A thermodynamic model for homogeneous charge compression ignition combustion with recompression valve events and direct injection: Part i - Adiabatic core ignition model

Abstract

This two-part article presents a model for boosted and moderately stratified homogeneous charge compression ignition combustion for use in thermodynamic engine cycle simulations. The model consists of two components: one an ignition model for the prediction of auto-ignition onset and the other an empirical combustion rate model. This article focuses on the development and validation of the homogeneous charge compression ignition model for use under a broad range of operating conditions. Using computational fluid dynamics simulations of the negative valve overlap valve events typical of homogeneous charge compression ignition operation, it is shown that there is no noticeable reaction progress from low-temperature heat release, and that ignition is within the high-temperature regime (T > 1000 K), starting within the highest temperature cells of the computational fluid dynamics domain. Additional parametric sweeps from the computational fluid dynamics simulations, including sweeps of speed, load, intake manifold pressures and temperature, dilution level and valve and direct injection timings, showed that the assumption of a homogeneous charge (equivalence ratio and residuals) is appropriate for ignition modelling under the conditions studied, considering the strong sensitivity of ignition timing to temperature and its weak compositional dependence. Use of the adiabatic core temperature predicted from the adiabatic core model resulted in temperatures within ±1% of the peak temperatures of the computational fluid dynamics domain near the time of ignition. Thus, the adiabatic core temperature can be used within an auto-ignition integral as a simple and effective method for estimating the onset of homogeneous charge compression ignition auto-ignition. The ignition model is then validated with an experimental 92.6 anti-knock index gasoline-fuelled homogeneous charge compression ignition dataset consisting of 290 data points covering a wide range of operating conditions. The tuned ignition model predictions of 50 have a root mean square error of 1.7° crank angle and R2 = 0.63 compared to the experiments.