* Corresponding author

Transcription

1 Characterization of Dual-Fuel PCCI Combustion in a Light-Duty Engine S. L. Kokjohn * and R. D. Reitz Department of Mechanical Engineering University of Wisconsin - Madison Madison, WI 5376 USA Abstract. This study uses a multi-dimensional CFD code, the KIVA-CHEMKIN code, and multi-zone modeling to investigate a dual-fuel PCCI combustion strategy. The present dual-fuel PCCI concept uses port-fuel-injection gasoline and direct-injection of diesel fuel to create an optimal fuel reactivity distribution. In this work, dual-fuel PCCI operation at 5.5 bar net IMEP and 23 rev/min in a light-duty engine is characterized through a parametric study. Sweeps of injection timing and premixed gasoline percentage were evaluated and an operating regime map to achieve high efficiency low NOx low noise combustion was generated. It was found that highly-efficient dual-fuel operation with moderate rates of pressure rise (5 to 8 bar/deg) and near zero soot and NOx could be achieved by using a single, early cycle direct-injection. Additionally, it was found that, by introducing a gradient in fuel reactivity, the combustion duration could be extended to reduce the rate of pressure rise. Using this method, premixed operation was achieved with low rates of pressure rise without using EGR. To understand the effects of fuel reactivity and equivalence ratio stratification, multi-zone modeling was used. It was found that the observed extended combustion duration was primarily due to the gradient in fuel reactivity. Introduction Many researchers have shown that premixed compression ignition combustion (PCCI and HCCI) strategies are capable of achieving low NOx and low soot emissions while maintaining diesel like efficiency. Due to the existing fuel infrastructure, most PCCI research has been conducted using either strictly gasoline or diesel fuel. However, in their neat forms, each fuel has specific advantages and shortcomings for PCCI operation. Gasoline has a high volatility; thus, evaporation is rapid and a premixed charge can be obtained using port-fuel-injection. However, because the autoignition qualities of gasoline are poor, it becomes difficult to achieve combustion at low-load conditions. Conversely, diesel fuel has superior auto-ignition qualities; however, this can result in difficulty controlling the combustion phasing as engine load is increased. Furthermore, because diesel fuel is difficult to vaporize, portfuel-injection cannot be used and charge preparation becomes a challenge. The authors previous work [1, 2] has focused on dualfuel operation using a heavy-duty diesel engine. Multidimensional modeling and engine experiments were performed and it was found that by optimizing the gasoline-todiesel fuel ratio, optimal combustion phasing could be achieved and fuel consumption could be minimized. Additionally, it was found that in-cylinder reactivity gradients extended the combustion duration and reduced the rate of pressure rise compared to single-fuel PCCI combustion. The improved control over combustion phasing and duration allowed an extension of the PCCI operating regime to higher engine loads, while maintaining low rates of pressure rise (<1 bar/deg), low NOx and soot, and low fuel consumption. For example, at 9 bar IMEP and 13 rev/min, controlled PCCI combustion was achieved with NOx and soot levels significantly below the US EPA 21 heavy-duty limits while reaching a net indicated thermal efficiency of 53% (net ISFC of 158 g/kw-hr) [1]. In this work, multi-dimensional modeling is used to build on previous dual-fuel research to explore dual-fuel operation in a light-duty engine. In-cylinder fuel blending using port-fuel-injection of gasoline and optimized directinjection of diesel fuel is used to control combustion phasing and duration. The objective of this work is to characterize dual-fuel PCCI operation in the light-duty engine operating at 5.5 bar net IMEP and 23 rev/min. Engine and Model Specifications The engine used in this study is a single cylinder version of the GM 1.9 L. Engine and injector specifications are given in Table 1. CFD modeling was performed using the KIVA-3v release 2 code [3] with improvements to many physical and chemistry models developed at the ERC [4-6]. The KIVA- 3v code is coupled with the CHEMKIN II solver for detailed chemistry calculations. A 45 species and 142 reac- * Corresponding author

