Improving Tailings Dam Safety via Soil Treatment

Abstract

Mine tailings are the byproduct of mining activities, which need to be disposed of once the minerals in the ore are extracted. They can be disposed of in either dry or wet forms. The latter is most common, with the tailings being disposed of in the form of slurry inside retention structures. The retention structure may be a natural or manmade dam, with a predominant use of the upstream method due to its cost-effectiveness. This study analyzes the stability of an upstream tailings dam considering its staged construction. A two-dimensional nonlinear finite element model was developed using the program Plaxis 2-D to investigate the potential for stabilizing the tailings dam by using emulsified polymer and a mixture composed of cement kiln dust (CKD) and re-cycled gypsum (B). The numerical model demonstrated that utilizing a CKD: B mix increased the overall stability of the tailings impoundment above the conventional 1.5 safety factor requirements and indicated its usefulness in constructing robust dams whilst still being environmentally friendly.

1. Introduction

Tailings are a by-product of mining activities. They result from the crushing and grinding of rocks to extract the “ore”. Typically, the ore percentages within the raw material are between 30 to 0.4%, which means massive amounts of mined materials are disposed of daily in tailings storage facilities (TSF) [1]. A single mine may produce tailings in the order of 1 to 6 million cubic meters per year. There are various types of TSFs, and they fall into two major categories, (1) manmade and (2) natural. The manmade are further divided into three major sub-divisions, upstream, downstream, and centerline methods.

Comprehensive risk assessment studies should be performed to evaluate the environmental and geohazard risk of these TSFs and help the decision making to mitigate the risk tailings pose to the surrounding environments and their inhabitants and to promote a sustainable development of mining [2,3,4,5,6]. One of the associated risks is slope instability for the TSF which, if it occurs, can unleash massive quantities of the stored tailings onto the nearby dwellings and ecosystems. These situations can be extremely problematic, and their consequences can be everlasting. An example of such failure is the Bent-Rodrigues mine tailings failure that took place in 2015. In that incident, an upstream tailings dam for iron ore failed, killing dozens of people and releasing millions of toxic materials into surrounding rivers, causing one of the biggest litigations in history.

When it comes to the three major construction methods, the upstream construction method is the dominant construction method employed in practice because it is the most efficient way to dispose of large amounts of tailings [7,8]. In this method, a dyke is initially formed to contain early operations wastes, which are then heightened progressively “upstream” towards the TSF center. A major disadvantage of this method is its negative records during earthquakes; hence, it was abandoned in some countries (e.g., Chile) due to this risk [9,10]. In such circumstances, mining companies are left to adopt other methods of construction, which makes the construction of containments much more expensive and unnecessarily time-consuming. The pros of the downstream and centerline methods are that they have the lowest risks of failure [11,12]. Other risks the TSFs pose to the environment include the leaching of toxic materials into groundwater, overtopping, and acid mine drainage (ACMD). Despite these issues, mine operators are inclined towards the upstream construction technique due to its cost-effectiveness and simplicity. The design process for the upstream method for a tailings storage facility is usually carried out using finite element numerical models. To evaluate the slope stability of such tailings dams, two main approaches are typically used. The first method is the limit equilibrium (LE) approach, which was presented by Petterson (1955) for the analysis of the possible failure mechanisms of Stigberg Quay in Gothenburg, Sweden [13,14]. The main assumption of the earliest LE method developed, the Swedish circle, is that the slip surface is circular, and the soil shear strength is given by τ = su, i.e., neglecting the frictional resistance. This approach has been further enhanced and extended since then by many researchers (e.g., Janbu; 1954; Bishop, 1955; Morgenstern-Price, 1965; Morgenstern, 1967; and Fredlund-Khran, 1970) [13,15]. LE analysis can generally yield satisfactory results for the overall safety factor (S.F), but it treats the slip mass as a rigid body and hence the strain field is not established. The second method for conducting slope stability analysis is through a proper FE analysis, considering the gravity in situ stresses and utilizing a suitable constitutive soil model that captures the behavior of soil under the complex loading conditions typically found in these TSFs. In these types of analysis, stress and strain fields are established, especially plastic strain. This information can help identify the failure mechanism and track progressive failure. In addition, overstressed regions can be assigned lower shear strength parameters, i.e., residual strength parameters, to account for the strain-softening behavior exhibited by some clayey soils. In Plaxis 2D, there are two approaches to calculating the safety factor in F.E-based stability analysis: in situ stresses-based S.F and strength reduction technique-based SF. Strength reduction technique, such as Plaxis 2D Φ′-C technique, resembles the limit equilibrium method in that the strain after “failure” is not realistic. However, the F.E.-computed strains without strength reduction can be accurate and useful. For example, Ormann et al. (2013) investigated the stability of a tailings dam considering the proposed construction/raising sequence. They studied the effect of the staged construction on the safety factor (S.F) using PLAXIS 2D. They concluded that the safety factor is affected by the cycles of loading/excess pore water pressure generation and the associated consolidation and proposed a new method of stabilization using rock fills. Several sequences of rock filling were investigated, and the results indicated a good performance when the rock filling was utilized. Ozcan et al. (2012) studied the downstream slope stability of a copper tailings dam to assess the proposed capacity increase. They utilized the program Slide 2D, which uses the limit equilibrium method for stability analysis. Yin et al. (2011) investigated the influence of staged construction of the stability of a tailings dam employing a laboratory physical model [16,17].

