Oxygenated fuels have beneficial effects for leaner lifted-flame combustion (LLFC), a nonsooting mode of mixing-controlled combustion associated with lift-off length equivalence ratios below approximately 2. A single-cylinder heavy-duty optical compression-ignition engine was used to compare neat methyl decanoate (MD) and T50, a 50/50 blend by volume of tripropylene glycol monomethyl ether (TPGME) and #2 ultralow sulfur emissions-certification diesel fuel (CF). High-speed, simultaneous imaging of natural luminosity (NL) and chemiluminescence (CL) were employed to investigate the ignition, combustion, and soot formation/oxidation processes at two injection pressures and three dilution levels. Additional Mie scattering measurements observed fuel-property effects on the liquid length of the injected spray. Results indicate that both MD and T50 effectively eliminated engine-out smoke emissions by decreasing soot formation and increasing soot oxidation during and after the end of fuel injection. MD further reduced soot emissions by 50–90% compared with T50, because TPGME could not completely compensate for the aromatics in the CF. Despite the low engine-out soot emissions, both fuels produced in-cylinder soot because the equivalence ratio at the lift-off length never reached the nonsooting limit. With respect to the other engine-out emissions, T50 had up to 16% higher nitrogen oxides (NOx) emissions compared with MD, but neither fuel showed the traditional soot-NOx trade-off associated with conventional mixing-controlled combustion. In addition, T50 had up to 15% and 26% lower unburned hydrocarbons (HC) and CO emissions, respectively, compared with MD.
Diesel engines have lower brake-specific carbon dioxide (CO2) emissions compared to spark-ignition gasoline engines, but require expensive after-treatment systems to comply with the emission regulations for nitrogen oxides (NOx) and particulate matter (PM) in most developed nations. Advanced combustion strategies such as globally premixed, low-temperature combustion (e.g., homogeneous charge compression ignition) can lower NOx and PM emissions, but they are less suited to high-load operating conditions usually characterized by mixing-controlled combustion due to control difficulty, high combustion noise, poor robustness to factors that influence the ignition delay, and difficulty in managing the mode transition due to the limited load range. One alternative is leaner lifted-flame combustion (LLFC). LLFC is a form of mixing-controlled diesel combustion that can eliminate soot if the equivalence ratio at the flame lift-off length is less than two [1,2]. The flame lift-off length is defined as the axial distance between the fuel injector orifice exit and the position where the standing premixed auto-ignition zone stabilizes during mixing-controlled combustion . Thus far, LLFC using conventional diesel fuel has been successful only at low-load operating conditions .
A strategy to increase LLFC applicability to higher engine loads is to use fuels that are known to curtail PM emissions, such as oxygenated fuels, as demonstrated in Ref. . An oxygenated fuel is defined as one that has oxygen chemically bound within the fuel molecule, such as esters, alcohols, ethers, and ketones. All other conditions being equal, the presence of oxygen in the fuel molecule would bring the fuel–air mixture closer to stoichiometric compared to a mixture of nonoxygenated fuel and air , lowering the PM emissions.
Two oxygenated compounds of interest are methyl decanoate (MD, C11H22O2) and tripropylene glycol monomethyl ether (TPGME, C10H22O4). MD has been used in numerical combustion studies as a suitable single-component surrogate for biodiesel [6–10], while TPGME was found to be the most promising oxygenate compound from 71 potential oxygenate candidates based on total PM emissions and other considerations, like other gaseous emissions (i.e., NOx, HC, and CO emissions), flash point, and toxicity [11,12]. It is believed that TPGME is so effective at lowering soot because the distribution of O atoms in the TPGME molecule ensures that each O atom leads to a CO molecule, thus preventing a greater fraction of the C atoms from contributing to the pool of available soot precursor species [13,14].
Dumitrescu et al.  investigated fuel-property effects on engine-out emissions using nonoxygenated fuels and found that the ignition delay had the greatest effect on PM emissions under conventional diesel conditions. PM emissions decreased with increased ignition delay under mixing-controlled combustion at similar fuel injection pressure and charge dilution. However, Ref.  also shows that a longer ignition delay could increase noise as well as unburned hydrocarbon emissions if it results in over-mixing, i.e., areas of fuel–air mixtures below the flammability limit. Further, a longer lift-off length can raise the probability of ignition products being affected by the piston or chamber walls. As a result, LLFC using oxygenated fuels could enable better combustion control and stability if the oxygenated compound has similar ignition delay and lift-off length as the diesel fuel.
Manin et al. , which compared various oxygenated fuels, alkanes, and their blends in a constant-volume combustion vessel, found that while oxygenated fuels had longer lift-off lengths relative to alkanes, they produce a longer ignition delay only if their cetane number (CN) is lower. In addition, the aromatic content of the base fuel in an oxygenate blend correlated with the equivalence ratio required to reach nonsooting conditions. These findings were confirmed in the studies of Cheng et al.  and Dumitrescu et al. . Using the same experimental setup and similar operating conditions (i.e., a heavy-duty diesel engine with optical access under low-load conditions), [17,18] found that MD and T50, a 50/50 blend by volume of TPGME and #2 ultralow sulfur diesel certification fuel (CF), had similar ignition delay and lift-off length with the baseline CF.
