| dc.description.abstract | Coal blends are increasingly utilised at power plants with significant savings and without breaking environment regulations. However, evidence of interaction among the coals requires the study of some parameters that affect combustion efficiency and related opacity of emissions. Actual plant data was available for the combustion of five families of binary blends (single coals and approximately 25%/5%, 50%/50% and 75%/25% blends) with variable contents of ash, volatiles and maceral composition. Size distribution of particles was determined for the coals fed to the plant boilers and the fly ashes, as well as for unburnt carbon in the latter. The almost homogeneously sized feed from different coals generates a size distribution in the fly ash where 250-mum particles vary up to 1.29%, while particles smaller than 38 gm vary between 21.74% and 62.41%. Unburnt carbon increases with size of ash particles from a maximum of 12.2% for fractions smaller than 38 mum up to 73.9% for the fraction bigger than 150 mum. Total content of unburnt carbon in the fly ash from combustion of coal blends show deviations from the expected weighted average of the constituent coals (K, L, T, P, F, S and N). These deviations are related to maceral composition and rank based on reflectance values. The smallest deviation is shown by the blend (T/P) with coals having low values of reflectance and homogeneity of maceral contents. Larger deviations were found for blends K/L, P/F and S/N with higher difference of rank and greater heterogeneity of maceral composition. The K/L, R/N and S/N blends show positive deviations with respect to the expected weighted average, that is, blending was detrimental to the combustion efficiency, while blend P/F showed an enhance of the combustion efficiency as measured by unburnt carbon in the fly ash. A "reactive maceral index" introduced in this work plays a useful role. If a ratio of reactive maceral index is established for a binary blend as the ratio of the respective values of the index for the single coals, then the magnitude of the deviation from the expected average is proportional to the ratio of reactive maceral index of the blend. Thus, the K/L and P/F blends, with values of 4.6 and 65.6 for the ratio of reactive maceral index, differ significantly from the expected average, while blends T/P and R/N, whose binary families have a ratio of reactive maceral index close to 1, show a behaviour (in terms of unburnt carbon) which closely approach the expected weighted average. (C) 2002 Elsevier Science B.V All rights reserved. | |
