The most important consideration for developing a ground control strategy is the rock mass
behavior. (Photo: Golder)

At depth, rock masses tend to be more massive and brittle, and often fail by fracturing of the rock itself. This may occur slowly in the form of spalling (e.g., the progressive slabbing of the rock) or rapidly in the form of a rock burst (e.g., a violent failure of the rock). Dr. Rob Bewick is a principal at Golder (a member of WSP). Bewick leads the company’s global deep and mass mining team and spoke about the challenges involved in designing ground support strategies for very deep operations.

“Many of the conventional rock engineering design systems or methodologies used were developed from shallow mining experience where the rock itself does not fracture,” he explained. “While there are deep mining assessment methods available, they are often overlooked. When brittle, rock mass design approaches are appropriate but not utilized, the resulting design is at risk of being unreliable, putting mine safety, production reliability, and stakeholder value at risk.”

The most important factor that needs to be considered in developing a ground control strategy is the rock mass behavior, or how it fails. This must be identified, and the appropriate design methods for that behavior adopted so that a safe, reliable design is developed.

Following this, a ground control strategy that incorporates the following three pillars can be developed: • Strategic: includes overarching aspects that can be designed up front such as the mine layout, mine sequence, mine equipment, mine-wide monitoring systems, etc. • Operational: aspects that can be controlled by the mine such as mining rate, re-entry protocols, etc.; and • Tactical: aspects such as ground support, proactive support maintenance, local monitoring, exclusion protocols, etc. “On June 20, 1984, around 200 workers were underground at the Falconbridge mine in Sudbury when a 3.5 Richter magnitude seismic event occurred, triggering a rock fall and trapping many of the miners, four of whom lost their lives,” Bewick said.

“In Canada, since this event, a focused amount of research, development and innovation has occurred to improve the understanding of rock-mass behavior under high stress and to develop mitigation strategies that can be grouped into the main pillars noted above. While improvements have been made, the main challenges of rock mass progressive and violent fracturing are still present.” Andy Thomas, principal rock mechanics engineer, and Jarek Jakubec, mining and geology practice leader, both based at SRK’s Vancouver office, joined the conversation. Thomas explained: “Many factors need to be considered [when developing a ground control strategy for deep operations], but understanding the stress setting and its response to mining are key. Perhaps the main risk in deep hardrock mines for ground control is dealing with the effects of seismicity.” Jakubec agreed: “There are notable recent examples where operations have experienced significant ground control problems because the traditional mining strategy was not adjusted for the deeper levels. In these cases, the issue was not with the capabilities of the ground control systems or technology, but rather the design fundamentals such as method selection, layouts, advance direction, sequencing, etc. Although new technology and systems will have a role, these types of fundamental considerations will continue to have the most influence as we mine deeper.”

Learning to Design for Depth
One concern with today’s acceleration of deep mining is that there are very few universities and colleges that teach deep mining ground control or engineering design. “In general, human capital is limited in the mining industry and the field of ground control is no different, making skilled and experienced personnel hard to find,” explained Bewick. “As a result, there is a trend of using conventional rock engineering design systems/methodologies to design at depth. Mine designs based on this conventional thinking have the potential to be not only unreliable but unsafe.”

Cave mining is a case in point. The method has a long history of application at shallow depths but is now being used more frequently at deep, high-tonnage operations. These operations often encounter more moderately jointed to massive rock masses instead of the highly jointed rock typically experienced with shallower caving operations. As a result, they may be exposed to dynamic brittle failure of the rock and associated hazards, like rock bursting. To counter this, mining companies are starting to incorporate knowledge and understanding from other mining methods, such as sub-level open stoping from Canada and deep reef mining in South Africa, where mining in high-stress conditions has been the norm for more than 30 years.

Bewick has contributed to this movement through innovations in rock-mass characterization and the application of deep mining design approaches to caving. In addition to authoring many of his own papers, he also sits on the Ph.D. committee for Justin Roy who is completing his research at the University of British Columbia sponsored by PT Freeport Indonesia. The purpose of Roy’s research is to investigate the dynamic response of rock that is massive, brittle, and veined when exposed to excessive loads through high-tonnage block and panel cave mining methods. It recognizes that the engineering experience base with block cave mines at these depths is extremely limited, and that new mechanistic understanding is needed to safely manage the operational hazards and risk that these uncertainties impose.

