Modeling of Gaseous Fuel Injection and Combustion Process in a CNG Direct Injection Engine

Abstract

This paper describes a methodology to model the gaseous fuel injection process in a CNG (compressed natural gas) DI (direct injection) engine. Simulations were conducted using KIVA-3V Release 2 code by modifying the liquid fuel injection model to function as a gaseous fuel injection model. Until now, a very fine mesh smaller than the injector nozzle has been required to resolve the gas-jet inflow boundary. However, the gaseous sphere injection model simulates gaseous fuel injection using a coarse mesh. This model injects gaseous spheres as in liquid fuel injection and the gaseous spheres evaporate together without the latent heat of evaporation. Therefore, it does not require a very fine mesh and reduced calculation time. The fuel injection and combustion process was simulated using a three-dimensional engine mesh with four valves from IVO (intake valve open) to EVO (exhaust valve open). A poppet valve type injector was centrally mounted and the simulations were conducted under various injection timing and pressure conditions. The conventional RNG k-ε turbulence model tends to over predict gas jet diffusion radially, which results in shorter penetration. The modified RNG k-ε turbulence model reliably predicts the distribution of gaseous fuel as well. The modified RNG k-ε turbulence model was used only for the injection period. After completing the fuel injection, the conventional RNG k-ε turbulence model was used. The simulated penetration profiles agree well with the experimental penetration profiles although the early stage of injection was somewhat over-predicted. This limitation was considered to be an unavoidable aspect of the gas sphere injection methodology, which assumes that the gaseous fuel is injected as gas spheres and does not evaporate. Therefore, the simulated penetration profiles of the early stage are somewhat over-predicted compared with those of the experimental penetration. Combustion simulations were performed under various injection timings and injection pressures. Additionally, the experimental results of the in-cylinder pressure were compared with the calculated results for investigation of the combustion characteristics. The injection pressure did not significantly affect the IMEP (indicated mean effective pressure) results which changed little under the various injection pressures. However, the injection timing greatly affected the IMEP results. The IMEP was the lowest when the injection timing was 270° BTDC. This was caused by the turbulence kinetic energy. It is thought that the momentum of the intake flow and gaseous fuel injection flow offset each other when the fuel is injected at 270° BTDC. The simulation was performed at a 4000 rpm (revolutions per minute) engine speed and in the WOT (wide open throttle) condition. The proper spark timing for MBT (maximum brake torque) is 45° BTDC at a 4000 rpm engine speed and in the WOT condition. Additionally, the in-cylinder peak pressures were positioned between 5° BTDC and 10° BTDC in the MBT condition. At a 1700 rpm engine speed, the IMEP was the lowest when the injection timing was 270° BTDC. This point was shifted to 220° BTDC at a 4000 rpm engine speed.