The design and behaviour of crush pillars on the Merensky reef

Abstract

Crush pillars are used as part of the stope support in intermediate depth tabular mining stopes. Crush pillar design should ensure that the pillars crush when formed at the mining face. This behaviour of the pillars is typically achieved when the pillars have a width to height ratio of approximately 2:1. Once crushed, the residual stress state of the pillars provides a local support function. Crush pillars are extensively used in the platinum mines of South Africa. In most cases effective pillar crushing is not achieved, resulting in pillar seismicity. The objective of the research was to determine the parameters which influence crush pillar behaviour. A limit equilibrium model was identified as being able to simulate the behaviour of the pillars. The model implemented in a displacement discontinuity boundary element code provided insights into the stress evolution of a pillar depending on its position relative to the mining face, the effect of over-sized pillars, the impact of geological structures, layout and rock mass parameters as well as mining depth. An underground mining trial was conducted at Lonmin Platinum to measure and visually observe the behaviour of crush pillars. This was the most comprehensive monitoring of these pillars ever conducted in the platinum industry. The visually observed behaviour of the pillars agreed well with the findings of the measurements and the pillar fracturing profiles obtained at various stages of the pillar forming cycle. A sequence and mode of pillar failure could be identified. The results indicated that a pillar reaches a residual stress state when separated from the mining face. The pillar experiences secondary, subsequent reductions in stress when new pillars are formed. This unloading phase, has in the past, typically only been referred to as continued strain softening behaviour. However, it was found that at some point the pillars experienced no further reduction in stress whilst the pillars continued to deform. This observation was verified by convergence measurements. After all mining stopped, continued convergence was recorded. A numerical model was used to back analyse the behaviour of the underground trial site which consisted of an approximately 22 000 m2 of mined area and 55 crush pillars. To date, no numerical modelling of a mine-wide tabular layout, which explicitly included a large number of crush pillars, had been reported in South Africa. This work is therefore considered a major novel contribution to this field of research. After model calibration, both the observed and measured behaviour of the crush pillars in the trial site could be replicated. This was especially useful in evaluating the stress conditions measured above the pillars as well as the total amount of convergence experienced adjacent to the pillars and at the panel mid-spans. The findings validated the use of the limit equilibrium model implemented in a displacement discontinuity boundary element code to simulate the behaviour of crush pillars on a large scale.