Fuel Processing Technology 159 (2017) 200–210

Contents lists available at ScienceDirect

Fuel Processing Technology

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

Research article

Analysis of the effects of diesel/methane dual fuel combustion on

nitrogen oxides and particle formation through optical investigation in a
real engine
Silvana Di Iorio, Agnese Magno ⁎, Ezio Mancaruso, Bianca Maria Vaglieco
Istituto Motori-CNR, Napoli, Italy

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

Article history: Great attention was paid to nitrogen oxides and particulate matter emitted by the compression ignition engines.
Received 5 September 2016 The adoption of methane dual fuelled with diesel could contribute to the reduction of these pollutants.
Received in revised form 15 December 2016 This paper aims to investigate the combustion phenomena occurring when a premixed methane/air charge is ig-
Accepted 8 January 2017
nited by the direct injection of diesel fuel. The research activity was performed on a production compression ig-
Available online xxxx
nition engine, three-cylinder, 1.0 L, equipped with a common rail injection system. In order to operate in diesel/
Keywords:
methane dual fuel mode, the intake manifold of the engine was modified to set an electronic port fuel injector
Dual-fuel engine suitable for gaseous fuels. Different engine speeds and loads were tested. For each engine condition, a small
Methane part of the total energy was provided by the direct injected diesel fuel while the remaining by the methane. Ther-
Two-colour pyrometry modynamics analysis of the combustion process was performed through conventional measurement involving
NOx emissions in-cylinder pressure acquisition. Endoscope based optical techniques were carried out for the combustion visu-
Particle emissions alization with high spatial distribution and temporal evolution. Two-colour pyrometry method was applied to
the flame images to evaluate the temperature and the soot concentration. This method allows to achieve a better
insight about the pollutant formation. Experimental results revealed that DF combustion occurs with lower tem-
perature and soot formation than diesel operation thus leading to lower nitrogen oxides and particle emissions at
exhaust.
© 2017 Elsevier B.V. All rights reserved.

1. Introduction prevents ignition at the temperature reached under compression in a

In the past decade, advances in exploration and production technol-
ogies have considerably increased recoverable supplies of methane
making it an attractive alternative fuel for internal combustion engines
in a wide range of applications, from stationary power production to
road and marine transportation [1]. Methane is the simplest hydrocar-
bon; it has no carbon-carbon bonds and it has high hydrogen to carbon
ratio resulting in clean burning qualities [2]. Therefore, the use of meth-
ane as supplement to diesel fuel in compression ignition (CI) engines
has achieved particular attention thanks to its potential to reduce pol-
lutant emissions from vehicles operating in the congested urban area
of some countries [3]. It is expected that by 2020, the methane share
in the bulk of motor fuels will reach 11% in Asia, 12% in Americas and
14% in Europe and the world gas consumption in transport is expected
to increase from 20 bcm in 2010 up to 40–45 bcm in 2030 [4].
The most suitable way to utilize methane in CI engines without sig-
nificant technological modifications is the dual fuel (DF) concept. The
poor ignition characteristics of the low cetane number (CN) gas fuel
⁎ Corresponding author at: Istituto Motori-CNR, Via G. Marconi 4, 80125 Napoli, Italy.
E-mail address: a.magno@im.cnr.it (A. Magno).
CI engine. In order to ensure ignition and sustain combustion of the
premixed methane/air charge, an appropriate amount of high CN liquid
fuel is injected in the cylinder [5].
DF combustion allows to reduce the main CI engine pollutants that
are nitrogen oxides (NOx) and particulate matter (PM) [6]. Moreover,
carbon dioxide (CO2) emissions are lower in DF mode with respect to
conventional diesel operation [7]. Emissions of carbon monoxide (CO)
and total hydrocarbon (THC) are increased when the engine runs in
DF mode due to the lower thermal efficiency [8]. Concerning THC emis-
sions from DF combustion, they consist mainly of methane hydrocar-
bons (MHC) that have negligible reactivity in the photochemical smog
cycle but, on the other hand, their global warming effect is 30 times
the one of CO2 [3]. In order to overcome the drawback of high CO and
THC emissions from DF combustion, some alternative strategies were
proposed such as air inlet preheating combined with the use of exhaust
gas recirculation (EGR) [9], intake air throttling [10], advanced diesel
pilot injection [11] and the use of biodiesel as ignition source instead
of diesel fuel [12]. Overall, several studies have investigated the effect
of methane addition to diesel fuel on pollutant emissions [13–17]. Nev-
ertheless, few studies deal with the particle size characterization from
DF combustion systems [18–20]. Hernandez et al. [18] analysed the

