DNS 1-2: Difference between revisions
(Modifications as per suggestions from HiFiTURB consortium) |
|||
| Line 8: | Line 8: | ||
= Abstract = | = Abstract = | ||
The archive comprises snapshots and time-average data produced via a high-fidelity computational simulation of turbulent channel flow, which is an fundamental canonical flow. The simulation was undertaken using the open source PyFR flow solver on Nvidia V100 GPUs on the [https://www.hpc.cineca.it/hardware/marconi100 Marconi] cluster of [https://www.cineca.it CINECA] and the Mystery cluster at Imperial College London under the [https://cordis.europa.eu/project/id/814837 HiFiTURB] project. The data can be used to train the next-generation of Reynolds Averaged Navier-Stokes turbulence models via a machine learning approach, which would have broad applicability to a wide range of science and engineering problems. | The archive comprises snapshots and time-average data produced via a high-fidelity computational simulation of turbulent (developed) channel flow, which is an fundamental canonical flow. The simulation was undertaken using the open source PyFR flow solver on Nvidia V100 GPUs on the [https://www.hpc.cineca.it/hardware/marconi100 Marconi] cluster of [https://www.cineca.it CINECA] and the Mystery cluster at Imperial College London under the [https://cordis.europa.eu/project/id/814837 HiFiTURB] project. The data comprises mean velocity, pressure, turbulent stress and all the correlations appearing in the budget equations for turbulent stress and turbulent kinetic energy. The data can be used to train the next-generation of Reynolds Averaged Navier-Stokes turbulence models via a machine learning approach, which would have broad applicability to a wide range of science and engineering problems. | ||
<!--{{Demo_UFR_Guidance}}--> | <!--{{Demo_UFR_Guidance}}--> | ||
<br/> | <br/> | ||
Revision as of 10:11, 7 December 2021
DNS Channel Flow =180
Abstract
The archive comprises snapshots and time-average data produced via a high-fidelity computational simulation of turbulent (developed) channel flow, which is an fundamental canonical flow. The simulation was undertaken using the open source PyFR flow solver on Nvidia V100 GPUs on the Marconi cluster of CINECA and the Mystery cluster at Imperial College London under the HiFiTURB project. The data comprises mean velocity, pressure, turbulent stress and all the correlations appearing in the budget equations for turbulent stress and turbulent kinetic energy. The data can be used to train the next-generation of Reynolds Averaged Navier-Stokes turbulence models via a machine learning approach, which would have broad applicability to a wide range of science and engineering problems.
Contributed by: Arun Soman Pillai — Imperial College London
© copyright ERCOFTAC 2024