Turbulent forced convective heat transfer is often encountered in engineering flows. A method of improving the heat transfer from a smooth surface is to add ribs and generate some turbulence. The precise nature and spacing of these ribs can have a considerable effect on the heat transferred. In this investigation the novel method of holographic interferometry is used to study the thermal fields over numerous two-dimensional ribbed geometries and to ascertain the Nusselt number distribution around them. Pull field information is provided as the fringe pattern generated is essentially an isotherm contour map of the flow situation. Hence, this makes it ideally suited to verify the theoretical solutions obtained from Large Eddy Simulations (LES) and Finite Element predictions. This latter predictive technique is used in this investigation and the solutions obtained are compared with the experimental results. Initially, a smooth surface geometry was investigated to verify the accuracy of the experimental technique. Excellent results were achieved, but a necessity to have momentum field information was identified. Ribbed geometries with a pitch to height ratio of 7.2 : I were then studied. Double exposure and real-time techniques enabled both detailed thermal measurements to be made and any time-dependency of the field to be identified. Flow rates up to nuclear reactor conditions (e+ = 600) were studied and typical interferograms are illustrated. A Reynolds number dependency for the heat transfer distribution for a rounded-rib geometry was identified while the Bquare-ribbed geometry distribution was observed to be Reynolds number independent. The addition of an insulating deposit to the surfaces is also studied. It led to smaller peak heat transfer rates and larger surface temperature variations. For large thermal gradients, usually at high flow rates, a ray crossing regime was identified. This led to a limitation of the technique as no information in this area could be extracted. However, a boundary layer extrapolation from the identifiable regions to the wall eradicates this limitation, but is only possible if momentum field information is available. Hence, this problem did not act as a limit to the smooth surface or square-ribbed geometry but did for all others. Numerical simulations of the flow were undertaken using the k-e and q-f models employed in the finite element code FEAT. Because a low flow rate was modelled the conventional wall functions were not considered appropriate and hence a low Reynolds number model was used. Incorrect modelling of the length scales in the near wall region using this model led to errors in the thermal field predictions. Hence, only a qualitative comparison with the experimental results is undertaken and a recommendation for improved modelling is proposed based on this comparison.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:378429 |
| Date | January 1987 |
| Creators | Lockett, John Francis |
| Publisher | City, University of London |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://openaccess.city.ac.uk/20856/ |
Page generated in 0.0014 seconds