hydrocarbons using close-coupled post injections at ltc optical ... · optical investigation of the...

INTRODUCTIONLow temperature combustion (LTC) has been proposed as

a strategy for meeting market demands for fuel efficiencywhile also satisfying regulated limits on NOx and sootemissions from heavy-duty diesel engines. In-cylinderstrategies designed to reduce engine-out emissions of thesetwo pollutants are of interest because they could reduce oreven eliminate the burden on aftertreatment systems.Compared to conventional diesel combustion, LTC strategiesgenerally can be characterized by high levels of intake chargedilution and increased mixing of fuel with the intake chargeprior to combustion.

One method for diluting the intake charge is exhaust gasrecirculation (EGR), which reduces the peak temperatures ofcombustion and hence the thermally produced NOx [1]. Theslowing of NOx formation is a result of two effects: anincrease in the heat capacity of the intake mixture, which

reduces peak temperatures, and reduction in intake oxygen,which slows NOx formation kinetics.

Increased pre-combustion mixing is typically achieved byaltering the fuel-injection timing. Conventional dieseloperation uses injection timings that occur several crankangle degrees before top-dead-center (TDC), so that theignition delay is short and combustion occurs near TDC. Thiscombustion phasing is favorable for thermodynamicefficiency, but much of the charge is fuel-rich duringcombustion, which can also lead to significant soot formation[2]. LTC techniques achieve less soot formation by placingthe injections either much earlier or much later than theconventional injection timing.

Early injection schemes - often termed partially-premixedcompression ignition (PPCI), premixed-charge compressionignition (PCCI), premixed compression ignition (PCI), or insome cases in the literature, even homogeneous chargecompression ignition (HCCI) - operate by providing

2013-01-0910Published 04/08/2013

Copyright © 2013 SAE Internationaldoi:10.4271/2013-01-0910

saeeng.saejournals.org

Optical Investigation of the Reduction of UnburnedHydrocarbons Using Close-Coupled Post Injections at LTC

Conditions in a Heavy-Duty Diesel EngineJacqueline O'Connor and Mark Musculus

Sandia National Laboratories

ABSTRACTPartially premixed low-temperature combustion (LTC) using exhaust-gas recirculation (EGR) has the potential to

reduce engine-out NOx and soot emissions, but increased unburned hydrocarbon (UHC) emissions need to be addressed. Inthis study, we investigate close-coupled post injections for reducing UHC emissions. By injecting small amounts of fuelsoon after the end of the main injection, fuel-lean mixtures near the injector that suffer incomplete combustion can beenriched with post-injection fuel and burned to completion. The goal of this work is to understand the in-cylindermechanisms affecting the post-injection efficacy and to quantify its sensitivity to operational parameters including post-injection duration, injection dwell, load, and ignition delay time of the post-injection mixture. Three optical diagnostics -planar laser induced fluorescence of OH radicals, planar laser induced fluorescence of formaldehyde, and high-speedimaging of natural combustion luminescence - complement measurements of engine-out UHC with parametric variationsof main- and post-injection timing and duration. Across all conditions tested, each at 1200 RPM, the optimal post-injectioncommand duration for UHC reduction was approximately 400 microseconds (2.9 °CA). Also, conditions with shorter (3.4°CA) post-injection ignition delays were over twice as effective on a percentage basis at reducing engine-out UHC as thosewith longer (5.5 °CA) post-injection ignition delays. Optical data at the post-injection “sweet-spot,” where UHC emissionsare minimized, indicate that the post injection promotes transition to second-stage ignition in the near-injector region, mostlikely by enriching the overly fuel-lean mixtures in the wake of the main injection.

CITATION: O'Connor, J. and Musculus, M., "Optical Investigation of the Reduction of Unburned Hydrocarbons UsingClose-Coupled Post Injections at LTC Conditions in a Heavy-Duty Diesel Engine," SAE Int. J. Engines 6(1):2013, doi:10.4271/2013-01-0910.

____________________________________

THIS DOCUMENT IS PROTECTED BY U.S. AND INTERNATIONAL COPYRIGHT.It may not be reproduced, stored in a retrieval system, distributed or transmitted, in whole or in part, in any form or by any means.

Downloaded from SAE International by Jacqueline O'Connor, Friday, April 12, 2013 04:03:16 PM

http://dx.doi.org/10.4271/2013-01-0910

http://saeeng.saejournals.org/

significant time between the injection and the thermodynamicstate during compression that induces ignition [3, 4, 5]. Giventhe long ignition delay of a mixture at early injection timings,these schemes create a positive ignition dwell, which is thetime between the end of the injection and the start ofcombustion. A positive ignition dwell gives all of the injectedfuel and air more time to mix before combustion, reducingrich regions where soot could form, while often creating afast, efficient premixed heat-release profile [6].

Late-injection schemes - such as “modulated kinetics”(MK) combustion [7, 8] - achieve similar ends as the early-injection schemes by injecting fuel very near TDC or eveninto the expansion stroke. Here, a positive ignition dwell ispossible because of high rates of EGR and the cooling of thecharge due to expansion. As with early-injection strategies,the positive ignition dwell provides more time for mixing andhence lower soot formation and a more premixed-typecombustion event.

While these methods of LTC can lead to substantialreductions in NOx and soot emissions, there are still issuesremaining. In particular, unburned hydrocarbon (UHC) andcarbon monoxide (CO) emissions can increase under theseconditions [9, 10]. Going from conventional dieselcombustion to LTC operation can be viewed as moving alonga trade-off between soot and UHC (and CO); as the ignitiondelay increases and more premixing is allowed before thestart of combustion, fewer regions of the reactants are fuel-rich enough to promote soot production, but UHC emissionsincrease, especially as the load decreases (for instance seeFigure 2). For this reason, we will focus primarily on low-load, LTC conditions where UHC emissions are particularlyproblematic.

Previous studies have investigated the sources of UHC atLTC conditions. Our own previous work has shown thatsome overly-lean mixtures in the region of the injectorformed after the end of injection do not achieve completecombustion and can be significant sources of hydrocarbons[11, 12]. Other UHC sources, including wall-wetting and fuelin the crevices, can also be important, especially in light-dutyengines [13, 14]. The current study focuses on UHCemissions from over-leaning in the bulk charge.

Overly-lean mixtures in the injector region have beenmeasured using a variety of techniques. In our previous work,we used a toluene fluorescence technique in a heavy-duty,single-cylinder, optical research engine as well laser-Rayleigh-scattering in an engine simulation spray facility[12]. For the tracer fluorescence measurements, fuel wasdoped with a known quantity of toluene and injected into anon-reacting charge to enable quantitative fuel concentrationmeasurements, at least up to the time that combustion wouldnormally commence. Using the fuel concentration data, anequivalence ratio map can be generated corresponding to agiven intake oxygen concentration. Equivalence ratio contourmaps for an intake-oxygen concentration of 12.7% are shownin Figure 1.

Figure 1. Equivalence ratio measurements in the opticalengine after the end of injection using toluene PLIF fora 12.7% O2 intake charge, from Ref. [12]. Crank angle

degrees are displayed in units of degrees after the end ofinjection. The lines indicate contours of constant

equivalence ratio. The injector is located on the left ofthe images and two jets of the 8-hole injector pass

through the measurement plane - one upward and to theright (near 2 o'clock) and the other downward and to theright (near 4 o'clock). Copyright © SAE International.

Reprinted with permission [12].

The equivalence ratio contours in Figure 1 show that themixtures near the injector (left side) become leaner veryquickly after the end of the injection, initially at a rate ofapproximately 4 equivalence ratio units per crank angledegree. By 5 °CA after the end of injection (AEI), theequivalence ratio is 0.5 or lower in the near-injector region.Other optical data showed that first-stage (cool-flame)ignition occurs throughout the jet, but second-stage ignitionoccurs only in the richer downstream mixtures (right side ofimages), near 5 °CA AEI [11]. At the conditions examined,the upstream fuel-lean mixtures never achieved second-stage

O'Connor et al / SAE Int. J. Engines / Volume 6, Issue 1(May 2013)



ignition in the time available before significant cylinderexpansion. Hence, incomplete combustion of those fuel-leanmixtures contributed to UHC emissions.

These results and others like them [15] motivated afollow-up study in the current engine, showing promisingresults for UHC reduction with the use of post injections [16].The rationale for using post injections is to enrich the overly-lean mixture near the injector, essentially “kicking” it intosecond-stage ignition and reducing unburned hydrocarbonsfrom the injector region. Because of limitations in the fuelinjection system, only two-injection schedules were tested,one with a short, close-coupled post injection (injection dwell≈ 1 °CA) and one with a longer post-injection duration anddwell (injection dwell ≈ 3 °CA). The injection schedules forthese two tests are in Figure 2a.

Figure 2. Results from Chartier et al. [16] that show a)injection schedule for a single injection and two post-

injection schedules and b) normalized UHCmeasurements for these three conditions. Copyright ©

SAE International. Reprinted with permission [16].

