Effect of Liquid Break-Up Model Selection on Simulated Diesel Spray and Combustion Characteristics

Abstract

Accurate modelling for spray vapour fields is critical to enable adequate predictions of spray ignition and combustion characteristics of non-premixed reacting diesel sprays. Spray vapour characteristics are in turn controlled by liquid atomization and the KH-RT liquid jet break-up model is regularly used to predict this: with the KH model used for predicting primary break-up given its definition as a surface wave growth model, and the RT model used for predicting secondary break-up due to it being a drag based, stripping model. This paper investigates how the alteration of the switching position of the KH and RT sub-models within the KH-RT model impacts the resulting vapour field and ignition characteristics. The combustion prediction is handled by the implementation of a 54 species, 269 reaction skeletal mechanism utilising a Well Stirred Reactor model within the Star-CD CFD code. Following on from the derivation and implementation of an Ohnesorge based switch between the KH and RT sub-models, this model is now tested in igniting cases for an ndodecane fuelled single holed injection representing the ECN "Spray A" condition, and is compared to the baseline Reitz-Diwakar model. Differences in flame behaviour, particularly within the temperature distribution, are seen and directly traced from the effect of liquid break-up position and model selection, through atomised droplet size distribution and mixture fraction distribution. Different criteria for judging the ignition delays and lift-off-lengths are compared, with all methods predicting very similar results for both models. The KH and RT sub-models are also tested against each other, with heavy instabilities seen when the RT model is solely applied to the spray. This correlates with the instabilities shown in the vapour fields, suggesting the enabling of the RT model near-nozzle is to be avoided.