The Detection and Quantification of Knock in Spark Ignition Engines 932759

A brief review is included of previous work aimed at quantifying the knock intensity from cylinder pressure measurements. This is used to identify some of the methods used in the current study. Digital signal processing techniques are also discussed, since their application to non-repetitive truncated signals can lead to results that are dependent on the techniques used. These problems are illustrated with some examples of windowing, and non-linear phase shift filters.

A good correlation is demonstrated between knock severity indices calculated with energy methods in the time domain and the frequency domain. It is argued that it is easier to implement such knock indices in the lime domain. Use has also been made of mass fraction burn calculations in conjunction with data for the onset of knock, for data recorded simultaneously by two different pressure transducers. Different values of the unburnt mass fraction at knock onset are obtained from each transducer, and it has been demonstrated that this is due to slight differences in the calculation of the mass fraction burnt from each transducer and the difficulty of defining the onset of knock. Finally, it is shown that energy methods give results that cross-correlate well with the unburnt mass fraction on an individual cycle basis with light knock, but the cross-correlation deteriorates as the knock intensity increases. However, a good cross-correlation has been found between the mean values of knock intensity with the mean values of the unburnt mass fraction at the onset of knock. . Since the energy methods are simpler to implement it is argued that these are an appropriate means of quantifying knock.