The engine-out UHC results in Figure 2b illustrate thesignificant effect that a short, close-coupled post injection canhave on UHC reduction. Whereas the longer post injectiondid not reduce UHC emissions relative to the single-injectionbaseline, the short, close-coupled post injection reduced theUHC by 27%. Optical investigations of the progression ofcombustion for these two conditions were consistent with

enrichment of the near-injector overly-lean mixture by thepost injection, bringing the combustion to completion toreduce UHC emissions.

Despite the obvious gains that were observed by using ashort, close-coupled post injection in this previous study, onlyone short, close-coupled post-injection timing was found tobe repeatable due to limitations of the heavy-duty injector.The goal of the current study is to further explore the effect ofa post injection on UHC reduction over a much wider rangeof injection schedules with a fast-responding, light-dutyinjector.

Fluid mechanical (mixing) and chemical kinetics effectsof the post injections are explored in two ways. First, thecommand duration of injection for close-coupled postinjection (DOI2C) is varied from very short (DOI2C=200microseconds) to relatively long (DOI2C=800 microseconds)for a given main injection duration and timing. This variationin post-injection duration not only changes the amount of fueldelivered in the post injection, but also the penetration andfluid mixing characteristics of the post jet. Short postinjections (DOI2C<350 microseconds) do not penetrate to thebowl wall of this engine within the time available forcombustion, and have different injection profiles and ignitiondwells than long post injections. Additionally, post injectionsthat do not penetrate to the bowl wall do not have fuel/airmixtures that “roll up” along the wall into the large siderollers in the jet-jet interaction regions that characterize low-temperature combustion in diesel engines [17]. Thesepossible differences in penetration, injection profile, andmixing time could significantly change how the post injectionenriches the overly-lean mixture left by the main injection.

Second, the ignition delay of the mixture created with thepost injection is varied by changing the combustion phasingof the entire injection schedule. Here, main- and post-injection events are moved as a unit to earlier and latertimings relative to a baseline timing. In doing so, the ignitiondelay of the near-TDC main injection changes only slightlybecause ambient conditions change minimally near TDC,when the piston velocity is low. The ignition delay of the postinjection, however, varies much more because the ambientgases change more significantly at the later injection timingof the post injection when the cylinder contents are starting toexpand. In this case, an increase (or decrease) in the ignitiondelay of the post injection would allow for more (or less)time for mixing between the post-injection fuel and theoverly-lean mixture from the main injection. Even though theschedule is shifted, the main- and post-injection durations anddwell between them are held constant, so the fluid mechanicsof injection remain largely the same. In this way, the fluidmechanical effects of post-injection penetration and mixingcan be separated from kinetics effects, to some degree.




EXPERIMENTAL SETUPOptical Engine Experiment

Table 1. Engine and fuel system specifications.

The engine used in this study is an optical, single-cylinderCummins N-series direct-injection, heavy-duty diesel engine.The engine has a bore of 139.7 mm and a stroke of 152.4mm, and is equipped with a Bowditch piston with an open,right-cylindrical bowl and a flat fused-silica piston-crownwindow providing imaging access to the bowl, viewing frombelow. A flat, round window is also installed in place of oneof the exhaust valves for imaging access to a portion of thesquish region above the piston (view not used in the currentstudy). A 30-mm wide curved window matching the contourof a portion of the bowl-wall allows laser access into the

bowl. Additional laser access is available through flatrectangular windows installed in a ring at the top of thecylinder. Because of thermal load limitations of the windows,the engine was skip-fired, with each fired cycle followed bynine motored cycles. Information about engine geometry is inTable 1 and a schematic of the experiment is in Figure 3.Further details about this engine can be found in Refs. [18,19].

Figure 3. Experimental setup of the single-cylinderengine, laser configuration, and three-camera optical

system. The camera field-of-view is shown in the upperright.

The injector used in this study is a Delphi DFI 1.5 light-duty, common-rail injector. A light-duty solenoid-driveninjector with 131 micron orifices was used in place of theprevious heavy-duty injector for its fast response time and itsability to deliver consistent, close-coupled, short-durationpost injections over a range of injection schedules. Becauseof the low lubricity and low viscosity of special fuels (such asn-heptane) selected for optical research, the productioncommon-rail fuel pump could not be utilized. Instead, acustom high-pressure diaphragm pump specially designed forlow-lubricity fuels pressurizes the fuel rail at up to 2000 bar.The delivery rate of the diaphragm pump is limited, however,and as a result it could only sustainably pressurize the fuelrail to 1200 bar at the static back-leak rate of this light-dutyinjector.




Engine Operating ConditionsThe engine is operated in a high-EGR (12.6% intake O2)

late-injection LTC mode summarized in Table 2, for whichthe start of main combustion begins slightly after TDC. Thelow intake temperature in Table 2 was intentionally selectedfor several reasons. First, diluting the intake with N2 ratherthan real EGR yields a higher specific-heat ratio of thecharge, so that lower intake temperatures are required toachieve the same compressed temperatures at TDC as withEGR. Even after accounting for differences in compression-heating effects, a relatively low TDC temperature allows fuelinjection close to TDC while still achieving a long enoughignition delay for LTC conditions. Injecting close to TDC isdesirable from a fundament perspective because the in-cylinder density remains nearly constant for a relatively longtime compared to injecting during the intake or expansionstrokes. As a result, the ignition delay for the main injectionremains nearly constant for small shifts of the injectionschedule, which facilitates the study on kinetic effectsdescribed at the end of the introduction. Furthermore,studying in-cylinder processes at nearly constant density isuseful for comparison to analysis and/or experiments inconstant volume combustion vessels. The cool conditionswith near-TDC conditions also provided combustion phasingfavorable for efficiency while still achieving pre-combustionmixing times characteristic of LTC. Finally, as discussedbelow, for more typical U.S. diesel fuels with cetane numbersnear 47 (rather than 56 for than n-heptane [20]), such coolintake conditions would not be required to achieve the sameignition delays at the high EGR conditions of this study.

As discussed above, UHC emissions are a particular issueat low-load conditions; the engine load-range in this study isvaried between one (1) and about six (6) bar gIMEP. Moredetails about the engine operating conditions are in Table 2.

Table 2. Engine operating conditions.

In this partially-premixed combustion mode, the heatrelease displays a small first-stage (cool-flame) ignition peakfollowed by a large second-stage peak and a small mixing-controlled combustion tail. The addition of a post injectiontypically adds another small peak to the heat release curveafter the main second-stage ignition. These features areillustrated in Figure 4.

Figure 4. Apparent heat release rate (AHRR) for singleinjection (a) with SOI1C=352 CAD and DOI1C=1000microseconds and a main plus post injection (b) with

SOI1C=352 CAD, DOI1C=1000 microseconds,SOI2C=363.5 CAD, and DOI2C=400 microseconds.

These operating conditions are designed to parallel thoseof the previous study by Chartier et al. [16] as closely aspossible. As described above in the Introduction, this is donepurposefully to further investigate the role of short, close-coupled post injections for UHC reduction.

Two important differences between the two studies arethe fuel and the fuel injector. In anticipation of fuelconcentration and other optical diagnostic measurements tofollow, we selected n-heptane as the fuel. While the dieselprimary reference fuels of the previous study better reflect thephysical characteristics (e.g., boiling point) of typical dieselfuel, as yet, a feasible and quantitative laser-induced fuel-




tracer fluorescence diagnostic has not been developed forsuch heavy fuels. Techniques for lighter fuels like n-heptanehave been developed and successfully employed (see Figure1 for instance), so it was chosen as the fuel. As will be shownlater in the results, the characteristic post-injection UHCbehavior observed in Chartier et al. [16] with high boiling-point fuels is retained here using n-heptane.

As shown in Table 3, however, the cetane number for n-heptane is significantly higher than for the diesel primaryreference fuel blend used in Chartier et al.. Using the samein-cylinder thermodynamic conditions would result in muchshorter ignition delays with n-heptane. To match ignitiondelay, either the temperature or the pressure of the intakestream could be adjusted. The chemical kinetics of ignitionand the following combustion processes (includingincomplete combustion) depend strongly on temperature, sowe were hesitant to decrease the temperature. With thechange of injectors, the fluid mechanical processes of fuelpenetration and mixing would already be somewhat different,so we chose to reduce the intake pressure (boost) to extendthe ignition delay. A lower intake pressure yields a lower in-cylinder density, which would yield richer mixtures at a givendownstream position in the jets, but the smaller nozzle size ofthe light duty injector offsets this effect, so that the mixturesare similar for the two injectors. The TDC conditionsrequired to match temperature and ignition delay are shownin Table 3, and the heat release for the two studies are inFigure 5.

Table 3. Comparison of operating conditions betweencurrent study and Chartier et al. [16].

The heat release profiles of the two studies in Figure 5,Ref. [16] and the current study, are qualitatively similar,though the timing is not exactly the same as a result of the 1°CA difference in SOI1C and slight changes in ignition delay.In the single injection case, a similar first-stage ignition heatrelease peak appears before the second-stage ignition heatrelease peak. In the case with a main- plus post-injection

schedule (Figure 5b), the post-injection heat release profile isdifferent between the two cases. In the previous study, thepost-injection heat release for the sole condition studied wassmall, so that it appeared on the main injection heat releaseprofile as a small shoulder rather than as a distinct peak. Inthe current study, the optimal post injection is larger, and themain-injection duration is longer. As a result, the post-injection, while still close coupled, is phased later relative tothe main-injection heat release, so that the post-injection heatrelease peak is later and more distinct.

