Energy Fuels 2010, 24, 1538–1551 Published on Web 03/03/2010
: DOI:10.1021/ef9011142
Effects of Fuel Volatility on Early Direct-Injection, Low-Temperature Combustion in an Optical Diesel Engine A. S. (Ed) Cheng,*,† Brian T. Fisher,‡ Glen C. Martin,‡ and Charles J. Mueller‡ †
School of Engineering, San Francisco State University, San Francisco, California 94132, and ‡Combustion Research Facility, Sandia National Laboratories, Livermore, California 94550 Received October 1, 2009. Revised Manuscript Received February 1, 2010
The effect of fuel volatility on early direct-injection, low-temperature combustion (LTC) was investigated using an optically accessible diesel engine. Five blends of conventional no. 2 diesel fuel and a high-volatility (HV) fuel mixture of n-heptane and toluene having approximately the same ignition quality were tested over a range of injection timings at steady-state speed-load operating conditions. Diagnostics included conventional heat-release analysis, the measurement of spatially integrated broadband light emitted during the combustion process (natural luminosity), and high-speed, in-cylinder imaging of both natural luminosity and laser light elastically scattered from liquid-phase fuel in the charge gas. Engine-out emissions of nitrogen oxides (NOx), smoke, unburned hydrocarbons (HC), and carbon monoxide (CO) also were monitored. Results show that, as the injection timing is advanced during LTC operation, liquidfuel films on in-cylinder surfaces are likely to form because of low in-cylinder gas and surface temperatures, particularly for the lower-volatility fuels. Such liquid-fuel films can lead to pool fires and higher smoke, HC, and CO emissions, as well as lower fuel-conversion efficiencies. Increasing HV fuel content was found to be an effective means of reducing or eliminating liquid-fuel films and pool fires, as well as their undesirable effects on efficiency and emissions. Small increases in the HV content produced large changes under conditions where pool-fire activity was significant. For the LTC conditions studied, an HV content of 78% eliminated pool fires and reduced smoke emissions to near-zero levels.
before combustion occurs near top-dead-center (TDC). This early direct-injection strategy is one approach for achieving low-temperature combustion (LTC),1-7 because it employs more premixed, fuel-lean (global equivalence ratio of approximately 0.4) conditions as well as high EGR levels targeted at eliminating the high-temperature NOx-formation regions associated with conventional direct-injected diesel combustion. In a previous study, Martin et al. established a strong connection between liquid-fuel impingement on in-cylinder surfaces and increased fuel consumption and emissions during early direct-injection, LTC operation.7 In cases where liquidfuel impingement produced surface fuel films that persisted beyond the onset of combustion, “pool fires” could occur and generate high levels of soot. Increased NOx emissions were also associated with vigorous pool fires and attributed to the near-stoichiometric regions around the fuel-rich pool-fire region. In addition, the level of pool-fire activity was linked to hydrocarbon (HC) and carbon monoxide (CO) emissions. The results emphasize the significant negative consequences of
1. Introduction Improving fuel-conversion efficiency while achieving current and future exhaust-emission targets is a serious challenge for engine designers. Researchers have been exploring the potential of advanced in-cylinder strategies to meet these goals. Among these strategies are those that employ compression-ignition for high efficiency, in combination with homogeneous (or near-homogeneous) combustion and exhaust gas recirculation (EGR) to minimize soot and nitrogen oxides (NOx) emissions. Strictly speaking, homogeneous-charge compression-ignition (HCCI) combustion relies on complete fuel-air mixing prior to ignition. However, direct fuel injection can also be used to produce a near-homogeneous charge if injection occurs early enough in the compression stroke so that the fuel has time to vaporize and premix in-cylinder *To whom correspondence should be addressed. Phone: 415-4053486. Fax: 415-338-0525. E-mail:
[email protected]. (1) Musculus, M. P. B. Multiple Simultaneous Optical Diagnostic Imaging of Early-Injection Low-Temperature Combustion in a HeavyDuty Diesel Engine. SAE Trans. 2006, 115 (3), 83–110 (SAE Technical Paper 2006-01-0079). (2) Genzale, C. L.; Wickman, D. D.; Reitz, R. D. A Computational Investigation into the Effects of Spray Targeting, Bowl Geometry and Swirl Ratio for Low-Temperature Combustion in a Heavy-Duty Diesel Engine. SAE Trans. 2007, 116 (3), 88–102 (SAE Technical Paper 2007-010119). (3) Huestis, E.; Erickson, P. A.; Musculus., M. P. B. In Cylinder and Exhaust Soot in Low-Temperature Combustion Using a Wide-Range of EGR in a Heavy-Duty Diesel Engine. SAE Trans. 2007, 116 (4), 860–870 (SAE Technical Paper 2007-01-4017). (4) Fang, T., Lin, Y.-C.; , Foong, T. M.; , C.-F. Lee. Spray and Combustion Visualization in an Optical HSDI Diesel Engine Operated in Low-Temperature Combustion Mode with Bio-diesel and Diesel Fuels. SAE Technical Paper 2008-01-1390; 2008. r 2010 American Chemical Society
(5) Genzale, C. L.; Reitz, R. D.; Musculus., M. P. B. Effects of Piston Bowl Geometry on Mixture Development and Late-Injection LowTemperature Combustion in a Heavy-Duty Diesel Engine. SAE Int. J. Engines 2008, 1 (1), 913–937 (SAE Technical Paper 2008-01-1330). (6) Kim, D.; Ekoto, I.; Colban, W. F.; Miles., P. C. In-Cylinder CO and UHC Imaging in a Light-Duty Diesel Engine during PPCI LowTemperature Combustion. SAE Int. J. Fuels Lubr. 2008, 1 (1), 933–956 (SAE Technical Paper 2008-01-1602). (7) Martin, G. C.; Mueller, C. J.; Milam, D. M.; Radovanovic, M. S.; Gehrke., C. R. Early Direct-Injection, Low-Temperature Combustion of Diesel Fuel in an Optical Engine Utilizing a 15-hole, Dual-Row, Narrow-Included-Angle Nozzle. SAE Int. J. Engines 2008, 1 (1), 1057– 1082 (SAE Technical Paper 2008-01-2400).
1538
pubs.acs.org/EF
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al. Table 1. SCORE Specifications engine type cycle valves per cylinder bore stroke intake valve opena intake valve closeda exhaust valve opena exhaust valve closeda connecting rod length piston bowl diameter piston bowl depth squish height swirl ratio displacement geometric compression ratio a
single cylinder four-stroke CIDI 4 125 mm 140 mm 32° BTDC exhaust 153° BTDC comp. 116° ATDC comp. 11° ATDC exhaust 225 mm 90 mm 16.4 mm 1.5 mm 0.59 1.72 L 12.8:1
All valve timings are for lift = 0.03 mm.
