Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

SAE TECHNICAL
PAPER SERIES 2006-01-0197

The Effect of Swirl Ratio and Fuel Injection

Parameters on CO Emission and Fuel Conversion
Efficiency for High-Dilution, Low-Temperature
Combustion in an Automotive Diesel Engine
Sanghoon Kook and Choongsik Bae
Korea Advanced Institute of Science and Technology

Paul C. Miles and Dae Choi

Sandia National Laboratories

Michael Bergin and Rolf D. Reitz

University of Wisconsin-Madison

Reprinted From: Compression Ignition Combustion Process 2006

(SP-2012)

2006 SAE World Congress

Detroit, Michigan
April 3-6, 2006

400 Commonwealth Drive, Warrendale, PA 15096-0001 U.S.A. Tel: (724) 776-4841 Fax: (724) 776-5760 Web: www.sae.org
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

The Engineering Meetings Board has approved this paper for publication. It has successfully completed
SAE's peer review process under the supervision of the session organizer. This process requires a
minimum of three (3) reviews by industry experts.

All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or
transmitted, in any form or by any means, electronic, mechanical, photocopying, recording, or otherwise,
without the prior written permission of SAE.

For permission and licensing requests contact:

SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Tel: 724-772-4028
Fax: 724-776-3036

For multiple print copies contact:

SAE Customer Service

Tel: 877-606-7323 (inside USA and Canada)
Tel: 724-776-4970 (outside USA)
Fax: 724-776-0790
Email: CustomerService@sae.org

ISSN 0148-7191
Copyright 2006 SAE International
Positions and opinions advanced in this paper are those of the author(s) and not necessarily those of SAE.
The author is solely responsible for the content of the paper. A process is available by which discussions
will be printed with the paper if it is published in SAE Transactions.

Persons wishing to submit papers to be considered for presentation or publication by SAE should send the
manuscript or a 300 word abstract to Secretary, Engineering Meetings Board, SAE.

Printed in USA
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
The Effect of Swirl Ratio and Fuel Injection Parameters
on CO Emission and Fuel Conversion Efficiency for
High-Dilution, Low-Temperature Combustion in an
Automotive Diesel Engine
Korea Advanced Institute of Science and Technology
Copyright 2006 SAE International
ABSTRACT
Engine-out CO emission and fuel conversion efficiency
were measured in a highly-dilute, low-temperature diesel
combustion regime over a swirl ratio range of 1.447.12
and a wide range of injection timing. At fixed injection
timing, an optimal swirl ratio for minimum CO emission
and fuel consumption was found. At fixed swirl ratio, CO
emission and fuel consumption generally decreased as
injection timing was advanced. Moreover, a sudden
decrease in CO emission was observed at early injection
timings. Multi-dimensional numerical simulations,
pressure-based measurements of ignition delay and
apparent heat release, estimates of peak flame
temperature, imaging of natural combustion luminosity
and spray/wall interactions, and Laser Doppler
Velocimeter (LDV) measurements of in-cylinder
turbulence levels are employed to clarify the sources of
the observed behavior. Mixing processes occurring after
the pre-mixed burn are found to be the likely source of
the optimal swirl ratio, while enhanced pre-combustion
mixing dominates the reduction in CO with earlier
injection. Liquid fuel films formed on the bowl lip are not
found to significantly impact CO emissions, and
increased injection pressure typically reduces CO
emissions at this high dilution rate. Fuel conversion
efficiency is examined in terms of component
efficiencies related to combustion phasing, heat loss,
and combustion efficiency. The influence of swirl and
injection timing on each of these efficiencies is
discussed.
INTRODUCTION
Diesel engines are an attractive option for automotive
powerplants due to their high fuel conversion efficiency
2006-01-0197
Sanghoon Kook and Choongsik Bae
Paul C. Miles and Dae Choi
Sandia National Laboratories
Michael Bergin and Rolf D. Reitz
University of Wisconsin-Madison
(often termed the thermal efficiency), which can exceed
the fuel conversion efficiency of modern SI engines by
as much as 40% [1]. However, for conventional diesel
combustion regimes, emissions of NOx and particulates
from these engines are excessive, and costly exhaust-
gas after-treatment systems are required to meet current
emission regulations.
Recently, diesel combustion systems employing low
combustion temperatures and enhanced fuel-air pre-
mixing have been proposed to drastically reduce NOx
and particulate emissions (e.g. [2]-[6]). In these systems,
low combustion temperatures are achieved by
employing high levels of exhaust gas recirculation (EGR)
or retarded injection timing, while pre-mixing is
enhanced by the use of split injections, high-injection
pressures, or measures that increase the ignition delay.
This work focuses on those systems that employ high
EGR rates to maintain low combustion temperatures,
with no special measures taken to increase pre-mixing
other than the increased ignition delay associated with
dilute mixtures.
Traditionally, low combustion temperatures are thought
to reduce NOx emissions but increase particulate
emissions due to a reduction in particulate oxidation
rates. However, as EGR rates are increased to very high
levels (approximately 60% or greater) simultaneous soot
and NOx reduction is observed [3, 5]. At these high
dilution levels, characterized by O2 concentrations of
about 11%, the flame temperatures are limited to levels
at which the soot and NOx formation rates are low.
Unfortunately, the region of applicability of these low
soot and NOx combustion regimes within the load-speed
operating map is often limited by high CO and UHC
emissions, with an accompanying fuel economy penalty.
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

2000 300
Skip-fired CO Emission [ppm]

290

1500 CO 280

ISFC [g/kW-hr]
270

1000 260

250

500 240

230
' ISFC | 4%
0 220
-30 -25 -20 -15 -10
SOI [CAD]

