Abstract
Abstract In this paper, a probabilistic design chart for coal pillar based on South African coal mine experience is proposed as a tool capable of estimating the pillar strength. A methodology based on the logistic regression is employed to identify failed and intact pillars with their corresponding probabilities of occurrence. The input parameters used for pillar design include the mine depth, pillar width, bord width, height of mining excavation and panel width. The results showed that 84.6% of the pillars, including stable and failed categories, were correctly classified. In addition, the probability associated with the pillar conditions was determined and plotted in form of a pillar strength graph. The results were in good agreement with the field data and existing studies. One of the merits of the current study is that the probability of having a coal pillar intact or failed can be easily determined. This is particularly useful for design purposes since a series of uncertainties are usually involved in collecting the rock mass parameters. It is concluded that the results of this study could improve the empirical design of coal pillars.