Rockburst, the sudden violent failure of rock, has been a significant hazard in hard rock mining for many decades. Andrieux et al. (2013) provided a series of rockburst case studies from several mining camps, highlighting the problem and documenting some of the risk mitigation strategies used. Despite of recent progresses, dynamic ground failure remains a major hazard for many mining companies in high-stress hard rock environments. One of the difficulties is that there are many different rockburst mechanisms in different geological environments.
This paper is organized as follows. After the Introduction, Section 2 describes some of the techniques used to combat rockburst hazards in deep hard rock mines. Section 3 uses field observations to discuss ground support performance and requirements under dynamic loading. Section 4 examines a deep mine’s database of rockburst/fall of ground incidents to illustrate the relative success of risk mitigation techniques. Finally, the paper concludes with a discussion of areas where continued improvements could be required to further reduce risk to worker safety in rockburst-prone mines.
Rock mechanics tools to mitigate rockburst risk
Given the erratic nature of rockburst, and the difficulty of exactly predicting when or where a burst may occur, ground support is one of the principal risk mitigation tools. Wherever possible, permanent infrastructure should be placed outside the influence area of ore extraction driven stress buildup. Remote access, e.g. non-entry stoping methods, is also extensively used. However, the nature of mining is that stresses will inevitably increase upon extraction, and faults or other geological structures will be intersected by larger mined-out volumes, both of which will gradually increase the risk of larger seismic events and rockbursts over time. Nevertheless, even if ground support is successful, the intent is not to have personnel exposed to the injury of large ground motion. In this case, avoidance strategies such as improved mine design (e.g. fewer or no converging mining fronts), barricading, and seismic re-entry protocols are used in conjunction with dynamic ground support. It is best to avoid a fault, or cross it at 90°, but you need to know where the fault is. Better mine design thus requires knowledge in, for example, structural definition.
Mining in yielded ground or stress shadowed areas is a proven method to reduce rockburst risk. The rockburst database shown in Section 4 is from a primary/secondary stoping mine. No bursts occurred in the secondary stopes. The mining method used leaves slender secondary pillars that yield when the adjacent primary stopes are mined. Accesses are only driven into the secondary stopes after the yielding process is complete. The support schemes need to cater for gravity, blast vibration, and corrosion (the yielded rock mass becomes more permeable and water infiltrates in mining process). Roughly 50% of the reserve falls into this category. There were 60 rockburst incidents over a 69-month period, which mainly occurred in development or primary stope access. There was only one case of relatively minor personnel injury in the 60 incidents; however, there was potential for more serious injury. The risk mitigation strategies in Section 4 illustrate the relative effectiveness in reducing rockburst exposure. The risk mitigation strategies are from a Canadian hard rock mining perspective, which implies a high degree of mechanization.
It is clear that an improved mine design can reduce rockburst potential. Converging mining fronts (sill pillars), for example, are notorious for increasing risk. Project economics dictates that fast ramp-up of production and early access to ore will improve the net present value of a deposit. Fast ramp-up often implies multiple working fronts, which can lead to multiple converging mining fronts in the future. More challenging environments will require more ingenuity in the mine design process.
A few general principles to lower rockburst risk from a mine design perspective are presented as follows:
Yielded ground does not store excessive strain energy. Timing and sequencing of excavations can take advantage of this principle.
Infrastructure placement needs to offset the active mining. Numerical models can estimate the required offset distance. Mining methods such as caving can have a very large zone of influence as extraction builds.
Take advantage of stress shadowed areas. Many deep mines in South Africa employ a strategy of following behind development. The reef mining is conducted ahead of the main haulages so they can be driven in a stress shadow.
Backfill does not burst. Underhand methods, or establishing blast-hole overcuts under backfill from adjacent top mining, are commonly used.
As noted above, sequencing and avoiding converging mining fronts can dramatically reduce excessive stress buildup, and thus lower rockburst risk.
Stope blasting can be designed to excavate in as few mining steps as practical. Each blasting may require drill-hole cleaning and another loading cycle, thus increasing worker exposure to high-stress conditions:
Pre-loading of drill rings to be fired in the next blast;
New wireless blasting caps are emerging which may reduce exposure even further.
Conversion of mining method to reduce worker exposure, for example, switching from narrow vein cut-and-fill mining to blast-hole methods:
High-grade narrow vein mine may have high recovery and less dilution with cut-and-fill methods, but high worker exposure.