Figure 5. Apparent heat release rate comparison betweenthe previous study (Chartier et al. [16]) and the current

study for single (a) and main plus post injection (b).Injection schedules are listed in tables in the figures.

Engine DiagnosticsSeveral diagnostics are used to investigate the origin of

UHC emissions at LTC conditions: cylinder pressuremeasurements, exhaust UHC emissions analysis, and threedifferent optical diagnostics.

Cylinder pressure is measured with an AVL QC34Dpressure transducer every quarter crank-angle degree. Theapparent heat release rate is calculated from the measured




cylinder pressure using standard techniques described in Ref.[21].

Engine-out UHC emissions are measured with aCalifornia Analytical Instruments (CAI) 600 serieshydrocarbon analyzer. This analyzer samples the exhauststream and uses a flame ionization detector to measure totalunburned hydrocarbons. The exhaust stream is pulled fromthe exhaust manifold, within 100 mm of the exhaust valve.The line from the engine to the analyzer is heated to 155degrees Celsius to prevent condensation of water andhydrocarbons. A pump provides a continuous flow of exhaustgases to the analyzer, and the transit time between the engineand the analyzer is about 4 seconds. All UHC data areaveraged over a run-time of 100 fired cycles. To account forthe skip-fired operation of the engine, the reported UHC datahave been multiplied by 10 to approximate what they wouldbe in a continuously-fired, single-cylinder engine. Thestandard deviation in UHC measurements between runs isapproximately 2 ppm (corresponding to 20 ppm aftercorrecting for a continuously-fired engine).

Three different optical diagnostic techniques areemployed to gain insight into the origins of UHC emissions.Two important radical species involved in ignition provideinformation about the evolution of in-cylinder unburned fuel.Formaldehyde (H2CO) is produced during first-stage ignitionand consumed during second-stage ignition. Detailedchemical kinetics simulations predict that after first-stageignition, the evolution of formaldehyde mirrors that of theoverall pool of unburned hydrocarbons [11]. Hence,formaldehyde serves as a marker of unburned fuel frommixtures that have achieved first-stage ignition but have notyet reached second-stage ignition. Formaldehyde does notserve as a good marker for all unburned fuel, such as thatfrom regions that have not achieved first-stage ignition (e.g.,wall wetting and potentially crevices), nor does it mark UHCin rich mixtures after second-stage ignition. However,previous studies [11, 12] have shown that in the near injectorregion, overly-lean mixtures are a primary source of UHCemissions, so that formaldehyde serves as a good marker forthose UHC sources.

The second radical species, the hydroxyl radical (OH),rises in concentration by orders of magnitude at second-stageignition in fuel-lean to stoichiometric mixtures [22], andhence provides an indication of second-stage ignition andrelatively complete combustion and hence low unburned fuelconcentration.

As described below, two laser-based techniques are usedto measure these two radical species, similar to previousstudies conducted at LTC conditions in this engine [11, 16].

Laser induced fluorescence of formaldehyde (H2CO-PLIF): The third harmonic output of a Spectra-PhysicsQuanta-Ray single-cavity Nd:YAG laser at a wavelength of355 nm is formed into a 30-mm wide sheet with a thicknessof 1 mm or less. A pulse energy of 80 mJ at the engine isused laser-induced fluorescence of formaldehyde within the

engine cylinder. The sheet is oriented at a 12° angle relativeto the firedeck to probe H2CO along the approximatesymmetry plane of one of the fuel jets. LIF emission iscollected at wavelengths longer than 310 nm with anintensified Princeton Instruments PI-MAX 3 intensifiedcharge-coupled device (ICCD) camera with a HQf (GEN-III)intensifier and a resolution of 1024×1024, a gate time of 50ns, and at maximum gain. The camera is fitted with a Nikkorglass f/2.5 lens. Two filters, a 385-nm long-wave-pass and a40-nm wide bandpass filter centered at 408 nm, reject laserscatter and other interference outside the formaldehydefluorescence emission band.

Laser induced fluorescence of OH (OH-PLIF): A Spectra-Physics Quanta-Ray single-cavity Nd:YAG laser pumps aSpectra-Physics optical parametric oscillator (OPO) toproduce the laser excitation for the OH-PLIF technique. TheOPO is tuned to near 568 nm and the light is doubled using aBBO crystal to approximately 284 nm to excite the Q1(9),Q2(8), and P1(5) transitions of OH that merge as a result ofpressure broadening. The 284-nm laser beam is co-alignedwith the 355-nm laser beam for H2CO-PLIF upstream of thesheet-forming optics. As a result, the laser sheet at 284 nm isapproximately the same size as that at 355 nm: 30 mm wideand up to 1 mm thick. The 284-nm laser pulse energy at theengine is 19 mJ per pulse. LIF emission is collected atwavelengths shorter than 385 nm with an intensifiedPrinceton Instruments PI-MAX 3 ICCD camera with a Super-blue, slow-gate intensifier (Gen-II) and a resolution of1024×1024, a gate time of 410 ns, and at maximum gain. Thecamera is equipped with an ultraviolet (UV) Nikkor f/4.5lens. Three filters, a 65-nm wide blocked bandpass filter at310 nm, a 15-nm wide unblocked bandpass filter at 310 nm,and a color-glass SWG305, reject laser scatter and otherinterference outside the OH fluorescence band near 308 nm.Detuning the laser off the transition at 284 nm producesalmost no signal through the filter pack, indicating that thesignal shown in the OH-PLIF images is almost entirelyfluorescence of OH.

For both laser-induced fluorescence techniques, the LIFdata are limited to one frame per cycle, due to repetition-rateconstraints of both the laser and camera systems. The OH-LIF signal is collected 1 microsecond before that of theH2CO-LIF signal, so that they are virtually simultaneousrelative to engine time scales. As shown in Figure 3, thefluorescence emission is collected through the large piston-crown window and directed to the two ICCD and one high-speed complementary metal oxide semiconductor (CMOS)camera using two beamsplitters. A long-wave-passbeamsplitter with a cut-off near 385 nm directs UV light tothe OH-PLIF camera. The remaining light is split by a 50%reflectance broadband beamsplitter between the H2CO-PLIFcamera and the high-speed CMOS camera (described below).

These two LIF techniques are used together tosimultaneously image intermediate species after first- andsecond-stage ignition. As described above, H2CO is a marker




of UHC remaining after first-stage ignition, while OH is amarker of second-stage ignition [11]. It has been shown inprevious studies as well as the current work that these twosignals generally do not overlap within the plane of the lasers,and that the location where they meet is an indication of thetransition from first- to second-stage combustion. Anexample of this is Figure 6, which is annotated to show eachregion. The figure shows a partial view from below of thecombustion bowl, with the injector on the left (white dot) andthe wall of the right-cylindrical bowl on the right (asindicated by a curved white line). The format of this figurewill be used throughout this manuscript to show thedifference between H2CO-PLIF and OH-PLIF signal.

Figure 6. Example of simultaneous OH-PLIF (green)and H2CO-PLIF (red) imaging at 373 CAD for an LTC,

single-injection case at 12.6% intake O2, SOI1C=352CAD, DOI1C=800 microseconds. The injector and bowlrim locations are indicated by the white dot and curved

line, respectively.

In Figure 6, the two false-colors indicate the two speciesunder consideration: red for H2CO and green for OHfluorescence. It is clear that within the plane of the laser, theH2CO and OH fluorescence overlap very little, which wouldappear as yellow in the image. The lack of overlap isconsistent with these two species marking the first- andsecond-stage ignition regions. Later in the cycle, generallyafter OH fluorescence is first detected, the 355 nm laser canexcite not only H2CO molecules, but also poly-cyclicaromatic hydrocarbon (PAH) molecules, which areprecursors to soot, as well as possible laser-inducedincandescence (LII) of soot itself. The emission spectrum andsignal intensity of fluorescence from these laser-excited PAHmolecules are different from H2CO, and hence can be used todiscriminate between the two sources.

Our previous studies using OH- and H2CO-PLIF imagingused a spectrometer to discriminate between H2CO and PAHfluorescence and/or soot LII [16]. The H2CO fluorescencespectrum shows a series of characteristic bands within the

filter bandpass, while both PAH fluorescence and soot LIIspectra are relatively broadband. Additionally, inspection ofthe images showed that the intensity of the broadband PAH/soot emission was almost universally much brighter than theH2CO emission, which remained relatively unchanged untilits disappearance at second-stage ignition. The closecorrelation between the spectral signatures and intensitypatterns of H2CO and PAH/soot allow discrimination byintensity, without using a spectrometer. For instance, thebroad, relatively moderate intensity fluorescence from 355-nm excitation on the left half of Figure 6 (red color in figure)is characteristic of formaldehyde fluorescence, while thesmaller, more isolated and intense pockets of emission on theright side of the image are typical of PAH fluorescenceand/or soot LII, as labeled in Figure 6. Hence, to allowsimultaneous imaging with all three diagnostics, theseintensity signatures, rather than fluorescence spectrameasured with a spectrometer, were used in the current studyto discriminate between formaldehyde and PAHfluorescence.