2 Table 1. Engine specifications Base engine type... GM1.9 L Bore x stroke x 9.4 cm Connecting Rod Length cm Displacement L Compression ratio :1 Swirl ratio Bowl type...reentrant Intake valve closing º ATDC Firing Exhaust valve opening...112º ATDC Firing Injector type...high-pressure solid-cone Manufacturer... Bosch Included angle Nozzle... 7 x 141 µm Table 2. Operating conditions Net IMEP bar Engine speed rev/min Total fuel mass mg/cycle EGR... % Premixed gasoline to 85% by mass Diesel injection timing to -6 ATDC Injection pressure...5 bar Intake surge tank temperature... 4 C Intake surge tank pressure bar IVC temperature C IVC pressure bar tion mechanism for primary reference fuels (PRF) [7] was used to simulate gasoline and diesel fuel chemistry. Many studies (e.g., Ra et al. [8]) have shown the combustion characteristics of gasoline and diesel are represented well by iso-octane (i.e., PRF 1) and n-heptane (i.e., PRF ), respectively. The physical properties (for the spray and mixing processes) of diesel fuel are represented by tetradecane. This modeling approach has been shown to yield acceptable results in numerous studies (e.g., Kong et al. [9]). Soot is predicted using a phenomenological soot model [9] based on the approach of Hiroyasu [1]. The soot model used in the present study uses acetylene as an inception species, which allows the soot model to be coupled to the chemistry solver through the addition of 13 reactions involving acetylene. NOx emissions are predicted using a reduced NO mechanism [11] consisting of 4 additional species and 12 reactions. The reduced NO mechanism is based on the Gas Research Institute (GRI) NO mechanism [12]. The suite of improved ERC spray models (e.g., the Gasjet model of Abani et al. [4]) have been employed to model the spray evolution. The computational grid is shown in Figure 1. Figure 1. Computational mesh used for the present study Results The effects of gasoline percentage and diesel start-ofinjection timing (SOI) were investigated through a parametric study. Operation was characterized at 5.5 bar IMEP and 23 rev/min. Table 2 shows the operating conditions. Although only the closed portion of the cycle (i.e., IVC to EVO) was modeled using KIVA, GT-Power cycle simulations were used to estimate the pumping work and operating conditions (e.g., intake temperature) required to reproduce the results presented in this work in the engine lab. Figure 2 shows the effect of injection timing on calculated cylinder pressure and heat release rate at 8% premixed gasoline. Advancing the injection timing resulted in a delay in combustion phasing. Since this result is unintuitive, the mixture preparation was investigated. Figure 3 shows contours of equivalence ratio at 8 BTDC for two cases, SOI-2 and SOI -4 ATDC. It can be seen that, near the onset of high temperature heat release, the SOI - 2 ATDC case has a region of significantly higher equivalence ratio than the SOI -4 ATDC case. This region is the result of the diesel direct-injection and also corresponds to a region of high diesel fuel concentration (i.e., high fuel reactivity). As the injection timing was advanced, the mixture became more homogeneous, with a peak equivalence ratio of ~.4. The low equivalence ratio extended the ignition delay of the diesel fuel and retarded the combustion phasing. Pressure [MPa] SOI -2 SOI -3 SOI -4 SOI -5 SOI Crank [ ATDC] 8% Gasoline SOI Advance Heat Release Rate [J/ ] Figure 2. Effect of injection timing on cylinder pressure and heat release rate for the 8% gasoline.

