Evaluation of CFD sub-models for in-cylinder light-duty diesel engine simulation

Abstract

An evaluation of the Computational Fluid Dynamics (CFD) sub-models for in-cylinder diesel engine combustion and emission simulations was conducted using an integrated numerical model from commercial CFD software FLUENT 6.3.26. As these simulations are sensitive to user-defined empirical parameters, the main aim of this investigation is to obtain an adjusted set of model parameters in order to achieve realistic results with the current version of FLUENT. Validation of simulation results is based on matching parallel experimental data to gauge the accuracy and applicability of each sub-model used. A Lagrangian Discrete Phase Model is employed to simulate spray atomisation/breakup process. Wave model is used, and the corresponding breakup time constant value, B1 that equals to 20 produces the most accurate results. Other sub-models such as drop distortion and dynamic drag, spray wall impingement and wall film with rebounding sliding and break-up, two-way turbulence coupling, collision and coalescences are integrated to model the dynamics of fuel spray. RNG k-ε is best suited for RANS approach to capture in-cylinder turbulent flow condition. Non-premixed combustion model which adopts the PDF approach provides good prediction of the in-cylinder diesel combustion process. The rich flammability limit of 0.1 produces good agreement with the experimental data. Extended Zeldovich, Fenimore and Turbulence-Interaction are incorporated to model NOx generation. One-step Khan and Greeves model is used for soot formation and oxidation prediction. Both soot and NOx models are parametrically calibrated to give reasonable predictions of the experimental measurements.