NEWS

Article by Carly Leonida

In the wake of a major tailings dam failure and, given the blanket nature of rather scathing coverage from the mainstream media, it is all too easy for those operating in the industry to take a knee-jerk approach in reaction; for governments impose prescriptive regulation quickly (too quickly perhaps?), investigatory bodies are setup, miners throw monitoring technology at their dams, and fingers are pointed. A blame culture, while understandable, is unhelpful, and it is certainly not conducive to learning or change for the better.

Change in the way that tailings storage facilities (TSFs) are designed, built and managed has been incremental and relatively minor over the past 20-30 years. The same types of containment structures are used today as they were decades ago, and the same issues and problems arise now as they did then.

Despite the introduction of thickened discharge technology and filtration, conventional thickened-slurry tailings deposition, mainly to valley surface TSFs, continues to be the most common method of tailings management. It is therefore important to look at the fundamentals of design and management, and see how the industry could improve the outcome for these types of facilities going forward.

Are current practices good enough?
The current state of practice in tailings management continues to focus on a degree of thickening of the tailings in the processing plant (depending on the nature of the tailings, and their clay mineral content and type), allowing their transport as a slurry using robust and relatively inexpensive centrifugal pumps to a surface TSF.

Thickened or paste tailings that require pumping using expensive and input-sensitive positive displacement pumps are rarely produced for disposal to a surface TSF (although cemented paste tailings delivered under gravity is increasingly used as underground stope backfill).

Filtration of tailings is mainly restricted to regions with a severe water shortage, such as the Atacama Desert in Chile, and metalliferous tailings (comprising a ground rock flour, with negligible clay mineral content), to maximise the recovery of water in the plant.

There are some moves towards tailings filtration and/or co-disposal with coarse-grained wastes; for example, the GeoWaste system being trialled by Goldcorp, but these applications are currently few and far between.

At this point it is worth noting that tailings management is typically the responsibility of the process plant manager whose training would likely not include tailings management.

“The plant manager’s primary concern is to place the tailings as cost-effectively as possible, and they will therefore tend to favour management methods that are inexpensive upfront (pumping a thickened slurry) but which could lead to higher risks, higher rehabilitation costs, and limited post-closure land use and/or ecological function options in the long-term,” explains David Williams Professor of Geotechnical Engineering at The University of Queensland in Australia, and an internationally recognised expert in tailings management.

On deposition in the TSF, the tailings undergo beaching, hydraulic sorting according to particle size and specific gravity, as well as settling, self-weight consolidation and desiccation if exposed to the elements. These physical processes serve to dewater and densify the tailings, although much of the water is lost to entrainment within the tailings voids, evaporation and seepage, with perhaps half recycled to the plant.

Tailings dams are fundamentally different to water dams; rather than being constructed in one go, they are built in stages in order to delay expenditure, making them more complex in structure.

Like water dams, tailings dams make use of valleys; the deeper the valley, the greater the volume of storage provided and the greater the forces the dam must withstand. Generally, homogeneous starter dams are constructed first, and the dam is then raised by one of the following methods:

Downstream: using natural fill or mine waste materials constructed in the downstream direction. These water-retaining dams require an increase in the volume of fill as the dam is raised. This method is required in climates that are wet and/or cold, such as the wet tropics and some parts of Canada, or in areas of high seismicity.
Upstream: this involves constructing raises in the upstream direction on beached and desiccated tailings using fill and/or dried tailings. This method is suited to dry climates, such as southern Africa, much of Australia and the south-western US, which facilitate desiccation of the tailings beach. It also requires that the rate of rise to be limited (typically < 1-2m/year, but this can vary depending on the consolidation characteristics of the tailings and climate), and that the tailings deposition be cycled in thin lifts that are allowed to consolidate and desiccate in between. The upstream method can fail when it is applied in seismic regions without proper drainage, those with high rainfall during wet season (such as Brazil), and where the rate of rise is too high and drying is inadequate.
Centreline: a cross between the former methods in which the centreline of the dam rises vertically with fill placed downstream and on top of tailings. The centreline embankment can be designed to be stable independently of the tailings characteristics.
It is worth noting that different regions have different tailings containment practices, and these are highly dependent upon the personnel involved and their experiences.