Motivations and Novelty of Work

This study aims to assess the effect of soil treatment on the stability of tailings dams in terms of the overall safety factor, and aid mine operators to design dams, using the upstream construction technique, that meets the safety requirements of the code enforced, and to protect the people and environment surrounding mining activities.

Various schemes of ground improvement techniques are used to enhance soil characteristics such as water retention, stiffness properties, and soil strength. One method of which is investigated to improve the stability of an upstream tailings dam. For instance, Nikbakhtan et al. (2007) presented a case study of the efficacy of jet grouting in increasing slope stability [18,19]. Their parametric study was done using the finite element program CLARA-W. They compared the factor of safety in the staged construction of the Shahriar Dam, Iran using 2D and 3D analyses. Their results proved the effectiveness of soil treatment. Shu-wei et al. (2013) and Poulos (1995) examined the usefulness of micro-piles and piles in increasing slopes’ stability. Their results indicated that piles can be an effective stabilizing measure to counteract the sliding forces [19,20,21]. Some have used geotextiles and geogrids to improve slopes, although not always successfully [22,23]. Other researchers have used soil mixing techniques for a variety of geotechnical engineering applications to strengthen the soil matrix [18,24,25,26,27,28,29,30,31]. For the mine tailings dams built using the wet method, consolidated tailings, or composite tailings (CT) and rock fills have been used to increase the whole dam’s stability. However, no attempt has been reported so far, to the author’s knowledge, to “mix” the wet tailings with additives before disposal. In the current research, traditional additive admixture of recycled Gypsum and Cement Kiln Dust (CKD) compelling performance with different soil types stimulated their selection to stabilize mine tailings impoundments built using the upstream method of construction [1,2,32].

The objective of this study is to evaluate the effectiveness of soil treatment for stabilizing mine tailings dams. To achieve this objective, a two-dimensional nonlinear finite element model was developed using the software PLAXIS V8. The developed model was verified by employing a case study of an upstream tailings dam. The verified numerical model was then utilized to analyze the stability of the tailings dam constructed using the upstream method through 11 construction stages. The performance of a tailings dam that is constructed using treated tailings with 7.5% admixture of (1 CKD:1B:1OC), where OC is ordinary cement, is analyzed and compared with a raw case. The effects of tailings treatment on the stability of the dam and its settlement were evaluated.

2. Materials and Methods

Oven-dried tailings sample was sieved through #40 U.S sieve, crushed gently with a pedestal for 10 min. The specimen was then tested to determine its liquid limit (LL) and plastic limit (PL) and plasticity index (PI) using Casagrande’s apparatus in compliance with ASTM standard (ASTM D4318). The soil was non-plastic with LL = 48%. The results are also listed in

Table 1. General physical properties of tailings.

