CFD validation in OECD/NEA t-junction benchmark.

Obabko, Aleksandr; Fischer, Paul; Tautges, Timothy J.; Karabasov, Sergey A.; Goloviznin, V. M.; Zaytsev, Mikhail A.; Chudanov, V. V.; Pervichko, V. A.; Aksenova, A. E.

doi:10.2172/1024601

Cited by 18 publications

(14 citation statements)

References 12 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The computations disclosed in the previous paragraph were expensive and required access to significant High Performance Computing (HPC) resource. It can be estimated that an LES simulation for a straight inlet T-Junction at a moderate Reynolds number of 10 4 with a maximum grid resolution of 20 million required, at minimum, [ 300 cpu-hours [63] to compute; with this increasing further if higher order statistics are desired. In contrast a RANS computation could be completed in less than one day, whilst still providing accurate predictions of the time-averaged thermal-hydraulic characteristics [55].…”

Section: Thermal Mixing In a T-junctionmentioning

confidence: 99%

A Review of Embedded Large Eddy Simulation for Internal Flows

Holgate

Skillen

Craft

et al. 2018

Arch Computat Methods Eng

View full text Add to dashboard Cite

When scale-resolving simulation approaches are employed for the simulation of turbulent flow, computational cost can often be prohibitive. This is particularly true for internal wall-bounded flows, including flows of industrial relevances which may involve both high Reynolds number and geometrical complexity. Modelling the turbulence induced stresses (at all scales) has proven to lack requisite accuracy in many situations. In this work we review a promising family of approaches which aim to find a compromise between cost and accuracy; hybrid RANS-LES methods. We place particular emphasis on the emergence of embedded large eddy simulation. These approaches are summarised and key features relevant to internal flows are highlighted. A thorough review of the application of these methods to internal flows is given, where hybrid approaches have been shown to offer significant benefits to industrial CFD (relative to an empirical broadband modelling of turbulence). This paper concludes by providing a cost-analysis and a discussion about the emerging novel use-modalities for hybrid RANS-LES methods in industrial CFD, such as automated embedded simulation and multi-dimensional coupling.

show abstract

Section: Thermal Mixing In a T-junctionmentioning

confidence: 99%

A Review of Embedded Large Eddy Simulation for Internal Flows

Holgate

Skillen

Craft

et al. 2018

Arch Computat Methods Eng

View full text Add to dashboard Cite

show abstract

“…calculations [10] were compared with experiment and the numerical calculations given in [2] (Figs. 3 and 4).…”

Section: Flow In T-shaped Joint Of Circular Tubesmentioning

confidence: 99%

“…The procedure was verified on numerical tests [7,9,10] for Rayleigh numbers in the range 10 6 -10 16 and Reynolds numbers in the range 10 2 -10 4 . In the present work, the numerical calculations were compared with experiments: flow in circular tubes in a T joint and ERCOFTAC experiment studying natural convection of an incompressible liquid at high Rayleigh numbers.…”

mentioning

confidence: 99%

Supercomputer Mathematical Modeling of the Fluid Dynamics in Elements of Nuclear Power Facilities