Table 2. SCORE Fuel-Injection System Specifications injector type injector model nozzle style outer row inner row oil rail pressure max injection pressure intensification ratio valve opening pressure
Figure 1. Cross-sectional schematic of the SCORE as configured for the collection of SINL data through the piston window and for high-speed imaging of NL through one of the cylinder windows. For high-speed spray visualization, laser light is directed through the piston window into the combustion chamber via the mirror. For the fuel-spray pattern shown, only one inner-cone jet is visible in the cross-sectional plane because the inner-cone jets are aligned with every other outer-cone jet. The two piston-bowl-rim windows (not labeled) immediately adjacent to the end of the outer-cone jets enable optical access to the piston bowl when the piston is near TDC.
Caterpillar HEUI A HIA-450 single-guided 10 103 μm 70° VCO 5 103 μm 35° minisac 20.8 MPa (3000 psig) 142 MPa (20 600 psia) 6.85:1 31 MPa (4500 psig)
enabling future engines to employ early direct-injection, LTC operating modes. 2. Experimental Setup and Procedures Sandia Compression-Ignition Optical Research Engine (SCORE). The SCORE is a single-cylinder version of a Caterpillar heavy-duty engine that has been modified to provide extensive optical access to the combustion chamber. A crosssectional schematic of the SCORE is shown in Figure 1, and engine specifications are provided in Table 1. The SCORE differs from a typical production engine in that the piston bowl has a flat bottom and vertical sidewalls (i.e., has the shape of a right cylinder) to facilitate imaging. In addition, the piston rings are positioned lower on the optical engine to allow for placement of piston-bowl-rim windows and to prevent the rings from riding over windows in the cylinder wall. These differences result in an increased clearance volume and decreased compression ratio in the SCORE relative to a typical production engine. To address this issue, the temperature and pressure of the intake-charge gas can be increased to better match the conditions in a higher-compression-ratio, production engine over a desired crank-angle range of interest (usually near TDC, where ignition and combustion are occurring). The SCORE is equipped with a Caterpillar hydraulically actuated, electronically controlled unit injector (HEUI) capable of injection pressures of up to 142 MPa. The injector axis is collinear with the cylinder-bore axis. General specifications of the fuel-injection system are provided in Table 2. For this study, the injector was equipped with a 15-hole, dual-row, narrowincluded-angle nozzle (10 holes 70° and 5 holes 35°) with 103-μm orifices. The alignment of one inner-row and two outerrow fuel sprays is illustrated in Figure 1. The 70° included angle of the outer row of 10 evenly spaced orifices is such that injected fuel is directed fully within the piston bowl as long as the engine is 65 crank angle degrees (CAD) or less from TDC. Accounting for the jet penetration rate allows for an actual start of injection (SOIa) as early as 70 CAD before TDC, or -70 CAD.1The inner
liquid-phase fuel impingement on in-cylinder surfaces during LTC operation. In this paper, the addition of high-volatility fuel components is investigated as a way to reduce or eliminate in-cylinder liquid-fuel impingement and fuel-film accumulation. The general effect of fuel volatility on combustion, emissions, and efficiency is assessed. While results have been presented in the literature on fuel volatility effects when using advanced in-cylinder combustion strategies, oftentimes the effect of fuel volatility is not decoupled from the effect of ignition quality (e.g., Zhong et al.8). In addition, much of the work has focused on true-HCCI combustion modes (e.g., Bunting9), where direct injection is not employed and the potential impact of fuel volatility is less significant. In the current study, an optical engine based on a Caterpillar heavy-duty engine is used to investigate blends of a conventional no. 2 diesel fuel with a high-volatility fuel mixture having approximately the same ignition quality as the baseline diesel fuel. Diagnostics include conventional heatrelease analysis, the measurement of spatially integrated broadband light emitted during the combustion process (natural luminosity, NL), and high-speed in-cylinder imaging of both NL and laser light elastically scattered from liquidphase fuel in the charge gas. Engine-out emissions of NOx, smoke, HC, and CO also were monitored. The results indicate that increased fuel volatility could play an important role in (8) Zhong, S.; Wyszynski, M. L.; Megaritis, A.; Yap, D.; Xu, H. Experimental Investigation into HCCI Combustion Using Gasoline and Diesel Blended Fuels. SAE Technical Paper 2005-01-3733; 2005. (9) Bunting, B. G. Combustion, Control, and Fuel Effects in a Spark Assisted HCCI Engine Equipped with Variable Valve Timing. SAE Technical Paper 2006-01-0872; 2006.
1 Throughout this paper, CAD will be referenced in a manner such that it ranges from -360 to þ360, with 0 CAD corresponding to TDC of the compression stroke.
1539
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Table 3. Selected Fuel Properties for HV0 and HV100 fuel description a
CN sulfur [ppm] aromatics [mass %] one ring twoþ rings density [kg/m3] lower heating value [MJ/kg] volumetric lower heating value [MJ/L] a
HV0
HV100
no. 2 diesel 45.9 11
20 vol % toluene in balance n-heptane 47.5
23.1 11.7 839 42.8
24.0 0 720 43.6
35.9
31.4
Cetane number derived via ASTM D6890.
Figure 2. Distillation curves via ASTM D86 for the five experimental test fuels.
row of five evenly spaced orifices is aligned such that the axis of every other outer-row jet is coplanar with the axis of an innerrow jet and the cylinder-bore axis. Further details of the SCORE can be found in refs 7 and 10-12. It should be noted that the geometric compression ratio reported in Table 1 is higher than that reported in ref 7 because of slight engine modifications that reduced the crevice volume. Test Fuels. The baseline diesel fuel used in the study was a 2007 ultra-low-sulfur no. 2 diesel certification fuel obtained from Chevron Phillips Chemical Company. The fuel had 11 ppm sulfur and 34.8 mass % aromatics. A high-volatility fuel with approximately the same ignition quality as the baseline fuel was prepared by blending 80 vol % n-heptane with balance toluene. The ignition quality was quantified by the derived cetane number (CN) via ASTM D6890. Normal (atmospheric pressure) boiling points of n-heptane and toluene are 98.4 and 110.7 °C, respectively,13 compared to a boiling point range of 194-351 °C for the baseline diesel fuel. The high-volatility fuel is denoted HV100, while the baseline diesel fuel is denoted HV0. Blends of intermediate fuels with varying levels of high-volatility content (HV content) were also prepared by mixing 28, 56, and 78 vol % HV100 in HV0. These fuels are denoted as HV28, HV56, and HV78, respectively. Selected fuel properties for HV0 and HV100 are provided in Table 3. It is noted that, while the monoaromatic content is wellmatched between HV0 and HV100, HV0 contains 11.7 mass % of di- and triaromatics that could increase its soot-formation tendency relative to HV100. The implications of this are discussed below in the Emissions and Fuel-Conversion Efficiency section of the paper. Fuel distillation curves obtained via ASTM D867 for each of the five experimental test fuels are shown in Figure 2. Engine Operating Conditions. The SCORE was operated in a skip-fired mode for a limited duration of 75 fired cycles per engine run to reduce the risk of window failure due to thermal and mechanical stresses. Engine operating conditions for this study were selected to facilitate comparison to the baseline conditions of the previous Martin et al. study, which also utilized the SCORE.7 The engine was operated at 1200 rpm, with the fuel injector fired on every 12th engine cycle (i.e., 11 motored cycles occur between one fired cycle and the next). Engine load for the fired cycles was maintained at 4.82-bar gross indicated mean effective pressure (gIMEP).
