown in Figs. 35. Some
parameters of the model are fit, with results shown in Fig.
3. Comparison between fuel consumption experimental
results monitored for Diesel buses and model predictions
is shown in Fig. 4. The toxic pollutants calculated are
presented in Fig. 5, but not compared with measurements, due to the lack of monitored data. Finally, in
Section 5, the conclusions are drawn.

2. Tank-to-wheel analysis tool

2.1. Description
As part of a research program, a numerical model was developed for simulation of fuel consumption and emissions
of road vehicles equipped with internal combustion engines, using conventional or alternative fuels. Hybrid and fuel
cell technologies are currently being added to the model. The model, named EcoGest (Silva et al., 2004), is coded in
Visual Basic programming language and is described with the minimum number of parameters required for a proper
simulation of the reality, allowing an easy use and a simpler calibration process. Fig. 1 presents EcoGest scheme for the
calculations.
Model calculations are based on the motor vehicle dynamics in order to estimate, at each second, the engine load and
speed. Engine load is obtained dividing the brake power by the maximum power at the same rotation speed (rpm).
Brake power (Pb) is generically obtained from:


1 1
2
d q d Cd d Af d m Cr d M d gd cosh M d gd senh Cir  1d Mc M d a d m Pac
1
Pb t
g 2 a
where is the transmission efficiency, a is the vehicle acceleration, Af is the vehicle frontal area, Cd is the drag
coefficient, Cr is the rolling resistance coefficient, Cir is the rotational inertia coefficient higher or equal to one, g is the
acceleration due to gravity, M is the loaded vehicle weight, Mc is the vehicle curb weight, Pac is the average power
consumption of vehicle accessories, is the road grade and a is the air density.
Engine speed (n) is obtained from the instantaneous vehicle speed, selected gear and tire dynamical radius (r), with
in m/s (Bosch, 2004):
60mi
2
n
2pr
where i is the gear ratio between engine and drive wheels.
text files
(engine maps, driving cycle,
ocupation, road)
rpm

input
model parameters
(options for calculation,
ambient temperature,
lubricating oil, vehicle
characteristics, ocupation,
etc.)

Dynamics
load

Engine
warm regime:
fuel consumption,
emissions;
exhaust
temperature

Catalitic
converter

Cold regime:
correction of fuel
consumption and
emissions

Fig. 1. Schematic of EcoGest model.

Fuel
consumptio
n, tailpipe
emissions
text and Excel
files

output

C.M. Silva et al. / Science of the Total Environment 367 (2006) 441447

up to 13.33

CO2 (g/s)

10

443

up to 14.39

8
6
4
2
0
0

200

400

600

800

1000

1200

1400

t (s)

CO2 (g/s)

Measured

EcoGest

CMEM

ADVISOR
up to 13.33

10
8
6
4
2
0
0

50

100

150

200

t (s)
Fig. 2. Example of measured and predicted instantaneous traces for CO2 emissions (light-duty vehicle with a 1.6l 4 in-line cylinder gasoline engine;
FTP75 cycle; start temperature of 20 C).

With this information (engine load and speed) bilinear interpolation of stationary and fully warm
characteristic maps of the engine allows to estimate fuel consumption and engine-out emissions. The submodel
for the spark ignition engine map generation calculates the hourly fuel consumption through the indicated
efficiency definition (Heywood, 1988). The frictional losses are estimated through the Patton et al. (1989)
model. The indicated efficiency is estimated through Wu and Ross (1999) methodology. Engine-out emissions
and exhaust temperature are estimated as a function of the fuel consumption though experimental correlations.
In case of a cold-start or part-warm start some factors concerning friction (oil temperature) and mixture
preparation (coolant temperature) are applied to the hot fuel and emissions allowing to estimate the cold fuel
consumption and engine-out emissions in such situations (Silva et al., 2004). The exhaust after-treatment
submodel (Silva et al., in press) allows estimating tailpipe emissions and is based on one-dimensional finite
difference analysis of the exhaust pipe and catalytic converter. In other models, like CMEM (Barth et al., 2001)
the catalyst efficiency is represented by parameters fitted to data, while in EcoGest it is calculated from
physical principles. An evaluation study of this submodel (Silva et al., in press) showed that the catalyst
temperatures are predicted with an error typically less than 10% and the conversion efficiencies errors are
typically of 6%. CO2 is calculated proportionally to fuel consumption deducting the CO2 not formed due to
tailpipe HC and CO formation.
2.2. Validation
Each submodel and the main model were validated using data measured or obtained from the literature. The
comparison between experimental data, including on-board measurements, and EcoGest model predictions
shows that fuel consumption and CO2 emission errors (relative to experimental data) are typically less than
10%. For the toxic pollutants (HC, CO and NOx) the level of error is higher and can reach up to 100%.
Comparison with other simulation models like the well known CMEM (Barth et al., 2001) and ADVISOR
(Brooker et al., 2002) revealed that these levels of errors are typical for the case studies analysed. Typically the
instantaneous traces concerning fuel consumption and emissions are correctly predicted (especially for fuel
consumption and CO2 emissions), although CMEM model predicts spikes not observed in reality (see Fig. 2
for an example).