http://dx.doi.org/10.1016/j.fuproc.2017.01.009
0378-3820/© 2017 Elsevier B.V. All rights reserved.
 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210
separate effect of hydrogen, methane and carbon monoxide addition on
the performance, combustion efficiency and pollutant emissions of a
supercharged 2 L diesel engine. Concerning the methane/diesel opera-
tion, they observed that the particle number concentration decreases
as the diesel fuel replacement increases while the particle mean diame-
ter does not change. Yang et al. [19] studied the particle emissions of
diesel/natural gas dual-fuel operation on a turbocharged four-cylinder
common rail (CR) engine with varying pilot injection timing. They
found out that from the view of reduction particle emissions, the
range of optimal pilot injection timing is from 17 crank angle degrees
(cad) after top dead center (ATDC) to 20 cad ATDC and thus it is not rea-
sonable to advance excessively the pilot injection timing. From experi-
ence of authors [20] that analysed the particles emitted by a small CI
engine modified to run in diesel/methane mode, at different operating
conditions, DF combustion allows to reduce the total particle number
with respect to diesel operation. On the other hand, the particle diame-
ter depends on the combustion evolution due to the interaction be-
tween diesel fuel and methane.
The topic of particle number and size is particularly relevant and be-
cause of the negative impact of ultrafine particles on environment and
human health, it requires intensive researches [21]. Great efforts were
paid to reduce the particles emissions from internal combustion engines
also because of the new emission legislations that include restrictions
on the particle number beyond the particle mass [22].
On the other hand, experimental investigations involving optical
visualization of in-cylinder phenomena could contribute to achieve a
better understanding of the effect of DF strategies on combustion
characteristics and pollutant formation mechanisms. Optically ac-
cessible engines, realized by replacing a part of metallic cylinder by
quartz or sapphire windows, allow a good visualization of injection
and combustion events [23,24]. Taniguchi et al. [25] analysed DF
combustion at different operating conditions by carrying out optical
measurements on a research engine. They observed that at low loads,
the gas fuel far from the diesel spray did not reach the flammability
limits. As consequence, methane was emitted at exhaust as un-
burned MHC. Dronniou et al. [26] performed simultaneous measure-
ments of natural luminosity, OH chemiluminescence and 2D PLIF
imaging on a light duty single cylinder research engine running in
Fig. 1. Test engine.
201
DF mode. They provided some fundamental combustion mecha-
nisms that occur when a premixed mixture is ignited by a pilot injec-
tion of reactive fuel. They observed that, independently from the
equivalence ratio, the ignition occurred in the region near the bowl
wall. However, they concluded that additional works are necessary
to evaluate the advantages of DF strategies. Optical investigation of
DF combustion was performed also by the authors [27] on a trans-
parent engine equipped with the head of a CR, 2 L production engine.
They found out that the combustion starts under the nozzle tip and
then the flames move towards the bowl wall.
It is worth underlining that transparent engines require geometrical
modifications of combustion chamber thus modifying the flow charac-
teristics. Moreover, limited strength of transparent window entails
speed and load limitations [28]. On the contrary, endoscope based opti-
cal techniques can be applied to production engines with the aim to ob-
serve combustion evolution in real engine geometries and in wide range
of operating conditions [29–31].
The objective of this paper is to investigate the cause-effect relation-
ship between combustion evolution and pollutants formation of diesel/
methane DF operation. The research was carried out on a production CI
engine modified to run in DF mode: a gas injector was installed in the
intake manifold and it was managed by a delay unit developed in
Istituto Motori. The original electronic control unit (ECU) was retained
for diesel injection control. Experiments were performed at different
engine speeds and loads. The engine was first operated in conventional
mode with neat diesel fuel and then in diesel/methane DF configuration.
Under DF mode, at a fixed engine speed, sufficient amount of diesel fuel
was injected to cover the 10% of the maximum torque output; the re-
maining percentage of the desired torque output was achieved by
methane. Combustion behaviour of DF combustion was analysed
through indicating data. Endoscope based optical techniques were im-
plemented to visualize the combustion event in the chamber. Flame im-
ages were post-processed by two-colour pyrometry method to provide
qualitative information about the flame temperature and the soot con-
centration that were correlated to NOx and PM emissions measured at
exhaust, respectively. Moreover, particular attention was paid to the
characterization of the particles in term of number and size at the
exhaust.
202 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210

Table 1 The injector was managed by a control unit developed in Istituto Motori
Engine specifications. that allowed the setting of the number of injections, the start (SOI) and
Engine Compression ignition the duration of injection (DOI). For diesel fuel, a commercial injector
Number of cylinders 3, in-line
was used; the injection strategies, i.e. SOI, DOI and injection pressure,
Bore [mm] 75.0 varied according to the operating condition.
Stroke [mm] 77.6
Displacement [cm3] 1028 2.1.2. Test facilities
Compression Ratio 17.5:1
The engine was coupled to an electrical dynamometer characterized
Max. power [kW] 15 @ 3600 rpm
Max. torque [Nm] 60 @ 2000 rpm by maximum power of 93.2 kW and maximum speed of 5300 rpm. The
Injection system Direct, common rail engine was instrumented with K type thermocouples to measure the
Max. injection pressure [bar] 1400 temperature of the coolant and of both the air inlet and the exhaust
Aspiration Naturally aspirated gas. A gravimetric balance was used to measure the diesel fuel con-
Intake valves opening [cad BTDC] 13
Intake valves closing [cad ATDC] 39
sumption while the methane consumption was provided by a thermal
Exhaust valves opening [cad BTDC] 38 mass flow meter. Quartz transducers were fitted in the glow plug
Exhaust valves closing [cad ATDC] 14 seats of the three cylinders in order to monitor the combustion evolu-
tion. The pressure signals were recorded for 500 cycles at every 0.2
cad by a high-speed data acquisition system.
2. Experimental setup and procedures NOx emissions were measured at exhaust by a Horiba MEXA 7100
DEGR analyser at a frequency of 10 Hz and with a resolution of 1 ppm.
2.1. Experimental setup The analyser employs a chemiluminescence device that measures nitro-
gen monoxide (NO) by detecting the light emitted when it reacts with
2.1.1. Test engine ozone. Other oxides of nitrogen such as nitrogen dioxide (NO2) are mea-
The investigation was carried out on a three-cylinder, six valves, sured by passing them through a converter located before the reaction
1.0 L, CI engine for quadricycles (Fig. 1). Details about engine specifica- chamber.
tions are listed in Table 1. Particle number and electrical mobility diameter were measured by
The engine was equipped with an electronically controlled CR injec- the Engine Exhaust Particle Sizer 3090 (EEPS) developed by TSI. The in-
tion system characterized by a unit pump operated by a three-lobe cam strument allowed to measure particle diameter from 5.6 to 560 nm with
that allows to reach a maximum injection pressure of 1400 bar. The en- a time resolution of 10 size distributions/s. For a proper particle charac-
gine was naturally aspirated; the intake air was drawn from a tank in terization the exhaust gas sample was diluted by the Dekati Engine Ex-
order to reduce the pressure oscillations. In the original configuration, haust Diluter (DEED), a Particle Measurement Program (PMP)-
the engine was provided with two emission reduction systems: cooled compliant conditioning system, with a dilution ratio of 100.
EGR for NOx reduction and a diesel oxidation catalyst (DOC) for CO More details about the test facilities are reported in [20].
and THC abatement.
Methane was charged in a compressed bottle at 200 bar and decom- 2.1.3. Optical setup
pressed to 4 bar through a regulator and, then, supplied to the intake Endoscope based optical investigation was performed to visualize
manifold. In order to carry out diesel/methane DF experiments, the in- the combustion process under real engine conditions. The hole for en-
take manifold was modified to set up a 4 holes injector suitable for gas- doscope was made in the first cylinder. The endoscope was inserted in
eous fuels characterized by dynamic flow rate of 2000 cm3/min at 4 bar. the cylinder through an adaptor equipped with a sapphire window. In

Head
Injector tip

Piston dome

Fig. 2. Endoscope field of view.