High-speed natural luminescence imaging: The thirdtechnique uses a Phantom 7.1 CMOS camera equipped with aNikkor 105 mm, f/1.8 lens to record natural luminescencefrom combustion reactions and combustion-generatedproducts. The camera framing is triggered by the shaftencoder at a resolution of one-half °CA, and the exposureduration is 50 microseconds. Although first-stage ignitionreactions yield some weak chemiluminescence thatsometimes can be detected with CMOS arrays, in the currentstudy the intensity is too weak after the two beam-splitters forthe laser diagnostics. Hence, any ignition chemiluminescencevisible in the high-speed images is from second-stageignition. Natural soot incandescence, when present, is alsorecorded by the camera. Exposure times were limited to 50microseconds to avoid over-exposure from the high-intensitysoot incandescence. The high-speed natural luminosityimages are not presented in the present study, but informationgleaned from inspection of the images (e.g., spray/jetpenetration and actual injection start/dwell times), isincluded.

Test MatrixThe test matrix includes engine operating conditions at a

variety of commanded injection timings to investigate theeffect of post-injection penetration and ignition delay. Anoverview of the text matrix is in Table 4.

As described in the Introduction, changes in post-injectionduration are used to investigate the effect of post-injectionmixing and penetration on UHC reduction, while changes inthe phasing of fuel-injection timing are used to vary theignition delay time of the post injection. The baseline main-injection command timing (SOI1C) is 352 CAD. Duringsingle-injection tests, the command duration of the injection(DOI1C) is varied to measure the baseline engine-out UHCemissions in the absence of post injections. The single-




injection DOI1C sweeps are tested at three injection timings,SOI1C=352 CAD, 349 CAD, and 355 CAD. Changes in theinjection timing alter the combustion phasing, and as a result,the unburned hydrocarbon emissions (and other emissions,such as NOx and potentially PM, though not measured here).

Table 4. Engine operating test matrix with commandedinjection timings and durations.

SOI2C is selected to place the post injection close to themain injection while maintaining a clear separation betweeninjections. Engine load variation, previous momentum-basedinjection rate measurements [16], and optical imaging of thesprays (not shown here) indicate that the separation betweenthe main and post injections becomes unstable when thecommand pulses are too close. Within a command resolutionof 0.25 °CA (∼35 microseconds at 1200 RPM), the closestpossible SOI2C yields consistently separate injections with areal dwell between injections of about 0.5 °CA. AdvancingSOI2C further can lead to unstable injector operation, withsome main injection events merging into the post injectionand yielding higher load and fuel delivery for those cycleswith merged injections. Conversely, if SOI2C is retarded byonly 0.25 °CA from the closest consistent timing, the realdwell between injections increases by 1.5 °CA, to about 2°CA. Thereafter, additional retarding of SOI2C producesequivalent increase in the real dwell between injections (i.e. afurther delay of SOI2C by 0.25 °CA also increases the dwellby 0.25 °CA). This initial non-linear relationship betweencommanded and real injection dwell is attributed to internal

injector dynamics that occur when the injection separation isvery small.

To provide a center point for study of the effects of thedwell between injections, we selected the baseline SOI2Ctiming that yields a real injection dwell of 2 °CA. Postinjections are first added to the baseline SOI1C=352 CADmain-injection timing with three different main-injectiondurations. A sweep of DOI2C at each of these three DOI1Ccases is used to observe the variation in UHC emissions withload; longer main-injection durations change the “baselineload” for the DOI2C sweeps. Changes in DOI2C are intendedto explore the role of mixing and penetration of the post-jeton UHC reduction. Additionally, SOI2C is varied holdingconstant SOI1C=352 CAD and DOI1C=1000 microseconds.These tests are used to investigate the effect of post-injectiondwell on engine-out UHC levels. Here, three SOI2C timingsare used, 363.25 CAD, 363.5 CAD, and 365 CAD. Asdescribed above, and according to high-speed video, thedwell times between the end of the main injection and thebeginning of the post injection for these commanded timesare 0.5 °CA, 2 °CA, and 3.5 °CA, respectively.

Finally, data are obtained for DOI2C sweeps at threedifferent SOI1C timings with the dwell between injectionsheld constant. These cases are used to illustrate the effect ofpost-injection ignition delay on UHC emissions. As discussedabove, these changes in SOI1C do not significantly changethe ignition delay of the near-TDC main injection, but theignition delays of the post injections do change significantlybecause of the later timings (after TDC) of the postinjections. An overview of the injection timings of thevarious sweeps, along with start of main combustion (SOMC)and start of post-injection combustion (SOPC) is in Table 4.Eight families of timings, all of which are described in thetest matrix in Table 4, are included.

Table 5. Start of combustion of both main and postinjections for eight injection-timing schemes.

The SOMC is defined as the minimum of the AHRR afterthe rise in heat release due to first-stage combustion, asindicated in Figure 7. The SOPC is calculated using thedifference between the single-injection and main- plus post-injection AHRRs, as shown in Figure 7. To calculate theSOPC, the AHRR of the single-injection schedule issubtracted from the AHRR of the post-injection schedule.




The resulting profile is relatively flat until the start of the postinjection, where it dips below zero due to the vaporization ofthe post-injection fuel. Thereafter, it rises and crosses zero asthe net heat release from combustion of the post-injectionfuel becomes positive. Here, we take this zero-crossing to beindicative of the beginning of heat release from the postinjection.

Figure 7. Definition of SOMC and SOPC using thedifference between the single-injection and main- plus

post-injection cases, for SOI1C=352 CAD, DOI1C=1000microseconds, SOI2C=363.5 CAD, DOI2C=400

microseconds.

Ignition delay times for both the main injection (IDmain)and post injection (IDpost) were estimated using the SOMC/SOPC calculated from the AHRR curves and the actual startof injection, taken from Mie scattering images in the high-speed videos. The results of this calculation are shown inTable 6. These videos have a resolution of 0.5 °CA, and sothe uncertainty in the ignition delay estimates is 0.5 °CA.

Table 6. Ignition delay estimates of both main and postinjections for eight injection-timing schemes.

The ignition delays in Table 6 for both the main and postinjection confirm that when the whole injection schedule isshifted by 6 °CA, the ignition delay of the near-TDC main-injection changes only slightly (total change of 0.5 °CA),while the ignition delay of the post-injection changes much

more, by over 2 °CA. These differences, as explained above,are used to test the effect of post-injection ignition delay onUHC reduction.

RESULTSThe results are presented in three sections. First, the single

injection results are shown to provide baseline UHCmeasurements without post injections. Sweeps of DOI1C atthree different SOI1C times are used as baselines in thefollowing two sections. Next, we present results fromvariation in DOI2C with SOI1C, DOI1C, and SOI2C fixed atvarious values. These data are intended to describe the effectof post-jet mixing and penetration on the efficacy of postinjections for reducing UHC emissions. Finally, both single-injection and main- plus post-injection data at multiple SOI1Cwith a constant dwell between injections and sweeps inDOI2C are presented to illustrate the effect ignition delay onthe efficacy of the post injection at reducing UHC. Theengine-out UHC emissions data are accompanied by imagesfrom the two laser diagnostic techniques to help explain thetrends.

Single-Injection BaselineUHC emissions data for single-injection duration sweeps

at three SOI1C timings of 349 CAD, 352 CAD, and 355 CADare plotted in Figure 8. These results show a significantvariation in UHC emissions with both SOI1C and DOI1C. Foreach SOI1C, the engine-out UHC level is quite sensitive toinjection duration. Very short injections, resulting in loadsdown to approximately 100 kPa gIMEP, produce highengine-out UHC, up to almost 1400 ppm (C1) after correctingfor continuous firing. High-speed luminosity imaging datasuggest this is in part due to partial misfires at these very low-load conditions. Injections at mid-length DOI1C produce theleast UHC emissions, down to about 300 ppm. After reachinga minimum, the engine-out UHC levels begin to increase withincreasing load. Curiously, long injections at all three SOI1Cproduce very similar levels of engine-out UHC; the threecurves in Figure 8 asymptote to nearly the same UHC levelfor the higher loads.

Also noticeable in Figure 8 is that the later main-injectiontiming (SOI1C=355 CAD) produces significantly more UHCthan the earlier injection timings, particularly at short DOI1C.This trend is likely driven by the phasing of the heat releasewith respect to TDC. For SOI1C=349 CAD, the start of maincombustion is approximately at TDC. As SOI1C is delayed,so is the SOMC. For SOMC that is timed after TDC, thepressure and temperature reduction occurring during theexpansion stroke likely leads to more incomplete burning andhigher UHC emissions, even though the main ignition delayis similar for all three timings.




Figure 8. UHC [ppm C1] emissions as a function of loadfrom single-injection tests at varying DOI1C for threedifferent SOI1C. UHC emissions have been corrected

from skip-fired operation to reflect those of continuous-fired operation.

