transient pressure fluctuations sufficiently strong to create liquid 

 tensions sufficient for microbubble growth. From the data of Figure 13, 

 for example, a fluctuation magnitude of about ten percent of the dynamic 

 pressure would be required to establish cavitation on the hemisphere body. 

 These fluctuations would be expected to occur at the reattachment of the 

 separation region, as can readily be appreciated from Figure 15. Rather 

 surprisingly, the literature is very scarce on this feature of real fluid 

 flows. It was possible, however, to install a single point pressure 

 transducer on the hemisphere body of Figure 15 (Arakeri 1975b). The 

 results were most interesting and show that rms and peak fluctuations are 

 certainly large enough in themselves to cause liquid tensions in the 

 regions of observed inception. Later, these findings were confirmed, with 

 even more dramatic and precise results on a larger body, by Huang and 

 Hannan (1975). Some of their findings are shown in Figure 16, to demon- 

 strate that the fluctuating pressure environment is sufficiently strong to 

 make it very likely for cavitation to occur on these bodies at local pres- 

 sures slightly exceeding the liquid vapor pressure. 



From these findings then, it is tempting to account for these tran- 

 sient fluctuations by putting the inception index in the form 



+c (11) 



'loc P t 



where c is a local pressure coefficient where inception is observed, 



P loc 

 and c is the fluctuating pressure-coefficient amplitude there. This type 



P t 

 of representation indeed makes the hemisphere data much more plausible be- 

 cause the c , measured near the end of the laminar separation bubble, 



P loc 

 (and perhaps the fluctuating component too) is strongly Reynolds number 



sensitive (Huang and Hannan ibid). This prescription ignores the physical 



state of the fluid, however, so that some caution should be exercised in 



seeking correlations based solely on Equation (11) , and does not explain 



the contrary finding of Figure 14, namely that a. < -c or -c 



s reattach 



28 