Williams explains that there is a bias towards conventional thickened tailings slurry disposal as it is perceived to be the least costly approach. Bruce Brown, director of Bruce Brown Consulting, who has 44-years experience in the design, construction, operation and closure of tailings dams, in both consulting and the mining industry rightfully points out: “The question is, should these operations not be allowed to operate, or should the design and operation of conventional slurry deposition be improved to result in an acceptable risk threshold?”

Historically, the monitoring of significant tailings dams was limited to the use of piezometers that were read monthly and settlement plates that were surveyed every six months, with little or no monitoring of smaller dams.

However, as has become apparent through recent events, tailings dams can fail in as little as 20 seconds, making less frequent monitoring ineffective at detecting signs of instability and rendering emergency response plans inadequate.

Increasingly, tailings dams are being monitored in real-time, using piezometers and radar, and the advent of IoT-enabled devices and cloud computing are providing an effective way to gather and, most importantly, process and interpret the vast quantities of data generated by these systems; a trend that will continue to grow as miners improve their monitoring programmes.

Failure: what have we learned?
According to the 2008 paper ‘Reported tailings dam failures’ by Rico et al., published in the Journal of Hazardous Materials, the average tailings dam failure rate over the past 100 years is 1.2%, or 2.2 failures per year.

“This has not improved over time, although tailings dams have increased in size,” says Williams. “The rate of tailings dam failures is greater than two orders of magnitude higher than that for water retention dams of 0.01%, which most consider is unacceptable.”

Rico et al. (2008) state that the tailings dams that have failed have been of moderate height – in the region of 5-20 m, which represent the majority of tailings dams and which are typically subject to less engineering and review – and the failure has been attributed to either (rather disturbingly) ‘unknown causes’, or water-related issues such as excessive surface water that leads to over-topping and erosion of the dam, or too high a phreatic surface within the dam, leading to geotechnical instability.

Rico et al (2008) also state that, over the past 100 years:

The US has reported the highest number of tailings dam failures of any country, with 39%;
18% were in Europe;
12% were in Chile; and
5% were in the Philippines.
It is also noting that many more events have occurred, for example in countries such as China, but have gone unreported.

So what, if anything, has the industry learnt from these types of events?

Firstly, that tailings dams fail, often without warning, quickly and catastrophically, and the failure is often not observed (sadly, those in a position to be able to observe them rarely survive). Dams are typically not equipped with real-time instrumentation, which results in no data being recovered in the period before, during or after the failure. Most involve the flow of tailings downstream, leaving little tailings in a state that could be tested to understand better what happened, hence forensic investigation of the mechanisms of failure are difficult to carry out.

Responses to tailings dam failures have included numerous reviews of existing dams by mining companies and regulators, more conservative designs for dams, using lowest bound design parameters, and changes to regulations. For example, upstream construction was outlawed in Chile following earthquake-induced failures in 1965, requiring flattening and compaction of cyclone sand dams and the application of impermeable geomembrane liners on the upstream face; with the result that Chile has not had a recent catastrophic failure of one of these sand dams.

What has become obvious is that the dominant type of tailings dam used in different geographical regions may continue to be used unless outlawed. That said, outlawing a particular method of tailings dam construction or type of tailings dam does not necessarily improve dam safety; operating and raising tailings dams located above towns in Brazil was outlawed following the failure at Samarco in 2016. However, the legislation failed to specify how decommissioned and abandoned facilities should be dealt with; an oversight that might have contributed to the Feijão dam collapse in Brumadinho in January this year.

