SIA - Détails de l'évenement

The LES for Internal Combustion Engine Flows (LES4ICE) conference aims at providing a forum for exchanges concerning recent advances in research on and the use of Large-Eddy Simulation (LES) applied to internal combustion engine (ICE) flows, injection and combustion. The aim is to bring together engine designers and players in LES research worldwide and have them compare views on the state of the art in LES applied to ICEs and on the bottlenecks future research will have to overcome to pave the way to practical use of LES in the piston engine design process.

LES has become a topic of central interest in Computational Fluid Dynamics (CFD) applied to turbulent flows. This reflects its potential to bridge the gap between the classical Reynolds Averaged Navier – Stokes (RANS) approach - which gives access to an ensemble average by modeling the whole turbulent spectrum – and Direct Numerical Simulation (DNS) - which resolves the whole spectrum of flow scales without any turbulence modelling. LES is aimed at resolving the large instantaneous flow scales that depend directly on the geometry studied and modelling only the small scales whose structure can be assumed universal. LES opens up new prospects for the use of CFD in the piston engine design process. It can be expected to definitively improve the description of engine aerodynamics, because the steady growth of computing power will make it possible to resolve the essential part of the flow energy in LES, yielding increasingly reliable results. In the next 10 years, this opens up the prospect of using LES for numerical investigations of engine aerodynamics capable of yielding reliable quantitative results that could potentially further reduce development times and costs for new concepts. Concerning liquid injection, mixing and combustion in piston engines, many phenomena will not be resolvable on meshes small enough to be practical, and LES predictions will therefore depend on the quality of sub-grid-scale models. Even so, given that the major input to these models – local turbulence conditions – is accurately predicted in LES, one can reasonably envisage increased levels of predictivity. This is all the more true in that model formulation for LES involves instantaneous local phenomena, which are more easily addressed in a general way than the multiscale statistical phenomena addressed in RANS models, which can not distinguish turbulence, intermittency and cyclic variations.