Modelling spontaneous combustion liability of carbonaceous materials

Abstract

This paper presents predictive models to determine spontaneous combustion liability of carbonaceous materials (coals and coal-shales) using statistical analysis. The intrinsic properties and spontaneous combustion liability index were determined by testing 14 coals and 14 coal-shales from Witbank coalfields, South Africa. The relationship between these intrinsic properties (obtained from proximate, ultimate and petrographic analysis) and spontaneous combustion liability indices (the Wits-Ehac Index and Wits-CT Index) were established. The influence of the intrinsic properties of coal-shales in relation to coal properties affecting spontaneous combustion has been established using a statistical method. The linear regression analysis indicates better linear relationships between some of the selected intrinsic properties and spontaneous combustion liability index and thus, identifies the major intrinsic factors affecting their liability toward spontaneous combustion. It was found that a definite positive or negative correlation coefficient exists between the intrinsic factors and spontaneous combustion liability. A set of models to predict the spontaneous combustion liability was derived. The best significant correlation along with the most appropriate model as indicated by R-squared values, the coefficient of correlations and standard error was used to predict the incident of spontaneous combustion.