MOMENTUM AND BUOYANCY REPARTITION IN TURBULENT MIXED CONVECTION

Kurzfassung

We present results from direct numerical simulations of asymmetrically heated turbulent vertical channel flow at Reτ = 00, Gr = 9.6 · 105 and P r = 0.71. The vertical channel represents a canonical geometry suitable for the analysis of turbulent wall-bounded flows. The differentially heated walls pose as sources of aiding and opposing buoyant forces altering the amplification and suppression of turbulent motions. In comparison with plane channel flow, this offers the possibility for analysing amplification and surpression of turbulent fluctuations close to the heated and cooled walls, respectively. Similar studies concerned with heat transfer in fluids deal with vertical pipe flows heated from the outside such as Metais & Eckert in 1964 [5]. Afterwards, many others followed and studied heated pipe flow in experiments, however they were not able to resolve the small flow scales near the walls. In 2006, Bae et al. [1] presented their DNS results of a strongly heated pipe flow with regard to large density variations. They showed that the flow relaminarizes under heating and analysed different divergence measures from the unheated reference case. In 1997, Kasagi & Nishimura (KN) [3] published detailed statistics obtained with DNS of vertical channel flow. References to earlier works can be found in their paper. The isothermally heated and cooled walls lead to an asymmetric flow structure. Towards the heated channel wall the buoyancy aids the mean flow rate and augments it while the mean flow rate is suppressed near the cooled wall, hereafter denominated as opposing flow side. The peculiarity of relaminarizing flow in the channel merges on the aiding flow side, where the locally higher Reynolds number should lead to increased turbulence intensities. Instead, the turbulence is suppressed there and augmented on the opposing flow side. KN presented an overview of statistical results including statistical moments of momentum and heat transfer as well as stress balances. Their reported turbulent kinetc energy budgets for both the aiding and opposing flow sides depict the dominance of the particular terms but not the interaction between the velocity components and the buoyancy term in particular. In the work presented at the DLES11 we want to further deepen the analysis of how buoyancy affects the flow behavior and structures and compare with results obtained from plane channel flow.