Transportation Research Part D 51 (2017) 29–42

Contents lists available at ScienceDirect

Transportation Research Part D

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

A method to measure the eco-efficiency of diesel locomotive

Bernardo Bicalho Carvalhaes a, Rodrigo de Alvarenga Rosa a,⇑, Márcio de Almeida D’Agosto b,
Glaydston Mattos Ribeiro b
a
Federal University of Espírito Santo, Brazil
b
Federal University of Rio de Janeiro, Brazil

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

Article history: Brazilian railroads transport over 490 million tons a year using diesel-electric locomotives.
These locomotives emit several pollutants into the atmosphere and because of that, the
railroads seek to reduce emissions and achieve global emission standards. Thus, it is impor-
Keywords: tant to analyze the environmental impact of the use of diesel and alternative fuels to reach
Railroad’s eco-efficiency environmental standards. This paper makes use of a method based upon the World
Biodiesel Business Council for Sustainable Development (WBCSD) metrics to evaluate the locomo-
LNG
tives’ eco-efficiency. The method was applied to Estrada de Ferro Vitória a Minas
Locomotive
Greenhouse gas emissions
(EFVM). Different scenarios representing the exchange of fuel sources and technologies
were developed, tested and analyzed. The impacts were evaluated by seven eco-
efficiency performance indicators and compared with United States Environmental
Protection Agency (EPA) standards. The results offered cost savings and emissions reduc-
tion opportunities.
Ó 2016 Elsevier Ltd. All rights reserved.

1. Introduction

Brazilian railroads transported over 490 million tons in 2013 and in 2016 it is forecasted that they will transport over 550
million tons (ANTF, 2014). Transporting such high volumes using diesel-electric locomotives means railroads will emit thou-
sands of tons of pollutants into the atmosphere every year. Transporting the same volume by trucks through roads would be
catastrophic, taking into consideration the already poor state that federal roads find their selves in and the high traffic they
support, causing even more degradation. When it comes to emissions, it is known that moving freight by rail instead of truck
reduces greenhouse gas emissions by 75 percent (AAR, 2015).
This paper presents a method to quantify the environmental impact caused by the use of locomotives, comparing petro-
leum diesel and also greener fuel alternatives. The method evaluates atmospheric emissions, energy efficiency and the costs
concerning the different fuels. Petroleum diesel, biodiesel from soybean and liquefied natural gas (LNG) were compared in
this paper. The eco-efficiency of the actual operation using petroleum diesel, was compared with the operation using alter-
native fuels.
The method is an approach to evaluate the cost/benefit to improve the economic and environmental aspects of the busi-
ness. It is an evolution over the traditional energy-efficiency programs used by companies focused strictly on reducing fuel

⇑ Corresponding author at: Av. Fernando Ferrari, 514, Goiabeiras, Vitória, ES CEP 29075-910, Brazil.
E-mail addresses: bbcarvalhaes@gmail.com (B.B. Carvalhaes), rodrigo.a.rosa@ufes.br (R.A. Rosa), dagosto@pet.coppe.ufrj.br (M.A. D’Agosto),
glaydston@pet.coppe.ufrj.br (G.M. Ribeiro).

http://dx.doi.org/10.1016/j.trd.2016.11.031
1361-9209/Ó 2016 Elsevier Ltd. All rights reserved.
30 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

consumption and can be used by railroads in all countries, with all types of locomotives and fuels – including railroads with
electric locomotives.
The method was applied to Vitoria a Minas Railroad (EFVM). EFVM transports 50% of the Brazilian railroads freight vol-
ume, has 905 km of length and a fleet of 265 locomotives and over 18,000 wagons, and transports over 100 million tons of
iron ore and 20 million tons of general cargo (soy, corn, fertilizers, coal, steel etc.) every year.
This paper is structured as follows: Section 2 describes the concept of eco-efficiency. Section 3 presents a literature
review. Section 4 presents the method to measure locomotives’ eco-efficiency. Section 5 describes the characteristics of
the operation of EFVM. Section 6 describes how the data were obtained, including the characteristics of the fuels, its con-
sumption and emission rates. It also presents scenarios to be evaluated. Section 7 presents results and analysis of the results
achieved. Finally, Section 8 presents the conclusions.

2. Eco-efficiency concept

The World Business Council for Sustainable Development (WBCSD) defines eco-efficiency as the skill in measuring the
evolution of an economic activity in an environmentally sustainable manner to meet human needs and upgrade the quality
of life, reducing environmental impacts and the consumption rates of natural resources, limited by the environmental capac-
ities of the planet keeping the competitiveness of the companies. There are seven goals related to eco-efficiency, some of
them are: (1) reduce the material intensity; (2) reduce the energy intensity; (3) reduce the dispersion of toxic substances;
(4) enhance capacity of recycling material; (5) maximize the use of renewable resources; (6) extend the product life cycles;
and (7) increased service intensity (WBCSD, 2000).
WBCSD proposes a framework containing three levels of organization for eco-efficiency information: categories, aspects
and indicators. The categories are broad areas of environmental influence or business value. Each one has several aspects,
which are general types of information related to a specific category. The aspects describe what should be measured. The
indicators are the specific measurements of an individual aspect that can be used to demonstrate the eco-efficiency
(WBCSD, 2000; D’Agosto and Ribeiro, 2004).
The Eco-Efficiency Measurement (EM) indicators are performance measurements obtained through the ratio of product/
service value (V) by the environmental influence caused by the generation or use of the product/service (EI) (Eq. (1)) (WBCSD,
2000).

