Energy Conversion and Management 164 (2018) 70–82

Contents lists available at ScienceDirect

Energy Conversion and Management

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

Performance and emissions of a compression-ignition direct-injected natural T

gas engine with shielded glow plug ignition assist
⁎
Ivan M. Gogolev , James S. Wallace
University of Toronto, 5 King’s College Road, Toronto, ON M5S 3G8, Canada

A R T I C L E I N F O A B S T R A C T

Keywords: Performance and emissions experiments were conducted on a compression-ignition direct-injected natural gas
Natural gas engine (DING) equipped with a shielded glow plug ignition assist system. Tests were conducted at three different
Combustion intake pressures (34.5 kPag, 68.9 kPag, 103.4 kPag), and four nominal (targeted) equivalence ratios (0.2, 0.3,
Emissions 0.4, 0.5). CH4, NOx, CO and PM emissions were measured and analyzed, showing that emissions levels are
Particulate matter
influenced by the DING engine’s combustion modes and intake pressure. Premixed combustion which dominates
Direct-injection
Compression-ignition
at low equivalence ratios resulted in higher levels of CH4 and NOx emissions, while mixing-controlled com-
bustion, which develops at higher equivalence ratios, resulted in elevated CO and PM levels. Higher intake
pressure was found to improve all emissions levels. The most significant effect was the reduction of PM and CO
emissions due to improved fuel charge mixing and air entrainment that results from a pressure-driven mo-
mentum increase of the engine’s air swirl field. Brake specific emissions and fuel consumption were estimated
and compared against levels reported in the literature for dual-fuel port-injection, and High-Pressure Direct-
Injection (HPDI) natural gas engines. The most significant finding was that the DING engine exhibits lower fuel
consumption and PM emissions levels when compared to values reported in the literature for HPDI engines. The
PM emissions advantage was attributed to a higher proportion of premixed combustion and the absence of a
diesel pilot in DING engine operation. Lastly, PM size distributions were analyzed, showing that the DING engine
produces PM that is smaller than PM of a conventional diesel engine, but similar to the PM reported in the
literature for HPDI engines.

1. Introduction
Natural gas has several advantages for use in automotive engines. It
is relatively inexpensive, widely available, has a well-established dis-
tribution infrastructure, and is thought to be a cleaner burning fuel
when compared to gasoline and diesel in terms of both greenhouse gas
(GHG) and regulated emissions. The transportation sector is a tre-
mendous source of GHG emissions. In the US, for example, transpor-
tation consumes 29% of primary energy, with 92% of that coming from
petroleum [1]. Much research has been focused on substituting biomass
derived fuels to make the sector’s energy consumption more sustainable
[2,3]. The use of natural gas in place of gasoline or diesel fuel produces
approximately 30% less CO2 emissions when compared on the same
energy content basis [4]. Although natural gas is sometimes viewed as
an interim solution to reducing GHG emissions – substituting a fossil
fuel with lower CO2 emission for another fossil fuel – the use of
renewable natural gas (RNG), which can be produced from biomass
sources by anaerobic digestion or by combining gasification and cata-
lytic processes, can provide long-term sustainable GHG reductions.1 A
study of the resource base of Canadian biomass waste estimated total
greenhouse gas reduction of 107 Mt CO2 eq/year for Canada, with
potential of replacing 130% of current residential and commercial use
of natural gas [5]. In the future, RNG may also be produced from hy-
drogen and carbon dioxide, where the hydrogen is generated by elec-
trolysis using electricity from renewable sources [6].
Natural gas use in spark ignition (SI) engines for heavy duty and
passenger vehicles is already quite established, with approximately 23
million natural gas vehicles on the road globally in 2016 [7]. SI natural
gas engines have favorable emissions performance, having been shown
to emit lower levels of carbon monoxide (CO), non-methane hydro-
carbons (NMHC), and particulate matter (PM) emissions when com-
pared to engines fueled by gasoline and diesel [8,9]. However, natural

⁎
Corresponding author.
E-mail addresses: imgogolev@gmail.com (I.M. Gogolev), wallace@mie.utoronto.ca (J.S. Wallace).
1
A wide range of gaseous fuels, collectively called biogas, can be produced from biomass. RNG refers specifically to a higher quality (high methane content) biogas that has the same
volumetric energy density as natural gas and is effectively a drop-in replacement for natural gas. Biogas containing large amounts of inert gases (CO2, N2) is not suitable for high pressure
direct injection due to increased compression work requirements, compared to natural gas or RNG, and the need for larger fuel injectors.

https://doi.org/10.1016/j.enconman.2018.02.071
Received 11 October 2017; Received in revised form 17 February 2018; Accepted 19 February 2018
Available online 05 March 2018
0196-8904/ © 2018 Elsevier Ltd. All rights reserved.
I.M. Gogolev, J.S. Wallace Energy Conversion and Management 164 (2018) 70–82

Nomenclature EPA environmental protection agency

FTIR Fourier transform infrared spectroscopy
ɸa actual equivalence ratio or equivalence ratio (determined FTP federal test procedure
through test results analysis) GHG greenhouse gas
ɸn nominal equivalence ratio – targeted test equivalence ratio HEPA high-efficiency particulate air
P1 34.5 kPag intake pressure test condition HPDI high pressure direct injection
P2 68.9 kPag intake pressure test condition IMEP indicated mean effective pressure
P3 103.4 kPag intake pressure test condition LNG liquefied natural gas
BMEP brake mean effective pressure NMHC non-methane hydrocarbons
BSFC brake specific fuel consumption OEM original equipment manufacturer
CAD crank angle degrees RNG renewable natural gas
CI compression ignition RPM revolutions per minute
CNG compressed natural gas SCR selective catalytic reduction
DING direct-injected natural gas SI spark ignition
DPF diesel particulate filter THC total hydrocarbon
EEPS engine exhaust particle sizer ULSD ultra-low-sulfur diesel
EGR exhaust gas recirculation

gas SI engines suffer from lower power output compared to their ga-
soline counterparts due to knock limitations [8,9]. A recent experi-
mental study examined the potential for improved efficiency and re-
duced CO2 emissions that could be achieved by downsizing and
optimizing a stoichiometric turbocharged SI engine for natural gas
operation [10]. Through a combination of downsizing, higher com-
pression ratio, EGR use and careful optimization, this study suggests
that reductions in specific CO2 emissions (g/kWh) on the order of
25–34% could be achieved at part load conditions and 45% at full load,
with full load equaling the full load achieved on gasoline [10]. Since
natural gas typically has an octane rating lower than that of the pure
methane used in this study, smaller reductions would be expected in
practice. Other recent studies of direct injection SI natural gas engines
have shown that use of partially stratified combustion, achieved by
control of fuel injection timing, can reduce fuel consumption and im-
prove the trade-off between knock and combustion stability [11,12].
Like gasoline engines, exhaust aftertreatment is needed for SI nat-
ural gas engines to meet emissions regulations. Use of a 3-way catalyst
together with stoichiometric engine operation is a proven and effective
technology. An example of this approach is the Cummins Westport ISL
G engine that has been certified to heavy-duty engine emissions stan-
dards. Table 1 compares the emissions of this engine operating on
natural gas with emissions of the Cumming LS 9 diesel engine from
which it is derived. Measured over the federal test procedure (FTP)
transient test cycle, the natural gas engine has comparable NOx
emissions (0.13 vs 0.14 g/bhp-hr) and much lower CO2 emissions (463
vs 564 g/bhp-hr, an 18% reduction). However, the emissions of CO and
methane are higher. A newer version of this engine, the ISL G Near
Zero, has been released to provide an engine offering that meets the
new EPA/CARB Near Zero NOx emissions standard. This newer version
has achieved emissions of 0.02 g/bhp-hr NOx, which is below the Near
Zero NOx standard. This newer version also meets the 2017 environ-
mental protection agency (EPA) greenhouse gas emission requirements
[13].
Utilization of natural gas in compression-ignition (CI) engines can
potentially achieve the high efficiencies of diesel engines, but with
improved PM emissions performance [8,9]. However, achieving igni-
tion in natural gas CI engines is more challenging than in diesel engines.
Natural gas is highly resistant to compression ignition and requires
some form of ignition assist [8,9,16–19]. The primary solution to this
issue has been to use a two-fuel system. Here, a diesel pilot injection is
used to achieve controlled compression ignition, while natural gas
provides most of the combustion heat release. There are two distinct
approaches to a two-fuel system. The most common, often termed
“dual-fuel”, uses throttle body or port injection of natural gas which is
essentially premixed with the intake air, and is ignited by a late-cycle
direct-injected diesel pilot. The diesel pilot burns in a diffusion flame,
igniting the premixed air/natural gas charge. Upon ignition, natural gas
burns in a premixed flame. This engine configuration generally pro-
duces lower CO and PM emissions, but higher levels of NOx and

Table 1
Comparison of natural gas engine technologies. All results from heavy-duty FTP transient test.