The objective of the current study is to use the data from Refs. [17,18] to investigate which of these two oxygenated compounds, MD and T50, has the highest potential to enable sustained LLFC under conditions representative of a modern heavy-duty diesel engine. No other experiments were performed for this study. The diagnostics in Refs. [17,18] included conventional heat-release analysis, engine-out emissions, and high-speed in-cylinder imaging of both natural luminosity and OH chemiluminescence. The experiments in these two studies were conducted under similar operating conditions at two fuel-injection pressures and three levels of charge-gas dilution. In addition, this study employed Mie scattering measurements to observe fuel-property effects on the liquid length of the injected spray. The comparison between MD and T50 is also of interest because they have relatively similar fuel properties, which should result in a similar combustion behavior. The goal is to identify the extent to which LLFC can be obtained at the highest engine efficiency and with the lowest NOx emissions.
Experimental Setup and Procedure
Engine and Fuel System.
Experiments were performed in a single-cylinder version of a heavy-duty engine modified to provide optical access to the combustion chamber. A schematic and the primary specifications of the engine are provided in Fig. 1 and Table 1, respectively. The main differences between the optical engine and a comparable production engine relate to the piston bowl, which has a flat bottom, vertical walls, and a larger land above the top compression ring. The larger top-ring land allows optical access through a notch in the piston bowl-rim and prevents the rings from riding over the cylinder-wall windows. The above differences lead to a lower geometric compression ratio (CR) of 12.3:1 for the optical engine versus ∼16:1 for a comparable production engine. To compensate for the lower CR, the intake temperature and pressure are elevated to closely match the conditions in the production engine over the crank-angle range during which combustion occurs .
For these experiments, the optical engine was skip-fired with one injection event every five engine cycles. This approach reduces the rate of soot accumulation on the combustion-chamber windows and lowers the risk of window failure due to thermal and mechanical stresses generated when switching between motored and fired operation . Skip-firing affects the exhaust-gas composition; thus, real exhaust-gas recirculation (EGR) was not used. Instead, EGR was simulated by adding N2 and CO2 to the dry intake air. A heater was used to maintain a constant coolant temperature as the heat rejection from the combustion chamber to the coolant is lower during skip-fired operation compared to a continuously fired mode.
Fuel delivery was accomplished using a prototype high-pressure common-rail fuel-supply system comprised of two diaphragm pumps (Newport Scientific, Jessup, MD, Model 46-16060) and a near-production Caterpillar electronically actuated solenoid-operated fuel injector fitted with a two-hole tip. The fuel-wetted internals of the pump are fabricated from Teflon, stainless steel, or Kalrez, making the system suitable for handling a wide range of conventional and unconventional fuels having high corrosivities, low viscosities, or other properties that can cause conventional piston-type pumps to fail.
Fuel-injector specifications are provided in Table 2. X-ray tomography was used to confirm the orifice diameter, length-to-diameter ratio, and included spray angle listed in Table 2. Figure 2 presents MD and T50 injection-rate profiles for the two-hole tip at injection pressures of 80 MPa and 180 MPa, as measured using a momentum-based rate-of-injection meter developed at Sandia .
A flush-mounted, water-cooled pressure transducer (AVL Model QC32C, Graz, Austria) was used to measure cylinder pressure every 0.5 crank-angle degree (CAD). Gas analyzers (California Analytical Instruments, Series 600, Orange, CA) measured CO, CO2, O2, NOx, and HC emissions. A smoke meter (AVL Model 415S) and a prototype exhaust laser induced incandescence (LII) instrument  measured engine-out soot emissions. The procedures used for cylinder-pressure data acquisition and how the skip-fired exhaust-gas measurements were converted to values similar to those that would be obtained if the engine was operated in a continuous-fired mode are described in Ref. . References  and  provide further details of the experimental setup and procedures.
The test fuels investigated were MD and a 50/50 blend by volume of TPGME and CF, called T50. The molecular structures of the oxygenates are shown in Fig. 3. MD is a single-component oxygenated fuel having the chemical formula C11H22O2. MD has an oxygen content of 17.2% by mass and can be produced from renewable feedstocks. TPGME is an oxygenate having the chemical formula C10H22O4. TPGME has an oxygen content of 31% by mass and is produced commercially as a mixture of up to eight isomers by reacting propylene oxide with methanol . It is used in the manufacture of polyester plastics, and as a coupling agent or solvent for household and industrial cleaners such as oven cleaner. It is also used for rust, paint, and varnish removers, as a solvent for coil and wood coatings, and as a solvent for ballpoint, felt-tipped pens, and inkpads (to prevent drying). Selected properties of CF, TPGME, T50, and MD are presented in Table 3 (refer to Refs. [24,25] for detailed CF properties and compositional characterization). The effect of the lower specific energy (energy per unit mass) of TPGME compared to MD is mitigated by blending it with CF.