EM ¼ V=EI ð1Þ

3. Literature review

A chronological report is made in this section considering some of the main papers about alternative fuels, locomotives
and railroad emissions, and environmental impacts of this mode of transportation.
Plakhotnik et al. (2005) presented an analysis of the ecological situation at railway transportation of Ukraine with a speci-
fic focus on the Prydniprovs’ka regional railways. Large-scale pollutant was found and a comparison of the environmental
impact of the different railway subdivisions was presented. Also, a computer simulator was developed to allow comparative
data analysis. The paper only considered stationary emission sources, such as wagon and locomotive depot. It did not con-
sider emissions from actual operation of the rolling stock.
In 2006 Rail Safety and Standards Board (RSSB) and the Association of Train Operating Companies (ATOC) investigated the
use of biodiesel on Britain’s railways and published a report on August 2010. The effects on the engine’s performance and
exhaust emissions was tested using increasing biodiesel blending with diesel fuel, in steps from 5% of biodiesel to 100%
of biodiesel. From the research project results, the Diesel Metering Group (DMG) concluded that B20 (a 20% blend of biodie-
sel mixed with 80% diesel) was sensibly the highest blend that could be accepted without significant expenditure to retune
engines. The use of B20 did not appear to cause any significant engine wear, but the fuel consumption performance was
worse.
Lapuerta et al. (2008) collected and analyzed papers published in scientific journals about diesel engine emissions when
using biodiesel fuels as opposed to conventional diesel fuels. The first section is dedicated to the effect of biodiesel fuel on
engine power, fuel consumption and thermal efficiency, while the second section focus on the comparison of engine emis-
sions from biodiesel and diesel fuels, paying special attention to the most concerning emissions: nitric oxides and particulate
matter, the latter not only in mass and composition but also in size distributions. In this case the highest consensus was
found in the sharp reduction in particulate emissions.
Papagiannakis and Hountalas (2003), Papagiannakis et al. (2010a, 2010b) conducted an experimental investigation to
examine the effects of the emissions of a high speed, compression ignition engine where liquid diesel fuel is partially sub-
stituted by natural gas in various proportions, with the natural gas fumigated into the intake air. The experimental results
disclose the effect of these parameters on nitric oxide, carbon monoxide, unburned hydrocarbons and soot emissions, with
the beneficial effect of the presence of natural gas being revealed. They conclude that dual fuel combustion using natural gas
as a supplement for liquid diesel fuel is a promising technique for controlling both NOx and particulate matter (PM) emis-
sions on existing diesel ignition diesel engines, requiring only slight modifications of the engine structure. The observed dis-
 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42 31

advantages are an increase in HC and CO emissions that can be possibly mitigated by applying modifications on the engine
tuning, e.g. injection timing of liquid diesel fuel mainly at part loads.
Johnson et al. (2013) described and applied a technique for analyzing exhaust emission plumes from unmodified locomo-
tives under real world conditions from railway trains servicing an Australian shipping port. The method utilized simultane-
ous measurements downwind of the railway line of the following pollutants; particle number, PM, mass fraction, SO2, NOx
and CO2, from which emission factors were then derived. Samples from 56 train movements were collected, analyzed and
presented. The quantitative results for emission factors were noted and the findings were compared with previously pub-
lished papers. Statistically significant correlations within the group of locomotives sampled were found between the emis-
sion factors for particle number, SO2 and NOx.
Xue et al. (2011) analyzed reports about biodiesel engine performance and emissions, published by highly rated journals
in scientific indexes since year 2000. The effects of biodiesel on engine power, economy, durability and emissions including
regulated and non-regulated emissions were analyzed. It was found that the use of biodiesel leads to substantial reduction in
PM, HC and CO emissions accompanying with a small power loss, increase in fuel consumption and increase in NOx emis-
sions on conventional diesel engines.
Poompipatpong and Cheenkachorn (2011) modified a diesel engine for natural gas operation and evaluated the emission
and power output effects of such modifications. They also mentioned that two of the advantages of natural gas are clean
combustion and attractive price. They tested the emissions of CO, THC and NOx for different compression ratios and compare
the results.
Abdelaal and Hegab (2012) tested a single-cylinder direct injection (DI) diesel engine on regular operation and dual-fuel
mode, with natural gas as the main fuel and diesel fuel as a pilot. Comparative results of exhaust emission were presented for
several operating modes. They mentioned natural gas as a partial supplement for diesel fuel as a very promising solution for
reducing pollutant emissions, particularly nitrogen oxides (NOx) and PM. The results showed reduction in NOx and CO2
emissions, while CO emissions increased.
In 2012, Clean European Rail-Diesel (CleanER-D, 2012) delivered a report on the impact and performance of alternative
fuels in rail applications. The main objective was to study the different types of fuel used in railway applications and their
effect on engine parameters. It was found that biodiesel blends up to 20% are technically feasible although increasing fuel
consumption compared to diesel.
Park et al. (2012) examined the particulate matter characteristics of diesel locomotive engine exhaust at various engine
ratings. Diesel engine exhaust was collected via a dilution tunnel and the concentration and size distribution of fine particles
were measured by a scanning mobility particle sizer. The results showed that the maximum carbon monoxide emission was
reached at 59% of the maximum rating, after which emissions decreased.
Esters and Marinov (2014) compared the methods used for calculating emissions of UK rolling stock based on their type
and mode of operation. The three modes under comparison were; diesel, electric and bi-mode. As well as comparing these
three modes of operation, a comparison between Conventional, Freight and High Speed Rail was made. Alternate fuels were
considered for diesel and bi-mode locomotives and compared based on their environmental impact. It was possible to con-
clude that diesel trains produce less emissions than electric trains when factoring in mechanical and air resistances and that
alternative fuels, such as biodiesel, should be a consideration for the future of rail, as emissions fall dramatically with content
of biodiesel in fuel blends.
Gangwar and Sharma (2014) analyzed the railroad transport sector in India. A well to wheel approach was adopted to
quantify the emissions from diesel and electric locomotives which concluded that the accumulated carbon footprint of run-
ning electric locomotives was higher. They suggest that there should be a judicious mix of both tractions to achieve a balance
in environmental efficacy, sustainability and equity.
Based on the literature review, it can be verified that there is no paper considering a method to evaluate the eco-efficiency
of locomotives considering the use of alternative fuels, their cost, emissions and maintenance impacts. Therefore, such
method is presented in the next section of this paper.