Figure 1 The influence of injection timing on CO

emissions and fuel economy for a highly-dilute
(O2=10.1%) intake charge measured in the same
engine of this study. 1500 RPM, 3 bar IMEP,
Pinj=800bar, Rs=3.77. Reproduced from Ref. [9] Figure 2 Optical engine geometry and experimental set-
up for velocity measurements. High-speed video
imaging of combustion luminosity and spray penetration
was also performed, observing the combustion chamber
CO emission generally stems from two primary sources. from below (via the 45 mirror) and from the side (via the
The first source, under-mixing of fuel, is the CO formed liner windows)
in the products of rich pre-mixed combustion. If mixing
rates are too low, this CO will fail to mix with sufficient
additional O2 to complete the oxidation process before
expansion quenches chemical reactions. The second In this paper, we focus on clarifying the causes of the
source, over-mixing, is associated with fuel that is CO emission and fuel economy trends observed in
mixed to very lean equivalence ratios during the ignition Fig. 1, through application of a combined approach
delay period. After ignition, these overly lean mixtures do employing conventional pressure-based diagnostics,
not reach sufficiently high temperatures to complete the optical diagnostics, and numerical simulation. In addition
oxidation process on engine time scales. to injection timing, we have also varied both the injection
pressure and the swirl ratio, in an effort to further clarify
For thoroughly pre-mixed, very lean combustion systems, both the dominant source of CO emission as well as the
the over-mixed fuel and subsequent low peak particular aspects of the mixing processes that influence
combustion temperature is the source of engine CO CO emission and fuel economy. Increased injection
emission [7]. Similarly, the results of numerical pressure can be expected to primarily influence mixing
simulations point to over-mixed fuel as the principal processes during the ignition delay period, while
source of CO emission from conventional diesel increased swirl is anticipated to influence the later
combustion systems [8]. However, as dilution levels mixing processes to a greater extent than the injection-
increase, the correlation between CO emission and dominated early mixing processes. Additionally, swirl is
ignition delay observed in dilute, low-temperature diesel a parameter which has been previously cited as
combustion systems becomes negative [9]that is, important for reducing emissions [2] in low-temperature
increased ignition delay is found to correspond to combustion systems, and is generally found to be
reduced CO emissions for highly-dilute mixtures. beneficial for soot emission and fuel economy in
Additionally, the average equivalence ratio at ignition is conventional diesel combustion systems (e.g. [10]-[12]).
estimated to exceed 2, and to increase with increasing
dilution level. Jointly, these observations indicate that for The dilution level employed is characteristic of the
highly-dilute systems the dominant CO emission source Toyota smokeless combustion regime [3] or the AVL
more likely stems from under-mixed fuel. HCLI regime [5], and represents a very low-NOx, low-
soot operating regime where CO emissions and fuel
More importantly, significant changes in CO emission economy have already begun to suffer [9]. In clarifying
and in fuel economy can be obtained by changing the the source of CO emission and changes in fuel economy,
injection timing, as is shown in Fig. 1. With early our objective is to identify potential pathways for
injection, i.e. SOI=-26 CAD ATDC, CO emission is improvement of the performance of these combustion
decreased 3-fold over emission levels measured with systems.
injection timing at -10 CAD ATDC. Clearly, mixing
processes can be significantly influenced by injection
timing, and potentially by other factors as well.
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
EXPERIMENT
RESEARCH ENGINE
A single-cylinder optically-accessible diesel research
engine was employed to study the effect of swirl and
injection timing on CO emission and fuel conversion
efficiency. The engine has characteristics typical of
small-bore engines intended for automotive applications,
i.e.: four valves; a central, vertical injector; a 6-hole;
minisac nozzle; and a concentric, re-entrant bowl. A
schematic of the engine is shown in Fig. 2, and its main
specifications are shown in Table 1. The floor of the
piston bowl was comprised of a quartz insert that
permitted in-cylinder fluid velocity measurements to be
obtained from below via laser Doppler velocimetry.
The engine is equipped with a common-rail fuel injection
system capable of a maximum injection pressure of
1350 bar. This system allows the injection pressure,
injected mass (duration of injection) and start-of-injection
(SOI) to be set independently. The injector is
instrumented with a needle-lift sensor that permits the
actual start of injection to be accurately determined.
The approximate fueling rate was determined by
measuring the mass injected into a small chamber which
seals against the engine head. The calibration was
performed with the head/injector at operating
temperature. Fuel was injected into the chamber for 575
injection events, and the average injected quantity
determined by dividing the total mass by the number of
injection events. This averaged value was used to
calculate the fuel conversion efficiency. Note that this
procedure accounts for thermal changes in the injector
performance, but does not account for the higher
ambient pressures encountered when the engine is
operating. Accordingly, the average injected mass is
likely to be overestimated, resulting in low values for the
fuel conversion efficiency. Neglect of the influence of
cylinder pressure will also introduce a bias into the
estimated fueling rate, wherein the injected fuel amount
at later injection timing will be less due to the higher
cylinder pressures.
The magnitude of the overestimation/bias in the injected
fuel mass can be estimated by assuming that, at the
baseline injection pressure of 800 bar, a mean effective
injection pressure of 400 bar is achieved. At the earliest
injection timing employed, the average cylinder pressure
during injection is 22 bar, resulting in an overestimation
of the injected fuel by 3%. At the latest injection timing,
the 47 bar cylinder pressure results in a 6%
overestimation. As will be seen below, the magnitude of
this bias is small as compared to the magnitude of the
changes in fuel conversion efficiency observed.
In-cylinder pressure measurements were acquired using
a water-cooled (KISTLER 6043A60) piezoelectric
pressure transducer. Measurements obtained over 50
engine cycles were averaged to calculate the indicated
Table 1 Engine specifications
Basic Geometry
Bore: 79.5 [mm] Stroke: 85.0 [mm]
Disp. Vol.: 422 [cm3] Comp. Ratio: 18.7
Valve Events
IVO: -360 CAD ATC EVO: 145 CAD ATC
IVC: -140 CAD ATC EVC: -350 CAD ATC
Fuel Injection Equipment
Bosch Flow No.: 320 Included Angle: 145
Number of Holes: 6 Hydro-erosion: 10 %
Nozzle Style: Cylindrical Minisac (Bosch DLLA)
mean effective pressure (IMEP) and the apparent heat
release rate. Care was exercised to ensure that
pressure-record filtering did not introduce artificial side-
lobes in the calculated apparent heat release rates.
Additional details of the calculation methodology are
presented in our earlier work [9]. In-cylinder pressure
data, and the velocity data described below, were
acquired with a resolution of 0.25 crank angle degrees
(CAD). Unless otherwise specified, all crank angles
presented in this paper are reported as CAD after TDC
compression.
Engine-out CO emission was measured using a
California Analytical Instruments NDIR (Non-Dispersive
Infra-Red) CO analyzer (model 300). Samples were
taken from the exhaust plenum and transferred to the
analyzer through a heated sample line. Moisture and
condensable hydrocarbons were removed from the
sample by a condenser prior to the analyzer.
OPTICAL DIAGNOSTICS
Spray and Luminosity Imaging
To provide information on spray-targeting, an expanded
514.5 nm Ar+ ion laser beam was directed into the
combustion chamber through one of the side liner
windows shown in Fig. 2. A high speed digital video
camera (Integrated Design Tools Inc., X-Stream Vision)
was employed to view the spray penetration and wall-
film formation from below, via the extended piston and
45 mirror. The frame rate was set to 21,593 frames per
second, corresponding to 0.4168 CAD between images
at 1,500 rpm. The image resolution is 100 x 120 pixels
and an exposure time of 30 s was employed.
High speed videos were also obtained of the natural
luminosity from soot emissions, viewing the combustion
chamber through a side window. In this case, the
framing rate employed was 9,000 frames per second,
corresponding to 1 image per crank degree.
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

Table 2 Operating Parameters Table 3 Variation in fueling rate with swirl ratio

Basic Geometry Swirl Ratio Inj. Duration Fueling Rate

Speed [rpm] 1500 [Psec] [mg/stroke]
Load 3 bar gross IMEP 1.44 409 8.325
2.59 409 8.325
TCoolant [C] 88
3.77 421 8.908
Simulated EGR [Mass %] 65
4.94 425 9.108
10.1% O2, 80.6% N2,
Intake charge composition 7.12 456 10.258
9.3% CO2
Typical In-Cylinder I 0.57
PInjection [bar] 600, 800, 1000, 1200
Swirl Ratio (Rs) 1.44, 2.59, 3.77, 4.94, 7.12 fuel injection occurring on only one of every four engine
PIntake [bar] 1.2 cycles. This skip-firing period allowed the species
concentrations in the exhaust system and the emission
TIntake [C] 90
analyzers to stabilize. Finally, data were acquired over
PExhaust [bar] 1.3 50 additional skip-fired cycles. For velocity
Start-of-Injection (SOI) measurements, data were acquired for 120 s after the
-30.25 to -7.75
[CAD ATDC] commencement of skip-fired operation. After data
acquisition, a fixed cool-down/cleaning period was
employed to ensure that the gross thermal state of the
engine was the same for all measurements.

Flow Velocity
Measurements of the tangential component of velocity
were acquired with a custom-designed, fiber-coupled
LDV. A laser power of approximately 260 mW per beam
was used at a wavelength of 514.5 nm. A 5.0 MHz
differential frequency shift removed directional ambiguity.
ZrO2 seed particles were introduced with a fluidized bed
seeder. The seed particles have a mean diameter of
0.48 m and a standard deviation of 0.15 m. Additional
details can be found in Ref. 21.
ENGINE OPERATION
The engine operating parameters are described in
Table 2. All data were acquired at a fixed engine speed
of 1500 rpm.
Recirculated exhaust gas was simulated by diluting the
intake air stream with N2 and CO2. The relative
proportions of CO2 and N2 were chosen to match the
mixture molar specific heat of real engine exhaust gas at
the selected load and O2 concentration. This matching
was performed with gas properties evaluated at 600 K.
The 10.1% O2 concentration selected corresponded to a
65% exhaust gas recirculation (EGR) rate. At Rs = 3.77
the typical in-cylinder equivalence ratio ( I ), was 0.57
(O= 1.75), corresponding to an equivalence ratio based
on intake air of 0.79 (O= 1.27).
Prior to obtaining in-cylinder pressure, exhaust
emissions and flow velocity, the engine was motored for
a minimum of 90 s in order to pre-heat the combustion
chamber walls and to allow the intake plenum pressure
to stabilize. For in-cylinder pressure and exhaust
emission data, the engine was skip-fired for 70 s, with
The swirl ratio is varied by throttling the low-swirl intake
ports. To achieve the large range of swirl ratios reported
here, the intake valve in the high-swirl port was fitted
with a 180 shroud, oriented to enhance the component
of the intake flow tangential to the cylinder wall. The
swirl ratios reported in Table 2 are the numeric values
resulting from flow-bench measurements performed by
Ricardo, Inc. Radial profiles of the tangential velocity at
three axial locations, measured with the LDV through the
liner windows, provide a more direct measure of the in-
cylinder angular momentum. At -55 CAD, integration of
these profiles resulted in swirl ratios of 1.10, 2.75, 3.86,
5.05, and 6.53. With the exception of the lowest swirl
ratio, these measurements are within 8% of the flow-
bench swirl ratios. The swirl ratios reported throughout
the remainder of this document are those determined via
flow-bench.
The fueling rate was controlled to provide a constant 3
bar gross IMEP (indicated mean effective pressure) for
each swirl ratio, at the injection timing providing best fuel
economy. Thereafter, as SOI was varied, the fueling rate
was held fixed. The most advanced injection timing
considered at each swirl ratio was determined by the
achievable IMEP. When the IMEP dropped significantly
from the desired load no earlier timings were considered.
As discussed below, excessive spray penetration due to
low in-cylinder gas density is thought to limit the earliest
practical injection timing. Injection timing was retarded to
the crank angle at which impaired fuel economy was
observed at the lower swirl ratios, typically -10.25 CAD.
The approximate fueling rates employed for each swirl
ratio are listed in Table 3. The fuel employed was a 2007
emission certification diesel with a cetane number of
47.1 and a lower heating value of 42.98 MJ/kg.
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
NUMERICAL SIMULATION
In previous studies, the three-dimensional intake flow
field was calculated with STAR-CD. At IVC, those results
were mapped to a sector grid and used to initialize the
KIVA-3V [14] simulations of the combustion process.
However, a comparison of the temporal history of the
tangential velocity and turbulent kinetic energy for
various initialization schemes revealed that a simpler
solid-body-rotation swirl initialization performed well
compared to the mapped results [15]. Due to the
computational expense of computing the intake flow for
the range of swirl velocities considered here (Rs = 1.44
7.12), solid-body-rotation was used for initialization of
the tangential velocity in this study. The usual practice
of initializing the axial velocities by linearly interpolating
between the piston surface and the head, and initializing
the radial velocities to zero was followed.
The spray and combustion sub-models employed are
described more fully elsewhere ([15],[16]). The
characteristic time combustion sub-model [17], is of
greatest relevance to the present study. In this model,
the instantaneous production and oxidation rates of the
important combustion species are not calculated directly
via either simplified or detailed chemical mechanisms.
Rather, the rate of change of each species is calculated
as proportional to the difference between the actual
species mass fraction and its instantaneous thermo-
dynamic equilibrium value. The constant of propor-
tionality (the inverse of the characteristic time) is a
weighted combination of a chemical timescale and a
turbulent timescale, with progressively greater weighting
toward the turbulent scale as the combustion proceeds.
RESULTS AND DISCUSSION
The results of this study are presented in four main parts.
In the first part, the variation in CO emissions with swirl
ratio and injection timing is reported. Second, the source
of the influence of swirl ratio on CO emissions is clarified
through examination of the heat release characteristics,
numerical simulations and in-cylinder imaging of
combustion luminosity. Third, the factors potentially
responsible for the observed dependency of CO
emissions on injection timing are examined. Finally, we
examine the influence of both swirl ratio and injection
timing on fuel conversion efficiency.
CO EMISSION TRENDS
Figure 3 shows the engine-out CO emissions as a
function of both swirl ratio and injection timing (SOI).
Two features are particularly notable:
First, at fixed injection timing, there is generally an
optimal swirl ratio that gives the lowest CO emission.
The minimum CO emission is typically measured near
the swirl ratio of 2.59. As noted in the Introduction, at
this high dilution rate the dominant source of CO
emission is thought to be under-mixed fuel. Thus, the
7 4 Skip-Fired
CO Emission 3500
6 3000
Swirl Ratio
5 2500
4 2000
3 1500
1000
2
[ppm]
-30 -25 -20 -15 -10
SOI [CAD ATDC]
Figure 3 Effect of swirl ratio and injection timing on CO
emission
observed optimal swirl ratio is somewhat unexpected.
Higher swirl ratios are typically thought to enhance
turbulence production and mixing processes, resulting in
more rapid, complete combustion [18]. Consequently, if
under-mixed fuel is the dominant source of CO
emissions, we would anticipate a continuous
improvement in CO emissions as the swirl level is
increased.
The existence of an optimal swirl level suggests that a
mechanism by which the highest swirl levels impede
mixing and combustion is present. However, the
traditional mechanism by which excessive swirl is
thought to impede mixing, by limiting jet penetration
(e.g. [19],[20]) is not believed to be significant for the
small bore diameters and high injection pressures that
characterize modern automotive diesel engines. This
belief is supported by Fig. 4, which shows the fuel mass
distributions predicted for various swirl ratios at -17 CAD
ATDC. In these simulations, SOI was fixed at -22.25
CAD and the crank angle shown occurs shortly after the
end of injection. A high fuel mass fraction is observed
near the bowl lip at all swirl ratios, indicating penetration
of the fuel jet to the maximum possible extent. At the
higher swirl ratios, more fuel is observed in the cut-plane
shown (22 down-swirl of the fuel jet) due to greater
convection by the swirl velocity.
The second dominant feature of Fig. 3 is the rapid
decrease in CO emission observed at fixed swirl ratio
when the injection timing is advanced. This behavior is
also apparent in Fig. 1. The decrease is larger at the
higher swirl ratios. For a swirl ratio of 7.12, CO emission
was reduced by over 83% as the injection timing was
advanced from -22.25 to -30.25 CAD ATDC.
In the following sections we investigate the causes of
these two behaviors.
OPTIMAL SWIRL RATIO
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