The development of the H2CO- and OH-PLIF for a singleinjection is shown in Figure A1 in the Appendix, whichshows LIF measurements from the SOI1C=352 CAD,DOI1C=1000 microseconds timing. At the first measuredcrank angle, 363 CAD, the spray (false-colored green) is stillvisible due to some Mie scattering of the 284-nm laser-lightthat leaks through the filters. Formaldehyde (false-coloredred) is formed within the vapor-fuel/air mixture of the spray,primarily in the downstream region, as the jet penetrates tothe bowl wall. After its initial appearance near 363 CAD, theformaldehyde fluorescence signal strength quickly increases,reaching is greatest brightness near 365 CAD. Immediatelythereafter, at 366 CAD, some mixtures near the bowl rim oneither side of the jet axis transition to second-stage ignition,as indicated by the OH-PLIF signal (in green). The OH-PLIFimages show that second-stage ignition begins in the rollerregions at the wall and grows toward the jet centerline,generally forming an unbroken band along the bowl rim by369 CAD. During the progression of the OH toward thecenterline, the H2CO-PLIF extends farther and fartherupstream, eventually reaching the injector by 369 CAD. Thistemporal upstream progression of H2CO-PLIF is differentthan the generally uniform initial appearance previouslyobserved in this facility [11], which may be due to the changeof injectors for this study. The downstream signal strength ofthe H2CO-PLIF also decreases somewhat as the OH-PLIFprogresses toward it.

The growth of the second-stage ignition region near thebowl rim, as indicated by the OH-PLIF signal, continuesthrough approximately 376 CAD. Small pockets of brightemission appear in some of the H2CO-PLIF images (red) asearly as 371 CAD. As discussed in the Engine Diagnostics

section, these bright regions can be attributed to PAHfluorescence and/or soot LII as indicated by the annotationson the Figure 6 (also see confirming spectral data in Ref. [22]for similar conditions). During the same time, the intensity ofH2CO-PLIF closer to the injector decreases, but does notcompletely disappear, as it broadens spatially, spreading overthe whole near-injector region.

Later in the cycle, the OH-PLIF emission generallyremains confined to the outer bowl, rarely appearing within25 mm from the injector, while the H2CO-PLIF remains inthe same region of the residual jet where OH is absent, nearthe injector. Both the persistence of H2CO and the absence ofOH in the near-injector region indicate that the fuel/airmixture near the injector tip never transitions to second-stageignition. Hence, this region is the origin of at least some, andpotentially most [12], of the UHC in the exhaust.

At each of the three SOI1C timings for the single-injectionconditions and over the entire range of DOI1C, the generalcharacter of H2CO- and OH-PLIF distributions (not shownhere) is similar to that in Figure A1, indicating that there isroom for improvement in the UHC emissions at a variety ofengine operating conditions. As discussed above, postinjections can help by enriching the fuel-lean mixtures nearthe injector to provide the additional “kick” needed to pushremaining fuel in the center of the bowl into second-stageignition. In the next two sections, we describe the behavior ofthese post injections and the mechanism by which theyreduce unburned hydrocarbons.

Post-Injection Penetration SweepIn this portion of the study, the commanded main-

injection timing (both SOI1C and DOI1C) is held constantand the duration of the post injection is varied to test theeffect of post-injection penetration and mixing on UHCemissions reduction. As was discussed in the introduction, weexpect that the post injection is helping to enrich the overly-lean fuel/air mixture near the injector. Here, we have chosenthree main-injection timings as baselines to which the post-injection effects on UHC emissions may be compared. Anexample result of this exercise can be seen in Figure 9, wherea cubic curve-fit has been added for reference to the singleinjection engine-out UHC data of the SOI1C=325 CADcondition.

As described earlier, SOI2C for the post injection is set toyield a real dwell between injections of 1.5 °CA, asevidenced by optical imaging. The engine-out UHCemissions shown in Figure 9 indicate that a range of post-injection durations, from DOI2C = 250 microseconds toDOI2C = 800 microseconds, leads to decreases in UHC. Theshortest post-injection command duration, 200 microseconds,does very little to reduce UHC, and this is probably becausethe real injection duration is very short and inconsistentduring these tests. High-speed Mie-scatter imaging of thepost-injection spray shows that penetration and even the




emergence of fuel from the injector varies significantly fromcycle to cycle for the very shortest post injections. Postinjections with longer durations, however, are much moreconsistent and are effective at reducing UHC. The greatestreduction in UHC, 27% from a single injection at the sameload, is achieved with DOI2C = 400 microseconds. Thisreduction agrees well with the value observed in the previouswork [16] at a single post-injection operating condition usinga fuel with more typical diesel volatility (Figure 2). Theagreement between the results in Figure 9 and Figure 2indicate that the characteristic behavior of UHC reduction bypost injections is not affected by the low volatility of the n-heptane fuel used here.

Figure 9. UHC [ppm C1] emissions as a function of loadfrom single-injection and main- plus post-injection tests

at varying DOI2C, where SOI1C=352 CAD, DOI1C=1000microseconds, and SOI2C=363.5 CAD. UHC emissionshave been corrected from skip-fired operation to reflect

those of continuous-fired operation.

Figure 10 shows the engine-out UHC levels for threesweeps of DOI2C starting from three baseline loads of 209,265, and 438 kPa gIMEP, corresponding to DOI1C=800,1000, and 1600 microseconds, respectively. For each sweepof post-injection duration in Figure 10 (open symbols), thepost-injections are added to a main injection with SOI1C andDOI1C held constant, with the durations noted above. In eachcase, the post injections are effective at reducing the UHClevels, and the greatest UHC reduction in all three cases wasrealized at DOI2C values of approximately 400microseconds, which corresponds to a load increase of about100 kPa gIMEP from the addition of the post injection.Hence, DOI1C (i.e., main-injection load) seems to have littlebearing on the optimal DOI2C for UHC reduction.

One explanation for the similarity in the DOI2C yieldingthe minimum UHC emissions at three loads may be thelength of post-jet penetration versus the bowl size. High-speed chemiluminescence imaging shows that with a post-

injection duration of 400 microseconds, the combusting post-jet penetrates to the wall. However, it does not have sufficientmomentum to flatten against the wall and roll up to the sideswithin the time available before rapid cylinder expansion(before approximately 380 CAD) to the same degree aslonger post injections do. Hence, the 400-microsecond postinjection may be sized correctly to enrich the overly-leanareas without any additional roll-up along the bowl wall,where it would deposit fuel outside the overly-lean regionnear the injector. If the bowl diameter was much larger,resulting in a larger overly-lean region in the center, a longerpost-injection duration might be optimal, at least with thesame injection characteristics. Although this scenario iscertainly plausible, there may be other factors that control theoptimal post-injection duration, including injection ramp-rateand other injector dynamics issues.

Figure 10. UHC [ppm C1] emissions from single-injection and main- plus post-injections tests at varying

DOI2C and three DOI1C, where SOI1C=352 CAD,DOI1C=800, 1000, and 1600 microseconds, and

SOI2C=362, 363.5, and 368 CAD, respectively. UHCemissions have been corrected from skip-fired operation

to reflect those of continuous-fired operation.

Finally, the dwell between the end of the main injectionand the start of the close-coupled post injection has littleeffect on the ability of the post injection to reduce engine-outUHC. Figure 11 illustrates this by showing a range of post-injection durations at three post-injection injection timings,SOI2C=363.25 CAD, 363.5 CAD, and 365 CAD. Asdescribed above in the Experimental Overview section, thesecommanded timings resulted in actual injection dwell timesof 0.5 °CA, 2 °CA, and 3.5 °CA. The reduction in engine-outUHC is nominally the same, regardless of SOI2C. Nearly allthe points lie along the same curve, the exception being theshorter DOI2C post-injections at the longest dwell betweeninjections.




Figure 11. UHC [ppm C1] emissions from single-injection and main plus post-injections tests at varying

DOI2C and three SOI2C, where SOI1C=352 CAD,DOI1C=1000 microseconds, and SOI2C=363.25, 363.5,and 365 CAD, respectively. UHC emissions have beencorrected from skip-fired operation to reflect those of

continuous-fired operation.

H2CO- and OH-PLIF data provide insight into the effectof the post injection on the combustion process at thecondition with SOI1C=352 CAD, DOI1C=1000microseconds, SOI2C=363.5 CAD, and DOI2C=400microseconds; this is the minimum UHC post-injectioncondition for this baseline load and timing (see Figure 9).

A sequence of simultaneous H2CO and OH-PLIF imagesin Figure A2 in the Appendix shows the effect of the postinjection on residual UHC from the main injection. Similar tothe single-injection case, first-stage ignition begins before theend of the main injection, as is evidenced by the Miescattering of the 284-nm laser by the liquid fuel and theformaldehyde signal present at the head of the jet at 363CAD. Again, the formaldehyde signal increases in strength asthe jet penetrates against the bowl wall and the mixture flowsalong the wall. The beginning of the post injection can beseen at 366 CAD via the Mie scattering of the 284-nm laserlight that leaks through the filter pack.

