Mine Water Environ

DOI 10.1007/s10230-017-0468-y

TECHNICAL ARTICLE

Numerical Study of the Effects of Drainage Systems on Saturated/

Unsaturated Seepage and Stability of Tailings Dams
Ali Pak1 · Mostafa Nabipour1 

Received: 22 December 2015 / Accepted: 5 June 2017

© Springer-Verlag GmbH Germany 2017

Abstract  The stability of tailings dams is affected by Keywords  Capillary · Drainage · Seepage modeling ·
seepage characteristics such as the location of the phreatic Phreatic surface · Unsaturated permeability
surface inside the dam, the effects of the capillary fringe,
and the unsaturated zone above the zero pore pressure
level. In this study, the performance of drainage systems in Introduction
tailings dams was investigated by analyzing saturated and
unsaturated seepage in the dam, considering the effects of Tailings dams are major hydraulic structures that can store
the construction method, tailings properties, and the type of millions tons of the mine tailings as well as substantial
drainage systems. First, general seepage characteristics in quantities of supernatant water. There are two basic types
tailings dams were studied and the effects of non-homoge- of earthen structures that are used to retain tailings in
neity were investigated. Our results show that in a silty tail- impoundments: retention dams and raised embankments.
ings dam with a height of 15 m, unsaturated plus capillary Retention dams are constructed at the full height before the
seepage flux can reach 13% of the total seepage. The total beginning of the disposal, whereas raised embankments are
head vs. discharge volume curves under various conditions constructed in phases according to the need for additional
were compared and their practical implications are pre- volume. Raised embankments can be constructed using
sented. Then, stability analyses were carried out using the upstream, downstream, or centerline methods (Vick 1983;
results of seepage analyses for different construction meth- UNEP 1996; Mittal and Morgenstern 1977). Seepage is one
ods, material properties, and drainage systems. Finally, a of the most important factors that affects the stability of the
number of practical conclusions are drawn regarding dam tailings dam on one hand, and the environmental impacts
stability and the efficiency of toe, blanket, and chimney on the reservoir area, on the other (UNEP 2001; US 1994;
drains in different construction methods. Using a blanket Abadjiev 1976; Shen et al. 2011).
and/or a chimney drain can increase the stability safety fac- Raised tailings dams fail approximately ten times more
tor by up to two times, depending on the type of material. often than retention dams. Based on a large database gath-
ered from 1970 to 2001, it can be concluded that there
have been roughly two to five major tailings dam failure
incidents per annum. It is interesting to note that hydraulic
Electronic supplementary material  The online version of this
aspects, such as seepage, were found to be the cause of 38%
article (doi:10.1007/s10230-017-0468-y) contains supplementary
material, which is available to authorized users. of tailings dam failures (Davies 2002). These failure statis-
tics are about physical failures alone. Environmental haz-
* Mostafa Nabipour ards, such as groundwater and surface water pollution from
[email protected]
tailings impoundments, is also an international problem
Ali Pak (Lottermoser and Ashley 2005; Sharma and Al-Busaidi
[email protected]
2001).
1
Civil Engineering Department, Sharif University Studies have been carried out to determine seepage dis-
of Technology, Tehran, Iran charge through earthen dams using various numerical and