“It’s important to remember that ground control is not just about ground support and monitoring,” Bewick said. “It encompasses the strategic, operational, and tactical pillars [outlined above]. And, if a program based on these elements is well developed and properly implemented then, from my experience, even the most challenging mining conditions can be safely operated in. “When mines have had to stop mining, the root cause is often found within the strategic ground control pillar. For example, mine layouts are fixed once created. If a mine layout promotes deep mining challenges, as opposed to mitigating the challenges, management of the conditions through operational and tactical means is often not enough.

“If we bring this line of thinking back to the training challenges facing our future engineers in universities and colleges and the limited skilled human capital available, the acceleration of deep mining is outpacing the ability of our academic training grounds to make modifications to course content so that it remains relevant to mining industry needs.”

Step Change Through Digital
While the fundamentals of ground support and reinforcement have not changed markedly over the past 20-30 years, the technologies available to investigate and monitor ground response have advanced significantly. Better and more reliable instrumentation such as smart cables, smart markers and seismic monitoring systems, provide excellent data for near real-time decision making and strategy adjustments. “The challenges are similar, but we are better equipped to evaluate and prepare for them,” Jakubec said. “There has been significant advancement in our experience of deeper mining environments as well as in engineering solutions (e.g., automated equipment) to operate in them.

“The technique of drone surveying of underground openings is becoming very impressive. The technology is allowing drones to operate autonomously and collect high-quality data remotely. I believe that robotization and AI will play a bigger and bigger role, eventually replacing people in high-risk areas and make mining safer.” The ever-increasing speed of computing and the sophistication of numerical modelling codes is also enabling the modelling of ever-more complex mining problems. Provided these models have realistic input parameters and are calibrated with ongoing monitoring data, they can be highly valuable for informing and testing ground control strategies. Pre-conditioning of the rock mass to induce seismic potential is another hot research topic at present.

The team at SRK collaborated with Elexon Mining to advance its wireless marker technology, Geo4Sight. This instrumentation is ideally suited for deep underground mine settings. “The prospect of incorporating more instruments into wireless markers for even more functionality, for example, instruments for seismic monitoring, is exciting,” Thomas said. “We are also involved with a client from Europe, evaluating safer mining strategies for deep mining levels where seismicity is making current mining layouts increasingly risky. Automation of the mining process is a big part of it.”

Bewick believes that disruptive change will be seen very soon in ground control for deep mining operations. “While things have looked similar over the past decade, I expect the progression of change over the next 10 years to be drastically evident,” he said. “This is because the context of mining is changing. There are more ultra-high capital investment mines with greater risk related to rock engineering (especially around the challenges of deep mining) and its impact on production reliability. The rapid digitization and simulation of the world that is already impacting equipment, material flow and processing will eventually impact rock engineering too.”

The main limitation for change in rock engineering today is the data stream; there is still insufficient data related to the rock-mass response to mining coming from the underground, but there are some mines getting close to, or reaching the required data flow. “This can and will be solved with current sensor technology on the market,” said Bewick. “It’s only a matter of time. New sensor technology, data analytics, image recognition, and visualization will all help move rock engineering into the connected, virtual world in operating mines.”

Two of the main technologies that create the ability for change are cloud computing access and capability, and machine-learning algorithms and platforms. While current modelling codes are not fully optimized for cloud computing efficiencies, they are in the process of being optimized and new tools are being developed specifically to take advantage of the cloud that do not have the constraints of current tools. Bewick explained: “With codes optimized for the cloud and current processing technology, the models of today that take weeks to run will be computed in a matter of hours or minutes. With the cloud storage and infinite virtual computers for rent, in a single day, we will see hundreds of mine-scale models simulated (as opposed to the handful that are simulated now) and true reliability-based designs will be developed because of probabilistic assessment. “The results could then be fed into discrete event simulation tools (if a coupled framework is not yet developed) and the rocks’ response to mining can be used to efficiently and more realistically design, schedule and plan a mine, adding immense safety and value to stakeholders. This is the short-term life of conventional simulation in mining.”

In the longer term, with the data streams from mines being stored and analyzed, ground response to mining will enter a data-rich environment allowing for machine learning to disrupt the industry and making conventional simulations (in some cases) redundant. Bewick believes that creating a true digital twin of each mine will eventually become common practice. The twin will be updated in near real time and provide forecasts of mining challenges for better safety management, planning, and scheduling of resources and the material supply chain.

