Grid Turbulence in Dilute High-Polymer Solutions 

 2[ I I I I 



500 



FREQUENCY, cps 



Fig. 19 - Spectral level shift for x/M = 10.7 

 due to a 30% viscosity increase, compared 

 with the predicted shift (dashed curve) fronn 

 Eq. (5) using values of m determined from 

 Fig. 18 



was lost in the noise. A detailed comparison shows that there are no significant 

 spectral changes other than those which can be attributed to raggedness. The 

 estimated viscosity increase was so small that the expected spectral shift due 

 to viscosity increase was insignificant. 



For higher concentrations of Polyox, two other types of spectral shift were 

 observed. The first type was a greater scatter in spectral measurements near 

 the grid. A possible cause of such behavior could be temporary sensor sensi- 

 tivity loss, due to intermittent coating. However, no abrupt changes of signal 

 level were ever observed. Another possible cause is poorer mixing of the wakes 

 of the grid bars, since the diffusion of turbulent wakes has been shown to be re- 

 duced even by low Polyox concentrations (10). The grid-turbulence experiment 

 is not appropriate for a study of the wake mixing, for which a large increase in 

 the number of runs would be necessary. With older solutions or greater x/M the 

 scatter of spectral measurements was the same as the scatter in water. 



The final type of spectral shift in the more concentrated Polyox solutions 

 was that due to the viscosity increase. An example of this type of shift can be 

 seen with Fig. 20, which gives the relative spectral curves for water and Solu- 

 tion F-5 at 66 hours, both for x/M = 39. Considering Fig. 6 which shows wave- 

 forms for x/M = 20.2 and 72 hours, and noting the similar effects of grid dis- 

 tance and solution age, we can assume that the waveforms in Fig. 6 for x m = 

 20.2 would have also applied, except in amplitude, to x m = 39 at 66 hours. Thus 

 no spectral effects due to raggedness are expected. In Fig. 20, the curve for 

 the solution departs from the data points due to the usual noise correction, and 

 the curve is stopped when the correction becomes 4 db. When correction of 



63 