13
Vol.:(0123456789)
 Mine Water Environ

experimental methods (Fu and Sheng 2009). McWhorter to as unsaturated in this paper (Houston et al. 2000; Kovács
and Nelson (1980) studied one-dimensional seepage 2011; Lu and Likos 2004).
through tailings inside a reservoir and its influence on As the tailings dams are made of fine-grained soil, seep-
groundwater. Stauffer and Obermeyer (1988) investigated age through capillary and unsaturated zones in tailings
pore water pressures in some tailings dams with differ- dams is more important than that in earthen dams, and
ent operational conditions using a piezometric network. saturation may occur due to the capillary rise above the
Rykaart et  al. (2002) modeled surface flux boundary con- phreatic surface. Hence, it is usually recommended that the
ditions for water balance problems of tailings impound- upper saturation line should not intersect the downstream
ments using SVFlux software. Zhang et al. (2011) proposed face of the tailings dams; otherwise, the whole body of the
a generalized method for numerical modeling of three- tailings dam would virtually become saturated. Chapuis
dimensional seepage in tailing ponds. Yi et al. (2011) and and Aubertin (2001) applied a two-dimensional finite ele-
Yin et al. (2011) studied the effect of the phreatic surface ment code to estimate saturated and unsaturated seepages
on tailings dam stability. However, published studies to through dikes with different heights under steady state con-
determine seepage discharge from tailings dams are mostly ditions. They reported that the steady state seepage flux is
about one or two dimensional seepage through tailings; greater than predicted by classical saturated seepage theory,
research about seepage through the body of tailings dams especially for smaller dikes. Zandarin et  al. (2009) inves-
is rare. Because the environmental problems and number tigated the role of capillary rise in the stability of a tail-
of tailings dams’ failures are important, seepage analysis ing dam in Cuba and showed that the stability of the dam
through tailings dams is important. Also, previous endeav- strongly depends on the capillary phenomena.
ors have not thoroughly studied the effects of different The general configuration of flow nets is different in
drainage systems for tailings dams. conventional water dams and in tailings dams. For con-
The goal of this numerical study was two-fold. First, the ventional dams, flow lines are concave up while in tailings
effects of different seepage mechanisms on the quantity of dams, especially in upstream embankments, flow lines are
seepage through tailings dams were considered. Second, concave down (Vick 1983).
the influence of seepage regime and the location of the Providing adequate internal drainage can improve tail-
phreatic surface on the stability of tailings dams was inves- ings consolidation, decrease the likelihood of liquefaction,
tigated by analyzing different drainage systems. and improve structural stability (Valipour 2012). Rico et al.
(2008) showed the importance that correctly sizing a tail-
ings dam’s drainage system can have on general perfor-
Unsaturated Seepage in Tailings Dams mance and safety.
The upstream construction method does not usually
Several experiments have shown that pore water can allow for incorporation of vertical drains. Horizontal drains
migrate from the saturated zone to the unsaturated zone and may be installed during construction of the starter dike to
vice versa. However, water cannot flow through unsaturated keep the phreatic surface low within the embankment. Pro-
soil as easily as through saturated soil, because the unsatu- viding a blanket drain that extends towards the upstream
rated hydraulic conductivity is less (Delleur 2006; Fred- of the starter dike can be effective in lowering the phreatic
lund and Rahardjo 1993; Thieu et al. 2000). The governing surface in the initial and subsequent embankment rises.
equation of saturated–unsaturated seepage can be derived However, special efforts must be taken to ensure against
using continuity equation and Darcy’s law. The seepage blockage of the blanket drains in upstream embankments
equation in homogeneous, anisotropic, unsaturated soil (UNEP 1994; Van Zyl et al. 1980).
under steady-state conditions is: The downstream construction method allows for incor-
poration of all types of drains to control the phreatic sur-
𝜕 2 hw 𝜕 2 hw face. For example, an inclined chimney drain near the
kwx + k = 0,
𝜕x2 wy
𝜕y2 upstream face of the dike, connected to a blanket drain at
the dike base, can be installed with each successive raise
where, kw is the coefficient of water permeability of unsatu- of the embankment. Drainage zones can be similarly incor-
rated soil which is a function of pore water pressure, and porated into the construction using the centerline method
hw is total head of water (Boufadel et al. 1999; Dogan and (Nabipour 2006). For centerline construction, a vertical
Motz 2005; Fredlund and Rahardjo 1993). Above the phre- chimney drain may be proper. Supplemental Fig.  1 illus-
atic level, two different zones can be recognized: the capil- trates the types of tailings dam construction methods with
lary fringe (with considerable moisture content), and a par- common drains, as discussed above. The location of the
tially saturated zone (with less moisture) that we will refer phreatic surface in the body of the tailings dams and the

13
Mine Water Environ

effects of various drainage systems on the stability of the A. For upstream dams, seepage was analyzed for two
slopes dams will be discussed later. cases: no drain and a 4 m high toe drain with an inlet
slope of 45°. The dimensions and properties of the toe
drain were kept constant for the various construction
heights.
Numerical Simulation of Seepage through Tailings B. For centerline dams, seepage was analyzed for three
Dams cases: no drain, with a blanket drain, and with a ver-
tical chimney drain. The blanket and chimney drains
The saturated–unsaturated seepage in tailings dams were were assumed to be 2 m thick.
analyzed (assuming 2D steady state), considering the C. For downstream dams, seepage was analyzed for
effects of dam construction (upstream, centerline, and three cases: no drain, with a blanket drain, and with
downstream), the material properties of the dam and the an inclined chimney drain. The blanket and chimney
tailings, and the type and performance of the drainage sys- drains were assumed to be 2 m thick.
tem. The tailings dams simulated in this parametric study
were assumed to be homogeneous dams built on an imper- Tailings and water levels in the dam reservoir were
meable bottom, with a 10  m wide crest and 1:4 slope of considered the same, i.e. 1  m below the dam crest. The
the faces, constructed in three stages (with heights of 15, boundary condition for the water table at the reservoir was
30, and 45 m). Changes in the permeability coefficient and assigned a constant head, equal to atmospheric pressure.
hydraulic anisotropy of the materials were considered in Therefore, heads of 14, 29, and 44  m were applied to the
the simulations. The general geometry, configuration of the upstream surface, for the first, second, and third construc-
mesh, and boundary conditions are shown in Supplemental tion stages, respectively.
Fig. 2. Flow field and streamlines should remain uniform
Seep/W software was used because of its capabilities upstream of the reservoir through the steady-state seepage
for simulating partially saturated flow in complex geom- analysis. Therefore, infinite elements (which are useful for
etry with different materials (Geo-Slope International Ltd. modeling far field boundary conditions) were placed at the
2002). This software solves the Laplace equation for steady left boundary, and then a zero flux boundary condition was
state saturated–unsaturated flow using the finite element applied to the right boundary nodes.
method. The geometry of the structure, boundary condi- Since the intersection of the phreatic line with the down-
tions, and the hydraulic properties of the materials are the stream slope of the dam had to be determined iteratively by
required input data. The most important material proper- the software, the downstream face of the dam was defined
ties are the curves that demonstrate variations of hydraulic as a “review boundary”.
conductivity vs. pore-water pressure and hydraulic conduc- Since the bottom of the dams was supposed to be imper-
tivity anisotropy (Geo-Slope International Ltd. 2002). The meable, a “no flow” boundary condition was designated
variation of hydraulic conductivity with pore-water pres- for the bottom line. The hydraulic boundary conditions of
sure makes the finite element equations nonlinear, and con- the models are presented in Supplemental Fig.  2. For sta-
sequently an iterative process. This software was used by bility analysis, the bottom nodes were assumed to be fixed
some researchers to model partially saturated flow (Leoni against vertical and horizontal movement.
et al. 2004; Lee et al. 2009; Bussière et al. 2003).
Quadrilateral and triangular elements (with 4 and 3 Tailings Materials
Gauss points, respectively) were used to mesh the geom-
etry. Based on mesh sensitivity analysis, the dimension of Tailings dams are usually constructed of tailings or
the quadrilateral elements was set at 1  m. In a 15  m high cycloned sand. In this study, three types of materials were
tailings dam, domain discretization was performed with selected for the body of tailings dams, based on the hydrau-
1734 nodes and 1572 elements (Supplemental Fig. 2). Over lic conductivity of typical mine tailings. In the field, the
9500 nodes were required for a downstream tailings dam ratio of the horizontal to the vertical hydraulic conductivity
with a height of 45 m and a chimney drain system. of tailings usually ranges from 2 to 10, the effects of three
anisotropic permeability ratios (1, 5, and 10) were investi-
gated. In all of the analyses, the tailings that had settled in
Types of Tailings Dams Drains, and Boundary the reservoir were assumed to be silt. Also, sand was cho-
Conditions sen as the material for the drains, because sand is usually
available at tailings dam sites (Table 1).
The common types of drainage, as described below, were The function of hydraulic conductivity vs. pore-water
considered for this numerical investigation: pressure of these materials must be specified for numerical

