Documente Academic
Documente Profesional
Documente Cultură
SAE TECHNICAL
PAPER SERIES 2006-01-0197
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.
SAE Permissions
400 Commonwealth Drive
Warrendale, PA 15096-0001-USA
Email: permissions@sae.org
Tel: 724-772-4028
Fax: 724-776-3036
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
2006-01-0197
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]
In-cylinder pressure measurements were acquired using High speed videos were also obtained of the natural
a water-cooled (KISTLER 6043A60) piezoelectric luminosity from soot emissions, viewing the combustion
pressure transducer. Measurements obtained over 50 chamber through a side window. In this case, the
engine cycles were averaged to calculate the indicated 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
Flow Velocity The swirl ratio is varied by throttling the low-swirl intake
ports. To achieve the large range of swirl ratios reported
Measurements of the tangential component of velocity here, the intake valve in the high-swirl port was fitted
were acquired with a custom-designed, fiber-coupled with a 180 shroud, oriented to enhance the component
LDV. A laser power of approximately 260 mW per beam of the intake flow tangential to the cylinder wall. The
was used at a wavelength of 514.5 nm. A 5.0 MHz swirl ratios reported in Table 2 are the numeric values
differential frequency shift removed directional ambiguity. resulting from flow-bench measurements performed by
ZrO2 seed particles were introduced with a fluidized bed Ricardo, Inc. Radial profiles of the tangential velocity at
seeder. The seed particles have a mean diameter of three axial locations, measured with the LDV through the
0.48 m and a standard deviation of 0.15 m. Additional liner windows, provide a more direct measure of the in-
details can be found in Ref. 21.
cylinder angular momentum. At -55 CAD, integration of
these profiles resulted in swirl ratios of 1.10, 2.75, 3.86,
ENGINE OPERATION
5.05, and 6.53. With the exception of the lowest swirl
ratio, these measurements are within 8% of the flow-
The engine operating parameters are described in
bench swirl ratios. The swirl ratios reported throughout
Table 2. All data were acquired at a fixed engine speed
the remainder of this document are those determined via
of 1500 rpm.
flow-bench.
Recirculated exhaust gas was simulated by diluting the
The fueling rate was controlled to provide a constant 3
intake air stream with N2 and CO2. The relative
bar gross IMEP (indicated mean effective pressure) for
proportions of CO2 and N2 were chosen to match the each swirl ratio, at the injection timing providing best fuel
mixture molar specific heat of real engine exhaust gas at economy. Thereafter, as SOI was varied, the fueling rate
the selected load and O2 concentration. This matching was held fixed. The most advanced injection timing
was performed with gas properties evaluated at 600 K. considered at each swirl ratio was determined by the
The 10.1% O2 concentration selected corresponded to a achievable IMEP. When the IMEP dropped significantly
65% exhaust gas recirculation (EGR) rate. At Rs = 3.77 from the desired load no earlier timings were considered.
the typical in-cylinder equivalence ratio ( I ), was 0.57 As discussed below, excessive spray penetration due to
(O= 1.75), corresponding to an equivalence ratio based low in-cylinder gas density is thought to limit the earliest
on intake air of 0.79 (O= 1.27). practical injection timing. Injection timing was retarded to
the crank angle at which impaired fuel economy was
Prior to obtaining in-cylinder pressure, exhaust observed at the lower swirl ratios, typically -10.25 CAD.
emissions and flow velocity, the engine was motored for The approximate fueling rates employed for each swirl
a minimum of 90 s in order to pre-heat the combustion ratio are listed in Table 3. The fuel employed was a 2007
chamber walls and to allow the intake plenum pressure emission certification diesel with a cetane number of
to stabilize. For in-cylinder pressure and exhaust 47.1 and a lower heating value of 42.98 MJ/kg.
