International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

Stability assessment and slope design at Sandsloot open pit, South Africa
A.R. Byea,*, F.G. Bellb
a

Potgietersrust Platinums Limited (AngloPlatinum), Private Bag X2463, Potgietersrus 0600, South Africa b Department of Geology and Applied Geology, University of Natal, Durban 4041, South Africa Accepted 12 February 2001

Abstract Sandsloot open pit is located on the northern limb of the Bushveld Igneous Complex. It is the largest open pit platinum mine in the world. Three major joint sets have been recognized at Sandsloot, which are related to the regional tectonic history. They have an important inuence on slope stability in the open pit, notably in terms of planar and wedge failures. Detailed geological and geotechnical data are often a notable unknown factor in the design and operation of an open pit, the lack of which may pose a signicant risk to the mining venture. As data are accumulated and used eectively, so the risk of unforeseen conditions is reduced, and accordingly safety and productivity is increased. Usually, the geotechnical work undertaken at an open pit mine is in connection with improving slope stability. At Sandsloot open pit geological and geotechnical data have been obtained by face mapping, scanline surveys, from exploration drillholes and from laboratory tests. Such data have been used to delineate dierent geotechnical zones in which dierent types of slope failure have occurred. These are the usual types of slope failure associated with rock masses, namely, planar, wedge, toppling and circular failures. Analysis of the data has allowed optimum design parameters to be developed for these zones which has led to improved slope stability. In other words, this has allowed slope management programmes to be initiated, as well as slope optimization of the hangingwall. The latter resulted in an improved slope conguration and an increase in the ultimate angle of the wall by 71. This has resulted in substantial savings, as well as an improvement in safety. r 2001 Elsevier Science Ltd. All rights reserved.
Keywords: Discontinuities; Slope stability; Geotechnical zones; Data analysis; Slope conguration

1. Introduction In order to minimize the amount of waste rock, which has to be removed in the recovery of an orebody, the ultimate slopes of an open pit mine generally are excavated to the steepest possible angle. Unfortunately, the economic benets gained can be negated by major slope failure. Consequently, continual evaluation of the stability of the ultimate slopes is a vital part of open pit planning. As the presence and character of discontinuities have an important inuence upon the stability of rock slopes, their assessment forms a critical part of any stability assessment. Sandsloot open pit (Fig. 1) was commissioned in 1992 and since then a number of slope designs have been
*Corresponding author. Tel.: +27-15-418-2000; fax: +27-15-4182159. E-mail address: abye@angloplat.com (A.R. Bye).

developed. The evolution of the designs has not only led to improvements in slope stability and safety but also has given rise to cost savings. The pit is situated on the northern limb of the Bushveld Complex, some 250 km north-east of Johannesburg. It currently is roughly 1500 m long, 800 m wide and trends in a northsouth direction. Sandsloot open pit is located at an average elevation of 1100 m and has a projected depth of 325 m. At present, Sandsloot is the largest open pit mining platinum in the world. In the past the majority of geotechnical work undertaken at open pit mines has focused on the slopes. Detailed geotechnical data are often a largely unknown factor in open pit design and therefore represents a signicant risk in any mining venture. As the amount of geotechnical data increases so the risk of unforeseen conditions is reduced, and therefore safety and productivity can be increased. In order to increase and improve the geotechnical data a detailed geotechnical

1365-1609/01/$ - see front matter r 2001 Elsevier Science Ltd. All rights reserved. PII: S 1 3 6 5 - 1 6 0 9 ( 0 1 ) 0 0 0 1 4 - 4

450

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

2. Geology of the area The Bushveld Igneous Complex represents the largest intrusive body in the world. The main period of igneous activity resulted in the emplacement of a mac phase (the Rustenburg Layered Suite) followed by an acid phase (the Bushveld Granophyre and the Lebowa Granite Suite). The Rustenburg Layered Suite has been sub-divided into various zones, namely: the Marginal, Lower, Critical, Main and Upper zones, and exceeds 9000 m in thickness. At least three large interlinked chambers, with separate feeder pipes, have been identied. A feeder pipe close to Potgietersrus resulted in the Potgietersrus Limb. The Potgietersrus Limb covers an area of roughly 2000 km2 and consists of a north striking trough-shaped body (Fig. 2). It is some 110 km in length and attains a maximum width of 15 km. A platiniferous horizon, known as the Platreef, is associated with the Potgiertersrus Limb. It forms the base of the Main Zone and hosts economic platinum group element mineralization within a sulphide-bearing pyroxenite body. It has an economic strike length of 40 km. The basal contact of the Platreef is highly irregular and has a transgressive relationship with the underlying sediments of the Transvaal Supergroup and Archaean granites as it progresses northwards. This transgressive relationship with dierent rock types has given rise to diering degrees of metamorphism, metasomatism and assimilation of the oor rocks. Hence, the interaction of the pyroxenite of the Platreef with dierent sedimentary sequences has resulted in a highly complex and unique suite of rock types. The Platreef is capped by a sequence of Main Zone gabbronorites, which in turn, is overlain by Upper Zone sequences of ferrogabbros (Fig. 3). Metamorphosed pyroxenite, that is, parapyroxenite hugs the footwall of the Platreef. At Sandsloot the outcrop of the Platreef exhibits a somewhat sinuous pattern with local changes in strike and dip. There is considerable variation in thickness of the Platreef at Sandsloot, ranging from 70 to 200 m (Fig. 4). The pyroxenites, which comprise the Platreef at Sandsloot, vary in texture from ne to coarse grained and the oor rocks consist of dolomite belonging to the Malmani Subgroup. Indeed, between Tweefontein and Sandsloot there is an embayment of dolomite known locally as the dolomite tongue, which thins the Platreef. The footwall to the Platreef at Sandsloot is represented by metadolomite referred to as calc-silicate. The dolomite has resulted in appreciable contamination at Sandsloot, with many xenoliths of dolomite being assimilated to form parapyroxenite. The parapyroxenites are essentially contaminated metamorphosed pyroxenite. They are medium grained and contain abundant serpentine, which represents a product of late

