The Southern African Institute of Mining and Metallurgy Slope Stability 2015 B.J. Hutchison, S. Naude, and J. Howarth Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia B.J. Hutchison*, S. Naude, and J. Howarth *Hillgrove Resources IDS Maptek Hillgrove Resources operates the Kanmantoo Copper Mine on the eastern side of the Adelaide Hills in South Australia. Three open pits were in operation when it was recognized that the south-eastern wall of the Nugent Pit was susceptible to a flexural toppling failure. Mining personnel were able to mine for five months without disruption as flexural toppling reverse scarps slowly developed. Slope monitoring during that time comprised surface extensometers and prism readings. On 5 October 2014 a ramp blast located at the base of the 45 m high slope caused a sudden jolt of movement, resulting in an immediate requirement to alter the pit design and the monitoring programme. These alterations allowed for the continued mining of the deposit as the flexural toppling failure developed and finally collapsed. This paper describes the failure mode and the management/monitoring regime adopted to continue mining with minimal disruption and ore loss. Initial monitoring was with crack extensometers and prisms; supplemented by an I-Site 8820 laser scanner trial being undertaken at the time. Following the blast, the main thrust of the management strategy to contend with the potential failure was the deployment of an IBIS slope stability radar for providing alerts in the event of progressive movements leading to slope failure. The use of an inverse velocity plot based on radar data allowed for the prediction of the final collapse from two days out to within two hours. Early detection of slope movements and precise identification of the onset of the failure as a result of the radar allowed mining personnel to plan and implement appropriate actions with sufficient notice to minimize the effect of the failure on personnel safety and mine productivity. Introduction The Kanmantoo copper deposits have been intermittently mined since the mid-1800s, when Cornish miners first exploited the oxidized copper lenses in shallow workings. In the 1970s the Kanmantoo open pit was developed to a depth of 115 m before a fall in the copper price shut operations. Hillgrove Resources reopened the mine in 2011; initially developing the Spitfire Pit and then expanding the original Kanmantoo pit to what was called the Kavanagh Pit (Figure 1). These two pits were mined in a similar massive rock mass domain (geotechnical domain GDM1) that allowed for three relatively steep walls (north, east, and south) and a moderately dipping west wall; the latter being susceptible to planar and wedge crest failures on a bench scale. 81
Slope Stability 2015 Figure 1 Kanmantoo mine site plan As the Kavanagh Pit was to reach the planned depth of 230 m towards the third quarter of 2014, two smaller satellite pits were started in early 2014. These two pits, called the Emily Star and Nugent pits, were to provide ore until a new life-of-mine cutback of the main Kavanagh deposit (referred to as the Giant Pit) was sufficiently developed to supply all the ore for processing. These two satellite pits are both located partially or fully outside the massive rock mass regime (GDM1) of the original pits (Figure 2 and Table I). It was on the eastern side of the Nugent Pit in GDM2 that a 40 m high toppling failure occurred in more ductile rocks. 82
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia Figure 2 Kanmantoo mine site cross-sections 83
Slope Stability 2015 Table I. Geotechnical domains The potential for a toppling failure was recognized several months before it occurred. Conventional surface extensometers and prisms were initially used to monitor the slope. A Maptek I-Site 8820 laser scanner (and accompanying Sentry software) had been deployed in the pit as a research project and also provided intermittent coverage of the slope. Following a toe blast that caused up to 58 mm of movement at the prisms on the crest, an IBIS slope stability radar was deployed to provide critical monitoring capabilities to allow the management of continued mining directly under the potential failure. This paper describes the toppling failure characteristics, the associated monitoring results, and the management strategies adopted. Nugent Pit design stages and rock conditions The Nugent Pit design was planned with an initial mini-pit phase while awaiting regulatory approval to expand to the east to allow for a cutback to deepen the pit to its final depth (Figure 3). The eastern side of the Nugent Pit is located in a highly foliated, relatively weak biotite schist rock mass overlain by a stiff red clay unit up to 15 m in depth, which peters out to the south. As most of the exploration for this pit was carried out with reverse circulation drilling, very little geotechnical information was available. It was not until the first phase of mining when the pit was excavated 30 40 m below the eastern crest that the true nature of the biotite schist was exposed. Although the rock mass did not feature the numerous foliation plane shears described in the Savage River flexural toppling failures (Hutchison et al., 2000) it was recognized that flexural toppling was the most likely failure mode that could develop, due to the 70 75 easterly dip of the foliation into the wall. The Savage River experiences showed that flexural toppling was generally a very slowdeveloping phenomenon taking many months before ultimate collapse of a slope. Such developments were easily 84
