Energy 170 (2019) 375e390

Contents lists available at ScienceDirect

Energy
journal homepage: www.elsevier.com/locate/energy

Auto-ignition of direct injection spray of light naphtha, primary

reference fuels, gasoline and gasoline surrogate
Libing Wang a, Zengyang Wu a, Ahfaz Ahmed b, Jihad A. Badra c, S. Mani Sarathy b,
William L. Roberts b, Tiegang Fang a, *
a
Department of Mechanical and Aerospace Engineering, North Carolina State University, Raleigh, NC 27695, USA
b
King Abdullah University of Science and Technology, Clean Combustion Research Center, Thuwal, Saudi Arabia
c
Fuel Technology Division, R&DC, Saudi Aramco, Dhahran, Saudi Arabia

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

Article history: In this work, the spray and auto-ignition characteristics of light naphtha (LN), primary reference fuels
Received 20 April 2018 (PRF65, PRF95), Haltermann gasoline (CARB LEVIII, 10 vol% ethanol), and a gasoline surrogate were
Received in revised form studied in an optically accessible constant volume combustion chamber. An outwardly opening hollow
13 November 2018
cone piezoelectric gasoline direct injection fuel injector was used. Five ambient temperatures from 650
Accepted 18 December 2018
Available online 23 December 2018
to 950 K with a 75 K step were selected with a fixed ambient density of 3.5 kg/m3, similar to the Spray G
density defined by the engine combustion network (ECN). Fuel auto-ignition was achieved with varying
ignition delays for the five investigated fuels depending on the selected experimental conditions. Results
Keywords:
Gasoline direction injection
show that the auto-ignition locations are randomly distributed in the combustion chamber. Differences
Auto-ignition in ignition delay times among the five fuels are more significant when the ambient temperature is lower
Naphtha than 750 K. When the ambient temperature is lower than 750 K, PRF95 always has the longest ignition
Gasoline surrogate delay and LN has the shortest. Ignition delays of the five fuels are almost identical when the ambient
Constant volume combustion chamber temperature exceeds 750 K. Meanwhile, the five fuels have a similar spray front penetration length and
spray angles before the occurrence of auto-ignition under all the investigated conditions.
© 2018 Elsevier Ltd. All rights reserved.

1. Introduction efficiency and emissions [6,7]. Two compression-ignition combus-

tion modes are the most widely used in research [8e11]: premixed
Gasoline direct injection (GDI) has been considered to have the charge compression ignition (PCCI) and homogenous charge
potential to satisfy the stringent emission requirements due to its compression ignition (HCCI). PCCI engines were investigated in
advantages, and is becoming more popular in recent markets of terms of fuel economy and NOx reduction in the 1990s by Aoyama
spark-ignition (SI) engines. The incorporation of direct injection and co-workers [8]. By using a single-cylinder light-duty
strategy into a spark-ignition engine offers several improvements compression ignition engine with PCCI combustion, Hildingsson
including enhanced fuel economy, higher compression ratio, et al. [9] investigated the necessity of high volatility of different
reduced knocking tendency and improved transient responses. fuels to achieve more premixed combustion with low NOx and soot
Compared with conventional port-fuel injection (PFI) engines, they emissions. HCCI is an alternative combustion approach for gasoline
can produce 5e10% higher miles per gallon (MPG) and lower fuel compression ignition studies with higher combustion effi-
greenhouse gas emissions [1e4]. Recent research has been focusing ciency as well as low NOx and particulate emissions [10]. Experi-
on employing new modeling methods [5] and new combustion mental studies showed that stable GCI engine operation was
concepts such as gasoline compression-ignition (GCI) with the achieved down to idle speed and load on a multi-cylinder
potential of achieving diesel-like efficiencies and gasoline-like compression ignition engine using gasoline [11]. GCI engines offer
emissions. The use of gasoline fuels in compression-ignition com- significant advantages with increased efficiency and reduced
bustion modes has shown encouraging results in both engine emissions compared to either conventional diesel compression
ignition or gasoline spark ignition engines.
While gasolines with higher octane numbers can enable smooth
* Corresponding author. and efficient spark ignition (SI) engines, low octane gasoline-like
E-mail address: [email protected] (T. Fang).

https://doi.org/10.1016/j.energy.2018.12.144
0360-5442/© 2018 Elsevier Ltd. All rights reserved.
376 L. Wang et al. / Energy 170 (2019) 375e390

