Abstract

This investigation focusses on the use of two wall-resolved LES approaches for the simulation of the buoyancy driven flow in a differentially heated square cavity, aiming to advance understanding of the flow phenomena involved. The study also assesses the predictive accuracy and cost-effectiveness of LES for natural convection flows. This assessment is pursued through comparisons of the resulting predictions, and also of their computational requirements, with those for DNS, produced in an earlier study, and those of a number of URANS models produced during the current investigation. URANS models tested here include both high- and low-Reynolds-number schemes. In the case of the high-Reynolds-number models a recently developed numerical variant of the Analytical Wall Function (AWF) has been employed to account for the effects of near-wall turbulence. Both LES approaches result in predictions that are close to those of the DNS but have considerably more modest resource requirements. The URANS approaches, which have the lowest resource requirements, for key parameters such as the local Nusselt number are not as accurate as the LES approaches and their performance is highly dependent on the turbulence model employed. Surprisingly, the simplest of the low-Re models employed, the Launder–Sharma two-equation model, is reasonably close to the DNS data. The approach that requires the least computational resource, the high-Re model with the recent version of the AWF, also offers a promising URANS route. Simulations were also conducted to investigate the flow behaviour at a Rayleigh number that is an order of magnitude higher than that of the available DNS. At such high Rayleigh number, at which the phenomena are of close relevance to cooling loops of nuclear reactors, comparisons are performed between the LES and the high-Re AWF combination. The findings show that as the Rayleigh number increases, the turbulence levels become stronger, but these are confined to more refined structures within the thinner near-wall boundary layers.

Full Text
Published version (Free)

Talk to us

Join us for a 30 min session where you can share your feedback and ask us any queries you have

Schedule a call