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia managed by crest extensometers and prism monitoring. At Savage River the final delays to mining were restricted to a week or so when the final collapse and re-establishment of ramps occurred. The Savage River slopes were managed, using horizontal drain holes, up to heights of 90 m without a final collapse of the walls, although these occurred on intermediate slope angles in the order of 37. Figure 3 Nugent Pit east and southeast wall sections Based on the new Nugent rock exposures, the relatively short pit life, a regulatory need to backfill the final pit, and senior staff experiences with flexural toppling failures, the decision was made to continue with an aggressive design in the new cutback once regulatory approvals were obtained. The intent was to get in and out as quickly as possible. Still somewhat restricted by regulatory boundaries and weak rock conditions, the eastern wall slope design was altered to comprise a 50 continuous slope broken up by the main ramp (Figure 3) such that the maximum intermittent 85
Slope Stability 2015 slope height was restricted to 42 m. The new design was to be transitioned from 50 to 54 into the southern wall, over heights of 42 m to 70 m. The batter configuration on the south wall comprised 24 m high 60 batters with 8 m wide berms. The overall angle of the transition slope where the toppling event occurred was 52. The transitional SE wall was mainly located in the GD3 regime of ductile biotite schist. Further to the south was the contact to the massive rock with the GDM1 regime. The topple event was restricted to GDM2. It was also known that the groundwater table was at least 70 m in depth at the southern end of the pit. Cutback mining and subsequent wall movements In July 2014 regulatory approvals were obtained allowing the commencement of the new eastern wall cutback, following the continuous 50 slope design. As the cutback progressed, cracking behind the east wall crest was noted but this was not considered to be anything out of the ordinary in a stiff clay scenario. At the top of the SE transition wall some crest and upper batter dilation and cracking had previously been observed. This was being monitored with crack extensometers, prisms, and the I-Site laser scanner (a more detailed description can be found in Hutchison and Howarth, 2015). The laser scan data indicated that the main area of dilation was restricted to an area within 15 20 m of the pit crest; where up to 60 mm of movement had occurred between zones 1 and 2 in the three months between June and August 2014 (Figure 4). (Green: Data not available) Figure 4 I-Site laser scanner movements (6 June to 7 September 2014) Eventually, minor reverse scarp cracking began to develop in the stiff clays on the east wall batter slopes below the crest. These cracks were the first indication of flexural toppling, with the prism and extensometer data indicating between 10 and 60 mm of total movement. On 5 October the final lower ramp blast was fired to get the new ramp back down onto the floor of the first phase of mining. This blast severely jolted the SE wall; with up to 58 mm of additional movement occurring at the crest, based on prism monitoring (Figure 5). 86
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia Figure 5 SE wall movements post ramp blast The laser scanner data also confirmed a similar amount of movement, with zone 1 movement increasing up to a total average of just over 105 mm (Figure 6). The scanner heat map actually showed the beginnings of the toppling event, with both the upper and lower areas moving outwards much more than the central area. Figure 6 Laser scanner data to 8 October 2015 Consideration was given to either cutting back the slope or continuing to carefully mine out the blasted material while monitoring movements. Due to the proximity of the cracking behind the crest to a public road, a third option of establishing a stabilization berm and adopting a three-stage mining strategy was chosen. The three-stage design is shown in Figure 7. After establishing the stabilization berm, the east wall ramp was to be switch backed to the northern end of the pit from where the complete floor would be stripped out by a ramp to the south (Figure 7A). Once completed, the southern end of the pit was to be backfilled, creating a down ramp system to the new floor (Figure 7B). This was to be followed by mining of the northern end of the pit (Figure 7C), taking out the northern ramp. The establishment of the stabilization berm and switchbacked ramp can be seen in Figure 8. 87
Slope Stability 2015 Figure 7 Three-stage mining sequence Figure 8 Stabilization berm 88
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia Slope monitoring and flexural toppling To enable safe mining under the SE wall, an IBIS M (Mk 2) radar was mobilized to site in late October. It was deployed in a position directly opposite the SE wall (Figure 9) to carry out critical alarm monitoring. In this position, approximately 165 m across the pit, the radar began scanning a 250 m length of wall every 5 minutes. (A subsequent upgrade to the hardware and software has decreased the scanning frequency to 3 minutes.) The radar data was sent to the IBIS Guardian software via a Telstra 3G link, allowing analyses and alarm settings to be undertaken. The pixel size of the radar scans was 0.75 m 1.5 m. Figure 9 IBIS M slope stability radar The I-Site laser scanner had been intermittently monitoring the pit from a position at the far north of pit. Following the blasting event of 5 October the laser scanner was moved to the NW wall about 30 m from the radar location. The original resolution (R8) of the scan had a pixel size of 0.5 m 0.5 m and a run time of 44 minutes. After being moved, the scanner was set at a lower (R2) resolution but at a faster scan time of 4 minutes. The pixel size reverted to approximately 1.7 m 1.7 m. Due to the lack of direct communication from the geotechnical office to the scanner, the laser scanner could not be used as a front-line critical monitoring device. It did, however, provide insight into the velocity alarm levels applied to the radar and was used to verify radar detected movements. Throughout the first three weeks of November both the radar and laser scanner detected movements of less than 0.1 mm/d diffused over the SE wall. The movements began concentrating into two main areas (Figure 10). The upper and fastest moving area (0.7 mm/d) was at and just below the crest; where the 58 mm of blast-related prism movement occurred. The second area of concentrated movement (0.4 mm/d) was located near mid-slope but with an elongated shape trending uphill to the south. 89
