Experimental Investigation of Cyclic Variation of Heat Release Dynamics of HCCI Combustion Engine

Homogenous charge compression ignition (HCCI) combustion emerged as a potential technique for reducing automotive pollution. Controlling the combustion timing at different engine operating conditions is one of the major challenges for the commercial application of HCCI combustion engines. To control HCCI ignition timing, it is often necessary to know the characteristics of HCCI cyclic variations. In this study, cyclic combustion variations in an HCCI engine are analyzed. Combustion stability and cycle-to-cycle variations of HCCI combustion parameters were investigated on a modified four-stroke diesel engine. The experiments were conducted by varying intake air temperatures and relative air-fuel ratios at constant engine speed. In the steady-state engine operating condition, in-cylinder pressure signals of 2000 consecutive engine combustion cycles are acquired for each test condition. From this large volume of experimental data collected, cyclic variations of various combustion parameters were analyzed. In this investigation, cycle-based time series analysis of total heat release is performed by constructing a recurrence plot for different air-fuel ratios and intake air temperatures. In this study, probability plots are used to investigate the distribution of combustion parameters. The results show that the qualitative change in combustion can be easily related to the patterns in recurrence plots. The deterministic nature exists in heat release time series for all operating conditions, which strengthens with the increase in values of relative air-fuel ratio and intake air temperature. The interactions between heat release and P_max time series are also explored by using cross recurrence plots. These interactions become stronger when HCCI engine operation shifts towards an unstable regime. The recurrence rate calculated on the diagonal basis is found to be a robust measure to capture the interactions between heat release and the P_max time series.