主催: The Japan Society of Mechanical Engineers
会議名: The Ninth International Conference on Modeling and Diagnostics for Advanced Engine Systems (COMODIA 2017)
開催日: 2017/07/25 - 2017/07/28
Cycle simulation plays increasingly important roles on either conceptual designs or operating parametric optimization in development of internal combustion engines. Knock is one of the major constraints in improving spark-ignition (SI) engines, and determination of the knock limited spark advance (KLSA) is very crucial for a successful simulation. In the experimental calibration of a spark ignition engine, KLSA is usually defined as a spark angle where slight knock may occur with a statistical intensity (i.e. maximum amplitude of pressure oscillation, MAPO) distribution. However, the state-of-the-art knock models are mostly limited to prediction of onset of auto-ignition of end gas and fail to predict the stochastic nature of knock, causing deviation of the predicted KLSA from the experimental result. In this paper, a phenomenological model for knock onset, taking multiple variables including the pressure, temperature, excess air ratio and EGR ratio into accounts, is firstly formulated and validated against the experimental results of an SI engine. Based on statistical analysis of the experimental data with the spark sweep around the KLSA, the knock intensities are found to follow the lognormal distribution, and then a knock factor (KF) using the ratio of knock intensities corresponding to 95% and 25% probabilities in the cumulative lognormal function is proposed to determine the KLSA for the engine cycle simulation. To obtain the expectation and standard deviation of the lognormal distribution function, a two-factor model, which considers several factors, including the hot spot size, energy density and heat release rate in the end gas, is developed. Finally, the newly developed knock model is evaluated and better performance is testified, compared to the other knock models used in the engine cycle simulation.