fuels might be more desirable in GCI engines. Compared to com- using a non-linear constrained optimization scheme detailed in
mercial gasoline and diesel fuels, blends of various refinery streams Ahmed et al. [35]. The formulated surrogate matches several
with low octane numbers (RON) in the 50e80 range have recently properties of Haltermann gasoline such as RON, MON, density,
been considered attractive alternatives to provide suitable chemical molecular weight, and volatility. These fuels and surrogates were
characteristics (longer ignition delay than diesel) in GCI engines investigated under different reactive spray conditions to measure
while lowering production cost and well-to-tank CO2 emissions. In various key spray and auto-ignition characteristics. By utilizing
addition, the utilization of the valuable high-RON streams and oc- fuels with a wide range of octane numbers, more insights can be
tane boosters is not a part of the refining process of producing low- obtained on the feasibility of using low octane number fuels and
octane fuels [12,13]. Hao et al. [12] found that compared with the naphtha fuels in gasoline compression ignition applications. The
conventional pathway, the low-octane gasoline-GCI pathway leads effects of fuel properties on auto-ignition and spray development
to a 24.6% reduction in energy consumption and a 22.8% reduction before ignition can also be investigated, facilitating further
in greenhouse gas (GHG) emissions. The research group in Saudi advanced study on gasoline direct injection spray under elevated
Aramco extensively investigated the combustion of various fuels in ambient conditions.
GCI engines [14e20]. Wang et al. [21] compared naphtha and
commercial gasoline in the multiple premixed compression igni-
tion (MPCI) mode. Their work demonstrated that naphtha MPCI can 2. Methods
operate stably in a wide load range. With the same experimental
system and combustion mode, they also found that the ignition Experiments were conducted in a constant volume combustion
delay of naphtha fuel is extended with the increase of injection chamber with six ports under simulated engine auto-ignition
pressure and the soot emission decreases at a high injection pres- combustion environments, as shown in Fig. 1. The advantage of
sure with higher CO and HC emissions [22]. Manente et al. [23] utilizing a constant volume combustion chamber is that one or
studied the effect of gasoline-like fuels on the performance and more key parameters can be controlled directly and precisely. This
emissions of an engine running in a partially premixed combustion enables us to better study the effect of investigated parameters on
(PPC) mode. Engines operating on lower octane number gasolines the spray and combustion. The optical access was enabled by a
were also capable of running idle without increasing the inlet quartz window installed on one port. Other ports were sealed by
temperature [24]. A more optimized combustion chamber design solid stainless steel plugs fitted with gas inlet, outlet, fuel injector,
was developed by Chang et al. [25] to improve the idle and light spark plug, pressure sensor, and other necessary hardware com-
load combustion stability of naphtha fuels. The application of low ponents. The inner diameter of the quartz window is 100 mm, and
cetane number (CN) naphtha fuels on diesel CI engines were the total chamber inner volume is 0.95 L. More details of the
evaluated by Won et al. [26] and Chang et al. [25] with different chamber can be found in the previous publications [36e38]. An
bowl and nozzle designs. Diesel CI engines can run with low CN outwardly opening piezoelectric fuel injector was employed in this
naphtha fuels without compromising the vehicles' drivability, study. The fuel injector can produce a hollow cone spray. The spray
emissions, and power requirements. CN 35 naphtha fuels were shape is determined by the design of the nozzle exit without any
recommended for their better robustness and lower HC and CO swirl motion. It was controlled by a piezo injector driver from Na-
emissions. Zhang et al. [27] compared naphtha fuel and ultra-low tional Instruments, Inc. Fuel injection pressure was maintained at
sulfur diesel (ULSD) under both conventional and low tempera- 10 MPa (100 bar) by a gasoline common rail fuel system through all
ture combustion modes. Naphtha fuels exhibited appreciably lower experiments. More details of the injector and fuel injection system
levels of soot compared to ULSD across the investigated engine can also be found in a previous publication [39]. Injection pulse
operation range while maintaining a diesel equivalent fuel duration for all investigated fuels was kept at 0.3 ms by using an
efficiency. external pulse generator. Temperature of both the chamber walls
Generally speaking, naphtha is composed of C5 to C11 hydro- and fuel injector body were controlled at 90  C. The fuel tempera-
carbons and has a low research octane number (RON) value, ture at the injector tip was assumed similar to the body tempera-
roughly within the 40e70 range [28]. Naphtha requires less pro- ture due to the small amount of fuel injected. In order to generate
cessing in the refinery than either gasoline or diesel [29], and the an auto-ignition environment, a premixed combustion of lean
greenhouse gas emissions of the well to pump process are reduced acetylene-air mixture was employed prior to the fuel injection.
from 18 g/MJ fuel energy of gasoline to only 12 g/MJ fuel energy for Three different gases, acetylene, O2/N2 (50%/50%) and dry air were
naphtha. Thus, there is an additional benefit in terms of the well-to- used to formulate appropriate mixtures to obtain the desired
wheel CO2 emissions and overall energy consumed, and it can also
be considered as a way of increasing the efficiency of a gasoline
powertrain system [19]. The auto-ignition quality of a fuel at a given
engine condition is described by an octane index defined as, OI ¼
(1-K) RON þ K MON, where RON and MON are the Research and
Motor Octane numbers respectively and K depends only on the
engine design and operating conditions [30]. Approaches like direct
injection, higher compression ratios, downsizing, and turbocharg-
ing can reduce the unburnt gas temperature for a given pressure
and push the value of K downwards [31]. Fuels with high ignition
resistance like gasoline has a significantly longer ignition delay and
hence results in lower NOx and smoke for a given load compared to
diesel fuels [32].
This study investigates the auto-ignition and spray character-
istics of several gasolines and surrogates with a wide range of oc-
tane numbers. The fuels studied include light naphtha, primary Fig. 1. Experimental system: 1. fuel injector, 2. exhaust line, 3. chamber body, 4. quartz
reference fuels, Haltermann gasoline [33] and a gasoline surrogate window, 5.plug/window retainer, 6. pressure transducer, 7. intake line, 8. metal plug, 9.
[34]. The surrogate fuel for Haltermann gasoline was prepared spark plug, 10. Combustion chamber, 11. high speed camera.
 L. Wang et al. / Energy 170 (2019) 375e390 377

ambient temperature and ambient oxygen concentration. Premixed binary mixture of iso-octane and n-heptane and is used as a sur-
reactants were ignited by a spark plug. In this study, the ambient rogate for gasoline. PRF65, a blend of 65% iso-octane and 35% n-
oxygen concentration was set as 21% for all experiments before fuel heptane by volume, was used here as a simple surrogate for LN.
injection. The compositions of the gas mixtures before and after the Similarly, PRF95 (95% iso-octane and 5% n-heptane by volume) was
premixed acetylene combustion are listed in Table 1. selected as a surrogate for the gasoline fuel. The Haltermann CARB
A Kistler 6041B transducer coupled with a Kistler 5004 charge LEV III RON 91 gasoline fuel was also used in this study. A gasoline
amplifier was utilized to measure the transient chamber pressure surrogate was formulated, prepared, and tested along with other
during the entire combustion process, and the ambient tempera- fuels. The surrogate fuel for Haltermann gasoline has been prepared
ture was calculated based on the pressure data using a corrected using methodology described in Ahmed et al. [35]. Haltermann
ideal gas law. A monochrome high-speed digital video camera gasoline is an oxygenated fuel (~17 mol%) with significant per-
(Phantom V4.3 camera from Vision Research Inc.) was used to centages paraffins (~40 mol%), aromatics (~23 mol%) and naphthenes
capture both the spray and the auto-ignition processes. The camera (~16 mol%). The selection of target properties for surrogate fuel is
was set at 5500 fps with a resolution of 320  312 pixels for auto- roughly based on conclusions of a recent study by Violi and co-
ignition experiments. A 50 mm fixed focal length lens was used workers [41] and our own works on gasoline surrogates
to collect signals. Camera exposure time and the aperture number [33,34,42e47], and includes RON, MON, density and H/C ratio. The
(F#) were selected based on the flame intensity under different detailed hydrocarbon analysis (DHA) of fuel was conducted at Saudi
ambient conditions for fuel auto-ignition experiments. The settings Aramco in compliance with standards ASTM 6733 and 6730 [48]
of exposure time and aperture number (F#) are listed in Table 2. The and the palette species for the surrogate fuel are chosen based on
auto-ignition combustion images for each condition were captured their structural similarities with species present in higher con-
using the high speed camera and saved for later processing. centrations in the DHA. The surrogate palette consists of toluene, n-
Because the focus of the current study was to measure the ignition heptane, iso-octane, cyclopentane, 1,2,4-trimethylbenzene, and
delay, therefore, the camera optical settings were adjusted for ethanol. The relevant values of the target properties were obtained
capturing early weak flames to identify the ignition locations and from the DHA and certificate of acceptance from the fuel manu-
time delays. Although there exists flame intensity saturation during facturer. The blend of the surrogate palette was then optimized
the late stage of combustion, the saturation has few effects on the using the approach in Ahmed et al. [35] to match the target prop-
flame development and shape for qualitative analysis. The time erties. The gasoline surrogate composition is shown in Table 3. The
between the timing of the first flame image with detected ignition properties of the five examined fuels are listed in Table 4 and the
points and the start of fuel injection timing was used as the ignition distillation curves are shown in Fig. 2. As shown in Table 4, the fuel
delay time in this work. The average ignition delay time from five properties depend highly on the compositions, and heat of vapor-
repeated measurements was used as the ignition delay for each ization (HOV), fuel volatility and viscosity can change significantly.
condition. For the fuel spray experiments, the spray development The vapor pressures can also correlate well with the flash points of
during the fuel injection process was visualized by applying a Mie- the fuels. During the experiments, the fuel injection pressure and
scattering technique, and all the camera settings were kept the injection duration were kept constant. Due to the different den-
same (F# 2.8, exposure time 37 mm, resolution 128  152, and sities of the fuels, the injected fuel mass quantities of gasoline,
21000 frames per second). In the current investigation, a pulse gasoline surrogate, LN, PRF65 and PRF95 are 10 mg, 9.85 mg,
delay generator was used to enable an external trigger to syn-
chronize the image capturing of the high-speed camera and the fuel
injection. Rising edge of the triggering pulse was used for all the Table 2
devices. Experiments were repeated five times for each case. In The F# and exposure time for the auto-ignition test conditions.
order to account for the uncertainty, data from the five runs were
F#/Exposure time (us)
averaged, and the standard deviation was then calculated and
reported. Ambient temperature (K) 650 725 800 875 950
*Ambient pressure (kPa) 657 733 809 884 960
For all the experiments, gas density in chamber was set as
Gasoline 2.8/178 2.8/178 2.8/120 2.8/120 2.8/120
3.5 kg/m3 to be consistent with the non-reacting spray G condition Gasoline surrogate 2.8/178 2.8/178 8.0/178 11/178 11/178
defined in the engine combustion network (ECN) by Sandia Na- LN 2.8/178 2.8/50 11/5 11/5 11/5
tional Laboratories [40]. While the ambient temperature and PRF65 2.8/166 2.8/160 8/20 8/1 11/5
pressure condition changes for different cases, the ambient air PRF95 2.8/166 2.8/160 5.6/20 5.6/1 11/5