4. Eco-efficiency method

Based upon the formula to calculate eco-efficiency proposed by WBSCD, Eq. (1), this paper employs a method that is suit-
able for calculating eco-efficiency of diesel locomotives. Thus, service value (V) is defined by seven eco-efficiency measure-
ments (EM) and to calculate them, the environmental influence (EI) must be first obtained. The seven Eco-Efficiency
Measurements (EM) proposed in this paper are: (1) Total energy consumption, (2) Total renewable energy consumption,
(3) Carbon dioxide emissions, (4) Carbon monoxide emissions, (5) Nitrogen oxides emissions, (6) Particulate matter emis-
sions and (7) Total cost (investment, maintenance and fuel).
The methods to calculate service value, environmental influence and eco-efficiency measurements are presented next.
A worldwide indicator used by railroads is the Gross Tonne Kilometer (GTKM) that represents the sum of the total weight
of the train multiplied by the distance traveled. GTKM includes the weight of locomotives, wagons and all freight trans-
ported. Since GTKM has been widely used by railroads it was used as the service value (V) for the method.
32 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

Service value GTKM (ton  km) is calculated by Eq. (2) where W is the set of wagons and T is the set of trips of each
wagon per year. In Eq. (2), GT wt is the gross tonnage (the dead weight plus the freight) of each wagon w 2 W in each trip
t 2 T (ton) and KM wt is the total kilometers traveled by each wagon w 2 W during trip t 2 T (km).
XX
GTKM ¼ GT wt KMwt ð2Þ
w2W t2T

For all environmental influence indicators (1) F is the set of fuels that can be used by the locomotives and (2) L is the set of
locomotives of the fleet.
For total energy consumption (eiE) (kJ), Eq. (3), FEfl represents the consumption of fuel f 2 F by locomotive l 2 L (l/h); WHfl
are the working hours of locomotive l 2 L using fuel f 2 F (h) and CC f is the calorie content of fuel f 2 F (MJ/l).
!
X X
eiE ¼ FEfl WHfl CC f ð3Þ
f 2F l2L

Total renewable energy consumption (eiRE) is calculated by Eq. (4), which differs from Eq. (3) because it considers only loco-
motives using renewable fuel, e.g., soybean biodiesel. It is defined that F r # F is the set of renewable fuels used by the loco-
motives, and Lr # L is the set of locomotives that use renewable fuels.
!
X X
eiRE ¼ FEfl WHfl CC f ð4Þ
f 2F r l2Lr

Total carbon dioxide emission (eiCO2) (kg) is calculated by Eq. (5), where FEfl represents the consumption of fuel f 2 F by
locomotive l 2 L (l/h); WHfl are the working hours of locomotive l 2 L using fuel f 2 F (h) and EC f is the CO2 fuel emission
factor of fuel f 2 F (kgCO2/l or kgCO2/m3).
!
X X
eiCO2 ¼ FEfl WHfl EC f ð5Þ
f 2F l2L

Carbon Monoxide (eiCO), Nitrogen Oxides (eiNOx) and Particulate Matter (eiPM) emissions are measured in grams per liter or
grams per cubic meter for NG. Eqs. (6)–(8), like Eq. (5), are used to calculate these three environmental influence indicators,
differing from one another by the emission factor for each pollutant, respectively, EOf , EN f and EP f .
!
X X
eiCO ¼ FEfl WHfl EOf ð6Þ
f 2F l2L

!
X X
eiNOx ¼ FEfl WHfl EN f ð7Þ
f 2F l2L

!
X X
eiPM ¼ FEfl WHfl EPf ð8Þ
f 2F l2L

The last indicator is cost efficiency (eiCE), which represents the total cost needed to operate the locomotive with a specific
fuel. It is the sum of fuel, maintenance and investment costs. How these three costs are calculated is presented next.
Fuel cost (FC), Eq. (9), represents the total fuel consumption cost (US$), which is the cost of the fuel used by the locomo-
tives; FEfl represents the consumption of fuel f 2 F by locomotive l 2 L (l/h); WHfl are the working hours of locomotive l 2 L
using fuel f 2 F (h) and FU f represents the cost of fuel f 2 F (US$/l or US$/m3).
!
X X
FC ¼ FEfl WHfl FU f ð9Þ
f 2F l2L

Extra maintenance is needed to operate with alternative fuels, i.e., it is necessary to repair the dual-fuel kits installed on the
locomotives that operate with LNG and some components need to be replaced more frequently when using biodiesel. The
extra maintenance cost (MC) is determined by Eq. (10), where MC represents the additional maintenance cost (US$); WHfl
are the working hours of locomotive l 2 L using fuel f 2 F (h) and MU f is the additional maintenance cost per working hour
when using fuel f 2 F (US$/WH).
!
X X
MC ¼ WHfl MU f ð10Þ
f 2F l2L
 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42 33

Some investments are necessary to operate with NG. So, the capital expenditure (CAPEX) to install the natural gas (NG) kits
in the locomotives must be considered. No CAPEX is necessary to operate with B25. The investment cost (IC) is calculated by
Eq. (11), where IC is the investment cost (US$); and IU f represents the total investment when using fuel f 2 F (US$/WH).
X
IC ¼ IU f ð11Þ
f 2F

After all costs determined by Eqs. (9)–(11), eiCE represents the cost efficiency (US$); where FC is the fuel cost; MC is the
maintenance cost and IC is investment cost. This leads to Eq. (12), by which the cost efficiency eiCE (another environmental
influence) can be determined.

eiCE ¼ FC þ MC þ IC ð12Þ

After environmental influence is calculated, it is possible to calculate the seven Eco-Efficiency Measurement (EM) indicators
related in Eqs. (13)–(19). They are: Total energy consumption (E), Total renewable energy consumption (RE), Carbon dioxide
emissions (CO2), Carbon monoxide emissions (CO), Nitrogen oxides emissions (NOx), Particulate matter emissions (PM) and
Cost Efficiency (investment, maintenance and fuel) (CE).