In Eq. (2), nO− and nO= represent the single and double-bonded oxygen atoms to C-atoms, respectively.
Engine Operating Conditions.
Table 4 presents the engine operating conditions. The engine was fired (i.e., fuel was injected) once every fifth cycle at 1500 rpm for a minimum of 180 fired cycles per engine run. To ensure good data repeatability, at least 540 engine cycles of pressure-based data were recorded at each operating condition during three separate optical-engine runs that generally occurred on different days. As a result, data points presented in the figures throughout this paper represent averages over the three replicate measurements at the same test condition; error bars denote minimum and maximum average values on those three separate optical-engine runs.
The experiments were performed at two injection pressures (80 MPa and 180 MPa), constant engine speed, intake pressure, and intake temperature. Three dilution levels—21, 18, and 16 mol % O2—were studied to simulate the effects of using zero to moderate EGR rates.
Table 5 shows the mole fractions of O2, N2, CO2, and argon (Ar) in each intake mixture. Although the intake-O2 mole fraction (XO2) varied, the gas-mixture compositions were selected so that the temperature, density, pressure, and specific heat at −10 CAD after top dead center (ATDC) remained constant at roughly 850 K, 26 kg/m3, 63 bar, and 35 J/mol K, respectively. This approach was taken to isolate the effects of XO2 variation on the results.
Injection timing and injection duration were chosen to obtain predominantly mixing-controlled combustion and valid lift-off length measurements. As indicated in Table 4, the start of combustion (SOC) also was held constant for all fuels and operating conditions. SOC was defined in this study as the crank angle where the high-temperature heat release (HTHR) first becomes positive after the start of injection (SOI). The similar injection duration for T50 and MD versus baseline CF operation resulted in a 25% load difference due to the lower T50 and MD heating values.
Two high-speed cameras were used to simultaneously record natural luminosity and OH* chemiluminescence during the first ten fired cycles of each engine run. High-speed visualization of Mie-scattered light from the liquid portion of the diesel spray was used in this study for the liquid-length measurements.
Natural luminosity (NL) refers to the broadband light emitted by the combustion process during a fired cycle and used to indicate the locations of high-temperature soot particles during mixing-controlled combustion . In this study, NL signal in the range of 380–1000 nm was collected every 0.5 CAD using a high-speed complementary metal oxide semiconductor (CMOS) camera (Vision Research, Wayne, NJ, Model Phantom v7.3) fitted with a 35-mm Nikon lens. The combustion chamber is visualized from below through the fused-silica piston window using a 45 deg mirror, as shown in Fig. 1. The field of view is 104 mm × 96 mm with a spatial resolution of 0.5 mm/pixel.
In Eq. (3), n and m are the number of pixels in the x- and y-directions, respectively; fs is the set lens f-stop; and texp is the set image exposure duration. SINLmax is the signal that would be measured if all pixels within the piston bowl were saturated when the lens aperture and exposure time were set to their minimum values (i.e., fs = 16 and texp = 1.0 μs). SINL values provide a time-resolved, quantitative measure of high-temperature soot that can be compared between different engine runs. Further details of the NL data-acquisition procedure can be found in Ref. .
Chemiluminescence (CL) signal from electronically excited hydroxyl radicals (OH*) at 308 ± 10 nm was used to determine the flame lift-off length (H). The CL signal was collected every 1.0 CAD using an intensified CMOS camera (Vision Research, Model Phantom v311, Gen II intensifier) and a Nikon 105-mm ultraviolet (UV) lens. A sequence of 114 mm × 114 mm images with a spatial resolution of 0.23 mm/pixel was captured during each fired cycle. CL images were analyzed following the procedure described in Ref. .
In Eq. (4), H(j, c, CAD) represents the lift-off length measurement from a single jet and cycle, at a specific crank-angle degree; njets is the number of injector orifices, and ncycle is the number of engine cycles imaged. For notational simplicity, ⟨H(CAD)⟩ will be referred to as H in the rest of this paper.
The equivalence ratio at the flame lift-off length, ϕ(H), has been shown to correlate with in-cylinder soot levels, with significant NL appearing only for ϕ(H) values greater than 2 . The methodology used to calculate ϕ(H) in this study is based on Naber and Siebers' relationship . Reference  details the methodology and discusses the validity of this approach in estimating ϕ(H) for the experimental setup used in this study.