Second-stage ignition of the main-injection mixturebegins shortly after the start of the post injection, evidencedby the appearance of OH-PLIF along the bowl wall at 367CAD, similar to the single-injection condition. The region ofsecond-stage ignition grows in the outer bowl and then at 370CAD, the upstream fuel jet itself shows signs of second-stageignition along the outer periphery of the jet. The inner regionsof the fuel jet are still dominated by H2CO-PLIF signal,indicating that the inner jet has not yet transitioned to second-stage ignition. By 374 CAD, second-stage ignition progressesto the center of the downstream fuel jet, and the reactionstarts to “race back” along the fuel jet, causing second-stage

ignition to occur closer to the injector than in the single-injection case (cf. Figure A1). Similar “racing back” has beenseen in other studies of UHC [16] and soot [17] in LTC.

Although we do not have equivalence ratio data for themixture near the injector after the end of the post injection,certain features of the PLIF images in the post-injection casesuggest that the enriching of the fuel/air mixture near in theinjector helps to reduce the UHC emissions. It is evident fromthe images that the formaldehyde signal after the end of thepost injection, between 370 and 373 CAD is clearly in theform of a jet and is indicative of the initial first-stagechemistry of the post injection itself. This is in contrast to thesame times in the single-injection case, where theformaldehyde signal is much more diffuse as the mixtureconsists of the overly-lean fuel/air mixture residual from themain injection.

The evidence for the efficacy of the post injection lies inthe greater extent to which the OH-PLIF signal appearsupstream in the jet; this indicates that more of the mixture is“kicked” into the second-stage ignition when a post injectionis present. For example, comparing 375 CAD in the single-injection case (Figure A1) and the same timing in the post-injection case (Figure A2) shows that much more widelydistributed second-stage combustion is occurring when a postinjection is added than when one is absent. This enhancedsecond-stage combustion continues throughout the rest of thecycle.

This mechanism for UHC reduction works independentlyof any obvious bulk-temperature effect. During combustionof the post-injection fuel, shown in the AHRR traces inFigure 4, the cylinder pressure during main- plus post-injection operation is less than that of single-injectionoperation at the same load, indicating that the bulk cylindertemperature is also less. The UHC reduction by postinjections is hence driven by the enrichment and subsequentignition of the overly-lean fuel/air mixture at the center of thebowl, and not by an increase in the bulk temperature.

These optical results indicate that the post injection has asignificant effect on the characteristics of the second-stagecombustion and hence the level of UHC at the end of thecycle. Engine-out UHC emissions measurements indicate,though, that not all post-injection durations are as effective atreducing UHC as others. There may be several reasons forthis, some of which are discussed here. While theseexplanations are certainly plausible, we do not have enoughexperimental evidence at this time to definitively confirm orreject any of them; evaluating these explanations is the focusof future work.

First, post injections of varying length may also have verydifferent injection profiles, a result of injector and raildynamics. For example, the two post-injection profiles shownin Figure 2a from Chartier et al. [16] are somewhat different,both in their height as well as their shape. Previous analysesof both free diesel jets [23] as well as engine experiments [24,25] have shown that injection rate profile can change in-cylinder mixing and exhaust emissions. The differences in




injection rate profile between short and long post injectionsmay affect mixing and thereby the equivalence ratiodistribution in the residual overly-lean region from the maininjection. Consequently, the completeness of combustion,which depends on the local equivalence ratio, may depend ondifferences in the rate profile with post-injection duration,which will ultimately affect the engine-out UHC levels.

Second, increasing DOI2C reduces the time available formixing prior to the SOPC (i.e., the ignition dwell for the post-injection). Much like the main injection, the mixture in thenear-injector region resulting from the post injection will beincreasingly fuel-rich as the ignition dwell decreases (seeFigure 1). Hence, increasing DOI2C moves the end ofinjection closer to SOPC, which may reduce UHC emissionsby creating richer mixtures at SOPC. Very long postinjections will even have a negative injection dwell, whichcould significantly reduce UHC emissions by avoidingoverly-lean, near-injector mixtures altogether, but likely atthe expense of increased soot formation.

Additionally, the source of UHC may vary as a functionof injection duration and load. As was noted above withreference to Figure 8, the engine-out UHC levels seem toasymptote to a single line, regardless of injection timing. Aswill be discussed in the next section, this trend continues withthe addition of post injections at each of these timings. Thismay be indicative of a switch from lean sources of UHC -overly-lean mixtures found near the injector - to rich sourcesof UHC - fuel-rich pockets that have insufficient time to mixand burn in the late-cycle mixing-controlled combustion.

The trade-off between these effects - UHC dependency onpost-injection duration and transition between lean and richsources of UHC - may explain the existence of an optimalpost-injection duration. At this duration, these effects maybalance to result in a lower UHC level with post injections.

Post-Injection Ignition Delay SweepIn addition to post-injection penetration, another

controlling factor in post-injection efficacy is the ignitiondelay time of the mixture formed by the post injection.Decreasing the ignition delay time provides less time for themixture in the near-injector region created by the postinjection to become overly lean, so that combustion mayproceed farther to completion in that region.

As discussed at the end of the previous section, changes inignition dwell by altering DOI2C may play a role UHCreduction, but the effect is confounded by changes in mixingand penetration with changes in DOI2C. In this section, weattempt to isolate the effects of post-injection ignition delayfrom spray penetration and mixing by keeping the injectiondurations and dwells fixed. Here, the entire injectionschedule, including the main injection, is advanced ordelayed by 3 °CA relative to the baseline point (SOI1C=352CAD, SOI2C=363.5 CAD). The UHC emission results forjust the single injections at the baseline, advanced, anddelayed times are discussed with reference to Figure 8. Figure

12 shows these same data, now with the UHC emissionresults for the injection schedules with post injections startingfrom baseline points with DOI1C=1000 microseconds.

Figure 12. UHC [ppm C1] emissions as a function ofload from single-injection and main- plus post-injection

tests at varying DOI2C for three different SOI1C, andDOI1C=1000 microseconds for the schedules with postinjections. UHC emissions have been corrected from

skip-fired operation to reflect those of continuous-firedoperation.

Overall, the post-injection trends with DOI2C at the threeSOI1C timings are very similar. Very short post injections donot produce significant reduction in UHC, but as the post-injection duration increases, the benefit of the post injectionincreases until a minimum UHC level is reached. After thatminimum level, longer post injections still reduce the UHCemissions relative to a single-injection at the same load, butto a lesser extent. Similar to the single-injection UHC resultsin Figure 8, the UHC levels for both single and main- pluspost-injection schedules seem to asymptote to a single curveat higher loads.

An alternative perspective to Figure 12 is presented inFigure 13, which shows the percentage reduction (a) as wellas the absolute reduction (b) in UHC compared to a singleinjection at the same load. These values are calculated by firstfitting a cubic function to the single-injection UHC results(see Figure 8, for instance) and calculating what a single-injection UHC level would be for the same load at each post-injection condition. Then the percentage or absolute reductionfrom that baseline UHC level is calculated.

Post injections at the earliest timing schedule, SOI1C=349CAD, were most effective at reducing the engine-out UHClevels on a percentage-wise basis. A post injection withduration of 400 microseconds reduced the UHC level themost, 34%, as compared to a single injection at the sameload. The late-timing schedule, SOI1C=355 CAD, inhibitedthe performance of the post injection, and the maximum




reduction in UHC levels was only 14% at a post-injectionduration of 350 microseconds.

Figure 13. Percentage reduction (a) and absolutereduction in ppm C1 (b) in engine-out UHC emissionswith the use of a post injection as compared to a singleinjection at the same load for three injection timings.

Additionally, Figure 13b shows the absolute reduction ofUHC (in ppm C1) at all three timing schedules. As was notedpreviously, the very short post injections do little to abateUHC emissions, but post injections with durations greaterthan 350 microseconds can have a significant impact onengine-out UHC. It is interesting to note that over a range ofpost-injection durations, from approximately 400microseconds to 700 microseconds, the post injection reducesengine-out UHC by approximately 80 ppm C1, regardless ofmain-injection timing (SOI1C). While the absolute UHCreductions among the three injection schedule phasings inFigure 13 are quite similar, the results are from a limited setof data, so the 80 ppm UHC reduction should not begeneralized as a limit for how much UHC may be addressedwith post injections. For these three cases, it may becoincidental that the enhanced-oxidation mechanismsdescribed with reference to Figure A2 result in roughly the

same absolute UHC reduction. Other post-injection strategiesnot examined here (e.g., multiple post injections) couldconceivably reduce UHC even further, especially for thehigher UHC conditions.

To visualize why one injection schedule was moreeffective (on a percentage basis) than another, we can look atthe effect of the post injection on combustion using the twoPLIF techniques. Figure A3 in the Appendix shows thebehavior of combustion with the post injection for all threeSOI1C timings at three points within the cycle. In each of thecases, the post-injection duration is 400 microseconds, whichis the approximate minimum in the engine-out UHC for eachmain-injection timing. The first row is for the advancedinjection schedule, the middle row is for the baselineinjection schedule, and the bottom row is for the retardedinjection schedule.

