This content is not included in your SAE MOBILUS subscription, or you are not logged in.

Knock Detection Method for Dual-Fuel Compression Ignition Engines Based on Block Vibration Analysis

Journal Article
03-14-02-0012
ISSN: 1946-3936, e-ISSN: 1946-3944
Published January 11, 2021 by SAE International in United States
Knock Detection Method for Dual-Fuel Compression Ignition Engines Based on Block Vibration Analysis
Citation: Rosas, M. and Amador, G., "Knock Detection Method for Dual-Fuel Compression Ignition Engines Based on Block Vibration Analysis," SAE Int. J. Engines 14(2):199-209, 2021, https://doi.org/10.4271/03-14-02-0012.
Language: English

Abstract:

A new knock detection method based on block vibration analysis, specially developed for dual-fuel compression ignition (CI) engines, is presented in this work. Experimental tests were carried out in a four-cylinder CI engine at full and 60% load, running at 2000, 2500, and 3200 rpm with different amounts of hydrogen and liquefied petroleum gas (LPG) injected in the air inlet hose. Fuel flow was increased in approximately 10% energy share steps until knock was detected for both fuels. For hydrogen, the maximum substitutions at full and 60% load were 38% and 54%, respectively, whereas for LPG were 57% and 63%, respectively. The component of the block vibration signal that is sensitive to knock was determined by studying the block’s resonant frequency, the influence of valve closing impacts, and comparing the block vibration recorded with knocking and non-knocking combustion. To quantify the knock intensity of a combustion cycle, four fast-computing metrics were tested, selecting the maximum amplitude of filtered vibration for knock detection since it was the least sensitive to crankshaft speed. Two knock indexes for knock evaluation were compared, concluding that the pondered deviation from the reference index, proposed in this work, has a better performance. The knock threshold was achieved when the knock index was greater than 5 regardless of the substitute fuel, crankshaft speed, and engine load. Finally, the method was optimized for real-time knock detection.