Figure 4 presents the optical setup used for the Mie scattering measurements. A Nd:YAG laser (Lee Laser, Orlando, FL, Model LDP-100MQG) provided the ∼0.5-μs pulse width, 532-nm laser light for the spray illumination. Spray-visualization images were acquired at 50 kHz (i.e., ∼0.2-CAD resolution at 1500 rpm), using a collection system consisting of a high-speed CMOS camera (Vision Research, Model Phantom v7.3) equipped with a 20-mm-diameter variable-focal-length borescope (Abakus Ltd., Stamford, UK) with a nominal 115 deg field of view.
Figure 4 shows that the 5-mm-thick laser sheet entered the combustion chamber from below through the piston window and was aligned in the same plane with the two fuel jets. The 50 kHz data-acquisition rate limited the area of the CMOS chip that could be used; thus, the camera field of view contained only one of the fuel jets. As a result, the laser sheet spanned the camera field of view from the injector tip to the piston bowl to ensure that the maximum liquid penetration distance was measured (see Fig. 4(b)). The light scattered from liquid-fuel droplets was collected through a window in the cylinder wall, which faced a notch cut from the bowl-rim to allow continuous optical access into the piston bowl through top dead center (TDC). The curved cylinder-wall window and wide-angle borescope lens distorted the acquired images, necessitating geometric correction. The spatial resolution of the corrected images was ∼0.3 mm/pixel. Acquired images were also corrected for nonuniformities in the laser sheet intensity profile and in the collection efficiency of the imaging system. The correction procedures are described in detail in a previous study .
The maximum measured signal in the raw images was kept at ∼50–75% of the CMOS sensor full scale (i.e., detector saturation was avoided). The liquid length was determined along the spray axis after application of the corrections described above, using a fixed threshold intensity of 2% of maximum intensity (Ref.  shows that liquid-length measurements are nearly independent of the threshold value when the threshold is lowered to ∼1–4% of full scale).
Exhaust-LII measurements and filter smoke numbers (FSN; measured with the AVL smoke meter and corrected for skip-firing) as a function of XO2 for the T50 and MD at the 180- and 80-MPa injection pressures are presented in Fig. 5. Exhaust-LII and FSN measurements were in excellent agreement for all operating conditions. FSN measurements for T50 were up to 10x higher compared to MD, even if Dumitrescu et al.  showed that T50 almost eliminated smoke emissions when compared with neat CF. Further, Fig. 5 indicates that FSN measurements for MD were near the detection limit of the AVL smoke meter, with values below 0.005 FSN independent of operating condition. Figure 5 also shows that the effect of charge dilution on smoke emissions depended on the equipment used to measure the emissions at such a low range. Exhaust-LII measurement shows that the relative difference in smoke emissions between T50 and MD was not affected by charge dilution (however, it was affected by the injection pressure: 4× and 6× higher exhaust-LII signal for T50 compared with MD, at 80- and 180-MPa injection pressure, respectively). In contrast, FSN measurements suggest that a higher charge dilution reduces the smoke-emission differences between T50 and MD, while the injection pressure influence is reversed, with the higher injection pressure decreasing the difference in smoke emissions between fuels. Figure 5, which generally shows a larger variation in FSN values at a specific operating condition compared to exhaust-LII measurements, suggests that this is probably due to different exhaust-LII and AVL smoke meter sensitivities near their lower detection limit (i.e., the near-zero detection range).
At a specific Pinj and independent of the oxygen concentration, the apparent heat release rate (AHRR) in Fig. 6 shows that MD and T50 had similar premixed-combustion fraction, engine crank angle corresponding to 50% mass fraction burned (CA50), and combustion duration even if T50 cases had 1–3% higher load than MD cases. Figure 6 suggests that the reduction in engine-out smoke emissions for MD operation compared with T50 was not the result of a change in combustion mode or phasing. Considering that MD and T50 have similar oxygen ratio, effective carbon-to-hydrogen ratio, stoichiometric air–fuel ratio, and lower heating values, the lower smoke level for MD cases can be due either to different air–fuel mixing history or the differences in detailed fuel composition.
Fuel volatility and latent heat of vaporization affect the vaporization and penetration of the diesel spray, and hence the air–fuel mixing at constant injection pressure and charge-gas density and temperature . Low-volatility sprays need more time to vaporize, which would result in a longer liquid length.
Figure 7 shows the measured noncombusting liquid-length results for T50 and MD, at 80 and 180 MPa injection pressure, and SOIactual injection timings from −50 CAD ATDC to 10 CAD ATDC in increments of 10 CAD. The effects of in-cylinder flows, such as swirl or bulk flow due to piston motion, have been neglected and all SOI and end-of-injection (EOI) transients have been omitted. The bowl radius limited the maximum measured liquid length under some conditions. References , , and  discuss in detail the engine and operating condition effects on measured liquid length.
