Ash Fusion Temperature Prediction Model And Test Based On Coal Ash Mineral Properties Of Blended Coal

One of the main factors affecting on the safety operation of the boiler and coal gasifier is coal ash slagging. The potential security risks such as slag formation and slag block may be happened in the process of burning blended coals because of inappropriate blending methods, kind of coals and blend ratios. So in order to solve the security problems, it is necessary to carry out the ash fusion point measurement, prediction and analysis.The analysis of ash fusion temperature(AFT) effecting factors including reaction atmosphere and coal ash chemical composition was summarized in this article. The AFT experiments in weak reducing atmosphere were carried on, and the results of experiments provided the evidence of slag formation and slag block. And the prediction model for AFT of blended coal, including mineral composition model and linear prediction model, based on coal ash mineral properties was established, then the prediction results comparison of two models was analysed to ensure safety operation. Mineral phases at a special temperature or variations of mineral phases in a certain temperature range could be got by mineral composition model. And based on mineral composition model, the fusion points of mineral phases were introduced to build the functional relation of AFT and mineral phases. The correction values of fusion points were obtained by linear regression.The prediction results from the comparison between the mineral prediction model and support vector machine model show us that the results of the former are better than the latter. The linear correlation coefficient r are all above 0.85 and the mean absolute percentage error are about 3%. So, the mineral prediction model has good prediction efficiency for AFT. Not only that, the correction values of mineral fusion points make the relationship between mineral phases and AFT be simple and clear, and the mineral prediction can depict the coal ash fusion characteristics more directly. As well, the mineral prediction model has good precision and could provide guidance for AFT test and blending plan in practice.