Stoichiometric SI NG comparisona HPDI NG comparisonb Dual-fuel NG comparisonc

Engine model Units Cummins-westport ISLG Cummins L9 330 Westport GX 475 Cummins ISX15 Retrofit NG fumigation module Mack MP8-505C
Fuel CNG ULSD LNG ULSD CNG + ULSD ULSD
Technology SI Natural gas Diesel HPDI Natural gas Diesel Dual-fuel NG Diesel
Aftertreatment 3-way catalyst DPF, SCR DPF, SCR DPF, SCR DPF, SCR DPF, SCR
Max. power hp 320 310 486 583 – 505
Max. torque ft-lbs 1000 1100 1750 2050 – 1810
NOx (g/bhp-hr) 0.13 0.14 0.11 0.16 0.08 0.14
PM (g/bhp-hr) 0 0 0.002 0.001 N/A N/A
NMHC (g/bhp-hr) 0.04 0 0.01 0.02 1.0 BDLd
Methane (g/bhp-hr) 1.97 0.02 No data No data 4.94 BDL
CO (g/bhp-hr) 7.1 0.1 0.04 0 3.96 0.23
CO2 (g/bhp-hr) 463 554 466 614 493.7 520.8
Brake thermal efficiency % N/A N/A N/A N/A 33.5 35.1

Table notes:
a
Data for ISL-G from 2016 certification data [14]; Data for L9 330 from 2017 certification data [14].
b
Data for GX475 and ISX15 from 2010 certification data [14] as the GX475 is no longer in production.
c
Data for dual-fuel comparison from Besch et al. [15].
d
Below detection limit.