13
 Mine Water Environ

Table 1  Main properties of Material Position Ksat (m/s) Air entry Specific grav- Cohesion Friction
materials used in seepage and value (KPa) ity (KN/m3) (KPa) angle
stability analyses (°)

Fine sand Dam body 4.3 × 10−6 4 20 0 28

Sandy silt (coarse tailings) Dam body 4.8 × 10−7 10 18 0 25
Silty clay (fine tailings) Dam body 3 × 10−8 40 17 10 25
Silt (tailings) Reservoir 5.8 × 10−8 10 17 5 25
Sand Drain 5.4 × 10−5 6 20 0 30

simulation in saturated–unsaturated seepage analysis. This Construction methods, types of drainage systems, and
function can be determined by experimental (Corey 1957) material properties of the dam body were compared in this
or predictive methods (Fredlund et  al. 1994; Green and study. Different cases considered for the parametric study
Corey 1971; Van Genuchten 1980). Because of air diffusion are shown in Table  2. After eliminating similar cases, the
and small flow quantities, measuring unsaturated hydraulic total number of simulations were reduced to 198. For fine
conductivity is complex (Brooks and Corey 1966; Green sand material with very steep hydraulic conductivity func-
and Corey 1971), and various procedures have been used tions, the problem may be extremely nonlinear and con-
to predict it (Fredlund et al. 1994; Green and Corey 1971; vergence problems can appear. In such cases, selection
Van Genuchten 1980). Green and Corey (1971) introduced of proper convergence parameters, such as the maximum
a practical method and reported that their approach was number of iterations, convergence tolerance, and mesh
fairly accurate for most field applications. Elzeftawy and refinement can eliminate problems.
Cartwright (1981) subsequently verified Green and Corey’s In most of the cases discussed below, results for the k­ h/kv
method for various soils. value of 10 were used to present the simulation outcomes
Based on the typical properties of tailings (Bruch 1993; in order to illustrate the most realistic case in practice.
Geo-Slope International Ltd. 2002; Gonzalez and Adams
1980; Swanson et  al. 1999; Vick 1983), five materials
were selected for the various parts of the tailings dams and Results and Discussion
hydraulic conductivity vs. pore pressure functions of these
materials were predicted using Green and Corey’s method. Seepage Characteristics with Respect to Dam
The main hydraulic properties of the materials are pre- Construction Method
sented in Table  1. As shown in Supplemental Fig.  3, the
pore-water pressure in the unsaturated zone above the water Figure 1 shows the total head of decant water vs. the total
table is negative and the hydraulic conductivity changes as amount of seepage discharged for tailings dams with no
a function of the negative pore-water pressure. In Table 1, drain constructed from silty clay, sandy silt, and fine sand
air entry pressure is the matric suction where air starts to material (with an anisotropy kh ratio of 10) for the three dif-
k

enter the largest pores in the soil. Coarse materials such as

v

ferent construction methods. Because of the low amount of

sand have a low air entry pressure, high saturated perme-
seepage flux in embankments, especially in tailings dams,
ability, and steep hydraulic conductivity curve.
and to better display the horizontal axis numbers, the unit

Table 2  Different cases of parametric study

Construction method Drainage system Dam height (m) Dam body material Anisotropy ­(kh/kv)

Upstream No drain 15, 30, 45 Fine sand, sandy silt, silty clay 1, 5, 10
Toe drain
Centerline No drain 15, 30, 45 Fine sand, sandy silt, silty clay 1, 5, 10
Blanket drain
Chimney drain
Downstream No drain 15, 30, 45 Fine sand, sandy silt, silty clay 1, 5, 10
Blanket drain
Chimney drain

13
Mine Water Environ

Fig. 1  Total head (H) vs. a 50

total seepage discharge ­(Qt)
of tailings dams with no drain
constructed from a silty clay, b Silty Clay
40
sandy silt, c fine sand mate-
rial (with the anisotropy of 10)
for upstream (black diamond), 30
centerline (black square) and

