Quasi-Two-Zone Modeling of Diesel Ignition Delay in Pilot-Ignited Partially Premixed Low Temperature Natural Gas Combustion

Abstract

Listen

Translate article

This paper presents simulated ignition delay (ID) results for diesel ignition in a pilot-ignited partially premixed, low temperature natural gas (NG) combustion engine. Lean premixed low temperature NG combustion was achieved using small pilot diesel sprays (2–3% of total fuel energy) injected over a range of injection timings (BOIs ∼ 20°–60° BTDC). Modeling IDs at advanced BOIs (50°–60° BTDC) presented unique challenges. In this study a single-component droplet evaporation model was used in conjunction with a modified version of the Shell autoignition (SAI) model to obtain ID predictions of pilot diesel over the range of BOIs (20°-60° BTDC). A detailed uncertainty analysis of several model parameters revealed that Aq and Eq, which affect chain initiation reactions, were the most important parameters (among a few others) for predicting IDs at very lean equivalence ratios. The ID model was validated (within ± 10 percent error) against experimentally measured IDs from a single-cylinder engine at 1700 rpm, BMEP = 6 bar, and intake manifold temperature (Tin) of 75°C. For BOIs close to TDC (e.g., 20° BTDC), the contribution of diesel evaporation times (Δθevap) and droplet diameters to predicted IDs were more significant compared to advanced BOIs (e.g., 60° BTDC). Increasing Tin (the most sensitive experimental input variable affecting predicted IDs), led to a reduction in both the physical and chemical components of ID. Hot EGR led to shorter predicted and measured IDs over the range of BOIs, except 20° BTDC. In general, the thermal effects of hot EGR were found to be more pronounced than either dilution or chemical effects for most BOIs. Finally, uncertainty analysis results also indicated that ID predictions were most sensitive to model parameters AP3, Aq, and Af1, and Eq, which affected chain initiation and propagation reactions and also contributed the most to overall uncertainties in IDs.