Benchmarking Reynolds-Averaged Navier–Stokes Turbulence Models for Water Clarification Systems

Abstract

Turbulence is inherent in clarification basin systems; turbulence is challenging to quantify, yet directly impacting particulate matter (PM) separation. In computational fluid dynamic (CFD), Reynolds-averaged Navier-Stokes (RANS) turbulence models are widely adopted for computational efficiency. However, the accuracy of RANS is less examined in clarification systems. RANS models without benchmarking can potentially cast doubt or false confidence for results. In this study, common RANS models are applied to steady-flow clarification. Results are examined against high-resolution large-eddy simulations (LES) by high-order spectral element method (SEM) Nek5000 and laser Doppler anemometry (LDA) for three deflector configurations: (1) one-sided, (2) two-sided, and (3) no deflector. RANS model’s relative mean differences with respect to LES ranges from 13.9% to 41.4% and 32.7% to 105.1% in streamwise and vertical velocities, varying with configurations. LES predictions were improved from RANS. LES and RANS differences transcend hydrodynamics and persist for PM. Across configurations, RANS models of PM separation are predominately lower than LES. RANS model variability is discerned across clarifier configurations, decreasing with increasing PM diameter.