Experimental and numerical study of a bio-methane heavy-duty engine equipped with a pre-chamber ignition system: a focus on flame kernel development and combustion process under lean conditions

In this work, in-depth experimental and numerical analyses are carried out on a single-cylinder bio-methane heavy-duty spark ignition engine, mounting a passive pre-chamber, to investigate the flame kernel formation, the combustion process and the overall performance. In the first stage, an extensive experimental activity is performed on the examined engine. Measurements are carried out in several engine operating points, exploring lean mixture conditions up to the lean burn limit and recording global engine parameters, combustion indicators, in-cylinder pressure traces and emissions. A 0D/1D engine model, integrated with user-coded phenomenological sub-models of flame kernel, turbulent combustion and emissions, is developed and combustion is properly tuned referring to the experimental stoichiometric points. A physics-based correlation of flame kernel duration for air/bio-methane charge is identified to forecast the kernel formation time at the spark event, which is a challenging task in presence of diluted mixtures and variable in-cylinder thermodynamic conditions. Flame kernel formation time predicted by the correlation considers the dependencies on pressure, unburned temperature, equivalence ratio and residual gas content of the air/fuel mixture. The proposed numerical approach demonstrates that a reliable prediction of the kernel duration remarkably improves the simulation of burn rates and pressure traces in the main chamber, especially under high mixture dilutions. The 0D/1D model accounting for the kernel formation satisfactorily replicates the main engine performance, including net Indicated mean effective pressure and Indicated specific fuel consumption with maximum absolute errors of 1.8% and 4%, respectively; main gaseous emissions (nitrogen oxides, carbon monoxide and unburned hydrocarbons) are reproduced with an overall acceptable accuracy; greater errors are observed only at high load, with absolute experimental/numerical deviations of about 200 ppm for nitrogen oxides and slightly greater than 400 ppm for carbon monoxide and unburned hydrocarbons.