71
I.M. Gogolev, J.S. Wallace
hydrocarbon emissions, especially methane, when compared to con-
ventional diesel engines [8,9]. To optimize emissions performance,
recent research on dual-fuel engines has explored the boundaries of
natural gas and diesel pilot proportions and their impact on emissions
[20,21]. As well, a number of researchers have investigated the addi-
tion of hydrogen to the natural gas to improve combustion character-
istics [22,23]. With respect to performance, natural gas dual-fuel en-
gines can achieve efficiencies closer to those of diesel engines due to the
ability to operate at high compression ratios. However, just like in port-
injected SI natural gas engines, natural gas dual-fuel engines suffer from
a significantly lower power output due to knock limitations, when
compared to diesel engines [8,9]. A key weakness of dual-fuel engines is
methane emissions [24]. Königsson et al. [24] have studied the con-
tribution to methane emissions of combustion chamber crevices and
quenching and found that at lean conditions > 70% of hydrocarbon
emissions could be attributed to entrapment of the premixed natural
gas charge in crevices. While some reduction could be achieved by
minimizing top land crevice volume, which is the largest crevice, me-
thane emissions make it challenging to meet the heavy-duty engine
emission limit of 0.1 g/bhp-hr (0.134 g/kWh) of the US GHG emission
regulations [25]. Besch et al. [15] fitted a heavy-duty engine with a
retrofit dual-fuel natural gas module that fumigates a controlled natural
gas flow upstream of the engine’s compressor. The engine was equipped
with a diesel particulate filter (DPF) and a selective catalytic reduction
(SCR) system. The retrofit module controlled only the natural gas flow,
and otherwise relied on the original equipment manufacturer (OEM)
engine control module for all other functions. Thus, dual-fuel operation
was not optimized. Table 1 contains the specifications for the base
diesel engine and the emissions over the FTP transient test for both the
base diesel engine and the same engine operating in dual-fuel mode
with the retrofit natural gas system. Operating in dual-fuel mode, the
engine had significantly lower NOx emissions (0.08 vs 0.14 g/bhp-hr)
and modestly lower CO2 emissions (493.7 vs. 520.8 g/bhp-hr). How-
ever, the methane emissions were very high, 4.94 g/bhp-hr vs. below
the detection limit for the base diesel engine, which would negate any
GHG benefit from the reduced CO2 emissions [15]. As dual-fuel op-
eration in that study was not optimized, application of knowledge
gained in more fundamental work [20–23] should help reduce emis-
sions and increase efficiency.
An alternative design that avoids the knock limitation and con-
sequent power penalty of homogeneous charge SI and dual-fuel engines
has been developed at the University of British Columbia and com-
mercialized by Westport Fuel Systems Inc. in collaboration with
Cummins [26–32]. The Westport High-Pressure Direct-Injection (HPDI)
system injects both fuels directly into the combustion chamber near the
end of the compression stroke from a custom two-fuel high-pressure
injector. The diesel pilot is injected first and ignites spontaneously from
compression after an initial ignition delay period. The natural gas is
injected shortly after and is ignited by the diesel pilot’s diffusion flame.
Since the natural gas is direct-injected, the HPDI system is not knock
limited and achieves diesel-like efficiencies due the ability to operate
high compression ratios. Also, power output levels are not sacrificed
since direct-injected natural gas does not displace the inducted air
charge [8,9,26–32]. Although the HPDI system generally performs
better than comparable diesel engines, it is still challenged with PM
emissions formed in the diesel pilot and natural gas diffusion flames, as
well as with NOx emissions [27,28,30,33,34]. Table 1 compares the
certification emissions measured during the FTP heavy-duty engine
transient test of the HPDI engine with the ISX15 diesel engine from
which it was derived. The full aftertreatment system of the diesel en-
gine (DPF, SCR) is retained for the HPDI engine. As a result of em-
ploying a DPF, the two engines have negligible PM emissions. The HPDI
engine has significantly lower NOx emissions (0.11 vs 0.16 g/bhp-hr)
compared to the ISX15 diesel, noting again that both have SCR after-
treatment. Finally, the HPDI engine has lower CO2 emissions (466 vs
614 g/bhp-hr, a 24% reduction) compared to the ISX15 diesel.
72
Energy Conversion and Management 164 (2018) 70–82
Another approach to enabling ignition in direct-injected natural gas
(DING) engines is to use a hot surface, such as a glow plug, to assist
ignition. If implemented successfully in DING engines, hot surface ig-
nition assist can offer an elegant solution for achieving effective natural
gas ignition without the need for a complex and costly two-fuel injector,
and the need to store two fuels onboard the vehicle. Glow plug ignition
has been previously explored for various fuels such as methanol
[35–39], propane [40], as well as natural gas [17,18,41,42]. In DING
applications, Aesoy and Valland found that the hot surface (or glow
plug) must reach temperatures of 1400 K to achieve timely ignition
[18]. With such high temperature requirements, the few early im-
plementations of DING engines with glow plug ignition assist systems
had limited success, largely due to glow plug reliability and longevity
issues [43]. However, glow plugs have shown tremendous improve-
ment in recent years. Early experiments utilized metallic sheathed glow
plugs (e.g. Denso DG-143) that required 55 W of power to achieve
1300 K temperature [44]. Recent experiments have utilized ceramic
glow plugs (e.g. Bosch GLP4) that required only 28 W of power to
achieve 1500 K temperature [44]. This improved performance sug-
gested that expected continued improvements in materials and glow
plug technology warranted revisiting the potential for using glow plugs
to assist ignition in DING engines.
Ignition by glow plug is a distinctly different process than the
normal compression ignition process. The experimental study of glow-
plug assisted ignition and combustion of 100% methanol carried out by
Mueller and Musculus [45] in an optically accessible 4 S heavy-duty
diesel engine provides a clear understanding of the ignition sequence.
Their engine employed a centrally located fuel injector with six injec-
tion holes, with the glow plug (unshielded) placed between two of the
jets. Their experiments showed that ignition occurs at the glow plug
and combustion propagates to the two surrounding fuel jets, continues
in both circumferential directions to the next two fuel jets, and finally
to the two farthest fuel jets. They also found that ignition was highly
sensitive to glow plug temperature and to its proximity to a fuel jet.
The Engine Research and Development Laboratory at the University
of Toronto has been developing a hot-surface natural gas ignition assist
technology based on a shielded glow plug [46,47]. Modelling work has
focused on understanding the mixing characteristics of the injected
natural gas jet, while experimental work conducted in a fixed-volume
combustion apparatus focused on glow plug selection and glow plug
shield design, as well as optimization of the fuel injector and fuel in-
jection pattern. Recently, the research program achieved the first suc-
cessful implementation of this technology in a full-featured DING en-
gine environment in a study focusing on optimizing the ignition assist
system geometry and on understanding the effects of equivalence ratio
and intake pressure on the ignition and combustion performance [48].
The DING engine was successfully operated at intake pressures of
34.5–103.4 kPa and indicated mean effective pressures (IMEP) of
3.4–10.3 bar. For comparison, the Westport HPDI engine operated at
brake mean effective pressure (BMEP) values of 18–20 bar at peak
power and peak torque (calculated from data in [14]). Combustion in
the DING engine was found to proceed in two distinct modes. At lower
nominal (targeted) equivalence ratio (ɸn) values of 0.2 and 0.3, ma-
jority of the combustion event occurred in a premixed fashion. At the
higher ɸn values of 0.4 and 0.5, the premixed combustion phase was
followed by a mixing-controlled combustion phase. For a given
equivalence ratio, operating at higher intake pressure increased the
proportion of the mixing-controlled combustion phase due to a longer
fuel injection duration [48]. Since these combustion characterization
experiments were conducted at conditions used in the present study,
the following discussion makes use of the combustion mode findings to
explain the observed performance and emissions trends.
The study presented in this paper is a follow-up study with the
objective to understand the emissions and performance characteristics
and processes of the DING engine. NOx and PM emissions are the main
focal points of this study, since reducing these two emission types
I.M. Gogolev, J.S. Wallace
comprise one of the most pressing issues in compression-ignition engine
operation. Performance and emissions experiments were conducted on
a DING engine equipped with a shielded glow plug ignition assist
system. The unique oil-less design of the experimental engine allowed
for measurement of PM and gaseous emissions resulting from non-
premixed combustion of natural gas. To the best of the authors’
knowledge, the PM data presented in this study is the first data set to
definitively describe direct-injected compression-ignited natural gas
sourced PM. Emissions results obtained from the DING engine are
compared to emissions reported in the literature for dual-fuel and HPDI
engines to highlight the potential advantages and shortfalls of the
shielded glow plug ignition-assisted DING engine.
2. Experimental setup
The DING engine used in this study is based on a crankcase from a
single-cylinder Ricardo Hydra research engine. The original Hydra en-
gine was modified to accommodate a custom built high-pressure nat-
ural gas injector [49] and a shielded glow plug ignition assist system to
enable pilot-free DING operation. The DING engine was also modified
for optical access (to allow future optical studies) by fitting the engine
with an elongated Bowditch piston housed in an extended cylinder
block. The Bowditch piston assembly was fitted with custom oil-less
polyamide piston rings (i.e. no splash lubrication of the piston rings).
This limits the engine’s safe operating speed, but since lubricating oil is
eliminated, enables the study of PM emissions that originate solely from
compression-ignition combustion of natural gas (i.e. PM that may form
from lubricating oil is excluded). The completed DING engine specifi-
cation are detailed in Table 2.
The original Ricardo Hydra test bed system was used to control the
engine coolant and oil temperatures. The engine head temperature was
regulated with a programmable controller that was connected to a re-
sistance heater and a cooling fan installed on the cylinder head. An
integrated Cussons DC dynamometer was used to control the engine
speed. Intake air was supplied from a building air compressor connec-
tion (700 kPa) and was set to the desired operating pressure using a
pressure regulator. With the intake pressure set, the air flow rate was
controlled by the engine speed. Since most of the air moisture was
knocked out by the air compression system, the air supply was dry and
had a constant moisture content with a dew point of 3.3 °C. A LabVIEW-
based data acquisition/control system was used to control fuel injection
(by commanding injector opening and closing) and to acquire key en-
gine parameters such as in-cylinder pressure (Kistler 6121 transducer),
crank angle degree (CAD) position at 0.2 CAD resolution (AVL optical
crankshaft encoder), and air flow (measured by a laminar flow meter).
A more detailed description of the DING engine (including combustion
chamber geometry), the dynamometer system, and process data ac-
quisition system can be found in Ref. [48].
To enable emissions sampling, the DING engine was built with a
custom exhaust sampling system shown in Fig. 1. Emissions samples
were drawn from a sampling manifold comprised of a 50.8 mm dia-
meter, 914.4 mm long stainless-steel tube equipped with multiple
sampling ports. To avoid condensation of the exhaust species on the
manifold wall, the entire length of the exhaust sampling manifold was
heat traced and insulated. Gaseous exhaust emissions were measured
using an MKS 2030 Fourier Transform Infrared Spectroscopy (FTIR)
analyzer. The sample was drawn from the exhaust sampling manifold
and routed to the FTIR analyzer through a heated filter (to remove
particulates) via a heated line. Both the heated filter and heated line
were maintained at 191 °C to prevent sample condensation. The FTIR
analyzer sampled the emissions once every second during each engine
test.
Exhaust gas for particulate matter (PM) analysis was sampled with a
TSI 379020A thermodiluter head connected to sampling port on the
exhaust sampling manifold. The diluter head uses a 10-cavity rotating
disk to withdraw a metered amount of exhaust gas and dilute it with
73
Energy Conversion and Management 164 (2018) 70–82
preheated, HEPA-filtered dry air supplied from the TSI 379020A diluter
control unit. Dilution of the raw exhaust sample is required for pre-
venting excessively high sample temperatures which can damage the
analyzer, as well as to keep the sample PM concentration within the
measurement limits of the Engine Exhaust Particle Sizer (EEPS). The
dilution ratio is controlled by setting the dilution air flow rate and the
rotational speed of the rotating thermodiluter disk. The diluted sample
is routed to the diluter control unit and then to a TSI 379030 thermal
conditioner, where the sample is heated to 300 °C and subjected to a
secondary dilution. Heating is required to evaporate volatile com-
pounds, which if condensed could erroneously register as solid particles
by the EEPS analyzer. From the thermal conditioner, the sample is
routed to a TSI 3090 EEPS analyzer. The EEPS records one-second
averages of the particle concentration as well as the particle size dis-
tribution for particles in the 5.6–560 nm mobility diameter range. The
exact total dilution ratio is calculated from measurements of CO2 con-
centration before and after dilution (CO2 is used as a tracer gas, mea-
sured by an independent analyzers). The calculated total dilution ratio
is used to correct the measured PM concentrations to reflect pre-dilu-
tion (raw exhaust) levels. The resultant PM data was also corrected
using a protocol developed by Zimmerman [50] which minimizes errors
attributed to soot morphology effects on the EEPS PM size and con-
centration measurement.
3. Experimental procedure and calculations
The experimental work in this study builds on the skip-fired DING
engine tests conducted to investigate the DING engine’s ignition and
heat release profiles [48]. In the present study, the DING engine was
operated in a continuously fired mode using the same ignition assist
system geometry settings and the same 12 operating conditions as
tested in the ignition study: intake pressures of 34.5 kPag (P1 condi-
tion), 68.9 kPag (P2 condition), and 103.4 kPag (P3 condition); and
four nominal (targeted) equivalence ratios (φn) of 0.2, 0.3, 0.4, and 0.5
[48]. Test results analyses confirmed that the actual equivalence ratio
(φa) values achieved in the experiments varied depending on test
conditions and thus differed from the targeted nominal values. The
plots presented in Section 4 (Results) show data plotted vs. actual
equivalence ratios (φa). For generalizing the observations, trends are
discussed with respect to the nominal equivalence ratio (φn) values.
DING engine operating conditions are summarized in Table 3.
For each of the 12 operating conditions, the DING engine was
warmed up to the operating conditions summarized in Table 3 and then
operated in continuously fired operation. Each test duration was limited
to 6 min to avoid breakage of the oil-less polyamide piston rings used in
the Bowditch piston assembly. Each test was repeated three times on
different test days.
Gaseous emissions, measured by the FTIR analyzer, were found to
change over the test duration. Methane emissions showed a gradual
decrease to steady levels, NOx emissions showed a gradual rise to steady
levels, and CO emissions remained relatively steady over the test
duration. To obtain the best estimate of steady state values, gaseous
emissions measurements were averaged over the last 30 s of each test.
Section 4 (Results) presents plots of gaseous emissions measurement on
ppm and estimated brake specific basis. Brake specific emissions were
Table 2
DING engine specifications.
Bore 935.3 mm
Stroke 88.9 mm
Type 4-stroke, single cylinder, boosted intake
displacement 633.46 cm3
Compression ratio 13.17–1
Piston bowl design Cylindrical bowl 50.9 mm diameter, 13.59 mm deep
Injector 9-hole, 0.2 mm orifice size
Fuel injection path In-plane, 6.22 mm offset depth from head deck
I.M. Gogolev, J.S. Wallace Energy Conversion and Management 164 (2018) 70–82

Fig. 1. DING engine emissions sampling system.

