Application of Adaptive Large Eddy Simulation Methodology in IC-Engine Related Problems

Abstract

Today, Reynolds Averaged Navier Stokes (RANS) simulation approach remains the most widely used method in computational fluid dynamic studies of IC-Engines because it allows a good prediction of the mean flow properties at an affordable computational cost. The main limit of the RANS approach resides in the method used to predict turbulence that fails in the reproduction of anisotropic turbulence conditions. It can result in a lack of accuracy in reproducing the main physical processes, as spray evolution (mixture formation), heat transfer, and combustion, governing the IC-Engine physics. To fix this problem, the large Eddy Simulation (LES) approach can be considered. In LES the governing equations are filtered in space, rather than time-averaged as in RANS. It allows the direct solution of all the turbulent scales up to a cut-off length defined by the filter dimension. Therefore, in LES a more accurate description of the turbulence and of all the physical processes correlated to it has to be expected. However, even if the LES method allows an irrefutable improvement in turbulent flow solution accuracy, today its application to industrial IC-Engine design is still rare because of its high computational cost. During the last few years, significant advances in numerical methods, sub-grid scale models, and hardware performance have supported LES applications in many industrial fields. This paper is intended to work in the same direction by presenting a new LES methodology based on the coupling between LES and an adaptive mesh refinement (AMR) procedure. The main goal of this procedure is to guarantee a good resolution of the turbulent flow field adapting the filter size to the local turbulence length scale. The developed procedure allows a significant reduction of the total mesh size and, therefore, of the computational cost. The LES-AMR method was tested on an IC-Engine geometry for which experimental results were available.