The ability to use cutting edge tools, such as numerical modelling, to predict seismically active volumes is a clear goal of today's geomechanical engineers. A knowledge of problem volumes in advance of drifting and production would lead to a safer working environment, where risk mitigating design can be implemented. While specific successful cases exist, a standard approach for numerical stress analysis of seismicity is not currently available in the literature.
This paper presents the use of numerical modelling to analyse seismically active volumes of Luossavaara- Kiirunavaara AB's (LKAB) Kiirunavaara Mine, a 4.5 km long, iron orebody extracted using sublevel caving. Crack initiation and slip along pre-existing discontinuities were evaluated using Itasca's FLAC3D software and compared to mine seismicity. Results were used to evaluate expected changes in seismically active volumes with planned production. Hanging-wall seismicity was correlated with the location of plastic failure in the models, as well as differential stress near the production front. Less clear relationships existed between footwall seismicity and model results, however, many orientations of discontinues have the possibility to slip, and therefore may contribute to seismicity. Patterns in orientations of discontinuities that can slip were consistent in the footwall drifts relative to the active production level, regardless of the depth of production evaluated. In general, the models did not indicate any expected changes in seismically active volumes with planned production.
It is undisputable that an understanding of the phenomena of seismicity in rock masses will improve mine safety. Although prediction of specific seismic events does not seem achievable (e.g. McKinnon, 2006), identification of areas that have a higher risk of seismic activity seems within reach. Numerical stress analysis models are the leading-edge tools to understand rock mass behaviour, and much effort has been put towards the modelling of seismicity (e.g. Diederichs, 2000; Andrieux et al., 2008; Beck et al., 2009; Sjöberg et al., 2011; Ghazvinian et al., 2014). However, to date, a standardised and accepted strategy to numerically model seismicity that successfully reproduces this behaviour for all cases does not exist.