f m = measured flux (g/cm 2 /ndn) 



f = predicted flux (g/cm 2 /min) , calculated using the threshold 

 power fit equations and midflow speeds 



N - number of data points 



92. The results of this analysis are presented in Table 8. Values of 



S y are considered reasonable, with the SUPERDUCK nozzle and 2.5-min basin 



data having the highest error of estimate, followed by the 7.5-min basin data, 



and the DUCK85 and C nozzle results, respectively. If values of predicted 



flux f p are taken to represent the true flux, then the error of estimate 



gives a measure of nozzle reliability. If transport processes in the vicinity 



of a particular nozzle occurred similarly for all flow conditions, then the 



nozzle would be easily modeled with a single equation, and the standard error 



of estimate would be lower in value. Transport processes with the SUPERDUCK 



nozzle were variable, with scour occurring approximately half the testing 



period. Scour consistently occurred at the DUCK85 nozzle; hence the lower 



error of estimate value. The low value of S y calculated for the C nozzle 



reinforces the observed consistency in transport processes in the vicinity of 



the nozzle. 



Table 8 

 Experiment Variability for Nozzle and Basin Data 







Standard Error 









of Estimate 



to. 025 



Number of 



Tvpe of Test 



g/cm 2 /min 



Statistic 



Data Points 



2.5-min Basin 



0.35 



2.306 



10 



7.5-min Basin 



0.22 



2.262 



11 



DUCK85 



0.20 



2.776 



6 



SUPERDUCK 



0.35 



2.064 



26 



CUBE 



0.17 



2.365 



9 



93. Assuming that the measured fluxes are normally distributed about 

 their respective mean values, the appropriate t- statistic can be utilized with 

 the standard error of estimate to obtain confidence limits for the linearized 

 empirical relationships, as follows: 



65 