H (m)
downstream (red triangle)
construction methods 20

10

0
0 5 10 15 20
Qt (L/day/m)

b 50

40 Sandy Silt

30
H (m)

20

10

0
0 50 100 150 200
Qt (L/day/m)

c 50

Fine Sand
40

30
H (m)

20

10

0
0 100 200 300 400 500 600 700 800 900
Qt (L/day/m)

of seepage flux was calculated as L/day through 1  m of The total seepage discharge increases with the total head
­ 3/s, seepage flux (L/
dam width. To convert this unit to m (H), and the amount and rate of seepage discharge vary
day/m) must be divided by 8.64 × 107. The total seepage based on the dam body materials and construction meth-
fluxes were measured at vertical section in the dam axis ods. In the case of silty clay, the slope of the curves for the
(from bottom to dam crest). upstream dam is less than the slope of centerline dam, and

13
 Mine Water Environ

in turn, the slope of the centerline dam is lower than that of Effects of Non‑homogeneity of Tailings Materials
the downstream dam (Fig. 1). This means that for a certain
H, the amount of seepage in an upstream dam is greater Gradual deposition of tailings inside a reservoir over the
than the amount of seepage in centerline and downstream lifetime of a mining operation can cause the consolida-
dams. However, Fig. 1 also indicates that for sandy silt and tion of previously deposited materials, reducing the per-
fine sand materials, the sequence of the slopes of the curves meability coefficient of the lower layers. To investigate
is the reverse. In other words, the amount of seepage for a the effects of such non-homogeneity, let’s suppose that
certain H in upstream dams with sandy silt and fine sand the saturated hydraulic conductivity of the lower (the
materials is less than for centerline and downstream dams. bottom 15  m) and middle layers (from 15 to 30  m) are
This is because sandy silt and fine sand are more permeable decreased based on the values in Table 3.
than the tailings inside the reservoir (i.e. silt), while silty Using sandy silt as an example, Fig.  3 shows how
clay is less permeable than silt. These results are compat- hydraulic conductivity is reduced as the weight of new
ible with the results of Chapuis and Aubertin (2001) with material consolidates previously deposited materi-
respect to the seepage flux with water head elevation. als, which in turn decreases seepage volume. This is
Moreover, the head vs. discharge curve is almost linear especially significant in the case of slurries with poten-
and passes approximately through the origin for center- tially toxic leachates that might threaten groundwater
line and downstream construction methods especially for resources. In other words, the performance of the tail-
silty clay and sandy silt materials. These phenomena con- ing dams can gradually improve over time. However, this
firm validity of Darcy’s law. But, in upstream construction reduction of seepage flux in sub-layers can also cause
method, the head vs. discharge curve is almost parabolic if excessive pore water pressure to develop (Ito and Azam
it continues through the origin. The reason may be attrib- 2013).
uted to the steep hydraulic conductivity function of sandy
material depicted in Supplemental Fig. 3.
Figure 2 illustrates seepage volume for a centerline con-
structed dam with an anisotropy ratio of 10. It is obvious
that the amount of the total seepage flux increases rapidly
as the permeability coefficient of the tailings increases
when one compares the slope of the curve for fine sand Table 3  Saturated hydraulic conductivity of lower and middle layers
(with a high permeability coefficient) and the slope of silty of inhomogeneous tailings in reservoir
clay (with a low permeability coefficient). It should be
Dam height Ksat of lower lay- Ksat of middle lay- Ksat of
mentioned that in all cases, the saturated hydraulic conduc- (m) ers (m/s) ers (m/s) upper layers
tivity coefficient of the stored tailings inside the reservoir (m/s)
was kept constant.
15 5.8 × 10−8 – –
30 10−8 5.8 × 10−8 –
45 5 × 10−9 10−8 5.8 × 10−8

Fig. 2  Total head (H) vs. 50

total seepage discharge ­(Qt) of
centerline tailings dams with
no drain for different materials 40
(with the anisotropy of 10)

30
H (m)

20

10
Silty Clay Sandy Silt Fine Sand

0
0 100 200 300 400 500 600
Qt (L/day/m)

13
Mine Water Environ

Fig. 3  Total head (H) vs. a 50

total seepage discharge ­(Qt) of
tailings dams with no drain con-
structed from sandy silt materi- 40 Upstream
als (with the anisotropy ratio of
10) for a upstream, b centerline,
30
c downstream construction

H (m)
methods with homogeneous
(black square) and non-homo- 20
geneous (red triangle) tailings
in reservoir
10

0
0 50 100 150 200
Qt (L/day/m)

b 50

40
Centerline

30
H (m)

20

10

0
0 50 100 150 200
Qt (L/day/m)

c 50

40 Downstream

30
H (m)

20

10

0
0 50 100 150 200
Qt (L/day/m)

Comparison of Saturated, Capillary, and Unsaturated total seepage flux, the proportion of (­Qc  +  Quns) is pre-
Seepage Flux sented as a percentage of Q­ t in Fig. 4. Capillary seepage
fluxes are indicated at a vertical section in the dam axis
As mentioned before, water flows in capillary and unsat- [from phreatic line to capillary line (from blue line to
urated zones as well as in the saturated zone. The cor- red line in Supplemental Fig.  4)] and unsaturated seep-
responding velocity vectors for different zones are illus- age fluxes are measured at the same location (from cap-
trated in Supplemental Fig.  4. To demonstrate how the illary line to dam crest). To obtain the precise amounts
capillary plus the unsaturated seepage fluxes can affect of the capillary and unsaturated seepage fluxes, the