Fig. 1. View of Sandsloot open pit looking north.

programme was implemented at Sandsloot open pit. The focus of the project was to make practical use of geotechnical information so as to improve slope stability, and slope congurations, and thereby optimize the nal wall angle and so reduce the cost of mining. Previously, one production and slope design method was applied to the entire open pit, as in the case in many mines. In order to improve production, the open pit was divided into geotechnically similar zones to which empirical slope designs have been applied. These were based on detailed structural and geological mapping which enabled any active or potential instability problems to be detected. The denition of each zone involved rock mass classication, a rock strength testing programme, scanline surveys and slope stability analyses. Five potential failure mechanisms were identied in terms of pit slopes, namely, planar, wedge, toppling, and circular failure zones, as well as a geotechnical nose zone. Specic slope design congurations, stability values and sensitivity parameters were estimated for these areas. A major feature plan and cross sections have been produced based on the structural and geotechnical data collected. The plan has proven very useful for slope design. Five slope designs have been used since the mine started. The changes were due in part to variations in mining method but mostly were due to the use made of the geotechnical data assembled to date. From a slope design perspective there are several congurations, which can be used to give the most stable and economic nal wall. The data gathered and designs formulated have been of considerable value in developing a slope management programme, improving safety and optimizing the nal open pit walls. The nancial benet associated with optimizing the nal walls in terms of extending the life of mine and reducing the stripping ratio is signicant.

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

451

Fig. 2. Regional geology of the Bushveld Complex.

stage alteration. The hangingwall norite unit generally exhibits a medium to coarsely crystalline texture with occasional disseminated sulphides. The norites exhibit a degree of layering and are normally leucocratic. The Platreef has been disturbed by several sets of faults, which strike at 0301 and at 0801 to the east. Most of these faults have a normal attitude and commonly have a downthrow on the southern limb. A series of strike faults, which trend north north-west also are evident in the region. Indeed, a strike fault marks the basal contact between the Bushveld Complex and the oor over certain sections of the Potgietersrus Limb. The open pit at Sandsloot is interrupted by three major north-east trending faults. There are two normal faults in the north pit, which form a graben structure with a downthrow of 30 m and a major oblique-sinistral fault, which has displaced the ore body to the south-east by approximately 400 m. This fault has caused extensive alteration and deformation in the adjacent rock types. For instance, highly deformed serpentinized pyroxenites occur on the south-east wall of the pit. Such serpentinization has weakened the pyroxenite signicantly. A highly sheared zone of pyroxenite, up to 5 m in width and dipping at 451, separates the hanging wall norites from the pyroxenite ore body. The zone provides a natural drainage path for local groundwater.

3. Geohydrology Sandsloot open pit is situated on the eastern side of the shallow valley of the Groot Sandsloot River, which falls within the catchment area of the Mogalakwena River. The Groot Sandsloot River basin has a mean annual runo of 1.731 106 m3 and contributes 5% of the surface runo to the Mogalakwena River. The recorded average annual rainfall generally varies from 380 mm to just over 700 mm. Precipitation occurs mainly in the form of thunderstorms during the summer months with a peak in January. Extensive borehole investigations in the mine lease area and the surrounding catchment indicated that the main water bearing zones lie at the base of the weathering prole at depths of between 2 and 43 m. The annual rate of groundwater recharge within the catchment area is approximately 1.86 106 m3. It has been estimated from boreholes that the open pit at Sandsloot has an inuence on the surrounding groundwater system for a radius of only 400 m. There are no groundwater users within this radius. The mine itself is heavily dependent on its well elds for a large portion of the approximately 5.0 106 m3 of water it requires per annum for its mining processes. The mine has two well elds, which are hosted within two types of

452

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

Fig. 3. (a) Geology of the Sandsloot locality. (b) Order of stratigraphic occurrence of the rocks in the Sandsloot locality.

aquifers. The semi-conned, weathered and fractured aquifers are the more frequent. They take the form of a series of elongated troughs approximately parallel to the

strike of the Bushveld rocks. Fracturing has enhanced the development of these weathered basins and higher yielding boreholes are associated with them. The

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

453

Fig. 4. Plan view showing the geology of Sandsloot open pit. The double lines reect the crest and toe positions of the active mining benches.