Index Parameters	Value
USCS of the tailings	ML
Mean particle size, D 50 (mm)	0.045
Effective particle size, D 10 (mm)	0.0175
Coefficient of uniformity, Cu	3.4
Coefficient of curvature, Cc	0.8
Liquid limit, LL (%)	48
Plastic limit, PL (%)	NP
Liquidity index, LI	1.34
Specific gravity, Gs	2.69

. The soil is classified as low plasticity silt, ML according to the ASTM D2487 standard.

The tailings were then treated by mixing with CKD and OC with recycled and pure gypsum at varying contents as well as emulsified polymer [1]. The results of the treated tailings with CKD/OC at 7.5% are selected in [7], the strength properties of treated and untreated tailings are provided in

Table 2. Tailings geotechnical properties.

Material	C’	Φ′	Eoed.	k (cm/s)
Untreated tailings	0	38	9400	9.1 × 10−8
Treated tailings (1:1:1)	184	38	300,284	6.01 × 10−9

. These properties can be used in developing proper treatment schemes and numerical models that can aid in assessing the stability of tailings dams treated with the proposed admixture combination.

3. Methodology

To evaluate the effectiveness of the proposed solution to enhance mine tailings stability, the results of the soil strengthening obtained from the experimental program reported in [1,2,32] were used in simulating the treated tailings as construction materials for an upstream tailings dam using finite element (F.E). The numerical investigation was conducted using the F.E program PLAXIS 2D, which allows analysis of staged construction and therefore enables accounting for excess pore pressure (EPP) development and subsequent dissipation and consolidation behavior of the staged construction used in the upstream tailings dam [28,33,34]. The numerical model was constructed following the procedure of staged construction, and the stability of the tailings dam was evaluated at each stage of construction. The numerical model was verified through the analysis of a well-documented case study of the Aitik upstream mine tailings dam in Sweden whose geometry and soil properties have been reported by many researchers [35,36,37,38]. This case study in particular was selected due to its complexity and authenticity.

4. Finite Element Modelling

For problems involving slope stability, PLAXIS 2D allows different computation types (i.e., plastic, consolidation, safety, and dynamic analyses) and incorporates different constitutive models that can describe different soil behavior. The used constitutive model for tailings is a simple elastic-perfectly plastic where the soil shear strength is represented by the Mohr–Coulomb failure criterion. In addition to the material’s unit weight, this model requires only five parameters, namely friction angle (ϕ′), cohesion (c′), angle of dilation (ψ′), Young’s modulus (E), and Poisson’s ratio (ν). Duncan (1996) conducted an extensive review of the different constitutive models employed for the analysis of earth dams. He concluded that the elastoplastic and elasto-viscoplastic stress–strain relationships more realistically model the behavior of soils close to failure, at failure, and after failure [39,40]. Similar investigations were conducted more recently to examine different available constitutive models [37,41,42,43]. Although the MC model can overestimate the shear strength for NC clays in undrained conditions, this study focuses on the comparative side between the improved and non-improved soils’ shear strengths. Moreover, it is expected the improved soil behavior will yield softening behavior of the improved soil due to the noticed overconsolidation ratio (OC) in Alsharedah and El Naggar’s (2023) paper [1]. In addition, most studies demonstrated that the Mohr–Coulomb (MC) criterion is suitable for simulating the behavior of soil in slope and dam stability problems. Therefore, it was used in the current study to simulate the behavior of both treated and untreated tailings. The tailings are considered to be a silty material and are assumed to be non-dilative due to their loose disposal method. The strength and deformation parameters of the treated and untreated tailings considered in the present study were established from direct shear (DS) and Oedometer tests. For all soil layers used in the numerical, ν and ψ′ were assumed to be 0.33 and 0°, respectively. In addition, the hydraulic conductivity for the tailings was established from the results of the Oedometer tests. The hydraulic conductivity was estimated based on the following equation [3]:

K = c_v m_v ɣ_w

(1)

where c_v is the soil coefficient of consolidation, mv is the inverse of the elastic Young’s modulus at specific stress range (stress dependent), and γ_w is water unit weight. Since the Oedometer is only vertically draining, it was assumed that hydraulic conductivity in both vertical and horizontal directions, ky and kx, are equal.