Table 3 engine load conditions yield similar effective PM particle density de-
DING engine test conditions. pendency on particle size for the particles lower than approximately
50 nm in diameter. As shown later in Fig. 8 of Section 4.5, the majority
Parameter Parameter set point Actual test range or control
tolerance
of the PM particles produced by the DING engine are well within this
range, so adoption of the density coefficients from the high load, 0%
Intake pressure 34.5, 68.9, ± 1% EGR HPDI condition is justified. The brake specific PM emissions cal-
103.4 kPag culation methodology presented here was implemented due to the lack
Intake temperature 25 °C 25–29 °C
Engine speed 1000 RPM 1000–1100 RPM
of availability of direct mass measurements and is only used to facilitate
Head temperature 120 °C 131–224 °C high-level discussion of DING PM emissions levels.
Coolant temperature 60 °C 50–80 °C
Oil temperature 44 °C 40–50 °C
Start of injection 335 CAD 335 CAD 4. Results and discussion
Natural gas pressure 12.4 MPag ± 0.5%
All results presented are engine-out emissions; no aftertreatment
was fitted to this single-cylinder research engine. Tables A1 and A2 in
calculated by multiplying the specie ppm concentration by the engine’s Appendix A present measured emissions values from representative
exhaust total molar flow rate (determined from air & fuel flow mea- literature, also engine out measurements, and summarize the current
surement and stoichiometry) and by each specie’s molar mass. The results for comparison. The following discussion of results makes use of
resultant figures were divided by the engine’s indicated power output the combustion modes identified during an earlier study of the DING
(determined from indicated torque and engine speed) and by an as- engine’s ignition and combustion characteristics and summarized in the
sumed mechanical efficiency of 90% [51] to convert to estimated brake introduction [48]. The figures presented in the following discussion
power basis. show emissions and fuel consumption values averaged across all three
PM emissions were measured with the EEPS analyzer which pro- test days. The error bars in the figures represent ± 2 standard devia-
vides a particle count concentration for 32 bin sizes that correspond to tions between each day’s parameter averages, thus denoting day-to-day
different particle mobility diameters. Due to high measurement varia- variability. Each emissions figure presents the measured DING engine
bility in PM emissions, measurements were averaged over the last 60 s emissions and includes a secondary scale that corresponds to the cal-
of each test. Total particle number concentrations were calculated by culated brake specific emissions values.
summing the concentrations for each of the bin sizes.
Brake specific PM mass emissions were estimated by multiplying the
4.1. Methane (CH4) emissions
particle count concentration within each EEPS particle size bin by the
particles’ respective volumes (spherical volume calculated from the
Methane is the primary unburnt hydrocarbon emitted by natural gas
mobility diameter) and then by the particles’ effective density, which
engines [8,9]. The combined methane emissions results for all three test
accounts for the particle morphology. The resultant PM mass con-
days are presented in Fig. 2.
centrations were multiplied by the exhaust flow rate and divided by the
Methane emissions are in the range of 120–1550 ppm-wet or
engine’s indicated power output and an assumed mechanical efficiency
0.39–8.79 g/bhp-hr. Considering that the DING engine has no diesel
of 90% [51] to convert to estimated brake power basis. The effective
pilot fuel and used a natural gas fuel with > 94% methane content, for
densities for the calculation of brake specific PM emissions were
comparison purposes, these DING methane emissions results can be
adopted from the regression of PM particle effective density vs. particle
considered equivalent to unburnt total hydrocarbon (THC) emissions.
size presented in a HPDI engine PM morphology study by Graves et al.
Compared to THC data in the literature (Appendix A), the DING me-
[34]. Since the majority of HPDI PM particles originate from mixing-
thane emissions levels at the two lower equivalence ratios are slightly
controlled combustion of natural gas [29,32], the DING engine is ex-
lower than THC levels in Table A2 emitted by dual-fuel natural gas
pected to have similar PM morphology and effective density depen-
engines (650–1550 ppm vs. 1700–6500 ppm) [52–55], and higher than
dence on particle size. Specifically, the effective particle density as-
THC levels in Table A1 reported for HPDI engines (2.65–8.79 g/bhp-hr
sumed in the calculation of brake specific PM emissions uses the
vs. 0.2–1.7 g/bhp-hr) [30,33,56]. At the higher two equivalence ratios
correlation and regression coefficients for particles produced at high
the DING methane emissions (0.39–2.37 g/bhp-hr) decline to levels
load, 0% EGR HPDI engine conditions [34]. Although this HPDI engine
comparable to THC emissions of HPDI engines (0.2–1.7 g/bhp-hr)
condition does not directly correspond to the range of loads tested on
[28,30,33,56].
the DING engine, the study by Graves et at. showed that different HPDI
The advantage of direct injection engines (DING and HPDI) is that

74
I.M. Gogolev, J.S. Wallace
Fig. 2. DING engine methane emissions.
crevice entrapment and quenching of the fuel/air mixture, which is the
main cause of methane emissions in dual-duel engines, is much less
prevalent. In direct injection engines, the primary cause of methane
emissions is the overmixing of the natural gas jet within the ignition
delay time frame. Overmixing forms over-lean elements of fuel/air
mixture that are below the lower flammability limit of methane. Some
of the over-lean fuel/air elements survive the combustion event, pro-
ducing methane emissions. The ignition delay period defines the
amount of premixing, and thus the extent of over-leaning and methane
emissions. For the DING engine, this relationship is corroborated since
the high variability of methane emissions at ɸn of 0.2 coincides with the
high variability of ignition delay values measured for this engine con-
dition in the previous ignition study [48]. Since the natural gas fuel in
the DING engine has a significantly longer ignition delay than the diesel
pilot in HPDI engines (ignition delay of approx. 1.7–1.9 ms [48] for
DING vs. approx. 0.6–0.8 ms for HPDI [57]), it is not surprising that at
the lower equivalence ratios, where premixed combustion dominates,
the DING engine exhibits higher methane emissions. At the lower
equivalence ratios, the surviving methane in not extensively oxidized
due to low in-cylinder temperatures in the late stages of the expansion
process. This is because the dominant premixed combustion phase is
short-lived and is followed by rapid charge cooling due to cylinder
expansion. At the higher equivalence ratios, the mixing-controlled
phase that follows premixed combustion keeps late-cycle temperatures
higher, thus increasing the oxidation rates of the over-lean methane
elements. This results in reduction of methane emissions at higher
equivalence ratios, as can be seen in Fig. 2.
4.2. Nitrogen oxides (NOx) emissions
The combined NOx emissions results for all three test days are
presented in Fig. 3.
Keeping in mind that the DING engine was operated without ex-
haust gas recirculation (EGR) or any exhaust after treatment, Fig. 3
shows that overall NOx levels range from 310 to 710 ppm-dry, or 2.01
to 5.24 g/bhp-hr. These levels are roughly similar to NOx levels in Table
A1 reported for HPDI engines for similar operating conditions
(1.2–13.3 g/bhp-hr) [28,30,33,56], and are similar to levels in Table A2
reported for dual-fuel engines (200–600 ppm) [52,53], although much
higher NOx emissions have also been reported for dual-fuel engines
[54,55].
The first notable trend in Fig. 3 is that NOx levels decrease with
higher intake pressure. This effect may be explained considering the
average in-cylinder temperature. Average in-cylinder temperature was
calculated using the ideal gas law, knowing the in-cylinder pressure, the
mass of air and fuel inducted (experimentally measured), and the cy-
linder geometry. Since the effect of mass lost due to piston blow-by
could not be incorporated into the calculation, the calculated average
in-cylinder temperature profiles shown in Fig. 4 should be slightly
75
Energy Conversion and Management 164 (2018) 70–82
lower than actual average temperatures and are presented for high-
level discussion of the relative temperature histories between the dif-
ferent operating conditions. Fig. 4 shows the three intake pressure
series at the lowest and highest ɸn values of 0.2 (a) and 0.5 (b).
At both the low and high equivalence ratio conditions, the peak in-
cylinder temperature conditions occur later with increasing intake
pressure. This is caused by longer injection durations required for a
given equivalence ratio at higher intake pressure. NOx formation rates
are expected to be highest at peak temperature, but the delay in
reaching peak temperature limits the time available for NOx formation
and results in lower NOx levels. For example, for the P1 condition at ɸn
of 0.5, peak temperature occurs approximately 115 crank angle degrees
(CAD) prior to exhaust valve opening. At the P3 condition, peak tem-
perature occurs at approximately 70 CAD prior to exhaust valve
opening. In the latter case, NOx has approximately 40% less time to
form before the cylinder mixture is exhausted.
The second notable trend from Fig. 3 is that NOx emissions rise from
ɸn of 0.2 to 0.3 and fall from ɸn of 0.3 to 0.5. Generally, NOx formation
is expected to rise continuously with increasing equivalence ratio in
lean operation due to higher temperatures resulting from higher heat
release at increased equivalence ratios [8,51,52,58]. In fact, this is
observed in Fig. 3 for the two lower equivalence ratios, where the
premixed mode of combustion dominates.
At the two higher equivalence ratios NOx levels decline. This trend
can be explained by considering NOx freezing, a phenomenon which
has significant implications on NOx levels in diesel engines [51] and has
been reported for dual-fuel natural gas engines [8]. NOx freezing occurs
when the high NOx emissions formed early in the combustion cycle are
not allowed to decompose to lower late-cycle equilibrium levels due to
the rapid charge cooling that occurs after the main combustion event
due to cylinder expansion and mixing of the hot combustion products
with cold excess air [51]. At the two lower equivalence ratios NOx
freezing is high since combustion occurs in a relatively short single
premixed phase and is followed by a steep temperature decrease, as
seen in Fig. 4(a). At the two higher equivalence ratios, the premixed
combustion phase is followed by a mixing-controlled combustion phase,
which provides a late-cycle heat release. This results in a more gradual
temperature decrease, as can be seen in Fig. 4(b), and mitigates NOx
chemistry freezing by allowing some of the early-formed NOx to de-
compose closer to late-cycle equilibrium levels. As ɸn increases from 0.4
to 0.5, this mitigating effect grows with the increased proportion of
mixing-controlled combustion and starts to dominate over the com-
peting effect of higher heat release increasing the NOx formation levels.
Another likely contributing factor to the NOx decline over the two
highest equivalence ratios is reduced oxygen availability. An increase in
equivalence ratio results in less oxygen available for oxidation of N2
since a higher proportion of oxygen is used up for oxidizing the fuel.
This is true for the entire range of the tested equivalence ratios but
Fig. 3. DING engine NOx emissions as a function of equivalence ratio. In these experi-
ments, overall equivalence ratio controls load.
I.M. Gogolev, J.S. Wallace Energy Conversion and Management 164 (2018) 70–82