E ¼ GTKM=eiE ð13Þ

RE ¼ GTKM=eiRE ð14Þ

CO2 ¼ GTKM=eiCO2 ð15Þ

CO ¼ GTKM=eiCO ð16Þ

NOx ¼ GTKM=eiNOx ð17Þ

PM ¼ GTKM=eiPM ð18Þ

CE ¼ GTKM=eiCE ð19Þ

For Total energy consumption (E), Total renewable energy consumption (RE), Carbon dioxide emissions (CO2), Carbon
monoxide emissions (CO), Nitrogen oxides emissions (NOx), Particulate matter emissions (PM) and Cost Efficiency (CE),
higher results are better, since for the same service value (GTKM) the railroad either emits less pollutants or spends less
money to make the shift to a cleaner fuel. For Total renewable energy consumption (RE) a lower value signifies a better
result, since the railroad would use more biofuels for the same service value.
WBCSD proposed seven goals to measure eco-efficiency. As shown before, they are: (1) Reduce the material intensity; (2)
Minimize the energy intensity; (3) Reduce the dispersion of toxic substances; (4) Enhance the capacity of recycling; (5) Max-
imize the use of renewable energy; (6) Extend the product durability and (7) Increase the service intensity. The proposed
metrics are linked with these goals as follows:
Metric 1, Energy consumption, is linked with Goals 1 and 2, because it seeks a more efficient use of the fuel by running
longer with less fuel and reducing energy intensity. Metric 2, Total Renewable Energy Consumption, is linked with Goals 3
and 5, because using renewable energy tends to emit fewer pollutants into the atmosphere while capturing CO2 from the
during the production phase. Metrics 3, 4, 5, and 6, respectively, Total carbon dioxide emission, Carbon Monoxide, Nitrogen
Oxides and Particulate Matter, are linked with Goal 3, since they seek to reduce pollutant emissions. Metric 7 deals directly
with the principles of the WBCSD philosophy that intends to keep the competitiveness of the companies at the same level,
while developing environmental friendly attitudes.
The proposed metrics are limited by the technology of the locomotive’s engine, since it is not possible to use less fuel than
what the engine requires to produce its power. It is also limited by the fuels accepted by the diesel engine without significant
expenditure to retune it – great interventions could be too expensive for the railroads, especially in developing countries.
Using the SWOT methodology, the strengths, weakness, opportunities and threats of the method were analyzed (Fig. 1).
The strengths are: the method can be easily adapted by railroads with any type of locomotive and fuel, different eco-
efficiency indicators can be easily added or excluded, no significant costs are associated, except the man-hour to raise
and process data used by the method, it is easy to use and the results can be quickly obtained, besides being easily taught
and understood by employees. The main weakness is that railroad owners might not be willing to disclose sensitive infor-
mation, such as consumption, emissions and operational data, therefore the method would most likely be used inside the
company at first, searching for more efficient models of operation. Thus, comparison between railroads indicators would
come on a further stage. An opportunity is that the method could be used by governments as a standard evaluation tool
of locomotives’ eco-efficiency and finally a threat is that the method is not yet sponsored or adopted by regulating agencies.
If that happens and the method is widespread through railroads around the world, contributions from all of them would help
to improve it.
34 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

Fig. 1. SWOT analysis of the methodology.

5. Case study

EFVM is one of the most important railroads in Brazil. It transports over 120 million tons/year through 905 km of lines
with a fleet of 265 locomotives and over 18,000 wagons that run on average 30 ton high axle load iron ore trains daily, pow-
ered by up to three diesel-engine locomotives. A full cycle iron ore train trip (port-mine-port) can take up to three days,
depending on which mine is visited (there are 18 mines reached by the railroad). The land topography makes it so that
49% of the railroad consists of curves, with average radius of 233 m and minimum radius of 110 m. It leaves the port at
sea level and reaches up to 750 m of altitude at the mines, with maximum slope of 2.1%. Of the 905 km of length, 601 consist
of duplicated tracks. The maximum allowed speed for iron ore trains is 65 km/h, but they reach an average of up to 40 km/h
considering the full cycle. The railroad signaling works by electrical pulses sent through the rails, and there are energy supply
stations across the tracks at every seven kilometers. There is a control center at the port from which all trains are monitored
all the way to the mines.
Nowadays, several locomotives use petroleum diesel engines that consume standard petroleum diesel oil sold in Brazil.
The standard diesel oil sold in Brazil is a blend of petroleum diesel and soybean biodiesel. From a blend of 95% of petroleum
diesel and 5% of soybean biodiesel (B5), the government has increased biodiesel share on petroleum diesel up to 6% (B6) by
July 2014 and 7% (B7) by November 2014 (DOU, 2014).
Most of the Brazilian biodiesel is produced from soybean. In early 2014, soybean oil was the source for over 70% of all
biodiesel production (ANP, 2014c). The biodiesel used by the railroad for the test phase was acquired in its pure form
(100% biodiesel, also called B100) and then mixed with the preexisting B5 diesel to achieve a ratio of 25% biodiesel and
75% petroleum diesel (B25).
The railroad has tested B25 in some of its locomotives during regular operation for over two years. Engine and fuel system
components were disassembled and monitored for wear signs, undertaking frequent thorough inspections. After severe anal-
ysis of the data, it was possible to confirm that there is no need to make expressive changes in locomotive engines and fuel
systems to run with B25 biodiesel. Maintenance plans must run more frequently, and special attention must be directed to
components most prone to suffer from the use of the new fuel mix. Hoses and rubber sealants from the fuel system must be
replaced to avoid being compromised by the biodiesel, which can damage ordinary plastic and rubber components. Since the
biodiesel heat content is lower than the petroleum diesel, an increase of 2.5% in consumption was observed during the per-
formance of the locomotives. The B25 operation follows the same pattern as the petroleum diesel operation and no signif-
icant differences were reported by the engineers when conducting the trains.
The natural gas (NG) market in Brazil is regulated by the National Agency of Petroleum, Natural Gas and Biofuels (ANP), a
national agency responsible for the standards of petroleum-based fuels, natural gas and biofuels (ANP, 2014a). It sets the
upper and lower thresholds for each component of every fuel that is sold in Brazil. Liquefied natural gas (LNG) is a way
to commercialize NG in liquid state. NG is a fossil fuel that consists mainly of methane. ANP defines that LNG must be com-
posed at least of 85% methane (ANP, 2014b).
LNG has also been tested by the railroad. Its track tests on locomotives began in 2008 and are still ongoing. To be able to
run with LNG, the locomotives must receive special devices called dual-fuel kits. The dual-fuel kits enable them to run with a
 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42 35

mixture of LNG and diesel, maintaining the same power output. During the locomotive operation, there is always diesel in
the blend. Diesel is required to ignite the gas because the compression ratio of the engine is not high enough to ignite the gas
and there are no spark plugs in the engine, so when the diesel ignites, it then ignites the gas in sequence, obtaining full power
of the engine requiring less diesel. A tender car coupled to the locomotive to store the LNG is also necessary. The LNG is
mixed with B5 in the engine, reaching up to 80% LNG (in its vaporized state) and 20% petroleum diesel.
The LNG used by the railroad during the tests originates from Bolivia and is transported through pipelines across the
country down to the state of São Paulo, where it is then liquefied and transported by trucks to the railroad fueling station
at Vitoria, a coastal city located in Espírito Santo State where the company’s port is located.