COLOR IMAGE GREYSCALE IMAGE BINARY IMAGE 100

90
a) b) c) d) 80
70
flame area [%]

60
50
40
30
20
10
0
0 10 20 30 40
crank angle [cad]

Fig. 3. Image processing sequence for flame image: (a) raw image, (b) grey scale image, (c) binary image, (d) temporal flame area evolution.
 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210 203

a) b) 2700 c) 0.20

Sootconcentration[mg/m3]
0.15
2400

temperature[K]
0.10
2100

0.05
1800

0
1500

Fig. 4. Image processing sequence for flame image: (a) raw, (b) temperature, (c) soot.

order to prevent lenses damaging due to the high temperatures, the cylinder soot concentration, the AVL-ThermoVision software provided
housing of the endoscope was cooled with compressed air. the KL factor, expressed in arbitrary unit, where K is the absorption co-
Images were acquired by a charge coupled device CCD camera char- efficient proportional to the soot concentration and L is the geometric
acterized by a resolution of 640 × 480 pixels. The camera was coupled to thickness of the flame along the optical axis. By making proper assump-
the endoscope through a c-mount adaptor and a 25 mm lens. Before tions, the KL values were converted in soot concentration expressed in
starting the tests, the optical fibre of an endoscope was connected to a mg/m3 [33,34].
strobe light and inserted in the glow plug seat in order to illuminate
the combustion chamber; another endoscope was connected to the 2.2.2. Operating conditions
CCD camera to look into the cylinder. This arrangement allowed a prop- Engine was fuelled with commercial diesel fuel and with diesel and
er focus of the camera as shown in the combustion chamber image in methane. The main physicochemical properties of the tested fuels are
Fig. 2 where the injector tip and the piston dome are distinguishable. shown in Table 2.
The endoscope used to light the chamber was then removed and re- The engine was warmed up until the coolant temperature reached
placed by piezoelectric pressure transducer. During the acquisition of 80 °C. For all test cases, the inlet air temperature and humidity were
the combustion images, no external lightening was required because 23 °C and 45%, respectively. In order to isolate the effect of combustion
of the natural flame luminosity. A delay unit was used to synchronize evolution on exhaust emissions, the EGR was mechanically deactivated
the camera with the engine through the signal provided by the shaft and the DOC was removed and replaced by an element capable to repro-
encoder. duce the same pressure drop. The measurements were carried out at
Images were detected with an exposure time of 1 cad properly cho- several engine speeds and loads; the conditions at 1600 and 2800 rpm
sen to allow a comparative analysis between diesel and DF combustion. and at 30, 50 and 70% of the maximum torque output were shown. Ex-
For each crank angle, five repetitions of flame images were detected in periments were first performed by running the engine with diesel fuel
order to take into account the cycle-to-cycle variability. Flame images in order to provide a baseline data set for comparing diesel/methane op-
were post-processed by the AVL-ThermoVision program that allows to erations. Under DF mode, 10% of the full load, about 5 Nm, was provided
apply the two-colour pyrometry to evaluate the flame temperature by diesel combustion. The remaining load, to match the desired torque
and the soot concentration. output, was supplied by methane. Table 3 reports some characteristic
parameters of the tested operating conditions, for both diesel (D) and
2.2. Procedures DF operation. At a given combination of engine speed and load, lower
air mass flow rate was entrained in the cylinder in DF mode since it
2.2.1. Image processing was replaced by methane. Experiments were done with a methane per-
Images of the combustion phase were post-processed to evaluate centage energy substitution (PES), representing the fraction of the total
the temporal evolution of the flame area with the aim to perform a com- fuel energy provided by methane, higher than 50%. At fixed engine
parative analysis between different operating conditions and fuelling speed, the PES value increases with load since the diesel energy input
modes. Fig. 3 shows the scheme of flame images processing. Firstly, is fixed and more energy was supplied by methane to reach the desired
true colour images were converted to 8 bit format with 255 greyscale torque output. DF combustion is characterized by higher brake specific
of intensities representing the luminous levels. Then, greyscale images fuel consumption (bsfc) due to a not efficient methane combustion in
were transformed to binary by thresholding level of 0.5. This conversion poor air-methane mixture (air/fuel ratio higher than 2) as confirmed
was made by replacing all pixels in the input image with luminance by the lower fuel conversion efficiency (ηf) in DF mode with respect
greater than level with the value 1 (white) and replacing all other pixels to the conventional diesel operation. Similar conditions were kept for
with the value 0 (black). Each image includes 131.769 pixels; a pixel
corresponds to a combustion chamber area of about 0.1 × 0.1 mm2. Table 2
For each images the number of pixel with the value 1, corresponding Test fuels physicochemical properties.
to flame presence, was counted and the sum was divided for the total
Properties Diesel Methane
number of pixels of the image in order to better investigate the flame
distributions in the chamber. Chemical formula CnH1.8n CH4
Density @ 15 °C [kg/m3] 834.5 0.678
In order to analyse the effect of the fuel properties and operating Viscosity @ 40 °C [mm2/s] 3.34 –
conditions on the mechanisms of pollutant formation, flame images Lower heating value [MJ/kg] 42.8 50.0
were post-processed by AVL-ThermoVision software through two-col- AFRst 14.2 17.2
our pyrometry method to quantify the flame temperature and the in- Auto-ignition temperature [°C] 220 650
Flammability limits [% vol] 0.6–7.5 5–15
cylinder soot concentration that provide information about the two
Cetane number 55.1 –
main exhaust emissions of a diesel engine, NOx and PM, respectively Octane number – N120
[32]. Fig. 4 shows a typical temperature and soot concentration distribu- C [% mass] 84.71 75
tion resulting from the post processing of the flame image. About tem- H [% mass] 13.89 25
perature images, the colour scale was set to 1500–2700 K in order to O [% mass] 1.24 –
N [% mass] 0.16 –
simplify the reading of the temperature gradients. Concerning the in-
204 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210

Table 3
Operating conditions.

