No abstract

In this work, we have studied cycle-to-cycle variation in a spark-ignited engine using large-eddy simulation in conjunction with the G-equation combustion model. A single cylinder of a four-cylinder port-fueled spark-ignited engine was simulated. A total of 49 consecutive full cycles were computed. The operating condition studied in this work is stoichiometric and stable and represents a load of 16 bar brake mean effective pressure and an engine speed of 2500 r/min. The computational fluid dynamics simulation shows good agreement in terms of in-cylinder pressure prediction with respect to the experiments and is also able to capture the range of cycle-to-cycle variation observed in experiments. Furthermore, neither the simulation nor the experiments show any distinguishable pattern in the sequence of high and low cycles. We numerically decoupled the effects of variations in equivalence ratio fields and velocity fields to isolate the effects of differences in the velocity field and differences in the equivalence ratio field on flame development and propagation. Based on this study, we inferred that for this engine, under the operating conditions studied, the differences in burn rates can be attributed to the differences in the velocity flow-field in the region around the spark gap during ignition. We then performed an analysis to identify the correlation between peak cylinder pressure and flame topologies over all the simulated cycles. We found that high cycles (higher peak cylinder pressure values) are strongly correlated to flatter flame volume shapes (flattened in the piston-to-head direction) and volumes that are more symmetric about the ignition axis. In addition, these kinds of flame volumes were found to correlate well with lower values of prior-to-ignition velocity going from the intake to the exhaust side (mean flow caused by tumble) at the spark and also higher values of prior-to-ignition velocity in the piston-to-head direction.

Soot production mechanism in multiple injections is complex since it involves its dependence on turbulent interactions of constituting injections and their combustion progress. A concise study was performed in a constant-volume combustion vessel by considering a double injection scheme of 0.3 ms pilot injection, 0.5 ms dwell time and 1.2 ms main injection (nomenclature: 0.3/0.5/1.2 ms) with n-dodecane as fuel and replicating the thermodynamic operating condition of a compression ignition (CI) engine. Experimental ambient temperature variations of 900 K and 800 K were performed at 15% ambient oxygen level. Simultaneous planar laser-induced fluorescence (PLIF) of formaldehyde and schlieren imaging techniques were employed to analyze the ignition and flame characteristics experimentally. These studies revealed almost similar heat release rates for a double injection at 900 K and 800 K ambient gas temperatures due to combustion of a longer main injection which is enhanced by pilot combustion event. A lower soot production for 800 K ambient condition over 900 K case was observed, which was concluded to be due to its higher lift-off length which would allow for a leaner combustion of fuel-air mixtures. Numerical simulations were performed using a Large Eddy Simulation (LES) approach by extensively validating the 900 K double injection condition with respect to non-reacting vapor penetration profiles of both injections, reacting jet heat release rate and spatial as well as temporal (qualitative) soot production. As part of LES work, a dwell time variation of 0.65 ms (0.3/0.65/1.2 ms) was performed to reveal the sensitivity of soot production to variations in dwell time. It was observed numerically that marginally higher quasi-steady lift-off length of the 0.3/0.65/1.2 ms injection causes increased entrainment of surrounding oxygen into the flame region. This leads to combustion of slightly leaner fuel-air mixture and hence relatively less soot when compared to a 0.3/0.5/1.2 ms injection.

Studies are performed in a constant volume preburn type combustion vessel over a range of ambient temperatures (750 K, 800 K, and 900 K) at constant density (22.8 kg/m3) with 15% O2 by volume in the ambient at 1200 bar (n-dodecane) fuel injection pressure. The influence of the pilot (first) spray flame on the ignition and combustion characteristics of the main (second) injection is investigated while varying injection pressure, dwell time, and injection strategy. Simultaneous schlieren (with soot luminosity imaging) and 355 nm planar laser-induced fluorescence (PLIF) imaging for formaldehyde (CH2O) and polycyclic aromatic hydrocarbons (PAH) visualization was performed. At both 900 K and 800 K ambient, main injection exhibits a reduction in ignition delay (ID) by a factor of 2 over their respective pilots. For the ambient temperature condition of 750 K, reducing injection pressure from 1500 bar to 1200 bar causes a significant increase in ignition delay (by ∼0.8 ms), which was attributed to the influence of injection pressure on spray-mixing and early development of cool flame. Also, at 750 K ambient condition, multiple injection schedule having two 0.5 ms injections separated by a 0.5 ms dwell was found to have a shorter ignition delay than a single 0.5 ms injection. Studies carried at an 800 K ambient show that by increasing the dwell time, main interaction with pilot reactive intermediates can be controlled to avoid an early rich ignition of the main spray and to reduce soot precursors.

In this work, we have applied a machine learning (ML) technique to provide insights into the causes of cycle-to-cycle variation (CCV) in a gasoline spark-ignited (SI) engine. The analysis was performed on a set of large eddy simulation (LES) calculations of a single cylinder of a four-cylinder port-fueled SI engine. The operating condition was stoichiometric, without significant knock, at a load of 16 bar brake mean effective pressure (BMEP), at an engine speed of 2500 rpm. A total of 123 cycles was simulated. Of these, 49 were run in sequence, while 74 were run in parallel. For the parallel approach, each cycle is initialized with its own synthetic turbulent field to generate CCV, as a part of another work performed by us. In this work, we used 3D information from all 123 cycles to compute flame topology and pre-ignition flow-field metrics. We then evaluated correlations between these metrics and peak cylinder pressure (PCP) employing an ML technique called random forest. The computed metrics form the inputs to the random forest model, and PCP is the output. This model captures the effect of all inputs, as well as interactions between them owing to its decision-tree structure. The goal of this work is to demonstrate (as a first step) that ML models can implicitly learn complex relationships between the pre-ignition flow-fields, the flame shapes, and the eventual outcome of the cycle (whether a cycle will be a high or a low cycle).

No abstract

As an alternative fuel to diesel in compression-ignition (CI) engines, dimethyl ether (DME) has gained interest in combustion research due to its high cetane number for fast ignition and ultra-low emission of particulate matter. In this study, ignition and important intermediate species including formaldehyde are experimentally investigated in a constant-volume combustion vessel facility that includes a fuel injection system for spray-like behavior of liquid DME. Experiments of different oxygen concentrations simulating various levels of exhaust-gas recirculation (EGR) were performed to study its corresponding effect on the flame structure and emissions from DME combustion. Results from the experiment were then used to validate a 3-D CFD simulation using a detailed chemistry solver. Different stages of ignition, characterized by temperature profile and certain species (e.g., CH 2 O for cool flame) after start of injection, are provided to conceptualize the DME combustion process under the effect of low-to-high oxygen ambient gas concentrations. Both simulations and experiments showed that there is a supporting link between CH 2 O formation and low-temperature combustion prior to diffusion-controlled flame uniquely for DME. By studying the temperature and equivalence ratio dependence of the reacting spray at different O 2 levels, different stages of ignition along with the formation of CH 2 O suggested that the start of the depletion of CH 2 O can be used as an ignition indicator.

No abstract