| OCR Text |
Show indication of the overall mass loss. This analysis is based on a mass balance around the reactor and depends on all of the sample being collected, unlike the analyses using the tracer. Systematic errors in the mass balance can only occur if all of the sample is not collected on the filter . This would have the effect of underestimating the residual mass in the char particles. The anticipated trend in errors from the mass balance is in the opposite direction of those for errors due to the tracer analysis. When the two analysis techniques agree, there is strong evidence that both are correct. Figure 1 illustrates the computed residual mass for each of the sampled data points by each of these methods and mean value of each of the estimates for particles with an initial size of 106-125 j.Lm. The top portion of the figure illustrates these data for the 6 percent oxygen environment and the bottom half for the 12 percent oxygen environment. The data are reported on a mineral-included basis, with the final mass fraction of mineral or ash shown in the lower right-hand corner of both figures . The initial concentrations of each of the elements in the coal were measured twice. The small amount of variation in the initial mlmo point shown at a residence time of 0 ms refiects the variation in these measurements. The remainder of the data refiect the total statistical variation, including variation due to sample inhomogeneity, random error, measurement error, and sample to sample variation. In many cases, the individual data points cannot be seen because they fallon top of one another. The residence time is calculated from the time the particles pass through the fiat fiame in the reactor and is based on measured particle velocities for a 100 j.Lm particle. Table 1 summarizes the statistical confidence intervals for the data illustrated in Figure 1. The symbols Tr , j.L, and (j represent residence time, mean value, and standard deviation, respectively. The table shows that the fraction of the mass remaining in the char is typically measured within ± 0.04 at 95 percent confidence for these data. The method of calculating this interval is outlined in standard statistics texts [Canavos, 1984]. The same experimental procedures were repeated using the smaller particles (75- 106 j.Lm) as feed stock. Similar results were obtained in terms of overall mass loss. These will not be illustrated here. But the calculation of iron mass loss determined from these measurements will be illustrated later. 4 Determination of Iron Mass Loss from the Char Particles The evolution of the elemental components of the char can be studied individually once the remaining fraction of the original particle mass has been determined for each sampling point. The fraction of the original amount of a given component which 5 |
