0.010 ,- 







A Rothermel's fuel bed, data 

 o Burning basket data 



0.01 0.02 



0.03 0.04 0.05 

 Packing ratio 



0.06 0.07 0.08 



Figuve 4. — A plot of the max-imum load-loss rate, w, versus the packing ratio, for 

 excelsior burned in a basket and in a fuel bed. The continuous curve was obtained 

 from the expression for w in appendix A. The upper and lower extreme are the standard 

 deviation from the mean. 



DISCUSSION 



The data contained in tables 1 and 2 for the two different methods of obtaining 

 the maximum load-loss rate do not lend themselves to a simple comparison at the same 

 packing ratios. However, assuming that a quadratic model fits these data, we can 

 approximate each data set with a quadratic regression curve, w regressed on 6, and com- 

 pare these curves to a quadratic regression of the composite of both data sets. The 

 composite curve expresses fuel loss rate as accurately as the individual data sets if 

 the composite curve has a sum of squared errors (SSE) (squared differences between 

 observed and predicted from quadratic model) not significantly greater than the sum of 

 SSE's from the separate curves. It then follows that the curves are not dissimilar. 

 A statistical test of the data from tables 1 and 2 (appendix C) shows that it is very 

 unlikely that the two data sets are different. 



The early load history (fig. 3) of the burning fuel follows a declining sigmoid 

 curve. Shape of the curve varies with packing ratio and is distinguished from other 

 curves through the maximum load-loss rate occurring at the inflection. It is important 

 to determine if the depth was sufficient to achieve the characteristic maximum load-loss 

 rate for each packing ratio. Fuel consumption efficiency is a means of examining this 



6 