emission data, the engine was skip-fired for 70 s, with
Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
Swirl Ratio
were mapped to a sector grid and used to initialize the
KIVA-3V [14] simulations of the combustion process. 5 2500
However, a comparison of the temporal history of the
tangential velocity and turbulent kinetic energy for 4 2000
various initialization schemes revealed that a simpler
solid-body-rotation swirl initialization performed well 3 1500
compared to the mapped results [15]. Due to the 1000
computational expense of computing the intake flow for 2
the range of swirl velocities considered here (Rs = 1.44 [ppm]
7.12), solid-body-rotation was used for initialization of -30 -25 -20 -15 -10
the tangential velocity in this study. The usual practice SOI [CAD ATDC]
of initializing the axial velocities by linearly interpolating
between the piston surface and the head, and initializing
the radial velocities to zero was followed. Figure 3 Effect of swirl ratio and injection timing on CO
emission
The spray and combustion sub-models employed are
described more fully elsewhere ([15],[16]). The
characteristic time combustion sub-model [17], is of observed optimal swirl ratio is somewhat unexpected.
greatest relevance to the present study. In this model, Higher swirl ratios are typically thought to enhance
the instantaneous production and oxidation rates of the turbulence production and mixing processes, resulting in
important combustion species are not calculated directly more rapid, complete combustion [18]. Consequently, if
via either simplified or detailed chemical mechanisms. under-mixed fuel is the dominant source of CO
Rather, the rate of change of each species is calculated emissions, we would anticipate a continuous
as proportional to the difference between the actual improvement in CO emissions as the swirl level is
species mass fraction and its instantaneous thermo- increased.
dynamic equilibrium value. The constant of propor-
tionality (the inverse of the characteristic time) is a The existence of an optimal swirl level suggests that a
weighted combination of a chemical timescale and a mechanism by which the highest swirl levels impede
turbulent timescale, with progressively greater weighting mixing and combustion is present. However, the
toward the turbulent scale as the combustion proceeds. traditional mechanism by which excessive swirl is
thought to impede mixing, by limiting jet penetration
RESULTS AND DISCUSSION (e.g. [19],[20]) is not believed to be significant for the
small bore diameters and high injection pressures that
The results of this study are presented in four main parts. characterize modern automotive diesel engines. This
In the first part, the variation in CO emissions with swirl belief is supported by Fig. 4, which shows the fuel mass
ratio and injection timing is reported. Second, the source distributions predicted for various swirl ratios at -17 CAD
of the influence of swirl ratio on CO emissions is clarified ATDC. In these simulations, SOI was fixed at -22.25
through examination of the heat release characteristics, CAD and the crank angle shown occurs shortly after the
numerical simulations and in-cylinder imaging of end of injection. A high fuel mass fraction is observed
combustion luminosity. Third, the factors potentially near the bowl lip at all swirl ratios, indicating penetration
responsible for the observed dependency of CO of the fuel jet to the maximum possible extent. At the
emissions on injection timing are examined. Finally, we higher swirl ratios, more fuel is observed in the cut-plane
examine the influence of both swirl ratio and injection shown (22 down-swirl of the fuel jet) due to greater
timing on fuel conversion efficiency. convection by the swirl velocity.
80
1.44
60 2.59
50 3.77
4.94
40 7.12
30
20
10
0
0.6
0.4
0.3
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
Figure 6 The mean r z plane flow structure and the Figure 7 Changes in the spatial distribution of O2 at
spatial distribution of CO at 3 CAD for various swirl 3 CAD for various swirl ratios, as predicted by numerical
ratios, as predicted by numerical simulation. simulation. SOI = -22.25 CAD, cut-plane 22 down-swirl
SOI = -22.25 CAD, cut-plane 22 down-swirl from the from the fuel jet
fuel jet
numerical simulation were examined. In our past work, crank angle is during the mixing-controlled portion of
we have shown that the mean (bulk) flow structures combustion, just after the secondary peak in the
predicted by the simulation capture the measured mean apparent heat release rates seen in Fig. 5. At this time,
flow structures remarkably well [21]. However, this the cumulative heat release rates seen in the lower
comparison was performed for a single, moderate swirl portion of Fig. 5 continue to diverge.
