Turbulent Free Convection Over Horizontal Surfaces

Abstract

This thesis deals with turbulent free convection in a horizontal fluid layer above a heated surface. The work can be roughly divided into two parts. In the &st part a model is proposed for the near wall dynamics in turbulent convection over heated horizontal surface. The second part is an experimental study of the heat transfer rates and the near wall flow structure under different conditions. Experimental evidence indicates that the near wall flow consists of randomly moving line plumes. Our model.for the near wall dynamics - based on this evidence - consists of periodic, steady 2-D line plumes and the associated boundary layers. The model predicts a value for the average plume spacing (A,). It is given by R ~=Y 52 Pr-0.012,w here Rab is the Rayleigh number based on A,. The predicted plume spacing is in reasonable agreement with the plume spacings that have been obtained in experiments. The model predicts, reasonably well, the distributions of mean temperature, rms temperature fluctuations and rms vertical velocity fluctuations close to the wall. There is disagreement between the predicted and experimental rms horizontal velocity fluctuations. This is probably due to the large scale flow present in real flows but which is not accounted for in the present model. Experiments have been carried out on a heated horizontal surface to obtain and analyse the planform structure and the heat transfer under different conditions. Water is the working fluid. The different conditions are the steady convection between parallel plates (Rayleigh-Bhnard convection), the convection with the top plate insulated (unsteady non-penetrative convection), the convection with the top fluid surface exposed to the ambient, and the convection with a small external flow imposed on the heated surface. The last experiment is an attempt to mimic the large scale flow obtained in high Rayleigh number convection. The Rayleigh-BCnard convection experiments and the external flow experiments have been for one heat flux value and one cell height, whereas the other two experiments have been done at four fluxes and three heights. r Visualisation studies confirm the existence of randomly moving line plumes close to the heated wall for all the conditions. The planform obtained for Rayleigh-Benard convection is one of randomly oriented lines in the form of cells. External flow on the heated surface alignes these lines in the flow direction. Further, the external flow reduces heat transfer rate for a given temperature difference compared to the Rayleigh-B enard case. In the experiments with top water surface open to the ambient, three types of planform structure were identifiedviz., cells, broken lines and aligned lines. At low fluxes, .. the cells resemble the type of planform we obtain in Rayleigh-Benard convection. At high fluxes we start getting broken lines. At the highest flux and, especially, at low aspect ratio, we get aligned plumes, presumably, due to a large scale velocity generated inside the convection cell by the top evaporating free surface. These aligned line plumes resemble the type of planform we obtain in our external flow experiments. The sideviews at high fluxes show inclined dye lines confirming the presence of a large scale flow. The non-dimensional average plume spacing, ~ a i ' ~va,ri es between 40 and 90. At the higher fluxes the plume spacings are close to the model predictions. In order to represent the heat transfer rate un'der different conditions we represent it (the heat transfer rate) by ~a;;'~, where RasT is Rayleigh number based on the conduction layer thickness. In the open convection experiments the Rad-1,1 3 lies between 0.1 and 0.2, and shows dependence on Ra, and aspect ratio. In the experiments with the top plate insulated, Ra&, is nearly constant with change in aspect ratio and Rayleigh number and -113 a E 0.13. For comparison this value of Ra&, is obtained at Ra -- 10" in Rayleigh-B6nard / convection.