Fuel Speed [rpm] T/Tmax [%] T [Nm] P [kW] PES [%] mair [kg/h] mdiesel [kg/h] mCH4 [kg/h] bsfc [g/kWh] ηf [%]

D 1600 30 16 2.7 – 47.0 0.87 – 312.9 26.9

DF 57.1 45.8 0.50 0.56 408.4 19.0
D 1600 50 26 4.4 – 46.5 1.15 – 261.2 32.2
DF 66.7 45.0 0.45 0.76 274.0 27.8
D 1600 70 36 6.0 – 40.0 1.51 – 254.0 33.1
DF 69.0 38.0 0.52 1.00 248.0 30.6
D 2800 30 15 2.5 – 77.8 1.56 – 354.4 23.7
DF 51.7 77.5 1.02 0.94 453.7 17.2
D 2800 50 25 7.3 – 81.0 2.03 – 271.7 30.9
DF 59.8 79.3 1.02 1.31 308.7 25.0
D 2800 70 35 10.0 – 80.8 2.6 – 259.4 32.4
DF 64.2 78.8 1.05 1.61 261.0 29.3

all the tests for a better comparison of the results. In order to prove the
accuracy of the experiments, the coefficient of variation (CoV) of each
measured quantity was calculated. In Table 4 the maximum CoV detect-
ed for each parameter during the experiments are reported [35] show-
ing a good repeatability of the tests.
3. Experimental results
The in-cylinder pressure measurement is an effective tool to analyse
the engine behaviour since the pressure evolution affects the power
output and the engine out emissions. Fig. 5 shows the pressure histories
detected in the three cylinders for DF operation at 1600 rpm and 30% of
load. The injection signals of diesel fuel and methane for each cylinder
are also shown. For all operating conditions, diesel injections consisted
of two events, a pilot and a main. Methane was injected in the intake
manifold at 10 cad ATDC for each cylinder. Pressure traces reveal two
peaks due to the combustion of the premixed air/methane charge in-
duced by the two different ignition sources, the pilot and main diesel in-
jection. It is worth noting that a negligible variability, lower than 2%, is
observed among the pressure traces of the three cylinders demonstrat-
ing that there are no significant variations in combustion evolution
among the cylinders. Optical measurements were performed by
inserting an endoscope in the first cylinder. Based on the above consid-
erations, it can be assumed that the findings from optical measurements
applied to the first cylinder are valid for all the cylinders.
In order to analyse the DF configuration, some characteristic param-
eters of the combustion phase were calculated from the indicating data.
Table 5 reports the maximum in-cylinder pressure (pmax) and the crank
angle at which it occurs (cad pmax). The cad for 50% of mass fraction
burned (cad50), calculated according to the Wiebe function [36], and
the combustion duration, defined as the time between the cad for 90%
and 10% of mass fraction burned (cad90–cad10), are also shown. It
can be observed that the maximum pressure in DF mode is lower and
delayed with respect to conventional diesel operation. At the engine
speed of 1600 rpm, in DF mode, pmax is reduced by 3% at low load;
lower reduction, about 1%, is detected at medium and high load. At
2800 rpm, pmax decreases from 2 to 10% when the load increases from
low to high. Moreover, at 1600 rpm, DF mode pmax occurs 1.5, 2.4 cads
Table 4
CoV of the measured quantities.
Measured quantity CoV [%]
Engine speed 0.2
Engine torque 0.7
Air mass flow rate 1.3
Diesel mass flow rate 2.5
Methane mass flow rate 1.6
Exhaust temperature 2.7
imep 4.0
Flame emission intensity 4.3
Particles number 3.0
later than diesel fuelling at 30 and 50% of load while less relevant differ-
ences are observed at 70% of load. At 2800 rpm, pmax in DF mode is de-
layed of about 0.5, 1 and 3 cad at low, medium and high load,
respectively. This trend could be ascribed to the higher specific heat ca-
pacity ratio of methane with respect to air that lowers the temperature
and, hence, the pressure during the compression stroke. As conse-
quence, in DF mode, the combustion peak is lower and delayed with re-
spect to conventional diesel operation. For the same reason, the center
of combustion is retarded in DF mode as evidenced by the cad50 values.
Moreover, the presence of methane results in a longer combustion du-
ration, as proved by the higher values of cad90–cad10, ascribable to
the slower burning rate of the gaseous fuel at lean mixture conditions.
The combustion evolution in DF engines is affected by the interac-
tion between the liquid diesel and the gaseous methane that are charac-
terized by different flammability ranges (Table 2). An effective method
to characterize the main features of DF combustion is the optical visual-
ization of flame development. Fig. 6 shows a selection of images of com-
bustion process, in the angular range 2–30 cad ATDC, for diesel and DF
operation at 1600 rpm and 50% of full load. For conventional diesel fuel-
ling, the first luminous spots (evidenced by red circles) are distinguish-
able at 2 cad ATDC due to the pilot combustion. At 12 cad ATDC, two jets
of the main event are visible. Unlike the jets of the pilot event, the main
jets are detectable because lightened by the pilot combustion occurring
in the bowl. The jets move from the injector tip towards the chamber
wall. At 14 cad ATDC, luminous flames, due to the main fuel, appear in
the endoscope field of view. At 16 cad ATDC, air motions drive the
flames from the bowl wall towards the head contributing to the burning
of the impinged fuel. From 20 cad ATDC, the flames cover the whole en-
doscope field of view showing high luminous intensity. From 30 cad
ATDC, the piston is positioned away from the top dead center (TDC)
and the developing flames exhibit lower luminosity. The flame luminos-
ity typical of diesel combustion is due to the thermal radiation emitted
by particles formed during combustion. The burning high-temperature
carbon particles appear yellow-white. As the flame cools, the radiation
from the particles changes colour through orange to red as observed
also by other authors [36].
Concerning DF combustion, the first flames, due to diesel pilot com-
bustion (in the red circles), appear 1 cad later with respect to diesel op-
eration. The thermal and kinetic interaction between the liquid spray
and the gaseous fuel, in fact, changes the ignition characteristics of die-
sel fuel thus resulting more delayed. At 12 cad ATDC, the jets of diesel
main event are distinguishable. The injected liquid spray ignites several
points in the gaseous fuel-air mixture, forming more nuclei flame that
move throughout the charge. When DF combustion is analysed, it is nec-
essary to consider the different physicochemical properties of the meth-
ane with respect to diesel fuel leading to different combustion
mechanisms. DF mode, in fact, includes both the premixed combustion
of methane and the diffusive phase of diesel fuel. In the images present-
ed in Fig. 6 the premixed phase combustion is not detectable; on the
other hand, the features of diffusive combustion due to the diesel
injected to ignite the gaseous mixture, are recognizable. In DF operation,
 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210 205

Fig. 5. In-cylinder pressure histories detected in the three cylinders in DF mode at 1600 rpm and low load.