80

App. Heat Release Rate [J/CAD]

R
70 s

1.44
60 2.59
50 3.77
4.94
40 7.12
30
20
10
0
0.6

Cumulative AHRR / mf QLHV

0.5

0.4

0.3

0.2 Increasing Swirl

0.1

0
-15 -10 -5 0 5 10 15 20
Crank Angle [CAD]
Figure 5 Variation in the apparent heat release rate and
the cumulative heat release rate as the swirl ratio is
varied. SOI = -22.25 CAD

Beyond TDC the cumulative heat release curves

Figure 4 Predicted fuel mass distribution for various
swirl ratios at -17 CAD ATDC: SOI = -22.25 CAD,
Pinj = 800 bar. The r-z plane velocity vectors are also
shown superimposed
continue to diverge, but by this time a clear ordering with
swirl ratio in the amount of heat released has been
established. This ordering persists until EVO. The
behavior of the cumulative heat release curves suggests
that the mixing processes responsible for this ordering
commence between -5 CAD and TDC, and endure for
several crank degrees thereafter.

The ordering in the cumulative heat release is also found

As a first step toward examining the cause of the optimal
swirl ratio exhibited in Fig. 3, the variation in the
measured apparent heat release rate and in the
integrated, cumulative heat release rate with flow swirl
level are shown in Fig. 5 for fixed SOI = -22.25 CAD.
Note that, as observed previously for conventional diesel
combustion over a more limited range of swirl ratios [12],
increased swirl is found to reduce the ignition delay and
to advance the phasing of the early portion of the heat
release.
A second, more important observation can be made
from the cumulative heat release behavior shown in the
lower part of Fig. 5. By about -4 CAD, the heat release
rate observed for Rs = 7.12 slows considerably, followed
by slowing of the heat release observed at Rs = 4.94 and
Rs = 3.77.
to correlate inversely with the engine-out CO emissions.
That is, lower CO emissions are observed for swirl ratios
providing the highest cumulative heat release.
Accordingly, insight into the mechanisms responsible for
the CO emission behavior are likely to be found by
examining the mixing processes occurring during the
latter portion of the combustion process, beyond -5 CAD.
From Fig. 5, it is also apparent that, in addition to high
CO emissions, the high swirl cases will suffer from
reduced fuel economya topic that is discussed in
greater detail below.
To help clarify the cause of the behavior seen in Figs. 3
and 5, the flow structures and spatial distributions of CO
(or partially-burned fuel) and O2 computed in the
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
Figure 6 The mean r z plane flow structure and the
spatial distribution of CO at 3 CAD for various swirl
ratios, as predicted by numerical simulation.
SOI = -22.25 CAD, cut-plane 22 down-swirl from the
fuel jet
numerical simulation were examined. In our past work,
we have shown that the mean (bulk) flow structures
predicted by the simulation capture the measured mean
flow structures remarkably well [21]. However, this
comparison was performed for a single, moderate swirl
ratio. To extend this validation to the higher swirl ratios
employed here, we have measured axial profiles of the
tangential velocity in the central region of the bowl at
each swirl ratio, and compare these to the simulation
results in the Appendix. Although some differences exist,
overall the comparison shows that the axial distribution
of the tangential velocity is similar to the computed
results, and evolves in a like manner.
Figure 6 shows the variation in the r z plane flow
structures and in the spatial distribution of CO mass
fraction predicted for various swirl ratios at 3 CAD. This
Figure 7 Changes in the spatial distribution of O2 at
3 CAD for various swirl ratios, as predicted by numerical
simulation. SOI = -22.25 CAD, cut-plane 22 down-swirl
from the fuel jet
crank angle is during the mixing-controlled portion of
combustion, just after the secondary peak in the
apparent heat release rates seen in Fig. 5. At this time,
the cumulative heat release rates seen in the lower
portion of Fig. 5 continue to diverge.
At the lowest swirl ratio, Rs = 1.44, the bulk flow
structure is dominated by a single clockwise-rotating
structure that fills the outer portion of the bowl. In the
central region of the bowl, above the pip, a smaller
secondary vortex rotating in the opposite direction is
seen. From Fig. 7, which shows the distribution of the
remaining O2 in the combustion chamber, it is seen that
this secondary vortex transports O2 to the same region
where the dominant vortex transports CO. In this
interfacial region between the two vortices, steep
gradients in swirl velocity (Fig. 8), as well as high rates
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
Figure 8 Changes in the spatial distribution of the swirl
velocity at 3 CAD for various swirl ratios, as predicted by
numerical simulation. SOI = -22.25 CAD, cut-plane 22
down-swirl from the fuel jet
of mean flow deformation in the r z plane ([22],[23])
lead to enhanced turbulence production. Thus, the
observed pair of counter-rotating flow structures forms a
mixing system, wherein large-scale structures transport
the partially-burned fuel and O2 to a common interface,
while simultaneously providing mechanisms to produce
the turbulence needed to achieve rapid small-scale
mixing.
As swirl ratio is further increased to 2.59, the same dual-
vortex flow structure is again apparent. In this case, the
centrifugal forces acting on the high angular momentum
fluid transported into the inner regions of the bowl are
larger, and the lower vortex is smaller due to the
tendency of this high angular momentum fluid to return
to the bowl periphery. Consequently, the upper vortex is
larger and encompasses a greater fraction of the
remaining O2. The mixing system created is thus
anticipated to be more effective than is seen for
Rs = 1.44, an expectation which is supported by the heat
release characteristics seen in Fig. 5. The dual-vortex
structures formed at these swirl ratios are long-lived; the
simulation results suggest that they continue to provide
effective mixing for at least another 5 CAD beyond the
crank angle illustrated in Figs. 68.
The formation of these structures correlates well with the
secondary peak in the apparent heat release rate, seen
near TDC in Fig. 5. Similar dual-vortex structures, whose
formation coincides with increased heat release rates,
have also been observed under late-injection operating
conditions [23].
Secondary peaks in the experimental heat release rate
near TDC are also seen in Fig. 5 for the higher swirl
ratios. Although small, these peaks are both
reproducible and of a magnitude that easily exceeds the
precision of the measured heat release during this
portion of the cycle, approximately 1 J/CAD [12].
Corresponding to the appearance of these peaks, the
numerical simulations also predict the formation of
counter-rotating vortex structures in the higher swirl
cases, though at slightly earlier crank angles. The
important distinction appears to be the lifetime of these
structures, and the manner in which they evolve and
transport the unburned fuel and oxidant later in the cycle.
Already by 3 CAD, Figures 6 and 7 indicate that for
Rs = 3.77 a third vortex structure is forming in the central
portion of the cylinder. The original upper vortex is
displaced outward, and rather than acting to bring the
partially-burned fuel (CO) and O2 to the same interface,
it now predominantly acts to transport the partially-
burned fuel away from the O2 in the center of the bowl.
This effect becomes even more pronounced as the swirl
level increases further. By Rs = 7.12 yet a fourth vortical
structure is apparent, and the original bowl vortex
dominant at Rs = 1.44is now barely visible in the
reentrant portion of the bowl.
Comparison of the CO spatial distributions in Fig. 6 with
the swirl velocity distributions in Fig. 8 reinforces the
above discussion. In general, the partially-burned fuel is
spatially co-located with the high-swirl fluid. As the swirl
ratio increases, this co-location results in greater
average radial locations of the CO, due to the return of
the high-swirl fluid to the combustion chamber periphery.
At moderate swirl ratios (e.g. Rs = 2.59) this is advanta-
geous: not only is the dual-vortex structure formed, but
the partially-burned fuel is positioned to exit the bowl into
the squish volume, where additional turbulence
generated by the exiting flow is anticipated to facilitate
mixing with the remaining O2 in the squish volume.
In contrast, at the highest swirl ratios, co-location with
high-swirl fluid both impedes mixing with the O2 in the
center of the bowl and interferes with the exit of the
partially-burned fuel from the bowl with the squish flow.
Fundamentally, the centrifugal forces acting on the high-
swirl fluid prevent the inward displacement required to
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