The first column depicts the H2CO and OH distribution 4°CA after the end of the main injection, which is close to thebeginning of the post injection. At this point in the cycle, thethree cases look very similar, most likely because the ignitiondelay of the main injection is quite similar. In each case,formaldehyde has formed along the jet and second-stagecombustion has begun along the wall on either side of the jetaxis, as indicated by the green OH signal on either side of thejet. The shape and size of these second-stage combustionregions are similar between the three injection schedules.

The next column shows images 7 °CA later,approximately 4 °CA after the end of the post injection. In thebaseline injection schedule (center row), there is still someformaldehyde in the jet, but the second-stage combustion,marked by green OH signal, has raced back along the jet axispart way to the injector, converting unburned hydrocarbons tofinal combustion products. For the retarded injection schedule(bottom row), the second-stage combustion has raced backalong the jet axis somewhat less; here, second-stagecombustion stays confined to the outer bowl nearer to thebowl wall. For either the baseline or the retarded timing, thesecond-stage combustion never extends into the overly-leanregion closer to the injector. Hence, remaining formaldehydein the vicinity of the injector (red color in images) is stillevident. By contrast, for the advanced injection schedule,almost no formaldehyde is detectable, and the second-stagecombustion has raced back along the entire jet axis, stretchingfrom the bowl back to the injector. Here, it is quite evidentthat the post injection helps to combust much of the UHC inthe overly-lean region.

Finally, in the late stages of combustion, shown in theright column of Figure A3, the differences in the efficacy ofthe post injection for each of the shifted injection schedules isalso clear. For the advanced injection schedule, almost noformaldehyde fluorescence is visible in the field of view andOH fluorescence is quite strong and extends over a relativelylarge region of the bowl. For the baseline injection schedule,the OH fluorescence is confined to the outer bowl, and weakformaldehyde fluorescence remains in the near-injector




region. This is indicative of UHC remaining near the injector.Finally, for the retarded injection schedule, a very strongformaldehyde signal remains in a larger area of the bowl thanfor the baseline schedule. This correlates well with theengine-out UHC results, showing that the retarded injectionschedule result in higher UHC levels.

This marked change in the combustion characteristics forthe retarded injection schedule is liked due to the increasedpost-injection ignition delay. As was discussed above, theadvanced injection schedule results in the shortest ignitiondelay of the post-injection mixture. As the ignition delayincreases for the baseline and then for the retarded injectionschedule, the mixture formed near the injector has more timeto lean-out again after the end of the post injection, resultingin a second occurrence of the over-leaning issues. Forconditions with short post-injection ignition delays, the postinjection is more able to “kick” the newly enriched mixtureinto the second-stage chemistry regime quickly due to shortignition delay and favorable thermodynamic conditions.

The long-ignition delay data in Figure A3 (bottom row)also indicate that significant reductions in UHC are stillpossible if a single or a main-plus-post injection can betailored to cause second-stage ignition in the near-injectorregion. Rate-shaping is one potential method to do so, andprevious analyses indicate that richer mixtures can bedesigned by slowing the ramp-down rate [23]. Traditionally,rich mixtures near the injector have been undesirable,especially from a soot-emission standpoint, so most injectorsare designed to ramp down rapidly. For partially premixedLTC conditions, however, it may be advantageous to end theinjection more slowly to create longer lasting rich mixturesthat will achieve second-stage ignition during the ignitiondwell before becoming too overly fuel-lean.

Finally, although accurate fuel consumption data are notavailable for the skip-fired optical engine, and we did notmeasure any engine-out emissions other than UHC, we canoffer our expectations on fuel efficiency and other emissions.Close-coupled post injections should have only minor effectson fuel efficiency and NOx emissions, but could increase sootemissions, depending on operating conditions. The closecoupling between the main and post injection keeps the heatrelease in a portion of the cycle that is beneficial for fuelefficiency, and the post-injection improves combustionefficiency (at least for UHC), so the efficiency should besimilar, or even improved with the post injection schemesinvestigated here. The high rate of EGR used at LTC is thesignificant controlling factor in NOx emissions, and smallchanges in the injection schedule, such as those from asingle-injection to main- plus post-injection schedule, will notchange NOx emissions significantly [26, 27]. But, due to theenrichment effects of the post injections, we expect that sootemissions could increase, especially for the longer postinjection durations. Although we did not measure exhaustsoot, we did observe that soot luminosity increased markedly

for the longer post-injections, so we would expect thatexhaust soot might also increase for long post injections.

CONCLUSIONS AND FUTURE WORKIn this work, we have investigated the effect of close-

coupled post-injection timing and duration on engine-outUHC emissions compared to single-injection operation atlow-load, EGR-diluted LTC conditions in a heavy-dutyoptical diesel engine. Planar laser-induced fluorescenceimaging of formaldehyde (H2CO) and the hydroxyl radical(OH) provided insight into the effects of post-injections onthe in-cylinder spatial and temporal progression of first- andsecond-stage ignition chemistry. Key observations andconclusions are:

1. Over the range of conditions studied, close-coupledpost injections reduce UHC emissions by as much as 34%compared to a single injection at the same load.

2. The most important parameter affecting close-coupledpost-injection UHC emissions reduction is the duration of thepost injection. Across ranges of start of main injection,duration of main injection, and dwell between main and postinjections, the minimum UHC emissions always occurrednear a post-injection command duration of 400 microseconds,corresponding to a load increase of about 100 kPa gIMEPfrom the addition of the post injection.

3. The second most important parameter affecting close-coupled post-injection UHC emissions reduction is theignition delay of the post-injection mixture. At the optimalpost-injection durations, conditions with a short (3.4 °CA)post-injection ignition delay achieved over twice thepercentage UHC reduction as that for a long (5.5 °CA) post-injection ignition-delay condition.

4. Optical measurements indicate that the optimal post-injection (duration of approximately 400 microseconds)causes second-stage ignition to extend farther upstream thanwith single injections. As a result, previously measuredoverly fuel-lean mixtures near the injector achieve morecomplete combustion, most likely because they are enrichedby the post-injection so that the chemical kinetics proceedmore rapidly to complete combustion.

5. The optical data also show that with the exception ofthe conditions with the shortest post-injection ignition delays,UHCs remain within about a 25 mm radius of the injector latein the cycle, even at the optimal post-injection duration. Thissuggests that much greater UHC reductions may be possiblefor all of those conditions if the post-injection can be tailored(e.g., by rate shaping) to promote the transition to second-stage ignition in those near-injector regions similar to thatachieved for the short post-injection ignition delay condition.

In future work, further analysis of OH and H2CO-PLIFimages will elucidate the effects of post injections on the fateof in-cylinder UHCs over a much broader range of injectiontiming and duration of both the main and post injections. This




work will help to define how in-cylinder mixing andchemical kinetic mechanisms can be altered by adjustinginjection schedules. Ultimately, the goal is to define rules fordesigning injection schedules for low UHC emissions at LTCconditions.

REFERENCES1. Alriksson, M., Rente, T., and Denbratt, I., “Low Soot, Low NOx in a

Heavy Duty Diesel Engine Using High Levels of EGR,” SAE TechnicalPaper 2005-01-3836, 2005, doi:10.4271/2005-01-3836.

2. Tree D R and Svensson K I, “Soot processes in compression ignitionengines” Prog. Energy Combust. Sci. 33(3):272-309 (2007)

3. Keeler, B. and Shayler, P., “Constraints on Fuel Injection and EGRStrategies for Diesel PCCI-Type Combustion,” SAE Technical Paper2008-01-1327, 2008, doi:10.4271/2008-01-1327.

4. Hardy, W. and Reitz, R., “A Study of the Effects of High EGR, HighEquivalence Ratio, and Mixing Time on Emissions Levels in a Heavy-Duty Diesel Engine for PCCI Combustion,” SAE Technical Paper2006-01-0026, 2006, doi:10.4271/2006-01-0026.

5. Kanda, T., Hakozaki, T., Uchimoto, T., Hatano, J. et al., “PCCIOperation with Fuel Injection Timing Set Close to TDC,” SAETechnical Paper 2006-01-0920, 2006, doi:10.4271/2006-01-0920.

6. Musculus M P B, Miles P C, Pickett L M, “Conceptual models forpartially premixed low-temperature diesel combustion” Prog. EnergyCombust. Sci. (in press) (2012)

7. Kimura, S., Aoki, O., Ogawa, H., Muranaka, S. et al., “New CombustionConcept for Ultra-Clean and High-Efficiency Small DI Diesel Engines,”SAE Technical Paper 1999-01-3681, 1999, doi:10.4271/1999-01-3681.

8. Kimura, S., Aoki, O., Kitahara, Y., and Aiyoshizawa, E., “Ultra-CleanCombustion Technology Combining a Low-Temperature and PremixedCombustion Concept for Meeting Future Emission Standards,” SAETechnical Paper 2001-01-0200, 2001, doi:10.4271/2001-01-0200.

9. de Ojeda, W., Zoldak, P., Espinosa, R., and Kumar, R., “Developmentof a Fuel Injection Strategy for Diesel LTC,” SAE Technical Paper2008-01-0057, 2008, doi:10.4271/2008-01-0057.