“Failure of two similar tailings dams in Brazil raises the question as to whether the design concept is fundamentally flawed,” notes Brown.

Terry Eldridge is a senior mine waste consultant with Golder Associates has nearly 40 years of international experience in design, construction and operation of tailings storage facilities including post-failure dam design work at Mount Polley and at Los Frailes in 1998. He adds: “Considering the recent failure in Brazil, I find it hard to argue against legislating that upstream construction of tailings facilities is banned in Brazil. That the industry knows how to design and operate these types of dams safely in other jurisdictions does not balance the performance record of these types of dams in Brazil.”

Williams raises the issue of media coverage. “What has changed in recent times is exposure by the media and via the internet to tailings dam failures, particularly those that occur in developed countries and those that involve major mining companies,” he says.

“Recent high-profile dam failures, particularly those that cause fatalities, have made such failures perhaps the key threat to the mining industry’s financial and social licences to operate. They pose a threat of loss of control by the industry.”

Concern about personal liability/prosecution following a tailings dam failure is also high, and this extends to increased liability for the ‘engineer of record’ or design engineer, although Brown points out that in order for the engineer to be held accountable, they must be given the authority and resources to ensure that the facilities are constructed and operated to the intent of the design.

“It should also be said that there is a critical and increasing shortage of experienced tailings engineers,” he adds; a situation that is likely to worsen in coming years as experienced engineers retire, and one that isn’t helped by the increased level of scrutiny the role carries.

Eldridge hits the nail on the head: “Collective learning is needed, but we need to ask what the individual designers, mine managers, site-based tailings engineers and tailings managers have learned, as well as what the individual regulators and government inspectors have learned from each of these events.”

He continues: “The industry has not passed along lessons from past failures to the next generation of engineers. There seems to be a cycle in which a cluster of failures results in investigation and identification of causes, and publication of design and operation guidelines that can be used to avoid the conditions leading to the type of failure.

“If the type of failure is not seen for some time, the lessons are not passed on to the next generation of engineers. The issue or mistake is then made again and the cycle repeats. As an example, drainage guidelines for upstream constructed tailings dams were formalised in ICOLD Bulletin 97 ‘Tailings Dam – Design of Drainage’ published in 1994 but few younger engineers are aware of this work.”

Like many others, Brown believes there is too much focus on the technical aspects of failures.

“The technical reason for the failure becomes the main focus, whereas the real root cause of the failure in every case is the system’s failure of governance,” he explains. “Poor design, construction, operations and monitoring only occur if poor governance allows it to happen.”

Implementing advances
The 2015 report on the Mount Polley dam failure, which took place in British Columbia, Canada, in 2014, concluded that incremental improvements in TSF methodology and practice would be insufficient going forward. The report recommended consideration of a new paradigm of tailings management for all new facilities which it termed ‘best available technology’ (BAT).

The fundamental principles of BAT are:

1. Eliminate surface water from the impoundment;

2. Promote unsaturated conditions in the tailings with drainage provisions; and

3. Achieve dilatant conditions throughout the tailings deposit by compaction.

The report went on to add, that “surface storage using filtration technology is a prime candidate for BAT”.

Keith Seddon, senior principal of ATC Williams, with a background of 45 years in tailings facility design, explains the flaw in this suggestion: “Many people have taken this as indicating that filtered tailings are necessarily the next step, and it is a fact that this technology is now being explicitly included in many options studies,” he says. “However, it is generally the case that this technology is prohibitively expensive, and it would be fair to say that there has not been a rush to adopt it. It is also the case that filtered tailings do not automatically produce a TSF that complies with the three BAT principles, which is also not well appreciated.”

Eldridge adds: “I did my first ‘dry stack’ design in the 1990s, so the concept that is being so actively pushed these days is not new. Filtration technology has advanced so that the concept can be applied at more sites. This will be an advance for the industry provided that proper assessment of the foundation conditions is carried out; there have been many failures of analogues of the ‘dry stack’.”