Because of the skip-fired operation of the optical engine, the intake mixture obtained by recirculating its exhaust would have lower carbon dioxide (CO2) and water (H2O) concentrations, and a higher oxygen (O2) concentration, than the intake mixture in a comparable continuously fired engine operated at the same EGR level. For this reason, EGR was simulated by diluting dry intake air with nitrogen (N2) and CO2 to achieve the same intake-O2 mole fraction and mixture constant-pressure specific heat as would occur in a continuously fired engine of the same compression ratio operating with 50% EGR. The desired mixture of air, N2, and CO2 was metered using sonic-flow orifices. Flow rates were determined from the orifice diameters and measured upstream pressures using established procedures.14 Because the simulated-EGR mixture lacks H2O, the mole fraction of CO2 is higher in the simulated-EGR mixture to better match the constant-pressure specific heat of the comparable mixture of real EGR. Fuel-injection timing was varied such that SOIa ranged from -70 to -30 CAD, in 10 CAD increments, with a set-point tolerance of (0.5 CAD. As indicated above, the -70 CAD set point is the earliest injection timing that allows fuel to be directed entirely within the piston bowl. For the current load condition, injection timings later than SOIa = -30 CAD were previously shown to result in ignition before fuel injection was completed.7 Note that SOIa represents the CAD at which fuel begins to enter the combustion chamber, as determined by spray-visualization imaging. It occurs later than the indicated start of injection (SOIi), which represents the CAD at which the fuel-injection trigger signal is sent to the electronic control module. The difference between SOIi and SOIa (i.e., the injector lag) is due to the finite time required for the injector and fuel to respond to the trigger signal. For the fuels used in this study, the injector lag ranged from 1436 to 1556 μs (from 10.3 to 11.2 CAD at 1200 rpm). For each test fuel and SOIa combination, the length of the fuel-injection trigger signal, or indicated duration of injection (DOIi), was selected to maintain the 4.82 gIMEP load set point. DOIi values ranged from 1257 to 1399 μs (from 9.1 to 10.1 CAD at 1200 rpm). Figure 3 shows the DOIi values used as a function of SOIa and the HV content. As will be the case with all similar figures presented in this paper, data points in the figure represent averages over all runs at the same test condition; error bars denote minimum and maximum run values. Note that, for many of the data, the run values are themselves averages over the 75 fired cycles that occur during the run. For each test condition, no fewer than three repeated runs were conducted. Additional details regarding engine operating conditions are presented in Table 4. The intake manifold pressure and temperature were originally selected by Martin et al.7 so that
(10) Mueller, C. J. M. P. Musculus. Glow Plug Assisted Ignition and Combustion of Methanol in an Optical DI Diesel Engine. SAE Technical Paper 2001-01-2004; 2001. (11) Mueller, C. J.; Martin., G. C. Effects of Oxygenated Compounds on Combustion and Soot Evolution in a DI Diesel Engine: Broadband Natural Luminosity Imaging. SAE Trans. 2002, 111 (4), 518–537 (SAE Technical Paper 2002-01-1631). (12) Mueller, C. J.; Martin, G. C.; Briggs, T. E.; Duffy., K. P. An Experimental Investigation of In-Cylinder Processes Under Dual-Injection Conditions in a DI Diesel Engine. SAE Trans. 2004, 113 (3), 1146– 1164 (SAE Technical Paper 2004-01-1843). (13) NIST Chemistry WebBook, http://webbook.nist.gov/chemistry/, accessed January 20, 2009.
(14) ASME/ANSI standard MFC-7M: Measurement of Gas Flow by Means of Critical Flow Venturi Nozzles, 1987.
1540
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
fuel penetration, mixing, and combustion. The high-speed camera images through a cylinder-wall window (see Figure 1) using an Abakus Ltd. 20-mm-diameter variable-focal-length borescope with a 115° field of view. The borescope can be fixed in a position to enable in-cylinder imaging with a clear view from the surface of the head to the bottom of the piston bowl from approximately -75 to þ75 CAD. For natural-luminosity movies, the high-speed camera was set up individually for each operating condition to optimize lowlight collection while allowing the peak NL levels to significantly saturate the camera under certain test conditions. Because of the varying camera and lens settings, the camera saturation, and the nonlinear grayscale map applied to the NL movie images, the images should be used primarily as a qualitative indicator of the areas where peak luminosity occurs; SINL data should be used for comparing relative luminous intensities. For spray-visualization movies, the combustion chamber was flood-illuminated using a 532-nm Nd:YAG laser beam passed through a diffuser and entering the combustion chamber through the piston window. The laser repetition rate (and therefore camera image acquisition rate) was 50 kHz, giving a temporal resolution of 20 μs (∼0.14 CAD at 1200 rpm). The laser pulse duration was ∼0.5 μs, effectively freezing the spray in each image. While NL movies and SINL data were recorded during the same engine runs, a slightly different experimental configuration was required to obtain spray-visualization movies because laser-light illumination is directed into the combustion chamber via the same optical path as that used to receive SINL. As a consequence, separate engine runs (at the same engine operating conditions) were conducted to record spray-visualization movies. Engine-Out Emissions. Exhaust-gas was drawn to emissions analyzers through a heated Teflon sampling line installed downstream of an exhaust surge tank. The surge tank has a volume of 125 L and serves to dampen pressure oscillations in the exhaust system as well as to mix the gases leaving the engine from both fired and motored cycles. NOx was measured using a California Analytical Instruments (CAI) model 600 heated chemiluminescence detector, HC was measured using a CAI model 600 heated flame ionization detector, and CO was measured using a CAI model 602-P nondispersive IR analyzer. Smoke emissions were measured using an AVL model 415S smokemeter. Because the SCORE is skip-fired for relatively short durations, pollutant concentrations in its exhaust do not match those that would be measured if the engine were continuously fired until emissions reached steady-state levels. The differences are due to (1) the large volume of the exhaust surge tank introducing a transient in the measured emissions such that steady-state concentrations are not reached until well into or