검색 상세

Effects of spatially non-uniform wall temperature on turbulent heat transfer inside a square duct

초록 (요약문)

Spatially non-uniform heat transfer is observed frequently in heat exchangers, gas turbines, etc. In this study, direct numerical simulations are performed for turbulent heat transfer inside a square duct for Reτ = 150, 227 and Pr = 0.2, 0.7, 2 to understand the effects of spatially varying temperature boundary conditions (T BCs) on local heat transfer and broken velocity temperature similarity. With spatially non-uniform T BCs, the wall-normal temperature profiles are significantly lower than the log law near the corners where the profiles with uniform T BC approximately follow the log law. Both the secondary flows and spatially non-uniform T BCs contribute to broken similarity between the Nusselt number and friction coefficient in terms of the time-mean, RMS, and their ratio. The RMS-to-mean ratio of the Nusselt number becomes less sensitive to Pr as it increases. The RMS temperature and turbulent heat flux fields vary greatly depending on the T BCs, while the latter does not depend much on the thermal boundary layer thickness. The correlation coefficients between instantaneous turbulent fluctuations show that spatially non-uniform T BCs promote the broken velocity-temperature similarity. A mechanism by instantaneous vortical structures (interacting with Pr) is found across the domain using the swirling strength of the turbulent momentum and heat fluxes. The T BC by itself is shown to be an important factor to break the similarity in the wall vicinity. Most results present global effects of the T BC. From the viewpoint of the Boussinesq approximation, the directional error of the modeled turbulent heat flux may increase significantly by spatially non-uniform T BCs, corner vortices, and increasing Pr. Under spatially non-uniform T BCs, the error in turbulent heat flux tends to be biased to the wall-tangential component. The fields of the isotropic and anisotropic turbulent diffusivities are compared for an idea on revised modeling.

more