6. Data acquisition

The method used for data acquisition of consumption and emissions are described next. The data was collected during
over eight years, after more than 100,000 km of running test and 1700 h of static tests on an appropriate test facility. The
emission data was collected in accordance to the requirements of US Code of Federal Regulations, 40 CFR Part 92. American
railroads followed the outcome of the tests, such as BNSF Railway and Union Pacific Railroad, as well as General Electric and
Electro-Motive Diesel, both locomotives manufacturers.
During regular operation, the locomotives consumed B5, thus all comparison in this paper consider this as the baseline
fuel.

6.1. Consumption

BB40 locomotives have a self-test mode, where the main generator itself exerts force contrary to the rotation of the diesel
engine simulated load, thus eliminating the need for a dynamometer. Thus, the company tests shed consists of a precision
scale for measurement of the locomotive fuel consumption. For the older locomotives that do not present this feature, the
test shed can measure their consumption by dissipating electrical energy through break grids. So, the tests were conducted
first for B5 during a consumption test of 90 min for each locomotive, passing through every engine loading stage for ten min-
utes each. Each loading stage is called notch, and there are nine of them, including idle and first to eighth notches. This test is
repeated four more times for each locomotive, adding to 7.5 h of test each.
Once the results were registered, the locomotives were fueled with another fuel (B25 or LNG) and once again submitted to
the same procedure. Table 1 presents part of the results achieved by these tests. The average value of the five measurements
is considered as the fuel consumption value of the locomotive model. Three locomotives were tested with B5 and B25, and
two locomotives were tested with B5 and LNG fuels at the EFVM facility.
Five measurements were undertaken at each notch, for 10 min each. They are shown on the second column of Table 1. The
initial and final weights for each measurement were recorded, leading to the consumption seen in the last column, in kilo-
grams. The result is the average of the five measurements (40.14 kg), with standard deviation of 0.5 kg.
Table 2 shows the fuel consumption per hour for each one of the five tested locomotives. The results represent the aver-
age consumption of the locomotives of the fleet. Locomotive BB40 consumes 272.91 L/h using B5, while locomotive BB36
consumes 262.87 L/h and DDM, 259.82 L/h. The difference in consumption between them is explained by the difference
of power output. BB40 has higher consumption rate but generates more power, being the newest locomotive model used
by the railroad. Running with LNG, BB40 consumes 127.23 L/h of B5 and 129.96 L/h of vaporized LNG simultaneously.
BB40, BB36 and DDM consume 280.35 L/h, 270.04 L/h and 267.01 L/h, respectively, running with B25.

6.2. Emissions

An emission analyzer, model TM-31 from manufacturer Tecnomotor was installed in the exhaustion chamber of the loco-
motives to capture emissions during B5, B25 and LNG operation. A partial view of the collected data can be seen in Table 3.
Table 4 shows the heat content, emission factors and costs for each fuel. B100 has a higher emission of NOx than LNG or
B5. On the other hand, it generates less CO2, CO and PM. LNG emits less CO2, less than 25% of NOx and virtually no PM, but

Table 1
Consumption test sheet (partial view – notch 4).

Notch Measurement Initial weight (kg) Final weight (kg) Consumption (kg)
4 1 (10 min) 1200 1160 40
2 (10 min) 733 693 40
3 (10 min) 290 249 41
4 (10 min) 1236 1196 40
5 (10 min) 733.45 693.75 39.7
Average 40.14
36 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

Table 2
Fuel consumption per hour for each type of locomotive.

Hourly fuel consumption (l/h)

Fuel/loco model Fuel
B5 LNGa B25
B5/BB40 272.91 – –
B5/BB36 262.87 – –
B5/DDM 259.92 – –
LNG/BB40 127.23 129.96 –
B25/BB40 – – 280.35
B25/BB36 – – 270.04
B25/DDM – – 267.01
a
Cubic meters, vaporized.

Table 3
B5 emissions sheet (BB40 – partial view).

B5 emissions (kg/h)
Notch CO (ppm) CO2 (%)
8 215 1.1
7 192 1.3
6 98 1.2
5 56 1.5
4 39 1.4
3 27 1.1
2 13 1.0
1 8 0.8
Idle 2.3 0.8

Table 4
Heat content, emission factors and costs of fuels.

B5 LNGa B100 B25

Heat content 37,711 kJ/l 40,039 kJ/m3 34,643 kJ/l 37,065 kJ/l
CO2 emission factor 2.70 kg/l 2.03 kg/m3 2.50 kg/l 2.65 kg/l
NOx emission factor 44.3 g/l 10.09 g/m3 48.73 g/l 45.23 g/l
PM emission factor 1.62 g/l 0.14 g/m3 0.86 g/l 1.46 g/l
CO emission factor 4.5 g/l 5.9 g/m3 2.34 g/l 4.05 g/l
Fuel cost 1.09 US$/l 0.69 US$/m3 1.36 US$/l 1.15 US$/l
a
Vaporized.

emits more CO than B5 or B100. Each cubic meter of LNG has more heat content than one liter of B5 or B100, which means
that it generates more energy during combustion.
The emission values of Table 4 were obtained through self-load tests performed by locomotives with the same character-
istics of the ones composing the railroad’s fleet. The heat content was measured from samples during regular operation, and
the costs were practiced by the company during the tests.
The investment costs considered the cost of the dual-fuel kits and labor to install them, practiced by the railroad at the
time of installation. The replacements of parts for the locomotives running with biodiesel were also considered.
Extra maintenance for the dual-fuel kits and biodiesel locomotives were also accounted for, based on the manufacturers’
guidelines and assisted by them. Biodiesel extra maintenance costs included earlier replacement of the following parts: tur-
bocharger, fuel injectors and pumps, fuel and oil filters, lubricant oil and fuel hoses. For the dual-fuel kits, the replacement of
parts included: valves, pipes, filters, hoses, electric components and gas diffuser. All costs considered the price of acquisition
practiced by the railroad and labor.

6.3. Standard emissions defined by United States environmental protection agency

United States Environmental Protection Agency (EPA) sets the standard emission for on-road and non-road fuel vehicles
and engines, which are used as reference in this study, since the Brazilian government does not provide its own standard.
EPA (EPA, 2012) sets emissions standards for line haul and switch locomotives – which can be seen in Table 5.
 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42 37

Table 5
EPA locomotive exhaust emission standard.

Duty cycle Tier NOx (g/bhp-hr) PM (g/bhp-hr) CO (g/bhp-hr)

Line haul 0 9.50 0.22 5.00
1 7.40 0.22 2.20
2 5.50 0.10 1.50
3 5.50 0.10 1.50
4 1.30 0.03 1.50

Source: EPA (2012).

