Data-Driven RANS Turbulence Modeling of Mixed Convection in Reactor Downcomer Geometry

Arsen S. Iskhakov, Cheng Kai Tai, Igor A. Bolotnov, Nam T. Dinh, Elia Merzari

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Abstract

Several advanced nuclear reactor designs consider the use of low-Prandtl fluids (liquid metals) as coolants. However, their Reynolds-averaged Navier-Stokes (RANS) modeling is hampered by the absence of reliable turbulence closure models for Reynolds stresses (RS) and turbulent heat fluxes (THF). Accurate modeling of RS and THF is especially important when buoyancy effects take place in a flow. Since traditional physics-based models exhibit large model form uncertainties, there is a strong interest in data-driven (DD) modeling techniques. In this work, DD models for RS and THF are developed using a high-fidelity direct numerical simulation (DNS) database. The DNS database consists of sodium flows in the reactor downcomer geometry in the presence of buoyancy effects (mixed convection). The buoyancy is considered through the Boussinesq approximation. The DD models are based on invariant tensor basis neural network architectures. Their performance is investigated and discussed for selected flow conditions.

Original languageEnglish (US)
Title of host publicationProceedings of the 20th International Topical Meeting on Nuclear Reactor Thermal Hydraulics, NURETH 2023
PublisherAmerican Nuclear Society
Pages5004-5017
Number of pages14
ISBN (Electronic)9780894487934
DOIs
StatePublished - 2023
Event20th International Topical Meeting on Nuclear Reactor Thermal Hydraulics, NURETH 2023 - Washington, United States
Duration: Aug 20 2023Aug 25 2023

Publication series

NameProceedings of the 20th International Topical Meeting on Nuclear Reactor Thermal Hydraulics, NURETH 2023

Conference

Conference20th International Topical Meeting on Nuclear Reactor Thermal Hydraulics, NURETH 2023
Country/TerritoryUnited States
CityWashington
Period8/20/238/25/23

All Science Journal Classification (ASJC) codes

  • Nuclear Energy and Engineering
  • Instrumentation

Fingerprint

Dive into the research topics of 'Data-Driven RANS Turbulence Modeling of Mixed Convection in Reactor Downcomer Geometry'. Together they form a unique fingerprint.

Cite this