the flame is less widespread with respect to conventional diesel opera- Flame area graphs give information about the flame extension in the
tion because of the lower amount of injected diesel fuel. chamber and, hence, the sites where particles are developing. Another
In order to better investigate the diesel/methane combustion devel- parameter to take into account is the in-cylinder soot concentration.
opment, the temporal flame distribution was calculated as described in In this regards, combustion phase images were processed by two-colour
the previous section. Fig. 7 shows the percentage of the flame area in the pyrometry method as shown in Fig. 4. The integral soot concentration at
chamber versus cads for both diesel and DF operation, in the investigat- each crank angle was calculated and the maximum value detected dur-
ed conditions. The time resolved flame area histories are characterized ing the whole combustion process was reported in Fig. 8. It can be ob-
by an increasing rate due to the presence of fuel rich locations were served that DF combustion is characterized by lower soot formation
soot is formed. Flame area curves reach a peak and then decrease as with respect to conventional diesel operation at all investigated condi-
the combustion goes to the end. By comparing the flame evolution of tions. It is known from literature that soot is formed from unburned
diesel and DF combustion, it arises that the first flames are detected ear- fuel, which nucleates from the vapour to a solid phase in fuel-rich re-
lier for diesel operation with respect to DF combustion at all investigat- gions at high temperatures. Hydrocarbons or other available molecules
ed conditions, as evidenced by the zoomed images, in good agreement could condense on or be absorbed by soot. Particle emissions are the
with the indicated data (Table 5). This behaviour can be correlated to combination of soot and other liquid or solid phase materials that are
the methane presence that influences the diesel pre-ignition in a com- collected during expansion and exhaust stroke [37]. Therefore, the
plex manner depending on the gas fuel concentration and operating lower the in-cylinder soot concentration the lower is expected the par-
condition. The ignition of the liquid fuel, in fact, is retarded by the meth- ticle emissions at exhaust. The lower soot formation in DF mode could
ane/air charge aspiration that lowers the temperature in the chamber be ascribed to the lower amount of injected diesel fuel. In this regards,
with respect to the conventional diesel mode. The increasing rate of Graves et al. [38] found out that soot aggregates from DF combustion
the flame area curves of DF operation is smoother with respect to the have primary particle size, aggregate size, and effective density similar
diesel fuelling revealing a slower combustion development in DF to diesel soot. This study suggested that the main source of soot produc-
mode. Moreover, DF operation is characterized by smaller flame area tion in DF mode is the liquid fuel.
with respect to conventional diesel fuelling for the whole combustion The effect of DF combustion on particle number and size was
evolution since in DF mode the amount of diesel fuel injected in the analysed. Fig. 9 presents the PNSDFs for diesel and DF mode at
chamber is lower. This behaviour affects the pollutant formation: the 1600 rpm and 50% of load. The 3D graphs (Fig. 9-a and b) show the
smaller is the flame area the smaller should be the locations where par- PNSDF variations in a temporal range of 60 s. Particles range from 8
ticles are forming. up to 300 nm and exhibit bimodal distributions: the nucleation mode
(particles smaller than 20 nm) and the accumulation mode (particles
larger than 20 nm) are, in fact, distinguishable. The nuclei particles are
Table 5
mainly due to the condensation of semivolatile components. Mean-
Indicating data. while, the accumulation mode consists of carbonaceous agglomerates
[39].
Fuel Speed T/Tmax pmax cad cad50 cad90–cad10
[rpm] [%] [bar] pmax[deg] [deg] [deg]
It can be observed from the 3D graph that the measurements are
rather stable in the range of accumulation mode while the nucleation
D 1600 30 61.2 1.7 13.2 21.6
mode particles exhibit higher temporal variation for both fuelling
DF 58.9 3.2 13.4 22.0
D 1600 50 62.3 1.8 14.1 21.1 modes. For greater clarity, the temporal mean PNSDFs were calculated
DF 61.4 4.2 14.3 24.3 (Fig. 9-c). Moreover, to better characterize the particles emitted from
D 1600 70 65.6 16.6 14.8 16.6 DF combustion, the mean particle diameter (Dm) and the particle num-
DF 64.7 16.8 14.9 22.0
ber (N) were taken into account. Dm was calculated as weighted average
D 2800 30 60.1 1.0 13.7 15.7
DF 58.8 1.4 15.1 20.0
of diameters at each particle size:
D 2800 50 63.4 13.6 13.1 16.7
DF 58.6 14.4 15.9 22.7 n
D 2800 70 69.0 11.9 12.8 19.2 ∑i¼1 Dp;i  Ni
Dm ¼ n ð1Þ
DF 61.7 15.2 15.9 21.9 ∑i¼1 Ni
206 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210

where n is the total number of particle size classes, Dp is the particle di- convert dN/dlogDp to N, it is necessary to divide by 16 as the concentra-
ameter and Ni is the particle number concentration at the ith particle tion was normalized to 16 channels per decade:
size class. As the particle size classes are logarithmically spaced, to
dNi =d logDp
Ni ¼ ð2Þ
16