Fig. 4. Calculated mean in-cylinder temperature for ɸn = 0.2(a) and ɸn = 0.5 (b).

becomes a more prevalent effect at the higher equivalence ratios as the
excess oxygen is much reduced. It is also likely that oxygen availability
is influenced by mixing. This second factor is specific to the two higher
equivalence ratio conditions since a large portion of the combustion
process is mixing-controlled. With increased equivalence ratio, the
natural gas jet portion injected during the mixing-controlled phase gets
larger and therefore more difficult to mix with the excess air. Effective
mixing is important for NOx formation since it delivers excess oxygen to
the region of highest temperature and high N2 concentration. Both the
lower oxygen availability and less effective mixing, can contribute to
lower NOx levels at the two highest equivalence ratios.
4.3. Carbon monoxide (CO) emissions
The combined CO emissions results for all three test days are pre-
sented in Fig. 5.
Except for the P1 condition at ɸn of 0.3 and 0.4, CO emissions are in
the range of 270–750 ppm-wet or 1.66–3.93 g/bhp-hr. These CO levels
are similar to levels in Table A1 reported for HPDI engines for similar
operating conditions (0.8–9.7 g/bhp-hr) [28,30,33,56], and are sig-
nificantly lower than levels in Table A2 reported for dual-fuel engines
(800–4000 ppm) [52–55].
With respect to CO concentration response to equivalence ratio, the
most notable trend is the rise observed from ɸn 0.3 to 0.5. The rise for
the P1 series is the steepest of the three intake pressure conditions.
Higher pressure at ɸn of 0.4 and 0.5 appears to significantly lower CO
emissions levels. Discussion on the cause of the P1 series excursion and
on the effects of equivalence ratio and intake pressure is deferred to
Section 4.4.
from tests conducted with polyamide piston rings, as this allowed for
longer test durations. Total PM particle number concentrations and
brake specific emissions for all three test days are presented in Fig. 6.
Except for the P1 test condition, at the two lower ɸn values of 0.2
and 0.3 brake specific PM emissions levels are estimated at
0.69–2.20 mg/bhp-hr. These levels are quite low and can be explained
by the fact that most of the combustion event proceeds in a premixed
mode, same as combustion of natural gas in SI engines and dual-fuel
engines with low pilot proportion [55]. PM levels start to rise for all
intake pressures at the higher ɸn values of 0.4 and 0.5. This trend is
consistent with the trend observed for CO emissions presented in Fig. 5.
In fact, CO and PM concentrations correlate with an R2 coefficient
of > 0.95 over ɸn values of 0.4 and 0.5. The strong correlation supports
the well accepted theory that both CO and PM emissions are controlled
by fuel/air mixing dynamics which govern the availability of oxygen
during the combustion event. This observation is consistent with the
DING combustion modes introduced in Ref. [48]. At the two higher
equivalence ratios, a large proportion of the combustion is mixing-
controlled. Incomplete mixing in the mixing-controlled combustion
phase results in high-temperature fuel-rich gas regions, where CO and
PM form from fuel oxidation in a hot and oxygen deficient environ-
ment. Since the proportion of the mixing-controlled combustion phase
grows with equivalence ratio, CO and PM levels increase accordingly.
The observed excursion of CO and PM concentrations for the P1
condition is quite dramatic, so much that it can be noticed in the plot of
combustion efficiency shown in Fig. 7. Combustion efficiency, the
fraction of the fuel energy supplied which is released in the combustion
process, was calculated from the measured exhaust emissions by di-
viding the difference of enthalpies of products and reactants by the
heating value of the fuel (as described in Eq. (3.27) of Ref. [51]).

4.4. Particulate matter (PM) emissions

An important feature of the PM data presented in this study is the

exclusion of lubricating-oil-sourced PM. This was enabled by using oil-
less polyamide piston rings that were installed in a cylinder that was
only lightly oiled for assembly purposes. To positively verify that this
assembly oil coating or the graphite used as filler in the polyamide
material did not contribute to the PM emissions measurements, all
emissions experiments were repeated twice with custom-designed
Teflon piston rings. The Teflon rings do not contain graphite filler and
were installed in a completely degreased cylinder. Although thermal
expansion of the Teflon rings limited the engine test duration to 2 min,
comparison of PM emission from the polyamide and Teflon piston ring
engine tests showed no discernable difference in PM count or size dis-
tributions (see Appendix B for comparison). This confirmed that the
assembly oil coating and the filler used in the polyamide piston rings do Fig. 5. DING engine CO emissions as a function of equivalence ratio. In these experi-
ments, overall equivalence ratio controls load.
not contribute to PM emissions. The results presented in this section are

76
I.M. Gogolev, J.S. Wallace
Fig. 6. DING engine PM emissions as a function of equivalence ratio. In these experi-
ments, overall equivalence ratio controls load.
Fig. 7. DING engine combustion efficiency as a function of equivalence ratio. In these
experiments, overall equivalence ratio controls load.
Typical combustion efficiency of SI engines ranges from about 95 to
98%, while 98% combustion efficiency is typical for diesel engines
[51]. The fact the DING engine achieves > 98% efficiency for the P2
and P3 condition is an indication of good combustion performance. The
combustion efficiency for the P1 condition falls quite sharply by about
2% over the higher two equivalence ratio conditions. This corresponds
to a noticeable proportion of the fuel forming CO and PM emissions
rather than achieving full oxidation.
Another noticeable trend observed in Fig. 6 at the two higher
equivalence ratios is that higher intake pressure decreases CO and PM
levels. The effect of ambient pressure on fuel air mixing dynamics was
investigated by Desantes et al. [59] in a study of fuel jets injected into
cross-flowing air. In their study, Desantes et at. showed that higher
momentum of the cross flowing air greatly increases mixing and air
entrainment in the fuel jet. In the DING engine, the natural gas fuel is
injected into a swirling air field [48,60], which is essentially a cross-
flow situation. Higher intake pressure increases the air density, re-
sulting in higher momentum of the air swirl field. The increased mo-
mentum leads to more vigorous mixing and air entrainment, conse-
quently leading to a lower proportion of rich fuel/air pockets where CO
and PM form. Improved mixing and air entrainment in the mixing-
controlled combustion phase can also increase the rates of oxidation of
CO and PM that had already formed. The observed CO and PM excur-
sions of the P1 test condition highlight the criticality of achieving
vigorous fuel/air mixing at all test conditions.
The DING engine’s estimated brake specific PM emissions for the
two higher equivalence ratio conditions are significantly lower than
those reported in the literature (Table A1) for HPDI engines
(0.69–16 mg/bhp-hr vs. 1–40 mg/bhp-hr) [29,30,56]. In fact, they are
approximately 70–90% lower than reported by Jones et al. for a HPDI
77
Energy Conversion and Management 164 (2018) 70–82
engine operated at mid-load conditions with no EGR [29]. This differ-
ence can be accounted for by the PM contribution from the engine oil
and diesel pilot of the HPDI engine. In the HPDI engine, engine oil was
determined to contribute up to 20% of total emitted PM [29]. This
source was eliminated in the DING tests by using oil-less polyamide
piston rings. The largest difference, however, can be accounted by PM
which originates from the diesel pilot. In their study, Jones et al.
showed that the diesel pilot contributes from 4 to 40% of the total PM
depending on load and EGR fraction. For the mid-load, no EGR HPDI
conditions referenced above, the pilot contribution was shown to be in
the range of 35–40%. Beyond direct contribution, the diesel pilot was
also found to significantly promote PM formation from natural gas [29].
These direct and indirect PM emissions effects of diesel pilot ignition
suggest that the shielded glow plug ignition assist system implemented
in the DING engine may have significant advantages over diesel pilot
ignition. Of course, confirming this potential advantage requires direct
PM mass measurements and substantial investigation of the interaction
of the many variables in the DING engine’s design and operation.
4.5. Particulate matter (PM) size distribution
PM size distributions measured by the EEPS analyzer are shown in
Fig. 8 on an absolute count and a percentage of total count basis. To aid
in the results analysis, Table 4 shows the ranges of particle sizes clas-
sified by their formation processes. These values are adopted from a
review of soot formation in compression ignition engines by Tree and
Svensson [61].
The majority of the DING engine’s particles are primary particles
formed by surface growth of the initial particle nuclei. Modal diameters
range from 22.1 to 39.2 nm, depending on test conditions. These modal
diameters are similar to HPDI engine particle modal diameters reported
for low to mid-load operation [34,62], but are much smaller than diesel
PM, which tend to have modal diameters of 70–100 nm [63]. A sum-
mary of the DING engine’s PM modal diameters is presented in Table 5.
A rightward shift in the PM size distribution occurs with increased
equivalence ratios. This results in larger particles and in modest levels of
agglomerate particle formation. The effect is most significant for P1 con-
dition, followed by the P2 and P3 conditions. The observed equivalence-
ratio-driven shift can be explained by considering the DING engine com-
bustion modes. In the premixed combustion mode of the two lowest
equivalence ratios, the overall particle levels are low since the bulk of the
natural gas premixes with the air charge. However, since mixing is limited
by the duration of the ignition delay, some small fuel-rich sites are ex-
pected. Since the premixed combustion phase is relatively short-lived, the
particles formed at these sites do not have sufficient time for extensive
surface growth or coalescence. Lack of growth results in small particles,
which are evident in the ɸn = 0.2 and ɸn = 0.3 data series. At the higher
ɸn values of 0.4 and 0.5, the addition of mixing-controlled combustion
phase presents greater opportunity for formation of larger fuel-rich sites
which provide ample residence time for particle formation and growth.
This results in overall larger particles.
Fig. 8 shows that higher intake pressure seems to counteract the
equivalence-ratio-driven shift in PM size. As mentioned prior, higher
intake pressure results in beter mixing and more air entrainment into
the natural gas jet and leads to lower rates of soot formation and higher
rates of soot oxidation. These effects result not only in lower particle
concentrations, but also smaller particles. Lower formation rates limit
particle growth, while higher oxidation rates can oxidize the chemical
species responsible for particle surface growth by absorption. Higher
oxidation rates can also partially oxidize the formed particles to a
smaller size. In general, increased pressure seems to improve mixing
and air entrainment levels to a point where the particles produced in
the mixing-controlled combustion phase start to resemble particles
formed in the premixed combustion phase. This is supported by the fact
that for ɸn values of 0.4 and 0.5, higher intake pressure progressively
I.M. Gogolev, J.S. Wallace Energy Conversion and Management 164 (2018) 70–82