Switching to a blast-hole method can facilitate remote mucking, extracting the ore in fewer steps. Smaller dips may require some waste extraction to make broken rock flow.
The switch to lower exposure methods may be required as mining fronts converge (sill pillars).
Kaiser et al. (1996) and Kaiser and Cai (2018) described the requirements of reinforcing the rock mass, retaining fractured rock, and holding the material in place. A significant update from Kaiser et al. (1996) is the recognition that many rockbursts are in fact induced by the strain energy around the opening being suddenly released by far-field seismic events such as a fault slip. The peak particle velocity (PPV) approach based on the seismic event itself often does not account for the excessive strain energy around the tunnel. It may just trigger the coiled spring scenario. The seismic event associated with the exact burst damage location is likely lost in the waveform of the larger event (Simser, 2017a).
Fig. 1 shows a brecciated dyke fragment (very stiff inclusion) that was ejected from a stope access wall. This was an example of a coiled spring being suddenly released. Support demand needs to consider this mechanism, and empirical observations show the ejection velocities of strain energy bursts in the range of several meters per second. PPV measurements from large events at some distance are often in the range of hundreds of millimeters per second.
Stiff support elements are good at reinforcing the rock mass, inhibiting dilation, and helping to preserve inherent rock mass strength. Stress-induced fracturing around openings can reduce rock mass cohesion, but friction can be mobilized when rock fragments are held tightly together. Large dynamic displacements, on the other hand, require support elements with yielding capacity. Very stiff support works well for gravity loading and inhibits dilation, but can rupture suddenly during a rockburst event. Current practice is generally to use relatively stiff tendons and highly compliant mesh. Yielding tendons are readily available in the Canadian market, but they are still much stiffer than the mesh used as surface/areal support. However, the tendon/mesh stiffness contrast leads to weak link failures (Simser, 2007), which can be reduced by careful mesh overlaps and well-designed plates. Examples are given in Section 3.
The complexity of the dynamic ground support problem has led to numerous researches, notably in the dynamic testing (Ortlepp and Stacey, 1997, Player et al., 2004, Hadjigeorgiou and Potvin, 2011, Roth et al., 2014). Field observation of in situ rockbursts is still one of the best sources of true support performance (Ortlepp, 1997, Simser, 2007, Li, 2010).
A significant problem is the ability to forecast the future performance of support. Analysis of rockburst incidents from one mine in Section 4 provides a good example. The original development utilized in-cycle shotcrete and rebar. After several years of mining, the buildup of mining-induced stresses started to increase the number of rockburst incidents. The percentage of large rockbursts was relatively small (with block mass less than 5 tonne, often less than 1 tonne). The number of this style of incident has dropped over time despite increasing overall mine extraction. The change of support standards in terms of mesh on the outside has dramatically decreased small shotcrete failures. During the first few years of operation, over 30 km of tunnels were excavated. Proactive support upgrades were completed in selected older areas prior to a specific stope extraction, and the newer development had mesh as the outer layer.
Morissette and Hadjigeorgiou (2017) created a support design strategy for dynamic loading based on statistical treatment of 14 years of Sudbury Basin rockburst records. Empirical experience remains one of the best ways for dynamic support design. Counter (2014) provided some excellent case histories of successful deployment of dynamic rock support in a very deep mining environment. Ground support solutions are always a balance between the need for economic application (ease and quality of installation, and productivity) and performance. Equipment selection also plays an important role; different practices in different mining camps are common, partly based on the local preference of equipment types.
Rockbursts are generally associated with or triggered by seismic events. Seismically-induced falls of ground can also occur. One mechanism is the momentary loss of clamping forces as the seismic wave passes through, for example, steep jointing in a high horizontal stress field. The digital transformation of many industries, including mining, has led to a steadily improved capacity to monitor mining-induced seismicity. Early seismic systems deployed large copper wire lines with multiple pairs, branching out into many different paths to cover different mining areas. More modern mines have fiber optic backbones, which allow for many short copper running to the nearest fiber hub, and easier deployment of dense arrays. Wireless data transmission has the potential to make further improvements. A modern deep hard rock mine arguably should not operate without a seismic monitoring system. Seismic hazard can be assessed using many different techniques (Mendecki, 2016), which can lead to more reliable deployment of enhanced ground support schemes.
