A probabilistic assessment method for sinkhole subsidence in the Northern Coalfield, NSW

Sinkholes occurring due to a progressive collapse of roof and superincumbent strata over shallow abandoned coal mine workings are a significant hazard to surface infrastructure in the Northern Coalfield of NSW. This is due predominantly to large areas of land, initially exploited for the extraction of coal, being progressively developed following exhaustion of the available coal reserves. As a result of the extensive utilisation of the bord and pillar mining method, an ongoing potential for surface subsidence due to progressive roof collapse remains, decades or centuries following the cessation of mining.

When compared with planned subsidence such as that occurring from near total extraction methods (such as longwall or mechanised pillar extraction), unplanned sinkhole subsidence is challenging from an engineering perspective. This is in part due to the unknown likelihood of sinkhole occurrence, requiring an assessment of sinkhole potential within a specified timeframe, as well as consideration of the potential surface effects and subsequent impacts on surface infrastructure that may result if it does occur. This challenge is increased by the low confidence in predictions of what the potential surface effects of a sinkhole may be, which by virtue of being a subsidence mechanism limited to shallow coal mines, has a higher inherent sensitivity to geological structures and localised variations in lithology.

This paper presents a review of the historical progress in the assessment and treatment of sinkhole risk in the Northern Coalfield of NSW, beginning with European settlement and the development of the first coal mining operations. It represents the results of a major 6-year study, completed in 2024, comprising the largest combined database of individual sinkhole events in NSW. The results are presented together with a probabilistic assessment method that is able to both quantify the likelihood of sinkhole occurrence for a variety of surface and mining conditions, as well as the likelihood that a sinkhole will be less than a certain maximum width. The resulting coupled probability model can be used to determine the annualised and total probability of exceedance for a variety of design sinkhole events over a given design/service life.

Finally, additional research and data developed following the completion of the initial study is provided, giving clarification on the effective use and application of the method, as well as the reliability of the probability distributions produced by the model with respect to particular environmental conditions.