Figure 9 Simulation results illustrating the co-location of CO (partially-burned fuel) and high-swirl fluid during the expan-
sion stroke. Rs = 7.12

Figure 10 Images of natural luminosity from soot emissions at three swirl ratios. At higher swirl, the visible soot is lower
and more uniformly fills the bowl. Clockwise from the upper-left corner of each figure are superimposed the transmissivity
of the neutral density filter employed, the exposure time, the crank angle relative to SOI, and the true crank angle at which
the image was obtained

move the fluid over the bowl lip. The co-location of the
CO and the high-swirl fluid therefore persists until well
into the expansion stroke, effectively trapping the CO in
the re-entrant portion of the bowl. This is illustrated in
Fig. 9 for the highest swirl case, Rs = 7.12.
To provide additional experimental support for this
picture, the spatial distribution of soot luminosity, and its
temporal evolution, was captured using the high speed
digital video camera. Figure 10 compares the side-view
images of the natural soot luminosity observed at
Rs = 1.44 with that observed at higher swirl ratios,
Rs = 3.77 and Rs = 7.12. Although the intensity of the
soot luminosity is less, images obtained at the higher
swirl ratios clearly show that the luminous soot (which is
anticipated to be co-located with partially-burned fuel) is
found deeper in the bowl and more uniformly fills the
bowl than at the lower swirl ratios. This condition
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

1.5 100

Rich Mass Fraction ( o > 1)

Apparant Heat Release Rate [J/CAD]
CO Mass [mg] SOI: -22.25 CAD ATDC
0.3
In-Cylinder

SOI: -30.25 CAD ATDC

1.0 80 SOI: -22.25 CAD ATDC

0.5 60
In-Cylinder CO mass 0.2
0.0
40 SOI: -30.25 CAD ATDC

Apparent Heat Release Rate

20 0.1

0
-30 -20 -10 0 10 20 0
Crank Angle [CAD ATDC] -40 0 40 80 120
Crank Angle [CAD ATDC]
Figure 11 Spatially-integrated in-cylinder CO mass from
the numerical simulations, and corresponding exper- Figure 12 Numerical simulation of the mass fraction of
imental apparent heat release for different injection the cylinder contents characterized by a (pre-
timings. Rs=7.12 combustion) equivalence ratio greater than 1. Rs=7.12