Rockburst prediction (exact time/location) remains elusive, but identifying areas with higher rockburst potential is a common practice. The recorded data are also one of the best sources of information to assist in determining what occurs, and why it occurs in rockburst investigations.
Spatial resolution of recorded seismicity is very important and often ignored. Hydraulic fracturing monitoring in oil industry has led to the development of robust sensors which can be installed in deep boreholes. When combined with traditional short hole sensors from mine excavations, a better three-dimensional (3D) array is possible/practical (Butler and Simser, 2017). Some seismic data analysis techniques (e.g. passive tomography or spatial contouring of source parameter data) attempt to identify high-stress anomalies. Stress buildup around mined-out areas occurs at the scale of a few meters to a few tens of meters. At the tunnel scale, the transition from fractured ground to high-stress anomalies can occur over a few decimeters. A common burst observation is that the fracture zone around the opening was the material that was ejected.
For stoping layouts, understanding the depth of fracture zone can guide the placement of the next stope access. Heavily fractured or yielded ground becomes aseismic. Many mines look for areas that were historically seismically-active, and have subsequently become very quiet as the evidence of rock mass yielding. Event location error can blur the lines. Precise locations can accurately delineate the transition between the fractured rock and the high-stress interface next to it. Accurate event locations can also reveal planar trends indicative of fault movement. Asperity detection may be possible, for example, if a portion of a fault is loaded. High-stress events, or simply clustering of events along the fault trace, can indicate a potential source of future larger magnitude activity.
Seismic network sensitivity and location accuracy are impacted by many different factors. Sensor density is perhaps the most influential factor, with tight arrays having sensor spacing in the range of 200 m3/sensor. Sensor type and the ability to mix sensors are important. High-frequency sensors can be used to detect smaller magnitude events (accelerometers), but low-frequency sensor response is required to quantify large events (geophones). Geological complexity affects wave propagation and velocity, and 3D velocity models instead of simpler single velocity models may be required. Mining creates voids, which often means that the fastest arrival times are not achieved in a straight-line path but in a curved path around the opening (3D velocity model incorporating voids). Better coverage of 3D models is practical to be achieved using long borehole sensor installations. Geological definition drilling often puts holes in good places for seismic monitoring. Sensitive arrays also record development blasting. For example, the first shot fired in a 5 × 5 m tunnel advance can be used as a location accuracy calibration. The survey coordinate of the central portion of the tunnel face is a reasonable estimate for the real location, which can be easily compared to the location calculated by seismic system.
Exclusion strategies (re-entry protocols, barricading off old areas), and remotely operated equipment can dramatically reduce exposure to potential rockbursts. Seismic monitoring data are a four-dimensional record of how the rock mass has responded to mining (time and location). Learning where the rock mass has elevated activity helps to determine where it will have future activity. If an area suddenly becomes seismically-active, the information can be used to temporarily evacuate portions of the mine.
Proactive exclusion strategies are often used. For example, pre-defined barricades are placed prior to blasting of a specific stope. Personnel access beyond the barricade is only allowed after a pre-defined time, and the seismic system is checked to assess the actual response. Determining the re-entry time can be as simple as excluding work for one shift, or using seismological tools to estimate when the activity has returned to background levels (Tierney and Morkel, 2017).
The data shown in Fig. 2 are from a small-tonnage narrow vein operation. In the specific region of the analysis, 96% of large events occur within 96 h of blasting. Remote equipment or a simple long duration exclusion period can dramatically reduce risk. However, not all the large events occur in this time period, thus in this case, rockburst support is still used. In most Canadian mines, exclusion strategies vary but are typically around one or two shifts (8–24 h). This example illustrates the simple fact that longer exclusion periods incrementally reduce exposure to large seismic events. The time lag of large events after blasting reflects the fact that uncertainty in structural responses to mining exists in this specific area.
Experienced miners can also gain a feel for the rock mass response, and evacuate an area based on local rock noise or other signs of poor ground conditions (support systems taking weight, suddenly wet conditions, and cracks dilating).
