Spottiswoode, SM, Linzer, LM & Majiet, S 2008, 'Energy and Stiffness of Mine Models and Seismicity', in Y Potvin, J Carter, A Dyskin & R Jeffrey (eds), SHIRMS 2008: Proceedings of the First Southern Hemisphere International Rock Mechanics Symposium, Australian Centre for Geomechanics, Perth, pp. 693-707, https://doi.org/10.36487/ACG_repo/808_152 (https://papers.acg.uwa.edu.au/p/808_152_Spottiswoode/) Abstract: We analysed seismicity in two deep gold mines in terms of strain energy release during mining. Mining took place with regularly-spaced dip pillars and backfill for regional support. The mines were chosen to represent mining of the Carbon Leader reef (CLR) and the Ventersdorp Contact reef (VCR). More than 500,000 m2 of reef were mined over several years and generated more than 10,000 events of magnitude greater than 0.0 in each case. Most seismicity was located close to faces that were actively mined, at least within the assumed location error of 50 m. Software was developed to associate seismic events with faces that were mined during the same month, or if not close to active faces, with stationary faces or abutments. Areas of mining that were active in each month and that were more than 100 m from any other active mining were grouped together to form over 300 discrete seismogenic regions, or polygons for each month of mining. Mining and seismicity data were cumulated in various ways including cumulating values for data sorted by increasing values of energy release rate (ERR) computed numerically and averaged over each polygon. The most obvious result was that cumulated seismicity was more clearly proportional to cumulated strain energy release than to cumulative area mined, supporting the use of minimisation of ERR as an effective rockburst control method when ERR is calculated from the ratio of strain energy release to area mined. The linear relationship between cumulated seismicity and cumulated strain energy was maintained even when the reef was assumed to yield at normal stress of 300 MPa and more. It was only when modelled pillar strength was reduced to 250 MPa or less at both mines, that the models released excessive strain energy release at high values of ERR. It is possible that the design criterion of 400 MPa for pillar strength applied on one mine was conservative. The role of geological structures on seismicity was not the focus of this paper as mining in the vicinity of these structures took place at all spans and therefore at all values of ERR. In a separate study we showed that geological structures contribute 50% or less of the total amount of seismicity. The apparent stress and the ratio of seismic moment derived from P-waves to that derived from S-waves were found to vary with ERR, with both decreasing with increasing ERR for data from mining of the CLR, while the apparent stress increased with increasing ERR for VCR mining, as expected for the generally increasing levels of stress ahead of the advancing faces. We suggest that the decrease in values for the CLR data might be attributable to a more extensive fracture zone resulting from bedding-plane slip than occurs around the VCR with its massive lava hangingwall. We propose a simple softening model to explain the CLR behaviour.