density was kept the same throughout this study. Five different
ambient temperatures ranged from 650 K to 950 K with a 75 K in-
terval were selected, with the ambient equivalent pressures of 657, Table 3
Haltermann gasoline surrogate composition.
733, 809, 884, and 960 kPa, respectively (listed in Table 2). Five
different fuels were investigated: light naphtha (LN), two primary Species Wt (%) mol (%)
reference fuels (PRF65, PRF95), gasoline (Haltermann CARB LEV III) Toluene 6.593 6.602
and gasoline surrogate (Haltermann surrogate). n-Heptane 14.286 13.159
The LN used in this study was obtained from Saudi Aramco, with iso-Octane 35.165 28.413
Cyclopentane 13.187 17.352
an AKI (the average of the research octane number (RON) and the
1,2,4-Trimethylbenzene 21.978 16.870
motor octane number (MON)) value of 64. Conventionally, PRF is a Ethanol 8.791 17.604

Table 1
Gas compositions before and after the premixed combustion.

States C2H2 (V%) O2 (V%) CO2 (V%) H2O (V%) N2 (V%) MW (g/mol) Density (kg/m3)

Before premixed combustion 4.1 44.2 0 0 51.1 28.6 3.5

After premixed combustion 0 21 9.8 4.9 64.3 29.9 3.5
378 L. Wang et al. / Energy 170 (2019) 375e390

Table 4
Selected fuel properties of investigated fuels.

PRF65 PRF95 LN Gasoline Gasoline surrogate

Density (gm/cm3) 687 692 654 755 743

Boiling point ( C) 98e99 98e99 50e73 41e176 71e165
Viscosity (N-s/m2) 4.32E-04 4.53E-04 2.74E-04 4.95E-04 5.62E-04
Surface tension (N/m) 1.87E-02 1.83E-02 1.63E-02 2.15E-02 2.23E-02
HOV (J/kg) 3.30Eþ05 3.14Eþ05 3.63Eþ05 5.01Eþ05 4.55Eþ05
LHV (MJ/kg) 44.40 44.32 45.08 41.92 41.78
RON 65 95 65 91 91
MON 65 95 63 83 83
AKI 65 95 64 87 87
Vapor Pressure at 300 K (kPa) 7.00 7.16 50.20 32.62 12.70

Fig. 2. Distillation curves of the five investigated fuels.

8.69 mg, 9.13 mg and 9.20 mg, respectively. Based on the spray G
condition [40], the ambient gas density is 3.5 kg/m3 and the oxygen
concentration before fuel injection is 21% by volume for spray auto-
ignition. The global average fuel/air mass ratio can be estimated as
0.003 (fuel lean condition) for the current experiment conditions.
But due to the direct injection nature, the local fuel/air mass ratio
can change significantly.

3. Results and discussions

In this section, auto-ignition characteristics of different fuels

and fuel spray development before auto-ignition will be discussed.
First of all, the auto-ignition combustion images are presented and
discussed. For each set of images, locations of the injector, pressure
transducer and the chamber view boundary are all the same, as
marked in Fig. 3. Fuel ignition delay, spray angle, spray front
penetration length, and time-averaged spray front velocity were
calculated and analyzed. Fig. 3. Locations of injector, pressure transducer and chamber view boundary.

3.1. Auto-ignition images features of the auto-ignition combustion of PRF65 under different
ambient temperatures is presented in Fig. 4 as an illustrative
In this study, fuel auto-ignition was observed for all the selected example. For each temperature, twenty consecutive images are
fuels under all the investigated ambient conditions. As mentioned selected to cover the ignition and combustion processes. Fuel type
above, experiments under each ambient condition were repeated and image timing after the start of the injection trigger are marked
five times. In this section, one set of images showing the typical on the upper left corner. The ambient temperature is noted at the
 L. Wang et al. / Energy 170 (2019) 375e390 379

Fig. 4. Auto-ignition images of PRF65 under different ambient temperatures and pressures with a constant density of 3.5 kg/m3.

bottom left corner. The related ambient pressure is also noted in the depends heavily on the ambient conditions. Under a lower ambient
figure captions. The time marked on the images corresponds to temperature, flame spreads more slowly than under a higher
each individual tests and may differ slightly from the averaged ambient temperature. Once the ambient temperature is higher
ignition delay values, since ignition delay results (presented in the than 800 K, the flame shifts towards the left and shows a more
next section) are an average of five experiment runs. elliptical shape, which is closer to the original spray cone shape.
From PRF65 auto-ignition images, it can be seen that the loca- This is due to the shorter ignition delay under higher ambient
tion of the occurrence of auto-ignition is rather random in the temperature conditions, and could be better explained in combi-
combustion chamber, especially under lower ambient tempera- nation with the discussion of ignition delay in later sections.
tures. For most of the experiments, the first detected flame pockets
are located at the upper-half region. Unlike HCCI combustion where 3.2. Ignition delay
fuel ignites almost simultaneously, a flame expansion process is
observed in this study for gasoline-like fuels. The flame expansion Ignition delay in this work is defined as the time interval
380 L. Wang et al. / Energy 170 (2019) 375e390

Table 5 between the start of the injection trigger and the occurrence of
Ignition delay of gasoline, gasoline surrogate, LN, PRF65 and PRF95 at different auto-ignition. The camera and the injector were triggered simul-
ambient temperatures (*: Ambient pressures are determined by the ideal gas law
under 3.5 kg/m3 density).
taneously by a pulse generator in this study. The timing of auto-
ignition occurrence is defined to be the first frame where com-
Ignition delay (ms) bustion flame appears. Therefore, the ignition delay could be ob-
Ambient temperature (K) 650 725 800 875 950 tained from the injection timing and captured high speed video
*Ambient pressure (kPa) 657 733 809 884 960 images. The final ignition delay value of each fuel under each
Gasoline 7.10 2.31 1.15 0.87 0.71
investigated condition is the average of the five repeats of experi-
Gasoline surrogate 5.76 2.08 1.25 1.10 0.89
LN 4.83 2.00 1.39 1.17 0.99 ments with the same experimental settings, and the standard
PRF65 10.06 3.01 1.65 1.16 0.95 derivation is calculated to evaluate the variation of the results, as
PRF95 12.27 3.81 1.17 1.37 1.09 show in Table 5. It is seen from the ignition delay measurements
that the ignition delay times range between 0.887 ms and
12.265 ms. The reported ignition delay times are all longer than the