overlying weathered horizon supplies most of the storage, with the underlying fractured zone forming the main transmissive zone. The second well eld is located within a highly fractured shear zone associated with north-eastsouth-west trending faults. Strike depths range between 20 and 45 m. An initial geophysical survey suggested that the weathered prole could vary from 20 to 35 m in depth within the pyroxenites. From subsequent eld observation it appears to be less well developed in the norites, extending to between 5 and 10 m below ground surface. The weathered horizon acts as a perched aquifer. Beneath the weathered horizon water occurs in discontinuities. It has been estimated that the total groundwater ow into the pit is about 112 m3 h@1, with some 22 m3 h@1 entering via the footwall, the remainder via the hangingwall. The phreatic surface exists at approximately 40 m below ground level and could have an aect on the factor of safety of the design slopes in the pit. Therefore dewatering by pumping from sumps is necessary. Fig. 5 illustrates the inuence of dry as opposed to wet conditions on the factor of safety of design slopes. A factor of safety of 1.25 was used in the analyses concerned, it allowing a 1520% probability of failure.

4. Discontinuities Discontinuities play a major role in the design and maintenance of opencast pits, their presence aecting the mechanical and hydrogeological properties of the rock masses. From a slope design perspective three major joint sets can be seen continuously throughout the Sandsloot pit area (Table 1, Fig. 6) and therefore were studied in detail to assess potential failure mechanisms. These three major joint sets reect the tectonic environment, being related to the emplacement of the Potgietersrus Limb. The remaining minor joints are thought to have formed by randomized contraction jointing on cooling of the Bushveld intrusive phase. The most prominent joint set (JS1) strikes north-west and has pronounced slickensides. Late stage quartzo-feldspathic (felsite) veins are associated with this joint set. They are steeply dipping, and are laterally and vertically continuous over hundreds of metres. As such, they pose slope stability problems. Scanline surveys were carried out in selected areas of the open pit at Sandsloot either where failure had occurred or where the potential for failure existed [1]. In fact, most of the surveys were conducted around the Satellite pit due to the occurrence of geotechnically poor areas associated with the Satellite pit fault and

454

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

Fig. 5. The eect of dry and wet conditions on the factor of safety of the design of slopes. Line A indicates a lower FOS/less stable slope than line B, for an equivalent slope height and angle, due to the higher level of the water table within the slope. For example a dry slope at 501 will have a FOIS of 2.1 while a wet slope will have a lower FOS of 1.6. Table 1 Averaged joint set data from Sandsloot open pit Dip JS1 JS2 JS3 JS4 72 73 78 82 70 62 72 63 Dip direction 088 263 357 183 310 125 237 065 Joint roughness (IV) Rough Irregular Undulating (II) Smooth Stepped (V) Smooth Undulating (VII) Rough/Irregular Planar Joint lling Calcite Serpentinite Calcite Serpentinite Calcite Joint spacing 0.5 m 0.4 m 0.3 m 0.15 m

serpentinized parapyroxenite. Numerous parameters were measured for each discontinuity such as roughness, dip and dip direction, spacing, continuity, joint aperture, and the presence of gouge. An example of the data sheets used to collect data during the scanline surveys is given in Fig. 7. Joint prole readings were taken [2] with a carpenters comb and compared with the joint roughness proles of [3]. Detailed structural mapping was also

undertaken of each face, the data collected including the lithology, dominant joint sets, critical joint spacing, veins and any faults. The DIPS software program was used to generate the stereonets from the scanline survey data collected. For each scanline survey a contoured pole data stereonet with joint set windows was constructed. The stereonets then were analysed according to

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

455

procedures given in [4] in order to assess potential modes of failure, thereby allowing potential failure zones to be identied.

Fig. 6. Stereonet showing the families of joint sets at Sandsloot open pit.

Determination of reliable shear strength values is a critical part of a slope stability analysis because relatively small changes in shear strength can result in signicant changes in the safe height or angle of a slope. Dierences in the shear strength of rock surfaces can occur because of the inuence of weathering, surface roughness, and the presence of water under pressure [4]. At each scanline survey samples were collected for each of the dominant joint sets. The samples were tested in a Golder shear box [5]. The Golder shear box is designed for testing samples at normal loads up to 2 MPa and the normal load is applied by means of a dead load system. It therefore remains constant throughout the test and is therefore more sensitive than the portable shear box. Moreover, the maximum eective normal stress acting across joints considered critical for stability lies, according to Barton [3], in the range 0.120 MPa. The average peak shear strength parameters for the three rock types primarily involved in slope stability assessments, namely, norite, pyroxenite and calc-silicate, are given in Table 2. Table 2 also gives the range of peak friction angles for normal loads of 0.54 and 1.08 MPa, which correspond to bench heights of around 20 and 40 m, respectively. Generally, the pyroxenite possesses the lowest shear strength along joints.

Fig. 7. Face mapping checklist for MRMR evaluation.