3 SOI -2 ATDC SOI -4 ATDC NOx [g/kgf] % Gasoline 7% Gasoline 75% Gasoline 8% Gasoline 85% Gasoline Figure 3. Equivalence ratio contours at 8 BTDC for two injection timings, SOI -2 and -4 ATDC. In these cases 8% of the total fuel is premixed gasoline at IVC. Notice that the equivalence ratio away from the diesel fuel jet is ~.3. This corresponds to the equivalence ratio of the premixed gasoline. Figure 4 shows the effect of injection timing and gasoline percentage on peak pressure rise rate and NOx emissions. It can be seen that the peak pressure rise rate (PRR) reaches a maximum at an SOI of -4 ATDC for cases with premixed gasoline less than 85%. For each of these cases, further advancing the injection timing past -4 ATDC caused a further delay in combustion phasing and reduced the peak pressure rise rate. Peak pressure rise rate also decreased with increasing gasoline percentage for each injection timing considered. NOx emissions decrease significantly as the injection timing is advanced. Figure 5 shows contours of temperature at 8 ATDC for cases with 8% premixed gasoline and diesel fuel injection timings of -2 and -4 ATDC. It can be seen that the higher equivalence ratio and advanced combustion phasing of the SOI -2 ATDC case resulted in significantly higher temperatures, thus explaining the increased NOx emissions with retarded injection timing. The results of the previous investigation were combined to generate a map of the operating range for dual-fuel PCCI combustion at this condition. Figure 6 shows the operating regime map for dual-fuel PCCI combustion at 5.5 bar IMEP and 23 rev/min. The numbers represent the net ISFC (with the pumping loop calculated from cycle simulations) at each injection timing and premixed gasoline percentage combination. The green shading corresponds to the stable low NOx low noise operating regime. The NOx and peak pressure rise rate limits were set to 1 g/kgf and 1 bar/deg, respectively. Peak PRR [bar/ ] Diesel SOI [ ATDC] Figure 4. Effect of gasoline percentage and diesel fuel SOI on peak pressure rise rate and NOx emissions. SOI -2 ATDC SOI -4 ATDC Figure 5. Temperature contours for cases with 8% premixed gasoline and diesel fuel injection timings of -2 and -4 ATDC. Figure 6. Operating regime map for dual-fuel PCCI combustion at 5.5 bar IMEP and 23 rev/min. The NOx limit is 1 g/kgf and the PRR limit is 1 bar/deg. The numbers indicate the net ISFC.

4 Figure 6 shows that the earliest injection timing investigated (i.e., SOI -6 ATDC) resulted in significantly improved fuel consumption compared to the other operating points. To understand the reduced fuel consumption with very advanced injection timing, the combustion inefficiencies were evaluated. Figure 7 shows the effect of injection timing on UHC and CO emissions for the 8% premixed gasoline cases. It can be seen that CO decreases with advancing injection timing, while UHC first shows an increase, reaching a maximum at an SOI of -4 ATDC, and then shows a steep decrease. To explain this trend, Figure 8 shows UHC contours at several times during the combustion process for cases with SOI s of -2, -4, and -6 ATDC. Initially hydrocarbons fill the combustion chamber due to the premixed charge of gasoline and direct injected diesel fuel. As discussed in the previous sections, the SOI -2 ATDC case ignites first with combustion initiating in the center of the piston bowl. At 2 ATDC, the SOI -4 ATDC case ignites in a similar location. The SOI -6 ATDC case follows with ignition also occurring in the center of the bowl at 4 ATDC. In each case, the premixed fuel and fragmented hydrocarbons created from low temperature reactions of the diesel fuel are then consumed as the reaction zone propagates outward from the ignition location. Late in the cycle (shown by the image at 25 ATDC), it can be seen that UHC is only present in the squish and near liner regions. Additionally, it can be seen that the extent of UHC into the combustion chamber (i.e., the distance between the wall and location of negligible UHC) is greatest for the SOI -4 ATDC case and smallest for the SOI -6 ATDC case. It appears that advancing the injection timing past -4 ATDC improves oxidation of SOI -2 ATDC SOI -4 ATDC SOI -6 ATDC Figure 8. Cut-planes colored by UHC mass fraction. The iso-surface shows a UHC mass fraction of.4. SOI -2 ATDC SOI -4 ATDC SOI -6 ATDC Figure 7. Effect of diesel SOI on UHC and CO emissions. Figure 9. Cut-planes colored by PRF number for cases with 8% premixed gasoline and diesel fuel injection timings of - 2, -4, and -6 ATDC.