Seddon says that another point that is often overlooked is that the fundamental principles of BAT can be achieved by other means, at far lower cost; the methodology of thickened discharge (central thickened discharge, or down valley thickened discharge) is well established, and provides the same outcomes. However, it works best in flat terrain, and climates with high levels of evaporation (as in most of Australia), and is difficult to apply in places like British Columbia (Mt Polley), and Minas Gerais (Samarco and Brumadinho).

A third option also exists. Seddon explains: “The current practice in Chile of compacting cyclone sand tailings to raise the dam in a downstream or centreline direction, as well as placing an impermeable membrane on the upstream face goes close to complying with BAT principles. At present it typically does not address principle one (elimination of water), but the system could be modified to achieve this.”

He adds that not all tailings types are suitable for dam construction by this method though.

“It is worth noting that because of the acid-generating potential of certain tailings, the practice of ‘wet closure’ (storing water over the top of the completed facility, in perpetuity) is common in Canada,” says Seddon.

“It is generally the case that additional processing of tailings at the production stage can remove most of the acid-generating component (for high-level disposal separately), which removes the need for wet closure. It is likely that increased environmental scrutiny will require this stage to increasingly be applied as part of routine ore processing.”

Eldridge believes that it is the inconsistent application of best practices that is the issue, rather than inadequate practices per se.

“The state of practice varies depending on the mining company, the consulting company doing the design work, the interaction between the individuals involved in the mining company and the design company, and the country in which the mine is located,” he explains.

“I see projects that are at the leading edge of the practice of engineering and that are advancing the state of practice. These projects are using design companies that have rigorous internal review processes and are owned by mining companies that implement rigorous independent review, either as one-off independent peer review or ongoing review by a board or panel.

“I also see projects that fall far short of what would be considered acceptable practice. These are typically carried out by educated but inexperienced engineers working for smaller mining or exploration companies with constrained budgets. That said, during the ‘boom times’ I saw many projects that were poorly executed by the large design companies working for the major mining companies. The issue was inexperienced engineers working without adequate review.”

Brown concurs: “I couldn’t agree more. I carry out numerous reviews and the range of competencies and engineering practice is huge.” He adds that a major issue at many sites is continuity of site staff who manage TSFs over their lifetime, noting that “staff turnover is usually high”.

If one sure thing has come from recent high-profile dam failures, it is the acknowledgment that better monitoring systems need to be installed at the majority of TSFs. IoT-enabled instrumentation is no longer expensive, and there really is no excuse for mines not to use real-time monitoring systems.

Technology has also benefitted the design of TSFs in recent years.

“What I see as a major advance is our ability to numerically model the structures we are designing,” says Eldridge. “The combination of economically available hardware and the advanced constitutive relations now coming to the forefront of design means that an engineer can assess the likely performance of a containment structure for a wide variety of load conditions, tailings parameters and foundation conditions.

“This allows the vulnerabilities of a particular type of design to be more clearly defined and communicated to all those involved in a tailings project. We are at the point where we could change to a performance-based design approach such as that set out by Dr Morgenstern in his 2018 Victor De Mello lecture ‘Geotechnical risk, regulation and public policy’.”

Perimeter spigots to a central decant in arid estern ustralia maintaining a very small decant pond Perimeter spigots to a central decant in arid Western Australia, maintaining a very small decant pond

Regulation, tech & education
Brown says that one positive change he has seen in recent years relates to miners paying greater attention to increased governance.

“Many mining houses have significantly increased their attention to governance and have implemented many of the recommendations of the ICMM and/or have adopted the MAC guidelines,” he says. “There is a significant increase in the awareness of the risks associated with tailings management at the senior management levels.”

Given the fallout from the recent dam collapse in Brumadinho and Brazil’s move to ban upstream TSFs altogether, it is likely that increased regulation will be imposed on miners globally.

