This study describes a review of the pillar design at the manganese mining operations in South Africa with specific reference to Nchwaning III Shaft at Black Rock Mine Operations. From a literature study, it is clear that there is uncertainty regarding hard rock pillar strength and the
appropriate formulae to use when designing hard rock bord and pillar layouts. The current design at Nchwaning III Shaft is based on the Hedley and Grant power-law formula and a K-value of 133 MPa. This K-value is currently used for all the shafts and all the different geotechnical areas. As a concern, extensive scaling of some of the pillars has been observed by the mine. Thin spray-on liners are currently used to stabilise these pillars and this practice has become a major operational expense. The literature survey indicated that extensive research has been conducted on the time-dependent failure of coal pillars in South Africa. These studies are exclusively empirical in nature and no attempt was made to conduct numerical modelling of the time-dependent scaling. In contrast, almost no work has been done on the time-dependent scaling of hard rock pillars. Complex numerical modelling to simulate the time-dependent scaling of hard rock pillars is described in the literature, but these models are difficult to calibrate and the results are typically not compared to underground pillar behaviour. Extensive underground observations and monitoring of pillar behaviour was conducted by the author in six different areas. The observations indicated that significant scaling of the pillars occur in some areas, while other areas are remarkably stable in spite of the slender pillars (width of 7 m, height exceeding 5.5 m). From the data collected, it is evident that the pillars with a low rock mass rating, which contain many closely spaced joints, are prone to excessive scaling. The pillar design at BRMO therefore needs to be extended to cater for the geotechnical areas with a low rock mass rating. This is an important new finding of this study. Numerical modelling was conducted to simulate the pillars stress in the areas of interest. The TEXAN displacement discontinuity code was used for this study as it was developed specifically to simulate a large number of small pillars in tabular layouts. The numerical modelling provided valuable data on pillar stress and variations of this stress based on pillar
size and position relative to abutments. This is a significant improvement over tributary areas stress calculations. The numerical modelling allowed a back-calculation of the “minimum” K-values for the pillar strength formula in the different areas. Values exceeding 100 MPa were
calculated for some of the stable pillars. A limit equilibrium constitutive model in the TEXAN code was used to simulate the pillar scaling observed underground. It was encouraging that the two different types pillar behaviours could be simulated using this approach. More precise
calibration of the model will be required in future, however, before it can be used as a design tool. Monitoring of selected pillars were also conducted by the author over a period of three months to quantify the rate of time-dependent scaling. Contrary to expectations, almost no additional scaling was recorded for the pillars over a three-month monitoring period. The existing scaling distances of the pillars could be measured, however. The amount of scaling for pillars mined three years before the measurement in the areas with a low rock mass rating was on average 0.8 m. This value was 0.5 m for pillars mined approximately five months before the measurements. It therefore seems as if the bulk of the scaling occurred soon after blasting and it may even be attributed to the effects of nearby blasting. A limited amount of additional time-dependent scaling seems to occur after this. Prominent time-dependent scaling is nevertheless present for some pillars as can be seen by the ongoing deterioration of pillars that were reinforced using thin spray-on liners. Numerical simulations of the time-dependent scaling were conducted using TEXAN and a limit equilibrium model. An exponential decay of rock strength at the edges of the pillars resulted in simulated behaviour that is qualitatively similar to underground observations. Similar to the actual pillar behaviour, the rate of pillar scaling becomes small after a long period of time.
Based on the data collected, the numerical modelling results and the observations that the pillar strength is affected by rock mass rating, it is proposed that the following revised pillar design should be adopted at Nchwaning III Shaft:
• Retain the current pillar design for areas where the RMR > 50. This rock mass rating is typically found in the high-grade areas.
• Adopt a revised pillar design for areas with an RMR < 50. This rock mass rating is typically found for the low-grade R5 product areas.
For the revised pillar design in low grade areas (RMR < 50), a new K-value was estimated using principles similar to that proposed for the design rock mass strength (DRMS). A value of K = 90 MPa was derived as a first estimate. This value is confirmed by the numerical modelling and the back-calculation of K-values for pillars subjected to extensive scaling. This revised design for areas with an RMR < 50 must be trialled at the mine. Monitoring of the pillar condition in these areas is required to determine whether this design ameliorates the scaling problem and negates the need for thin spray-on liners. Additional future work should include an improved calibration of the time-dependent limit equilibrium model and a better quantification of the rate of pillar scaling. This will involve the monitoring of pillars over a longer period of time.