The nature and cause of cyclic variability in engines is not yet fully understood but their importance in developing advanced engine concepts keeps increasing. Giving automotive engineers design tools to predict cyclic variability, and more generally unsteady engine phenomena like warm-up or engine transients, can be viewed as a major challenge for a more effective design of combustion engines of the future. Experimental efforts alone are not sufficient to fully explore the many facets of cycle-to-cycle fluctuations, and therefore Computational Fluid Dynamics (CFD) simulations appear as the best candidate for providing such a design tool. A suitable candidate is the Large Eddy Simulation (LES) technique. Identifying and solving the underlying fundamental flow physics will be the key to success in understanding the stochastic nature of in-cylinder flow – this, in turn, is required to develop practical first-principles-based predictive computational LES design tools for robust and reliable industrial use.
In this study, low-, medium-, and high-resolution LES approaches are being pursued in parallel to address these needs. High-resolution LES (minimal subgrid-scale modeling, approaching DNS; >108 cells per cylinder) is being used for physics discovery and model development/validation. Low-resolution LES (RANS-like resolution; currently 105-106 cells per cylinder) is being used for engineering development and applications. Medium-resolution LES (currently 106-107 cells per cylinder) bridges these two extremes. Mesh sizes and the number of cycles that are simulated will increase with time.
Optical single-cylinder engines were specifically designed, built, and characterized for LES model development and validation. Optical, including laser-based, imaging tools are available to measure flow and scalar quantities with high spatial and temporal resolution. In particular, the use of high-speed imaging tools enables recording of the evolution of flow structures through a cycle and from cycle to cycle.
The intent of this program is to create an interactive collaboration between the modeling and experimental efforts that will identify and execute the experiments and models needed to produce physically based LES models with sufficient fidelity for research and engineering applications.