Numerical modelling of mining subsidence

Subsidence due to the implementation of underground mining has caused significant damage to the environment. Subsidence is an important issue in the mining community and is caused largely by the longwall mining techniques employed to extract coal. Subsidence has always been associated with longwall mining, and has become the predominate issue. Due to the high extraction rate surrounding the longwall method, it invariably causes rapid subsidence within the geological strata (Booth et al. 1998). The stress fractures that occur as a result of the collapsed goaf, propagates to the surface producing a depression or dip in the soil profile. A case study was completed on a Rio Tinto mine called Kestrel which is located near Emerald, QLD. Being provided some the required information allowed for a numerical model to be generated in FLAC 2D. This project aimed to develop a numerical model within FLAC to accurately measure and model subsidence. This is due to the fact that numerical models are extremely important when dealing with large complex problems. The department of mineral resources requires numerical models for subsidence to accurately assess the viability of mining in that associated area. Through the use of FLAC 2D it was possible to develop a methodology that was applicable to all single longwall scenarios. The geometry and material properties of the model are the only values required to be changed within the methodology to suit a new model. The graphical interface in FLAC was used as it is user friendly and more applicable for initial analysis of subsidence of soils. This methodology can be extended to the use of other geotechnical subsidence applications. It was found that the results obtained from the model do not line up with the published work on subsidence. The magnitude of the subsidence of the FLAC model does not predict subsidence nearly to the magnitude that has been observed by (Keilich 2009). This was due to the FLAC program not being able to model discontinuities of the bedding layers effectively. The methodology, however was proved to be accurate as the models produced followed the profiles that of published work. This provides evidence to the fact that the material properties entered may need to be investigated further. For future work a ubiquitous model within the FLAC interface should be used as it has been found to model joints, discontinuities and bedding layers far more effectively. In summary, - The results obtained were not of the magnitude as provided by published works and real time monitoring data. - The methodology was proved to be accurate with very similar subsidence and displacement profiles. - FLAC cannot measure layers and bedding discontinuities effectively and therefore the ubiquitous constitutive model should be used for future analysis of this problem