Tele remote equipment is an obvious area that more focus should be put on as mining goes into more challenging deeper environments. Reducing worker exposure is perhaps a more pertinent driver than the potential productivity gains. Current technology allows for both lines of sight remote mucking, or the load-haul-dump (LHD) machines are routinely equipped with cameras and proximity detection so they can be operated in a control room. Drilling machines (jumbos, blast-hole drilling rigs) and shotcrete sprayers are routinely operated remotely. Remote ground support installation (tendons and mesh) is more difficult, as is the remote explosive loading for development rounds.
Stress levels that store energy or drive structures can be estimated via numerical modeling techniques; however, the models are often based on limited data. Numerical models often assume average rock properties, partially for ease of computation, and/or partially due to lack of precise geological domains. Local variations may change behavior, from strain softening/rock degradation to strain energy storing and eventual violent failure.
Most burst observations uncover some unique ingredients, for example, a joint that does not appear in the tunnel controlling the fracture pattern around it (Section 3.1), or a brecciated dyke fragment that is not explicitly delineated by the geological model (Section 3.2). Detailed geological domain in complex environments may not be available, and thus not be input in numerical models.
The models can be useful to back analyze a burst, for example, to test the mechanism hypothesis. From a proactive planning perspective, mine models can be used to estimate elevated stress regions within the mine as the extraction sequences progress. These may or may not result in rockburst, but generally mean that higher stress areas imply higher risks.
New methods for evaluating rockburst risk
The rockburst risk to personnel still remains. Current management plans reduce risk dramatically, but there is still a need for improvements. Rock mass characterization, for example, can benefit from borehole geophysical surveys. Simser and Hall (2018) and Simser et al. (2019) showed examples of borehole breakout as measured by acoustic tele-viewer (ATV) surveys correlated to rockburst in a ramp development. The data were collected using scout holes ahead of the advancing development headings. More intense breakouts (stress notching) occurred in an area that had a rockburst in the ramp. There were also correlations between higher breakout intensity and higher energy release. Seismic sensors installed in the scout holes after the ATV survey were used to monitor the ramps. There is clearly a benefit in obtaining early information, for example, in the development stages of a mine prior to stoping induced stress changes. Overall extraction can increase overall rockburst risk (e.g. higher abutment stresses, and large geological structures intersected by more voids).
Managing rockburst risk in mining tunnel development
In some geological environments, strainburst risk exists when excavations are first driven. Risk is elevated when mining-induced stresses occur, but risk mitigation is often required in the tunnel access development prior to stoping operations. The current drilling, blasting, mucking (removal of broken rock), and supporting cycle are not routinely done remotely. Different equipment facilities are used in different mines, with varying amount of worker exposure. Some options are listed as follows in order for the lowest worker exposure:
Two boom automated jumbos with enclosed cab (low exposure);
Single boom narrow vein jumbos, sometimes with canopy (moderate exposure); and
Hand held drills, jacklegs (also commonly referred to as airlegs), hand-held drills for wall or face drilling, stopers designed for back-hole drilling and stopers (high exposure).
String loaders for emulsion, and basket loaders for ANFO (lower exposure), which can have canopy or cage for loading basket; and
Hand loading of cartridge explosives or small ANFO pump (high exposure).
Mucking (removal of broken rock)
Remotely operated LHD machines (low exposure); and
Slushers (higher exposure).
Boom style bolters with carousel, and enclosed cab (low exposure);
Jumbo style (Australian) bolting (low exposure);
Platform bolters (scissor decks with automated drill, moderate exposure); and
Hand-held drills (stopers, jacklegs – high exposure).
Many mines install bolts and mesh in the face to protect the loading crews.
De-stressing or pre-conditioning
Additional blast-holes are added to the pattern, outside planned excavation perimeter, and they are designed to fracture the rock so that it cannot store much strain energy.
Seismic re-entry protocols, allowing time for seismic activity to decay after blasting.
Other simple measures to reduce risk in the development cycle include smaller openings, shorter advances, and staged excavations for wider spans (which open up incrementally and add long support prior to widening the span).
Empirical evidence such as underground observations remains one of the most important aspects of rockburst support design. Support systems have various elements with different stiffnesses. Mesh is compliant and can deform in decimeters. Yielding tendons can absorb kilojoules of energy and deform in centimeters. Weak links (Simser, 2007) can render the system ineffective, for example, a sharp-edged bolt plate guillotines the mesh. Soliciting the capacity of elements, especially tendons, is not obvious in the field. Rock outward crushing does not necessarily transfer load effectively to a yielding bar with small cross-sectional area (e.g. 22 mm diameter tendons). The plate connection has to both transfer load efficiently and prevent mesh punching. The following sections provide examples of rockburst in hard rock mines. Additional risk mitigation strategies were in place: mainly exclusion periods (re-entry protocols) or restricted access (failure areas were behind barricades) and no injuries occurred in the cases shown.