et al. 2015

Self Cite

View full text Add to dashboard Cite

A three-dimensional CFD code CONV-3D is being developed for mathematical modeling of the fluid dynamics of elements of nuclear power facilities. The code is based on developed algorithms with low artificial diffusion, for which discrete approximations are constructed using finite-volume and staggered grid methods. A regularized nonlinear monotonic operator splitting scheme has been developed to solve advection problems. Richardson's iteration method with a fast Fourier transform for the Laplace operator is used to solve the pressure equation. Compared with the conventional conjugate gradient method, this approach to the solution of elliptic equations with variable coefficients gives multiple acceleration. Quasidirect numerical modeling is used to model three-dimensional turbulent flows in single-phase streams. The CONV-3D code is completely parallelized and is effective on multiprocessor cluster computers, such as Chebyshev and Lomonosov (MGU).One avenue of our country's energy program is the construction and assimilation of fast reactors with a liquid metal coolant. Such reactors make expanded breeding of nuclear fuel possible, which opens the possibility of development with no limit on fuel resources.The high operational efficiency and reliability of fast reactors are determined by the level of their thermohydrodynamic validation. Multidimensional modeling of thermohydrodynamic processes for different design-basis accidents by means of modern computers and numerical methods makes this possible. Such calculations require, among other things, informativeness, complexity, and high reliability, especially for local hydrodynamic and thermal characteristics. Multidimensional methods of mathematical modeling of the velocity and temperature fields in the fuel assemblies of fast reactors are now under intensive development. A reliable thermohydrodynamic calculation of high-flux modern fast reactors requires verification of the numerical algorithms and computational modules.The experimental data do not cover the critical phenomena that are important from the standpoint of the safety of NPPs with fast reactors and liquid metal coolant (BREST, SVBR) and are included in the verification matrix [1]. Experimental studies including mathematical computational tracking of the results, local experiments studying the individual phenomena, and high-precision experiments studying fundamental processes are required. The following directions of study are included among the latter: investigation of the thermophysical properties of liquid-metal coolants, modeling of through flows in the elements of nuclear power facilities, and modeling of the development of natural convection at high Rayleigh numbers.The 3D-unified CFD method developed for the analysis of NPP safety includes the following: 1) methods and algorithms, as well as the software based on them, for automatic generation of computational grids with local densification near the boundaries of a body;

show abstract

“…The Argonne code Nek5000 participated in this benchmark using this mesh [17]. The mesh was generated with CUBIT 10.2, then modified to extend the (top) inflow and (side) outflow pipes.…”

Section: Hex-meshed T-junction Benchmark Problemmentioning

confidence: 99%

PostBL: Post-mesh Boundary Layer Mesh Generation Tool

Jain

Tautges

2014

Proceedings of the 22nd International Meshing Roundtable

View full text Add to dashboard Cite

Abstract.A boundary layer mesh is a mesh with dense element distribution in the normal direction along specific boundaries. PostBL is a utility to generate boundary layer elements on an existing mesh model. PostBL supports creation of hexahedral, prism, quad, and tri boundary layer elements. It is formulated as an algorithm in MeshKit, which is an open-source library for mesh generation functionalities. Typically, boundary layer mesh generation is a premeshing process; in this effort, however, we start from a model that has already been meshed. Boundary layer elements can be generated along the entire skin or on selected exterior or internal surface boundaries. PostBL mesh operation can be coupled with other MeshKit meshing operations such as Jaal, NetGen, TetGen, and CAMAL and custom meshing tools such as RGG. Simple examples demonstrating generation of boundary layers on different mesh types and the OECD Vattenfall T-Junction benchmark hexahedral mesh are presented. IntroductionBoundary layers are typically used for CFD applications, in regions with strong gradients-turbulent flow, diffusion-type equations, laminar flow or no-slip boundary with strong gradient of velocity [23]. Most other tools for generating boundary layers insert them before bulk mesh generation; in this effort, however, we develop a tool called PostBL and start from an already meshed model. This strategy is particularly useful when the meshing of the original problem is complicated and involves geometry decomposition and other geometry modifications prior to meshing. A utility for adding layers of elements along the exterior or interior (shared by two materials) boundary is developed. It relies on the edge or surface (2D or 3D) mesh on which the boundary layer elements are desired. This method is also useful for creating boundary layer elements in the wake regions along the direction of fluid flow. After boundary layer insertion, smoothing can optionally be used to improve mesh quality. Typically in external aerodynamic simulations, the type of turbulent modeling decides the thickness and number of boundary layer elements; PostBL would save time and eliminate the complexity involved in remeshing the entire model.

show abstract

CFD validation in OECD/NEA t-junction benchmark.

Cited by 18 publications

References 12 publications

A Review of Embedded Large Eddy Simulation for Internal Flows

A Review of Embedded Large Eddy Simulation for Internal Flows

Supercomputer Mathematical Modeling of the Fluid Dynamics in Elements of Nuclear Power Facilities

PostBL: Post-mesh Boundary Layer Mesh Generation Tool

Contact Info

Product

Resources

About