EPA (EPA, 2012) defines five standards, called Tiers. Tier 0 is the less restrictive, while Tier 4 is the most restrictive, allow-
ing the least emissions. United States federal government defines that all locomotives must comply with Tier 4 standards
from January 2015. In Section 6, the emissions from EFVM were compared to EPA standards.
To compare EPA emissions with the railroad in case, they were brought to the same measuring unit. Table 6 presents
emissions per gross horsepower per hour, and to calculate the gross horsepower produced by all the locomotives of the fleet
during the period of one year it is necessary to multiply the emission factors of NOx, PM and CO by 24 h times 365 days, this
way obtaining the maximum allowed emissions for each Tier category. That is how the Tier horizontal lines presented in
Figs. 5–7, in Section 7 were calculated.

6.4. Scenarios

Based on the proposed eco-efficiency indicators, Section 4, five different scenarios with distinct blends of fuels were pro-
posed to measure environmental impacts and investments needed to implement each of them. These five scenarios are
described in Tables 6 and 7.
The results achieved by the method and their analysis are described in the next section.

7. Results and analysis

Fig. 2 shows the result of Total Energy Consumption (E). Scenario 2 is the most efficient due to the use of LNG, and pre-
sents a small difference (0.06%) from Scenario 3. Scenario 5 is the least efficient because biodiesel used increases fuel con-
sumption. Scenario 1 is better than Scenario 5 (difference of 0.97%) since less biodiesel is used. The standard deviation of the
results is 0.84%, influenced by the consumption measurements for each fuel. This means that technically Scenarios 2, 3 and 4
are tied, presenting no significant difference.
Fig. 3 shows the results achieved for renewable energy from each scenario. Renewable energy leads to a better environ-
mental performance and Scenario 5, which uses the largest amount of biodiesel, presents the best results. Scenario 4 comes
second, with half the performance of Scenario 5. Scenario 2 has the lowest renewable energy use due to the substitution of
B5 for LNG, which is not considered a renewable fuel. Scenario 3 is better than Scenario 2 due to the use of B25 on locomotive
models BB36 and DDM.
Fig. 4 shows the results concerning CO2 emissions. Scenario 4 presents the best result since the LNG used on the BB40
locomotives combined with biodiesel provide the lowest CO2 emissions. Sceneries 2 and 5 show very similar results and Sce-
nario 1 presents the highest emissions. Since biodiesel is considered 100% renewable, its CO2 emissions can be reported sep-
arately from the total emissions (GHG Protocol, 2004). The emissions considered refer only to the combustion of fuel.
Emissions generated during the production and transportation of the fuel were not considered.
CO emissions are presented in Fig. 5. Scenario 5 presents the best result since biodiesel CO emissions are lower than B5 or
LNG emissions. Scenarios 2, 3 and 4 have higher CO emissions than Scenario 1 since natural gas emits more CO than B5 or
B25. Scenarios 2 and 3 are almost even, with a difference of 499 GTKM per gram of CO, which has no statistical significance,
since the standard deviation consists of 1871.14 GTKM/g. The lower horizontal line represents CO emissions for TIER 0
(16.437 GTKM/g), the middle line for TIER 1 (37.356 GTKM/g) and the top line for TIERs 2, 3 and 4 (54.788 GTKM/g). All sce-
narios achieve TIER 4 requirements for CO emissions.

Table 6
Proposed scenarios to evaluate locomotives’ performance.

Scenario Description
Scenario 1 All locomotives use B5 (actual operation)
Scenario 2 All GE BB40 locomotives use LNG and B5, the others use B5
Scenario 3 All GE BB40 locomotives use LNG and B5, the others use B25
Scenario 4 All GE BB40 locomotives use LNG and B25, the others use B25
Scenario 5 All locomotives use B25
38 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

Table 7
Resume of fuels and locomotives for each scenario.

B5 LNG + B5 LNG + B25 B25 Total

BB40 BB36 DDM BB40 BB40 BB40 BB36 DDM
Scenario 1 216 14 35 – – – – – 265
Scenario 2 – 14 35 216 – – – – 265
Scenario 3 – – – 216 – – 14 35 265
Scenario 4 – – – – 216 – 14 35 265
Scenario 5 – – – – – 216 14 35 265

Fig. 2. Total energy consumption (E).

Fig. 3. Renewable energy efficiency (RE).

Fig. 4. CO2 emissions efficiency (CO2).

 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42 39

Fig. 5. CO emission efficiency (CO).

Fig. 6. PM emission efficiency (PM).

The results of PM emission are shown in Fig. 6. Scenario 4 presents the best result since LNG and biodiesel PM emissions
are lower than petroleum diesel. Scenario 1 shows the highest emissions of PM, followed by Scenario 5.
The lower horizontal line represents the PM emission standard for TIER 0 and 1 (373,557 GTKM/g of CO) and the top line
for TIERs 2 and 3 (821,826 GTKM/g). TIER 4 can only be achieved with (2,739,421 GTKM/g) – none of the scenarios can
achieve it. Scenarios 2, 3 and 4 achieved TIER 1 requirements, and Scenarios 1 and 5 cannot achieve even TIER 0 standard
for PM emissions.
Fig. 7 presents the results for NOx. Scenario 2 shows the best result since LNG emissions are much lower than petroleum
diesel or biodiesel. Scenarios 3 and 4 have slightly higher emissions than Scenario 2 because biodiesel has higher NOx emis-
sions than petroleum diesel, but Scenario 3 presents no significant difference when compared to Scenario 2 since the differ-
ence of 102 GTKM/g lies within the standard deviation of 198.80 GTKM/g. Scenario 5 presents the highest NOx emission, even
higher than Scenario 1, due to the increase of fuel consumption by the locomotives when using biodiesel and its higher emis-
sion factor.

Fig. 7. NOx emission efficiency (NOx).

