Understanding and detecting misfire in an HCCI engine fuelled with ethanol

• Published in: Applied energy 2013-08, Vol.108, p.24-33
• Main Authors: Bahri, Bahram, Aziz, Azhar Abdul, Shahbakhti, Mahdi, Muhamad Said, Mohd Farid
• Format: Article
• Language: English
• Subjects: Applied sciences; Artificial neural network; Combustion; Cylinders; emissions; Energy; Energy. Thermal use of fuels; Engines; Engines and turbines; Equipments for energy generation and conversion: thermal, electrical, mechanical energy, etc; Ethanol; Ethanol combustion; Ethyl alcohol; Exact sciences and technology; fuels; HCCI; heat; Learning theory; Mathematical models; Misfire; Neural networks; pressure
• ISSN: 0306-2619; 1872-9118
• DOI: 10.1016/j.apenergy.2013.03.004

•Investigates the effects of misfire on HCCI engine emissions and operating regions.•Examines major engine combustion parameters for misfire detection.•Proposes and validates a new model to detect misfire in HCCI engines (the first work on misfire detection in HCCI literature). Homogeneous charge compression ignition (HCCI) with ethanol as a renewable fuel offers a promising solution to tackle some of major challenges before realizing green powertrains. Misfire limits HCCI engine operation and can damage exhaust after treatment system. This article aims to understand the effect of misfire on the operation of an ethanol fuelled HCCI engine. The experimental data from a 0.3 liter converted-diesel HCCI engine was used to investigate the effect of misfire on exhaust emissions, in-cylinder pressure trace, indicated mean effective pressure (IMEP), heat release and combustion phasing metrics. It was found that variation of combustion parameters such as start of combustion (SOC) and crank angle of maximum in-cylinder pressure are not effective parameters for HCCI misfire detection. However, there is a strong correlation between the occurrence of misfire and variation of cylinder pressure at 5, 10, 15 and 20 CAD aTDC. These experimental findings were then used to design an artificial neural network (ANN) model to detect misfire in the HCCI engine. The model was tested on the experimental data for a mix of 7800 normal and misfire cycles. The results indicated that the ANN misfire detection (AMD) model can detect HCCI misfire with 100% accuracy. In addition, the AMD model was found to be capable of successfully detecting the onset of the transition from normal to misfire operation region.