Study of fuel injection and mixture formation for a gasoline direct injection engine

Abstract

Future requirements for lower automotive emissions have lead to the development of new internal combustion (IC) engine technologies. Gasoline Direct Injection (GDI), for example, is one of these promising new IC engine concepts. It offers the opportunity of increased efficiency through unthrottled operation. However, the realisation of this concept is critically dependent on the in-cylinder mixture formation, especially in the late injection/lean operation mode. Ideally, this would require a precise stratification of the in-cylinder fuel-air mixture in 3 distinct zones: an ignitable pocket located at the spark plug, surrounded by a stoichiometric mixture of fuel and air, encompassed by air. To enable this stratification, the GDI concept utilises advanced injector technology. Phase Doppler Anemometry (PDA), Planar Laser-Induced Fluorescence (PLIF) and the combination of PLIF and Mie scattering in the Laser-Sheet Dropsizing (LSD) technique, have been applied to sprays in the past to obtain dropsize information and study the mixture formation process. These new GDI sprays are denser, their droplet sizes are smaller and they evaporate faster, and as such, place us at the limit of the validity of these measurements techniques. The diagnostics were applied to a GDI spray in a pressure vessel for realistic in-cylinder conditions, ranging from supercooled to superheated environments. Tracer evaporation issues in the PLIF technique were resolved by using a dual tracer system. The study showed that the LSD technique provided good quantitative data in low evaporation regimes. In highly evaporating regimes, the technique still gave reliable dropsize data for the early stages of the injection, but was limited afterwards by vapour-phase contribution to the fluorescence signal. Variations between PDA data and LSD results also suggested a deviation of the Mie scattering signal from the assumed d2 dependence. This was further investigated and was found to be true for small droplets (d/?. <0.2). This source of error might be improved by using a different observation angle. High density seriously compromises the accuracy of PDA, whilst its effect through multiple scattering is of second order for the LSD technique. In low evaporating regimes, LSD has the overall advantage of being a 2-D measurement technique, and will yield data with a maximum error of 30% in dense parts of the spray where PDA data is totally unreliable. If the spray evaporates quickly, PLIF by itself is an appropriate tool for following the air-fuel mixture, because short droplet lifetimes limit the 2-phase flow behaviour of the spray. Particle Image Velocimetry (PIV), the LSD technique and equivalence ratio LIF measurements were applied to a BMW single cylinder optical GDI engine. The early injection operation showed no particular issues. However, the results obtained in the late injection highlighted the poor mixing and inappropriate stratification.