Testing and validation of coal combustion prediction indices from conventional laboratory analyses

Abstract

This study reports on the testing and validation of combustion prediction indices derived from the conventional laboratory analyses of coal. This involved the actual testing of coal samples using the pilotscale combustion test rig, the Drop Tube Furnace (DTF), petrographic analysis and the normal conventional laboratory analysis. The indices covered in the study were Fuel Ratio (FR), Hydrogen to Carbon ratio (H/C ratio) and maceral-based indices. The anomalies encountered in the study were also investigated. The investigation was based on ten low-grade Bituminous coal samples (A-J). Each of the ten samples was subdivided into two; the bulk sample of approximately 1.2 tons was taken for the pilot-scale combustion tests and the remainder for the laboratory bench-scale tests. A portion of the laboratory sample was used for DTF and petrographic analyses. All ten bulk samples were fired in the combustion test rig. An attempt was made to keep all major combustion parameters constant for all samples. The fineness of all samples was kept at approximately 70% -75μm particles and the excess oxygen at approximately 6%. The burnout times of both the parent feed coals and their char products were used to test the indices. There was no clear correlation between the burnout times of the parent coals in the combustion test rig and the indices obtained from the conventional laboratory test results, viz. FR and H/C ratio. There was also no clear correlation between the burnout times 4 of chars in the DTF and the above indices. This can be attributed to the fact that the conventional laboratory analyses, on which these indices are based, are performed at lower temperatures as compared with combustion conditions in the combustion test rig and the DTF. However, the char burnout times in the DTF showed a strong linear correlation (R2 = 0.89) with the Fuel Ratio obtained from DTF volatile matter, which was corrected by subtracting the mineral volatiles. This relationship was not obtained from the burnout of parent coals in the combustion test rig. This proved that the high temperature combustible volatile matter relates better to high temperature combustion performance in the DTF. The petrographic nature and rank of most parent coal samples had a strong influence on the combustion performance in the combustion test rig. Namely, there was a strong correlation between the burnout times of the parent coals and the maceral-based indices, i.e. Burnout times versus vitrinite content showed a linear correlation (R2) of 0.91, burnout times versus Reactivity Index (RF) showed a linear correlation (R2) of 0.96 and burnout times versus Maceral Index (MI) showed an exponential correlation of 0.85. However, there was no clear correlation between the above indices and the char burnout times in the DTF. Of the ten coals tested, three showed consistently anomalous correlations including burnout results, i.e. coals E, H and J. 5 The relatively slow burnout time of Coal E (1.20 seconds) in the combustion test rig was not commensurate with the high vitrinite content of 42.2% and the volatile matter of 22.5%. The full survey of the analytical results indicated that Coal E, apart from having the highest proportion of -75micron fines in the feed sample (85%), the lowest Abrasive Index (22 mgFe), the highest Hardgrove Grindability Index (67) and possessing a significant proportion of weathered and oxidised particles, most notably exhibits a range of rank that extends from Sub-bituminous to well into the Mid Bituminous coking range of rank (0.5 to 1.3 RoVr%). This evidence indicates (a) that the coals which fall normally in the Low Rank Bituminous C range, have been heated relatively significant and (b) most importantly, have passed into the range in which vitrinite macerals become “cokified”. This infers that the vitrinite in the higher ranking coking range would soften, swell, become porous, fuse with other particles and then harden. During this process the texture of the walls of the gas pores become semi-crystalline, developing mosaic structures (a form of semigraphitisation) on exposure to high temperatures Thus, due to the high vitrinite content and extended rank into the coking range, it has been concluded that a significant proportion of the coaly material has become “cokified” and semi-graphitised. Such lower rank material, which is normally highly reactive to combustion in the presence of oxygen at lower temperatures, becomes inert above a specific temperature after undergoing severe molecular reordering and resolidification. This unusual condition, 6 in addition to the presence of oxidised materials in this coal, is considered to be responsible for the anomalously long burnout time of a supposedly reactive coal. Coal H exhibited a longer burnout time of 1.55 seconds in the combustion test rig. This has been attributed to the fact that this coal has a significant quantity of long-term weathered and oxidised material as indicated by the highest proportion of discoloured coaly particles (9.2%) along with the relatively high cracking and fissuring (20.6%) that arises with weathering. This is commensurate with the fact that this sample was derived from an old stockpile as indicated in Table 3.1. The long burnout time of this coal is considered to be due to the presence of weathered material which gives rise to limited volatile release, slow ignition and slower rates of combustion. Coal J possessed the shortest burnout time in the combustion test rig (0.75 seconds) but it also possessed, by far the highest proportion of coaly material exhibiting abnormal conditions (55.6% compared to the next highest value, 37% in sample H) of which 29.6% was heat-affected and 17.4% cracked and fissured and possibly desiccated. This sample also possessed a low vitrinite content (11.2%), the lowest MI value (0.028), the lowest volatile matter content (17.7% based on proximate analyses) and the widest range of vitrinite reflectance (0.5 to 2.2 RoVmr%) thereby confirming the extensive levels to which the coal had been burnt. The unexpectedly fast burnout time of what would otherwise have been considered a relatively slow 7 burning, difficult to ignite coal appears, in this case, to be attributable to the combustion response to high temperature exposure of the extensively heataffected, desiccated and cracked material in the combustion test rig. Thermal shock is therefore considered to have taken place, namely, material such as that described above has been known to explode and shatter into smaller particles when exposed to instant high temperatures. The result is the provision of small particles with higher surface areas which leads to rapid and efficient combustion. This process has been identified in other similar South African coals, as reported by Falcon (1992). This has been termed ‘deflagration’. The occurrence of the highest flue gas temperature as reported in the combustion test rig tests when burning this coal further confirms the presence of higher rank (burnt) coals. The results of this work therefore indicate that coal combustion performance in the combustion test rig is most closely correlated to the petrographic parameters, i.e. vitrinite content, Maceral Index (MI) and Reactivity Index (RI), except when coals are oxidised, burnt, high ash or liable to potential deflagration due to incipient cracking in the original coal. The conventional laboratory analyses, and the Fuel Ratio (FR) and Hydrogen to Carbon Ratio (H/C) derived from them, cannot be correlated with coal combustion performance in either the DTF or the large scale combustion test rig. Furthermore, the burnout results of the chars tested in the DTF cannot be correlated with the burnout results of the normal parent coals obtained in the combustion test rig. This is considered to be due to the differences in sample preparation prior to testing and to variations in combustion conditions between the two test units.