Development and implementation of a new hybrid RANS/LES model for transitional boundary layers in OpenFOAM

  • Abhi Beechook

    Student thesis: Master's ThesisMaster of Science by Research

    Abstract

    Boundary layer transition occurs in a wide range of engineering applications and accurately modelling transition has been a challenge for over a century. In recent years, hybrid RANS/LES
    modelling approaches have gained significant attention by the research community. In essence,hybrid RANS/LES approaches employ Reynolds-averaged Navier-Stokes (RANS) in wall regions
    (i.e. to model the boundary layer) whilst applying subgrid-scale large-eddy simulation (LES) models to separated flow regions. A new hybrid RANS/LES model was developed as part of the
    current work for transitional boundary layers. The hybrid RANS/LES model was developed using the detached-eddy simulation (DES) hybridisation approach [1] and using the kT −kL −ω [2] transition model as the background RANS model. The model was implemented in OpenFOAM, an open source CFD package written in C++ programming language. The model development process starts with the formulation and implementation of the DES version of the hybrid RANS/LES model, the kT − kL − ω DES model. The model when tested on a 2D flat plate at zero-pressure-gradient, failed to capture laminar-turbulent transition, and the solutions remained fully laminar. This was due to a serious problem faced by the DES approach. In order to alleviate the problem faced by the DES model, the delayed detached-eddy simulation (DDES) approach was used to develop the kT − kL − ω DDES model, which responded very well when tested on the 2D flat plate configuration. The DDES results were compared with the RANS (kT − kL − ω transition model) results, the kT − kL − ω DDES demonstrated approximately a 30% improvement in the predicted transition location. The newly developed kT − kL − ω DDES model was also tested on a 2D cylinder and the results obtained showed reasonably good agreement with experimental data, which also indicated the need to calibrate the model for important parameters such as the DES model constant, CDES.
    Date of Award2015
    Original languageEnglish
    Awarding Institution
    • Coventry University
    SupervisorStephen Benjamin (Supervisor), Humberto Medina (Supervisor) & Remus Cirstea (Supervisor)

    Cite this

    '