D DF By integrating the area under the curves dN/dlogDp over Dp, the total
particle number can be obtained.
Fig. 10 summarizes the particle number and diameter calculated for
2 cad ATDC diesel and DF combustion in the investigated conditions. It can be ob-
served that DF combustion allows to reduce the total particle number
with respect to conventional diesel fuelling. At 1600 rpm, Np is reduced
by 80% and 50% at low and medium load, respectively; smaller reduc-
tion, about 10% is observed at high load. At 2800 rpm, 50% reduction
of particle number emitted by DF operation is observed at both low
and medium load while higher reduction, N70%, is detected at high load.
The replacement of diesel fuel by methane allowed the particle
3 cad ATDC
emissions reduction, since the lower carbon atoms and no C\\C bonds
of the gaseous fuel with respect to diesel, thus reducing the soot
formation.
This trend is consistent with optical measurements that showed
lower flame area for DF combustions (Fig. 7). On the other hand, DF
combustion leads to the formation of larger particles with respect to
diesel operation. At 1600 rpm low load, particles emitted by diesel
12 cad ATDC and DF operation have almost the same size. Dm increases of about 10
and 17% at medium and high load, respectively, when the engine is
run with diesel/methane. Same trend is observed at 2800 rpm: DF com-
bustion emits particles larger than diesel operation of about 6, 14 and
10% at low, medium and high load, respectively. The bigger size of par-
ticles produced in DF mode could be ascribed to a different mechanism
of particle formation with respect to diesel one. The cooling effect of
14 cad ATDC methane, in fact, could worsen the pilot liquid fuel vaporization and
its combustion efficiency, as also evidenced by the delayed pilot com-
bustion (Fig. 6). Moreover, an important aspect to take into account is
that the particles size depends on both the surface growth and soot ox-
idation. As observed by Nithyanandan et al. [40], the addition of gaseous
fuel to diesel promotes particle growth while suppressing particle oxi-
dation. The result is a lower number of particles of bigger size emitted
16 cad ATDC by DF combustion with respect to conventional diesel operation.
Temperature images (Fig. 4) were processed to calculate the average
temperature during the combustion process, as shown in Fig. 11. The
symbols represent the temperature values obtained by averaging
pixel-by-pixel the temperature images; for a better understanding of
the data, the interpolation curves are also shown.
Concerning the temperature evolution at 1600 rpm and 30% of the
load, DF and diesel operation have almost the same temperatures in
20 cad ATDC the angular range 5–11 cad ATDC. Then, DF combustion exhibits a
peak of temperature of about 2540 K at 16 cad ATDC versus 2480 K de-
tected at the same time for conventional diesel fuelling. However, from
20 cad ATDC up to the end of combustion process, DF combustion shows
lower temperature than diesel mode. At 1600 rpm and 50% of load, in
the first phase of combustion, up to 18 cad ATDC, DF and diesel mode
have similar temperatures. Then, DF combustion is characterized by
22 cad ATDC
lower temperature (even 100 K lower), with respect to diesel operation.
Similar consideration can be done at 1600 rpm and 70% of load where D
and DF operation exhibit similar temperatures up to 15 cad ATDC and,
then, DF combustion proceeds with lower temperature (temperatures
differ by N 200 K). By looking at the flame temperature at high engine
speed, the trend of lower temperature detected for DF combustion
with respect to diesel operation is always confirmed. It is known from
30 cad ATDC literature [36] that the reactions of thermal NO formation occur at
high temperature in presence of oxygen. Tests with diesel and DF
mode were done with relative air-fuel ratio higher than 2 revealing
high oxygen availability for both fuelling. It can be inferred that the in-
cylinder temperature is the main parameter affecting the NO formation
Fig. 6. Flame images for D and DF mode at 1600 rpm and 50% of full load. in the analysed test cases. Due to the triple bond of nitrogen molecule
 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210 207

Fig. 7. Flame area for D and DF mode at 1600 (left) and 2800 rpm (right) at low, medium and high load.

(N2), temperatures of about 2000 K must be reached in order to split the
molecules and to enable the formation of NO according to the Zeldovich
mechanism. However, as soon as these temperatures are reached, the
NO production increases exponentially with temperature. Therefore,
the higher the flame temperature, the higher is expected the NO forma-
tion. This observation is confirmed by the exhaust measurements.
Fig. 12 shows the NO, NO2 and NOx emissions measured for diesel and
DF combustion in the investigated conditions. In the test points at
1600 rpm, NO emissions in DF mode are reduced with respect to diesel
operation. This trend could be ascribed to the lower in-cylinder temper-
ature reached in DF operation with respect to diesel fuelling, as found
out by the optical measurements. It is known from literature that NO
formed in the flame zone can be rapidly converted to NO2. Subsequent-
ly, the conversion of this NO2 to NO occurs unless the mixing with the
cooler fluid quenches the NO2 formed in the flame. This explanation is
consistent with the highest NO2/NO ratio occurring at lower loads
when cooler regions could quench the conversion back to NO. The
quenching effect could be more relevant in presence of the methane,
as proved by the higher NO2 emissions measured in DF mode. Therefore,
NOx emissions, as sum of NO and NO2, are reduced when the engine
runs with diesel/methane. However, NO2 emissions are higher in DF
mode and have a major weight on the total NOx with respect to NO.
NOx emissions are regulated but there is not distinction between NO
and NO2 so the increase of NO2 is not taken into account even if it is
more toxic than NO [41].
4. Conclusions
It is expected that internal combustion engines will still play, in the
near future, a relevant position in transportation and power generation
208 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210

7.0E+03 6.0E+03
D D

Max in-cylinder soot concentration

Max in-cylinder soot concentration
6.0E+03
DF 5.0E+03 DF
5.0E+03
4.0E+03

[mg/m3]
[mg/m3]

4.0E+03
3.0E+03
3.0E+03
2.0E+03
2.0E+03

1.0E+03 1.0E+03

0.0E+00 0.0E+00
1600 rpm-16 Nm 1600 rpm-26 Nm 1600 rpm-36 Nm 2800 rpm-15 Nm 2800 rpm-25 Nm 2800 rpm-35 Nm

Fig. 8. Max in-cylinder soot concentration for D and DF mode at 1600 (left) and 2800 rpm (right) at low, medium and high load.

a) b) c) 8.0E+6

8e+6 8e+6
6.0E+6

dN/dlogDp [#/cm3]
6e+6 6e+6
dN/dlogDp [#/cm3]
dN/dlogDp [#/cm3]

4.0E+6
D DF

4e+6 4e+6
2.0E+6

2e+6 2e+6
60 60
40 0.0E+0
] 40 ]
[s [s 10 100
20 e 20 e
0 im 0 m
10 T 10 Ti Diameter [nm]
100 0 100 0
Diameter [nm Diameter [nm
] ]

Fig. 9. PSDF versus time (a and b) and temporal average (c) PSDF for D and DF mode at 1600 rpm and medium load.

thus raising the necessity to use cleaner fuels in place of or as supple- to monitor the combustion development. Optical measurements of
ment to petroleum fuels. Diesel/methane dual fuel operation could be flame evolution were performed to analyse the cause-effect relationship
an effective way to face environmental problems ascribable to the ex- between combustion process and pollutant formation. NOx and particle
haust emissions. However, it is important to gain a better understanding emissions were measured at the exhaust.
of such aspects of the dual fuel combustion in order to help the further It was found out that DF combustion is characterized by higher brake
development and optimization of dual fuel engine strategies. In this specific fuel consumption than diesel operation. However, this draw-
regards, endoscopic imaging applied to production engines could be a back is compensated by the reduced fossil fuel consumption.
valid tool to analyse the combustion evolution and the pollutant DF mode exhibits lower and delayed pressure peak than diesel oper-
formation. ation as evidenced by indicating data. Moreover, the methane combus-
This study deals with an experimental investigation carried out on a tion phase is longer due to the slower flame propagation with respect to
1.0 L compression ignition engine modified to run with diesel/methane. the diesel fuel.
Tests were carried out at steady state conditions at different engine Optical investigation has evidenced that DF combustion is character-
speeds and loads. In-cylinder pressure measurements were performed ized by less extended flame area and lower in-cylinder soot

40 40
Dm Dm
50 [nm] 50 [nm]

60 60

70 70
Np
1.6E+08 1600 rpm 80 [#/cm3] 2.0E+08 2800 rpm 80
Np D DF D DF
[#/cm3]
1.3E+08 1.7E+08

9.5E+07 1.3E+08

6.5E+07 9.5E+07

3.5E+07 6.0E+07
30 50 70 30 50 70
T/Tmax [%] T/Tmax [%]

Fig. 10. Particle number and mean diameter for D and DF mode at 1600 (left) and 2800 rpm (right) at low, medium and high load.
 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210 209

Fig. 11. Average combustion temperature for D and DF mode at 1600 rpm (up) and 2800 rpm (down) at low, medium and high load.