444

C.M. Silva et al. / Science of the Total Environment 367 (2006) 441447

3. Well-to-tank analysis tool

For the well to tank analysis, a commercial life cycle
analysis software was used, GaBi (IKP and PE Europe
BmbH, 2003). This tool allows the user to incorporate in
each product all the emissions and resources used during
its production. In the present case it was used to assess to
total energy consumption and emissions associated with
the production of fuels, but it is also possible to analyse
the use of raw materials in the production of goods.
In the case of fuels, the results are presented in MJ per
MJ and CO2 per MJ of fuel in the tank. Taking the
example of natural gas production (Table 2), in order to
have 1 MJ of natural gas in the tank at 200 bar it is
necessary to use 1.029 MJ of energy: one MJ of natural
gas plus 0.029 MJ of electricity for the compression. The
same reasoning applies to all the processes.

Two real buses of the bus company Sociedade de

Transportes Colectivos do Porto (STCP), in Porto,
Portugal were selected (Diesel Mercedes Citaro, EuroIII,
205 kW, 11460 kg bus and CNG MAN NL 310, EuroIII,
180 kW, 12300 kg bus). STCP has a fleet comprising ca.
400 Diesel buses and 175 compressed natural gas buses.
The selected route was line 20 of the network. Line 20 is
about 7.8km long, with road grades varying between
15% and +10%. The Diesel bus was monitored within
this line, along 1 day. The monitoring devices consist of a
flowmeter installed in the bus to measure fuel consumption and an altimeter to give the topography of the route.
These devices were connected to an acquisition system
and to a portable computer along with the speed
information sensor from the bus tachometer. Passengers
were counted manually along the trips.
4.1. Tank-to-wheel analysis: EcoGest model

4. Case study
The Diesel bus driving cycles (speed against time),
the bus characteristics (including engine maps), the bus
occupation and the topography of line 20 were
introduced as inputs in EcoGest. Then the parameters
Pac, Cir and Cr (see Eq. (1)) were adjusted to one
measured trip and used for the remains. Their values
were found to be, respectively, 8kW, 1.05, 0.006, that
are typical values for this kind of vehicles. Fig. 3 shows
the comparisons, for some trips, of the predicted and
measured fuel consumption, at different hours of the day.
EcoGest was then used to simulate the same trips for the

6
5

8:25 am

14:56 pm

3
2

Predicted

Measured

3
Predicted

Measured

0
0

1000

2000

1000

t (s)

2000

t (s)
Fuel consumption (l)

Fuel consumption (l)

In order to compare the fuel consumption, green

house gas (GHG) emissions, and toxic pollutant
emissions (HC, CO, NOx) between a Diesel bus and a
CNG-Compressed Natural Gas bus, EcoGest model was
used to simulate the fuel utilization in the buses and the
GaBi model (IKP and PE Europe BmbH, 2003) was
used to analyse the well-to-tank path of both fuels. The
integration of the results of the two models allowed
performing a fuel life cycle assessment concerning fuel
consumption and GHG emissions.

6
5

18:55 pm

4
3
2

Predicted

Measured

0
0

1000

2000

3000

t (s)
Fig. 3. Example of accumulated fuel consumption trip results for several time periods.

3000

120

2500

100

2000

CO2 g/km

l/100km

C.M. Silva et al. / Science of the Total Environment 367 (2006) 441447

80
60
40
20

CNG

0
00:00

04:48

Diesel

09:36

14:24

1500
1000
500

Flow meter

CNG
0
00:00

00:00

19:12

445

04:48

09:36

Hour

14:24

Diesel
00:00

19:12

Hour

Fig. 4. Fuel consumption (diesel equivalent) and CO2 emission prediction (EcoGest) for the diesel and CNG bus. Flow meter stands for the monitored
data.

CNG bus. Figs. 4 and 5 show the average trip based

results for the whole sample of trips.