13
 Mine Water Environ

Fig. 4  Total head (H) vs. the 50

percentage of flow discharges
through capillary and unsatu- 40
rated zones comparing to the
total discharge [(Qc + Quns)/Qt]
for isotropic fine sand, sandy 30

H (m)
silt, and silty clay materials in
downstream tailings dams 20

10
Fine Sand Sandy Silt Silty Clay
0
0 2 4 6 8 10 12 14
(Qc + Quns)/Qt (%)

computational elements of these zones were refined in dominates. This effect becomes more significant with
the analysis. anisotropy, as discussed below.
Figure  4 illustrates the total head vs. the percentage of
discharge through the capillary and unsaturated zones for
isotropic fine sand, sandy silt, and silty clay in a down- Effect of Anisotropy
stream dam with no drain. It shows that finer-grained
material allows more capillary and unsaturated flow to dis- The vertical permeability coefficient decreases with
charge due to its higher capillary fringe. The flow discharge increasing hydraulic conductivity anisotropy. As a result,
through these zones is about 5 and 13% of the total seepage the total seepage flux decreases with respect to the ani-
for 15  m high sandy and silty clay tailings dams, respec- sotropy ratio, while the percentage of capillary plus
tively. These phenomena are compatible with the results of unsaturated flow discharge increases (Fig.  6). The vari-
Freeze (1971), who concluded that unsaturated seepage in ation of the location of the phreatic surface with anisot-
low-height dams is more important than that in high dams, ropy ratio agrees with Vick (1983). For isotropic sandy
and the effect of unsaturated flow in fine-grained materials silt material, the total seepage flux is approximately 1.2
is more than in coarse-grained materials. times greater than that of material with an anisotropy
Generally, the total, capillary, and unsaturated seepage ratio of 10, but the percentage of capillary plus unsatu-
discharges increase with total head, but as the height of rated flow discharge is ≈2.4% less than that of material
a dam gradually increases, the amount of capillary and with anisotropy ratio of 10. Comparison of Figs.  5 and
unsaturated discharges becomes less significant. Figure 5 6a reveals that the total seepage flux curve is linear and
illustrates the trend of variation of the total head vs. cap- roughly passes through the origin. However, the capillary
illary and unsaturated discharges, separately. It can be and unsaturated discharge curves are nonlinear, reflecting
inferred that the ratio of capillary discharge to unsatu- the nonlinear nature of permeability variation in unsatu-
rated discharge is approximately 2 for sandy silt with an rated soils.
anisotropy ratio of 10, indicating that capillary seepage

Fig. 5  Total head (H) vs. 50

capillary ­(Qc) and unsaturated
discharges ­(Quns) in downstream Quns
40
tailings dams with no drain for
sandy silt materials (for the case Qc
of anisotropy of 10) 30
H (m)

20

10

0
0 2 4 6 8 10 12
Q (L/day/m)

13
Mine Water Environ

Fig. 6  Total head (H) vs. a total a 50

seepage discharge (­ Qt), b the
percentage of capillary and
unsaturated flow discharges 40
[(Qc + Quns)/Qt], in downstream
tailings dams with no drain
constructed from sandy silt 30

H (m)
material for the anisotropy ratio
(kh ) of 1 (black diamond), 5
k

20
v

(black square) and 10 (red

triangle)
10

0
0 50 100 150 200 250
Qt (L/day/m)

b 50

40

30
H (m)

20

10

0
0 2 4 6 8 10 12 14
(Qc + Quns)/Qt (%)

Stability Analysis of Tailings Dams with Respect the tailings dam’s safety factor. Supplemental Fig. 5a indi-
to the Phreatic Line cates that the phreatic lines in upstream dams are similar
with and without a toe drain. However, blanket and chim-
One of the factors contributing to the stability of embank- ney drains dramatically affect the location of the phreatic
ments in general and tailings dams in particular is the posi- line, allowing most of the dam to remain unsaturated, which
tion of the phreatic line. A properly functioning drainage protects against liquefaction-induced failure. In Supple-
system can lower the phreatic surface and thereby increase mental Figs. 5b, c, blanket and chimney drains are shown in

Fig. 7  Unsaturated area to 60
dam body area ratio (­ Auns/At)
vs. total head (H) of centerline
50
tailings dams constructed from
sandy silt material (with the
anisotropy of 10) for different 40
A uns / At (%)

drainage systems
Blanket & Chimney
30

Blanket Drain
20

No Drain
10

0
0 10 20 30 40 50
H (m)