Brown shares his concerns; concerns that were echoed by every consultant interviewed for this article. “Government regulation should focus on ensuring that mining companies have adopted and implemented good governance practices,” he says. “The dangers of technical prescriptive regulation are that it encourages practitioners to rely on the prescriptive regulations and use them as justification for aspects of the design without critical thought. I see this constantly with the ANCOLD (Australian National Committee for Large Dams) guidelines. Prescriptive regulations can also stifle innovation.”

The reality is that government regulation has not reduced the rate or severity of tailings dam failures worldwide thus far and, given that the scale of tailings dams has increased over the past 20-30 years, and that this trend will continue as concentrator throughput grows, other avenues for change need to be explored.

Williams agrees: “Government regulators should not regulate or prescribe outside their expertise, or too broadly, and not based on community perceptions and (understandable) emotional outrage.

“Government regulators should collaborate with companies, consultants, contractors, universities and start-ups to ensure the appropriate level of expertise and experience among those responsible for tailings management; educate and train tailings practitioners (designers, constructors and operators); develop and apply improved, real-time monitoring and responses; and, advance alternatives to thickened tailings slurry deposition.”

One area in which this approach has proven successful is in the development and installation of smart tailings-dam monitoring systems. GroundProbe was a start-up out of The University of Queensland in Australia that developed and sold radar systems for monitoring movements of open-pit walls. The technology has more recently been applied to monitoring the movement of tailings dams; the company was purchased by Orica in 2017 and has since been helping BHP and Vale to monitor the Samarco site as part of the clean-up operation following the dam collapse in 2016.

Reutech of South Africa and IDS of Europe have joined GroundProbe in offering radar systems, and tailings deposition models such as Muck3D by MineBridge, and water balance models such as GoldSim have also proved their worth in tailings applications. These are just a few examples of developments in the technology space – search the web and you will find many more.

Sophisticated software is all very well, but it is no good if engineers do not have the expertise and experience to interpret and test the models generated; better technology and education must go hand in hand.

Brown cautions: “A significant risk is that the sophistication of the software – FLAC, Plaxis etc. – can outstrip the practitioner’s ability to understand how it works, what parameters are required and whether the parameters are valid. There is a tendency to ‘believe’ the output from these programs without critical review.”

Eldridge reports similar findings: “We now have analytical tools that allow a design engineer to assess performance of a dam under a variety of loading conditions, pore pressure conditions and foundation conditions. We also have investigation and laboratory techniques that provide the high-quality input data for these assessments,” he says.

“When these tools or models were first developed, the physical basis of the model and methods of validation and calibration of the model were widely discussed and publicised so that the few practitioners using and advancing the models understood the limitations of the models.

“As these tools become prevalent and routine in the industry, the focus has changed to how to use the model or software. The physical basis and limitations of the model are demoted to appendices, and the ability for most engineers to properly question and really understand the limitations of the output has been lost or not developed.

“Young analysts can easily run the software but cannot interpret the results into the context of engineering design which would occur during a rigorous engineering review process. They do not understand the need to identify the vulnerabilities and sensitivities of a design.”

Eldridge adds that, on the surveillance side, technology is providing instantaneous performance data and the quantity of this data is forcing the use of intelligent systems to interpret performance. This is a large step ahead, but if engineers are not actively involved in surveillance and interpretation activities, performance issues or failure modes that were not considered when the algorithms were developed will be missed and failures will result.

“I see technology as reducing the time I spend compiling information and increasing the time I have to interpret dam performance based on the data,” he says.

Better education is at least part of the answer. And not just for engineering students; going forward, investment in appropriate education and training needs to be directed to all those responsible for tailings management, from designers, through contractors to operators. And all mining company employees, from the CEO and board down to those operating tailings-disposal systems, need to be instructed about the importance of sound and safe tailings management.

Knowledge is, after all, power.

Source: Mining Magazine