Liquid length depended on in-cylinder conditions (pressure, temperature) and was relatively independent of injection pressure. Liquid lengths for overlapping injections do not agree perfectly in the overlapping crank-angle range, and there is a similar degree of asymmetry with liquid length being longer in the expansion stroke (i.e., after TDC) compared with the equivalent crank-angle position in the compression stroke (i.e., before TDC) . However, there are no obvious liquid-length differences between T50 and MD under noncombusting conditions. Consequently, it is expected that liquid-length under combusting conditions would also be similar, considering the similar premixed-combustion fraction and AHRR. This would suggest similar air–fuel mixing at a particular axial distance from the injector nozzle, which can be described by the uniform-profile model of Naber and Siebers  or the variable-profile model of Musculus and Kattke .
As a result, the differences in smoke emissions observed in Fig. 5 appear to be mostly the effect of the fuel molecular structure. This underlines the importance of detailed fuel chemistry in lowering particulate emissions. While TPGME chemistry would result in a more efficient usage of each oxygen atom in the molecule compared with MD (see the “Introduction” section), TPGME cannot totally compensate for the sooting propensity of other species contained in the T50 blend, such as the ∼30% aromatics-by-mass content of CF.
Although near-zero engine-out smoke measurements suggest that soot oxidation was almost complete for these particular conditions, this may not be the case in a real diesel engine operating at similar intake-manifold conditions but higher loads. A larger number of injector nozzle holes could result in jet-to-jet interactions that could lower the overall soot oxidation (especially in the late-combustion stages) and increase smoke emissions (see Fig. 12 of Ref.  which reports an order of magnitude smaller smoke levels for the two-hole experiments compared to those in six- and ten-hole experiments). However, it is important to know in detail how the fuel composition affects in-cylinder soot formation and oxidation without the jet-to-jet interaction, in order to decouple the fuel-composition effects from the fluid-mechanics effects on in-cylinder mixing and combustion processes. Insights into the in-cylinder mechanisms leading to soot formation and oxidation can be obtained by using combustion visualization techniques, such as imaging of natural luminosity (NL) and chemiluminescence (CL).
Figure 8 shows sample instantaneous and simultaneous NL and CL images at different crank angles during MD combustion (after the SOC, midway through fuel injection, immediately before the end of injection (EOI), and a few degrees after EOI). The reader is reminded that these two combustion visualization techniques provide a qualitative idea of the hot-soot and OH* concentrations present during the combustion process. In Fig. 8, the sensitivity of the detection system was separately optimized for each image sequence to use the full dynamic range of its corresponding camera.
To assist in visualizing NL and CL signals that otherwise would be too difficult to distinguish, NL and CL signals in each image were scaled to their respective maximum values, as shown in Fig. 9. Further, these NL and CL images were superimposed on one another to simultaneously show the temporal and spatial evolution of both the flame zone and the hot-soot regions during the combustion event.
The hot-soot- and chemiluminescence-containing areas are colored in shades of magenta and green, respectively, that are proportional to signal intensity. The areas where NL and CL signals overlapped are colored in shades of gray (i.e., a darker shade of gray indicates an area with simultaneously strong NL and CL signals). To make it even clearer, contours of the areas containing soot and chemiluminescence are also marked, as following: the black and red lines outline the NL signal above 5% and 30% of the maximum NL signal in that specific image, respectively, while the cyan lines outline the CL signal above 50% of the maximum CL signal in that specific image (the percentages were chosen based on a compromise between the amount of the details and the figure clarity). The NL contours in Fig. 9 have different signal levels; therefore, they must be analyzed with respect to the actual SINL measurements at a particular crank angle. Finally, the yellow lines represent measured H values.
The chemiluminescence and soot evolution in Fig. 9 follow Dec's conceptual model of conventional diesel combustion . The model suggests that as the fuel-rich premixed autoignition zone stabilizes at H, near-stoichiometric combustion takes place in a diffusion flame surrounding the fuel jet, and soot is generally formed on the rich side of the diffusion flame (i.e., toward the spray centerline).
The first column in Fig. 9 presents images taken at TDC, which is immediately after the end of the premixed-combustion phase seen in Fig. 6. The increase in jet momentum at the higher injection pressure moved the soot-containing areas closer to the edge of the piston bowl and increased the distance between H and the first evidence of soot inside the jet plume. The hot-soot-containing areas at a particular Pinj are similar in size (there is a small decrease in size for the MD cases at 180-MPa injection pressure), but there are differences between the NL levels and their evolution during the combustion process.
The second column in Fig. 9 shows images midway through the fuel injection, at 11 CAD ATDC. The jet plume is more developed and, after reaching the bowl edge, expands circumferentially parallel to the bowl edge as the incoming fuel upstream of the reaction zone pushes the combusting mixture forward. More importantly, the OH*-containing areas completely overlap the hot-soot-containing areas, evidence that at this stage in the combustion process all the visible soot is contained within the envelope of the reaction zone.