persists over all later crank angles during which the bowl experimentally-determined apparent heat release rates
is visible. for each case. There is a noticeable difference in the
amount of CO present in the cylinder, in agreement with
In summary, the numerically predicted bulk flow the experimental results shown in Fig. 3. For both cases,
structures provide a mechanism that explains the CO formation started in the middle of the premixed burn
observed optimal swirl ratio. At the lowest swirl ratios, a phase and the peak CO mass is found near the start of
dual-vortex structure forms that enhances the mixing of the mixing-controlled burn phase. Examining the
partially-burned fuel and air. For moderately increased evolution of the CO mass, it is apparent that the latter
swirl ratios, this dual-vortex system appears to be more oxidation rates are similar and that the lower CO mass
effective in enhancing the mixing processes. However, observed for the earlier injection timing is principally due
as swirl is increased further, the bulk structures created to lower initial CO formation or more rapid early
actually impede the late-cycle mixing, even though early oxidation of CO.
mixing and heat release appears to be enhanced.
Ultimately, too high a swirl ratio leads to trapping of The nature of the characteristic-time combustion sub-
partially-burned fuel within the bowl. model is such that CO must be formed from rich
mixtures. For lean mixtures, instantaneous equilibrium
RAPID DECREASE IN CO EMISSION WITH CO concentrations are small and the model will not
ADVANCED SOI predict large CO production rates; i.e., CO emissions
stemming from incomplete combustion of over-mixed
The potential causes of the second major observation fuel will not be captured. Thus, the higher CO mass
made from Fig. 3, a decrease in CO emissions as SOI is formed at the later injection timing must be caused by
advanced, are investigated in this section from several the presence of a greater amount of rich mixture. The
perspectives. Once again we rely heavily on the results fraction of rich mixture found in the cylinder for both
of numerical simulations, which suggest that the initial injection timings, shown in Fig. 12, supports this
degree of fuel-air mixing achieved is a critical factor. To conclusion. At the earlier injection timing, the peak
support this finding, we examine the measured influence fraction of the in-cylinder mass characterized by fuel-rich
on CO emission of injection pressure and ignition delay. conditions is smaller, and the fuel-rich mass fraction is
The correlation of CO emissions with peak flame already decreasing at the onset of combustion near -12
temperature is also examined for evidence that early CO CAD. For the later injection timing, however, the peak
oxidation rates could also be an important factor mass of rich in-cylinder mixture is not reached until after
influencing CO emission. Finally, we scrutinize other TDCwell into the mixing-controlled portion of the
processes that influence the early mixture preparation combustion process.
processincluding wall-wetting, spray-targeting, and
flow turbulence. The better mixture preparation characteristics observed
at the earlier injection timing in Fig. 12 are not the result
Simulation Results of an increased ignition delay. While increased ignition
delay is undoubtedly beneficial, the mixing process itself
Figure 11 shows simulation results for the total, in- is enhanced with the earlier injection timing, such that for
cylinder mass of CO for two different injection timings at equal times after the start of injection, a smaller fraction
the highest swirl ratio, Rs = 7.12. Also shown are of the in-cylinder mass is fuel-rich for the earlier injection
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
4500
P :
inj
CO Emission [ppm]
3500
4 Skip-Fired
2500
1500
500
-30 -25 -20 -15 -10
SOI [CAD ATDC]
Figure 13 Effect of injection pressure and injection timing
on CO emission at swirl ratio of 3.77
timing. This enhanced mixture preparation is reflected in
the higher peak heat release rates seen in Fig. 11 for
SOI=-30.25, behavior which is well-captured in the
simulation results, as are the ignition delay periods
observed for both injection timings.
Examination of the spatial distributions of fuel predicted
numerically suggest that the enhanced mixing observed
with earlier injection is not due to a significant difference
in the flow structures existing during the ignition delay
period, but rather to initial deposition of a significant
fraction of fuel vapor within the squish volume. As the
piston rises, this fuel is returned to the bowl with the
squish flow and results in a better mixed, broader spatial
distribution of fuel vapor than is seen with later injection.
Peak fuel vapor concentrations (and subsequent CO
concentrations), for both injection timings, are found
deep in the bowl, as can be seen in the lower portion of
Fig. 6. In both cases, the CO remains trapped in the
lower regions of the bowl during expansion. For the
earlier injection timing, however, there is far less of it.
Thus, the simulation results suggest that the dominant
cause of the reduced CO emissions with advanced
injection timing is associated with reduced peak CO
mass during the early combustion process. Early mixing
processes, occurring prior to or during the first portion of
the premixed burn, are at least partly responsible for the
reduced CO mass.
The Influence of Injection Pressure and Ignition Delay
The numerical simulations of the previous section
suggest that more rapid initial mixing is an effective
method of reducing CO emissionprovided that CO
emission stems predominantly from under-mixed fuel. A
straight-forward test of this prediction can be made
through increasing injection pressure, a measure that
will enhance initial mixing rates but which is less likely to
significantly impact the late-cycle mixing. As
demonstrated by Fig. 13, increased injection pressure is
found to generally reduce CO emissions, in agreement
Rs=3.77
600 bar 1800
CO Emission [ppm]
800 bar
1000 bar
4 Skip-Fired
1200 bar SOI=-22.25 CAD ATDC
1400
1000
SOI=-28.25 CAD ATDC
SOI=-26.25 CAD ATDC
600
1.4 1.6 1.8 2 2.2
-5
[ms]
id
Figure 14 Correlation of CO emission with the ignition
delay at swirl ratio of 3.77
with the numerical predictions. At the earliest injection
timings, however, increased injection pressure may be
detrimental.
Ignition delay is well-known to increase as injection
timing is advanced, due to the lower ambient
temperature and pressure at earlier SOI [24]. Here, the
ignition delay is defined as the time from SOI until the
time at which 10% of the total heat release has occurred.
The 10% heat release timing correlates well with the
beginning of rapid, high-temperature heat release. As
noted in the introduction, we have previously observed
an inverse correlation at high dilution rates between CO
emission and ignition delay. In general provided spray
targeting, wall impingement, or spray/flow interactions
do not vary significantly with SOI increased ignition
delay should correlate well with a higher degree of fuel-
air premixing. In turn, more premixing will correlate
inversely with CO emission, provided CO emission is
dominated by the combustion of rich mixtures.
Conversely, CO emissions resulting from over-mixed
fuel would be expected to increase with increasing
ignition delay.
The correlation between ignition delay and CO emission
measured for Rs=3.77 is shown in Fig. 14. This
intermediate swirl ratio was selected for detailed
examination because a minimum in CO emission is
observed near SOI =-26 CAD in both Figs. 1 and 3. In
keeping with the predictions of the numerical simulations,
the variable injection pressure tests, and our previous
results, Fig. 14 shows a strong negative correlation of
CO emission with increasing ignition delay, indicating
the benefits of increased early mixing for reducing CO
emission.
At the earliest injection timing shown in Fig. 14,
increased CO emission is observed despite an increase
This view must be adopted with caution. As shown in
the previous section, under some circumstances mixing-
rates can be affected significantly by changes in SOI.
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
Rs=3.77
2000
CO Emission [ppm]
SOI=-22.25 CAD ATDC
4 Skip-Fired
1000
900
SOI=-28.25 CAD ATDC
800
SOI=-26.25 CAD ATDC
700
600
2000 2010 2020 2030
Peak Flame Temperature [K]
Figure 15 Correlation of CO emission with peak flame
temperature at a swirl ratio of 3.77
in ignition delay. This behavior is both repeatable and
reflected in the CO emission behavior seen at other swirl
ratios (see Fig. 3). Furthermore, it is consistent with the
behavior seen at the earliest injection timings of Fig. 13,
where increased injection pressure was no longer
beneficial. It would appear that at very early injection
timing, factors other than ignition delay are influencing
pre-mixing and subsequent CO emissionbehavior that
will be examined more closely below.
Flame Temperature
The evidence presented thus far suggests that
decreased peak CO mass early in the combustion
process is primarily responsible for the decrease in CO
emissions observed with advanced injection timing.
Simulation results suggest that the decreased CO mass
is due to enhanced mixing. However, earlier injection
results in advanced combustion phasing which results in
higher peak cylinder pressures and higher peak
combustion temperatures. Accordingly, it seems
plausible that the reduction in peak CO mass may be
partly associated with the more rapid early oxidation of
CO due to these higher temperatures.
We have selected the peak adiabatic flame temperature
to characterize the CO oxidation rates early in the
combustion process (i.e., during the premixed burn
period). The peak flame temperature is the adiabatic
flame temperature achieved by combustion of a
stoichiometric mixture at the peak core charge
temperature, which is estimated assuming adiabatic
compression of the in-cylinder charge. Although CO
oxidation will always occur at temperatures lower than
this, the relative magnitude of the peak flame
temperature is likely a good indicator of the relative
magnitude of CO oxidation temperatures.
The correlation of CO emission with peak flame
temperature is shown in Fig. 15, where the CO
emissions measured at Rs = 3.77 are plotted on a
logarithmic axis. As anticipated, the peak flame
temperature increases monotonically as injection is
advanced. Notably, as SOI is advanced to -22.25 CAD,
there is little change in CO emission despite a near 30 K
change in peak flame temperature. As SOI is further
advanced to -26.25 CAD, however, a significant re-
duction in CO emission is observed, although the flame
temperature only increases by 5 K. Overall, the
observed correlation between peak flame temperature
and CO emission is not consistent with the proposition
that early CO oxidation significantly influences the
emission level. The subsequent increase in CO emission
as SOI is further advanced to -28.25 CAD, while the
2040 peak flame temperature continues to increase,
reinforces this view.
Spray Penetration and Targeting
An additional important factor related to early mixture
formation processes in small-bore diesels is spray
penetration and targeting. Excessive penetration will
clearly lead to piston wall-wetting, and in extreme cases
can lead to liquid fuel impacting the cylinder liner (e.g.
[25]). Spray targeting is expected to influence the initial
mixture formation process by affecting the initial
splitting of the spray upon striking the bowl lip, and to
further influence the size and location of liquid films that
may be formed on the piston surface.
Figures 16 and 17 illustrate the liquid spray penetration
and wall-film formation, as determined from high-speed
imaging of the liquid sprays. Also shown, below the
experimental images, is the evolution of the gas-phase
fuel distribution predicted in the numerical simulations.
The computed fuel distributions are shown for a vertical
cut-plane 22 downstream of the spray axis. The center
of the piston and location of the bowl lip are scaled to
the spray image and are designated with dashed lines.
The time after the start of injection and the current crank
angle are shown at the right-hand side of each gas-
phase fuel mass image.
For the injection timing of -26.25 CAD seen in Fig. 16,
where the lowest CO emission was observed at the
Rs = 3.77 swirl ratio, the spray tip clearly impacts the
bowl lip and creates a liquid film. In contrast, with
SOI = -22.25 CAD, Fig. 17 shows that the spray tip
reaches to the bowl lip, but little or no liquid film can be
seen. Formation of liquid films is generally thought to
increase UHC emissions (e.g. [26]), and would likely
increase CO emissions as well due to an overall
retarded fuel preparation process. The lower CO
emission observed at the earlier injection timing
suggests, therefore, that formation of wall films is not a
dominant process influencing CO emission here.
A second interesting observation can be made from
Figs. 16 and 17. The predicted gas-phase fuel mass
distributions show that for the earlier, -26.25 CAD
injection timing, some of the fuel penetrates into the
squish area. However, the squish flow induced by the
rising piston pushes this fuel back into the bowl. By the
end of the sequence shown, the fuel distribution is
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
Figure 16 Visualized spray impingement on the bowl lip
and predicted fuel mass distribution: SOI = -26.25 CAD,
Rs = 3.77
almost identical for both injection timings. Accordingly,
the initial distribution of the vapor phase fuel does not
appear to be a dominant factor in determining the CO
formation and ultimately the CO emission level. This
stands in contrast to our previous discussion of the fuel
vapor distributions observed for Rs = 7.12. Recall that, in
that case, significantly more advanced injection timing
was considered.
Figure 17 Visualized spray impingement on the bowl lip
and predicted fuel mass distribution: SOI = -22.25 CAD,
Rs = 3.77
At these more advanced injection timings, liquid films
can be expected to form not only on the bowl rim, but
also on the piston top. Figure 18 shows an idealized
view of the spray targeting on the piston for various
crank angles. With a start of injection at -26 CAD and
later, we predominantly expect wetting of the bowl rim,
as was seen in Fig. 16. Minor wetting of the piston top
may occur with SOI = -26 CAD, but only from the earliest
portion of the injection event. For earlier injection timings,
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
-5 CAD
-15 CAD
-22 CAD
-26 CAD
-30 CAD
Figure 18 Idealized depiction of the un-deflected spray
targeting for various crank angles
wetting of the piston top is expected to become
increasingly important, due to both increased
penetration and the spray targeting. At high swirl ratios,
we anticipate that vaporization of these films will be
more rapid, and their potential influence on emissions
diminished. Piston top wall films are thus a possible
explanation for the increased CO emissions seen in Fig.
3 for the earliest injection timings and lower swirl ratios.
Turbulence Enhancement
A final factor that may influence the early mixture
preparation process and increase mixing rates during
the ignition delay period is enhanced flow turbulence. In
earlier work we have shown that, over the course of the
latter compression stroke, production of turbulence by
bulk compression acting on the isotropic normal stresses
can often dominate [16]. For a uniform density flow, this
production can be expressed as:
1 dV
Piso 2 k
3 V dt
Here, the production of turbulence is directly proportional
to the level of pre-existing turbulent kinetic energy k, and
the negative of the logarithmic rate-of-change of cylinder
0
TDC
D
-5
z [mm]
-10
-15
-20 -25 CA
AD ATDC
0
r = 13.6
3 mm
Figure 19 Velocity measurement location and piston
position at -25 CAD ATDC: r = 13.6 and various axial (z)
locations
greater period of time, and could significantly enhance
the early mixture preparation process.
To examine this potential source of enhanced mixing,
tangential velocity data were acquired at a radius of
r = 13.6 mm and axial locations of z = 2 20 mm below
the cylinder head. The data discussed below were
obtained at the locations z = 4 10 mm, which are
shown in Fig. 19, along with superimposed bowl outlines
which show the piston position relative to these
measurement locations at various crank angles. At the
early injection timings considered here, these
measurement locations were approximately centered
along the path of a fuel spray during the injection event
(cf. Fig. 18). Later, these locations are characteristic of
locations within the central region of the bowlwhere
much of the later mixing occurs. All measurements were
obtained in a plane 22 down-swirl of a fuel jet.
The measured temporal evolution of the mean flow
velocity and RMS fluctuations are shown in Figs. 20 and
21, respectively. Velocities measured under both
motored and fired operation are presented in each figure.
The swirl ratio shown is 7.12. Recall that, as the injection
timing is advanced, the maximum reduction in CO
emission was observed at Rs = 7.12 (see Fig. 3).
Measurements obtained at injection timings of -22.25
and -30.25 CAD are presented.
The mean flow evolution, shown in Fig. 20, is consistent
with both physical expectations and with the results of
numerical simulations. For motored operation, Fig. 20
shows that the tangential velocity increases in the latter
(1) part of the compression stroke as high angular
momentum is transferred into the bowl by the squish
flow. With fuel injection, little influence of the injection
event is observed until after injection has ended. After
the fuel injection event, a distinct perturbation in the
tangential velocity is observed at all injection timings and
volume - 1 V dV dt  a quantity which is large between measurement locations, associated with a displacement
about -30 and -5 CAD. Thus, turbulent energy injected of the high angular momentum fluid near the bowl rim by
earlier is amplified by the compression process over a the fuel spray. This perturbation rapidly subsides,
however, and beyond about -10 CAD there is little
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