Fig. 5. Comparison of auto-ignition images of five fuels under 650 K and 657 kPa ambient conditions with a density of 3.5 kg/m3.
 L. Wang et al. / Energy 170 (2019) 375e390 381

injection duration (0.3 ms) used in this study. As will be discussed ambient temperature. The uncontrolled spray development and
in the next section, the measured physical injection process ends at droplet vaporization after the end of injection, as well as the fuel
about 0.43 ms after the injection trigger pulse. Therefore, a vapor turbulent interaction with the ambient air, may be the key
partially-premixed combustion mode is expected here and the reasons for the random auto-ignition locations and flame shape.
mixture stratification is dependent on the ambient temperatures Finally, no obvious spray diffusion combustion is observed even for
and pressures. At low ambient temperatures, the injected fuel has the 950 K ambient temperature condition due to the relatively
more time to mix with the air and hence the stratification is lower short fuel injection duration. Comparison of auto-ignition flame
at the time of autoignition. The stratification level is significantly images of all the five fuels under the 650 K and 950 K conditions are
higher when operating at high ambient temperatures because of presented in Figs. 5 and 6. It is found that flame characteristics
the relatively short time between the end of injection and the discussed above could also be found for the other four fuels in this
autoignition of the fuel/air mixture. In addition, the short mixing study. Flames structures share almost the same patterns for all five
time helps explain why flames generally shift to the left towards the fuels at different ambient temperatures.
injector nozzle and becomes more elliptical with increasing The measured ignition delay times for all the fuels at all the

Fig. 6. Comparison of auto-ignition images of five fuels under 950 K and 960 kPa ambient conditions with a density of 3.5 kg/m3.
382 L. Wang et al. / Energy 170 (2019) 375e390

conditions are presented in Fig. 7. As expected, the ignition delay phenomenological model for diesel spray combustion by Dec [50],
times of all the examined fuels decrease when increasing the which concluded that ignition occurs at locally rich equivalence
ambient temperatures and pressures. This is due to the higher ratio regions (e.g., f ¼ 2e4).
reactivity of the fuels at higher temperatures. However, the As can be seen from Fig. 8, the ignition delay times of the three
decrease in ignition delay times as a function of ambient temper- fuels are stronger functions of temperatures at lower than 750 K
ature can be divided into two distinct regimes. The first regime is temperatures. However, a clear negative temperature coefficient
for temperatures between 650 K and 800 K where the decrease of (NTC) behavior is overserved at temperatures between 750 K and
the ignition delay times is a strong function of temperature. The 900 K where the ignition delay time dependency on temperature is
second regime is for temperatures higher than 800 K where the weak. Moreover, the similarities in the ignition delay times pre-
ignition delay times of all the fuels collapse to similar values and sented in Fig. 7 at temperatures higher than 800 K can also be
their decreasing rates become almost negligible when further explained by the presented homogeneous ignition delay times
increasing the ambient temperature. where the ignition timings of the three fuels are similar outside the
To further analyze the discussed trends, the homogeneous gas- NTC region. These observations are in line with the findings from
phase ignition delay times of PRF65, PRF95 and the gasoline sur- recently reported experimental ignition delay time measurements
rogate are calculated at 800 kPa (8 bar) and equivalence ratios of of LN and its surrogates [49]. Comparing the ignition delay among
0.5, 1.0, and 2.0 with CHEMKIN software using the mechanism in the five investigated fuels, PRF95 always has the longest ignition
Ref. [49]. The homogeneous gas-phase ignition delay times for the delay and LN has the shortest ignition delay when the ambient
three fuels under different ambient temperatures from 650 K to temperature is lower than 750 K. Gasoline surrogate has a slightly
1200 K are shown in Fig. 8. It should be noted that these simulations shorter ignition delay than neat gasoline, and as expected, the
are intended only to qualitatively understand the observed exper- ignition delay of PRF65 is lower than PRF95.
imental trends, and quantitative agreement is not the objective. A The fuel reactivity, defined as the homogenous ignition delay
comparison between the absolute values of the measured and time, is not the only controlling parameter when it comes to the
calculated ignition delay times cannot be made here because they ignition delay times of reacting sprays. Many other physical and
were obtained with different physical boundary conditions. As chemical fuel properties affect the autoignition. For example, the
mentioned earlier, the experimental ignition delay times, by defi- octane numbers (RON and MON) are very important indicators of
nition, have spray injection, breakup, evaporation, mixing and the fuel's reactivity. Higher ON indicates lower reactivity and hence
autoignition embedded in them. However, the calculated homog- more resistance to autoignition under different operating condi-
enous ignition delay times are from a 0-D homogeneous reactor tions. The correlation between AKI (average between RON and
where the vapor phase fuel and air are perfectly premixed awaiting MON) of the tested fuels and the reported ignition delay times is
for autoignition. In general, the calculated ignition delay times are obvious at low temperatures. LN has the lowest AKI and PRF95 has
significantly higher than the measured ones. There are many fac- the highest AKI. These correspond to the shortest and longest re-
tors that can explain these differences, however, the main one is the ported ignition delay times at 650 K for LN and PRF95, respectively.
equivalence ratio at which autoignition occurs. Fig. 8 shows that the However, the ignition delay times cannot be explained by only
ignition delay time is a strong function of equivalence ratio and looking at the ON (or AKI) because the gasoline fuel and its surro-
richer mixtures are more reactive. In fact, the calculated ignition gate have similar AKI but different ignition delay times at 650 K.
delay time of PRF65 at f ¼ 2 and T ¼ 650 K is shorter than the The equivalence ratio-temperature distribution is a key deter-
measured one for the same fuel at the same temperature condi- mining factor in compression-ignition engines [51e53] and like any
tions. In the experiments, the fuel-air mixing is highly stratified compression-ignition combustion system, the ignition delay times
because of injection, breakup, evaporation, and mixing. We there- of reacting sprays are dependent on the local temperature and
fore believe that the experimental autoignition is mostly initiated equivalence ratio distribution in the constant volume combustion
at locally rich fuel-air pockets leading to a significantly lower chamber. Therefore, any fuel properties that affect the equivalence
ignition delay times compared to the calculated ones at lean and ratio-temperature distribution should be considered when
stoichiometric conditions. This is consistent with the analyzing the trends observed here. Below is a list of the properties

Fig. 7. Ignition delay of the five investigated fuels at the five tested ambient temperatures and pressures: 650 K (657 kPa), 725 K (733 kPa), 800 K (809 kPa), 875 K (884 kPa), and
950 K (960 kPa).
 L. Wang et al. / Energy 170 (2019) 375e390 383

low-temperature heat release rates and hence the main com-

bustion event.
 Viscosity: the droplet breakup is inversely proportional to the
viscosity of the fuel. The droplet breakup affects its evaporation
and hence mixing.
 Surface tension: the droplet breakup is inversely proportional to
the surface tension of the fuel.