40 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

The lower horizontal line represents NOx emission standards for TIER 0 (8651 GTKM/g), the middle line for TIER 1
(11,106 GTKM/g). TIERs 2 and 3 (14,942 GTKM/g) are represented by the top line. TIER 4 can only be achieved with
(63,217 GTKM/g) – none of the scenarios achieved it. Scenarios 2, 3 and 4 can achieve TIER 3 requirements, and Scenarios
1 and 5 comply with TIER 1 standard for NOx emissions.
The results of Cost Efficiency (CE) are shown in Figs. 8 and 9. Fig. 8 considers only the fuel cost. Scenario 2 is the best, since
LNG is cheaper than biodiesel or petroleum diesel in Brazil, but within the standard deviation if compared to Scenario 3. Sce-
nario 5 presents the worst result since biodiesel is more expensive than petroleum diesel and has lower calorific value,
increasing the locomotive’s consumption. Scenario 1 is better than Scenario 5, because petroleum diesel is cheaper than
biodiesel.
Fig. 9 considers the costs of installation and maintenance of the LNG dual-fuel kits, besides extra maintenance required
for locomotives using B25. The rank is the same as before, with Scenario 2 presenting the best result, although with a lower
value. This means that even considering the costs of maintenance and investment to replace diesel for LNG and biodiesel,
they still reach better results than the current operational model.
Scenarios 1 and 3 do not show the best performance in any of the indicators. Scenario 3 presents lower costs of fuels and
maintenance than Scenario 1. Scenario 2 shows the best result for E, NOx and CE. Scenario 4 is the best for CO2 and PM. Sce-
nario 5 is the best for RE and CO. Since Scenario 1, the actual scenario of EFVM, does not show lower costs than Scenarios 2, 3
and 4, the company should implement the scenario that best fits its needs, whether it is cost efficiency or emission reduction.
Table 8 summarizes the results (the best results are in bold numbers). Scenarios 2, 3 and 4 are cheaper than Scenario 1 due to
the high share of LNG. Even with Biodiesel being more expensive, the high amount of LNG used by these scenarios make
them worth it.
A sensitivity analysis was carried out considering price fluctuations from 30% to +30% for B5, B25 and LNG. The upper
and lower results for each scenario are shown in Table 9.
Scenario 1 reaches 74.71 GTKM/US$ if B5 price rises by 30% and 138.75 GTKM/US$ if it falls by 30%. If the LNG and B5
prices would fall by 30%, Scenario 2 would reach 161.93 GTKM/US$. On the other hand, if both fuel prices rise by the same
amount, the result would fall to 90.87 GTKM/US$. For Scenarios 3 and 4, after fluctuations of all fuel prices, the combination
that generated the lowest and highest results were 30% and +30% fluctuation of B5 and LNG, resulting respectively in 90.52
and 88.73 on the low side, and 158.98 and 142.34 GTKM/US$ on the high side. Scenario 5 resulted in 68.48 GTKM/US$ if B25
price rose by 30% and 126.11 if it fell by the same amount.
To bring the results to a comparable base, the following criteria were applied to the results:

– E, RE, CO2, and CE: all scenarios that showed better results than Scenario 1 were assigned 1 point;
– CO: all scenarios that complied with TIER 4 standards were assigned 1 point, since TIER 4 is the highest regulation stan-
dard in United States; and
– PM and NOx: all scenarios that complied with TIER 3 standards were assigned 1 point since TIER 4 standard is too high for
any scenario to achieve it.

Table 10 shows the normalized results. Column ‘‘Scenario” shows the number of the scenario, columns 2–8 represent the
normalized results obtained by each of the eco-efficiency indicators analyzed and finally, the column ‘‘Total” represents the
sum of each normalized value for the eco-efficiency.
Scenario 4 ends up being the best to be implemented by the railroad. Scenario 4 mixes the use of LNG and Biodiesel, pre-
senting lower costs than the actual operational model and at the same time, lower emissions – except CO emissions, even
though it achieves TIER 4 standard emissions.

Fig. 8. Cost efficiency (CE), considering only fuel cost.

 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42 41

Fig. 9. Cost efficiency (CE), considering fuel cost, investment and extra maintenance required for LNG and B25.

Table 8
Summary of results.

Scenarios E RE CO2 CO PM NOx CE

Scenario 1 14.20 352 208.95 119,086 330,913 12,085 97.12
Scenario 2 14.60 176 244.97 111,128 614,718 20,152 116.41
Scenario 3 14.59 260 248.42 111,627 620,083 20,050 115.36
Scenario 4 14.53 843 275.53 115,248 660,392 19,362 109.32
Scenario 5 14.06 1685 244.76 128,956 357,504 11,521 88.76

Table 9
Sensitivity analysis results.

Scenarios Min. Actual Max.

Scenario 1 74.71 97.12 138.75
Scenario 2 90.87 116.41 161.93
Scenario 3 90.52 115.36 158.98
Scenario 4 88.73 109.32 142.34
Scenario 5 68.48 88.76 126.11

Table 10
Normalized results.

Scenario E RE CO2 CO PM NOx CE Total

Scenario 1 0 0 0 1 0 0 0 1
Scenario 2 1 0 1 1 1 1 1 6
Scenario 3 1 0 1 1 1 1 1 6
Scenario 4 1 1 1 1 1 1 1 7
Scenario 5 0 1 1 1 0 0 0 3

If Scenario 4 was implemented, compared to Scenario 1, the locomotives could gain 2.3% in energy efficiency, 139.0% in
renewable energy efficiency, 32.0% in CO2 emissions, 100.0% in PM, 60.0% in NOx and 12.6% in cost efficiency. The only indi-
cator that shows worse result when comparing Scenario 4 to Scenario 1 is CO with 3.2%, due to the use of LNG.
The results showed that the method can be used as a support tool to help the railroad to decide what is the best mix of
fuel and technology to minimize the environmental impacts and the operations cost.

8. Conclusions

A method to measure locomotives’ eco-efficiency, evaluating emissions and costs, was applied to Estrada de Ferro Vitória
a Minas (EFVM), one of the most important Brazilian railroads. Different scenarios representing the exchange of fuel sources
and technologies were developed, tested and analyzed. The emissions were evaluated by seven eco-efficiencies measure-
ments: (1) Total energy consumption, (2) Total renewable energy consumption, (3) Carbon dioxide emissions, (4) Carbon
monoxide emissions, (5) Nitrogen oxides emissions, (6) Particulate matter emissions. The costs measurements considered
investments and operational costs.
42 B.B. Carvalhaes et al. / Transportation Research Part D 51 (2017) 29–42

The results achieved by the method applied to EFVM were compared with United States Environmental Protection Agency
(EPA) standards. The best scenario to be implemented by the studied railroad would be Scenario 4, which mixes the use of
LNG and Biodiesel, presenting lower costs than the actual operational model and at the same time, lower emissions – except
CO emissions, but it achieves TIER 4 standards nonetheless. If Scenario 4 is implemented, compared to Scenario 1 there are
gains of 2.3% in energy efficiency, 139.0% in renewable energy efficiency, 32.0% in CO2 emissions, 100.0% in PM, 60.0% in NOx
and 12.6% in cost efficiency. The only indicator that shows worse result when comparing Scenario 4 to Scenario 1 is CO with
3.22%.
The method showed to be robust, offering cost savings and emissions reduction opportunities. Railroads can use this
method to bring their operations to a better environmental level. The results achieved present opportunities to improve
the performance of all eco-efficiency indicators, depending on the main interest of the company. It can also reduce opera-
tional costs and emissions, increase energy efficiency and renewable energy use.