ratio. To extend this validation to the higher swirl ratios
employed here, we have measured axial profiles of the At the lowest swirl ratio, Rs = 1.44, the bulk flow
tangential velocity in the central region of the bowl at structure is dominated by a single clockwise-rotating
each swirl ratio, and compare these to the simulation structure that fills the outer portion of the bowl. In the
results in the Appendix. Although some differences exist, central region of the bowl, above the pip, a smaller
overall the comparison shows that the axial distribution secondary vortex rotating in the opposite direction is
of the tangential velocity is similar to the computed seen. From Fig. 7, which shows the distribution of the
results, and evolves in a like manner. remaining O2 in the combustion chamber, it is seen that
this secondary vortex transports O2 to the same region
Figure 6 shows the variation in the r z plane flow where the dominant vortex transports CO. In this
structures and in the spatial distribution of CO mass interfacial region between the two vortices, steep
fraction predicted for various swirl ratios at 3 CAD. This 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
As swirl ratio is further increased to 2.59, the same dual- In contrast, at the highest swirl ratios, co-location with
vortex flow structure is again apparent. In this case, the high-swirl fluid both impedes mixing with the O2 in the
centrifugal forces acting on the high angular momentum center of the bowl and interferes with the exit of the
fluid transported into the inner regions of the bowl are partially-burned fuel from the bowl with the squish flow.
larger, and the lower vortex is smaller due to the Fundamentally, the centrifugal forces acting on the high-
tendency of this high angular momentum fluid to return swirl fluid prevent the inward displacement required to
to the bowl periphery. Consequently, the upper vortex is
larger and encompasses a greater fraction of the
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 digital video camera. Figure 10 compares the side-view
CO and the high-swirl fluid therefore persists until well images of the natural soot luminosity observed at
into the expansion stroke, effectively trapping the CO in Rs = 1.44 with that observed at higher swirl ratios,
the re-entrant portion of the bowl. This is illustrated in Rs = 3.77 and Rs = 7.12. Although the intensity of the
Fig. 9 for the highest swirl case, Rs = 7.12. soot luminosity is less, images obtained at the higher
swirl ratios clearly show that the luminous soot (which is
To provide additional experimental support for this anticipated to be co-located with partially-burned fuel) is
picture, the spatial distribution of soot luminosity, and its found deeper in the bowl and more uniformly fills the
temporal evolution, was captured using the high speed 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
0.5 60
In-Cylinder CO mass 0.2
0.0
40 SOI: -30.25 CAD ATDC
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 Rs=3.77
P : 600 bar 1800
CO Emission [ppm]
inj 800 bar
CO Emission [ppm]
1000 bar
3500
4 Skip-Fired
1200 bar SOI=-22.25 CAD ATDC
4 Skip-Fired
1400
2500
1000
SOI=-28.25 CAD ATDC
1500
SOI=-26.25 CAD ATDC
600
500
1.4 1.6 1.8 2 2.2
-30 -25 -20 -15 -10 -5
[ms]
SOI [CAD ATDC] id
Figure 13 Effect of injection pressure and injection timing Figure 14 Correlation of CO emission with the ignition
on CO emission at swirl ratio of 3.77 delay at swirl ratio of 3.77
timing. This enhanced mixture preparation is reflected in with the numerical predictions. At the earliest injection
the higher peak heat release rates seen in Fig. 11 for timings, however, increased injection pressure may be
SOI=-30.25, behavior which is well-captured in the detrimental.
simulation results, as are the ignition delay periods
observed for both injection timings. Ignition delay is well-known to increase as injection
timing is advanced, due to the lower ambient
Examination of the spatial distributions of fuel predicted temperature and pressure at earlier SOI [24]. Here, the
numerically suggest that the enhanced mixing observed ignition delay is defined as the time from SOI until the
with earlier injection is not due to a significant difference time at which 10% of the total heat release has occurred.