These fuel properties need to be considered altogether when

trying to analyze the similarities and differences between the
ignition delay times of fuels with similar reactivity (RON, MON and
AKI). It was shown earlier that the physical and chemical properties
of fuels with the same octane ratings affect their combustion in
gasoline compression ignition mode [52e54].
PRF65 and PRF95 have very similar physical and chemical
properties except their ON. The differences between their ignition
delay times is attributed to the 30 RON difference between these
two fuels. LN and PRF65 have similar ON but different ignition delay
times. The differences in the ignition delay times are caused by
other physical and chemical properties of both fuels. The density of
LN is lower than that of PRF65 and hence the spray exit velocity of
LN is higher. This higher velocity will facilitate the droplets breakup
and the mixing with air resulting in overall leaner mixtures. Lean
mixtures are less reactive than rich ones and hence LN should have
had longer ignition delay times compared to PRF65 if density was
the only contributing parameter. However, it is not the case here
and other parameters such as the boiling characteristics are
affecting the ignition delay times of LN and PRF65. LN has much
lower final boiling point (FBP) compared to PRF65 and hence the
injected fuel quickly evaporates and mixes with the air close to the
injector. This sequence of events results in richer fuel/air pockets
that are located closer to the injector when compared with the
PRF65 case. These richer pockets cause the ignition delay times to
be shorter than those of PRF65. This finding is in agreement to what
was recently observed in engine simulations and experiments us-
ing LN and PRF65 surrogate [51e53]. PRF 65 has significantly higher
reactivity compared to the gasoline fuel and its multi-component
surrogate, however, it has longer ignition delay times at low tem-
peratures (Fig. 7). The density differences between PRF65 and the
gasoline fuel as well as its surrogate are noticeable and certainly
plays an important role in the observed trend. The gasoline fuel and
its surrogate have higher densities compared to PRF65 and hence
the spray exit velocity is lower resulting in richer fuel/air pockets.
These localized rich/air mixtures control the ignition event and
Fig. 8. Calculated ignition delay time of the three surrogate fuels (PRF65, PRF95, and
Gasoline surrogate) under different temperatures and an ambient pressure of 800 kPa
partially explain why the more reactive PRF65 has ignition delay
(8 bar): (a) equivalence ratio of 0.5; (b) equivalence ratio of 1.0; and (c) equivalence times that are longer than those of RON91 fuels. In addition to the
ratio of 2.0. differences in densities, the gasoline fuel and its surrogate has
wider boiling curve compared to PRF65 and hence some of the light
components from the gasoline fuel and its surrogate evaporate
before PRF65 and cause rich fuel/air pockets to form in the vicinity
that might affect the mixing and hence combustion in the pre-
of the injector. The boiling range effect also contributes towards
sented spray setup:
aiding the RON91 fuels to have lower ignition delay times. Inter-
estingly, the contributions of different factors yielded shorter
 Boiling curve: the distillation curve of the fuel is an indicator of
ignition delay times for RON91 fuels despite PRF65 has 26 RON
the evaporation of the fuel and hence affects its mixing with the
advantage. Finally, the ignition delay times of the gasoline surro-
surrounding air.
gate are close to those of the gasoline fuel except at 650 K where the
 Density: the spray exit velocity is inversely proportional to the
gasoline surrogate is slightly more reactive. This again can be
square root of the fuel density. The spray exit velocity affects the
explained by the small differences in the boiling characteristics.
spray penetration, breakup, and mixing with air.
However, the gasoline surrogate generally successfully reproduces
 Heat of vaporization (HOV): the temperature of the evaporated
the ignition quality of the neat gasoline fuel.
fuel is inversely proportional to the heat of vaporization of the
fuel. This directly affects the temperature at which the fuel is
3.3. Fuel spray characteristics before ignition
distributed at.
 Lower heating value (LHV): the lower heating value is a measure
In order to further analyze the resulting auto-ignition process
of the energy contained in a fuel. This energy density affects the
and the variation in the ignition delay of different fuels,
384 L. Wang et al. / Energy 170 (2019) 375e390

macroscopic spray features of each fuel before auto-ignition were ended at around 0.43 ms after the start of injection trigger (ASOIT)
studied to help understand the spray characteristics under com- due to the delay caused by opening and closing injector nozzle. To
busting environments. Generally speaking, the fuel spray angle and ensure result consistency, the ambient conditions were kept similar
the fuel front penetration length can be used to evaluate physical to the fuel auto-ignition experiments. In other words, the presented
spray characteristics. Therefore, the time averaged fuel spray angle, spray images exactly represent the spray development before the
spray front penetration length and spray penetration velocity were occurrence of fuel auto-ignition under investigated ambient con-
calculated and discussed. The fuel spray angle is defined as the ditions. Since the spray liquid signal becomes negligible right after
averaged angle between the two spray edges at pre-determined the end of injection, the spray angle and spray front penetration
locations near the injector tip at a time after the start of fuel in- length calculation were limited to the time from first spray image
jection [55]. In this study, the effective locations are selected from timing to the end of fuel injection.
the nozzle orifice to the spray front. The spray front penetration A set of spray images of PRF65 under different ambient tem-
length is defined as the average distance between the entire spray peratures are selected as an example, as shown in Fig. 9. For each
front and the nozzle exit plane. The spray penetration velocity is image, the two spray edges and the spray front were marked red,
calculated based on the spray front penetration length results. In blue and green, respectively. The ambient temperature ranges from
order to calculate the spray angle and the spray front penetration 650 K to 950 K with a constant in-chamber gas density of 3.5 kg/m3.
length, the two spray edges and the spray front were first detected It has been discussed in Ref. [55] that a higher ambient pressure
and marked with colored lines using a Matlab code developed in- lead to a shorter spray penetration length and stronger edge vortex
house with the same threshold for all transient images in each at downstream under the same ambient temperature. In this work,
single injection. The electronic injection pulse duration was set as it turns out that the spray under different ambient temperature
0.3 ms. For all the experiments, the physical injection process conditions has almost the same structure and development for the

Fig. 9. PRF65 spray images before ignition under different ambient temperatures and pressures with a constant density of 3.5 kg/m3.
 L. Wang et al. / Energy 170 (2019) 375e390 385

Fig. 10. Comparison of fuel spray before auto-ignition of five fuels at 650 K and 657 kPa ambient conditions with a density of 3.5 kg/m3.