456

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

Table 2 Geotechnical properties of the dominant rock types at Sandsloot open pit Hanging wall Norite Peak Shear Strength (MPa) IRS (MPa) Schmidt Hammer RQD% FF MRMR MRMR Class Density (Mg m@3) Youngs modulus Poissons ratio Tensile strength (MPa) jp(1) jb(1) JRC JCS (MPa) Slope angles (1) 0.93(sn=0.54) 1.78 (sn=1.08) 190 (150210) 80 (40100) 9 (2030) 53 (4860) (III A) Fair 2.9 10.24 GPa 0.2 9.5 (7.510.5) 5764 3538 610 87100 5659 Ore Zone Pyroxenite 0.78 (sn=0.54) 1.64 (sn=1.08) 160 (120180) 65 (25100) 13 (293) 48 (3552) (III B) Fair 3.2 F F 8.0 (69) 4058 2832 212 62100 4850 Parapyroxenite Serpentinized parapyroxenite Foot wall Calc-silicate 1.09 (sn=0.54) 1.76 (sn=1.08) 140 (100180) 55 (2080) 16 (329) 42 (3250) (III B) Fair 2.9 F F 7.0 (59) 50 32 810 50 51

200 (180250) 75 (55100) 10 (1530) 56 (5061) (III A) Fair 3.2 F F 10.0 (912.5) 55 3436 810 75 5159

270 (200310) 70 (50100) 11 (1630) 61 (5170) (II B) Good 3.2 F F 13.5 (1015.5) F F 812 100150 5659

jp1FPeak friction. snFNormal stress. JRCFJoint roughness coecient. jb1FBase friction angle. FFFFracture frequency. JCSFJoint wall compressive strength. IRSFIntact rock strength. RQDFRock Quality Designation. MRMRFMining Rock Mass Rating.

Tilt tests were undertaken in the open pit to obtain an approximate value of the basic friction angle for the rock types in the pit. An average value of 321, with a range of 28381, was obtained. In fact, Ref. [6] suggested that tilt tests provided a more reliable means of estimating the joint roughness coecient, JRC, than comparison with typical roughness proles. Subsequently, Ref. [7] supported the use of the tilt test, especially in heavily jointed rock masses.

5. Delineation of failure zones Rock slopes generally fail along existing geological defects, notably discontinuities. It is only in very high slopes or weak rocks that failure through intact material becomes signicant. Most rock slope problems therefore require consideration of the geometrical relationships between discontinuities, the slope and force vectors involved. In order to assess the stability of individual benches, so as to aid stack and overall slope design, detailed kinematic failure analyses were undertaken. Kinematic failure zones representing a dominant or critical failure mechanism were delineated for the pit from multiple scanline surveys and detailed eld mapping. Five

kinematic failure zones were identied, namely, planar, wedge, toppling and circular failure zones, as well as a nose zone (Fig. 8). Each zone represented a geotechnically similar part of the pit, the modes of failure being predominantly structurally controlled rather than controlled by rock type. The exception is where circular failure occurs, instability there being due to the weathered rock-soil material. Using the structural data presented in Fig. 6, a kinematic failure analysis [4] was used to dene the dominant modes of failure for the various zones of the pit. The dominant modes will vary according to the intersection of the pit slope orientation and dominant joint orientation. Plane failures are a common type of translational failure and occur by sliding along a single plane. In order for sliding to occur on a single plane the following geometrical conditions must be satised.

The plane on which sliding occurs must strike parallel or nearly parallel (7201) to the face. The failure plane must daylight in the face, that is, the dip of the failure surface must be less than that of the face. The dip of the failure plane must be greater than the angle of friction.

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

457

Fig. 8. Failure zones identied at Sandsloot of pit.

Release surfaces, which provide negligible resistance to sliding, must be present in the rock mass to dene the lateral boundaries of the slide.

All these geometrical conditions are satised by the major joint sets on the western highwall (Fig. 9). The three critically aligned joint sets are presented in Fig. 10 with joint set one, JS1, trending within 201 of the slope. Joint set one strikes at 3521 and dips at 501, while the face strikes at 3551 and dips at 70751. It also daylights

in the face and is inclined at an angle greater than the friction angle, which is 311. Additionally, joint sets two, JS2, and three, JS3, provide release surfaces for failure to occur. The relevant data relating to the plane failure attributable to the major joint sets are given in Table 3. The critical joints showed signicant water ow after heavy rainfall, which drastically reduces the slope stability. This reduction in stability is evident from the sensitivity analyses given in Table 4. In particular, continuous joints, which are slickensided and contain

458

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466 Table 3 Data relating to the critical joint sets causing planar failure JS1 Dip Dip direction Spacing (m) Critical joint spacing (m) Continuity (m) Joint width (mm) Joint lling 1755 073114 0.2 1.5 170 260 Thick calcite, serpentinite, some inactive clay 260 Planar rough (VII) JS2 6590 329018 0.3 4 410 220 Calcite, serpentinite, inactive clay 220 (Planar rough) (VII) JS3 7090 291351 0.4 3 450 220 Calcite, serpentinite

Fig. 9. Planar failures on the western highwall.