in the flow structures existing during the ignition delay The 10% heat release timing correlates well with the
period, but rather to initial deposition of a significant beginning of rapid, high-temperature heat release. As
fraction of fuel vapor within the squish volume. As the noted in the introduction, we have previously observed
piston rises, this fuel is returned to the bowl with the an inverse correlation at high dilution rates between CO
squish flow and results in a better mixed, broader spatial emission and ignition delay. In general provided spray
distribution of fuel vapor than is seen with later injection. targeting, wall impingement, or spray/flow interactions
Peak fuel vapor concentrations (and subsequent CO do not vary significantly with SOI increased ignition
concentrations), for both injection timings, are found delay should correlate well with a higher degree of fuel-
deep in the bowl, as can be seen in the lower portion of air premixing. In turn, more premixing will correlate
Fig. 6. In both cases, the CO remains trapped in the inversely with CO emission, provided CO emission is
lower regions of the bowl during expansion. For the dominated by the combustion of rich mixtures.
earlier injection timing, however, there is far less of it. Conversely, CO emissions resulting from over-mixed
fuel would be expected to increase with increasing
Thus, the simulation results suggest that the dominant ignition delay.
cause of the reduced CO emissions with advanced
injection timing is associated with reduced peak CO The correlation between ignition delay and CO emission
mass during the early combustion process. Early mixing measured for Rs=3.77 is shown in Fig. 14. This
processes, occurring prior to or during the first portion of intermediate swirl ratio was selected for detailed
the premixed burn, are at least partly responsible for the examination because a minimum in CO emission is
reduced CO mass. observed near SOI =-26 CAD in both Figs. 1 and 3. In
keeping with the predictions of the numerical simulations,
The Influence of Injection Pressure and Ignition Delay the variable injection pressure tests, and our previous
results, Fig. 14 shows a strong negative correlation of
The numerical simulations of the previous section CO emission with increasing ignition delay, indicating
suggest that more rapid initial mixing is an effective the benefits of increased early mixing for reducing CO
method of reducing CO emissionprovided that CO emission.
emission stems predominantly from under-mixed fuel. A
straight-forward test of this prediction can be made At the earliest injection timing shown in Fig. 14,
through increasing injection pressure, a measure that increased CO emission is observed despite an increase
will enhance initial mixing rates but which is less likely to
significantly impact the late-cycle mixing. As This view must be adopted with caution. As shown in
demonstrated by Fig. 13, increased injection pressure is the previous section, under some circumstances mixing-
found to generally reduce CO emissions, in agreement rates can be affected significantly by changes in SOI.
Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
Figure 16 Visualized spray impingement on the bowl lip Figure 17 Visualized spray impingement on the bowl lip
and predicted fuel mass distribution: SOI = -26.25 CAD, and predicted fuel mass distribution: SOI = -22.25 CAD,
Rs = 3.77 Rs = 3.77
almost identical for both injection timings. Accordingly, At these more advanced injection timings, liquid films
the initial distribution of the vapor phase fuel does not can be expected to form not only on the bowl rim, but
appear to be a dominant factor in determining the CO also on the piston top. Figure 18 shows an idealized
formation and ultimately the CO emission level. This view of the spray targeting on the piston for various
stands in contrast to our previous discussion of the fuel crank angles. With a start of injection at -26 CAD and
vapor distributions observed for Rs = 7.12. Recall that, in later, we predominantly expect wetting of the bowl rim,
that case, significantly more advanced injection timing as was seen in Fig. 16. Minor wetting of the piston top
was considered. 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
0
-5 CAD TDC
D
-5
z [mm]
-10
-15 CAD
-15
-20 -25 CA
AD ATDC
-22 CAD 0
r = 13.6
3 mm
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
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
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 The gross indicated fuel conversion efficiency is the ratio
of the gross indicated work W to the chemical energy in
In the previous sections we have examined how swirl the injected fuel, calculated from the product of the fuel
ratio and combustion phasing (injection timing) influence mass and the lower-heating value QLHV of the fuel
CO emission, which is an indication of complete,
efficient combustion. Complete combustion is a pre- W
K fc (2)
requisite for obtaining a high fuel conversion efficiency m f Q LHV
Kfc . However, other factors also influence Kfc . In this
section our goal is to clarify the broader impact of As an aid to understanding the influence of various
variations in swirl ratio and injection timing on Kfc . In factors on Kfc (including CO emission and combustion
pursuing this goal, we consider only the gross indicated efficiency), we have previously proposed [9] that Kfc be
fuel conversion efficiency. The net fuel conversion decomposed as a product of three separate efficiencies
efficiency may be influenced through other factors (e.g.
