Investigation on the knock characteristics in a gasoline direct-injection engine port-injected with water-methanol blends

Combustion knock is one of the key factors limiting the performance of spark-ignition engines, and thus great emphasis is placed on developing methods for effective and efficient knock suppression. The majority of works in this area are focused on the influence of these methods on the engine parameters, while maintaining the same “knock level” represented by a certain statistical parameter of knock combustion. The knock level defined as the mean or other statistical parameter of a stochastic process does not represent its characteristics. Thus, it is not evaluated on a sufficient level to show how different solutions for the knock reduction affect the knock probability distribution; nor is it known whether a universal treatment to characterise this influence can be developed. In this work we aim to fill this gap, and by the analysis of the knock combustion in a single-cylinder direct-injection gasoline engine with co-injection of water, methanol, and their blends, conclude on the effects of different additives on the probability distribution of in-cylinder pressure-related parameters. We also applied a method to normalise the knock phenomenon by a model representation of its probability distribution. The analysis was performed for knock-limited spark advance conditions for different amounts of co-injected water, methanol, and their blends. The method to normalise the knock distribution was based on the log-normal probability distribution, and it was successfully applied to obtain the knock characteristics at a quantified characteristic limit of knock, which was impossible to precisely meet directly in the tests. The results revealed that depending on the co-injected additive and its amount, the knock distribution undergoes slight but systematic changes, in terms of distribution skewness and kurtosis. At the selected knock limit, the injection of methanol resulted in knock characteristics more skewed to the knock limit comparing to the baseline case. Water addition resulted in opposite trends. Blending water and methanol gave a similar knock peak-to-peak distribution as in the gasoline-only case