13
 Mine Water Environ

red and brown, respectively. Figure 7 represents the ratio of to those of a downstream tailings dam. It is clear that toe
the unsaturated area to dam cross section area vs. total head drains do not significantly increase a dam’s factor of safety
for a centerline tailings dams constructed from sandy silt for higher dam levels (Fig. 9a). On the other hand, blanket
(with an anisotropy ratio of 10) for the different drainage and chimney drains increase the minimum safety factor sig-
systems. The unsaturated area in this figure is considered to nificantly for all heights (Fig. 9b). The minimum safety fac-
be the area above the capillary zone. The ratio of the unsat- tor for a 15 m high downstream tailings dam with a blan-
urated area to total cross section of the dam with no drain is ket drain is comparatively low because the blanket drain is
less than the ratios with drains and does not change consid- short during the first stage of construction.
erably with an increase of total head. But, with blanket and Figure 10 illustrates the minimum safety factor of 30 m
chimney drains, the unsaturated area can reach up to about high tailings dams vs. the hydraulic conductivity of dam
60% of the dam’s cross section area. body materials (with an anisotropy ratio of 10) for upstream
Figure  8 shows the unsaturated area to dam body area and downstream construction methods with different drain-
ratio vs. the total head of centerline tailings dams con- age systems. Again, the minimum safety factor depends on
structed of sandy silt (with an anisotropy of 10) for differ- the phreatic line position. With increased hydraulic con-
ent construction methods and drainage systems. The mean ductivity, the phreatic line drops and the minimum safety
value of the ratio of the unsaturated area to dam cross sec- factor increases. However, in downstream tailings dams
tion area for downstream tailings dam with a blanket drain with blanket and chimney drains, the minimum safety fac-
is 6% less than with a chimney drain. For upstream con- tor does not change with hydraulic conductivity because, in
struction, a toe drain does not notably lower the phreatic these cases, the phreatic line position is already very low.
line. Also, drainage systems perform better in centerline
tailings dams than in downstream dams. Effects of Drainage Systems
In this study, stability analyses were carried out using
Slope/W software, which is fully integrated with Seep/W For all three construction methods, the milder the slope
and can accommodate negative pore-water pressures just as of the head vs. discharge curve, the better the drainage
well as positive pressures. Soil suction (negative pore-water systems perform (Fig.  11). Figure  11a shows that with
pressure) increases soil strength. Slope/W uses the limit upstream construction, seepage discharges with no drain
equilibrium method to calculate the safety factor against and toe drains are similar, especially with higher elevations.
failure. Spencer’s method was used to compute the safety As can be seen, the effectiveness of toe drains decreases
factor because of its complete force and moment equilib- in the second and third construction stages. This observa-
rium procedure. Coulomb’s equation was used to describe tion is consistent with the results of Saad and Mitri (2011),
the shear strength of materials. No foundation failures were who reported that installation of a drainage system at the
considered. starter dyke did not sufficiently lower the phreatic surface
Figure  9 shows the minimum safety factor vs. the total in upstream dams.
head of tailings dams constructed of sandy silt (with an ani- In centerline tailings dams with blanket drain, the seep-
sotropy ratio of 10) for upstream and downstream construc- age discharge is greater than with no drain (Fig.  11b).
tion with different drainage systems. The centerline tail- Because of the increased length of a blanket drain in higher
ings dam curves are not presented because they are similar centerline tailings dams, this seepage discharge difference

Fig. 8  The ratio of unsaturated 60

area to dam cross section area
­(Auns/At) vs. total head (H) of
50
tailings dams constructed from Toe Drain
sandy silt material (with the (Upstream)
anisotropy ratio of 10) for dif- 40 Blanket Drain
A uns / At (%)

ferent construction methods and (Centerline)

drainage systems
30 Blanket & Chimney
(Centerline)

20 Blanket Drain
(Downstream)

Blanket & Chimney

10 (Downstream)

0
0 10 20 30 40 50
H (m)

13
Mine Water Environ

Fig. 9  Minimum factor of a 2
safety (MFOS) vs. total head
(H) of tailings dams constructed
from sandy silt material (with 1.8
the anisotropy of 10) for a
upstream and b downstream
construction method with differ- 1.6
ent drainage systems

M FOS
Toe Drain
1.4
No Drain

1.2

1
0 10 20 30 40 50
H (m)

b 2

1.8
Blanket & Chimney

1.6
Blanket Drain
M FOS

1.4 No Drain

1.2

1
0 10 20 30 40 50
H (m)

increases with greater heads. This increased seepage flux be environmentally desirable, they improve tailings dams
due to blanket drain lengthening is also reported for earthen stability, as discussed above.
dams (Mishra and Singh 2005). Figure  11b shows that
chimney drains perform better than blanket drains in cen-
terline constructed dams. Conclusion
Figure  11c demonstrates that the performance of blan-
ket and chimney drains are almost the same for down- Tailing dams are major hydraulic structures that differ from
stream tailings dams. This is because of the extra length water storage dams in various ways. Seepage characteris-
of the blanket drain in downstream construction. Thus, in tics play a vital role in the performance of tailing dams,
the downstream method, if the blanket drain is constructed especially when the seepage contains potentially toxic
from free draining material with a proper thickness, there contaminants. Two issues regarding this seepage are espe-
seems to be no need to construct a chimney drain. cially important: the volume discharged, which is environ-
Figure  11 reveals that without a drain, the discharge mentally important, and the position of the phreatic surface
responds linearly to water head, in accordance with Darcy’s inside the dam, which strongly affects dam stability. In
law. However, when drains are used, nonlinearity appears, this paper, we used a series of numerical models to inves-
such that flow through coarse-grained drains does not nec- tigate the effects of the dam construction method, drain-
essarily follow Darcy’s law. Finally, although drainage sys- age system, saturated/unsaturated seepage, and anisotropy
tems increase the amount of total seepage, which may not of hydraulic conductivity on the discharge volume and the

13
 Mine Water Environ

Fig. 10  Minimum factor of a 2
safety (MFOS) of tailings dams
(with 30 m height) vs. hydraulic Toe Drain
conductivity of dam body mate- 1.8
rials (with the anisotropy of 10)
for a upstream and b down- No Drain
stream construction method 1.6