increased pumping work associated with high-swirl port K
designs) that are not addressed here.
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
blow-by some of this work is inevitably lost. In the latter SUMMARY AND CONCLUSION
case, heat release is sufficiently retarded that much of
the volume expansion has already occurred. CO emission and fuel conversion efficiency are
examined in a highly-dilute, low-temperature diesel
A second observation is the low K wc in the upper-central combustion regime as the injection timing and swirl ratio
region of Fig. 24, which coincides with a minimum in the are varied. To clarify the dominant factors influencing the
combustion efficiency. As discussed in the section emission and efficiency behavior, the cylinder pressure,
describing the optimal swirl ratio, the cause of this engine-out emissions, and flow velocities are measured.
minimum in Kc is thought to be trapping of CO in the High speed video imaging and multi-dimensional numer-
bowl. Apparently, this trapping also retards the heat ical simulations of the combustion process are also
release thereby influencing K wc . performed.
Finally, K wc also exhibits a general decreasing trend as The CO emission behavior exhibits two interesting
the swirl level increases. As discussed elsewhere [28], trends:
the work conversion efficiency is influenced by heat
transfer losses that occur in a non-combusting cycle. First, an optimal swirl ratio exists at which the lowest CO
emission and best fuel conversion efficiency are
The higher heat transfer losses incurred in high swirl
observed. Heat release analysis, the results of the
flows result in a decreased K wc . Consequently, the
numerical simulations, and imaging of combustion
influence of heat loss is not fully-confined to the heat-
luminosity indicate that this behavior is likely due to
loss efficiency. mixing processes occurring after the premixed burn
period. At the lowest swirl ratios, a dual-vortex vertical
Although Q chem and Q hl have not been measured plane structure forms that enhances the mixing of
directly, the heat-loss efficiency K hl can be estimated partially-burned fuel and air. This structure is most
from the remaining efficiencies via Eq.(3). The results effective at a moderate swirl ratio (| 2.5). At higher swirl
must be employed cautiously, however, as the ratios, the flow structures formed impede mixing, and
combustion efficiencies calculated from Eq.(4) will likely can lead to trapping of partially-burned fuel within the
be overestimated, due to the neglect of the energy bowl.
content in the unburned hydrocarbons and other
unmeasured species. Provided that these species follow Second, CO emission generally exhibits a rapid
the same trends as the CO emissions, their neglect decrease from the maximum as SOI is advanced,
implies that the changes observed in Kc are particularly at the highest swirl ratios. The numerical
underestimated. simulations indicate that, at a fixed swirl ratio, earlier
injection timing leads to enhanced pre-combustion
Fig. 25 illustrates the resulting estimates of K hl . Note mixing and hence lower peak in-cylinder CO mass. The
that the overall variation in K hl is larger than is seen in enhanced mixing is due not only to increased ignition
either K wc or Kc . However, because the variation in Kc is delay, but also to increased mixing rates under high-
underestimated, the variation seen in K hl is likely swirl conditions. A reduction in CO emission with
excessive. Nevertheless, the magnitude of the true increased pre-mixing CO implies that CO emission
variation is likely to be significant, and the trends stems predominantly from under-mixed fuel (rich
observed are physically plausible and bear consideration. mixtures), a finding which is supported experimentally by
First, a general trend of decreasing heat-loss efficiency a strong tendency toward reduced CO emission with
is seen as swirl is increaseda result that is not increased ignition delay and with increased injection
pressure. Correlation of CO emission with peak
unexpected. More interestingly, the heat loss efficiency
adiabatic flame temperature suggests that temperature
shows a greater variation with injection timing. This
related changes to the early CO oxidation rates are not a
variation might be expected based on the simple
significant factor influencing peak in-cylinder CO mass
observation that there is more time available for heat and subsequent engine-out emission. Spray penetration
transfer at the earlier injection timing. At least three other and spray targeting, with concomitant liquid film
factors also contribute to this behavior, though. First, formation was also examined as a potential source of
with earlier combustion phasing the average surface modified premixing. Significant liquid film formation
area-to-volume ratio of the hot, in-cylinder gases is along the bowl lip was observed at early injection timings.
