LES and DES for complex geometries
Title: LES and DES for complex geometries
DNr: SNIC 2022/5-22
Project Type: SNIC Medium Compute
Principal Investigator: Lars Davidson <lada@chalmers.se>
Affiliation: Chalmers tekniska högskola
Duration: 2022-02-01 – 2023-02-01
Classification: 20306
Homepage: http://www.tfd.chalmers.se/~lada/
Keywords:

Abstract

The project will be used for several purposes: Airframe noise, generated by the interaction of turbulent flows with aircraft landing gears (LG) and high-lift devices (HLD), The most commonly used CAA approach is a hybrid method combining Computational Fluid Dynamics (CFD) simulations and acoustic analogies. Among different CFD methods, Detached Eddy Simulation (DES) is considered to be sufficiently accurate and computationally affordable for aeronautical industry, and thus is the best trade-off between predictive accuracy and computational cost. In the presently proposed project, for the LG configuration, numerical simulation and modelling will be carried out to explore LG noise reduction with a perforated fairing. Instead of resolving the detailed flow past the fairing (mesh screen), a novel numerical model will be used, which will be verified against experimental data. Implementation and evaluation of numerical boiling models within a homogeneous mix- ture multiphase framework to analyse nucleate boiling occurring in high power density applications, such as coolant jacket of internal combustion engines. This numerical modelling approach includes a wall boiling model combined with a model for interphase mass transfer. The numerical methodology is implemented in a high-resolution three dimensional Conjugate Heat Transfer (CHT) model of a Volvo engine in production. Evaluating different methods for imposing velocity fluctuations in a compressible finite-volume solver are evaluated for embedded hybrid RANS-LES applications. The first method is a volume source term derived from an expansion of the time discretization of the fluctuating velocity component, suitable for an unstructured solver. The second method stems from an expansion of the convective flux of the fluctuating velocity component normal to the flow direction. The third methods considers a combination of both aforementioned methods. Additionally, a commutation term is derived for the SA turbulence model in order rapidly reduce the turbulent viscosity across the RANS-LES interface.