near the end of the 90-s duration of engine firing and (2) dilution of the exhaust gas from the fired cycles by that from the interspersed motored cycles. To address transients in gaseous emissions levels due to the first issue noted above (i.e., to determine concentrations that appropriately represent those that would be measured during steady-state skip-fired operation), the entire time history of each gaseous species is integrated (to provide a ppm 3 s result), and the result is divided by the duration of the engine firing. This yields the measured average mole fraction of the species in the skipfired exhaust. To address the difference in gaseous emissions due to skipfiring (i.e., the second issue noted above), the average skip-fired mole fraction is used to compute an indicated specific emission level (e.g., in g/hp 3 h) by dividing by the indicated power output during the fired cycles only. Indicated specific NOx emissions are calculated using the molar mass of NO2 (46 g/mol) as per conventional practice.17 The filter smoke number (FSN) was
Figure 3. Indicated duration of injection (DOIi) for each of the 25 conditions in the test matrix. Table 4. Engine Operating Conditions speed load (gIMEP) intake manifold pressure intake manifold temp simulated EGR intake O2 mole fraction SOIa
1200 rpm 4.82 bar 1.418 bar 42 °C 50% 15.0% -70 to -30 CAD
motored temperature, pressure, and density at -30 CAD were matched to a Caterpillar 3401E single-cylinder oil test engine operating in an HCCI mode with a 12:1 compression ratio.15 Diagnostics. In-Cylinder Pressure. In-cylinder pressure was measured with 0.5 CAD resolution using a water-cooled quartz piezoelectric pressure transducer (AVL model QC32C). Cylinder-pressure data were used to calculate parameters such as gIMEP, the apparent heat-release rate (AHRR), and the bulkgas-averaged in-cylinder temperature. Additional details of the cylinder-pressure data acquisition system for the SCORE can be found in ref 10. Optical Measurements. NL refers to the broadband light emitted by the combustion process during a fired cycle. This luminosity arises from both soot incandescence and chemiluminescence, but the contribution from soot incandescence has been shown to be 4-5 orders of magnitude higher than that from chemiluminescence during mixing-controlled combustion.16 The instantaneous flux of NL visible through the piston window was collected every 0.5 CAD using a New Focus model 2031 high-speed, large-area photoreceiver (400-1070 nm spectral sensitivity) equipped with a Nikkor 50 mm, f/1.4 lens. Although temporally resolved, this measurement is spatially integrated over the region of the combustion chamber visible through the piston window. Therefore, the measurement is denoted as spatially integrated natural luminosity (SINL). Under certain conditions, SINL can serve as a relative measure of the average in-cylinder soot volume fraction;11 however, in this study, SINL is used simply as a relative indicator of the presence of hot soot. Assuming similar soot particle properties and optical thicknesses of radiating soot clouds within the cylinder, higher SINL at a given crank angle indicates more radiant flux from hot soot from the viewable region of the combustion chamber into the spectral collection bandwidth of the photoreceiver. In-cylinder natural-luminosity and spray-visualization movies were acquired using a Phantom V7.3, 14-bit, high-speed camera to provide information about in-cylinder liquid-phase (15) Bessonette, P. W.; Schleyer, C. H.; Duffy, K. P.; Hardy, W. L.; Liechty., M. P. Effects of Property Changes on Heavy-Duty HCCI Combustion. SAE Trans. 2007, 116 (3), 242–254 (SAE Technical Paper 2007-01-0191). (16) Dec, J. E.; Espey., C. Chemiluminescence Imaging of Autoignition in a DI Diesel Engine. SAE Trans. 1998, 107 (3), 2230–2254 (SAE Technical Paper 982685).
(17) Code of Federal Regulations. Title 40, Part 86: Control of Emissions from New and In-use Highway Vehicles and Engines.
1541
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 4. CAD-resolved in-cylinder pressure, AHRR, and SINL for representative runs for each test fuel at SOIa = -70 CAD. Curves shown are averages over 75 fired cycles.
Figure 5. CAD-resolved in-cylinder pressure, AHRR, and SINL for representative runs for each test fuel at SOIa = -60 CAD. Curves shown are averages over 75 fired cycles.
corrected for dilution due to skip firing by assuming the sample volume is that which would pass through the smokemeter over the duration of the fired cycles only. In other words, duration of fired cycles corrected sample volume ¼ sample duration
volume of approximately 16 mL was injected into the fuel capture vessel (corresponding to a mass of 11.5-13.4 g, depending on the density of the fuel). An Acculab AL-1502 balance with a resolution of 0.01 g was used to record the difference in the weight of the vessel before and after the injector firings. This result was then used to quantify the average mass of fuel introduced into the combustion chamber per injection event. Several measurements were made for each fuel at each duration of injection. The calculated average mass of fuel per injection was highly repeatable, with a relative standard deviation of less than 1% between measurements.
actual sample volume For the smokemeter’s sample duration of 120 s and with 75 fired cycles at 1200 rpm, the correction factor becomes 0.0625. Quantity of Fuel Injected. Fuel consumption was quantified by measuring a priori the mass of fuel injected as a function of the indicated duration of fuel injection (DOIi). This calibration procedure was carried out for each test fuel by using a stainless steel capture vessel placed under the cylinder head and sealed around the tip of the injector. With the engine static but heated to normal operating temperature, the fuel injector was fired a known number of times into the vessel at approximately atmospheric pressure. The rate of fuel injector firing was identical with that used during actual skip-fired engine operation. The number of injections was selected such that a total
3. Results and Discussion Pressure-Derived and SINL Data. Figures 4-8 show suites of plots of CAD-resolved data from representative runs at each of the five injection timings. Individual line patterns in each plot designate the different test fuels. In all cases, the actual duration of fuel injection is ∼12 CAD and thus fuel injection into the cylinder has completed prior to the start of 1542
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 6. CAD-resolved in-cylinder pressure, AHRR, and SINL for representative runs for each test fuel at SOIa = -50 CAD. Curves shown are averages over 75 fired cycles.