M FOS
with different drainage systems

1.4

1.2

1
0 0.000001 0.000002 0.000003 0.000004 0.000005

hydraulic conductivity (m/s)

b 2

1.8

Blanket & Chimney

1.6
M FOS

Blanket Drain

1.4
No Drain

1.2

1
0 0.000001 0.000002 0.000003 0.000004 0.000005

hydraulic conductivity (m/s)

position of the phreatic surface. Commonly used materials are used because the thickness of the capillary fringe
for tailing dam construction (fine sand, sandy silt, and silty increases in fine-grained soils.
clay) were considered. The results indicate that: 3. Gradual deposition of materials causes self-weight
consolidation of previously deposited materials which,
1. The total, capillary, and unsaturated seepage all in turn, reduces the coefficient of permeability. There-
increase with the total head (or with the height of the fore, seepage volume can be expected to gradually
dam), but the relative percentage of discharge through decrease over time.
the capillary and unsaturated zones relative to the total 4. Despite their common use, toe drains do not signifi-
seepage gradually diminishes. Therefore, the capillary cantly affect the seepage regime of tailings dams. As
and unsaturated seepage seem to be more important for the height of a dam increases, the performance of a
low tailings dams. toe drain becomes less significant, even for upstream
2. Unsaturated and capillary seepage fluxes are strongly dams. For better performance of toe drains in upstream
influenced by material types. In 15 m high sandy tail- embankments, the toe drain dimensions should be
ings dams, the unsaturated plus capillary seepage flux based on the final dam configuration.
constitute only 5% of the total seepage, while in silty 5. For downstream dams, blanket and the chimney drains
tailings dam, it can be up to 13%. This indicates the perform approximately the same. In other words, if a
importance of unsaturated seepage analysis in low tail- blanket drain is constructed of free-draining mate-
ings dams constructed with less permeable material. rial with the proper thickness, the additional cost for
On average, the seepage through the capillary zone is a chimney drain may not be justified. For a centerline
twice the seepage through the unsaturated zone. This dam, a blanket drain connected to a vertical chimney
becomes more significant when finer grained materials drain performs the best. Also, it seems that from the

13
Mine Water Environ

Fig. 11  Total head (H) vs. a 50

total seepage discharge ­(Qt) of
tailings dams constructed from
sandy silt material (with the ani- 40
sotropy of 10) for a upstream,
b centerline, c downstream No Drain Toe Drain
construction methods 30

H (m)
20

10

0
0 20 40 60 80 100 120
Qt (L/day/m)

b 50

40

30
H (m)

20

10
No Drain Blanket Chimney

0
0 50 100 150 200 250 300 350
Qt (L/day/m)

c 50

40

30
H (m)

20

10
No Drain Blanket Chimney

0
0 100 200 300 400 500 600
Qt (L/day/m)

seepage point of view, construction of centerline dams which reduces the possibility of liquefaction-induced
is more advisable. failure in tailing dams. This effect is more pronounced
6. Blanket and/or chimney drains increase the unsaturated for low and medium height dams; for higher dams, the
part of the dam body by lowering the phreatic surface, rate of improvement gradually decreases.

