Development and Assessment of Large Eddy Simulation Methodology for Internal Combustion Engines

Large Eddy Simulation (LES) represents nowadays one of the most promising techniques for the evaluation of the dynamics and evolution of turbulent structures characterizing Internal Combustion Engines (ICE). The demand for a high level of resolution accuracy as well as the need to evaluate different scenarios and system configurations lead to considerable computational and economic costs for both the hardware infrastructure and the licensing fees of commercial codes. In such context, the present Doctoral project has the objective to define the most suitable numerical methodology to perform LES analysis of ICE flows and to implement such methodology in an efficient, accurate and robust CFD code, based on open-source components. An evaluation of freely available CFD codes has led to the choice of the open-source CFD package OpenFOAM as the most suited code for the project's objective. The LES modeling of interest for ICE applications has been then studied and three Sub-grid scale models particularly suited for such flows have been implemented and assessed into OpenFOAM. Moreover, Python scripts have been developed in order to automate and speed-up both pre-processing and post-processing phases. The CFD methodology has been then applied to a real world ICE systems such as a stationary flow bench, for which prior RANS simulations had shown some predictive deficiencies. The quality of the analyses has been assessed through specific LES quality estimators and the computational results have been validated against measurements, showing pretty good agreement. Finally, LES simulations have allowed the accurate investigation of the flow bench fluid-dynamic behavior and, thanks to the insights gained, an alternative RANS approach based on the Reynolds Stress Tensor Modeling has been proposed and tested in order to alleviate the aforementioned predictive deficiencies.