Predicting Coal Pillar Life In South Africa

Merwe, J. N. van der
Organization: The Southern African Institute of Mining and Metallurgy
Pages: 10
Publication Date: Jan 1, 2003
The data base of failed coal pillars in South Africa was analysed with regard to the time of failure in an effort to develop a method to predict how long after their formation coal pillars will fail. It was found that meaningful results could only be obtained by splitting the data base up into groups that displayed similar behaviour. It was postulated, based on observation and experience, that coal pillars fail by a process of progressive weakening progressing into the pillar from the edges. Minimum pillar dimensions to ensure failure were calculated?these were the dimensions that corresponded to a minimum ?real? factor of safety. From the scaling distance required to reach the minimum safety factor, rates of scaling for the failed pillars were calculated based on the observed real lives of the pillars. It was found that the rates of scaling were unique for different mining areas and that they correlated directly with mining height and inversely with time. Using the derived scaling rates, the predicted pillar lives were back calculated for the pillars in the data base of failed pillars. In two of the three groups, the predicted lives correlated well with the observed lives on a one-to-one comparison. In all cases, the predicted lives were in the same time range as the observed lives. Overall, the frequency of predicted lives closely replicated the observed frequency of observed lives of the pillars in the data base of failed pillars. The method was successful in distinguishing between the data base of failed and a representative sample of intact pillars. While the predictions matched the observed lives of failed pillars, the predictions of expected lives for intact (hitherto unfailed) pillars were orders of magnitude greater. It is now possible to estimate the frequencies of future pillar failures for the purposes of mine closure and undermining of surface structures within broad bands.
Full Article Download:
(1211 kb)