Introduction

The main cause of failure is slope instability (Cenderelli 2000). The presence of water within and beneath the TSF reduces the stability of the face (the slope of the dam wall) by modifying the shear stresses on the potential failure surfaces. Once the slope of a tailings storage facility (TSF) has been established, either during the building of an early section or completion of a final lift, the only significant factors that can be controlled to reduce the risk of failure are the weight of water in the slope of the faces and the upward pressure of groundwater.

Water pressure is measured by recording pore pressures within the dam wall and beneath the TSF. This paper describes the distribution of pore pressures within a TSF face and gives a recommended methodology for measurement and control of pore pressures in the faces at risk of failure and the areas below and above the TSF that contribute to its water regime. Modern techniques for tailings dam monitoring emphasise the measurement of slope movement using radar, lidar, and prisms. However, these techniques only measure the reaction of slopes to instability factors and do not address the causes of failure or reduce the risk. Addressing the cause of failure is essential to managing the risk and preventing or reducing failure. Modern big data management techniques, artificial intelligence (AI), and robotics controlled via local or remote monitoring using satellite links can be used to increase control and reduce risk. Integration of geotechnical design, accurate monitoring, and engineering can substantially improve management and reduce risk. The efficiency of remote monitoring has increased during the Covid-19 reduction in human interaction; robotics and satellite controls offer the opportunity to continue this improvement. The recent publication of the global standards on tailings (Oberle 2020) describes 15 principles; accurate monitoring of causation plus transparency in reporting addresses Principles 2–10 and supports Principles 12 and 15. Understanding the opportunities to improve monitoring design and reporting objectives requires a basic understanding of the anatomy of a TSF. This paper describes typical construction, monitoring techniques, network design, and the opportunities for remote sharing of information to reduce risk.

Typical TSF Construction

The main types of TSFs are classified by either the method of construction or purpose: 1. upstream; 2. downstream; 3. centreline; and 4. water retention. Figure 1 shows the upstream, downstream, and centreline methods of construction. These use a combination of the tailings material and other materials for construction. Figure 2 shows the intermediate or final construction of a centreline TSF with a large pond.

Fig. 1
figure 1

Types of sequentially raised tailings dams or TSF (Vick 1983, 1990)

Full size image
Fig. 2
figure 2

Final construction of a tailings dam or TSF

Full size image

Most old tailings dams were constructed by piling material into depressions in the ground or riverbeds and therefore have weak foundations and no lining and are thus hydraulically connected with underlying aquifers. The TSFs located on riverbeds can have an underlying instability and it is very important that the water pressure beneath and upstream of the TSF be monitored, though each TSF is unique and has its own characteristics. Older TSFs may have had several phases of construction and therefore need a carefully designed monitoring network that measures each causative factor that increases risk of failure.

Few TSFs are established on a flat plain (as usually depicted in textbooks). Figure 3 shows examples of TSFs established on mountainous terrain.

Fig. 3
figure 3

Examples of TSFs established on mountainous terrain (Courtesy: A. Copeland)

Full size image

Although every TSF is different and requires its own monitoring design, they all hold water, in a pond and/or held in sediments; both require accurate monitoring to reduce the risk of failure. For accurate control, saturated sediments require measurement of the distribution and changes in pore pressure and effective stress.

Pore Pressures and Effective Stress

Pore pressure monitoring is essential because failure, known as volume deformation, in a slope will present (under various pressures and stresses) as three possible scenaria:

  1. 1.

    Compression of water in the pores of the material.

  2. 2.

    Compression of individual particles (sediments etc.)

  3. 3.

    Rearrangement of particles, usually to a more compact configuration.

As described in an earlier paper (Morton et al 2008), the stress state for any point in a slope is governed by the principal stresses and the acting water pressure (measured as pore pressure). Figure 4 shows the directions of stress acting on a hypothetical plane within a saturated medium.

Fig. 4
figure 4

Directions of stress acting on a hypothetical plane

Full size image

The primary effect of groundwater pressures is through effective stress. Rearrangement of particles (slope failure) is caused by changes in effective stress. The stress state at any given point and any time is governed by the principal stresses and acting pore pressure. The Mohr–Coulomb equation (Eq. 1), derived from the Coulomb-Terzaghi equation, states:

$$\tau =c+\left(\sigma -p\right)tan\varnothing$$
(1)

where: τ = shear strength; σ = total normal stress; p = pore water pressure; ∅ = internal angle of friction; and c = cohesion. The strength, and therefore stability of the slope, is created by the cohesion and weight of the materials making up the slope. Weakness comes from excessive pore water pressures and the internal angle of friction. Any reduction in the effective stress (σ′ = σ − p) will reduce the shear strength of the slope material. Therefore, in stability analysis, it is essential to know the distribution of pore pressures within the slopes of a TSF and beneath or upstream of a TSF. Once pore pressures are plotted, the pressure gradient within a homogenous slope will look like Fig. 5.

Fig. 5
figure 5

Anatomy of the pressure gradients within a slope of a tailings dam (TSF)

Full size image

This figure shows that the flow lines are orthogonal to the equipotentials plotted from the pressure readings. This flow line information can be used to model the stability of the slope and establish the location of the optimum phreatic surface for the dam wall. This information can then be used to improve safety and reduce risk. For non-homogenous materials, the pattern will be more complex; however, with accurately placed and well-distributed point piezometers, the equipotentials can be plotted.

Measurement of Pore Pressures

Pore pressures can be measured in three dimensions (four if you include time), using pressure transducers. These are often of the sealed vibrating wire point type and installed at specific points in a slope or dam wall. These point piezometers are grouted into boreholes drilled into critical sections of the slopes of the tailings dam retaining walls. The best placement of the point piezometers for critical areas of the TSF can be established using flow net plotting or numerical modelling based on what is known or assumed about the construction of the dam walls and underlying groundwater movement. Figure 6 shows a typical pressure transducer of the vibrating wire piezometer type.