Fig. 8. Measured DING engine PM size distribution for all test conditions.

Table 4
Characterization of particle types.

Formation process Particle type Particle size

range

Pyrolysis + Nucleation = (P + N) Nuclei Down to

1.5–2 nm
P + N + Coalescence = (P + N + C) Primary 30–70 nm
P + N + C + Agglomeration = (P + N + C + A) Agglomerates > 70 nm

Table 5
Measured particle modal diameters for all test conditions.

Nominal equivalence ratio P1 condition P2 condition P3 condition Fig. 9. Calculated brake specific fuel consumption as a function of equivalence ratio. In
these experiments, equivalence ratio controls load.
ɸn = 0.2 22.1 nm 22.1 nm 22.1 nm
ɸn = 0.3 25.5 nm 22.1 nm 22.1 nm
ɸn = 0.4 39.2 nm 29.4 nm 25.5 nm
ɸn = 0.5 39.2 nm 29.4 nm 25.5 nm

78
I.M. Gogolev, J.S. Wallace
forces a PM size distribution similar to that of the premixed series for ɸn
values of 0.2 and 0.3. This is evident in comparison of PM size dis-
tribution plots in Fig. 8.
4.6. Brake specific fuel consumption (BSFC)
Brake specific fuel consumption was calculated from the DING en-
gine’s operating parameters and is presented in Fig. 9.
DING engine’s estimated average brake specific fuel consumption is
in the range of about 127–136 g/kW-hr. Since the data exhibits large
variability, no apparent dependence on equivalence ratio or on the
intake pressure can be noted. An important high-level observation is
that the fuel consumption of the DING engine is substantially lower
than fuel consumption (186–485 g/bhp-hr) of dual-fuel engines from
the literature (Table A2) [52,55], and lower than the fuel consumption
figures (145–172 g/bhp-hr) reported for HPDI engines (Table A1)
[30,33,56]. This is a promising result, but it should be noted in this
initial study the DING engine was not optimized to balance emissions,
performance, and fuel consumption. Operational strategies, such as the
use of EGR to improve NOx emissions, or implementation of different
fuel injection strategies to curtail PM levels will likely result in effi-
ciency and fuel consumption trade-offs. Future investigation is required
to understand the emissions and performance interdependencies of the
DING engine.
5. Summary and conclusions
Performance and emissions experiments were conducted on a
compression-ignition direct-injected natural gas engine (DING)
equipped with a shielded glow plug ignition assist system. Experimental
results showed that the DING engine’s distinct combustion modes
control the engine’s emissions processes.
At the lower ɸn values of 0.2 and 0.3, premixed combustion is
dominant, resulting in low CO and PM emissions, but high CH4 and NOx
emissions. The high degree of air/fuel mixedness and the high avail-
ability of oxygen impede formation and increase oxidation of PM and
CO. High methane emissions result from local over-leaning of the fuel,
while high NOx emissions result from NOx chemistry freezing at super-
equilibrium levels due to a steep in-cylinder temperature decline after
the premixed combustion event.
At the higher ɸn values of 0.4 and 0.5, mixing-controlled combus-
tion follows the initial premixed combustion phase. As in conventional
diesels, the mixing-controlled combustion phase produces elevated le-
vels of CO and PM due to formation of fuel-rich high-temperature re-
gions within the core of the mixing fuel jet [61,64]. However, the
mixing-controlled combustion phase has beneficial effects for CH4 and
NOx emissions. Since the heat release from mixing-controlled combus-
tion occurs later in the expansion cycle, in-cylinder temperature falls at
a slower pace. The higher late cycle in-cylinder temperatures promote
oxidation of surviving methane and aid in decomposition of early-
formed NOx.
Higher intake pressure was found to decrease NOx, CO and PM
emissions levels. Lower NOx emissions result from an indirect effect of
intake pressure. Achieving a given equivalence ratio at a higher intake
pressure requires a longer fuel injection, which delays peak in-cylinder
temperatures and limits time available for NOx formation. The effect of
intake pressure on CO and PM emissions is quite different. Higher in-
take pressure increases the momentum of the engine’s swirl field and
Appendix A. DING emissions summary and comparison
The following tables summarize the comparison of the DING emissions results of this work with values reported in literature for HPDI and dual-
fuel engines. All data reported are engine-out measurements (no exhaust aftertreatment). Where literature values were reported on indicated basis, a
mechanical efficiency of 90% was used to convert the values to brake specific basis.
79
Energy Conversion and Management 164 (2018) 70–82
Table 6
Comparison of DING emissions and BSFC to those of dual-fuel and HPDI engines.
Metric DING vs. dual-fuel DING vs. HPDI
CH4 • Lower at ɸ of 0.2–0.3 • Higher at ɸ of 0.2–0.3
• Much • Similar
n n
lower at ɸ of 0.4–0.5 at ɸ of 0.4–0.5
• Similar • Similar
n n
NOx
• Much • Similar
CO lower
PM • Similar • Lower
BSFC • Much Lower • Lower
promotes mixing and air entrainment in the injected natural gas fuel jet.
Increased mixing and air entrainment helps break up the fuel-rich
pockets where CO and PM form and helps in oxidation of formed CO
and PM.
PM size distributions reveal that DING PM is primarily formed by
nucleation and surface growth, resulting in particles that are much
smaller than typical diesel PM, but similar in size to PM produced by
HPDI engines. An increase in equivalence ratio in mixing-controlled
combustion was found to increase average PM size. This is thought to
occur because a larger fuel jet (at higher equivalence ratios) is more
difficult to mix, thus giving PM particles more time to form and grow to
a larger size. Higher intake pressure, on the other hand, was found to
reduce PM size by enhancing fuel-jet mixing and air entrainment.
Due to the unique oil-less design of the experimental engine, the PM
data presented in this study, to the best of the author’s knowledge, is
the first data set to definitively describe direct-injected compression-
ignited natural gas sourced PM. Natural gas is the only source of par-
ticles in this study, with no contribution from either lubricating oil or a
diesel pilot.
A high-level comparative summary of the DING engine’s emissions
and performance vs. values reported for dual-fuel and HPDI engines is
presented in Table 6.
The major advantage of the DING engine is that it is capable of
significant PM emissions reductions vs. HPDI engines due to a sig-
nificant proportion of soot-free premixed combustion and the absence
of a diesel pilot. Beyond being a definite direct advantage, the DING
engine’s low base PM emissions may allow for more aggressive strate-
gies for NOx reduction by EGR than may be possible in HPDI engines
due to the limitations of their PM vs. NOx emissions trade-off char-
acteristics. Implementation of EGR in the DING engine will be explored
in upcoming studies.
Acknowledgment
The authors gratefully acknowledge the previous work on DING
engines carried out by Daniel Chown, as well as helpful suggestions by
Kang Pan and technical assistance from Sean Kieran, Osmond Sargeant,
Keith Hockley, and Ryan Mendell.
Funding
This work was supported by the Natural Sciences and Engineering
Research Council of Canada Discovery Grant Program (Grant No.
GRPIN 201403720). I.G. was financially supported by an Ontario
Graduate Scholarship and the NSERC Collaborative Research and
Training Experience (CREATE) Program on Clean Combustion Engines.
I.M. Gogolev, J.S. Wallace Energy Conversion and Management 164 (2018) 70–82

Table A1
DING emissions vs. HPDI emissions comparison.