same fuel. Downstream vortices occur at similar locations for all the angle decreases with increasing ambient temperature, which is
investigated conditions. The only obvious difference is that liquid believed to be caused by a faster fuel evaporation rate under a
signal decreases with increasing ambient temperature caused by higher ambient temperature environment. Under higher ambient
increasing fuel evaporation rates. This indicates that the ambient temperatures the droplets at the periphery of the cone-shaped
gas density might be the dominant factor that determines the spray spray turn into vapor faster, leading to a smaller spray angle.
structure. In addition, the spray structure comparison of the five Comparing the time-averaged spray angle among the five selected
fuels at 650 K and 950 K is show in Figs. 10 and 11. It is seen that the fuels, PRF65 and PRF95 always have larger spray angles than the
differences in the fuel spray structure and development among all other three fuels for all investigated ambient conditions, as show in
the investigated fuels are small. Fig. 13. The difference of the time averaged spray angle among
different fuels also becomes smaller with increasing ambient
temperature. Meanwhile, it should be noted that the variation of
3.3.1. Time-averaged spray angle the time-averaged spray angle is quite small among different fuels.
The time-averaged spray angle of each fuel was calculated. In All the time averaged spray angle results are around 100 , and the
this study, the time interval from the start of the injection trigger to maximum difference is less than 3 .
the end of the injection was used for calculating the time-averaged
spray angle. The time-averaged spray angle of each fuel under
different ambient temperatures is shown in Fig. 12. The effect of the 3.3.2. Transient spray penetration length and front penetration
ambient temperature on the time-averaged spray angle can be then velocity
presented. It is found that for each fuel the time averaged spray The spray front penetration length of the five fuels under
386 L. Wang et al. / Energy 170 (2019) 375e390

Fig. 11. Comparison of fuel spray before auto-ignition of five fuels at 950 K and 960 kPa ambient conditions with a density of 3.5 kg/m3.

Fig. 12. Time averaged spray angle of five fuels under different ambient temperatures.
 L. Wang et al. / Energy 170 (2019) 375e390 387

Fig. 13. Comparison of time averaged spray angle among five fuels under different ambient temperatures.

Fig. 14. Spray front penetration length of five fuels under different ambient temperatures.
388 L. Wang et al. / Energy 170 (2019) 375e390

different ambient temperatures were calculated and are shown in

Fig. 14. For each fuel, a higher ambient temperature leads to a
shorter penetration length during the entire injection process, and
the penetration length difference at different ambient tempera-
tures increases with time. Three factors may contribute to this
phenomenon. First, as mentioned above, a higher ambient tem-
perature directly relates to a higher ambient pressure. A higher
ambient pressure generates faster momentum exchange, which
could lead to a larger drag force experienced by the moving fuel
droplets. As a result, the spray penetration is more prohibited with
increasing ambient temperature. Secondly, a higher back pressure
reduces the pressure difference between the common rail system
(10 MPa (100 bar) injection pressure in this study) and the in-
chamber environment. Based on Bernoulli's equation, the flow
exit velocity is proportional to the root square of the pressure dif-
ference. Therefore, the flow exit velocity under a higher ambient
temperature/higher ambient pressure condition is slightly lower
than flow velocity under a lower ambient temperature/lower
ambient pressure, which also affects the spray penetration length.
The third and the most important factor is the faster evaporation
for spray at higher ambient temperature, which causes the reduc-
tion in spray penetration. However, it should be noted that even
though the tendency of the spray front penetration length at
different ambient temperatures is clear and consistent for all the
five fuels, the absolute variation in the penetration length at
different ambient temperatures is quite small. In order to better
understand the effect of fuel types on the spray characteristics, the
spray penetration length was compared among the five investi-
gated fuels, as shown in Fig. 15. It turns out that the five fuels have
similar spray front penetration length for the entire injection pro-
cess at each ambient condition. No clear correlation is found on the
spray front penetration length among these fuels. The similar time-
averaged spray angle (as discussed earlier) and spray front pene-
tration length results of different fuels may indicate that the effect
of fuel physical properties may not be the dominant factor in
affecting the spray structure and development, where the injection
momentum and geometric parameters of the injector might play Fig. 15. Comparison of spray front penetration length among five fuels at: (a) 650 K, (b)
950 K.
more important roles. But the difference in fuel properties should
be significant in affecting the microscopic spray structure such as
fuel atomization and droplet size distribution.
The time-averaged spray penetration velocity was calculated conditions. Locations of the occurrence of auto-ignition are
based on the spray front penetration length results, as presented in randomly distributed in the combustion chamber. Occasionally,
Fig. 16. Similar to the time-averaged spray angle result, a lower more than one auto-ignition occurrence locations could take
time-averaged spray penetration velocity is observed at a higher place in an experimental test.
ambient temperature for all the five fuels. However, the effect of the 2. For all the experiments, the injection event ends at 0.43 ms after
ambient temperature weighs differently for different fuels. For trigger signal, and ignition delay ranges from 106% to 2750%
example, the penetration velocity of PRF95 decreases very slow longer than injection event. For each fuel, a longer ignition delay
when the ambient temperature is lower than 750 K, then drops is found under a lower ambient temperature condition. The
much faster when the ambient temperature is higher than 750 K,
ignition delay decreases rapidly with increasing ambient tem-
while the penetration velocity of LN behaves in an opposite perature when the ambient temperature is lower than 750 K,
manner. For the time-averaged spray penetration velocity among
but the ignition delay decreasing rate becomes smaller once the
the five selected fuels, no definite pattern can be found under a ambient temperature exceeds 750 K.
certain ambient temperature.
3. The octane number plays an important role when the ambient
temperature is low. When the ambient temperature is lower
4. Summary than 750 K, PRF95 always has the longest ignition delay, while
LN has the shortest ignition delay which is about 50% of that of
This work investigates the auto-ignition of sprays of different PRF95. Gasoline surrogate has slightly shorter ignition delay
fuels from a piezoelectric fuel injector under different ambient than neat gasoline. Ignition delay of five fuels are very similar
temperatures with a prescribed ambient air density of 3.5 kg/m3. when the ambient temperature exceeds 750 K.
Both the combustion and spray characteristics were measured and 4. Five fuels have similar spray front penetration length develop-
analyzed under auto-ignition conditions. Important findings can be ment and spray angles before the occurrence of auto-ignition
summarized as follows: under all the investigated conditions. The effect of fuel phys-
ical properties may not be the dominant factor in affecting the
1. Fuel auto-ignition was observed with noticeable ignition delay spray structure and development, where the injection
for the five investigated fuels under all the selected experiment
 L. Wang et al. / Energy 170 (2019) 375e390 389

Fig. 16. Comparison of spray front penetration velocity among five fuels (a) and under different temperatures (b).

