Spray combustion with multiple-injection in modern engine conditions

Sammanfattning: Combustion of fuel in diesel engines emits substances harmful to the environment such as soot. These emissions can be reduced by either in-cylinder treatments or after-treatments. One of the common in-cylinder treatments is multiple-injection, which divides a single fuel injection to multiple smaller injections. There are many open questions on the physical processes of the ignition, combustion and emissions of diesel spray flame with multiple injections. The current PhD project aims at studying these processes using large-eddy simulations (LES) and strives to answer some of the open questions. To develop a fast and robust LES tool for this study, a new method is formulated for spray combustion simulation. This method is developed based on the flamelet-generated manifold (FGM) method and the Eulerian stochastic fields (ESF) method. The new ESF/FGM method relaxes some of the substantial assumptions in conventional FGM, while it still keeps the computational costs at a reasonable level for engineering applications. Additionally in this work, a new reaction progress variable for FGM models is proposed by using local oxygen consumption, and the advantages and limitations of this progress variable are explored. Spray-A from Engine Combustion Network (ECN) which is designed to mimic modern engine conditions is chosen as the baseline case for simulations. In this case, liquid n-dodecane, which is a diesel surrogate, is injected into a high-pressure constant-volume vessel. The comparison of simulation results with experimental measurements shows that the ESF/FGM method with the new progress variable can predict the spray combustion characteristics such as ignition delay time, ignition location, lift-off length, pressure rise and thermochemical structure of the spray flame, accurately. After validation of simulation results against experimental measurements, the new ESF/FGM and other available turbulence-combustion simulation tools are applied to simulate multiple-injection spray combustion. Different multiple-injection strategies are investigated by systematically changing the injection timing. The effects of applying each strategy on the ignition, combustion, mixing and emissions are investigated. The results show that in split-injection and post-injection strategies the major physical reason for reduction of soot is better air entrainment and lower local equivalence ratio. It is shown that increasing the dwell time and retarding it toward the end of injection can enhance this effect. On the contrary, for the pre-injection strategies, shortening the ignition delay time of the main injection reduces its pre-mixing and increases its soot formation. In these strategies, the high-temperature region from the pre-injection combustion can increase soot oxidation of the main injection fuel, only if this region is not cooled down as a result of air entrainment during dwell time. Therefore, in such cases shortening the dwell time decreases net soot emissions.