Engine type & data CH4 (g/bhp-hr) THC (g/bhp-hr) NOx (g/bhp-hr) CO (g/bhp-hr) PM (mg/bhp-hr) BSFC (g/bhp- Varied parameter
source hr)

DING 2.65–8.79 – 3.74–5.24 1.96–4.59 0.75–1.00 170–182 Equivalence ratio, intake pressure
φn 0.2–0.3
DING 0.39–1.37 – 2.01–4.12 1.66–20.9 0.69–16.0 170–183
φn 0.4–0.5
HPDI [30] – 0.4–1.7 2.5–13.3 1.9–4.1 15.0–17.0 145–172 Combustion timing and pilot/natural gas
phasing
HPDI [56] – 0.2–0.5 3.6–4.0 1.3–8.5 10.0–40.0 152–162 Fuel injection pressure
HPDI [28] 0.3–0.4 0.45–0.55 1.2 – 1.5 0.8–1.0 – – Fuel injection timing
HPDI [33] – 0.4–0.6 2.2–8.9 2.2–9.7 – 151–164 Diesel pilot rail pressure and injection
timing
HPDI [29] – – – – 1.0–21.0 – Diesel pilot proportion and equivalence
ratio

Table A2
DING emissions vs. dual-fuel emissions comparison.

Engine type & data source CH4 (ppm) THC (ppm) NOx (ppm) CO (ppm) PM (mg/bhp-hr) BSFC (g/bhp-hr) Varied parameter

DING 650–1550 – 310–710 270–450 0.75–1.00 170–182 Equivalence ratio, intake pressure
φn 0.2–0.3
DING 120–400 – 380–660 280–4030 0.69–16.0 170–183
φn 0.4–0.5
Dual-Fuel [52] – 1700–5600 200–600 1500–3200 – 186–485 Natural gas proportion
Dual-Fuel [53] – 2000–4000 100 – 400 3500–4000 – – Natural gas proportion
(NO only)
Dual-Fuel [54] – 2000–6500 1200–1600 800–1100 10.4–26.1 – Injection timing and pilot amount
Dual-Fuel [55] – 2000–4300 450–1300 1000–1500 0.22–0.30 186–373 Natural gas proportion

Appendix B. PM emissions polyamide vs. teflon piston rings

PM emissions for experiments performed with polyamide and Teflon piston rings are presented in Fig. B1. Each point represents the mean of the
average PM concentration collected from 1 min 30 s to 2 min into each test. The tests for the “Polyamide” series were repeated three times, while the
“Teflon” series tests were repeated twice. The error bars of the “Polyamide” series represent ± two standard deviations between three test values,
while for the “Teflon” series the error bars represent the maximum and minimum of the two test values.
Fig. B1 shows that within the same test conditions, there is no significant difference in PM emissions between the “Polyamide” and “Teflon”
series. Inspection of the PM size distributions (not shown here) at each of the test condition also showed no significant deviation of particle size
distributions between the polyamide and the Teflon piston ring tests. It was concluded that the assembly lube oil used with polyamide piston rings
and the graphite filler of the polyamide piston rings does not contribute to the PM emissions produced during DING engine tests.

Fig. B1. PM emissions polyamide vs. teflon piston rings.