1600 rpm-16 Nm 1600 rpm-26 Nm 1600 rpm-36 Nm 2800 rpm-15 Nm 2800 rpm-25 Nm 2800 rpm-35 Nm
NOx NOx
[g/kWh] [g/kWh]
NO2 NO2

NO NO

Fig. 12. NO and NO2 emissions for D and DF mode at 1600 rpm (left) and 2800 rpm (right) at low, medium and high load.

concentration than the diesel operation resulting in a reduction of total cad10 cad of 10% Mass Fraction Burned
particle number at the exhaust. This result is due to the lower carbon cad50 cad of 50% Mass Fraction Burned
atoms in the methane molecule and the absence of carbon-carbon cad90 cad of 90% Mass Fraction Burned
atoms with respect to diesel fuel. CH4 Methane
Moreover, DF combustion is characterized by lower flame tempera- CCD Charge Coupled Device
ture than diesel operation thanks to the lower heat capacity of methane. CI Compression Ignition
As consequence, DF combustion emits lower NO than conventional die- CO Carbon Monoxide
sel mode. On the other hand, for DF combustion, NO2 emissions increase CoV Coefficient of Variation
even if total NOx emissions are reduced in DF mode. CO2 Carbon Dioxide
Therefore, DF combustion benefits of simultaneously PM and NOx CN Cetane Number
reduction by overcoming the trade-off relationship typical of diesel CR Common Rail
engines. D Diesel
DEED Dekati Engine Exhaust Diluter
Abbreviations DF Dual Fuel
Dm Mean Particle Diameter
AFRst Stoichiometric Air Fuel Ratio DOC Diesel Oxidation Catalyst
ATDC After Top Dead Center DOI Duration of Injection
a.u. Arbitrary Unit Dp Particle Diameter
bsfc Brake Specific Fuel Consumption ECU Electronic Control Unit
cad Crank Angle Degree EEPS Engine Exhaust Particle Sizer
210 S. Di Iorio et al. / Fuel Processing Technology 159 (2017) 200–210