Slope Stability 2015 Figure 10 Radar displacement The elongated shape was recognized to be parallel to the trace of the steeply dipping foliation (70 east). A few days later a reverse scarp crack was identified on the slope, and when surveyed it plotted just above the mid-level movement area, as can be seen in Figure 10. The crack proved to be difficult to distinguish unless the sunlight was oriented such that the shadows delineated it. The prism surveys showed the overall crest movement of the SE wall up to 22 November 2014 to be in the order of 335 mm (Figure 11). The mid-level movement was approximately 100 mm. 90
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia 22 nd November Rain Event 5 th October Ramp Blast Figure 11 Prism movements On 22 November there was a 7 mm rainfall event over a 9-hour period, which triggered accelerating slope movements see Figure 10 (radar data) and Figure 11 (prism data), leading up to final collapse of the wall one week later on 29 November (Figure 12 and Figure 13). The flexural toppling mode is very evident in the photograph in Figure 13. The total crest movement just prior to the failure, as measured by the prisms, was 450 mm. The radar and laser scanner movements were significantly less, but they had base reading dates from late October and early November in comparison to early July for the prisms. Figure 12 SE wall collapse 91
Slope Stability 2015 Figure 13 Toppling structures at toe of failure Of considerable interest was the monitoring of surface movements with the progressive development of the flexural toppling. Both the radar (Figure 14) and the laser scan (Figure 15) data showed distinct bands of movement that correlated with the reverse scarp locations. The reverse scarp locations in the lower image of Figure 15 were surveyed from the laser scanner image using Maptek I-Site Studio software. 92
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia Figure 14 Spatial distribution of radar displacements 93
Slope Stability 2015 Figure 15 Spatial distribution of laser scanner displacements 94
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia Radar critical monitoring The critical monitoring of the slope by the IBIS slope stability radar was based on two levels of velocity alarms for preselected areas; a geotechnical alarm and a critical alarm. The geotechnical alarm was set at 1 mm/h and the critical alarm at 2 mm/h. When these levels were reached, emails and SMS messages were automatically sent to the geotechnical staff, senior mining personnel, pit supervisors, and leading hands. For a geotechnical alarm, the trigger action response plan (TARP) required crews to stand off the SE wall and await an assessment by the geotechnical staff. For a critical alarm, the pit was to be immediately evacuated in accordance with the emergency egress plan developed for that pit. As can been seen in Figure 16, the geotechnical alarms began on the night of 27 November. The mine supervisor on duty and the geotechnician on call assessed that the blast-hole drilling being carried out in the centre and north of the pit could proceed through the night. Figure 16 Velocity plot This assessment was also based on the use of the inverse velocity tool in the Guardian software package. Late in the afternoon of 27 November, the geotechnical staff started seeing the first indications of a predicted failure late on 29 November. By the early hours of 28 November the inverse velocity plot (Figure 17) was consistently indicating that the failure would occur between 21:00 22:00 hours on 29 November. Figure 17 Inverse velocity plot 95
Slope Stability 2015 Late in the morning of 28 November the decision was made to stop all activity in the Nugent Pit and to fully inform the mining workforce of the situation. Although the workforce had been regularly informed of the minor amounts of movement, they had not previously been exposed to large wall failures. With very little evidence of large-scale movements on the slope itself, there was considerable scepticism from the workforce about the failure prediction. The wall failed at 20:00 hours on Saturday 29 November, just at sunset. Unfortunately, a member of the geotechnical staff was just returning to site to film the wall when the failure occurred and no-one actually observed the failure. The size and runout of the failure was as had been predicted in the workforce presentation. The successful alarming and failure time prediction greatly enhanced appreciation of the geotechnical staff s efforts at providing a safe workplace. The I-Site laser scanner also picked up the failure and subsequent analysis indicated that it would have been capable of providing similar critical alarming capabilities in this instance. For more information on the laser capabilities, refer to Hutchison and Howarth (2015). The laser scan data was used to estimate the volume of the slip as 64 000 t, using the I- Site Studio software. Post-failure mining and assessment The failure continued to be monitored by the radar, which allowed drilling to recommence within 48 hours. Other than fine scree rilling, no medium- or large-scale movement occurred on the failed