The model analyzed the construction of the dam, which involved 11 construction stages (rises) executed over the course of 11 years. A two-dimensional plane strain model of the tailings dam was formulated using the required information reported in the literature on its geometry and soil properties. The model was then used to explore the effect of soil improvement on the dam behavior through the different construction stages.

Validation

Aitik is an open pit copper mine located close to Gällivare in northern Sweden. The mining activities at Aitik started in 1968 with the 2011 production of this mine reaching 31.5 million tons of ore. The tailings dam is 450 m wide and 70 m in height. The tailings are pumped to the tailings disposal area and discharged by spigots from the dam crest. Figure 1 shows the impoundment which covers a 13 km² area and is limited by the topography and four dams: A-B, C-D, E-F (including E-F2 extension), and G-H.

The settling pond is situated downstream of dam E-F. A view of dam E-F is depicted in Figure 2. The section E-F of the tailings dam failed in year 2000 over a length of 120 m. Consequently, 2.5 million cubic meters of water leaked to the settling pond. Therefore, the leaked water filled up the settling pond, whereby the water flowed to nearby rivers. No definite conclusion concerning the cause of the failure was reached [35,36,38,44]. The finite element analysis in this case study has been performed on section E-F of the dam, shown in Figure 2.

The failure of section E-F had serious consequences for humans and the surrounding environment. In addition, its failure may lead to a failure of dam I-J, which is located downstream of dam E-F and the settling pond (Figure 1). Section E-F was constructed using the upstream method. The different material zones are presented in Figure 3. Material zones 2, 3, 5, 6, 7, and 8 (Figure 3) represent the coarse tailings particles. These material zones can be classified as silty sands according to the unified soil classification system. A field testing campaign was launched to retrieve samples for lab testing to determine the geotechnical parameters that were required for the analysis of the dam stability. The numerical model was established using Plaxis 2D employing 15 nodded triangular elements, which give a fourth order interpolation on displacement [33,34]. The elastic-perfectly plastic model and MC failure criterion were used to assess the variation of dam stability, should the proposed plan of construction be followed. Standard fixities in PLAXIS 2D were utilized, which are fixed base to constrain horizontal and vertical movements and constrained horizontal displacements at the left vertical boundary. Water was allowed to seep in all direction except the left and bottom boundaries [35,36,38]. Each layer was constructed in 10 days and was allowed to consolidate for 355 days. This short period of construction generates excess pore pressures that dissipate with time. Not all the EPP is expected to have dissipated by the end of construction due to the existing hydraulic boundaries and low hydraulic conductivities of the materials. Therefore, the stability of the dam is expected to decrease during the construction period and increase afterwards, since the effective stresses increase at the same rate at which the EPP decays.

Duncan (1996) suggested that the short-term condition is most critical for embankments under multistage loading conditions because this loading leads to increases in pore pressures, which dissipate over time, and consequently, the effective stresses and soil strength. Therefore, consolidation analyses along with safety analyses were conducted in order to establish the safety factors fluctuation of the Aitik tailings dam with loading/consolidation cycles of 11 years in length. To enhance the stability of the dam, Ormann et al. (2013) investigated the use of rock fills to stabilize the tailings dam. They concluded that the rock fills option utilized was successful in enhancing the dam stability. Figure 3 shows the details and dimensions of the model developed to simulate the Aitik dam. Prior to using the numerical model for evaluating the effectiveness of the tailings treatment technique investigated and reported in a companion paper (Alsharedah et al., 2016) for enhancing the dam stability, the numerical model was first verified against the published results of the Aitik dam [2,35]. For this purpose, a simplified version of Ormann et al.’s (2013) model geometry was considered in the analysis. The loading, consolidation, and safety analyses were carried out similar to the original model of Ormann et al. (2013) [35,36].