increased. Second, mixture that burns first and then is However, low CO emission was observed under these
compressed reaches a higher temperature than mixture conditions, signifying that wall-film formation in this
that burns after compression. Third, convective heat region is not a dominant factor influencing CO emission.
transfer coefficients are expected to be larger at the At the earliest injection timings, the likely existence of
higher peak cylinder pressures observed with early wall films on the top of the piston is identified as a
injection. Each of these factors is expected to increase potential source of increasing CO. Finally, the possibility
heat transfer losses as injection is advanced. 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 8. Adomeit, P., Pischinger, S., Becker, M., Rohs, H.,
mixing taking place in the ignition delay period. and Greis, A., Laser Optical Diagnostics and
Numerical Analysis of HSDI Combustion, THIESEL
The impact of swirl level and injection timing on the fuel 2004: Conference on Thermo- and Fluid Dynamic
conversion efficiency is also assessed. Qualitative Processes in Diesel Engines, Sept. 8-10, Valencia,
differences are observed between the fuel conversion Spain, 2004.
efficiency and the combustion efficiency derived from the 9. Kook, S., Bae, C., Miles, P. C., Choi, D, and Pickett,
CO emissions. These differences can be accounted for L. M., The Influence of Charge Dilution and
by changes in the work conversion efficiency Injection Timing on Low-Temperature Diesel
(associated primarily with combustion phasing) and by Combustion and Emissions, SAE Paper 2005-01-
heat transfer losses.
3837, 2005.
10. Timoney, D.L., Smoke and Fuel Consumption
ACKNOWLEDGMENTS Measurements in a Direct Injection Diesel Engine
with Variable Swirl, SAE Paper No. 851542, 1985.
Support for this research was provided by the U.S.
11. Van Gerpen, J.H., Hwang, C.-W., and Borman, G.L.,
Department of Energy, Office of FreedomCAR and
The Effects of Swirl and Injection Parameters on
Vehicle Technologies. The research was performed at
Diesel Combustion and Heat Transfer, SAE Paper
the Combustion Research Facility, Sandia National
Laboratories, Livermore, California. Sandia is a No. 850265, 1985.
multiprogram laboratory operated by Sandia Corporation, 12. Miles, P. C., The Influence of Swirl on HSDI Diesel
a Lockheed Martin Company, for the United States Combustion at Moderate Speed and Load, SAE
Department of Energys National Nuclear Security Paper 2000-01-1829, 2000.
Administration under contract DE-AC04-94AL85000. The 13. Miles, P.C., Choi, D., Pickett, L.M., Singh, I.P.,
BK21 and Future Vehicle Technology Development Henein, N., RempelEwert, B.H., Yun, H., and Reitz,
Corps. of Korea supported Sanghoon Kook's visiting R. D., Rate-Limiting Processes in Late-Injection,
research. The authors express their appreciation to Mark Low-Temperature Diesel Combustion Regimes,
Musculus and Lyle Pickett for providing the high speed THIESEL 2004: Conference on Thermo- and Fluid
camera and the Matlab source code to calculate the Dynamic Processes in Diesel Engines, Sept. 8-10,
adiabatic flame temperature. Valencia, Spain, 2004.
