16 



liquid is sufficiently precise for most engineering applications, in the lab- 

 oratory and in some technical applications, it has become important to be able 

 to predict more precisely the pressure at which cavitation may begin. It is 

 clear that the problem of inception in a flowing liquid may be much more com- 

 plicated than that in a fluid at rest because of the uncertainties both of the 

 role of turbulence and of the processes in the presence of dissolved and en- 

 trained air and solid particles. 



SOME REMARKS ON THE ROLE OP TURBULENCE IN CAVITATION INCEPTION 



With regard to the role of turbulence in cavitation inception, it is 

 of some interest to know whether the velocity fluctuations in a turbulent flow 

 can give rise to pressure fluctuations that are large enough to bring about 

 momentary inception at a mean pressure higher than that required by a pre- 

 diction based on average velocity. An estimate of the effect of turbulence 

 can be made in the following way. Suppose that for a given system, it has 

 been predicted that cavitation will begin when the cavitation number 



*=^-£ [6] 



IpU 2 



drops to a predetermined value. Here p is the local pressure. With a turbu- 

 lent pressure fluctuation, the instantaneous cavitation number will be 



, _ P - (p + P') _ Pi. m 



Thus, depending on the value of p ' , inception may occur momentarily at a val- 

 ue of a higher than that predicted as critical. It remains, then, to estimate 

 the magnitude of p'/^pU 2 . 



From computations based on models of isotropic turbulence,* G.I. 

 Taylor 32 found that 



[8] 



where u' is the local velocity fluctuation. The value of K as given by Taylor 

 and by Green 33 for a number of models, is of the order o f one . Now for a 

 Gaussion distribution, p' may be e xpected to be about 3yP' 2 for a small 

 percentage of the time and 2yp' 2 for a rather large percentage of the time. 



*The writer is indebted to Dr. J.V. Wehausen for his suggestion of these results in estimating the 

 pressure fluctuations in preference to a method originally used by the writer. On the basis of similar 

 considerations, Wehausen had also reached the conclusions given herein. 