5 UHC in the near wall region. To investigate the improved UHC oxidation in the near wall region as the injection timing is advanced beyond -4 ATDC, Figure 9 shows contours colored by PRF number (i.e., the volume percentage of iso-octane in the fuel blend). First comparing the SOI -4 ATDC and -6 ATDC cases, it can be seen that the PRF number in the squish region is much lower for the SOI -6 ATDC case. This is explained by the diesel fuel penetration. That is, the fuel penetrates furthest towards the liner for the SOI -6 ATDC and thus increases the fuel reactivity (i.e., lowers the PRF number) in the squish region. The increased fuel reactivity results in improved UHC oxidation in the near liner region, reduced engine-out UHC, and reduced fuel consumption. Fuel Reactivity vs. Equivalence Ratio Stratification The previous work has shown that the dual-fuel PCCI combustion strategy presented in this work is capable of achieving premixed operation with low rates of pressure rise without using EGR. In this strategy, a gradient in fuel reactivity, generated by the port-fuel-injection of gasoline and direct-injection of diesel fuel, is used to extend the combustion duration to reduce the rate of pressure rise. However, accompanying the fuel reactivity gradient is a gradient in equivalence ratio. Equivalence ratio stratification has been shown to extend HCCI combustion duration and allow higher load operation (e.g., Dec et al. [13]). Thus, it is of interest to understand the effects of both equivalence ratio stratification and fuel reactivity stratification on dual-fuel PCCI combustion. To separate these effects, a multi-zone model of the previously described combustion process was developed. The case with an SOI -6 ATDC and a 8% premixed gasoline from Table 2 was selected for further analysis. First, the spray event was simulated using CFD as previously described. The CFD simulation was stopped at 2 BTDC and the combustion chamber was binned by equivalence ratio. Figure 1 shows the spatial location of each equivalence ratio zone and Figure 11 shows the equivalence ratio, zone mass fraction, and zone PRF number distributions. Notice that, due to the small quantity of diesel fuel and early direct-injection timing, the range of equivalence ratios present at 2 BTDC is relatively small. Additionally, it can be seen that the largest portion of the combustion chamber (~45% by mass) is at the lowest equivalence ratio. This is expected since 8% of the total fuel (by mass) was premixed at IVC. Furthermore, notice the strong inverse correlation between equivalence ratio and PRF number. That is, the richest regions of the chamber are at the lowest PRF number (i.e., the highest concentration of diesel fuel). Therefore, when fuel reactivity Figure 1. Equivalence ratio at 2 BTDC for the case with an SOI -6 ATDC and 8% premixed gasoline from Table 2. stratification is added, the most reactive regions in the combustion chamber become more reactive. Simulations were run using the multi-zone model of CHEMKIN-Pro [14]. Five adiabatic zones were used and, to eliminate the complexity added by temperature stratification, each zone was initialized with the same temperature at intake valve closure. No mixing was considered between zones and each zone was allowed to expand or contract to maintain equal pressure across the combustion chamber. That is, the zones only interact through pressure work. Similar to the previous investigations, the ERC PRF mechanism developed by Ra et al. [7] was used for the chemistry calculations. The effects of equivalence ratio and fuel reactivity stratification were evaluated by running two simulations. First, to evaluate the effects of only equivalence ratio stratification, the multi-zone model was run with the equivalence ratio distribution shown in Figure 11; however, the PRF blend of each zone was set to 88 (i.e., the average PRF blend shown in Figure 11). Next, to evaluate the combined effects of equivalence ratio and PRF stratification, the PRF distribution presented in Figure 11 was applied to the existing equivalence ratio gradient. Note that, consistent with the first study, the global (average) PRF number is 88. Figure 12 shows the rate of heat release for each simulation. It can be seen that, due to the small range of equivalence ratios present (see Figure 11), when only equivalence ratio stratification is considered, each zone reacts nearly simultaneously. This does not imply that equivalence ratio stratification cannot be used to control combustion duration, but rather, for the dual-fuel strategy presented in this work, the early injection timing and small quantity of directinjected diesel fuel results in a relatively small gradient in equivalence ratio. Conversely, it can be seen that when a fuel reactivity gradient (i.e., PRF stratification) is added to the existing equivalence ratio gradient, the zones with lower PRF numbers (i.e., more reactive) ignite significantly earlier in the cycle. Although combustion initiates earlier

6 Figure 11. Equivalence ratio, mass fraction, and PRF of each CHEMKIN zone at 2 BTDC. in the cycle in the most reactive zones, the energy release between zones is staged such that the combustion duration is significantly increased. The primary purpose of adding stratification in fuel reactivity (or equivalence ratio) is to reduce the rate of pressure rise by extending the combusting duration. Figure 13 shows the calculated cylinder pressure for both cases. It can be seen that, when only equivalence ratio stratification is considered, the short combustion duration results in a rapid rate of pressure rise. However, when fuel reactivity stratification is added to the existing equivalence ratio stratification, the pressure rise rate is reduced by a factor of 3 even though combustion occurs slightly earlier in the cycle. Conclusions First, dual-fuel PCCI combustion was studied in a light-duty engine through a computational parametric study. Sweeps of injection timing and premixed gasoline percentage were evaluated and an operating regime map to achieve high efficiency low NOx low noise combustion was generated. Several conclusions can be drawn from this work: Advancing the injection timing resulted in retarded combustion phasing due to increased mixture homogeneity. That is, as the injection timing was Figure 12. Heat release rate for cases with and without a fuel reactivity gradient. Note that both cases have the equivalence ratio gradient shown in Figure 11 and the global (average) PRF number of both cases is 88. Figure 13. Calculated cylinder pressure for cases with and without a fuel reactivity gradient. Note that both cases have the equivalence ratio gradient shown in Figure 11 and the global (average) PRF number of both cases is 88.