momentum and geometric parameters of the injector might injection gasoline engine using the MMT fuel additive gasoline. Fuel
2015;144:380e7. https://doi.org/10.1016/j.fuel.2014.12.064.
play more important roles.
[2] Imoehl W, Gestri L, Maragliulo M, Del-Frate L, Klepatsch M, Wildeson R. A DOE
approach to engine deposit testing used to optimize the design of a gasoline
Acknowledgement direct injector seat and orifice. SAE Int J Fuels Lubr 2012;5:1078e95. https://
doi.org/10.4271/2012-01-1642.
[3] Überall A, Otte R, Eilts P, Krahl J. A literature research about particle emissions
This research was supported in part by the Saudi Aramco R&D from engines with direct gasoline injection and the potential to reduce these
Center through the Clean Combustion Research Center of the King emissions. Fuel 2015;147:203e7. https://doi.org/10.1016/j.fuel.2015.01.012.
[4] Zhu R, Hu J, Bao X, He L, Lai Y, Zu L, et al. Tailpipe emissions from gasoline
Abdullah University of Science and Technology under the FUELCOM
direct injection (GDI) and port fuel injection (PFI) vehicles at both low and
program. Any opinions, findings, and conclusions or recommen- high ambient temperatures. Environ Pollut 2016;216:223e34. https://doi.org/
dations expressed in this material are those of the author(s) and do 10.1016/j.envpol.2016.05.066.
[5] Rahimi-Gorji M, Ghajar M, Kakaee A-H, Domiri Ganji D. Modeling of the air
not necessarily reflect the views of the funding agencies.
conditions effects on the power and fuel consumption of the SI engine using
neural networks and regression. J Braz Soc Mech Sci Eng 2017;39:375e84.
References https://doi.org/10.1007/s40430-016-0539-1.
[6] Kalghatgi G, Hildingsson L, Johansson B. Low NOx and low smoke operation of
a diesel engine using gasolinelike fuels. J Eng Gas Turbines Power 2010;132.
[1] Geng P, Zhang H. Combustion and emission characteristics of a direct-
390 L. Wang et al. / Energy 170 (2019) 375e390