Joint thickness (mm) Joint roughness (After Barton, 1978)

220 Planar rough (VII)

Fig. 10. Stereonet depicting plane failure on the western highwall.

serpentinite and calcite vein lling are potentially critical joints, which may fail, especially after heavy thunderstorms. Water tends to ow along such joints after heavy rainfall. From eld observations and stereographic analyses, it appeared that the hangingwall (western highwall) was prone to planar failures developing from tension cracks whereas in the footwall (eastern highwall) planar failures occurred without a tension crack. As an example, when the following parameter values were used: face height (20 m), cohesion (100 kPa), friction angle (311), tension crack distance (3 m), inclination of failure plane (501) and rock unit weight (27 kN m@3) to determine the factor of safety of the western highwall, then it was 1.38 dry and reduced to 0.8 wet. The results of sensitivity

analyses are given in Table 4 and suggest that the plane failure mechanism is most sensitive to slope height and inclination of the failure plane but insensitive to tension crack distance, cohesion and friction angle. If the failure plane occurs in a double bench with a height of 40 m, then the factor of safety drops to 0.97 dry and 0.27 wet. Therefore, it is unwise to create benches greater than 40 m on the western highwall. The most critical failure plane inclination is 501, which more or less corresponds with the dip of JS1. Sensitivity analyses also were undertaken to ascertain the nature of the failure mechanisms on the eastern highwall where failure occurred without the development of tension cracks. The results are illustrated in Fig. 11. Potential wedge failures exist where two discontinuities, which dip out of a face intersect and the line of intersection daylights into the face and is steeper than the friction angle of the joint surfaces. There are two zones in the pit where wedge failure occurs. In the north pit wedge failures develop due to the intersection of the hanging wall contact fault and JS1. In the south pit wedges are formed due to the intersection of the sympathetic jointing related to the satellite pit fault and the relict bedding in the calc-silicate. Observed wedge failures generally occur in association with a nearby production blast and normally are smaller than 100 tonnes (Fig. 12). Wedge failure analysis was undertaken using stereonets (Fig. 13), as well as the SWEDGE software package. The data related to the critical joint sets responsible for wedge failure are given in Table 5. In addition, a sensitivity analysis was undertaken to assess the eect of face height, cohesion and friction angle on slope stability in relation to the factor of safety in dry and wet conditions. The results are given in Table 6. Wedge stability appears to be relatively insensitive to these three parameters with the factor of safety barely

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466 Table 4 Variation of parameters used in sensitivity analyses Factor of safety Parameter Face height (m) Cohesion (kPa) Friction angle (1) Tension crack distance (m) Inclination of failure plane (1) 1050 60140 3035 0.510 3070 Dry 2.270.88 1.031.73 1.361.46 1.531.53 1.815.69 Wet 1.930.16 0.511.08 0.790.81 0.441.47 0.675.92

459

Fig. 12. An example of a small wedge failure.

Fig. 11. Sensitivity analysis for planar failure (relationship between factor of safety on the one hand and rock mass saturation, cohesion, c, and joint angle, jp on the other). The term saturation refers to the height of the water table in the slope as a percentage of the slope height. For example, 100% saturation indicates a water table that is at the surface.

changing with parameter variation, suggesting that small wedge failures are inevitable. Toppling failure involves the rotation or overturning of blocks of rock about some xed base and is associated with steep slopes and sub-vertical joints dipping back into the slope. The areas in the open pit where toppling failure has been identied are shown in Fig. 8. The joints causing potential failures are associated with JS1. Flexural type toppling most closely approximates the mechanism identied in Sandsloot open pit. In this, continuous columns of hard rock are separated by well-developed steeply dipping joints which break in exure as they bend forward. Dilation of the joints over time is associated with sliding along shallow dipping joint surfaces. Toppling failures are evident

Fig. 13. Stereonet indicating potential wedge failure. Joints 1 and 2 intersect within the friction circle (joint intersection dip angle>friction dip angle) and will daylight within the slope face, resulting in a high probability of wedge failure.

460

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

Table 5 Data relating to the critical joint sets causing wedge failure on the footwall ramps JS1 Dip Dip direction Continuity (m) Joint lling Joint spacing (m) Joint thickness (mm) Joint roughness (After Barton, 1978) 851 (801901) 751 (601901) 0871 (07510981) 2531 (23312721) 110 Calcite 0.40 220 IV Undulating rough JS2 861 (811901) 791 (681901) 0141 (35510321) 1961 (17712151) 210 Serpentinite, calcite 0.50 220 VII Planar rough JS3 611 (421801) 1241 (09811491) 110 Calcite 0.30 220 V Undulating smooth

