The characteristics of combustion knock metrics over a number of engine cycles can be an essential reference for knock detection and control in internal combustion engines. In a Spark-Ignition (SI) engine, the stochastic nature of combustion knock has been shown to follow a log-normal distribution. However, this has been derived from experiments done with gasoline only and applicability of log-normal distribution to dual-fuel combustion knock has not been explored. To evaluate the effectiveness and accuracy of log-normal distributed knock model for methane-gasoline blended fuel, a sweep of methane-gasoline blend ratio was conducted at two different engine speeds.
Experimental investigation was conducted on a single cylinder prototype SI engine equipped with two fuel systems: a direct injection (DI) system for gasoline and a port fuel injection (PFI) system for methane. The experiments were conducted at 1500 rpm and 2000 rpm, 12.0 bar net indicated mean effective pressure wherein the engine was boosted using compressed air.
E10 gasoline and methane were used in this study. The results from blending two fuels show that the log-normal distribution provides a good fit to the measured distribution and captures the knock characteristics. The independency of log-normal distribution to the knock distribution at different spark timings was examined. The distribution parameters (log normal mean (μ) and standard deviation (σ)) show a linear correlation with the spark timing from knock borderline (BD) to 1.75° crank angle retarded. A μ and σ fit based log-normal (calculation-based log-normal) distribution model was proposed. The coefficients of multiple determination (CoMD) for the calculation-based log-normal model are all above 0.8 over the tested conditions. The validation of calculation-based log-normal was also conducted for all blending ratios and speeds.