Figure 7. CAD-resolved in-cylinder pressure, AHRR, and SINL for representative runs for each test fuel at SOIa = -40 CAD. Curves shown are averages over 75 fired cycles.
combustion (onset of heat release). The AHRR curves in the figures display a smaller, initial heat release, which will be denoted as the cool-flame heat release.2 The cool-flame heat release is followed by the larger, main heat release. Across all injection timings, the effect of the HV content on cool-flame heat release was relatively small. At SOIa = -30 CAD, the HV content also had a limited effect on the main heat release. At all other injection timings, however, increasing the HV content significantly shortened the duration and increased the peak of the main heat release. In addition, the start of the main heat release was advanced with higher HV-content fuels. The combination of shorter duration and earlier start for the main heat release resulted in advanced combustion phasing with increased HV content. These effects are shown more quantitatively in Figures 9
and 10. Figure 9 presents maximum AHRR (AHRRmax) values, while Figure 10 presents the change in combustion phasing with the HV content, as measured by the CAD at which 50% of the injected fuel energy has been released (CA50). The observed trends may be due to the slightly higher CN for HV100 relative to HV0 (see Table 3) and/or more-complete premixing with the higher volatility fuels. The latter mechanism could result in a portion of the main heat release being shifted from a more mixing-controlled combustion to a more volumetric autoignition. Aside from their impacts on the combustion process and emissions, AHRRmax and combustion phasing can have important impacts on engine noise and durability. In terms of the injection-timing effect on heat release (comparison across Figures 4-8), it is notable that the AHRR curves become flatter and wider as the injection timing is advanced, except at SOIa = -70 CAD, where the AHRR curves once again rise to a sharp peak. This seemingly counterintuitive result could arise from the interaction between
2 In the language of HCCI, this initial heat release is commonly termed the low-temperature heat release, or LTHR. However, this term will not be used in this paper to avoid confusion with the use of LTC to describe the overall combustion strategy.
1543
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 10. Combustion phasing as measured by CAD for 50% heat release (CA50).
Figure 11. Recorded levels of SINLmax. For SOIa = -70 CAD, SINLmax decreases as the HV content increases because of the elimination of pool fires. For SOIa = -30 CAD, the increase in SINLmax with the HV content is presumably due to inhibited mixing and/or higher soot temperatures.
vertical piston-bowl wall. Under these conditions, advancing injection timing allows more time for in-cylinder mixing and results in fuel-lean homogeneous combustion in the bulk gas, precisely what is intended with an early direct injection, LTC combustion strategy. For SOIa = -70 CAD, however, highspeed spray-visualization movies reveal that while the inner row of fuel jets is directed toward the piston-bowl bottom, the outer row of fuel jets is directed toward the vertical piston-bowl wall. This results in a converging of injected fuel near the bottom of the piston bowl. Thus, despite the very early fuel injection, the SOIa = -70 CAD case could lead to inhibited in-cylinder mixing, more regions with a nearstoichiometric mixture, and therefore greater AHRR values. The SINL data plotted in the bottom portion of Figures 4-8 reveal that significant luminosity (SINL > 1 mW) occurred only for the earliest and latest SOIa timings of -70 and -30 CAD, respectively. Some initial observations will be noted here, but a more detailed discussion (including insights from the in-cylinder imaging experiments) is presented in the following subsection. For SOIa = -30 CAD, increasing the HV content resulted in higher SINL levels. However, for the other injection timings, higher volatility is associated with lower SINL levels. The effect is shown more directly in Figure 11. Inspection of the SINL and corresponding AHRR curves in Figures 4-8 also reveals that maximum SINL (SINLmax) levels typically occur well after AHRRmax, which is consistent with pool-fire combustion of liquid-fuel films.7 As shown in Figure 12, the delay between AHRRmax and SINLmax ranged from ∼9 to 21 CAD, except for the highest volatility test fuels for injection timings other than
Figure 8. CAD-resolved in-cylinder pressure, AHRR, and SINL for representative runs for each test fuel at SOIa = -30 CAD. Curves shown are averages over 75 fired cycles.
Figure 9. AHRRmax.
the injected fuel and the piston-bowl surfaces. For injection timings of SOIa = -60 CAD and later, all fuel jets (both inner and outer rows) are directed toward the flat bottom of the piston bowl, and the mixture of fuel and entrained intakecharge gas is convected outward and then upward along the 1544
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 12. Delay (measured in CAD) between the location of AHRRmax and SINLmax. The sharp drops in the curves for injection timings other than SOIa = -30 CAD are due to a transition from SINLmax resulting from pool fires to SINLmax occurring from chemiluminescence in the bulk gas during the main heat release. The nearly horizontal line for SOIa = -30 CAD is indicative of the less homogeneous combustion that occurred at this injection timing.
Figure 14. Spray-visualization images for SOIa = -50 CAD. Each column shows images from an individual cycle for the indicated fuel. Images within each column progress forward in time from top to bottom.
bulk-gas mixing dominating that of any potential pool fires at this injection timing. Images from High-Speed Movies. As outlined in the Experimental Setup and Procedures section, high-speed movies were acquired to visualize the injected fuel spray as well as the in-cylinder NL. Figure 13 presents a panel of images from a high-speed spray-visualization movie obtained for SOIa = -70 CAD. The images shown are from a single cycle of a single engine run and are representative of the spray behavior at this condition. The three columns correspond to images from three different fuels: HV0, HV56, and HV100. Within a given column, viewing the images from top to bottom corresponds to a progression in time. The images were recorded at fixed intervals (time delays) after SOIi, with the time after SOIa indicated in the top-left corner of each image. The corresponding CAD is shown in the top-right corner of each image. The small differences in CAD values shown across a given row of images are associated with the (0.5 CAD set-point tolerance for SOIa and the fixed 20-μs timebased image intervals. Figures 14 and 15 present similar panels of images for SOIa = -50 CAD and SOIa = -30 CAD, respectively. High-speed spray-visualization movies were also acquired for the other injection timings and fuels tested; however, because of space
Figure 13. Spray-visualization images for SOIa = -70 CAD. Each column shows images from an individual cycle for the indicated fuel. Images within each column progress forward in time from top to bottom.
SOIa = -30 CAD. For SOIa = -30 CAD, the HV content did not significantly affect the delay between AHRRmax and SINLmax. The disparate effect of the HV content on SINL for SOIa = -30 CAD likely results from the effect of poor 1545
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 16. Calculated bulk-gas-averaged, in-cylinder temperatures for motored operation. Atmospheric-pressure boiling-point and boiling-point ranges, as well as crank angles over which the earliest and latest injections occur, are indicated.