80
I.M. Gogolev, J.S. Wallace
References
[1] US Energy Information Agency. U.S. primary energy consumption by source and
sector; 2016 [available from: https://www.eia.gov/totalenergy/data/monthly/pdf/
flow/css_2016_energy.pdf].
[2] Atabani Abdelaziz E, Silitonga Arridina S, Anjum Badruddin Irfan, Mahlia TMI,
Masjuki HH, Mekhilef S. A comprehensive review on biodiesel as an alternative
energy resource and its characteristics. Renew Sustain Energy Rev
2012;16(4):2070–93.
[3] Bergthorson Jeffrey M, Thomson Murray J. A review of the combustion and emis-
sions properties of advanced transportation biofuels and their impact on existing
and future engines. Renew Sustain Energy Rev 2015;42:1393–417.
[4] Cathles Lawrence M. Assessing the greenhouse impact of natural gas. Geochem
Geophys Geosyst 2012;13(6).
[5] Abboud S, Aschim K, Bagdan B, Sarkar P, Yuan H, Scorfield B, et al. Potential
production of methane from Canadian wastes. Final report by the Alberta research
council and the Canadian gas association; September 2010 [available from: https://
www.researchgate.net/publication/268341359_Potential_Production_of_Methane_
from_Canadian_Wastes].
[6] Manuel Götz, Lefebvre Jonathan, Mörs Friedemann, Koch Amy Daniel, Graf Frank,
Bajohr Siegfried, et al. Renewable power-to-gas: a technological and economic re-
view. Renewable Energy 2016;85:1371–90.
[7] Natural gas vehicle statistics. International association for natural gas vehicles;
online statistics. Available on: www.iangv.org, visited on 2017 June 06, at http://
www.iangv.org/category/stats/.
[8] Korakianitis T, Namasivayam AM, Crookes RJ. Natural-gas fueled spark-ignition
(SI) and compression-ignition (CI) engine performance and emissions. Prog Energy
Combust Sci 2011;37(1):89–112.
[9] Wei Lijiang, Geng Peng. A review on natural gas/diesel dual fuel combustion,
emissions and performance. Fuel Process Technol 2016;142:264–78.
[10] Adlercreutz L, Cronhjort A, Andersen J, Ogink R. Optimizing the natural gas engine
for CO2 reduction. SAE Technical Paper No. 2016-01-0875; 2016.
[11] Zoldak Philip, Naber Jeffrey. Spark ignited direct injection natural gas combustion
in a heavy duty single cylinder test engine-AFR and EGR dilution effects. No. 2015-
01-2808. SAE Technical Paper; 2015.
[12] Zoldak Philip, Naber Jeffrey. Spark ignited direct injection natural gas combustion
in a heavy duty single cylinder test engine-start of injection and spark timing ef-
fects. No. 2015-01-2813. SAE Technical Paper; 2015.
[13] Cummins-Westport. ISL G Near Zero Fact Sheet [Available from http://www.
cumminswestport.com/content/679/CWI%20ISL%20G_NZ%20Fact%20Sheet
%202017.pdf].
[14] US EPA annual certification data for engines and equipment [available from
https://www.epa.gov/compliance-and-fuel-economy-data/annual-certification-
data-engines-and-equipment].
[15] Besch M, Israel J, Thiruvengadam A, Kappanna H, et al. Emissions characterization
from different technology heavy-duty engines retrofitted for CNG/diesel dual-fuel
operation. SAE Int J Engines 2015;8(3):1342–58. http://dx.doi.org/10.4271/2015-
01-1085.
[16] Fraser RA, Siebers DL, Edwards CF. Autoignition of methane and natural gas in a
simulated diesel environment. SAE Paper No. 910227; 1991.
[17] Aesoy V, Valland H. The influence of natural gas composition on ignition in a direct
injection gas engine using hot surface assisted compression ignition. SAE Paper No.
961934; 1996.
[18] Aesoy V, Valland H. Hot surface assisted compression ignition of natural gas in a
direct injection diesel engine. SAE Paper No. 960767; 1996.
[19] Naber JD, Siebers DL, Di Julio SS, Westbrook CK. Effects of natural gas composition
on ignition delay under diesel conditions. Combust Flame 1994;99(2):192–200.
[20] Papagiannakis RG, Rakopoulos CD, Hountalas DT, Rakopoulos DC. Emission char-
acteristics of high speed, dual fuel, compression ignition engine operating in a wide
range of natural gas/diesel fuel proportions. Fuel 2010;89(7):1397–406.
[21] Rochussen Jeremy, Yeo Jeff, Kirchen Patrick. Effect of fueling control parameters
on combustion and emissions characteristics of diesel-ignited methane dual-fuel
combustion. No. 2016-01-0792. SAE Technical Paper; 2016.
[22] Korakianitis T, Namasivayam AM, Crookes RJ. Diesel and rapeseed methyl ester
(RME) pilot fuels for hydrogen and natural gas dual-fuel combustion in com-
pression–ignition engines. Fuel 2011;90(7):2384–95.
[23] Gatts Timothy, Liu Shiyu, Liew Chetmun, Ralston Bradley, Bell Clay, Li Hailin. An
experimental investigation of incomplete combustion of gaseous fuels of a heavy-
duty diesel engine supplemented with hydrogen and natural gas. Int J Hydrogen
Energy 2012;37(9):7848–59.
[24] Königsson F, Kuyper J, Stalhammar P, Angstrom H-E. The influence of crevices on
hydrocarbon emissions from a diesel-methane dual fuel engine. SAE Technical
Paper No. 2013-01-848; 2013.
[25] US code of federal regulations. Greenhouse gas emissions and fuel efficiency stan-
dards for medium- and heavy-duty engines and vehicles - phase 2. 49 CFR Parts
523, 534, 535, and 538 [available from https://www.gpo.gov/fdsys/pkg/FR-2016-
10-25/pdf/2016-21203.pdf].
[26] Hodgins BK, Philip GH, Ouellette P, Hung P. Directly injected natural gas fueling of
diesel engines. SAE Paper No. 961671; 1996.
[27] Dumitrescu S, Hill PG, Li G, Ouellette P. Effects of injection changes on efficiency
and emissions of a diesel engine fueled by direct injection of natural gas. SAE Paper
No. 2000-01-1805; 2000.
[28] Harrington James, Munshi Sandeep, Nedelcu Costi, Ouellette Patric, Thompson
Jeff, Whitfield Stewart. Direct injection of natural gas in a heavy-duty diesel engine.
No. 2002-01-1630. SAE Technical Paper; 2002.
81
Energy Conversion and Management 164 (2018) 70–82
[29] Jones HL, McTaggart-Cowan GP, Rogak SN, Bushe WK, Munshi SR, Buchholz BA.
Source apportionment of particulate matter from a diesel pilot-ignited natural gas
fuelled heavy duty DI engine. SAE Paper No. 2005-01-2149; 2005.
[30] McTaggart-Cowan GP, Bushe WK, Rogak SN, Hill PG, Munshi SR. PM and NOx
reduction by injection parameter alterations in a direct injected, pilot ignited, heavy
duty natural gas engine with EGR at various operating conditions. No. 2005-01-
1733. SAE Technical Paper; 2005.
[31] Faghani E, Patychuk B, McTaggart-Cowan G, Rogak S. Soot emission reduction from
post injection strategies in a high pressure direct-injection natural gas engine. SAE
Technical Paper 2013-24-0114; 2013. doi: http://dx.doi.org/10.4271/2013-24-
0114.
[32] Mabson Christopher WJ, Ehsan Faghani, Pooyan Kheirkhah, Patrick Kirchen, Rogak
Steven N, McTaggart-Cowan Gordon. Combustion and emissions of paired-nozzle
jets in a pilot-ignited direct-injection natural gas engine. No. 2016-01-0807. SAE
Technical Paper; 2016.
[33] Qiang Zhang, Li Menghan, Shao Sidong. Combustion process and emissions of a
heavy-duty engine fueled with directly injected natural gas and pilot diesel. Appl
Energy 2015;157:217–28.
[34] Graves Brian, Olfert Jason, Patychuk Bronson, Dastanpour Ramin, Rogak Steven.
Characterization of particulate matter morphology and volatility from a compres-
sion-ignition natural-gas direct-injection engine. Aerosol Sci Technol
2015;49(8):589–98.
[35] Hilger U, Jain G, Scheid E, Pischinger F, Bruetsch R, Bernhardt W, et al.
Development of a direct injected neat methanol engine for passenger car applica-
tions. SAE Paper No. 901521; 1990.
[36] Mitchell WL, Litzinger TA, Santavicca DA. Neat methanol combustion in a D.I.
diesel engine using catalytically coated glow plugs. SAE Paper No. 912418; 1991.
[37] Reynaldo AJ, Abata DL, Mullins MA. Catalytic ignition of methanol in a diesel
engine with a platinum-coated glow plug. SAE Paper No. 911737; 1991.
[38] Mueller CJ, Musculus MP. Glow plug assisted ignition and combustion of methanol
in an optical DI diesel engine. SAE Paper No. 2001-01-2004; 2001.
[39] Ghabi NR, Wallace JS. Bluff-body stabilized glow plug ignition of a methanol-fueled
IDI diesel engine. SAE Paper No. 930935; 1993.
[40] Oester U, Wallace J. Liquid propane injection for diesel engines. SAE Paper No.
872095; 1987.
[41] Araki M, Dong H, Obokata T, Shiga S, Ishima T, Aoki K, et al. Characteristics of CNG
direct injection with auto-ignition. SAE Paper No. 2005-26-358; 2005.
[42] Ikegami M, Shioji M, Umekita K, Hashimoto H. A glow-plug assisted direct-injection
diesel engine using low-ignitability fuels. In: 23rd international federation of au-
tomobile engineers’ and technicians’ associations international congress, Trino,
Italy, May 7–11, Paper No. 905024; 1990.
[43] Mello JP, Bezaire D, Sriramulu S, Weber R. Performance and economics of catalytic
glow plugs and shields in direct injection natural gas engines for the next generation
natural gas vehicle program. No. NREL/SR-540-34286. National Renewable Energy
Laboratory (NREL), Golden, CO; 2003.
[44] Chown D, Habbaky C, Wallace JS. An experimental investigation of combustion
chamber design parameters for hot surface ignition. ASME Paper No. ICEF2014-
5646; 2014.
[45] Mueller CJ, Musculus MP. Glow plug assisted ignition and combustion of methanol
in an optical DI diesel engine. SAE Technical Paper No. 2001-01-2004; 2001.
[46] Fabbroni M, Wallace JS. Flame propagation in natural gas fueled direct injection
engines. ASME Paper No. ICEF2010-35182; 2010.
[47] Cheng SX. Modeling injection and ignition in direct injection natural gas engines.
Ph.D. dissertation, University of Toronto, Toronto, ON, Canada; 2008.
[48] Gogolev IM, Wallace JS. Study of assisted compression ignition in a direct injected
natural gas engine. ASME J Eng Gas Turbines Power 2017;139(12). http://dx.doi.
org/10.1115/1.4036968. 122802-122802-12.
[49] Green CJ, Wallace JS. Electrically actuated injectors for gaseous fuels. SAE Paper
No. 892143; 1989.
[50] Zimmerman Naomi, Jeong Che-Heon, Wang Jonathan M, Ramos Manuel, Wallace
James S, Evans Greg J. A source-independent empirical correction procedure for the
fast mobility and engine exhaust particle sizers. Atmos Environ 2015;100:178–84.
[51] Heywood J. Internal combustion engine fundamentals. New York: McGraw-Hill;
1988.
[52] Papagiannakis RG, Hountalas DT. Combustion and exhaust emission characteristics
of a dual fuel compression ignition engine operated with pilot diesel fuel and nat-
ural gas. Energy Convers Manage 2004;45(18):2971–87.
[53] Papagiannakis RG, Hountalas DT. Experimental investigation concerning the effect
of natural gas percentage on performance and emissions of a DI dual fuel diesel
engine. Appl Therm Eng 2003;23(3):353–65.
[54] Liu Jie, Yang Fuyuan, Wang Hewu, Ouyang Minggao, Hao Shougang. Effects of pilot
fuel quantity on the emissions characteristics of a CNG/diesel dual fuel engine with
optimized pilot injection timing. Appl Energy 2013;110:201–6.
[55] Lounici Mohand Said, Loubar Khaled, Tarabet Lyes, Balistrou Mourad, Niculescu
Dan-Catalin, Tazerout Mohand. Towards improvement of natural gas-diesel dual
fuel mode: An experimental investigation on performance and exhaust emissions.
Energy 2014;64:200–11.
[56] McTaggart-Cowan GP, Jones HL, Rogak SN, Bushe WK, Hill PG, Munshi SR. The
effects of high-pressure injection on a compression–ignition, direct injection of
natural gas engine. J Eng Gas Turbines Power 2007;129(2):579–88.
[57] Brown BS, Laforet CA, Rogak SN, Munsh SR. Comparison of injectors for com-
pression ignition of natural gas with entrained diesel. Int J Engine Res
2011;12(2):109–22.
[58] Abd-Alla GH, Soliman HA, Badr OA, Abd-Rabbo MF. Effects of diluent admissions
and intake air temperature in exhaust gas recirculation on the emissions of an in-
direct injection dual fuel engine. Energy Convers Manage 2001;42(8):1033–45.
I.M. Gogolev, J.S. Wallace
[59] Desantes JM, Arre‘gle J, Lopez JJ, Garcıa JM. Turbulent gas jets and diesel-like
sprays in a crossflow: a study on axis deflection and air entrainment. Fuel
2006;85(14):2120–32.
[60] Provenaz P. Measurement of swirl and tumble generated by several different cy-
linder heads. Engine Research and Development Lab, University of Toronto,
Toronto, ON, Canada.
[61] Tree Dale R, Svensson Kenth I. Soot processes in compression ignition engines. Prog
82
Energy Conversion and Management 164 (2018) 70–82
Energy Combust Sci 2007;33(3):272–309.
[62] Soewono Arka, Rogak Steven. Morphology and Raman spectra of engine-emitted
particulates. Aerosol Sci Technol 2011;45(10):1206–16.
[63] Lepperhoff Gerhard. Influences on the particle size distribution of diesel particulate
emissions. Top Catal 2001;16(1):249–54.
[64] Dec John E. A conceptual model of DI diesel combustion based on laser-sheet
imaging. No. 970873. SAE Technical paper; 1997.