In the current study, it is proposed to use improved tailings in lieu of rock fills to enhance the overall dam stability. The properties of the treated and untreated tailings determined from the experimental study reported in a previous paper (Alsharedah et al., 2016) are used in the numerical model [2]. The program PLAXIS 2D was utilized for the numerical modeling, and the θ′-c′ reduction technique was employed to estimate the safety factor. Both the internal angle of friction and the cohesion were reduced at the same proportion to weaken the soil and bring it into a non-equilibrium state. This method is termed strength reduction technique [2,45]. The safety factor in this comes is defined as follows:

$$\mathrm{S}.\mathrm{F}=\frac{c^{\prime }actual }{c^{\prime }residual }=\frac{{\theta }^{\prime }actual }{{\theta }^{\prime }residual }$$

(2)

where c′ is the soil cohesion intercept (drained) and ${\theta }^{\prime }$ is the soil friction angle (drained). Figure 4 shows the case I in Ormann et al.’s (2013) model, in which only the existing rock fills were activated. Ormann et al. (2013) conducted an analysis of the model where the 11 lifts shown in the top left corner of Figure 4 were successively added over the course of 11 years [35,36]. Each layer was constructed in a period of 10 days followed by a period of 355 days for consolidation. Then, the following layer is constructed over 10 days followed by 355 days of consolidation, etc. Afterwards, the safety factors analyses were established for each layer twice, one once the layer was constructed and the other after the end of the consolidation period. This process was repeated for different configurations, and the best results are shown in Figure 5. Since the objective of this study is to show how the safety factor is increased for improved tailings over un-improved tailings, three cases from those shown in Figure 5 were selected for verification purpose (Cases I, V, and IX).

Table 3. Materials properties of the replicated model (Ormann et al., 2013).

Material Type	γunsat. (kN/m3)	γsat. (kN/m3)	kx (m/s)	ky (m/s)	E (kN/m2)	c′, (kN/m2)	Φ
Moraine	20	22	9.9 × 10−8	4.5 × 10−8	20,000	1	35
Sand tailings soft at bottom	18	18	9.9 × 10−8	1 × 10−8	9800	6	18
Layer sand tailings	17	18.5	5.5 × 10−7	5.5 × 10−7	9312	9.5	22
Moraine (dikes)	20	22	4.9 × 10−8	1 × 10−8	20,000	1	37
Compacted sand tailings	16	19	1 × 10−6	9.9 × 10−8	8790	13	26
Sand tailings soft at top	18	18	9.9 × 10−8	1 × 10−8	3048	6	18
Compacted sand tailings (dikes)	16	19	1 × 10−6	9 × 10−6	7200	13	26
Sand tailings (top)	17	18.5	5.7 × 10−7	5.5 × 10−8	3895	9.5	22
Filter	18	20	1 × 10−3	1 × 10−3	20,000	1	32
Rockfill	18	20	1 × 10−1	1 × 10−1	40,000	1	42

presents the rest of the materials properties used in the validation of Ormann et al. (2013). Figure 6 shows the developed model. The complex geometry features and layers whose presence would not affect the accuracy were excluded. The dam body and its foundation were modeled, employing 15-noded triangular elements. Mesh refinement at highly stressed/strained zones was necessary to ensure the results’ accuracy. Accordingly, a series of models were developed where the mesh was incrementally refined, and the results were then compared. When the difference between the results of two consecutive models (i.e., refinements) became less than 2.5%, the most refined of the models was considered. The used material properties for the model verification are shown in Table 1. In the study of Ormann et al. (2013), the excess pore pressures immediately after the 2nd and 11th layers were constructed and after they were allowed to consolidate were given and their values matched the results presented here [36].

