【正文】 a blockage of about % of the entire width of the test tunnel. The mainstream velocity distribution during the tests was the same (177。2%), regardless of the freestream turbulence conditions. Incident mainstream velocity (Ux) was 21 m/s at X/D = (X/b = 48 for grid 1 and Xlb = for grid 2) from the grid location that transforms into a mainstream Re = 100,900. Oning freestream turbulence intensity for the nogrid case was and % for grid 1, and % for grid 2, based on levels at X/D = . Corresponding dissipation length scales at the same location for the gridgenerated turbulence were estimated to be about and cm for grids 1 and 2, respectively. Based on the analysis for flow around a circular cylinder,16 the boundarylayer displacement and momentum thickness were calculated at the upstream edge of the spallation location S4 (35 deg from the leading edge). The displacement thickness to spallation depth ratios were obtained to be and for depths of and cm, respectively. Similarly, the momentum thickness to spallation depth ratios were obtained to be and . The spallation depth modeled in this study was much larger than the local boundarylayer thickness at each location. This indicates that the boundary layer will be significantly disturbed by the spallation. This may cause strong effects on local heat transfer distributions. On the actual turbine airfoil, the boundary layer was formed at the leading edge of the airfoil. Therefore, the boundarylayer thickness will be very small pared with the actual TBC layer thickness in the near leadingedge region. The boundarylayer thicknesses to spallation depth ratios modeled in this study may be similar to that existing on actual airfoils.Smooth Surface Heat TransferHeat transfer coefficient tests were run for a smooth surface at three freestream turbulence levels. The local Nusselt number is normalized by the mainstream Reynolds number to obtain the Frossling number (Nu/). Figure 3 presents the spanaveraged Frossling number distribution for all three freestream conditions. The Frossling solution for stagnation region heat transfer for zero freestream turbulence intensity3 is also shown for parison. The Frossling number distribution increases with an increase in freestream turbulence intensity. There is a 50% increase with grid 2 (Tu≈%) and a 30% increase with grid 1 (Tu≈%), pared with no grid (Tu≈%) at the stagnation location.Heat Transfer with SpallationFigure 4 presents the detailed heat transfer distributions (Nu/) for all of the spallation locations at Tu = 1% and aspallation depth of cm. From the distributions for Tu = 1% it can be observed that the spallations closer to the leading edge (SI and S2) produce higher Nusselt numbers immediately downstream of the spallation pared to spallations farther from the leading edge (S3 and S4). The highest heat transfer region inside the spallation appears to be at different locations for each spallation location. As the spallation location moves away from the leading edge (SI to S4), the reattachment location inside the spallation appears to move closer to the downstream edge of the spallation. The detailed distributions also indicate low Nusselt numbers inside the spallation caused by the flow separation near the upstream edge of the spallation. The heat transfer distributions for higher freestream turbulence show similar trends. The detailed distributions for the spallation depth of cm (not shown) also show similar heat transfer patterns for all four spallation locations. Figure 5 presents the effect of freestream turbulence for each spallation location on the spanaveraged Nusselt numbers (Nu/) for a spallation depth of cm. The Frossling solution is also plotted for parison. The effect of freestream turbulence is clearly evident for each spallation location. Higher freestream turbulence produces higher Nusselt numbers inside and outside the spallation. For spallations SI and S2, the Nusselt number distributions inside the spallation show strong separation and reattachment effects. The thin boundary layer in this region may be the reason for higher Nusselt numbers downstream of the spallation for SI and S2. Spallations S3 and S4 may be in the region where there is minimal effect on the boundary layer. However, there is strong flow separation after the upstream edge of spallation S4 that causes a very low heat transfer zone immediately inside the spallation. The flow reattached inside the spallation closer to the downstream edge, reducing the separation at the upstream edge. Note that the Nu/Re176。.5 values upstream of spallation S2 appear to be enhanced. The Nusselt numbers upstream of spallation S2 are affected by the presence of spallation S1 on the other half of the cylinder. Spallations SI and S2 were on the same test cylinder on opposite quadrants of the cylinder. This may have caused the enhancement of Nusselt numbers upstream of spallation S2. However, in theory, spallation location is not expected to enhance upstream Nusselt numbers as clearly indicated for the S3 and S4 cases. Figure 6 presents the effect of freestream turbulence for each spallation location on the spanaveraged Nusselt numbers (Nu/) for a spallation depth of cm. The Frossling solution is also plotted for parison. The effect of freestream turbulence is clearly evident for each spallation location. Higher freestream turbulence produces higher Nusselt numbers inside and outside the spallation, as in the case for a depth of cm. For this case, spallations SI and S2 do not produce strong increases in Nusselt numbers downstream of the spallation. The Nusselt numbers inside and outside the spallations are nearly the same. However, spallations S3 and S4 produce strong variations inside and outside the is flow separation after the upstream edges of spallations S3 and S4, which causes low heat transfer zones immediately inside the spallation. However, in this case, spallation S3 ha slight variatio