Table 6 Sensitivity analysis of the pertinent parameters controlling wedge failure Factor of safety Parameter Face height (m) Cohesion (kPa) Friction angle (1) 1040 5500 2040 Dry 0.600.57 0.590.87 0.400.84 Wet 0.150.13 0.180.18 0.090.19

along the crests of the footwall and hangingwall. Potential failures are destabilized by crest damage attributable to poor presplit blasting and inadequate crest protection measures. However, owing to the steep nature of JSI (70751), and narrow joint spacing, the resultant toppling failures are small in size and are retained by the underlying catchment berm. Circular failures may occur in rock masses, which are highly weathered or are so intensely fractured in relation to the scale of the slope that they may be considered as randomly jointed and isotropic. Unlike wedge and planar failure in competent rock, circular failure is not dened by discontinuities and the failure surface is free to nd the line of least resistance. Along the eastern highwall of the Satellite pit the serpentinized pyroxenite is weathered, slickensided and jointed to such a degree that it approaches the conditions necessary for circular failure (Fig. 14). Indeed, the entire highwall underwent multiple circular failures to the extent that benches were carried away. However, the ore body within the Satellite pit was mined out in June 1998 and back lled with waste to surface elevation. From a geometrical perspective it can be assumed that concave slopes are more stable than convex slopes. The nose zone is therefore potentially a problem area as it has a prominent convex geometry (Fig. 15). Small scale planar and wedge failures occur around the entire nose zone. The nose zone consists predominantly of calc-silicate material, which is highly jointed. The area therefore requires continual monitoring, as well as bench inspections so that any potential failure can be predicted.

Fig. 14. Circular failure on the south-east highwall of the satellite pit.

Fig. 15. The nose zone between the main pit and the satellite pit.

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

461

6. Design zones A number of design zones in which the geotechnical conditions are similar were recognized at Sandsloot open pit mine. Their identication was dependent upon the data obtained from extensive scanline surveys, geotechnical face mapping and exploration drillhole logs, which were evaluated in terms of the mining rock mass rating (MRMR), rock quality designation (RQD) and intact rock strength (IRS). Particular strategies and mining methods can be adopted to work each of these zones. As noted, the mining rock mass rating (MRMR) system of [8] was used for rock mass quality assessment. When compared with the rock mass rating (RMR) system [9] and the Q-system [10] the results were very similar. The comparison was undertaken purely to ensure the MRMR assessments were providing reliable data. The MRMR system is suitable for both face mapping and drillhole logging, which ensures consistency in data capture. The averaged values of the geotechnical data collected are given in Table 2. The MRMR system involves the assignment to the rock mass of a rating. Each parameter used in the system is weighted according to its importance and the sum of the individual weightings gives the rating. The system is outlined in Fig. 16, which shows the weighting for each parameter. A range of ratings from 0 to 100 is

obtained and allows distinction of a number of classes from very poor to very good. Adjustment factors are used in the MRMR system to suit particular conditions. These include adjustments for weathering, mining induced stresses or change in stress eld, joint orientation and blasting eects. The site specic mining adjustments for Sandsloot open pit are given in Table 7. The concept of rock quality designation (RQD) was introduced in [11]. It is based on the percentage of core recovery when drilling rock with NX (57.2 mm) or larger diamond core drills and is the sum of the core sticks in excess of 100 mm expressed as a percentage of the total length of core drilled. A Schmidt hammer was used in the eld to obtain a rough estimate of the intact rock strength for all rock types within the pit. The point load strength of the rock types also was determined in the eld. Standardized eld mapping sheets for capturing rock mass rating, geological and structural data are developed (Fig. 7). The bench face to be mapped was visually divided into zones based on a change in rock type or major structural features. Any structural zones wider than a metre were classied separately. Each individual zone then was mapped using a separate mapping sheet to which all the relevant information was captured. The mapping included a scanline survey and recorded data on joint spacing, orientation, roughness, continuity and condition. As mentioned, Schmidt hammer readings

Fig. 16. Flow diagram illustrating the MRMR system of Laubscher [8].

Table 7 Mining adjustments for Sandsloot open pit Sandsloot MRMR mining adjustments Weathering Joint orientation Stresses Blasting Total adjustment 100% 95% 96% 95% 87% All rock types have UCS>140 MPa and do not weather signicantly within a 5 year period One unfavourably orientated joint set (JSI) Accounts for relaxation and de-stressing of the open pit slopes Accounts for smooth wall blasting and high blast activity in narrow pit The total adjustment is large to ensure a conservative rating

462

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

were taken to obtain an estimated value of intact rock strength. The data collected in the pit was transferred to computer for digital storage. A customized AutoCAD system was used to image the mapped faces (Fig. 17). In addition, the MRMR rating was calculated by AutoCAD. Structural features, lithologies and RMR zones, then were exported to a plan, which incorporated all relevant information for a particular bench. Dempers [12] discusses how the MRMR system was customized for a drillhole logging procedure, which allows rapid assessment of core for geotechnical evaluation. In conjunction with the logging technique, a spreadsheet was designed whereby the relevant parameters that are required for input into the MRMR system can be analyzed and interpreted. The results are used to produce nal MRMR ratings and preliminary slope angles, from the charts illustrated in [13], for each geotechnical zone. The principle of this core logging technique diers from conventional logging in which each core-run drilled is individually assessed for each geotechnical parameter required for engineering classication systems. The drillhole is separated into similar geotechnical zones based on rock type and any major structural features wider than one metre. After the drillcore has been grouped into geotechnical zones each relevant parameter required for geotechnical evaluation is determined within a particular geotechnical zone. Relevant geotechnical information subsequently can be exported directly to DATAMINE or AutoCAD for interpretation. The interpreted data are used to estimate stable slope congurations and predict rock mass conditions. Slope monitoring is necessary to ensure that open pit slopes are safe and that any movements, which do occur are within acceptable limits. It also is important to record excessive slope displacements associated with total or partial failure. Records of displacement data can be used to predict the behaviour of a failure and enable a mine to take steps to minimize production and equipment losses and, most importantly, danger to personnel. A comprehensive slope monitoring network