2220 μs after SOIa (ASOIa). For HV100 (right column), however, the liquid-fuel jets almost completely vaporized before entering the piston bowl, with the 2220-μs-ASOIa image showing almost no liquid fuel remaining. For HV56, the images are not dissimilar to those of HV0. However, close inspection reveals that the liquid-fuel jets are somewhat broader in their thickness (diameter), with the 2220-μsASOIa image showing slightly more uniformity across the unvaporized fuel cloud within the piston bowl. While a larger jet-spreading angle would be expected with the lower density (higher HV content) fuels, the reader should also recall that the HV mixtures (HV28, HV56, and HV78) are blends of the baseline diesel and the HV n-heptane and toluene mixture. Thus, the images for HV56 are consistent with the lessvolatile diesel fuel portion remaining largely in the liquid phase but being more widely dispersed because of vaporization and expansion of the lower volatility n-heptane and toluene components. As previously discussed, the images for SOIa = -70 CAD show that the outer row of fuel jets impact the vertical piston-bowl wall and are thus directed further into the piston bowl, likely limiting in-cylinder mixing as compared to the other injection timings. Figure 14 illustrates the change in outer-row fuel-jet impingement for SOIa = -50 CAD. For this injection timing, the piston is higher in its compression stroke, and the outer-row fuel jets are directed at the bottom of the piston bowl, as shown most clearly by the 620-μs-ASOIa image for HV0. This is likewise the case for SOIa = -30 CAD (Figure 15), with the piston now much closer to TDC during fuel injection. Note that, in images after ∼-45 CAD, increasing portions of the fuel jets are being viewed through the piston-bowl-rim window, as the piston rises up and across the cylinder window. The effect of the HV content on fuel-spray vaporization for SOIa = -50 and -30 CAD is essentially similar to that described for SOIa = -70 CAD, above. However, as the injection timing is retarded, the higher piston position during fuel injection causes significant changes in the in-cylinder thermodynamic conditions. This is illustrated in Figure 16, which shows calculated bulk-averaged charge-gas temperatures during motored operation for CAD values spanning the injection timings investigated. The temperatures are calculated assuming adiabatic compression and using the mole fractions and temperature-dependent specific heats of
Figure 15. Spray-visualization images for SOIa = -30 CAD. Each column shows images from an individual cycle for the indicated fuel. Images within each column progress forward in time from top to bottom.
and simplicity considerations, they are not presented in this paper. For any of the cases not shown, the liquid-fuel penetration behavior falls between that of two bracketing cases presented, in terms of both injection timing and HV content. In the images of Figures 13-15, the initial stages of fuel injection can be observed in the first (topmost) row of images. As time progresses and fuel injection continues, the fuel jets enter the piston bowl, which is positioned higher at more retarded injection timings. Both the outer and inner rows of injected fuel jets can be seen, although the inner-row fuel jets are significantly obscured by the outer-row jets in the foreground. In some of the images, fuel-jet reflections can be seen in the bottom of the piston bowl (top of the piston window in Figure 1) and from the top of the piston-bowl-rim window. For all of the sequences shown, fuel injection into the cylinder ceases between the fourth and fifth (lowest) rows of images. As would be expected, the spray-visualization movies show significant differences in fuel-spray vaporization as the HV content is changed. In Figure 13, for instance, the images for HV0 (left column) show liquid jets penetrating well into the piston bowl, leaving a cloud of unvaporized fuel that remains largely contained within the piston bowl at 1546
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 17. NL images for SOIa = -70 CAD. Each column shows images from an individual representative cycle for the indicated fuel. Images within each column progress forward in time from top to bottom. Luminosity levels should not be compared across fuels because of the different image exposure settings used.
Figure 18. NL images for SOIa = -50 CAD. Each column shows images from an individual representative cycle for the indicated fuel. Images within each column progress forward in time from top to bottom. Luminosity levels should not be compared across fuels because of the different image exposure settings used.
the intake-gas-mixture components. Also shown in Figure 16 are the atmospheric-pressure boiling points of n-heptane and toluene, and the atmospheric-pressure boiling point range of the baseline HV0 diesel fuel (from ASTM D86; see Figure 2). The figure provides an indication of how readily injected fuel or fuel components are expected to vaporize at specific injection timings. It is acknowledged that a strict interpretation of the atmospheric-pressure boiling point data in Figure 16 should not be made because (1) in-cylinder pressures are significantly higher than atmospheric pressures, (2) fuel vaporization is governed by partial pressures as the fuel spray mixes with entrained intake-charge gas, and (3) local cooling of the intake-charge gas occurs as injected fuel vaporizes. Despite these qualifiers, Figure 16 offers useful guidance for interpreting the spray-visualization images. For instance, for the earliest injection timing of SOIa = -70 CAD, nearly all of the fuel injection takes place prior to the point where the bulk-averaged charge-gas temperature reaches the initial boiling point of the HV0. The corresponding spray-visualization movies (Figure 13) reveal that much of the HV0 (and the diesel component of HV56) remains in the liquid phase even following the end of injection. For SOIa = -30 CAD,
however, injection primarily occurs beyond the end boiling point of HV0, and much of the injected fuel vaporizes by the end of injection (Figure 15, bottom row). The in-cylinder temperature, density, and piston position also dramatically affect the likelihood of liquid-fuel impingement and subsequent liquid-fuel-film accumulation on piston surfaces. While liquid-fuel-film accumulation may seem more likely at more retarded injection timings (with the piston closer to the injector nozzle), the higher in-cylinder temperatures and densities (and possibly higher piston surface temperatures) limit liquid-fuel penetration and might preclude liquid-fuel-film accumulation. Spray-visualization images beyond the time intervals shown in Figure 13, as well as NL images to be presented, reveal that the persistence of liquid fuel and liquid-fuel films is most significant for SOIa = -70 CAD, for those fuels containing lower HV content (HV0, HV28, and HV56). For these lowervolatility fuels at other injection timings, the higher chargegas temperatures and densities (and presumably higher pistonsurface temperatures), along with the absence of outer-row fuel-jet impingement on the vertical piston-bowl wall, enable more complete fuel vaporization prior to the start of combustion. 1547
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
