Large-eddy simulation coupled with wall models for turbulent channel flows at high Reynolds numbers with a lattice Boltzmann method — Application to Coriolis mass flowmeter

• Published in: Computers & mathematics with applications (1987) Vol. 78; no. 10; pp. 3285 - 3302
• Main Authors: Haussmann, Marc, BARRETO, Alejandro CLARO, KOUYI, Gislain LIPEME, Rivière, Nicolas, Nirschl, Hermann, Krause, Mathias J.
• Format: Journal Article
• Language: English
• Published: Oxford Elsevier Ltd 15.11.2019; Elsevier BV; Elsevier
• Subjects: Benchmarks; BGK; Channel flow; Computational fluid dynamics; Computer simulation; Coriolis mass flowmeter; Coupled walls; Direct numerical simulation; Domain names; Domain walls; Engineering Sciences; Finite element method; Fluid flow; Fluids mechanics; High Reynolds number; Industrial applications; Large eddy simulation; Mass flowmeters; Mathematical analysis; Mathematics; Mechanics; Numerical prediction; OpenLB; Pressure drop; Simulation; Turbulence models; Turbulent flow; Vortices; Wall function; OpenLB; Coriolis mass flowmeter; Turbulent flow; Wall function; Large-eddy simulation; BGK
• ISSN: 0898-1221; 1873-7668
• DOI: 10.1016/j.camwa.2019.04.033

The numerical prediction of pressure drop within wall-bounded flow domains at high Reynolds numbers (Re) using a large-eddy simulation (LES) approach is a challenging task for industrial applications because the fluid domain is usually underresolved. A lattice Boltzmann method (LBM) with Bhatnagar, Gross and Krook (BGK) collision operator coupled with the Smagorinsky–Lilly turbulence model is used to model these wall-bounded turbulent flows. The near wall region is modelled using wall functions to decrease the required mesh resolution for high Re. The influence of different velocity boundary approaches and wall functions is investigated for the benchmark bi-periodic fully developed turbulent channel flow for friction Reynolds numbers (Reτ) of 1000, 2000 and 5200. This benchmark case is validated against direct numerical simulation (DNS) results for turbulent statistics of 1st and 2nd order. Based on this validation, the pressure drop of an industrial Coriolis mass flowmeter is compared to experimental data for Re up to 127 800. The error of the pressure drop calculation in underresolved grids is reduced by two orders of magnitude in comparison to a no-slip approach for curved boundaries.