One of the difficulties in predicting future rockbursts is that the lead-up to the event may be very subtle. The example shown in Section 3.1 was triggered by a stope blasting. There were no obvious visual signs of support systems taking load prior to the blast (bolt plates bending, mesh with accumulations of loose rock and so on). The example shown in Section 3.2 is a stope cross-cut that had been developed six years earlier. Again, visual observations of the local ground support did not provide indication for the pending rockburst. Simser, 2017a, Simser, 2017b showed, in hindsight, that signs of stress buildup could be inferred from the seismic history, but more work is required to make reliable forecasts.
Rockburst damage of weld mesh and rebar support
Tunnels in high-stress/brittle rock can develop a tightly spaced stress fractured zone around the opening. Often this material ejects during a burst. Strain energy around the opening is concentrated at the interface between the fractured rock and the intact rock just beyond it. The fracture zone dilates, and shears, using up some of the ground support system capacity over time. The inherent rock mass strength is also reduced: cohesion loss, and dilation-induced frictional loss. Waveform amplification from far-field seismic events occurs at free surfaces and in the fracture zone around the openings (Van Sint Jan and Alvina, 2008). The damage shown in Fig. 3 is induced by a near-by stope blast and a moment magnitude 1.1 seismic event a few milliseconds later.
The violent failure occurred predominantly in stress fractured material in the back of the drive. High horizontal stress and a flat structure coming over the tunnel were present. The flat structure effectively traps stress between it and the opening, creating a more intense and deeper fracture zone than that normally observed in this part of the mine. The burst material was high-strength granitic rock (unconfined compressive strength (UCS) of ∼250 MPa) surrounded by a low-strength ore zone (UCS of ∼120 MPa). The stiffer/stronger rock concentrated load, and the stored strain energy was suddenly released by the stope blast and/or seismic event.
Fig. 4 shows a zoomed-in view of the burst. Numerous instances of weld mesh failure occurred, which was often triggered by sharp edged plates guillotining the wire strands, as did both tensile and shear failures of the rebar support. Several bars did not break but had the stress fractured rock being unraveled around them.
The mine used central blasting and did not allow re-entry into nearby areas until seismic monitoring results have been evaluated (re-entry protocol). In this case, these measures adequately managed the risk of rockburst. The knowledge gained from the incident led to more widespread use of rockburst support, with local waste rock gaps being identified as higher risk locations.
Dyke fragment burst
Geological dykes can be notoriously fine-grained and exceptionally strong. The dyke material shown in Fig. 1 is fine-grained with high-strength (UCS > 400 MPa), but very erratically distributed in the overall rock mass. Brecciated dyke fragments in this mine vary in sizes from tens of cubic centimeters to tens of cubic meters. The seismic event that triggered the burst was a moment magnitude 1.6 event located close to the observed damage, and within a blast re-entry protocol. The original development took place 6 years ago before the observed damage. The coiled spring was released when the seismic event wave passed (Simser, 2017a, Simser, 2017b). The small fragmentation and high ejection velocity require a safety net to catch the flying rock. Mesh with high deformation capacity is necessary. Transferring the dynamic load to the yielding tendon is also a challenge – large footprint plates or plate enhancements (e.g. plates over straps) improve the overall system performance.
Dynamic support enhancements
Fig. 5, Fig. 6, Fig. 7 show the examples of adding support system enhancements to lower the risk of weak links failing. Support design can be misleading if the details of a well-designed support system are not carefully considered. For example, a dynamic bolt with capacity of 40 kJ installed in a 1 m × 1 m pattern could produce a support system capacity of 40 kJ/m2. If the material flies out between the bolts, the plates prematurely buckle and pull over the bolts. The mesh retaining system fails, and then the design is a false sense of security.
Empirical field observation of ground support behavior during rockbursts remains one of the best ways to improve support systems. Lower personnel exposure is the first line of defense for a given mine design, but uncertainty in the timing and location of rockburst remains, thus rockburst prone support systems still play an important role in risk mitigation.
