The Effect of Droplet Temperature Model Choice on Gasoline Droplet and Spray Simulation

Prediction of spray droplet temperature is routinely performed within automotive engineering CFD simulation. Important applications include fuel spray modelling for in-cylinder combustion studies and urea-water spray modelling for SCR aftertreatment studies. Transient droplet surface temperature strongly influences droplet and spray evaporation rate and subsequent system performance. In this paper, different droplet temperature models are presented and compared for both single droplet modelling and a single-hole Lagrangian spray CFD simulation in OpenFOAM. The aim is to determine the complexity of implementation of the models and their effect on prediction accuracy for droplet surface temperature, evaporation rate and lifetime. A non-ideal multi-component droplet evaporation model using UNIFAC and NRTL for activity coefficient calculation is used. Model gasoline fuels containing varying quantities of ethanol are studied at conditions appropriate to gasoline direct injection. Three droplet temperature models are compared: a lumped thermal model typically employed in CFD packages; an effective thermal conductivity model (ETC) using the full heat equation; and an ETC model tracking the surface and centre temperature only. For droplet simulations, discrepancies of up to 13% in calculated droplet lifetime are found between the lumped thermal model and the effective conductivity models. The greatest discrepancy in lifetime is observed at conditions and compositions where rapid liquid thermal transients are encountered. This effect is characterized for the first time using the wet-bulb temperature, initial liquid temperature and fuel ethanol content, to guide modellers to an appropriate temperature model for the level of accuracy desired. The effect of droplet temperature modelling in spray CFD simulation is smaller, as measured by maximum changes in spray penetration (2%), evaporation rate (2%) and Sauter Mean Diameter (1.2%). A non-uniform temperature model is required when significant transient temperature change is expected, near-nozzle spray is studied or sub-models such as droplet collision and break-up are not in use.