Register now After registration you will be able to apply for this opportunity online.
This opportunity is not published. No applications will be accepted.
Development of a new RANS/TLES hybrid turbulence model
Despite the advances in computational performance LES is still expensive and RANS therefore preferred in the industry. However, RANS provides only averaged results. The idea of hybrid solvers is to combine the advantages of both. The coupling is a major difficulty since LES is spatially filtered.
Keywords: RANS LES Hybrid Programming OpenFOAM Turbulence Modelling
A first TLES (temporal LES) model was proposed by Pruett et al. in 2003, which is based on an approximate deconvolution (reverting the filtering operation). Our TLES model on the other hand is
analytically closed, therefore the filtered fields can be deconvoluted exactly. This means that the
results match with those obtained from DNS, given the resolution is equal.
The basic principle of TLES is to employ filtering in the time domain rather than in the space
domain. As damping high-frequency motions leads to a damping of high-wavenumber motions, simulations can be performed on a coarser temporal and spatial resolution. As in conventional LES the motivation is to save computational costs however, from our perspective the main advantage of TLES is its natural linkage to RANS which also exploits time-domain filtering. In two previous master theses a RANS-TLES hybrid solver was implemented and validated in OpenFOAM.
In order to damp high-frequency motions to make the use of a coarser resolution applicable, the
filter width needs to be several orders of magnitude above the time step. This poses difficulties as stability problems arise for such filter widths since the filtering is not dissipative.
Currently the source of the instabilities is being investigated.
A first TLES (temporal LES) model was proposed by Pruett et al. in 2003, which is based on an approximate deconvolution (reverting the filtering operation). Our TLES model on the other hand is analytically closed, therefore the filtered fields can be deconvoluted exactly. This means that the results match with those obtained from DNS, given the resolution is equal. The basic principle of TLES is to employ filtering in the time domain rather than in the space domain. As damping high-frequency motions leads to a damping of high-wavenumber motions, simulations can be performed on a coarser temporal and spatial resolution. As in conventional LES the motivation is to save computational costs however, from our perspective the main advantage of TLES is its natural linkage to RANS which also exploits time-domain filtering. In two previous master theses a RANS-TLES hybrid solver was implemented and validated in OpenFOAM. In order to damp high-frequency motions to make the use of a coarser resolution applicable, the filter width needs to be several orders of magnitude above the time step. This poses difficulties as stability problems arise for such filter widths since the filtering is not dissipative. Currently the source of the instabilities is being investigated.
Not specified
Interested candidates please send an email with a recent transcript of records to doberle@ifd.mavt.ethz.ch
Interested candidates please send an email with a recent transcript of records to doberle@ifd.mavt.ethz.ch