Example of rockburst database
A database of rockburst incident occurring over a 69-month period was back analyzed to highlight the relative effectiveness of rockburst risk reduction strategies. The data are from a relatively deep, mid-sized producer with characteristically hard rock. Most country rock types are of high strength (>150 MPa) and can be prone to bursting in the right set of circumstances.
The details behind the data shown in Fig. 8 help to illustrate the relative success of risk mitigation strategies. Thirty three of the 38 incidents, where fiber reinforced shotcrete was selected as the surface support system, occurred before 2015. The mine had started as an in-cycle shotcrete operation with no mesh. Gradually, as extraction built up and strainburst increased, the mine standard changed to mesh on the outside. Many developed areas had proactive support upgrades (with mesh on the outside) prior to firing stopes where significant stress changes were anticipated. The legacy of the original project development meant that some areas were vulnerable to small-scale rockburst damage. As the support changes took place and the retro-fitting vulnerable areas were caught up, those types of incidents were dramatically reduced.
The contained support categories in Fig. 8 are under-represented. Loose fragments contained in mesh may have come from pseudo-static fracturing/deterioration, or small strainbursts. Differentiating the two is not always possible. There are most likely cases of contained damage that was not directly linked to a strainburst.
Ten of the damage locations were in the unsupported face area. This relates to the development cycle, i.e. drilling, blasting, mucking, and then installing the ground support. Reducing the exposure to unsupported ground is the key risk mitigation for this category. Remote drilling and mucking are routinely possible. Remote loading and installation of ground support are more challenging. The development tunnel face is supported with splitsets and mesh at this operation. However, that process only occurs after the walls have been supported.
The risk reduction potential can be described as follows: the total number of incidents – the number of incidents where risk mitigation would have been successful. For this dataset, always having mesh on the outside, barricading off risky areas, and keeping people away from unsupported ground would have reduced the risk of injury from a potential of 60 incidents to 2.
Discussion and conclusions
A multi-tiered risk reduction strategy can dramatically reduce the number of potentially injuring rockbursts in underground mines. Ground support systems must recognize the need to survive rapid loading and high deformations. Weak links need to be engineered out of the system. Load transfer to tendons, which typically accounts for the largest portion of the potential energy absorption capacity, needs to be considered. A quandary exists between the need to keep wedges tightly in place and limit fracture zone dilation versus the need to yield during dynamic loading. Even amongst yielding bolts, significant stiffness contrasts exist. A softer bolt that can go for the ride more easily will arguably be less damage to surface support. Mines prefer to the simplicity of one bolt type to cater for most conditions. As more challenging environments are encountered, this may not be a luxury anymore (which may need mixed mode tendons, e.g. softer + stiffer).
Seismic monitoring remains one of the best tools for estimating rockburst risk. The effect of location accuracy is often ignored. Volumes of rock with low risk include yielded zones or stress shadowed zones around openings. The source of high stress (strain energy) can be right next to the fractured interface. At the tunnel scale, this is in the decimeter range, but at the stope or mining front scale, it is in the meter range. For example, a primary/secondary stope sequence may have a full panel width of yielding for moderate strength rock. This would imply that the next primary stope access in the abutment would drive at the high-stress interface. A switch to pillarless mining may be appropriate. A stronger rock mass may have ∼1/2 panel width of yielding. Pillarless mining would put the next stope access in the high stress interface. Seismic events that can be located accurately (errors in the range of ∼5 m) can delineate these zones of yielded ground (Simser, 2017a, Simser, 2017b). These accurate seismic data of how the rock mass responds can be used to calibrate numerical models. The modeling approach is good at forecasting the future, so long as it is bounded by good observational data.
Exclusion strategies are not likely to eliminate all potential risks, tunnels need to be developed, and man entry activity is still required for some aspects of the mining cycle. However, they can clearly reduce exposure. For example, a blast-hole undercut needs to be developed, but once established, all future activities can be completed with remote equipment. Mines that use tele-remote scooptrams can put on raise-bore reamers for slot raises on remote, and backfill barricades can be remotely placed and then remotely shotcreted.
Other mine design tools such as numerical models and scheduling packages can be used to optimize sequences and avoid converging mining fronts. The challenges of going deeper in many mature mining districts will be safely met by combining best practices with improvements in remote technology.
Source: Rockburst management in Canadian hard rock mines