is being installed in critical areas of Sandsloot open pit and monitoring is carried out on a regular basis. Both permanent and temporary monitoring points are used, as well as those that are accessible to GPS monitoring and those that require conventional survey monitoring. The temporary points are on walls that will be removed within two years. Specic monitoring of individual structures as they are exposed in the pit also is undertaken using portable monitoring equipment. Portable monitoring points are located in areas that are below unstable highwalls. These points utilize a telemetry system to relay displacement information in order to help predict any sudden failures. Traditionally, geotechnical information in an open pit environment has been collected for slope applications, however, using the data to dene unique geotechnical zones, to which specic designs can be applied, is becoming standard practice. The information collected has resulted in more stable slope congurations, substantial savings in stripping costs, improved safety and extension of the life of mine. The data obtained from the geotechnical zones are used in a slope management programme, as well as for slope design. Delineated kinematic failure zones highlight problem areas and allow adjustment of stack congurations in order to contain identied failure mechanisms. The recognition of similar geotechnical zones has aorded the opportunity to move away from one design recipe for the entire pit, and customization of slope designs and congurations has been developed to allow for local variations in the rock mass conditions. Hence, slope stability has been improved. FLAC was used to model the pit slopes as it can take into account ubiquitous joints and variable water tables. The FLAC modelling of the pit slopes was undertaken based on the data derived from the in-pit geotechnical programme. A rigorous analysis was carried out on the hangingwall in order re-evaluate aspects of stability on bench stack and overall slope angles. A sensitivity analysis was carried out using a number of ubiquitous joint model runs to take into account the eect of critical joints on slope stability. Model runs with variations in joint cohesion and critical joint angle were

Fig. 17. An example of a geotechnical face map.

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

463

Table 8 Rock mass input parameters for the FLAC ubiquitous joint model Youngs modulus Friction angle Joint friction angle Density Run 1 2 3 4 5 6 10 GPa 351 431 2.7 Mg m3 c (kPa) 0 100 300 MODEL MOHR 50 0 Poissons ratio Cohesion Joint cohesion Joint angle Joint angle (1) 65 65 65 F 65 50 0.2 30 kPa 0300 kPa 501 and 651 Joint tension (kPa) 0 10 30 F 5 0

Fig. 18. Stability envelopes for norite of the hangingwall developed by FLAC analyses.

undertaken in order to determine parameter sensitivity and a stability envelope for the norite hangingwall. A critical joint dip variation of 50651 was modelled, with a range of cohesion from 0 to 300 kPa. Fig. 18 outlines the stability envelopes of the norite hangingwall developed from the series of FLAC analyses. The input data used in the modelling is given in Table 8. It can be seen from Fig. 18 that for a 60 m stack inclined at an angle of 731, stability can be achieved with cohesion values between 25 and 35 kPa, while for a slope of approximately 200 m inclined at 611, the required cohesion increases to between 50 and 100 kPa. The results of FLAC modelling showed that the proposed overall angle for a 300 m slope could be increased by 71, from 511 to 581. Additionally, the optimized slope design has extended the life of mine by two full benches, thereby allowing the mine to accrue over R1.2 billion in additional income. Slope congurations at Sandsloot open pit have evolved as more geotechnical data have become available and variations in slope design have been applied to accommodate kinematic failure mechanisms. Fig. 19 illustrates the evolution of the stack congurations at Sandsloot open pit from two angled bench designs to a vertical double bench design. By moving to a double bench presplit design, bench preparation, presplit holes

and presplit drilling time is considerably reduced. The double bench conguration has improved slope design at the open pit as the 20 m catchment berms capture any planar failures on the hangingwall. From the viewpoint of overall slope stability a double bench vertical presplit design is as stable as an angled presplit design. Angled presplits, however, result in less crest damage and cleaner walls, thereby reducing rockfall. The problem of rockfall can be dealt with by designing double benches with large catchment berms as opposed to a 60 m stack, without catchment berms. Double bench presplitting has resulted in an improved batter condition by eliminating the mini-berm, which occurred at the toe of each 15 m bench. The mini-berms could retain some loose material so creating a potential rockfall hazard. Empirical design charts were developed from the MRMR data obtained from the geotechnical zones for stable slope height versus slope angle for each dominant rock type occurring in the pit. As a result, a unique stack conguration and optimized slope angle can be developed for each geotechnical zone, taking into account practical constraints. For example, Fig. 20 illustrates the stable slope height, with a factor of safety of 1.2, for the dierent rock types present at Sandsloot. As the variability of measured parameters must be taken into account when using the MRMR data for design purposes sensitivity analyses were undertaken to determine the upper and lower range of MRMR values per rock type. Figs. 21ae illustrate the range of slope design angles for each major rock type at Sandsloot open pit.