Fig. 6
figure 6

Example of a vibrating wire piezometer prior to insertion

Full size image

Plotting three-dimensional distribution of pore pressures requires installation of nested piezometers, typically three piezometers sealed at designed intervals in each drill hole. The location and construction of the nested piezometers depends on the construction and dimensions of the slopes, but often three rows of three nested piezometers can be used in most TSF’s. Upstream and downstream installations can be used to monitor upstream and downstream pressure distributions.

The borehole diameter for piezometer installation depends on the type of piezometer installation being constructed. The drilled diameter needs to be large enough to accommodate the gravel pack, the seal, and the diameter of the tremie pipe used to deliver the sealant (grout) to the correct location during installation.

The head measurement represents the vertical hydraulic head at the measuring point. The cables transmitting the pressure readings are connected at the surface to a data logger or sim card so measurements can be sent in real time to a central data monitoring device (Fig. 7). This data can then be converted into information such as trend graphs and used to create knowledge for TSF management.

Fig. 7
figure 7

Wireless transmitter sim card installation

Full size image

Nested grouted-in piezometers can be used in vertical boreholes to monitor water pressures around and beneath the TSF. Figure 8 shows installations using either one borehole or three.

Fig. 8
figure 8

Nested grouted-in pressure transducers a in one borehole and b in a cluster of boreholes (Morton 2020)

Full size image

Piezometers can also be installed in angled holes, preferably sub-horizontal, when installed from the face of the monitored slope. Figure 9 shows the construction of an angled point piezometer installation in the face of a TSF.

Fig. 9
figure 9

Installation of an angled piezometer (Read and Stacey 2009)

Full size image

The same construction design is used for nested piezometers installed at multiple locations in a TSF face and below the foundations. Figure 10 illustrates a possible layout of multiple sub-horizontal nested piezometers in a TSF that was built in a steep valley.

Fig. 10
figure 10

Nested piezometers installed in a TSF face and foundation

Full size image

Design of a Monitoring Network

Monitoring network design depends on the age, shape, and construction of the TSF, as well as its location within the overall catchment that contributes to its water balance. A TSF cannot be monitored accurately without monitoring the effect of the catchment it occupies since the water balance of a TSF is affected by rainfall input to its pond and on surface runoff upstream. The groundwater levels around, below, and upstream affect the water balance of the TSF, and these too need to be monitored and managed to reduce upward pressure and increase safety.

Figure 11 shows the typical movement of groundwater flow below a TSF located in a valley. Even TSFs located on flat terrain can have upward pressure from groundwater, depending on the local geology and depth of weathered material below the TSF.

Fig. 11
figure 11

The schematic diagram for a tailings dam with safety monitoring

Full size image

Accurate monitoring depends on understanding all of the water pressures that act on a TSF. Using a conceptual model like Fig. 11 enables an initial assessment of the flow lines and pressures that should be measured to give a good indication of where to install point piezometers. More advanced analysis can be done using more detailed geological cross-sections, geological structures, numerical modelling, sensitivity analysis, and calibrated using existing groundwater level records from open boreholes. Figure 12 shows the layout of an accurate monitoring network around an individual tailings dam (in plan view).

Fig. 12
figure 12

Location of monitoring points for a typical valley TSF

Full size image

At this site, rainfall, evaporation, pond water levels, downstream pond water levels, overflow, point piezometers in the slopes, regional groundwater level, and movement are all monitored. All information is collected in real time and transmitted via satellite to multiple dashboards designed for multiple users. Cloud storage is possible, as is on-site storage or the use of smart phones for on-site audits. The details can be used by geotechnical engineers to support their designs and set safety targets. Engineers on site can use the water levels and pressure monitoring to control water levels and pressures.

The advantage of this system is that the design can be modified for each unique TSF. The monitoring results can be linked to interpret trends and used to warn of water pressure build-up, which can cause failure. Once a trend in water pressure build-up is noticed or a target level is breached, then mitigation measures using a TARP (trigger action response programme) can be immediately applied to prevent or delay failure. Actions include increasing pumping from boreholes or activating more drain holes. Barry (2020) states that “AI and robotics have the ability to both enhance resilience and manage and act upon data more efficiently than humans”. Many software companies are working on the platforms and linkages to ensure seamless connectivity. Sharing and visualisations of systems can enhance transparency and understanding of TSF management with all stakeholders.

The initiatives fit well with the new global guidelines on tailings standards published in August 2020 (Oberle 2020) and are a firm move to transparency in tailings management. The transparency will benefit all stakeholders and improve the insurability of TSFs.

A regional monitoring network is recommended for accurate prediction and management of a TSF’s water balance and movement. Figure 13 is the same area as Fig. 12 but expanded to include monitoring stations throughout the catchment. This is particularly important in semi-arid environments, which often have rainfall events high up in the catchment that then raise ground- and surface-water levels and create flooding lower down in the catchment, with adverse impacts on a TSF.

Fig. 13
figure 13

Example of layout for a TSF’s entire catchment

Full size image

Conclusions

All TSFs are unique and require a unique monitoring network that addresses the cause of failure and not just the the movement of slopes. The main cause of failure is an increase in water pressure in the slopes of the TSF or beneath the TSF. This can be measured by the judicious placement of point piezometers in critical areas of the slopes. Pore pressure data can be used to understand and predict instability, thereby enabling corrective action and risk reduction. This information can be incorporated in the geotechnical design packages to set maximum water level or pressure limits. Modern remote data sharing techniques can then be used to enable observation and visualisation in real time, thus encouraging transparency in tailings management.