Search

Local convective heat transfer coefficients were measured from a smooth NACA 0012 airfoil having a chord length of 0.533 m. Flight data were taken for the smooth airfoil at Reynolds numbers based on chord in the range 1.24 to 2.50 million and at various angles of attack up to 4 deg. During these flight tests, the freestream velocity turbulence intensity was found to be very low. Wind tunnel data were acquired in the Reynolds number range 1.20 to 4.52 million and at angles of attack from -4 to +8 deg. The turbulence intensity in the IRT was 0.5-0.7 percent with the cloud-generating sprays off. A direct comparison between the results obtained in flight and in the IRT showed that the higher level of turbulence intensity in the IRT had little effect on the heat transfer for the lower Reynolds numbers but caused a moderate increase in heat transfer at the higher Reynolds numbers. Turning on the cloud-generating spray nozzle atomizing air in the IRT did not alter the heat transfer. The present data were compared with leading-edge cylinder and flat plate heat transfer correlations that are often used to estimate airfoil heat transfer in computer codes.

The present study numerically demonstrates how small spanwise variations in velocity upstream of a body can cause relatively large increases in the spanwise-averaged heat transfer to the leading edge. Vorticity introduced by spanwise variations, first decays as it drifts downstream, then amplifies in the stagnation region as a result of vortex stretching. This amplification can cause a periodic array of 3D structures, similar to horseshoe vortices, to form. The numerical results indicate that, for the given wavelength, there is an amplitude threshold below which a structure does not form. A one-dimensional analysis, to predict the decay of vorticity in the absence of the body, in conjunction with the full numerical results indicated that the threshold is more accurately stated as minimum level of vorticity required in the leading edge region for a structure to form. It is possible, using the one-dimensional analysis, to compute an optimum wavelength in terms of the maximum vorticity reaching the leading edge region for given amplitude. A discussion is presented which relates experimentally observed trends to the trends of the present phenomena.

Local heat transfer coefficients were obtained on irregular cylindrical shapes which typify the accretion of ice on circular cylinders in cross flow. The shapes were 2, 5, and 15 min accumulations of glaze ice and 15 min accumulation of rime ice. These icing shapes were averaged axially to obtain a nominal shape of constant cross section for the heat transfer tests. Heat transfer coefficients were also measured around the cylinder with no ice accretion. The models were run in a 15.2 x 68.6 cm (6 x 27 in.) wind tunnel at several velocities. The models were also run with a turbulence producing grid which gave about 3.5 percent turbulence. The effect of roughness was also simulated with sand grains glued to the surface. Results are presented as Nusselt number versus angle from the stagnation line for the smooth and rough models for both high and low levels of free stream turbulence. Roughness of the surface in the region prior to flow separation plays a major role in determining the heat transfer distribution. Free stream turbulence does not affect the distribution of heat transfer in this region but raises the level by a nearly uniform amount. For the rime shape, roughness had a larger effect in the near wedge shaped region past the initial separation point.