6
Rs = 7.12 Rs = 7.12
10 Pinj = 800 bar Pinj = 800 bar
r = 13.6 mm 5 r = 13.6 mm
8 z = 4 mm
4
6 3 Motored
Fired, -22.25 SOI
z = 4 mm Fired, -30.25 SOI
4 2
Fuel Injection Event
Motored
2 Fired, -22.25 SOI 1
Fired, -30.25 SOI
Fuel Injection Event
0
10
5
8 z = 6 mm
4
Motored
6 Fired, -22.25 SOI
3 Fuel Injection Event
Fired, -30.25 SOI

Tangential RMS Velocity [Sp]

z = 6 mm
4 2
Tangential Velocity [Sp]

Fuel Injection Event Motored

2 Fired, -22.25 SOI 1
Fired, -30.25 SOI

0
10
5
8 z = 8 mm
4
Fuel Injection Event Motored
6
3 Fired, -22.25 SOI
Fired, -30.25 SOI
z = 8 mm
4 2
Motored
2 -22.25 SOI 1
Fuel Injection Event
-30.25 SOI

0
10
5
8 z = 10 mm
4 Motored
Fired, -22.25 SOI
6 Fired, -30.25 SOI
3
Fuel Injection Event
z = 10 mm
4
2
Motored
2 Fuel Injection Event Fired, -22.25 SOI 1
Fired, -30.25 SOI

0 0
-30 -20 -10 0 10 -30 -20 -10 0 10
Crank Angle [CAD ATDC] Crank Angle [CAD ATDC]
Figure 20 Measured mean velocity in tangential direction Figure 21 Measured tangential velocity RMS fluctuations
at r = 13.6 and various axial (z) locations for Rs =7.12 at r = 13.6 and various axial (z) locations for Rs =7.12
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

evidence that the mean flow evolves in a significantly

different manner than the motored flow. For SOI = - 7 Fuel Conversion Efficiency 34
30.25 CAD combustion begins at about -11 CAD, while
for SOI = -22.25 CAD combustion begins near -8 CAD 6 32

Swirl Ratio
(see Fig. 11). Hence, at these locations, we see little
effect of combustion on the mean flow evolution. 5 30
The corresponding RMS fluctuations are shown in
Figure 21. Changes in RMS velocity fluctuations could
4 28
be due to cycle-to-cycle fluctuations in the mean flow
structures or to variability of the fuel injection event, 3 26
rather than additional turbulence. However, previous
cycle-resolved analysis has shown that enhanced RMS 2 24
fluctuations observed after fuel injection correlate well
with enhanced small-scale, high-frequency fluctuations -30 -25 -20 -15 -10 [%]
characteristic of turbulent motions [27]. Consequently,
we believe that the increased RMS fluctuations seen in SOI [CAD ATDC]
Fig. 21 represent, at least in part, increased flow
turbulence. Figure 22 Effect of swirl ratio and injection timing on fuel
conversion efficiency
In general, increased RMS fluctuations are observed
coincident with the period in which the largest
perturbation in the mean flow structure is seen. At the 7
highest axial location, z = 4 mm, these fluctuations are Combustion Efficiency
particularly large, and are both larger in magnitude and
94
of a longer duration for the earlier injection timing. Later Swirl Ratio 6 92
in the cycle, a more moderate increase in fluctuations
over the motored fluctuations is seen, which typically 5 90
persists for at least 10 CAD. Considering the -11 and -8
CAD start-of-combustion for the -30.25 and -22.25 CAD 4 88
injection timings, respectively, we see that this more
moderate increased turbulence, which appears earlier 86
for the earlier injection timing, may also increase pre- 3
combustion mixing. Overall, the measurements indicate 84
that with earlier injection, turbulence levels are higher 2 82
and act on the pre-combustion mixture for a longer
period. Increased turbulent mixing may therefore assist -30 -25 -20 -15 -10 [%]
in the early mixture formation process, complementing
the longer ignition delay. However, it must be recognized
SOI [CAD ATDC]
that the spatial locations sampled with LDV are sparse,
and increased turbulence will not increase mixing rates if Figure 23 Effect of swirl ratio and injection timing on
it occurs in regions with a homogeneous mixture combustion efficiency
composition.

FUEL CONVERSION EFFICIENCY
In the previous sections we have examined how swirl
ratio and combustion phasing (injection timing) influence
CO emission, which is an indication of complete,
efficient combustion. Complete combustion is a pre-
requisite for obtaining a high fuel conversion efficiency
Kfc . However, other factors also influence Kfc . In this
section our goal is to clarify the broader impact of
variations in swirl ratio and injection timing on Kfc . In
pursuing this goal, we consider only the gross indicated
fuel conversion efficiency. The net fuel conversion
efficiency may be influenced through other factors (e.g.
increased pumping work associated with high-swirl port
designs) that are not addressed here.
The gross indicated fuel conversion efficiency is the ratio
of the gross indicated work W to the chemical energy in
the injected fuel, calculated from the product of the fuel
mass and the lower-heating value QLHV of the fuel
W
K fc (2)
m f Q LHV
As an aid to understanding the influence of various
factors on Kfc (including CO emission and combustion
efficiency), we have previously proposed [9] that Kfc be
decomposed as a product of three separate efficiencies
K
 Kwc