14. Amsden, A. A., KIVA-3V: A Block Structured KIVA
REFERENCES Program for Engines with Vertical or Canted Valves,
Los Alamos National Laboratory Report No. LA-
1. Cowland, C., Gutmann, P. and Herzog, P.L., 13313-MS, 1997.
Passenger Vehicle Diesel Engines for the US, SAE 15. RempelEwert, Bret H., The Influence of Swirl on
Paper No. 2004-01-1452, 2004. Flow, Fuel Injection, and Emissions in an HSDI
2. Kimura, S., Aoki, O., Kitahara, Y. and Aiyoshizawa Diesel Engine, Masters Thesis, Dept. of
E., Ultra-Clean Combustion Technology Combining Mechanical Engineering, University of Wisconsin
a Low-Temperature and Premixed Combustion Madison, 2004.
Concept for Meeting Future Emission Standards, 16. Miles, P., Choi, D., Megerle, M., RempelEwert, B.,
SAE Paper No. 2001-01-0200, 2001. Reitz, R. D., Lai, M. D., Sick, V., The Influence of
3. Akihama, K., Takatori, Y., Inagaki, K., Sasaki, S., Swirl Ratio on Turbulent Flow Structure in a Motored
and Dean, A.M., Mechanism of the Smokeless Rich HSDI Diesel Engine A Combined Experimental
Diesel Combustion by Reducing Temperature, SAE and Numerical Study, SAE Paper 2004-01-1678,
Paper No. 2001-01-0655, 2001. 2004.
4. Walter, B. and Gatellier, B. Development of the 17. Kong, S.-C., Han, Z., and Reitz, R.D., The
High Power NADI Concept Using Dual Mode Diesel Development and Application of a Diesel Ignition
Combustion to Achieve Zero NOx and Particulate and Combustion Model for Multidimensional Engine
Emissions, SAE Paper No. 2002-01-1744, 2002. Simulation, SAE Paper 950278, 1995.
5. Weibck, M., Csat, J., Glensvig, M., Sams, T. and 18. Ikegami, M., Role of Flows and Turbulent Mixing in
Herzog, P., Alternative Brennverfahren ein Ansatz Combustion and Pollutant Formation in Diesel
fr den Zuknftigen Pkw-Dieselmotor, MTZ: vol 64: Engines, COMODIA 90: Proceedings of the
pp 718-727, 2003. International Symposium on Diagnostics and
6. Hasegawa R. and Yanagihara, H., HCCI Modeling of Combustion in Internal Combustion
Combustion in DI Diesel Engine, SAE Paper No. Engines, 35 September, Kyoto, Japan, 1990.
2003-01-0745, 2003. 19. Khan, I.M., Wang, C.H.T., and Langridge, B.E.,
7. Sjberg, M., and Dec, J. E., An Investigation into Effect of Air Swirl on Smoke and Gaseous
the Lowest Acceptable Combustion Temperatures Emissions from Direct-Injection Diesel Engines,
for Hydrocarbon Fuels in HCCI Engines, SAE Paper No. 720102, 1972.
Proceedings of the Combustion Institute, Vol. 30, 20. Binder, K. and Hilburger, W., Influence of the
pp.2719-2726, 2005. Relative Motions of Air and Fuel Vapor on the
Downloaded from SAE International by Indian Institute of Technology - Chennai, Monday, July 17, 2017
NOMENCLATURE
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
-10 -10 -5 -5
-10 -10 -5 -5
-20 CAD
z [mm]
0 0
-20 -20 TDC
z [mm]
-20 CAD
z [mm]
-5 0 0
-5
-10 -10 -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
-5 -5 -5 -5
-20 CAD
z [mm]
-20 CAD
z [mm]
z [mm]
0 0 0 0
-5 -5 -5 -5
-5 -5 -5 -5
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