The fuel is then supplied by a road tanker to the STCP

depot, where the buses are filled.
According to Galpenergia (fuel supplier), the crude
imported in 2002 was supplied by South America
(19.7%), Western Africa (28.5%), CIS (Commonwealth
of Independent States) (5.3%), Middle East (27.2%),
North Sea (13.2%) and Mediterranean (6.1%). According to existing studies (Argonne National Laboratory,
1999; General Motors, 2002), the efficiency of crude
extraction is assumed to be 98% (this efficiency refers to
the amount of energy spent in extracting one unit of
energy in the form of crude, in this case, 0.02 MJ of
energy are needed to extract 1.00 MJ of crude). The
emissions associated with the extraction were calculated
according to the same sources, providing a value of
3245 g CO2 equiv per MJ of crude extracted. The
transport of the crude includes transport from oil wells
to central storage, from there to terminal storage, to

4.2. Well-to-tank analysis: GaBi model

For the well-to-tank analysis we consider information available in GaBi database and country specific
information when available (Calhau et al., 2004). Only
emissions of gases that contribute to the greenhouse
effect, CO2, CH4 and N2O expressed as CO2 equivalent
according to IPCC-Intergovernmental Panel on Climate
Change conversion factors (Calhau et al., 2004), were
considered.

0.2
0.15
0.1
0.05
0
00:00

CO2 g/km

0.4
0.35
0.3
0.25

CNG
04:48

09:36

14:24

Diesel
19:12

00:00

4
3.5
3
2.5
2
1.5
1
0.5
0
00:00

CNG

04:48

09:36

Hour

14:24

Diesel

19:12

Hour

100

NOx g/km

HC g/km

4.2.1. Diesel path way

For the production of diesel fuel it is assumed that all
the crude is supplied to the Leixes refinery (ca. 10km
north of Porto) through sea transport, where it is refined.

10
1
0.1
CNG
0.01
00:00

04:48

09:36

14:24

Diesel
19:12

00:00

Hour
Fig. 5. Pollutant emissions of HC, CO and NOx predicted for the diesel and CNG bus (EcoGest).

00:00

446

C.M. Silva et al. / Science of the Total Environment 367 (2006) 441447

Table 1
Energy consumption, losses and GHG emission for diesel fuel
production (Calhau et al., 2004)

Crude
extraction
Crude
transport
Crude
refining
Diesel fuel
distribution
Vehicle
filling
Total

Energy
supplied

Energy losses

GHG emission

(MJ/
MJDiesel)

(MJ/
MJDiesel)

(g/MJDiesel)

1.154

0.020

12.9

3.25

22.8

1.134

0.005

3.6

0.38

2.8

1.129

0.124

9.71

68.4

1.005

0.002

1.4

0.14

1.0

1.003

0.003

2.1

0.72

5.0

1.154

0.154

80

100

14.20

Table 3
Diesel and natural gas (NG) life cycle assessment

Fuel (MJ/km)
GHG (kg CO2/km)

100

refinery and storage in refinery. From the terminal

storage to the refinery the crude is transported by sea
tanker. The refining of the crude takes place in the
Leixes refinery, the values for energy consumption and
emissions were taken from literature sources (Argonne
National Laboratory, 1999). From the same source the
overall efficiency of refining is 89%. The distribution of
the diesel fuel includes transport to terminals through
pipeline, storage and distribution by road tanker to the
filling stations. The distance by road is assumed to be of
75km, with the same distance in the return trip. The
electricity consumption of refilling the bus is of
0.0009 kWhel/MJ of fuel. The overall well-to-tank
analysis is presented in Table 1.
4.2.2. Natural gas path way
The natural gas used is extracted in Algeria and
transported through pipeline to the STCP depot.

Well-to-tank
(GaBi)

Tank-to-wheel
(EcoGest)

Well-towheel

Diesel

NG

Diesel

NG

Diesel

NG

3.9
0.35

2.0
0.29

24.6
1.85

31.4
1.75

28.5
2.2

33.4
2.0

According to General Motors (2002), extraction efficiency is 99.4%. Natural gas processing includes
removal of contaminants as CO2 and heavy hydrocarbons. In the process leaks occur, and this adds to the
total emissions of CO2 equivalent.
The natural gas is transported by pipeline, at a
pressure of ca. 70 bar, through Algeria, North Africa up
to Tangiers, where it crosses near Gibraltar to Spain.
Inside Portugal, the gas is distributed in the main
network at a pressure of 70 bar. The gas is considered to
reach the final customer at a pressure of 7 bar. The total
distance is assumed to be 1250 km. Some natural gas is
used as fuel in the intermediate compression stations,
the overall efficiency is assumed to be 97.0% according
to (Argonne National Laboratory, 1999). This value
evolves all processes after the processing station up to
end user (7 bar distribution network).
The natural gas vehicles of STCP store the natural
gas onboard at a pressure of 200 bar, so a compression
stage is needed. The final pressure is 250 bar in order to
guarantee enough pressure differential to fill the tanks.
The isentropic efficiency of the process is considered to
be 68% (General Motors, 2002). The final energy
consumption (electricity) for the filling process is
0.008 kWhel/MJ of CNG. The global result for natural
gas processing is presented on Table 2.
4.3. Life cycle assessment