SOIa timings or with different fuels does not always accurately represent the difference in their relative intensities (i.e., a brighter image does not necessarily correspond to higher NL). A quantitative comparison of luminosity should depend on the SINL data previously presented in Figures 4-8 and 11. Nevertheless, the high-speed NL data can provide useful qualitative spatial information. As previously indicated, NL results primarily from soot incandescence. In-cylinder locations having significant NL in the images therefore represent locations where hightemperature soot is present. Inspection of the NL images for SOIa = -70 CAD (Figure 17) shows regions of intense luminosity at 10.0 and 36.0 CAD for the diesel-containing fuels HV0 and HV56. These regions are on and immediately adjacent to the piston-bowl-bottom surface, as is most clearly seen in the 10.0-CAD images, where the plane of the piston-bowl bottom is nearly orthogonal to the field of view of the borescope. The images reveal that, for these fuels and this injection timing, the accumulation of liquid-fuel films on the piston-bowl bottom is significant, resulting in surface pool fires that occur and continue well after the main heat release (Figures 4 and 12). As will be shown, the persistence of pool fires late into the expansion stroke has dire consequences with respect to measured engine-out emissions. Figure 18 reveals pool-fire activity for HV0 and HV56 for SOIa = -50 CAD as well. However, the intensity of the NL is much lower, as shown by a comparison of SINL data in Figures 4 and 6 and in Figure 11. For all injection timings other than SOIa = -30 CAD, the high-speed NL movies and SINL data reveal a general decreasing trend in the NL associated with pool fires as the HV content is increased. For HV78 and HV100, essentially no pool-fire activity was observed. Without the intense luminosity from soot incandescence, the weak NL with HV78 and HV100 appears to be primarily from chemiluminescence distributed throughout the combustion chamber (e.g., 1.0-CAD image for HV100 in Figure 17). This absence of pool fires is also evident in Figure 12, with the delay in SINLmax from pool fires replaced by much shorter delays associated with SINLmax from chemiluminescence directly associated with the main heat release. For SOIa = -30 CAD, different phenomena are observed. In Figure 19, intense NL is evident throughout the bulk gas in the region below the injector tip rather than being limited to the bottom of the piston bowl. The luminosity at this injection timing is being emitted from high-temperature soot created in fuel-rich combustion regions within the bulk gas. Although an injection timing of SOIa = -30 CAD allows for the completion of fuel injection prior to the cool-flame heat release, not enough in-cylinder mixing occurs to produce a significantly fuel-lean, homogeneous combustion process (due, in large part, to the narrow included angles of the inner and outer spray cones). The AHRR curves in Figure 8 show that the sharp peaks associated with more rapid combustion are followed more closely in time by the peaks in SINL, which is consistent with the hypothesis that the NL is associated with soot from the main combustion event rather than from lingering pool fires. This latter observation, also evident in Figure 12, reinforces the hypothesis that the combustion has entered a new (and less desirable) regime for SOIa = -30 CAD relative to the other injection timings. As is evident in Figures 8 and 11, NL at SOIa = -30 CAD increases with increasing HV content. In addition, the high-speed NL movies from this injection timing
Figure 19. NL images for SOIa = -30 CAD. Each column shows images from an individual representative cycle for the indicated fuel. Images within each column progress forward in time from top to bottom. Luminosity levels should not be compared across fuels because of the different image exposure settings used.
Figures 17-19 show panels of images from the high-speed movies of in-cylinder NL. The images are from individual representative cycles at each test condition. As is the case with the spray-visualization figures, each NL figure is associated with a specific injection timing, with columns within a figure representing different fuels and images within a column progressing forward in time from top to bottom. NL images were recorded at fixed CAD intervals rather than indexed relative to SOIa; thus, the engine position is described only in terms of CAD. The images shown are actually superpositions of NL images onto images taken of the engine (at the same CAD) during motored operation with laser flood illumination. This allows the piston and cylinder surfaces to be visible even in cases where they are too dark to be seen with the camera gains used during NL imaging. Note that, in most of the images, NL is being transmitted through the piston-bowl-rim window and that the piston is moving downward with increasing positive values of CAD (i.e., after TDC). Also, the reader is reminded that the camera gain, lens aperture, and exposure time used for NL imaging vary from one operating condition to another to best capture the details of the emitted luminosity. Therefore, comparing the brightness of NL images acquired at different 1548
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
Figure 21. Indicated fuel-conversion efficiency.
Figure 22. Calculated fired-cycle bulk-gas-averaged in-cylinder temperature.
behavior and not to identify specific conditions that meet regulated emissions targets. The optical engine has larger crevice volumes and a simplified bowl geometry, and it has been operated only at steady-state conditions rather than over a transient emissions test cycle. It is connected to an emissions bench that, while carefully calibrated, has not been certified for emissions testing per ref 17. For these reasons, the reported emissions data are not necessarily equal to those that would be expected from a production-like engine operated at comparable conditions. Nevertheless, prior work has shown that the optical engine can accurately capture the emissions trends of a comparable production-like engine.7 Fuel-conversion efficiency data are presented in Figure 21. Examination of the NOx data reveals that, in general, increasing HV content is associated with slight increases in NOx emissions across all injection timings. These results are consistent with the pressure-derived data, which show higher and more advanced AHRR peaks (and more advanced combustion phasing) as the HV content is increased (see AHRR curves in Figures 4-10). This would result in longer residence times at higher in-cylinder temperatures for those combustion regions that produce NOx. The effect of the HV content on the in-cylinder temperature is more directly shown in Figure 22, which shows calculated bulk-gas-averaged in-cylinder temperatures, assuming ideal-gas behavior and a constant number of moles (the actual number of moles in the combustion chamber changes by no more than 2% as a result of fuel injection and combustion). The effect of injection timing on NOx can also be interpreted based upon the AHRR data, with higher NOx emissions again
Figure 20. Engine-out emissions. Indicated specific emissions for NOx, HCs, and CO. Smoke measurements are reported in terms of FSN.