The third-column images in Fig. 9 were acquired at 22 CAD ATDC, which is immediately before EOI. The soot-containing area expanded further parallel to the bowl edge, and the SINL signal was still robust at 80-MPa injection pressure. At the same time, the stronger shade of magenta near the bowl edge indicates a lower level of OH* chemiluminescence, probably a combination of a larger soot volume in that area, reduced mixture reactivity due to the bowl edge proximity, and trapping of the CL signal by soot that started convecting inward from the bowl edge.
The images in the fourth column of Fig. 9 were taken at 27 CAD ATDC, which is 4–5 CAD after EOI. There is evidence of relatively robust OH-chemiluminescence in all images in the areas previously occupied by the injected fuel, which might be expected based on the entrainment wave effect . At the same time, Fig. 9 suggests that smoke emissions are mainly coming from the soot-containing areas near the bowl edge that show little or no evidence of OH chemiluminescence, possible evidence of reduced soot oxidation. The injection pressure had an important effect in soot location and intensity at this CA. If both fuels presented evidence of soot at 80-MPa injection pressure, at the 180-MPa injection pressure the soot-containing areas are barely visible in the MD images, and there are just a few isolated soot pockets in the T50 images.
The soot-containing areas are more fragmented at higher injection pressure, suggesting that soot was contained in isolated “pockets” compared to the lower injection pressure where the soot-containing areas were more continuous and uniform. Furthermore, the relatively strong OH* chemiluminescence and the low NL levels in Fig. 9 would suggest curtailed soot production and robust soot oxidation for MD relative to T50, regardless of the injection pressure.
Figures 10 (average SINL levels at each crank angle) and 11 (the SINL ratio between the two fuels at similar operating conditions and crank angle, SINLMD/SINLT50) provide additional insight into the in-cylinder soot evolution illustrated in Fig. 9. The data plotted in Figs. 10 and 11 were used to determine the reduction in SINL for MD compared with T50 shown in Table 6, which is in agreement with the difference in smoke emissions between the fuels seen in Fig. 5.
Figure 10 shows that SINL lowered with increased injection pressure, regardless of charge dilution. In addition, Figs. 10 and 11(a) show that compared to T50, MD reduced average SINL at all operating conditions, regardless of their similar oxygen ratio, effective carbon-to-hydrogen ratio, air–fuel ratio, and lower heating values. Further, Fig. 10 shows that SINL was on a decreasing slope after 5–10 CAD ATDC, even if the bulk temperature (not shown here) was higher than the corresponding bulk temperature at TDC, which supports the hypothesis that MD curtailed soot production and increased soot oxidation, regardless of the injection pressure. A higher temperature would increase SINL signal for similar soot volume fraction and soot optical properties .
In addition, Fig. 11(a) also shows that the injection pressure affected how the SINLMD/SINLT50 ratio changes with charge dilution. The SINLMD/SINLT50 ratio increased with charge dilution at 80-MPa injection pressure, but at 180-MPa injection pressure the SINLMD/SINLT50 ratio was the largest for the 18 vol % O2. It will be shown later in the text that this behavior correlated well with the change in ϕ(H).
Figure 11(a) also indicates that the SINLMD/SINLT50 ratio decreased at similar rates after reaching its highest value. This would suggest a similar evolution of SINL with CAD after reaching the maximum value, probably due to similar rates of soot oxidation and that the design of experiments resulted in similar in-cylinder bulk temperatures (Ref.  provides a detailed explanation of the important effect of in-cylinder and soot temperature on SINL signal). Further, the increase in injection pressure had an effect on the SINL signal similar in magnitude and direction as changing the fuel from T50 to MD, as shown in Fig. 11(b) by the SINLHP/SINLLP ratio. A higher charge dilution generally increased the injection pressure effect on SINL (e.g., at TDC, MD SINLHP/SINLLP = 0.16 at 21 vol % O2, compared to SINLHP/SINLLP = 0.07 at 18 vol. % O2). At similar charge dilution, MD had a higher SINL decrease with injection pressure compared with T50 (e.g., at TDC and 16 vol % O2, MD SINLHP/SINLLP = 0.04, compared to SINLHP/SINLLP = 0.10 for T50). This would suggest that fuel composition dictates the effectiveness of a higher injection pressure in reducing SINL (and corresponding smoke emissions), regardless of the similar oxygen ratio, effective carbon-to-hydrogen ratio, air–fuel ratio, and lower heating values. On the other hand, the SINLHP/SINLLP ratio decreased with charge dilution, confirming that higher fuel-injection pressure is a solution to limit smoke emissions, especially for operating conditions that create large amounts of soot, such as those at higher charge dilution.
Equivalence Ratio at the Flame Lift-Off Length, ϕ(H).
Lift-off length (H) measurements provide additional insight into the effect of fuel composition on soot formation and oxidation. Specifically, ϕ(H) dictates the temperature and composition of the fuel-rich partially reacted mixture downstream of H that is responsible for the creation of soot precursors and subsequently for the rate of soot formation. T50 is expected to be more sensitive to equivalence ratio in terms of soot production than MD due to its aromatic content, which would favor the creation of larger polycyclic aromatic hydrocarbons compared to MD .