 K hl

c 

W Qchem  Qhl Qchem
K fc (3)
Q  Q Q m Q
chem hl chem f LHV
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
7 Work Conversion Efficiency
40
6 emissions (see Fig. 3) using the expression
Swirl Ratio
5 38
4 36
3 34
2 32
[%]
-30 -25 -20 -15 -10
SOI [CAD ATDC]
Figure 24 Effect of swirl ratio and injection timing on
work conversion efficiency
7 Heat Loss Efficiency 95
6 90
Swirl Ratio
5 85
4 80
3 75
2 70
-30 -25 -20 -15 -10 [%]
SOI [CAD ATDC]
Figure 25 Effect of swirl ratio and injection timing on
heat-loss efficiency
Because the difference between the chemical heat
release Q chem and the energy losses due to heat transfer
and crevice flows Q hl is the apparent net heat release,
the first term on the right-hand-side (RHS) represents
the fraction of the apparent heat release that is
converted into useful workthe work conversion
efficiency K wc . Changes in this efficiency are primarily
due to changes in the combustion phasing. The center
term on the RHS of Eq.(3) is a straight-forward
representation of the fraction of the chemical heat
release that is manifested as apparent heat releasea
heat-loss efficiency Khl , and the last term on the RHS is
the combustion efficiency Kc .
The fuel conversion efficiency, calculated from Eq.(2), is
shown in Fig. 22. For comparison, an estimate of the
combustion efficiency is calculated from the CO
'h 0f ,CO  'h 0f ,CO2
Kc 1  f (4)
'H R
In Eq.(4), f represents the fraction of fuel carbon found in
the CO emissions, 'H R is the negative of the fuel lower
heating value, and 'h 0f is the heat of formation of the
sub-scripted species. Additional details can be found in
Ref. 9. Combustion efficiencies estimated from Eq.(4)
are shown in Fig. 23.
Although many similarities exist between Figs. 22 and
23, there are also distinct differences that bear noting:
1) While the combustion efficiency remains high in the
upper-left quadrant of Fig. 23, K fc is decreasing at
the higher swirl ratios.
2) The best K fc is observed at somewhat lower swirl
ratios and more retarded injection timing than the
best Kc
3) In the lower-right quadrant of Fig. 22, a rapid
decrease in K fc is observed. No similar decrease
can be seen in Kc .
4) In the lower-left quadrant, a local minimum in K fc is
seen that is not present for Kc
To assist in the explanation of these differences, the
work conversion efficiency is presented in Fig. 24. All of
the differences between K fc and Kc identified above can
be seen to correlate well with the variation of K wc with
SOI and with swirl ratio. First, the work conversion
efficiency can be seen to decrease gradually in the
upper-left quadrant of Fig. 24. Thus, decreased K fc at
early SOI and high swirl ratio can be, at least partially,
explained by decreased K wc . Second, peak work
conversion efficiencies are observed at lower swirl ratios
and at more retarded injection timing than peak Kc , an
observation that helps explain the second difference
between Figs. 22 and 23 enumerated above. Third, the
behavior of K fc in both the lower-left and the lower-right
quadrants is mirrored in K wc .
In addition to explaining the differences between K fc and
Kc , the trends in K wc seen in Fig. 24 also serve to
illustrate the mechanisms by which changes in injection
timing and swirl affect K fc . For example, a general
characteristic of Fig. 24 is a decrease in K wc for both
early and late injection timings. In the former case, early
heat release results in negative work as the combustion-
induced pressure rise occurs before compression has
ended. Although some of this work is recovered during
the subsequent expansion, due to heat transfer and
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
blow-by some of this work is inevitably lost. In the latter
case, heat release is sufficiently retarded that much of
the volume expansion has already occurred.
A second observation is the low K wc in the upper-central
region of Fig. 24, which coincides with a minimum in the
combustion efficiency. As discussed in the section
describing the optimal swirl ratio, the cause of this
minimum in Kc is thought to be trapping of CO in the
bowl. Apparently, this trapping also retards the heat
release thereby influencing K wc .
Finally, K wc also exhibits a general decreasing trend as
the swirl level increases. As discussed elsewhere [28],
the work conversion efficiency is influenced by heat
transfer losses that occur in a non-combusting cycle.
The higher heat transfer losses incurred in high swirl
flows result in a decreased K wc . Consequently, the
influence of heat loss is not fully-confined to the heat-
loss efficiency.
Although Q chem and Q hl have not been measured
directly, the heat-loss efficiency K hl can be estimated
from the remaining efficiencies via Eq.(3). The results
must be employed cautiously, however, as the
combustion efficiencies calculated from Eq.(4) will likely
be overestimated, due to the neglect of the energy
content in the unburned hydrocarbons and other
unmeasured species. Provided that these species follow
the same trends as the CO emissions, their neglect
implies that the changes observed in Kc are
underestimated.
Fig. 25 illustrates the resulting estimates of K hl . Note
that the overall variation in K hl is larger than is seen in
either K wc or Kc . However, because the variation in Kc is
underestimated, the variation seen in K hl is likely
excessive. Nevertheless, the magnitude of the true
variation is likely to be significant, and the trends
observed are physically plausible and bear consideration.
First, a general trend of decreasing heat-loss efficiency
is seen as swirl is increaseda result that is not
unexpected. More interestingly, the heat loss efficiency
shows a greater variation with injection timing. This
variation might be expected based on the simple
observation that there is more time available for heat
transfer at the earlier injection timing. At least three other
factors also contribute to this behavior, though. First,
with earlier combustion phasing the average surface
area-to-volume ratio of the hot, in-cylinder gases is
increased. Second, mixture that burns first and then is
compressed reaches a higher temperature than mixture
that burns after compression. Third, convective heat
transfer coefficients are expected to be larger at the
higher peak cylinder pressures observed with early
injection. Each of these factors is expected to increase
heat transfer losses as injection is advanced.
SUMMARY AND CONCLUSION
CO emission and fuel conversion efficiency are
examined in a highly-dilute, low-temperature diesel
combustion regime as the injection timing and swirl ratio
are varied. To clarify the dominant factors influencing the
emission and efficiency behavior, the cylinder pressure,
engine-out emissions, and flow velocities are measured.
High speed video imaging and multi-dimensional numer-
ical simulations of the combustion process are also
performed.
The CO emission behavior exhibits two interesting
trends:
First, an optimal swirl ratio exists at which the lowest CO
emission and best fuel conversion efficiency are
observed. Heat release analysis, the results of the
numerical simulations, and imaging of combustion
luminosity indicate that this behavior is likely due to
mixing processes occurring after the premixed burn
period. At the lowest swirl ratios, a dual-vortex vertical
plane structure forms that enhances the mixing of
partially-burned fuel and air. This structure is most
effective at a moderate swirl ratio (| 2.5). At higher swirl
ratios, the flow structures formed impede mixing, and
can lead to trapping of partially-burned fuel within the
bowl.
Second, CO emission generally exhibits a rapid
decrease from the maximum as SOI is advanced,
particularly at the highest swirl ratios. The numerical
simulations indicate that, at a fixed swirl ratio, earlier
injection timing leads to enhanced pre-combustion
mixing and hence lower peak in-cylinder CO mass. The
enhanced mixing is due not only to increased ignition
delay, but also to increased mixing rates under high-
swirl conditions. A reduction in CO emission with
increased pre-mixing CO implies that CO emission
stems predominantly from under-mixed fuel (rich
mixtures), a finding which is supported experimentally by
a strong tendency toward reduced CO emission with
increased ignition delay and with increased injection
pressure. Correlation of CO emission with peak
adiabatic flame temperature suggests that temperature
related changes to the early CO oxidation rates are not a
significant factor influencing peak in-cylinder CO mass
and subsequent engine-out emission. Spray penetration
and spray targeting, with concomitant liquid film
formation was also examined as a potential source of
modified premixing. Significant liquid film formation
along the bowl lip was observed at early injection timings.
However, low CO emission was observed under these
conditions, signifying that wall-film formation in this
region is not a dominant factor influencing CO emission.
At the earliest injection timings, the likely existence of
wall films on the top of the piston is identified as a
potential source of increasing CO. Finally, the possibility
that enhanced pre-combustion turbulence was
promoting mixing under early-injection conditions was
examined. Increased turbulence was observed with
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
earlier injection timing, and likely assists in the initial
mixing taking place in the ignition delay period.
The impact of swirl level and injection timing on the fuel
conversion efficiency is also assessed. Qualitative
differences are observed between the fuel conversion
efficiency and the combustion efficiency derived from the
CO emissions. These differences can be accounted for
by changes in the work conversion efficiency
(associated primarily with combustion phasing) and by
heat transfer losses.
ACKNOWLEDGMENTS
Support for this research was provided by the U.S.
Department of Energy, Office of FreedomCAR and
Vehicle Technologies. The research was performed at
the Combustion Research Facility, Sandia National
Laboratories, Livermore, California. Sandia is a
multiprogram laboratory operated by Sandia Corporation,
a Lockheed Martin Company, for the United States
Department of Energys National Nuclear Security
Administration under contract DE-AC04-94AL85000. The
BK21 and Future Vehicle Technology Development
Corps. of Korea supported Sanghoon Kook's visiting
research. The authors express their appreciation to Mark
Musculus and Lyle Pickett for providing the high speed
camera and the Matlab source code to calculate the
adiabatic flame temperature.
REFERENCES
1. Cowland, C., Gutmann, P. and Herzog, P.L.,
Passenger Vehicle Diesel Engines for the US, SAE
Paper No. 2004-01-1452, 2004.
2. Kimura, S., Aoki, O., Kitahara, Y. and Aiyoshizawa
E., Ultra-Clean Combustion Technology Combining
a Low-Temperature and Premixed Combustion
Concept for Meeting Future Emission Standards,
SAE Paper No. 2001-01-0200, 2001.
3. Akihama, K., Takatori, Y., Inagaki, K., Sasaki, S.,
and Dean, A.M., Mechanism of the Smokeless Rich
Diesel Combustion by Reducing Temperature, SAE
Paper No. 2001-01-0655, 2001.
4. Walter, B. and Gatellier, B. Development of the
High Power NADI Concept Using Dual Mode Diesel
Combustion to Achieve Zero NOx and Particulate
Emissions, SAE Paper No. 2002-01-1744, 2002.
5. Weibck, M., Csat, J., Glensvig, M., Sams, T. and
Herzog, P., Alternative Brennverfahren ein Ansatz
fr den Zuknftigen Pkw-Dieselmotor, MTZ: vol 64:
pp 718-727, 2003.
6. Hasegawa R. and Yanagihara, H., HCCI
Combustion in DI Diesel Engine, SAE Paper No.
2003-01-0745, 2003.
7. Sjberg, M., and Dec, J. E., An Investigation into
the Lowest Acceptable Combustion Temperatures
for Hydrocarbon Fuels in HCCI Engines,
Proceedings of the Combustion Institute, Vol. 30,
pp.2719-2726, 2005.
8. Adomeit, P., Pischinger, S., Becker, M., Rohs, H.,
and Greis, A., Laser Optical Diagnostics and
Numerical Analysis of HSDI Combustion, THIESEL
2004: Conference on Thermo- and Fluid Dynamic
Processes in Diesel Engines, Sept. 8-10, Valencia,
Spain, 2004.
9. Kook, S., Bae, C., Miles, P. C., Choi, D, and Pickett,
L. M., The Influence of Charge Dilution and
Injection Timing on Low-Temperature Diesel
Combustion and Emissions, SAE Paper 2005-01-
3837, 2005.
10. Timoney, D.L., Smoke and Fuel Consumption
Measurements in a Direct Injection Diesel Engine
with Variable Swirl, SAE Paper No. 851542, 1985.
11. Van Gerpen, J.H., Hwang, C.-W., and Borman, G.L.,
The Effects of Swirl and Injection Parameters on
Diesel Combustion and Heat Transfer, SAE Paper
No. 850265, 1985.
12. Miles, P. C., The Influence of Swirl on HSDI Diesel
Combustion at Moderate Speed and Load, SAE
Paper 2000-01-1829, 2000.
13. Miles, P.C., Choi, D., Pickett, L.M., Singh, I.P.,
Henein, N., RempelEwert, B.H., Yun, H., and Reitz,
R. D., Rate-Limiting Processes in Late-Injection,
Low-Temperature Diesel Combustion Regimes,
THIESEL 2004: Conference on Thermo- and Fluid
Dynamic Processes in Diesel Engines, Sept. 8-10,
Valencia, Spain, 2004.
14. Amsden, A. A., KIVA-3V: A Block Structured KIVA
Program for Engines with Vertical or Canted Valves,
Los Alamos National Laboratory Report No. LA-
13313-MS, 1997.
15. RempelEwert, Bret H., The Influence of Swirl on
Flow, Fuel Injection, and Emissions in an HSDI
Diesel Engine, Masters Thesis, Dept. of
Mechanical Engineering, University of Wisconsin
Madison, 2004.
16. Miles, P., Choi, D., Megerle, M., RempelEwert, B.,
Reitz, R. D., Lai, M. D., Sick, V., The Influence of
Swirl Ratio on Turbulent Flow Structure in a Motored
HSDI Diesel Engine A Combined Experimental
and Numerical Study, SAE Paper 2004-01-1678,
2004.
17. Kong, S.-C., Han, Z., and Reitz, R.D., The
Development and Application of a Diesel Ignition
and Combustion Model for Multidimensional Engine
Simulation, SAE Paper 950278, 1995.
18. Ikegami, M., Role of Flows and Turbulent Mixing in
Combustion and Pollutant Formation in Diesel
Engines, COMODIA 90: Proceedings of the
International Symposium on Diagnostics and
Modeling of Combustion in Internal Combustion
Engines, 35 September, Kyoto, Japan, 1990.
19. Khan, I.M., Wang, C.H.T., and Langridge, B.E.,
Effect of Air Swirl on Smoke and Gaseous
Emissions from Direct-Injection Diesel Engines,
SAE Paper No. 720102, 1972.
20. Binder, K. and Hilburger, W., Influence of the
Relative Motions of Air and Fuel Vapor on the
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