Table 2
Energy consumption, losses and GHG emission for natural gas (NG)
production (Calhau et al., 2004)

Extraction of
natural gas
Processing of
NG
Transport
of NG
Compression/
Filling
Total

Energy
supplied

Energy losses

GHG emission

(MJ/
MJNG)

(MJ/
MJNG)

(g/MJNG)

1.065

0.006

9.2

0.62

6.8

1.059

0.007

10.8

0.61

6.7

1.052

0.023

35.4

2.53

27.7

1.029

0.029

44.6

5.37

58.8

1.065

0.065

100

9.13

100

Table 3 shows the values of fuel consumption and

CO2 equivalent emissions obtained for each part of the
fuels life cycle. The results are expressed as a function of
kilometres travelled by the buses. The integration of the
results of the two models (EcoGest and GaBi) suggests
that the diesel fuel use in the bus (from its production) is
more energy efficient (ca. 14%) although globally emits
more green house gases (ca. +1%).
5. Conclusions
A numerical model for simulation of fuel consumption and emissions of road vehicles equipped with
internal combustion engines, using conventional or

C.M. Silva et al. / Science of the Total Environment 367 (2006) 441447

alternative fuels, was presented. The integration of the

tank-to-wheel results of the EcoGest model with the
well-to-tank results from the software GaBi allowed to
perform a life cycle assessment concerning the utilization of diesel and natural gas fuels in a specific network
line of a bus transit company operating in Porto, Portugal. The integration of the results of the two models
(EcoGest and GaBi) suggests that although the diesel
fuel use in the bus (from its production) is more energy
efficient (ca. 14%), it globally emits more green house
gases (ca. + 1%).
This case study demonstrated the potential of
integrating an instantaneous fuel consumption and
emission results with a life cycle software for fuel life
cycle analysis assessments.
Acknowledgments
The first author (C. M. Silva) is supported by a PhD
scholarship of the National Foundation for Science and
Technology (FCT) and European Social Fund (FSE),
within the Third Framework Program. The second
author (G. A. Gonalves) is supported by research scholarship in the framework of the CUTE project, co-funded
by the European Commission. Special thanks to the bus
operator from Porto in Portugal, STCP (Sociedade de
Transportes Colectivos do Porto) and DGTT (Direco
Geral dos Transportes Terrestres) for the support.
References
Argonne National Laboratory. GREET 1.5transportation fuel-cycle
model; 1999.

447

Barth Matthew, An Feng, Younglove Theodore, Scora George, Levine

Carrie, Ross Marc, et al. Comprehensive modal emissions model
(CMEM), version 2.02 user's guide; 2001.
Bosch. Automotive handbook, 6a Edition. Bosch; 2004.
Brooker Aaron, Haraldsson Kristina, Hendricks Terry, Johnson
Valerie, Kelly Kenneth, Kramer Bill, et al. ADVISOR documentation. NREL; 2002.
Calhau KR, Gonalves GA, Farias TL. Environmental impact of
hydrogen in urban transports. Renewables 2004new and
renewable energy technologies for sustainable development.
vora, Portugal, 28 June1 July; 2004.
General Motors. Well-to-wheel analysis of energy use and greenhouse
gas emissions of advanced fuel/vehicle systemsa European
study. L-BSystemtechnik GmbH; 2002.
Heywood JB. Internal combustion engine fundamentals. McGrawHill; 1988.
IKP-Institute for Polymer Testing and Polymer Science of the
University of Stuttgard and PE Europe BmbH (2003). GaBi 4
software [http://www.gabi-software.com/].
Patton KJ, Nitschhe RG, Heywood JB. Development and evaluation of
a friction model for spark ignition engines. SAE paper 890836;
1989.
Silva CM, Farias TL, Mendes-Lopes JMC. Cold start, part warm start
and warm up simulation of vehicles in EcoGest. FISITA 2004
30th FISITA world automotive congress, Barcelona, 2327 May
(F2004V045); 2004.
Silva CM, Farias TL, Costa M, Santos Hlder. Evaluation of SI engine
exhaust gas emissions upstream and downstream of the catalytic
converter. J Energy Convers Manag (in press) [May].
Wu Wei, Ross Marc. Spark-ignition engine fuel consumption
modelling. SAE Paper 1999-01-0554; 1999.