Figure 12 shows that while decreased intake-O2 concentration increased H, ϕ(H) decreased when the oxygen content was lowered from 21 mol % to 18 mol %, then increased or remained relatively unchanged when the mixture was diluted further to 16 mol % O2, independent of the injection pressure. This was probably the result of the nonlinear combined effects of mixing time (i.e., ignition delay), mixing efficiency (injection pressure), and oxygen availability. However, the injection pressure effect on ϕ(H) is obvious. Lower injection pressure raised ϕ(H) at specific dilution level. A higher ϕ(H) has been shown to increase the soot volume fraction and soot particle size  in fundamental experiments, which would be expected to boost the SINL signal. This could explain the shorter distance between H and the first evidence of soot inside the jet plume observed in Fig. 9 at the lower injection pressure.
T50 had similar or slightly lower ϕ(H) than MD at many operating conditions. The exception is the 21-mol % O2 case, independent of the injection pressure. For this case, ϕ(H)T50 is greater than ϕ(H)MD after 10–12 CAD ATDC. Further, as shown in Fig. 12, there are some differences in how ϕ(H) changed with CAD between T50 and MD, with ϕ(H)MD decreasing faster toward EOI than ϕ(H)T50. The two fuels had similar SOI timing and ignition delays. The discussion around Fig. 7 and similar heat release rates suggest similar liquid-length under combusting conditions (Lc) and air–fuel mixing. As a result, the different ϕ(H) temporal behavior can be the result of the distinct fuel chemistry differently affecting the local temperature near the lift-off location. Besides the aromatic content of T50, Table 3 shows that T50 had a slightly higher (C/H)eff ratio than MD. This could result in a higher adiabatic flame temperature for T50 compared to MD , which could decrease H and increase ϕ(H). This observation is confirmed in the study of Dumitrescu et al. , which found that combustion natural luminosity and smoke emissions correlated with fuel aromatic content at similar Lc and ϕ(H) values.
SINL captured the different temporal evolution of ϕ(H) toward EOI. Figure 11(a) shows that the SINLMD/SINLT50 ratio decreased after 10–17 CAD ATDC. The minimum SINLMD/SINLT50 ratio observed for the 21-mol % O2 case is confirmed by the smoke measurements in Fig. 5, which also showed the largest differences in smoke emissions between T50 and MD for the no-dilution case.
NOx, HC, and CO Emissions
Indicated-Specific NOx (ISNOx) Emissions.
Higher ISNOx emissions for T50 at all operating conditions, shown in Fig. 13, support the hypothesis of slightly higher flame temperatures for T50 cases compared with MD. Figure 13 shows that the decrease of in-cylinder oxygen content from 21 to 16 mol % O2 drastically lowered ISNOx emissions and minimized the absolute differences between fuels. This is likely due to a decrease in Tad with decreasing XO2, as shown in Fig. 14 of Ref. . However, the relative difference in NOx emissions between T50 and MD was almost independent of dilution. As NOx emissions in diesel combustion are mostly the result of the thermal-NOx mechanism, this suggests that the dilution level had little effect on the relative difference in flame temperature and global in-cylinder temperature between fuels. On the other hand, injection pressure affected the relative difference in ISNOx emissions, with a 5–6% and 13–16% increase at 80 and 180 MPa injection pressure, respectively, for T50 compared with MD. In addition, each fuel had a different effect on ISNOx emissions change with the injection pressure. ISNOx emissions for T50 increased 0–8% compared with a 0–7% decrease for MD, when the injection pressure changed from 80 to 180 MPa. The lowest increase in T50 occurred at the highest dilution level. MD showed similar ISNOx emissions at 21 mol % O2 and 18 mol % O2, and a 7% reduction in ISNOx emissions at the highest dilution level. This supports the importance of detailed fuel composition on the combined effects of dilution and injection pressure on NOx emissions.
Indicated-Specific HC (ISHC) and CO (ISCO) Emissions.
Figure 14 indicates that ISHC emissions were higher for MD compared to T50 by 6–15% and 9–11%, at 80 and 180 MPa injection pressure, respectively. ISCO emissions had a similar trend, being higher for MD compared to T50 by 13–26% relatively independent of injection pressure. The higher ISHC and ISCO emissions for MD cases independent of injection pressure and charge dilution were likely the result of the 1–3% lower loads and lower flame temperatures (as suggested by the NOx emissions data). The lower combustion temperatures and slightly lower reactivity of MD compared to T50 likely lowered HC and CO oxidation rates, leading to higher ISHC and ISCO emissions, and decreasing its combustion efficiency.