EGR Exhaust Gas Recirculation [14] B. Yang, M. Yao, W.K. Cheng, Y. Li, Z. Zheng, S. Li, Experimental and numerical study
on different dual-fuel combustion modes fuelled with gasoline and diesel, Appl. En-
ηf Fuel Conversion Efficiency ergy 113 (2014) 722–733.
imep Indicated Mean Effective Pressure [15] B. Yang, X. Wei, C. Xi, Y. Liu, K. Zeng, M.C. Lai, Experimental study of the effects of
m_ Mass Flow Rate natural gas injection timing on the combustion performance and emissions of a
turbocharge common rail dual-fuel engine, Energy Convers. Manag. 87 (2014)
MHC Methane Hydrocarbons 297–304.
N Total Particle Number [16] W.M. Yang, H. An, L. Duan, Impact of methane addition on the performance of bio-
N2 Nitrogen molecule diesel fuelled diesel engine, Appl. Energy 160 (2015) 784–792.
[17] A. Yousefi, M. Birouk, B. Lawler, A. Gharehghani, Performance and emissions of a
NO Nitrogen Monoxide
dual-fuel pilot diesel ignition engine operating on various premixed fuels, Energy
NOx Nitrogen Oxides Convers. Manag. 106 (2015) 322–336.
ON Octane Number [18] J.J. Hernandez, M. Lapuerta, J. Barba, Separate effect of H2, CH4 and CO on diesel en-
gine performance and emissions under partial diesel fuel replacement, Fuel 165
PES Percentage Energy Substitution
(2016) 173–184.
P Brake Power Output [19] B. Yang, L. Wang, L. Ning, K. Zeng, Effects of pilot injection timing on the combustion
PM Particulate Matter noise and particle emissions of a diesel/natural gas dual-fuel engine at low load,
pmax Maximum pressure Appl. Therm. Eng. 102 (2016) 822–828.
[20] S. Di Iorio, A. Magno, E. Mancaruso, B.M. Vaglieco, Characterization of particle num-
PMP Particle Measurement Programme ber and mass size distributions from a small compression ignition engine operating
PSDF Particle Size Distribution Function in diesel/methane dual fuel mode, Fuel 180 (2016) 613–623.
SOI Start of Injection [21] P. Geng, C. Yao, L. Wei, J. Liu, Q. Wang, W. Pan, J. Wang, Reduction of PM emissions
from a heavy-duty diesel engine with diesel/methanol dual fuel, Fuel 123 (2014)
T Brake Torque Output 1–11.
Tmax Maximum Brake Torque Output at fixed engine speed [22] www.dieselnet.com.
TDC Top Dead Center [23] J.V. Pastor, J.M. Garcia-Oliver, A. Garcia, C. Mico, S. Moller, Application of optical di-
agnostics to the quantification of soot in n-alkane flames under diesel conditions,
THC Total Hydrocarbons Combust. Flame 164 (2016) 212–223.
[24] E. Mancaruso, B.M. Vaglieco, Premixed combustion of GTL and RME fuels in a single
cylinder research engine, Appl. Energy 91 (2011) 385–394.
Acknowledgments
[25] S. Taniguchi, M. Masubuchi, K. Kitano, K. Mogi, Feasibility study of exhaust emis-
sions in Natural Gas Diesel Dual Fuel (DDF) engine, SAE Paper 2012-01-1649
Authors thank Mr. Carlo Rossi and Mr. Bruno Sgammato for the en- (2012).
gine assessment and for the support in the experimental activity. [26] N. Dronniou, J. Kashdan, B. Lecointe, K. Sauve, D. Soleri, Optical Investigation of Dual-
fuel CNG/Diesel Combustion Strategies to Reduce CO2 Emissions, SAE Paper 2014-
01-1313 (2014).
References [27] A. Magno, E. Mancaruso, B.M. Vaglieco, Combustion Analysis of Dual Fuel Operation
in Single Cylinder Research Engine Fuelled with Methane and Diesel, SAE Paper
[1] R. Folkson, Alternative Fuels and Advanced Vehicle Technologies for Improves Envi- 2015-24-2461 (2015).
ronmental Performance, Woodhead Publishing, UK, 2014. [28] S.C. Kong, L.M. Ricart, R.D. Reitz, In-Cylinder Diesel Flame Imaging Compared with
[2] B.B. Sahoo, N. Sahoo, U.K. Saha, Effect of engine parameters and type of gaseous fuel Numerical Computations. SAE Paper 950455, 1955.
on the performance of dual-fuel gas diesel engines-a critical review, Renew. Sust. [29] A. Magno, E. Mancaruso, B.M. Vaglieco, Analysis of combustion phenomena and pol-
Energ. Rev. 13 (2009) 1151–1184. lutant formation in a small compression ignition engine fuelled with blended and
[3] T. Korakianitis, A.M. Namasivayam, R.J. Crookes, Natural-gas fueled spark-ignition pure rapeseed methyl ester, Energy 106 (2016) 618–630.
(SI) and compression-ignition (CI) engine performance and emissions, Prog. Energy [30] J. Vattulainen, V. Nummela, R. Hernberg, J. Kytola, A system for quantitative imaging
Combust. Sci. 37 (2011) 89–112. diagnostics and its applications to pyrometric in-cylinder flame-temperature mea-
[4] IGU and UN ECE Joint Report, Natural Gas for Vehicles (NGV), The 25th World Gas surements in large diesel engines, Meas. Sci. Technol. 11 (2000) 103–119.
Conference “Gas: Sustaining Future Global Growth”, Kuala Lumpur, Malaysia, June [31] A.P. Carlucci, D. Laforgia, R. Saracino, G. Toto, Combustion and emissions control in
4–8 2012. diesel-methane dual fuel engines: the effects of methane supply method combined
[5] L. Wei, P. Geng, A review on natural gas/diesel dual fuel combustion, emissions and with variable in-cylinder charge bulk motion, Energy Convers. Manag. 52 (2011)
performance, Fuel Process. Technol. 142 (2016) 264–279. 3004–3017.
[6] N.N. Mustafi, R.R. Raine, S. Verhelst, Combustion and emissions characteristics of a [32] H. Zhao, N. Ladommatos, Optical diagnostics for soot and temperature measurement
dual fuel engine operated on alternative gaseous fuels, Fuel 109 (2013) 669–678. in diesel engines, Prog. Energy Combust. Sci. 24 (1998) 221–255.
[7] M.S. Lounici, K. Loubar, L. Tarabet, M. Balistrou, D.C. Niculescu, M. Tazerout, Towards [33] Quoc HX, Vignon JM, Brun M. A New Approach of the Two-color Method for Deter-
improvement of natural gas-diesel dual fuel mode: an experimental investigation mining Local Instantaneous Soot Concentration and Temperature in a D.I. Diesel
on performance and exhaust emissions, Energy 64 (2013) 200–211. Combustion Chamber. SAE Paper 910736.
[8] S. Di Iorio, A. Magno, E. Mancaruso, B.M. Vaglieco, Performance, Gaseous and Particle [34] A. Pungs, S. Pischinger, H. Backer, G. Lepperhoff, Analysis of the Particle Size Distri-
Emissions of a Small Compression Ignition Engine Operating in Diesel/Methane Dual bution in the Cylinder of a Common Rail DI Diesel Engine during Combustion and
Fuel Mode, SAE Paper 2016-01-0771 (2016). Expansion, SAE Paper 2000-01-1999 (2000).
[9] R.G. Papagiannakis, Study of air inlet preheating and EGR impacts for improving the [35] J.P. Holman, Experimental Methods for Engineers, McGraw-Hill, New York, 1994.
operation of compression ignition engine running under dual fuel mode, Energy [36] J.B. Heywood, Internal Combustion Engine Fundamentals, McGraw-Hill, New York,
Convers. Manag. 68 (2013) 40–53. 1988.
[10] L. Su, L.B. Zhou, S.H. Liu, H. Zhong, Decreasing hydrocarbon and carbon monoxide [37] D.R. Tree, K.I. Svensson, Soot process in compression ignition engines, Prog. Energy
emissions of a natural-gas engine operating in the quasi-homogeneous charge com- Combust. Sci. 33 (2007) 272–309.
pression ignition mode at low loads, Proc. Inst. Mech. Eng. D 219 (D9) (2005) [38] B. Graves, B. Patychuk, R. Dastanpour, S. Rogak, Characterization on particulate mat-
1125–1131. ter morphology and volatility from a compression-ignition natural-gas direct-injec-
[11] B. Yang, C. Xi, X. Wei, K. Zeng, M.C. Lai, Parametric investigation of natural gas port tion engine, Aerosol Sci. Technol. 49 (2015) 589–598.
injection and diesel pilot injection on the combustion and emissions of a turbo- [39] D.B. Kittelson, Engines and nanoparticles: a review, J. Aerosol Sci. 29 (5) (1998)
charged common rail dual-fuel engine, Appl. Energy 143 (2015) 130–137. 575–588.
[12] S.H. Yoon, C.S. Lee, Experimental investigation on the combustion and exhaust emis- [40] K. Nithyanandan, Y. Lin, R. Donahue, X. Meng, J. Zhang, C.F. Lee, Characterization of
sion characteristics of biogas-biodiesel dual-fuel combustion in a CI engine, Fuel soot from diesel-CNG dual-fuel combustion in a CI engine, Fuel 184 (2016)
Process. Technol. 92 (2011) 992–1000. 145–152.
[13] K. Ryu, Effects of pilot injection pressure on the combustion and emissions charac- [41] J. Czerwinski, J. Péterman, P. Comte, A. Mayer, Diesel NO/NO2/NOX emissions - new
teristics in a diesel engine using biodiesel-CNG dual fuel, Energy Convers. Manag. experiences and challenges, SAE Paper 2007-01-0321 (2007).
76 (2013) 506–516.