7. Summary and conclusions Sandsloot open pit is the largest platinum mine in the world and it is situated on the northern limb of the Bushveld Igneous Complex. The major discontinuities

464

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

Fig. 19. Stack conguration for the hangingwall.

Fig. 20. An example of a slope design chart for rock types at Sandsloot.

at Sandsloot reect the tectonic history, the three main joint sets having no obvious relation with rock types. The principal joint sets are continuous throughout the pit. Geological and geotechnical data at the pit has been obtained by mapping faces and from exploration drillholes and from blastholes. Once the geotechnical data set is collected, it is then evaluated in AutoCAD

and SABLE to give rock mass quality ratings in terms of the MRMR classication system. This has aided the recognition of dierent geotechnical zones with dierent slope failure mechanisms within the pit. Hence, a major features plan of the pit together with geological sections has been produced and is continuously updated. Groundwater and its ow through a rock mass, with associated pore water pressures, adversely aect potential failure surfaces lowering the factor of safety of the slope. Therefore saturated joint zones are mapped in detail and incorporated into the major features plan. Plane failure appears not to be a serious problem unless the slope becomes saturated. Heavy rainfall with a rising water table signicantly reduces the stability of potential failure planes. Hence, the critical joints on the western highwall require monitoring. Wedge failures have tended to be small and are only likely to cause failure of a single bench. By contrast, potential multibench circular failure poses a threat to production in the pit. Unfortunately, there is little that can be done to stabilize these areas of potential circular failure and so monitoring is important in that it helps to predict slope movement. It also has to be borne in mind that vibrations generated by blasting subject slopes to dynamic loading and that blasting itself can weaken rock masses, especially weathered and/or highly discontinuous rock masses. Recognition of variations in rock mass quality and dierent types of slope failure has meant that slope design has been tailored to suit the geological conditions. In this connection, a rigorous analysis of the data obtained from mapping and drillholes, by means of the

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466

465

Fig. 21. Range of slope design based on MRMR, slope height and slope angle for (a) norite, (b) pyroxenite, (c) parapyroxenite, (d) serpentinized pyroxenite and (e) calc-silicate.

FLAC modelling, was undertaken on the hangingwall to assess aspects of stability on bench, stack and overall slope angles. This proved that the overall slope angle for a 300 m slope could be increased by 71, from 511 to 581.

The signicance of this is demonstrated by the fact that a single degree saves R12 million in stripping costs. In addition, optimization of slope design has extended the life of mine by two benches, thereby allowing the mine

466

A.R. Bye, F.G. Bell / International Journal of Rock Mechanics & Mining Sciences 38 (2001) 449466 [4] Hoek E, Bray JW. Rock slope engineering,3rd ed.. London: Institution Mining and Metallurgy, 1981. [5] Hencher SR, Richards LR. Laboratory direct shear testing of rock discontinuities. Ground Engineering 1989;22(2):2431. [6] Barton N, Choubey V. The shear strength of rock joints in theory and practice. Rock Mechanics 1977;10:154. [7] Barton N, Bandis S. Some eects of scale on the shear strength of joints. International Journal Rock Mechanics Mining Science and Geomechanical Abstracts 1980;17:6976. [8] Laubscher DH. A geomechanics classication system for the rating of rock mass in mine design. Journal South African Institution Mining and Metallurgy 1990;90:25773. [9] Bieniawski ZT. Engineering rock mass classications. New York: Wiley-Interscience, 1989. [10] Barton N, Lien R, Lunde J. Engineering classication of rock masses for design of tunnel support. Rock Mechanical 1974;6:189236. [11] Deere DU. Technical description of cores for engineering purposes. Rock Mechanics and Engineering Geology 1964;1: 1722. [12] Dempers GD. Optimal usage of exploration core for geotechnical purposes. Unpublished Report, Steen, Robertson and Kirsten, Consulting Engineers, Johannesburg, 1996. [13] Haines A, Terbrugge PJ. Preliminary estimation of rock slope stability using rock mass classication systems. Proceedings of the 7th International Congress International Society Rock Mechanics, Aachen,vol. 2. Rotterdam: A.A. Balkema, 1991. p. 88792.

to generate over R1.2 billion in additional revenue. Empirical design charts have been developed for stable slope angle in relation to stable slope height for the geotechnical zones. As a result, stack congurations have been adjusted to prevent or accommodate failures in these zones. Slope monitoring is necessary to ensure that the open pit slopes are safe and movements are within acceptable limits. A comprehensive monitoring system is being implemented and data gathered therefore helps to provide a rational basis for evaluating stability conditions. In other words, a record of displacement data can be used to help predict the behaviour of a potentially unstable slope.

References
[1] Sen Z, Kazi A. Discontinuity spacing and RQD estimates from nite length scanlines. International Journal Rock Mechanics Mining Science and Geomechanical Abstracts 1984;21:20312. [2] Stimpson B. A rapid eld method for recording joint roughness proles. International Journal Rock Mechanics Mining Science and Geomechanical Abstracts 1982;19:3456. [3] Barton N. The shear strength of rock and rock joints. International Journal Rock Mechanics Mining Science and Geomechanical Abstracts 1976;13:25579.