7 advanced, the peak equivalence ratio prior to autoignition decreased significantly. The decrease in equivalence ratio corresponded to a decrease in the peak fuel reactivity and thus caused an increase in the ignition delay. The earliest injection timing investigated (i.e., SOI -6 ATDC) resulted in significantly improved fuel consumption compared to the other operating points. It was found that advancing the injection timing improved UHC oxidation in the near wall region and reduced fuel consumption. This finding shows the importance of appropriate fuel (reactivity) distribution [15]. Next, multi-zone modeling was used to isolate the effects of equivalence ratio and fuel reactivity stratification. It was found that fuel-reactivity stratification works in union with equivalence ratio stratification to extend the combustion duration. Additionally, it was found fuel reactivity stratification plays a larger role in the observed extended combustion duration than equivalence ratio stratification. References 1. Hanson, R. M., Kokjohn, S. L., Splitter, D. A., and Reitz, R. D. "An Experimental Investigation of Fuel Reactivity Controlled PCCI Combustion in a Heavy- Duty Engine", SAE , Kokjohn, S. L., Hanson, R. M., Splitter, D. A., and Reitz, R. D. "Experiments and Modeling of Dual Fuel HCCI and PCCI Combustion Using In-cylinder Fuel Blending", SAE , Amsden, A. A., Kiva-3v, release 2, Improvements to KIVA-3v", LA-UR , Abani, N., Kokjohn, S. L., Park, S. W., Bergin, M., Munnannur, A., Ning, W., Sun, Y., and Reitz, R. D. "An Improved Spray Model for Reducing Numerical Parameters Dependencies in Diesel Engine CFD Simulations", SAE , Abani, N., Munnannur, A., and Reitz, R. D., "Reduction of Numerical Parameter Dependencies in Diesel Spray Models", Journal of Engineering for Gas Turbines and Power, vol.13, no.3, Han, Z. and Reitz, R. D., "Turbulence Modeling of Internal Combustion Engines Using RNG k-ε models", Combustion Science and Technology, vol.16, Ra, Y. and Reitz, R. D. "A Reduced Chemical Kinetic Model for IC Engine Combustion Simulations with Primary Reference Fuels", Combustion and Flame, Vol. 155, pp , Ra, Y., Yun, J. E., and Reitz, R. D., "Numerical Parametric Study of Diesel Engine Operation with Gasoline", Combustion Science and Technology, vol.181, pp. 1-29, Kong, S.-C., Sun, Y., and Reitz, R. D., "Modeling DI- Diesel Spray Flame Lift-off, Sooting Tendency, and NOx Emissions Using Detailed Chemistry with a Phenomenological Soot Model", ASME Journal of Gas Turbines and Power, vol.129, pp , Hiroyasu, H. and Kadota, T., "Models for Combustion and Formation of Nitric Oxide and Soot in DI Diesel Engines", SAE 76129, Sun, Y., "Diesel Combustion Optimization and Emissions Reduction Using Adaptive Injection Strategies (AIS) with Improved Numerical Models", PhD Thesis in Mechanical Engineering, University of Wisconsin- Madison, Frenklach, M., Bowman, T., Smith, G. and Gardiner, B., Dec, J. and Sjoberg, M, Smoothing HCCI Heat- Release Rates Using Partial Fuel Stratification with Two-Stage Ignition Fuels, SAE , CHEMKIN-Pro, Reaction Design, Reitz, R.D., Hanson, R., Splitter, D. and Kokjohn, S.L, Engine Combustion Control via Fuel Reactivity Stratification, Feb 1, 21. University of Wisconsin WARF Patent Application P154US Contact Information Sage Kokjohn 15 Engineering Dr. ERB Room 111 Madison, WI USA Tel: kokjohn@wisc.edu