Cycle-to-cycle variability in spark-assisted compression ignition engines near optimal mean combustion phasing

Abstract

Stoichiometric spark-assisted compression ignition (SACI) combustion with exhaust gas recirculation (EGR) dilution has demonstrated higher part-load thermal efficiencies compared to spark-ignited engines, while maintaining ultra-low tailpipe emissions. However, SACI is often characterized by high cyclic variability in heat release or torque output, which poses a challenge to its implementation in light-duty vehicles. This paper presents an experimental investigation of cyclic variability in stoichiometric SACI combustion under EGR-dilute conditions, while maintaining mean combustion phasing (θ50) near optimal timing for thermal efficiency. The present work focuses on SACI conditions where the flame-based heat release fraction is between approximately 10% and 40% of the overall heat release, and therefore contributes significantly to the combustion process. For the SACI conditions examined, the variability in θ50 was driven by variability in autoignition timing, which in turn correlated with the start of measurable heat release (θ02). High variability in θ50 caused unstable work output due to very late combustion with poor or no end-gas autoignition. The use of a high ignition energy dual-coil offset ignition system had negligible impact in reducing θ50 variability. Analysis of experimental data from close to 1000 operating conditions showed that the magnitude of θ50 variability correlates with the flame-based heat release fraction ([Formula: see text]), for a large range of intake pressures, spark timings and exhaust gas recirculation levels. For the SACI conditions examined, combustion phasing variability was largely determined by flame-based combustion and particularly the initial flame formation (θIGN-02), and minimally by the end-gas autoignition heat release. The analysis also demonstrated that θ50 variability is amplified as the contribution of the flame to the overall heat release increases.