Summary and Conclusions
Oxygenated fuels have been reported to have beneficial effects for leaner lifted-flame combustion (LLFC), a nonsooting mode of mixing-controlled combustion associated with equivalence ratios below approximately 2. A single-cylinder heavy-duty optical compression-ignition engine was used to compare two oxygenated fuels with similar oxygen ratio, effective carbon-to-hydrogen ratio, air–fuel ratio, and lower heating values: methyl decanoate (MD) and a 50/50 blend by volume of tripropylene-glycol monomethyl ether (TPGME) and #2 ultralow sulfur emissions-certification diesel fuel (CF), called T50. The experiments were conducted at two fuel-injection pressures (80 and 180 MPa) and three levels of charge-gas dilution (21, 18, and 16 mol % O2). Conventional heat-release analysis, engine-out emissions measurements, and high-speed optical diagnostics (simultaneous natural luminosity and chemiluminescence imaging) investigated the ignition, combustion, and soot formation/oxidation processes. Additional Mie scattering measurements were employed to observe fuel-property effects on the liquid length of the injected spray. The major observations and conclusions of this study are:
The two fuels exhibited similar combustion behavior such as similar premixed-combustion fraction, CA50, and combustion duration.
MD and T50 reduced considerably the engine-out smoke emissions by decreasing soot formation and/or increasing soot oxidation during and after the end of fuel injection. MD further lowered soot emissions by 50–90% compared with T50. As MD and T50 had similar ϕ(H) and similar air–fuel mixing (suggested by the similar liquid-length measurements), the variation in smoke emissions is likely due to the differences in detailed chemical composition between the fuels. While TPGME chemistry would result in a more-efficient usage of each oxygen atom to lower soot compared with MD, it is believed that this benefit of TPGME cannot totally compensate for the sooting propensity of other species contained in the T50 blend, such as the ∼30% aromatics-by-mass content of CF.
T50 had up to 16% higher NOx emissions compared with MD, likely the result of the higher flame temperatures, but neither fuel showed the traditional soot-NOx trade-off associated with conventional mixing-controlled combustion.
T50 had up to 15% and 26% lower HC and CO emissions, respectively, compared with MD, likely the result of the 1–3% higher loads and higher flame temperatures.
Neither fuel enabled LLFC because the equivalence ratios at the lift-off length, ϕ(H), never reached the nonsooting limit. Nevertheless, while both MD and T50 have the potential to enable LLFC under different experimental conditions that would further decrease ϕ(H) to ∼2 and below; this study showed that it would be easier to reach the nonsooting limit with MD. On the other hand, the lowest NOx and soot emissions for MD cases were penalized by a lower combustion efficiency due to higher HC and CO emissions. Another factor that could lead to the selection of T50 over MD is the fact that T50 contains only 50 vol % oxygenate (i.e., it is a blend with conventional diesel fuel), whereas MD is 100 vol % oxygenate. Hence, the perturbation to diesel-fuel cost, distribution infrastructure, and end-use could be larger with MD, for a given level of soot benefit.
This material is based upon work supported by the U.S. Department of Energy under Award No. DE-EE0005386. The authors gratefully acknowledge: DOE Office of Vehicle Technologies Program Manager Kevin Stork for support of the optical-engine laboratory at Sandia; Bill Cannella of Chevron for providing the chemically and physically well-characterized #2 ultra-low-sulfur diesel emissions-certification fuel used in this work; and Sandia technologists Sam Fairbanks, Chris Carlen, and Gary Hubbard for their assistance with comprehensive/mechanical, electronic, and data-acquisition hardware/software systems, respectively. The research was conducted at the Combustion Research Facility, Sandia National Laboratories, Livermore, CA. Sandia is a multiprogram laboratory operated by Sandia Corporation, a Lockheed Martin Company, for DOE's National Nuclear Security Administration under Contract No. DE-AC04-94AL85000. The experimental data were acquired while author Dumitrescu was employed by Sandia National Laboratories and funded by the U.S. Department of Energy (DOE). Preparation of this manuscript was performed while author Dumitrescu was employed and funded by West Virginia University.
- AHRR =
apparent heat release rate
- ATDC =
after top dead center
- BTDC =
before top dead center
- C =
- CAD =
crank angle degree
- CA50 =
CAD corresponding to 50% mass fraction burned
- CF =
ultra-low-sulfur, #2 diesel certification
- CL =
- CMOS =
complementary metal oxide semiconductor
- CN =
- CO =
- CR =
- CO2 =
- DOI =
duration of injection
- EGR =
- EOI =
actual end of injection
- FSN =
filter smoke number (AVL)
- H =
- HC =
- HTHR =
high temperature heat release
- LLFC =
leaner lifted-flame combustion
- MD =
- NL =
- NOx =
- O =
- OH* =
electronically excited hydroxyl radical
- Pinj =
fuel injection pressure
- PM =
- SINL =
spatially integrated natural luminosity
- SOC =
start of combustion
- SOI =
start of injection
- TDC =
top dead center
- TPGME =
tripropylene glycol monomethyl ether
- T50 =
50/50 blend by volume of TPGME and CF