Measurements of the convective heat transfer coefficient from ice roughened surfaces in parallel and accelerated flows

Values for the heat transfer coefficient from different ice roughened surfaces in parallel and accelerated flows were experimentally sought for use in de-icing computer codes. Aluminum castings of different ice accretions in an icing wind tunnel were obtained from which heat transfer models were constructed. Each model was a large composite of heat flux gages to which heat was supplied from the bottom using themofoil heating elements. The heat supplied to each gage was allowed to convect upward from the rough surface to an air stream in a dry wind tunnel. Other heat losses were eliminated and each gage was insulated from surrounding ones. Average values of the heat transfer coefficient were computed from an energy balance for each gage with the known electrical power of the heating elements.; Results were obtained for local Reynolds numbers ranging from {dollar}5.3times10sp4{dollar} to {dollar}1.3times10sp6{dollar}, and for tilt angles of {dollar}0spcirc, 14spcirc, 23spcirc, 32spcirc,{dollar} and {dollar}41spcirc{dollar}. The results were in general qualitative agreement with those of uniform roughness with the different behaviors being more drastic in the case of stochastic roughness. The Stanton number for random roughness was higher than that for uniform roughness and was directly proportional to both roughness element height and area increase and inversely proportional to spacing. The effect of free-stream velocity diminished at high enough Reynolds number and the Stanton numbers collapsed onto a single curve. Acceleration caused Stanton number to start at lower values close to the leading edge, Stanton numbers then increased as the flow accelerated along each tilted model. In the fully rough region and for parallel and mildly accelerated flows, up to 23{dollar}spcirc{dollar}, Stanton number was a function of Reynolds number only and followed a power law. The multiplier and the exponent of Reynolds number in this power law were found to correlate well with the newly defined parameter, Index of Random Roughness, and the roughness height, respectively.