Mixture Formation Processes of the Direct Injection

Diesel Engine, SAE Paper No. 810831, 1981.
21. Miles, P., Megerle, M., Sick, V., Richards, K., Nagel,
Z., and Reitz. R. D., The Evolution of Flow Struc-
tures and Turbulence in a Fired HSDI Diesel Engine,
SAE Paper 2001-01-3501, 2001.
22. Miles, P.C., RempelEwert, B.H., and Reitz, R.D.,
Squish-Swirl and Injection-Swirl Interaction in Direct
Injection Diesel Engines, ICE 2003: 6th International
Conference on Engines for Automobiles, Capri,
Naples, Italy, Sept. 14-19, 2003
23. Miles, P.C., In-Cylinder Turbulent Flow Structure in
Direct-Injection, Swirl-Supported Diesel Engines,
Ch. 1 in Flow and Comb. in Automotive Engines, C.
Arcoumanis, ed. Springer-Verlag, in press.
24. Heywood, J. B., Internal Combustion Engine
Fundamentals, International Edition, McGraw-Hill,
Inc., 1988.
25. Akagawa, H., Miyamoto, T., Harada, A., Sasaki, S.,
Shimazaki, N., Hashizume, T., and Tsujimura, K.,
Approaches to Solve Problems of the Premixed
Lean Diesel Combustion, SAE Paper 1999-01-0183,
1999.
26. Henein, N. Analysis of Polllutant Formation and
Control and Fuel Economy in Diesel Engines, Prog.
Energy Combust. Sci., Vol. 1, pp.165-207, 1978.
27. Miles, P., Megerle, M., Hammer, J., Nagel, Z., Reitz,
R. D., Sick, V., Late-Cycle Turbulence Generation
in Swirl-Supported, Direct-Injection Diesel Engines,
SAE Paper 2002-01-0891, 2002.
28. Choi, D. and Miles, P.C., A Parametric Study of
Low-Temperature, Late-Injection Combustion in an
HSDI Diesel Engine, COMODIA 2004: 6th Intl.
Symp. On Diagnostics and Modeling of Combustion
in IC Engines, Aug 2-5, Yokohama, 2004.

NOMENCLATURE

ATDC After Top Dead Center

CAD Crank Angle Degrees

EGR Exhaust Gas Recirculation

EVC Exhaust Valve Closing

EVO Exhaust Valve Opening

IMEP Indicated Mean Effective Pressure

IVC Intake Valve Closure

IVO Intake Valve Opening

SOI Start of Injection

Sp Mean Piston Speed

 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

Rs = 2.59, SOI = -22.25 CAD ATDC

0 0
APPENDIX: COMPARISON OF MEASURED AND
-5 -5
SIMULATED TANGENTIAL VELOCITIES
-10 -10

Figures A1 ~ A5 compare the evolution of the measured

-15 -15
mean tangential velocities with the numerical predictions Experiment
pe
Computation
m
for Rs = 1.44 ~ 7.12. The velocity data are normalized -20 -25 CAD -20 -5 CAD
with the piston speed Sp: 4.25 m/s at 1500 rpm. At each 0 0

crank angle in the figures, axial profiles of the tangential -5 -5

mean velocities are depicted on a schematic showing
the relative position of the piston. The velocities were -10 -10

measured at a fixed radial location of r = 13.6 mm, which -15 -15

is represented by the vertical, dashed lines. The crank
angle range investigated commences well before -20 -20 TDC

z [mm]
-20 CAD

z [mm]
0 0
injection occurs at -22 CAD ATDC and continues to the
end of the premixed combustion at around 10 CAD -5 -5
ATDC.
-10 -10

The overall development of the measured and predicted -15 -15

mean velocity fields is very similar, though several
-20 -15 CAD -20 5 CAD
discrepancies in the detailed shape of the profiles exist.
0 0
For example, the measured velocities are lower than the
computation at many locations lower in the bowl. These -5 -5
data are currently being re-assessed for noise
-10 -10
contributions arising from scattering from the bowl floor.
-15 -15

A second discrepancy is under-predicted velocities at -

-20 -10 CAD -20 10 CAD
15 CAD ATDC for Rs = 3.77 ~ 7.12. At this crank angle, Measurement
e Location Measurement
e Location
0 2 4 6 8 10 12 0 2 4 6 8 10 12
the calculations show a marked influence of the fuel
Tangential Velocity [Sp] Tangential Velocity [Sp]
injection event (SOI=-22.25 CAD ATDC) near z = -6 mm,
resulting in a minimum in the tangential velocities. This
Figure A2 Comparison of the axial profile of the
measured mean tangential velocities with the results of
Rs = 1.44, SOI = -22.25 CAD ATDC the numerical simulation. Rs = 2.59.
0 0
Rs = 3.77, SOI = -22.25 CAD ATDC
-5 -5 0 0

-10 -10 -5 -5

-15 -15 -10 -10

Experiment
pe
Computation
m
-20 -20 -15 -15
-25 CAD -5 CAD Experiment
pe
0 0 Computation
m
-20 -25 CAD -20 -5 CAD
-5 -5 0 0

-10 -10 -5 -5

-15 -15 -10 -10

-20 -20 -15 -15

TDC
z [mm]

-20 CAD
z [mm]

0 0
-20 -20 TDC
z [mm]

-20 CAD
z [mm]

-5 0 0
-5

-10 -10 -5 -5

-15 -10 -10

-15

-20 -20 -15 -15

-15 CAD 5 CAD
0 0
-20 -15 CAD -20 5 CAD
0 0
-5 -5

-5 -5
-10 -10

-10 -10
-15 -15

-15 -15
-20 -10 CAD -20 10 CAD
Measurement
e Location Measurement
e Location
0 2 4 6 8 10 12 0 2 4 6 8 10 12 -20 -20
-10 CAD 10 CAD
Measurement
e Location Measurement
e Location
Tangential Velocity [Sp] Tangential Velocity [Sp]
0 2 4 6 8 10 12 0 2 4 6 8 10 12

Figure A1 Comparison of the axial profile of the Tangential Velocity [Sp] Tangential Velocity [Sp]
measured mean tangential velocities with the results of
the numerical simulation Rs = 1 44 Figure A3 Comparison of the axial profile of the
measured mean tangential velocities with the results of
the numerical simulation. Rs = 3.77.
 Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017

Rs = 4.94, SOI = -22.25 CAD ATDC Rs = 7.12, SOI = -22.25 CAD ATDC
0 0 0 0

-5 -5 -5 -5

-10 -10 -10 -10

-15 -15 -15 -15

Experiment
pe Experiment
pe
Computation
m Computation
m
-20 -20 -20 -20
-25 CAD -5 CAD -25 CAD -5 CAD
0 0 0 0

-5 -5 -5 -5

-10 -10 -10 -10

-15 -15 -15 -15

-20 -20 TDC -20 -20 TDC

z [mm]

-20 CAD

z [mm]
-20 CAD
z [mm]

z [mm]
0 0 0 0

-5 -5 -5 -5

-10 -10 -10 -10

-15 -15 -15 -15

-20 -20 -20 -20

-15 CAD 5 CAD -15 CAD 5 CAD
0 0 0 0

-5 -5 -5 -5

-10 -10 -10 -10

-15 -15 -15 -15

-20 -20 -20 -20

Measurement
e Location
-10 CAD 10 CAD Measurement
e Location
-10 CAD 10 CAD
Measurement
e Location Measurement
e Location

0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12 0 2 4 6 8 10 12
Tangential Velocity [Sp] Tangential Velocity [Sp] Tangential Velocity [Sp] Tangential Velocity [Sp]

Figure A4 Comparison of the axial profile of the Figure A5 Comparison of the axial profile of the
measured mean tangential velocities with the results of measured mean tangential velocities with the results of
the numerical simulation. Rs = 4.94 the numerical simulation. Rs = 7.12.

mm, resulting in a minimum in the tangential velocities. Despite the discrepancies mentioned above, the
This behavior is not seen in the measurements. Finally, important point to note is that, overall, the measured
the overall trends in the axial velocity profile prior to velocities and their evolution correlate well with the
injection differs for Rs = 7.12the measured tangential numerical simulations.
velocity has a maximum near z = -12 mm, while the
simulation predicts a maximum near the cylinder head. CONTACT
This behavior was repeatable. Additional experiment
and analysis are needed to clarify this discrepancy. Sanghoon Kook, kooks@kaist.edu