5. Results

Figure 7, Figure 8, Figure 9 and Figure 10 depict the EPPs for those raises from the simplified model; an excellent match between the results of the two models (Ormann et al., 2013 and the simplified model used herein) were found [35]. It can be noted that there is a concentration of excess pore pressure in the bottom left corner. This is attributed to the presence of the impervious base and closed water boundary on the vertical left boundary. The EPP of the simplified model after the second layer placement was 59 kPa, which compares favourably with the EPP of 56 kPa predicted by Ormann et al. (2013) [35]. The same observation can be made for the EPP after consolidation occurs for the 2nd layer, in which case the difference between the two predictions was only 1kPa. For the 11th raise, the EPP values were 59–110 kPa and 95–50 kPa right after the construction of the layer and after the consolidation, for the simplified model and for Ormann et al.’s (2013) model, respectively [35]. Figure 11 displays the variation of safety factor with time for both the simplified model and the model by Ormann et al. (2013) [35]. The solid squares–dotted lines in Figure 11 pertain to the results obtained by Ormann et al. (2013) whereas the hollow triangles–dotted lines are those obtained from the model developed herein. As can be seen, good agreement was established between the two models for the three selected cases (Cases I, V, and IX; Figure 6), knowing that there was difference in geometry interpretation and piezometric line location. The lower results for the first five years were because a large part of the slip surfaces developed within the sixth layer that happened to be of lower shear strength values than that of the layer it replaced. The model was also verified via hand calculations where hydrostatic pore pressures and the total and effective stresses estimated at the bottom left corner were compared to the results obtained from the F.E.A.

6. Discussion

Effect of Tailings Treatment on Safety Factor

Figure 12 demonstrates clearly that the stability of the dam for Case I represented the worst-case scenario among the cases considered. The calculated average S.F for Case I, untreated, is 1.25, which is less than the required long-term safety factor of 1.5 [3,39]. Therefore, a soil treatment scheme is considered to enhance the stability of the dam. The analysis was repeated for the dam with identical geometry, but the tailings layers were assigned the properties of the treated tailings. The results obtained from this analysis are also presented in Figure 12. The calculated S.F for this case increased substantially to an average value of 1.55. This significant increase in S.F demonstrates the effectiveness of the soil treatment method proposed and its suitability to enhance the stability of the upstream tailings dams. The achieved average S.F of 1.55 meets the conventional requirement for long-term slope stability. These results imply that the treatment scheme considered in the analysis resulted in a 25% increase in the S.F of the dam. Additionally, Figure 12 shows the S.F variations for untreated tailings when rock fills were used for two cases of Ormann et al.’s (2013) model [35,36]. From the figure, it can be seen that using the treatment scheme proposed will yield an S.F much higher than when rock fills were used. These results are supported by other research on using additives with soil in general [8,38,46].

Figure 13, Figure 14 and Figure 15 show the progressive movement of the slip surfaces as the construction proceeds. As the construction continues, the volume of tailings enclosed within the critical slip surface increases (shown in gradient red tone colors). Thus, the contribution of the tailings to the shearing resistance along the slip surface increases, and thereby the contribution of the treated tailings to the stability of the tailings dam increases.

Effect of Tailings Treatment on Settlement

The stiffness of the treated tailings increased substantially compared to the untreated tailings [1]. This stiffness increase could have an important effect on the settlement of the tailings dams. In addition, the treatment reduces the hydraulic conductivity of the tailings, which could reduce the consolidation settlement. The effect of the tailings treatment on the settlement of the tailings dam is demonstrated in Figure 16 and Figure 17. As can be noted from these figures, the effect of this treatment method on the settlement was profound. For example, the settlement after the completion of the consolidation of the 11th layer for Case I of untreated soil was 3.60 m (Figure 12), while the settlement of the same case but with treated tailings’ stiffness was only 2.16 m (Figure 13), i.e., dam settlement decreased by almost 40% due to the tailings treatment.

Indeed, the tailings treatment investigated in the current study proved successful in not only increasing the overall stability of Aitik tailings dam for the proposed staged construction plan, around a 25% increase, meeting the conventional minimum safety factor of 1.5, but also in decreasing the amount of settlement by around 40%. Similar results were obtained by other researchers [29,46].

Effects of Tailings Stiffness on Safety Factor

The mechanical properties of the tailings treated with the mix proportion (1CKD:1B:1OC) were used in the numerical model to investigate the effect of the elastic modulus of the treated tailings, E, on the performance of the dam. The stiffness of the treated tailings was evaluated from the Oedometer test, E_oed, for the applied pressure range 100–400 kPa and was found to be 30.0 MPa. This value was used to represent the stiffness of the tailings in the numerical model of Case IX. Additionally, the analysis was repeated considering a reduced stiffness of 3.0 MPa. Figure 18 presents the variation of the safety factor with time for the two cases. It can be seen from Figure 18 that varying E_oed from 3.0 to 30.0 MPa had little to no observed effect on the safety factor. The effect of stiffer tailings was a small reduction in the EPP, which resulted in a small increase in S.F.

