This paper outlines a new methodology for modelling caveability and subsidence using bi-directional coupling between the continuum code FLAC3D and the cellular automata code CAVESIM. FLAC3D, using the CaveHoek constitutive model, simulates the progressive failure and disintegration of the rock mass from an intact/jointed to a caved material. CAVESIM simulates gravity flow, in particular the collapse, bulking and movement of caved rock. The coupled method captures many important aspects of caveability affecting cave design such as hang-up formation, material recovery, timing of surface breakthrough or interaction with other lifts, crater development, and surface subsidence. The key to improved modelling of many of these aspects is the ability to accurately capture the impact of draw and gravity flow on cave propagation and subsidence.

Large magnitude seismic events are a concern for most hard rock underground mines extracting minerals at depth. At the Kiirunavaara Mine, a long-term chain of events was initiated with a 3.0 magnitude event in 2008 in production block 19 on mining level 907 m. In the aftermath of the event, this production block started to trail behind the adjacent mining blocks creating an upward pointing wedge over three blocks, with block 19 at the top. In 2020 another large magnitude event (4.2 Mw) occurred in the same block, now designated block 22, while opening level 1022 m. The event caused significant damage to the mine infrastructure over a large volume affecting overall mine output for months.Because of the significant damage to the infrastructure, it was not possible to resume mining on the topmost levels in block 22. Mining resumed instead from the lower levels, creating a remnant pillar between the existing sublevel and the resumed cave. The behaviour of this pillar has been analysed in detail using a coupled FLAC3D-CAVESIM numerical model. The modelling used a global-local approach where the mine scale stress field from sublevel cave mining until 2020 was superimposed on a local model where the resumption of mining activities in block 22 was explicitly simulated for several alternative layouts and sequences.To alleviate the seismic hazard from an uncontrolled pillar collapse after resumed mining, the best option sequence was sought to facilitate controlled pillar caving. Variation studies were performed to find high impact parameters controlling the pillar behaviour. Parameters studied included looking at rock mass quality, horizontal sequencing and alterations in the footprint of the resumed cave.From the modelling effort, it was concluded that the confinement of the remnant pillar was the major factor controlling the pillar caving and seismic potential. With support from underground damage assessments and the numerical models, it was decided to attempt to resume the mining from level 1079 m (abandoning the damaged levels 1022 and 1051). A new set of footwall drives were excavated inside the orebody to replace the damaged drives at the footwall contact. From the new footwall drives, the existing crosscuts were rehabilitated towards the old footwall drive as far as possible to reduce the confinement of the remnant pillar above. Resuming production at 1079 will attempt to reset the global mining sequence, with the production block no longer lagging compared to adjacent blocks.

The Southern Ridge Cutback 3 (STR3) at the Tom Price mine site will be the highest and steepest slope in Rio Tinto Iron Ore's Pilbara operation. Initial geotechnical assessment of the STR3 western slope using two-dimensional limit equilibrium methods recommended a substantial flattening of the design. This would have resulted in the deferral of 3.2 Mt of high grade ore. Given the good performance of the preceding STR2 cutback, it was considered that the two-dimensional (2D) analysis results were not representative of the expected stability and were overly conservative. Structures constraining the dominant mode of instability strike oblique to the slope. This aspect and the effects of 3D lateral confinement are not considered by 2D analysis. In order to address this, a 3D modelling project was initiated with development of a 3D model by Itasca Australia Pty Ltd.The 3D model method utilises both 3DEC™ (Itasca 2013a) and FLAC3D™ (Itasca 2012) software to develop a constitutive model that takes into account the dominant bedding anisotropy within the slope. As this was the first such model developed for Rio Tinto Iron Ore (RTIO), an external review board was appointed to provide technical guidance during the project. Sensitivities were carried out to address questions regarding in situ stress regime, pervasive joint orientations relative to bedding, potential for large scale wedges, and ore friability.Three-dimensional modelling results were favourable and indicated that the existing design exceeded stability acceptance criteria. In addition, further optimisation of the slope was possible and would realise an additional 1 Mt of high grade recovery. In order to achieve this, revision of the slope design configuration was required. Work supporting this included assessment of the viability of 90 degree batter face angles and a re-routed haulage design. Batter-berm configurations and placement of wide geotechnical berms were tested for inherent stability and rockfall risk management effectiveness.Overall stability of the revised slope design was confirmed by the Itasca modelling. Some areas of potential local instability were identified and have been addressed by detailed design changes. This project demonstrates the potential value add that can be realised by 3D analysis, when compared with traditional 2D methods. Due to the high value of the STR3 ore as a blending material, this slope is being mined at a relatively high strip-ratio when compared with other RTIO Pilbara pits. This emphasises the potential impact of the protect plan and optimisation outcomes.