10. Han M, Assanis D N, Bohac S V, “Sources of hydrocarbon emissionsfrom low-temperature premixed compression ignition combustion froma common rail direct injection diesel engine” Combust. Sci. Technol.181(3):496-517 (2009)

11. Lachaux T and Musculus M P B, “In-cylinder unburned hydrocarbonvisualization during low-temperature compression-ignition enginecombustion using formaldehyde PLIF” Proc. Combust. Inst. 31(2):2921-2929 (2007)

12. Musculus, M., Lachaux, T., Pickett, L., and Idicheria, C., “End-of-Injection Over-Mixing and Unburned Hydrocarbon Emissions in Low-Temperature-Combustion Diesel Engines,” SAE Technical Paper2007-01-0907, 2007, doi:10.4271/2007-01-0907.

13. Kim, D., Ekoto, I., Colban, W., and Miles, P., “In-cylinder CO and UHCImaging in a Light-Duty Diesel Engine during PPCI Low-TemperatureCombustion,” SAE Int. J. Fuels Lubr. 1(1):933-956, 2009, doi:10.4271/2008-01-1602.

14. Koci, C., Ra, Y., Krieger, R., Andrie, M. et al., “Detailed UnburnedHydrocarbon Investigations in a Highly-Dilute Diesel Low TemperatureCombustion Regime,” SAE Int. J. Engines 2(1):858-879, 2009, doi:10.4271/2009-01-0928.

15. Hildingsson, L., Persson, H., Johansson, B., Collin, R. et al., “OpticalDiagnostics of HCCI and Low-Temperature Diesel Using Simultaneous2-D PLIF of OH and Formaldehyde,” SAE Technical Paper2004-01-2949, 2004, doi:10.4271/2004-01-2949.

16. Chartier, C., Andersson, O., Johansson, B., Musculus, M. et al., “Effectsof Post-Injection Strategies on Near-Injector Over-Lean Mixtures andUnburned Hydrocarbon Emission in a Heavy-Duty Optical DieselEngine,” SAE Int. J. Engines 4(1):1978-1992, 2011, doi:10.4271/2011-01-1383.

17. Bobba, M., Genzale, C., and Musculus, M., “Effect of Ignition Delay onIn-Cylinder Soot Characteristics of a Heavy Duty Diesel EngineOperating at Low Temperature Conditions,” SAE Int. J. Engines 2(1):911-924, 2009, doi:10.4271/2009-01-0946.

18. Espey, C. and Dec, J., “Diesel Engine Combustion Studies in a NewlyDesigned Optical-Access Engine Using High-Speed Visualization and2-D Laser Imaging,” SAE Technical Paper 930971, 1993, doi:10.4271/930971.

19. Dec, J., “A Conceptual Model of DI Diesel Combustion Based on Laser-Sheet Imaging,” SAE Technical Paper 970873, 1997, doi:10.4271/970873.

20. Murphy M J, Taylor J D, McCormick R L, “Compendium ofExperimental Cetane Number Data” N.R.E. Laboratory NREL/SR-540-36805. (2004)

21. Heywood J B, Internal combustion engine fundamentals, McGraw-Hill,New York, NY (1988)

22. Genzale, C., Reitz, R., and Musculus, M., “Effects of Piston BowlGeometry on Mixture Development and Late-Injection Low-Temperature Combustion in a Heavy-Duty Diesel Engine,” SAE Int. J.Engines 1(1):913-937, 2009, doi:10.4271/2008-01-1330.

23. Musculus, M. and Kattke, K., “Entrainment Waves in Diesel Jets,” SAEInt. J. Engines 2(1):1170-1193, 2009, doi:10.4271/2009-01-1355.

24. Hwang, J., Kal, H., Kim, M., Park, J. et al., “Effect of Fuel InjectionRate on Pollutant Emissions in DI Diesel Engine,” SAE Technical Paper1999-01-0195, 1999, doi:10.4271/1999-01-0195.

25. Rottmann, M., Menne, C., Pischinger, S., Luckhchoura, V. et al.,“Injection Rate Shaping Investigations on a Small - Bore DI DieselEngine,” SAE Technical Paper 2009-01-0850, 2009, doi:10.4271/2009-01-0850.

26. Ladommatos N, Abdelhalim S, Zhao H, “The effects of exhaust gasrecirculation on diesel combustion and emissions” Int. J. Eng. Res. 1(1):107-126 (2000)

27. Dronniou, N., Lejeune, M., Balloul, I., and Higelin, P., “Combination ofHigh EGR Rates and Multiple Injection Strategies to Reduce PollutantEmissions,” SAE Technical Paper 2005-01-3726, 2005, doi:10.4271/2005-01-3726.

CONTACT INFORMATIONFor more information, please contact:

Jacqueline O'ConnorSandia National [email protected]

Mark MusculusSandia National [email protected]

ACKNOWLEDGMENTSThe optical engine experiments were performed at the

Combustion Research Facility, Sandia National Laboratories,Livermore, CA. Support for this research was provided by theU.S. department of Energy, Office of Vehicle Technologies.Sandia is a multi-program laboratory operated by SandiaCorporation, a Lockheed Martin Company for the UnitedState Department of Energy's National Nuclear SecurityAdministration under contract DE-AC04-94AL85000. Theinjector was provided by Delphi Diesel Systems, and theauthors would like to thank Delphi and Philip Dingle for theirsupport of this system. The authors gratefully acknowledgethe contributions of Keith Penney and Dave Cicone for theirassistance in maintaining the lasers and research engine usedin this study, and Dipankar Sahoo for assistance with theUHC analyzer.

DEFINITIONS/ABBREVIATIONSAEI - After end of injectionAHRR - Apparent heat release rateATDC - After top dead centerBDC - Bottom dead centerBTDC - Before top dead centerCAD - Crank angle degree position (360 CAD is TDC ofcompression Stroke)°CA - Degrees crank angle (duration)




http://www.sae.org/technical/papers/2005-01-3836

http://dx.doi.org/10.4271/2005-01-3836


http://dx.doi.org/10.4271/2008-01-1327


http://dx.doi.org/10.4271/2006-01-0026


http://dx.doi.org/10.4271/2006-01-0920


http://dx.doi.org/10.4271/1999-01-3681


http://dx.doi.org/10.4271/2001-01-0200


http://dx.doi.org/10.4271/2008-01-0057


http://dx.doi.org/10.4271/2007-01-0907

http://dx.doi.org/10.4271/2008-01-1602

http://dx.doi.org/10.4271/2009-01-0928


http://dx.doi.org/10.4271/2004-01-2949

http://dx.doi.org/10.4271/2011-01-1383

http://dx.doi.org/10.4271/2009-01-0946

http://www.sae.org/technical/papers/930971

http://dx.doi.org/10.4271/930971

http://www.sae.org/technical/papers/970873

http://dx.doi.org/10.4271/970873

http://dx.doi.org/10.4271/2008-01-1330

http://dx.doi.org/10.4271/2009-01-1355


http://dx.doi.org/10.4271/1999-01-0195


http://dx.doi.org/10.4271/2009-01-0850


http://dx.doi.org/10.4271/2005-01-3726

mailto:[email protected]

mailto:[email protected]

DOI1C - Commanded duration of main injection (inmicroseconds)DOI2C - Commanded duration of post injection (inmicroseconds)EGR - Exhaust gas recirculationgIMEP - Gross indicated mean effective pressureH2CO-PLIF - Formaldehyde planar laser inducedfluorescenceLII - Laser induced incandescenceLTC - Low temperature combustionMK - Mixing kineticsOH-PLIF - OH planar laser induced fluorescenceOPO - Optical parametric oscillatorPLIF - Planar laser induced fluorescencePAH - Poly-cyclic aromatic hydrocarbonPCCI - Premixed charge compression ignitionPCI - Premixed compression ignitionPPCI - Partially-premixed compression ignitionSOI1C - Commanded start of main injection (in crank angledegrees)SOI2C - Commanded start of post injection (in crank angledegrees)SOMC - Start of main combustion (in crank angle degrees)SOPC - Start of post combustion (in crank angle degrees)TDC - Top dead centerΔtID - Ignition delay time

UHC - Unburned hydrocarbon




Figure A1. Evolution of combustion as seen through single-shot simultaneous OH-PLIF (green) and H2CO-PLIF (red)imaging for a single-injection case at SOI2C=352 CAD, DOI1C=1000 microseconds at 265 kPa gIMEP.


APPENDIX



Figure A2. Evolution of combustion as seen through simultaneous single-shot OH-PLIF (green) and H2CO-PLIF (red)imaging for a main plus post-injection case at SOI1C=352 CAD, DOI1C=1000 microseconds, SOI2C=363.5 CAD, and

DOI2C=400 microseconds at 371 kPa gIMEP.




Figure A3. Comparison of single-shot OH-PLIF (green) and H2CO-PLIF (red) evolution at three timings for three main pluspost-injection cases at 4 CAD (left), 11 CAD (middle), and 18 CAD (right) after end of the main injection at 371, 371, and 353

kPa gIMEP.




hydrocarbons using close-coupled post injections at ltc optical ... · optical investigation of the...

Documents