Effects of Changing Tailings k Value

Alsharedah (2015) demonstrated that the hydraulic conductivity is affected by the value of void’s ratio e and the applied effectives stress [2,47]. As the tailings containment is raised, effective stresses increase and the void ratio decreases with depth. Therefore, the hydraulic conductivity decreases as well. This interaction between the three parameters is not accounted for in the Mohr–Coulomb model in Plaxis 2D package. Therefore, it is interesting to explore the effect of using a constant value of k across the depth throughout the analysis on the calculated response of the dam. For this purpose, Case I of the untreated soil was re-analyzed considering different values of k. In the first analysis, k = 7.866 × 10⁻⁷ cm/s, which corresponds to effective stress of 5 kPa, was used, and in the second analysis, k = 2.11 × 10⁻⁸ cm/s, which corresponds to an effective stress of 800 kPa, was used. Figure 19 displays the variation of safety factors versus time for both cases. As expected, using a lower hydraulic conductivity yielded a lower safety factor.

Figure 20 shows the variation of excess pore pressure for the two cases at the end of sixth year. The results presented in Figure 20 demonstrate that the lower hydraulic conductivity resulted in higher excess pore pressure that is as much as twice that of the upper k value, 140 and 76 kPa, respectively. Moreover, when the low k value was used, the distribution of EPP changed and the contours of EPP expanded into the critical zone within which potential slip surfaces exist. This resulted in reduced effective stresses in this region and thus a reduced safety factor. Therefore, when a high k value is used, slope stability analysis may yield overly optimistic safety factors that are non-conservative. In fact, when the lowest k value was used in this study, failure is indicated to have initiated only six years from the construction beginning (Figure 19). The quick loading and the low hydraulic conductivity combined may cause static liquefaction to take place [11,14,15,45,48]. This demonstrates how important the selection of this parameter is when the slope stability of tailings dams is concerned [16,43,49,50,51]. This is because as the hydraulic conductivity increases, excess pore water dissipates quickly. Therefore, the value of k used should be cautiously selected to model tailings dams’ stability and should represent the range of stresses anticipated in the field. For a conservative estimate of the safety factor of the tailings dam, a low value of k should be used in the analysis [31,37,38,52,53]. However, it is much more appropriate and effective to simulate the change in k with effective stress, which is something the MC failure criterion in PLAXIS 2D is lacking. It is believed that renewed interests in recycling, coupled with the efforts of enforcing bodies such as governments and organizations for zero carbon emissions, such as that of the Paris Accord 2015, will help implement the increased usage of recycled materials for building new TSFs [17,54].

7. Conclusions

The effect of tailings treatment on the stability of tailings dams was investigated numerically. The outcomes of the numerical analyses can be summarized as follows:

• The soil treatment scheme adopted produced promising results regarding slope stability enhancement. The S.F of the treated tailings was consistently between 20 and 25% higher than that of untreated tailings.
• The proposed soil treatment reduced the maximum settlement of the TSF by almost 40% by the end of the 11th year. This can, in turn, help reduce the contouring cost by reducing the settlement and therefore the amount of borrowed materials required for final leveling for closure. The reduced contouring cost will help pay off the cost spent on the proposed soil treatment.
• The proposed soil treatment may eliminate the need for other stabilizing measures such as berms, rock fills, and compaction.
• The soil stiffness has a small effect on the safety factor but has a profound impact on the settlement.
• The hydraulic conductivity value should be cautiously selected for conducting any slope stability problem as its value will dictate how much water seeps out due to staged construction or any hydraulic energy change in the soil. This aspect is most important when analyzing the stability of tailings dams, where the k value can vary substantially from one point/time to another.

8. Future Research

Future research should study the effects of the tailings’ flow in cases of slope instability and the limit it reaches, as well as other important factors for zoning, nearby dwellings, and ecosystem safety. The effect of dam rising is a key factor, and a comparative study should be undertaken to investigate the effects of the construction sequence on the dam stability.