Building Orebody Knowledge
Change is also coming to data collection and, as orebody knowledge is absolutely crucial for project and mine risk management, this could play a big part in better ground support strategies. Bewick said soon conventional core logging will be replaced with smaller initial algorithm-training data sets for a rock mass with reliance on core photograph and scanning data for near continuous data extraction. “Every borehole drilled will provide fast, cheap, and full continuous data,” he said. “Consulting firms need to be ready for this change.”

As Jakubec mentioned earlier, image assessment of the underground environment using LiDAR, photos, and other scanning type data will also provide for a fully reconciled underground environment. This will change conventional classification and characterization at the mine. Bewick believes these tools will be used to relate the ground environment, mining, and rock-mass response in near real time and provide forecasts, making the need for classification systems and conventional characterization after the project stage unnecessary. “Even at the project stage, once mines have enough data flow and correlations have been processed to form common mine response machine-learned models based on ore deposit type and geologic rules, mine simulation work may no longer require any of our conventional approaches,” he said.

Linked to all of this, there is a significant risk of going backward in rock engineering; an experience transition is currently under way with many previous thought leaders retiring. “Systems developed for rock engineering in the past have stated specific limitations of applicability that were well known to those at the time but are now at risk of being used when they should not be,” Bewick added. “Also, many of today’s universities and colleges are not keeping up with industry needs around deep mining knowledge transfer.

“If we are not careful, the step change that is imminent or currently happening may get stalled due to loss of our history, and steps back may be taken because conventional rock engineering thinking is propagated as opposed to deep mining learnings and thinking.”

Optimizing Ground Support Systems
One educational institution with its finger firmly on the pulse of deep mining is the Australian Centre for Geomechanics (ACG) at the University of Western Australia. There, Professor Yves Potvin and his team are developing knowledge and tools that can be used to optimize ground support systems. While these are currently being developed with standard ground conditions in mind, the team’s research will be vital in helping mines balance the cost of ground support with their tolerance for both safety and operational risk as they chase deeper orebodies in more extreme conditions too.

Phase one of the Ground Support Systems Optimization (GSSO) project completed in 2019, and its learnings were distilled into a comprehensive guide; Ground Support for Underground Mines was published in March 2020, and Potvin is currently leading phase two of the project. He joined E&MJ to discuss its aims and lessons thus far. “Phase one of the GSSO project started in 2013,” he began. “There was still a mining boom following the 2008 dip and, because of that, there was very little tolerance for production interruptions across the industry. Mines were very profitable and could afford to install a lot of ground support. So, for a period of 10-15 years, there was little focus on optimizing this element of strategies.

“I realized that there were very few optimization tools and techniques available to operators. Unless mines can demonstrate to inspectors or the social environment that they can modify or even reduce ground support without compromising on safety, they will not be able to optimize. That’s when we began to discuss the possibility of developing a probabilistic approach to quantifying the risk of failures. “We also discovered that, when it comes to ground support system design, the mining industry relies heavily on civil engineering techniques. So, we used the project to develop an empirical system that is mining based, to give mines a tailored first-pass approach for their ground support design.”

Probable Improvements
Phase two of the GSSO will be complete in September 2021, and it has two main themes: firstly, designing guidelines for ground support in extreme conditions, and second, further developing the probabilistic approach for system optimization identified in phase one.

“The term ‘extreme conditions’ refers primarily to rock bursting and squeezing ground,” Potvin explained. “Mines typically manage those through monitoring using seismometers and ‘dynamic’ ground support systems. The third way is to reduce exposure, using exclusion zones, protocols and/or autonomous processes, which move people away from areas where seismic events could potentially occur.” Currently, there are two ways to understand how ground support will behave or respond to seismic events: either through a drop test (a heavy weight is literally dropped on an element of the support system and its response measured) or through blasting to test the support as a whole system. Both techniques are expensive and have limitations.