13
 Mine Water Environ

7. From a safety perspective, the use of blanket and/or Lottermoser BG, Ashley PM (2005) Tailings dam seepage at the
chimney drain can increase the stability safety factor rehabilitated Mary Kathleen uranium mine, Australia. J Geo-
chem Explor 85(3):119–137
up to two times, depending on the material type. Lu N, Likos WJ (2004) Rate of capillary rise in soil. J Geotech
Geoenviron 130(6):646–650
McWhorter DB, Nelson JD (1980) Seepage in the partially
saturated zone beneath tailing impoundments. Min Eng
32(4):432–439
References Mishra GC, Singh AK (2005) Seepage through a levee. Int J
Geomech 5(1):74–79
Abadjiev CB (1976) Seepage through mill tailings dams. Trans, 12th Mittal HK, Morgenstern NR (1977) Design and performance of
ICOLD, Mexico City, vol 1, pp 381–393 tailings dams. In: Proceedings of ASCE geotechnical specialty
Boufadel MC, Suidan MT, Venosa AD, Bowers MT (1999) Steady conference, pp 475–492
seepage in trenches and dams: effect of capillary flow. J Hydraul Nabipour M (2006) Investigation and analysis of seepage through
Eng ASCE 125(3):286–294 body of tailings dams using numerical modeling. MSc Thesis,
Brooks RH, Corey AT (1966) Properties of porous media affecting Sharif Univ of Technology, Tehran, Iran (in Persian)
fluid flow. J Irrig Drain E-ASCE 92(2):61–88 Rico M, Benito G, Salgueiro AR, Diez-Herrero A, Pereira HG
Bruch PG (1993) A laboratory study of evaporative fluxes in homoge- (2008) Reported tailings dam failures: a review of the Euro-
neous and layered soils. MSc Thesis, Dept of Civil Eng, Univer- pean incidents in the worldwide context. J Hazard Mater
sity of Saskatchewan, Saskatoon, Canada 152(2):846–852
Bussière B, Chapuis RP, Aubertin M (2003) Unsaturated flow mod- Rykaart M, Fredlund M, Stianson J (2002) Modelling tailings dam
eling for exposed and covered tailings dams. In: Proceedings of flux boundary condition with 3D seepage software. Ground Eng
international symposium on major challenges in tailings dams, 35(7):28–30
pp 57–70 Saad B, Mitri H (2011) Hydromechanical analysis of upstream tail-
Chapuis RP, Aubertin M (2001) A simplified method to estimate satu- ings disposal facilities. J Geotech Geoenviron 137(1):27–42
rated and unsaturated seepage through dikes under steady state Sharma RS, Al-Busaidi TS (2001) Groundwater pollution due to a
conditions. Can Geotech J 38(6):1321–1328 tailings dam. Eng Geol 60(1):235–244
Corey AT (1957) Measurement of water and air permeability in unsat- Shen L, Luo S, Zeng X, Wang H (2011) Review on anti-seepage
urated soil. Soil Sci Soc Am J 21(1):7–10 technology development of tailings pond in China. Proced Eng
Davies MP (2002) Tailings impoundment failures: are geotechnical 26:1803–1809
engineers listening. Geotech News, pp 31–36 Stauffer PA, Obermeyer JR (1988) Pore water pressure condition in
Delleur JW (2006) The handbook of groundwater engineering, tailings dams. Hydraulic fill structures, Proceedings of ASCE
2nd edn. CRC Press, New York special conference, pp 924–939.
Dogan A, Motz LH (2005) Saturated-unsaturated 3D groundwater Swanson DA, Savci G, Danziger G, Mohr RN, Weiskopf T (1999)
model, І: development. J Hydraul Eng ASCE 10(6):492–504 Predicting the soil-water characteristics of mine soils. In: Pro-
Elzeftawy A, Cartwright K (1981) Evaluating the saturated and unsat- ceedings of 6th international conference on tailings and mine
urated hydraulic conductivity of soils. Permeability and ground- waste conference, pp 345–349
water contaminant transport, ASTM STP, pp 168–181 Thieu NT, Fredlund DG, Hung VQ (2000) General partial differen-
Fredlund DG, Rahardjo H (1993) Soil mechanics for unsaturated tial equation solvers for saturated-unsaturated seepage. In: Pro-
soils. Wiley, New York ceedings of Asian conference on unsaturated soils, Singapore,
Fredlund DG, Xing A, Huang S (1994) Predicting the permeability pp 201–206
function for unsaturated soils using the soil-water characteristic UNEP (1994) Tailings dams-design of drainage. Bull 97, Interna-
curve. Can Geotech J 31(4):533–546 tional Commission on Large Dams, Paris
Freeze RA (1971) Influence of the unsaturated flow domain on seep- UNEP (1996) A guide to tailings dams and impoundments, design,
age through earth dams. Water Resour Res 7(4):929–941 construction, use and rehabilitation. Bull 106, International
Fu JF, Sheng JI (2009) A study on unsteady seepage flow through Commission on Large Dams, Paris
dam. J Hydrodyn 21(4):499–504 UNEP (2001) Tailings dams risk of dangerous occurrences. Bull 121,
Geo-Slope International Ltd. (2002) SeepW User’s Guide. Geo-Slope International Commission on Large Dams, Paris
International Ltd, Calgary US EPA (1994) Technical report: design and evaluation of tail-
Gonzalez PA, Adams BJ (1980) Mine tailings disposal (No. UTCE- ing dams. Office of Solid Waste, US Environmental Protection
80-06). University of Toronto, Canada Agency, Washington DC
Green RE, Corey JC (1971) Calculation of hydraulic conductivity: a Valipour M (2012) A comparison between horizontal and vertical
further evaluation of some predictive methods. Soil Sci Soc Am drainage systems (include pipe drainage, open ditch drainage,
J 35(1):3–8 and pumped wells) in anisotropic soils. IOSR J Mech Civil Eng
Houston WN, Houston SL, Al-Samahiji D (2000) Water content and 4(1):7–12
soil suction in the capillary zone. In: Proceedings of Asian con- Van Genuchten MT (1980) A closed-form equation for predicting the
ference on unsaturated soils, Singapore, pp 387–391 hydraulic conductivity of unsaturated soils. Soil Sci Soc Am J
Ito M, Azam S (2013) Large-strain consolidation modeling of mine 44(5):892–898
waste tailings. Environ Syst Res 2(1):7 Van Zyl D, Robertson AM, Mac G (1980) Subsurface drainage of tail-
Kovács G (2011) Seepage hydraulics, vol 10. Elsevier, New York ing impoundments some design, construction and management
Lee LM, Gofar N, Rahardjo H (2009) A simple model for prelimi- considerations. In: Proceedings of symposium on uranium mill
nary evaluation of rainfall-induced slope instability. Eng Geol tailings management, Colorado State University, pp 153–177
108(3):272–285 Vick SV (1983) Planning, design and analysis of tailings dams.
Leoni GL, Almeida MD, Fernandes HM (2004) Computational mod- BiTech Publ, Vancouver
eling of final covers for uranium mill tailings impoundments. J Yi Y, Wei S, Shufen L (2011) Tailings dam stability analysis of the
Hazard Mater 110(1):139–149 process of recovery. Proced Eng 26:1782–1787

13
Mine Water Environ

Yin G, Li G, Wei Z, Wan L, Shui G, Jing X (2011) Stability analysis Zhang LT, Qi QL, Xiong BL, Zhang J (2011) Numerical simulation
of a copper tailings dam via laboratory model tests: a Chinese of 3-D seepage field in tailing pond and its practical application.
case study. Miner Eng 24(2):122–130 Proced Eng 12:170–176
Zandarin MT, Oldecop LA, Rodriguez R, Zabala F (2009) The role
of capillary water in the stability of tailing dams. Eng Geol
105(1):108–118

13