9. Future work

This study does not take into consideration emissions generated during the production of the fuels, only those generated
as consequence of their consumption. As next steps for the study, it would be valid to consider the emissions of the whole
chain of production, including the Life Cycle Assessment (LCA) specific for transportation, which is called Well-to-Wheel
(WTW), and their effect on the results achieved so far.

Acknowledgements

The authors thank FAPES (processes 458/2013 and 75528452/2016) and CNPq (process 313408/2014-9) for the financial
support.

References

AAR, Association of American Railroads, 2015. Freight Railroads Help Reduce Greenhouse Gas Emissions Available at: <https://www.aar.org/
BackgroundPapers/Railroads%20and%20Greenhouse%20Gas%20Emissions.pdf> (Accessed: 15/03/2016 at 15:21hs).
Abdelaal, M.M., Hegab, A.H., 2012. Combustion and emission characteristics of a natural gas-fueled diesel engine with EGR. Energy Convers. Manage. 64,
301–312.
ANP, Agência Nacional do Petróleo, Gás Natural e Biocombustíveis, 2014a. Institutional Folder. Available at: <http://www.anp.gov.br/?dw=2447> (Accessed:
18/07/2014 at 16:52hs).
ANP, Agência Nacional do Petróleo, Gás Natural e Biocombustíveis, 2014b. Resolução.
ANP, Agência Nacional do Petróleo, Gás Natural e Biocombustíveis, 2014c. Boletim Mensal do Biodiesel Available at: http://www.anp.gov.br/?dw=70554.
April 2014 (Accessed: 01/08/2014 at 09:48hs).
ANTF, Associação Nacional dos Transportes Ferroviários, 2014. Balanço do Transporte Ferroviário 2013 Available at: <http://www.antf.org.br/index.php/
noticias/5139-balanco-do-transporte-ferroviario-2013> (Accessed: 16/07/2014 at 17:20hs).
CleanER-D. Clean European rail-diesel, impact and performance of alternative fuel in rail applications, June 2012. Available at: <http://secure.cnc.it/cleaner-
d/Docs/CLD-D-UNE-011-02.pdf> (Accessed: 16/03/2016 at 15:22hs).
D’Agosto, M., Ribeiro, S.K., 2004. Eco-efficiency management program (EEMP) – a model for road fleet operation. Transport. Res. Part D: Transp. Environ. 9
(6), 497–511.
Diário Oficial da União – DOU. (29/05/2014) Available at: <http://pesquisa.in.gov.br/imprensa/jsp/visualiza/index.jsp?jornal=1&pagina=1&data=29/05/
2014> (Accessed: 18/07/14 at 14:40hs).
EPA, United States Environmental Protection Agency, 2012 Available at: <http://www.epa.gov/otaq/standards/nonroad/locomotives.htm> (Accessed: 29/
04/2015 at 18:25hs).
Esters, T., Marinov, M., 2014. An analysis of the methods used to calculate the emissions of rolling stock in the UK. Transport. Res. Part D: Transp. Environ.
33, 1–16.
Gangwar, M., Sharma, S.M., 2014. Evaluating choice of traction option for a sustainable Indian Railways. Transport. Res. Part D: Transp. Environ. 33, 135–
145.
Johnson, G.R., Jayaratne, E.R., Lau, J., Thomas, V., Juwono, A.M., Kitchen, B., Morawska, L., 2013. Remote measurement of diesel locomotive emission factors
and particle size distributions. Atmos. Environ. 81, 148–157.
Lapuerta, M., Armas, O., Rodriguez-Fernandez, J., 2008. Effect of biodiesel fuels on diesel engine emissions. Prog. Energy Combust. Sci. 34 (2), 198–223.
Papagiannakis, R.G., Hountalas, D.T., 2003. Experimental investigation concerning the effect of natural gas percentage on performance and emissions of a DI
dual fuel diesel engine. Appl. Therm. Eng. 23 (3), 353–365.
Papagiannakis, R.G., Kotsiopoulos, P.N., Zannis, T.C., Yfantis, E.A., Hountalas, D.T., Rakopoulos, C.D., 2010a. Theoretical study of the effects of engine
parameters on performance and emissions of a pilot ignited natural gas diesel engine. Energy 35 (2), 1129–1138.
Papagiannakis, R.G., Rakopoulos, C.D., Hountalas, D.T., Rakopoulos, D.C., 2010b. Emission characteristics of high speed, dual fuel, compression ignition
engine operating in a wide range of natural gas/diesel fuel proportions. Fuel 89 (7), 1397–1406.
Park, D., Yoon, Y., Kwon, S.B., Jeong, W., Cho, Y., Lee, K., 2012. The effects of operating conditions on particulate matter exhaust from diesel locomotive
engines. Sci. Total Environ. 419, 76–80.
Plakhotnik, V.N., Onyshchenko, J.V., Yaryshkina, L.A., 2005. The environmental impacts of railway transportation in the Ukraine. Transport. Res. Part D:
Transp. Environ. 10 (3), 263–268.
Poompipatpong, C., Cheenkachorn, K., 2011. A modified diesel engine for natural gas operation: performance and emission tests. Energy 36 (12), 6862–
6866.
World Business Council for Sustainable Development, 2000. Measuring Eco-Efficiency. A Guide to Reporting Company Performance. World Business Council
for Sustainable Development, Geneva.
Xue, J., Grift, T.E., Hansen, A.C., 2011. Effect of biodiesel on engine performances and emissions. Renew. Sustain. Energy Rev. 15 (2), 1098–1116.