slope, even when three subsequent blasts occurred on the pit floor within three weeks of the failure. Mining directly under the failure was completed in March 2015 (Figure 18). The change back into the geotechnical domain GDM1 allowed the 50 m high-near vertical wall to be mined directly in front of the topple. The final 22 m was carried out in two goodbye cuts, carefully monitored by the IBIS radar. Figure 18 Post-failure mining A review of the failure and the monitoring concluded that: The potential for a toppling failure in the GDM2 geotechnical domain of the east wall of Nugent Pit was recognized several months prior to the failure The continuous 50 batter slope design, with a maximum height of 40 m, was successful at preventing toppling failures on the east wall The toppling failure occurred on the transitional SE wall, where the slope varied between 50 and 55, over heights of 42 70 m. 96
Management of a toppling failure wall collapse at the Kanmantoo Copper Mine in South Australia The initial movements were instigated by a blast of a ramp shot on 5 October, which toed out at 50 m below the crest in the centre of the topple area Following the blast, a stabilization berm was placed on blasted ground such that only a 40 m high wall was left exposed Movements continued at a rate of up to 4 mm/d until the stability berm was completed, and then gradually slowed to approximately 1 mm/d until 21 November A 7 mm, nine-hour duration rain event on 22 November instigated accelerating movements, before the final collapse on 29 November The final topple occurred above the stabilization berm, with some material rilling down on the southern end to the working floor 19 m below the top of the stabilization berm. Material reached a distance of 28 m from the toe of the slope on that working floor The groundwater table was well below the failure extent, so groundwater was not a factor In the early stages of the wall movement, prisms and surface extensometers were quite adequate for monitoring movements An IBIS slope stability radar was used to monitor the wall during the four weeks prior to the final collapse, and it successfully provided critical alarms and predicted the time of failure within two hours using an inverse velocity prediction tool A Maptek I-Site 8820 laser scanner had been deployed as a research development project and the data collected provided complementary information that was also used in managing the slope failure Average velocity alarm levels of 1 and 2 mm/s were set for establishing geotechnical (assessment) and critical (evacuation) alarms in the ductile rock conditions. Subsequent analysis indicated that 1.5 and 3 mm/s levels would have been adequate and would have reduced the number of alarms prior to the actual collapse The inverse velocity tool in the IBIS software successfully and continually predicted the failure to within two hours from 38 hours prior to the actual event. The pit was closed to all personnel 32 hours prior to the event. Although the slope did not exhibit classical extensive reverse scarp development (such as the 1 m high scarps at Savage River), the radar and laser scanner data and observations of the slope did confirm that the failure mode was flexural toppling, with reverse scarp heights in the order of 100 200 mm The final 40 m high failure involved 64 kt of rock, most of which was retained on a stabilization berm, with some material rilling out 28 m onto the lower working floor Although the stabilization berm did not prevent the wall failure, it did restrict the extent of the failure and allowed mining to continue well below the toe of the failure. It also prevented the head scarp from potentially extending outside the mining lease. Summary The Kanmantoo Copper Mine experienced a 64 kt wall collapse in the Nugent Pit that was initially monitored by conventional extensometers and prisms. A Maptek I-Site laser scanner was also used on an intermittent basis. Following a blast at the toe of the slope, which caused up to 58 mm of movement, an IBIS slope stability radar was deployed to provide critical alarm monitoring capability to safely manage the slope until and after the eventual collapse of the wall. The inverse velocity capabilities of the radar provided consistent predictions of the failure time within two hours of the actual time from up to 38 hours before the event. References Hutchison, B.J., Dugan K., and Coulthard, M.A. 2000. Analysis of flexural toppling at Australian Bulk Minerals, Savage River Mine. Proceedings of GeoEng2000, Melbourne, Australia, November 2000. Hutchison, B.J. and Howarth, J. 2015. Kanmantoo Mine - Rockfall and rock wall failures - I-Site 8820 laser scanning applications. Proceedings of Slope Stability 2015, Cape Town, South Africa, October 2015. Southern African Institute of Mining and Metallurgy, Johannesburg. 97
Slope Stability 2015 The Author Bruce Hutchison, Principal Engineer, Hillgrove Resources Bruce has 40 years of experience in the civil and mining engineering industries. His civil experience has been gained worldwide in Canada, China, the Philippines and Australia where he specialized in hydro-electrical and tunnelling investigations and construction. In the mining industry Bruce has been involved in open pit production specializing in the geotechnical fields of slope stability, slope monitoring and in mine planning; and in civil engineering related projects such as tailings dams, river crossings and haul roads. He has been involved with acid mine drainage rehabilitation and dump designs. Bruce was the Senior Geotechnical/Principal Engineer at Grange Resources Savage River Mine for 15 years before joining Hillgrove Resources in April 2014. His mining career has also involved in operations and consulting on mining projects in Western Australia, Queensland and Mongolia. 98