https://doi.org/10.1115/1.4000602. 092803-092803-9. fuels in HCCI engines. Warrendale, PA: SAE Technical Paper; 2003. https://
[7] Hildingsson L, Kalghatgi G, Tait N, Johansson B, Harrison A. Fuel octane effects doi.org/10.4271/2003-01-3215.
in the partially premixed combustion regime in compression ignition engines. [31] Kalghatgi GT. Auto-ignition quality of practical fuels and implications for fuel
Warrendale, PA: SAE Technical Paper; 2009. https://doi.org/10.4271/2009-01- requirements of future SI and HCCI engines. Warrendale, PA: SAE Technical
2648. Paper; 2005. https://doi.org/10.4271/2005-01-0239.
[8] Aoyama T, Hattori Y, J’ichi Mizuta, Sato Y. An experimental study on [32] Richter M, Engstro €m J, Franke A, Alde
n M, Hultqvist A, Johansson B. The in-
premixed-charge compression ignition gasoline engine. Warrendale, PA: SAE fluence of charge inhomogeneity on the HCCI combustion process. Warren-
Technical Paper; 1996. https://doi.org/10.4271/960081. dale, PA: SAE Technical Paper; 2000. https://doi.org/10.4271/2000-01-2868.
[9] Hildingsson L, Johansson B, Kalghatgi GT, Harrison AJ. Some effects of fuel [33] Lee C, Ahmed A, Nasir EF, Badra J, Kalghatgi G, Sarathy SM, et al. Autoignition
autoignition quality and volatility in premixed compression ignition engines. characteristics of oxygenated gasolines. Combust Flame 2017;186:114e28.
SAE Int J Eng 2010;3:440e60. https://doi.org/10.4271/2010-01-0607. https://doi.org/10.1016/j.combustflame.2017.07.034.
[10] Sjo€berg M, Dec JE. Influence of fuel autoignition reactivity on the high-load [34] Sarathy SM, Farooq A, Kalghatgi GT. Recent progress in gasoline surrogate
limits of HCCI engines. SAE Int J Eng 2009;1:39e58. fuels. Prog Energy Combust Sci 2018;65:67e108. https://doi.org/10.1016/
[11] Kolodziej C, Kodavasal J, Ciatti S, Som S, Shidore N, Delhom J. Achieving stable j.pecs.2017.09.004.
engine operation of gasoline compression ignition using 87 AKI gasoline [35] Ahmed A, Goteng G, Shankar VSB, Al-Qurashi K, Roberts WL, Sarathy SM.
down to idle. Warrendale, PA: SAE Technical Paper; 2015. https://doi.org/ A computational methodology for formulating gasoline surrogate fuels with
10.4271/2015-01-0832. accurate physical and chemical kinetic properties. Fuel 2015;143:290e300.
[12] Hao H, Liu F, Liu Z, Zhao F. Compression ignition of low-octane gasoline: life https://doi.org/10.1016/j.fuel.2014.11.022.
cycle energy consumption and greenhouse gas emissions. Appl Energy [36] Wu Z, Jing W, Zhang W, Roberts WL, Fang T. Narrow band flame emission
2016;181:391e8. https://doi.org/10.1016/j.apenergy.2016.08.100. from dieseline and diesel spray combustion in a constant volume combustion
[13] Viollet Y, Chang J, Kalghatgi G. Compression ratio and derived cetane number chamber. Fuel 2016;185:829e46. https://doi.org/10.1016/j.fuel.2016.08.022.
effects on gasoline compression ignition engine running with naphtha fuels. [37] Jing W, Wu Z, Roberts WL, Fang T. Spray combustion of biomass-based
SAE Int J Fuels Lubr 2014;7:412e26. https://doi.org/10.4271/2014-01-1301. renewable diesel fuel using multiple injection strategy in a constant volume
[14] Badra JA, Elwardany A, Sim J, Viollet Y, Im H, Chang J. Effects of In-Cylinder combustion chamber. Fuel 2016;181:718e28. https://doi.org/10.1016/
Mixing on Low Octane Gasoline Compression Ignition Combustion. 2016. j.fuel.2016.05.039.
SAE Technical Paper; Warrendale, PA, https://doi.org/10.4271/2016-01-0762. [38] Jing W, Wu Z, Roberts WL, Fang T. Effects of fuel quantity on soot formation
[15] Badra JA, Sim J, Elwardany A, Jaasim M, Viollet Y, Chang J, et al. Numerical process for biomass-based renewable diesel fuel combustion. J Eng Gas Tur-
simulations of hollow-cone injection and gasoline compression ignition bines Power 2017;139. https://doi.org/10.1115/1.4036292. 102803-
combustion with naphtha fuels. J Energy Resour Technol 2016;138. https:// 102803e5.
doi.org/10.1115/1.4032622. 052202e052202. [39] Wang L, Badra JA, Roberts WL, Fang T. Characteristics of spray from a GDI fuel
[16] Zhang Y, Voice A, Tzanetakis T, Traver M, Cleary D. An evaluation of com- injector for naphtha and surrogate fuels. Fuel 2017;190:113e28. https://
bustion and emissions performance with low cetane naphtha fuels in a doi.org/10.1016/j.fuel.2016.11.015.
multicylinder heavy-duty diesel engine. J Eng Gas Turbines Power 2016;138. [40] Engine Combustion Network j “spray G” Operating Condition n.d. https://ecn.
https://doi.org/10.1115/1.4032879. 102805-102805e10. sandia.gov/gasoline-spray-combustion/target-condition/spray-g-operating-
[17] Chang J, Kalghatgi G, Amer A, Viollet Y. Enabling high efficiency direct in- condition/(accessed October 25, 2018).
jection engine with naphtha fuel through partially premixed charge [41] Kim D, Martz J, Violi A. Effects of fuel physical properties on direct injection
compression ignition combustion. Warrendale, PA: SAE Technical Paper; spray and ignition behavior. Fuel 2016;180:481e96. https://doi.org/10.1016/
2012. https://doi.org/10.4271/2012-01-0677. j.fuel.2016.03.085.
[18] Chang J, Viollet Y, Alzubail A, Abdul-Manan AFN, Arfaj AA. Octane-on-demand [42] Javed T, Ahmed A, Lovisotto L, Issayev G, Badra J, Sarathy SM, et al. Ignition
as an enabler for highly efficient spark ignition engines and greenhouse gas studies of two low-octane gasolines. Combust Flame 2017;185:152e9.
emissions improvement. Warrendale, PA: SAE Technical Paper; 2015. https:// https://doi.org/10.1016/j.combustflame.2017.07.006.
doi.org/10.4271/2015-01-1264. [43] Selim H, Mohamed SY, Hansen N, Sarathy SM. Premixed flame chemistry of a
[19] Chang J, Viollet Y, Amer A, Kalghatgi G. Fuel economy potential of partially gasoline primary reference fuel surrogate. Combust Flame 2017;179:300e11.
premixed compression ignition (ppci) combustion with naphtha fuel. War- https://doi.org/10.1016/j.combustflame.2017.02.008.
rendale, PA: SAE Technical Paper; 2013. https://doi.org/10.4271/2013-01- [44] Chen B, Togbe  C, Selim H, Dagaut P, Sarathy SM. Quantities of interest in jet
2701. stirred reactor oxidation of a high-octane gasoline. Energy Fuels 2017;31:
[20] Viollet Y, Abdullah M, Alhajhouje A, Chang J. Characterization of high effi- 5543e53. https://doi.org/10.1021/acs.energyfuels.6b03193.
ciency octane-on-demand fuels requirement in a modern spark ignition en- [45] Ahmed A, Waqas M, Naser N, Singh E, Roberts W, Chung S, et al. Composi-
gine with dual injection system. Warrendale, PA: SAE Technical Paper; 2015. tional effects of gasoline fuels on combustion, performance and emissions in
https://doi.org/10.4271/2015-01-1265. engine. SAE Int J Fuels Lubr 2016;9:460e8. https://doi.org/10.4271/2016-01-
[21] Wang B, Wang Z, Shuai S, Yang H, Wang J. Combustion and emission char- 2166.
acteristics of Multiple Premixed Compression Ignition (MPCI) fuelled with [46] Chen B, Togbe  C, Wang Z, Dagaut P, Sarathy SM. Jet-stirred reactor oxidation
naphtha and gasoline in wide load range. Energy Convers Manag 2014;88: of alkane-rich FACE gasoline fuels. Proc Combust Inst 2017;36:517e24.
79e87. https://doi.org/10.1016/j.enconman.2014.08.045. https://doi.org/10.1016/j.proci.2016.05.040.
[22] Wang B, Wang Z, Shuai S, Xu H. Combustion and emission characteristics of [47] Sarathy MS, Kukkadapu G, Mehl M, Javed T, Ahmed A, Naser N, et al.
Multiple Premixed Compression Ignition (MPCI) mode fuelled with different Compositional effects on the ignition of FACE gasolines, Combust Flame 169,
low octane gasolines. Appl Energy 2015;160:769e76. https://doi.org/10.1016/ 2016, 171-193,. Combust Flame 2016;169:171e93. https://doi.org/10.1016/
j.apenergy.2015.01.115. j.proci.2016.05.040.
[23] Manente V, Johansson B, Tunestal P, Cannella W. Effects of different type of [48] Nadkarni R. Guide to ASTM test methods for the analysis of petroleum
gasoline fuels on heavy duty partially premixed combustion. SAE Int J Eng products and lubricants, 2nd edition. 2ND ed. ASTM International; 2007.
2009;2:71e88. https://doi.org/10.4271/2009-01-2668. https://doi.org/10.1520/MNL44-2ND-EB.
[24] Manente V, Zander C-G, Johansson B, Tunestal P, Cannella W. An advanced [49] Javed T, Nasir EF, Ahmed A, Badra J, Djebbi K, Beshir M, et al. Ignition delay
internal combustion engine concept for low emissions and high efficiency measurements of light naphtha: a fully blended low octane fuel. Proc Combust
from idle to max load using gasoline partially premixed combustion. War- Inst 2017;36:315e22. https://doi.org/10.1016/j.proci.2016.05.043.
rendale, PA: SAE Technical Paper; 2010. https://doi.org/10.4271/2010-01- [50] Dec JEA. Conceptual model of DI diesel combustion based on laser-sheet
2198. imaging*. 1997. https://doi.org/10.4271/970873.
[25] Chang J, Kalghatgi G, Amer A, Adomeit P, Rohs H, Heuser B. Vehicle demon- [51] Atef N, Badra J, Jaasim M, Im HG, Sarathy SM. Numerical investigation of
stration of naphtha fuel achieving both high efficiency and drivability with injector geometry effects on fuel stratification in a GCI engine. Fuel 2018;214:
euro6 engine-out nox emission. SAE Int J Eng 2013;6:101e19. https://doi.org/ 580e9. https://doi.org/10.1016/j.fuel.2017.11.036.
10.4271/2013-01-0267. [52] Badra J, Viollet Y, Elwardany A, Im HG, Chang J. Physical and chemical effects
[26] Won HW, Bouet A, Duffour F, Francqueville L. Naphtha fuel on a light duty of low octane gasoline fuels on compression ignition combustion. Appl Energy
single cylinder compression ignition engine with two different compression 2016;183:1197e208. https://doi.org/10.1016/j.apenergy.2016.09.060.
ratios. Warrendale, PA: SAE Technical Paper; 2016. https://doi.org/10.4271/ [53] Badra J, Bakor R, AlRamadan A, Almansour M, Sim J, Ahmed A, et al. Stan-
2016-01-2302. dardized gasoline compression ignition fuels matrix. Warrendale, PA: SAE
[27] Zhang Y, Kumar P, Traver M, Cleary D. Conventional and low temperature Technical Paper; 2018. https://doi.org/10.4271/2018-01-0925.
combustion using naphtha fuels in a multi-cylinder heavy-duty diesel engine. [54] Naser N, Jaasim M, Atef N, Chung SH, Im HG, Sarathy SM. On the effects of fuel
SAE Int J Eng 2016;9:1021e35. https://doi.org/10.4271/2016-01-0764. properties and injection timing in partially premixed compression ignition of
[28] Kalghatgi G, Gosling C, Wier M. The outlook for transport fuels: Part 1. Pe- low octane fuels. Fuel 2017;207:373e88. https://doi.org/10.1016/
troleum Technology Quarterly; 2016. j.fuel.2017.06.048.
[29] Wang M, Lee H, Molburg J. Allocation of energy use in petroleum refineries to [55] Hung DLS, Harrington DL, Gandhi AH, Markle LE, Parrish SE, Shakal JS, et al.
petroleum products. Int J LCA 2004;9:34e44. https://doi.org/10.1007/ Gasoline fuel injector spray measurement and characterization - a new SAE
BF02978534. J2715 recommended practice. SAE Int J Fuels Lubr 2008;1:534e48.
[30] Risberg P, Kalghatgi G, Ångstrom H-E. Auto-ignition quality of gasoline-like