reveal that regions of high-temperature soot become larger when the HV content is increased. Higher volatility therefore appears to enhance in-cylinder soot production at this mostretarded injection timing. This could be due to fuel vaporization and mixing differences that produce local stoichiometries that promote soot formation with increased HV content. Emissions and Fuel-Conversion Efficiency. Figure 20 presents the engine-out emissions data for NOx, smoke (as represented by FSN), HC, and CO. These data are provided to link emissions trends with observed spray and combustion 1549
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
associated with higher, more advanced AHRR peaks. The AHRR peaks generally decrease with more advanced injection timing, with the notable exception of SOIa = -70 CAD, because of the in-cylinder mixing issues previously discussed. Smoke emissions for injection timings other than SOIa = -30 CAD follow trends that would be expected, based upon the previously discussed impact of the HV content on poolfire activity. The high-speed NL movies revealed that sootgenerating pool fires persisted well into the expansion stroke, when bulk-gas temperatures and mixing rates would be insufficient to oxidize the soot produced. Thus, because increased HV content limits pool-fire activity, smoke emissions decrease. Reductions in smoke were significant with each step increase in the HV content, until at HV78 essentially no smoke emissions are measured. For SOIa = -30 CAD, the NL movies showed that soot production occurred in the in-cylinder bulk gas during the main (and at least partially mixing-controlled) combustion phase. Thus, pool fires were not a significant contributor to smoke emissions, and the effect of the HV content was small. In addition, because changing the HV content produced a minimal effect on smoke emissions at this injection timing, the smoke reductions at other injection timings cannot simply be attributed to changes in the chemical composition of the fuel. That is, if the reduction of the multiring aromatic content in the higher-HV-content fuels was of significant importance, the apparent effect of fuel volatility on smoke emissions would be similar for injection timings with or without significant pool-fire activity, but this is not observed. Emissions of HC and CO, along with the fuel-conversion efficiency results, can generally be explained by considering that the presence of liquid-fuel films would limit the fuel from participating in properly phased (i.e., near-TDC) combustion. With low HV content, liquid-fuel films result in pool fires or late-cycle vaporization and incomplete combustion. This corresponds to increased HC and CO emissions and lower fuel-conversion efficiency.7 The HC and CO emissions are lowest for SOIa = -30 CAD presumably because of the limited amount of charge-gas premixing at this injection timing and the resultant heat release occurring at hightemperature conditions that support more complete oxidation of HC and CO. Discussion of Aggregate Data. The data presented above serve to present a general picture of the effect of fuel volatility during early direct-injection, LTC operation. As injection timing is advanced during LTC operation, liquid-fuel films on in-cylinder surfaces are more likely to form because of low in-cylinder temperatures and densities, and for the engine configuration used in this study, this could be exacerbated by a difference in fuel-spray targeting (onto the piston-bowl wall rather than the piston-bowl bottom). Liquid-fuel films can lead to pool fires that continue well into the expansion stroke and produce high levels of soot that do not fully oxidize before the exhaust process. Liquid-fuel films also can result in higher HC and CO emissions and lower fuelconversion efficiencies. Even if pool fires do not ignite under certain conditions, liquid-fuel films prevent some of the injected fuel from participating in the main combustion process and thus should be avoided.7 With an increase in fuel volatility, liquid-fuel films and pool fires that occurred with the baseline HV0 diesel fuel were reduced or eliminated. The effect of fuel volatility was
more dramatic for those engine operating conditions at which pool fires were the most problematic. For the conditions studied, an HV content of 78% (HV78) eliminated pool fires and reduced smoke emissions to near-zero levels (for all injection timings except SOIa = -30 CAD). The elimination of liquid-fuel films with increasing HV content also greatly increased fuel-conversion efficiency, presumably because of the higher combustion efficiencies possible without liquid fuel being “trapped” on in-cylinder surfaces. Increasing the HV content did result in moderately higher NOx emissions because of longer residence times at higher in-cylinder temperatures for those combustion regions that produce NOx.18 Nevertheless, the NOx increases associated with higher HV content were small compared to the NOx reduction benefit of early direct-injection, LTC operation. The effect of fuel volatility was dissimilar at the most retarded injection timing of SOIa = -30 CAD. At this injection timing, liquid-fuel films and pool fires were not evident. Because of the absence of these in-cylinder phenomena, no benefits were gained by increasing fuel volatility. The behavior at this injection timing appears to represent a shift to an undesirable combustion regime, likely because of poor mixing resulting from the combination of the short ignition delay at this retarded injection timing and the narrowincluded-angle nozzle. 4. Summary and Conclusions An optically accessible heavy-duty diesel engine was used to investigate the impact of fuel volatility on early directinjection, LTC combustion. Five blends of conventional no. 2 diesel fuel and a high-volatility fuel mixture of n-heptane and toluene having approximately the same ignition quality were tested over a range of injection timings. In addition to conventional pressure-based and engine-out emissions measurements, optical diagnostics were used to measure SINL and record high-speed sprayvisualization and NL movies. The current work is a follow-up study to that of Martin et al.,7 who conducted similar experiments using only the baseline diesel fuel. The results of the current work are consistent with and support the major conclusions of ref 7, namely, the following: (i) Liquid-fuel films can produce highly sooting pool fires that persist into the expansion stroke (well after the main heat release). Because bulk-gas temperatures and mixing rates are insufficient to oxidize the soot being produced, high smoke emissions result. (ii) Increases in HC and CO emissions, as well as reductions in fuel-conversion efficiency, can be attributed to liquid-fuel films that limit complete combustion and/or optimal combustion phasing. The following additional observations and conclusions can be made based upon the results of this study: (a) The level of the HV content had a small impact on the cool-flame heat release. However, for the main heat release, increasing the HV content significantly advanced combustion phasing and increased the peak (18) Mueller, C. J.; Boehman, A. L.; Martin., G. C. An Experimental Investigation of the Origin of Increased NOx Emissions when Fueling a Heavy-Duty Compression-Ignition Engine with Soy Biodiesel. SAE Int. J. Fuels Lubr. 2009, 2 (1), 789–816 (SAE Technical Paper 2009-011792).
1550
Energy Fuels 2010, 24, 1538–1551
: DOI:10.1021/ef9011142
Cheng et al.
heat release, resulting in slightly higher NOx emissions. (b) SINL levels were the highest at SOIa = -70 and -30 CAD but resulted from different phenomena. At SOIa = -70 CAD, luminosity originated from large pool fires on the bottom of the piston bowl. At SOIa = -30 CAD, luminosity arose from nonhomogeneous combustion in the bulk gas that occurred because of the short time available for in-cylinder mixing prior to ignition. (c) The HV content reduced pool fires and SINL for all injection timings except at SOIa = -30 CAD, where significant pool-fire activity was not observed. (d) High-speed spray-visualization movies showed notable differences in fuel-spray vaporization as the HV content was increased. More rapid fuel vaporization and less liquid-fuel impingement were observed with higher-HVcontent fuels, with the impact more significant at earlier injection timings. (e) High-speed NL movies revealed that the level of poolfire activity can be associated with the extent of liquidfuel-film accumulation. At a given injection timing, less pool-fire activity occurred as the HV content was increased.
(f) Increases in the HV content greatly reduced smoke emissions and greatly increased fuel-conversion efficiency under those early injection conditions where pool-fire activity was most significant. For the LTC conditions studied, an HV content of 78% (HV78) eliminated pool fires and reduced smoke emissions to near-zero levels. On the basis of this work, it appears that the detrimental effects of liquid-fuel films will make it extremely challenging to achieve desired efficiency and emissions targets with early direct-injection, LTC strategies using conventional diesel fuel. The use of higher-volatility blendstocks or fuels may be an important means of limiting the problems resulting from fuelfilm formation. Acknowledgment. Funding for this research was provided by the U.S. Department of Energy, Office of Vehicle Technologies. The authors thank program manager Kevin Stork for supporting this study. The research was conducted at the Combustion Research Facility, Sandia National Laboratories, Livermore, CA. Sandia is a multiprogram laboratory operated by Sandia Corp., a Lockheed Martin Company, for the U.S. Department of Energy’s National Nuclear Security Administration under Contract DE-AC0494AL85000.
1551