OCR Text |
Show loss in the presence of a narrow distribution is not significantly different from the values for uniform excess air (<rEA = 0). A relatively narrow distribution of excess air, such as that shown in Figure 5, has little influence on carbon burnout at high average excess air. The effect of further increasing the width of the excess air distribution is shown in Figure 7 by the carbon loss curves for standard deviations of 7 and 14%. The increase from 7 to 14% approximately doubles the calculated carbon loss at 25 to 30% excess air, but the loss is still far lower than observed. Carbon loss is insensitive to the width of the distribution of excess air as long as the average excess air is high. Suppose that inleakage of air into the convection section causes the furnace to be richer than expected from the flue gas oxygen measurement. Imagine that the burners are closely controlled, so each delivers the same ratio of air to fuel, and that mixing in the flame zone is very good. The furnace gas is then richer than expected, but has uniform excess air throughout. Consider the case in which this effect is great enough to be entirely responsible for the discrepancy between B& W' s experience and the observed carbon losses. The horizontal line in Figure 7, originating at the carbon loss data point associated with the size distribution in Figure 3, shows, at its intersection with the curve for <rEA = 0, the excess air required to satisfy this condition, 3.6%. A calculation of the size distribution of unburned carbon for this case is compared with the measurements in Figure 9 (dashed curve). Even at such low excess air no char particles smaller than 90 Jlm are predicted to be present in the furnace exit gas, in the absence of any deviations of excess air from the set point. Now, imagine that inleakage is not limited to locations downstream from the furnace, but also occurs through the furnace enclosure as well. If there is any inleakage to the convection section the average excess air in the furnace is still lower than indicated by the flue gas analysis, although less so than in the case considered above, but the width of the distribution of excess air is increased because air leaking into the furnace is poorly mixed with fuel. Two combinations of average excess air and its standard deviation which might produce the observed carbon loss are shown in Figure 7: (1) average excess air = 5.3% with standard deviation = 3.5%, and (2) average excess air = 15.4% with standard deviation = 14%. These distributions are shown in Figure 8, and compared with the case of uniform excess air considered above. The calculated size distributions of unburned carbon in the 15 |