Although receptive to research, most mines aren’t keen on the idea of purposely destroying one of their hard-earned tunnels, so GSSO’s proposal to use statistical methods to gauge system’s probable response to certain rock conditions was well received. However, given the large amount of natural variability rock masses demonstrate, this was no easy task. “We have to try and account for this variability, estimate the types of loads that will need to be sustained, and the capacity of the support system to generate a probability of failure,” explained Potvin. “If a mine wants to be very conservative, then they would use all the worstcase conditions in their design. If their appetite for risk is higher, they use best-case conditions. But really, they need to look at the full range of ground conditions and use statistics to properly assess the risk.” The value of this approach is that it allows mines to adjust designs according to their risk tolerance. “The power of this approach is that, rather than using a factor of safety, which is a rather crude assessment, you can use all of your data that reflect what’s going on at each site to make informed decisions,” Potvin said.

The GSSO team has spent a lot of time on developing new tools — software programs essentially — that are powerful, yet easy to use and implement within current processes. This is crucial for adoption. “We’ve got a few tools being tested now in mines, but before we can develop them further and make them more sophisticated, we need to get people using them,” Potvin said. “We’re not quite there yet.” Does that mean there’ll be a phase three of GSSO? E&MJ asked. “We will certainly try to get a phase three, yes,” Potvin replied. “We need to do some further work on the rock burst component and our work in shotcrete also deserves to be continued, as well as developing the probabilistic tools. It’s taken a long time to make a small but important step there.”

It is very important, because by investing the time and research now into bettering ground support systems, we are, quite literally, doing the groundwork for future operations. The challenges encountered are only going to increase as deeper operations become the norm and, as we get to a lower part of the metals cycle and prices dip, then the push for optimization in every area, including ground support will be felt.

‘Seeing’ Into the Rock Mass
As mines head deeper and deplete higher grade, easy-to-access resources, they will encounter extreme ground conditions more frequently. Although the modular nature of traditional ground support technologies like rock bolts, mesh and shotcrete allow flexibility, both in their application and through repairs, there is the question of cost and time involved. At some point, installing more mesh, stronger bolts or more layers of shotcrete will render operations unprofitable. “The issue is knowing where to put the effort in,” Potvin explained. “You can put a lot of ground support in some places, but you cannot put it everywhere in a mine. “At a new mine, if you take longer to develop the tunnels, then production could be delayed by, say, a year, which delays your income by a year and erodes your net present value. While at an existing mine, the cost of ground support failure comes if you have to stop production to rehabilitate tunnels, or worse, suffer casualties.”

As explained earlier, the preferable way to address this is through improving mining techniques to reduce the concentration of stress within the rock mass. However, this isn’t always possible, particularly at longstanding operations, and this is where monitoring systems are key. Potvin echoed Jakubec and Thomas’ thoughts on LiDAR technology; it is proving useful for monitoring rock-mass deformation, and regular surveys open the possibility for preventative maintenance in ground support systems. However, there is a caveat… “A lot of mines use LiDAR these days, but there are still gaps in our knowledge because it only measures deformation at the surface,” Potvin said. “You don’t really know what’s going on inside the rock, past the surface. “I think the potential, and where the research should be going, is in combining the LiDAR with cheap instrumentation on ground support system elements. For example, if we can put cheap sensors in the rock bolts, together with the LiDAR, we might get to a point where we’re really good at preventative maintenance in terms of our ground support. The work we’re doing with shotcrete will also assist.”

Shotcrete for Mines (Not Tunnels)
Many of the recommended guidelines that mines work with today when applying shotcrete, have their roots in early civil engineering guidelines. However, the two industries and environments are fundamentally different. For example, civil projects often use circular tunnel cross sections or horseshoe shapes which keep the concrete in compression, while mines often use square or rectangular cross sections with vertical walls which keep the shotcrete layers in tension. The former is a much stronger state, while the latter promotes bending and cracking.

“In civils, the shotcrete is often applied all around the tunnel and even on the floor,” said Potvin. “In mining, sometimes we stop in the middle of the wall. Sometimes there’s shotcrete at the bottom of the wall but the layers are thinner. There’s just not the same level of quality control, and the deformation mechanisms are very different. So why should we use the same approach?

“It’s worth seeing how we can optimize shotcrete for mining, what we can do to make it better and possibly less costly,” said Potvin. “During our investigations, we realized that we didn’t know enough to design a good test, so we did some modelling. That’s allowing us to investigate different thicknesses, tunnel shapes, different combinations with bolts etc. to try and get a better understanding of its application.

“We’re at the point now of summarizing the modelling exercises and making recommendations. We may need to do further modelling, but we’ll soon be able to design a test to examine the mechanisms and design optimal parameters for shotcrete in mines.”