See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/352559655

WATER RETENTION BEHAVIOUR OF TAILINGS IN UNSATURATED

CONDITIONS

Article in Geomechanics and Engineering · July 2021

DOI: 10.12989/gae.2021.26.2.117

CITATIONS READS

27 2,588

1 author:

Gianluca Bella
Politecnico di Torino
65 PUBLICATIONS 160 CITATIONS

SEE PROFILE

All content following this page was uploaded by Gianluca Bella on 03 August 2021.

The user has requested enhancement of the downloaded file.

Geomechanics and Engineering, Vol. 26, No. 2 (2021) 117-132
https://doi.org/10.12989/gae.2021.26.2.117 117

Water retention behaviour of tailings in unsaturated conditions

Gianluca Bella1,2
1
Politecnico di Torino, C.so Duca degli Abruzzi 24, Torino, Italy
2
Pini Swiss Engineers, Via Besso 7, Lugano, Switzerland

(Received April 20, 2020, Revised June 18, 2021, Accepted June 26, 2021)

Abstract. Tailing dams are complex geotechnical systems comprising of an embankment and a basin containing the waste
products from the mining processes. These structures are characterized by a wide surface exposed to the atmosphere whose
interaction governs the position of the phreatic surface within the basin. A detailed knowledge of the hydro-mechanical
properties of the tailings is fundamental to reliably assess the stability of the tailing dams. While most of the previous studies
have dealt with the response of tailings in saturated conditions, this research provides an extension of the hydraulic behaviour in
unsaturated and nearly saturated state of tailings collected after the failure of the Stava basins. The hydraulic behaviour in
unsaturated conditions was investigated by means of tests where the suction was imposed and the water content was monitored
(axis translation technique and vapour equilibrium technique), and tests where the water content was imposed and the suction
was measured with psychrometer (dew point method). To account for the in-situ heterogeneity of tailings, the dependency of the
water retention relationship on the grain size distribution, the preparation technique and on the initial density / void ratio was
studied. Denser tailings showed a higher water retention behaviour than that given in looser specimens. Similarly, the increase of
the fine content was demonstrated to improve the water retention capability. As for standard soils, also statically compacted
Stava tailings reveal lower retention capability than the slurry samples, thus confirming the importance of the preparation
method in determining the hydro-mechanical response of such soils.
Keywords: tailing dams; soil water retention curve; unsaturated soil; void ratio; fine content

1. Introduction amount of surface runoff, infiltration rate, consolidation

process, or capillarity rise (Fig. 2(a)). These factors, caused
Tailings are the waste products resulting from the by interactions with the atmosphere, govern the position of
chemical and the mechanical separation processes of the phreatic surface within the basin, leading to variations in
mineral extraction. Comprising of mixtures of crushed the height of the unsaturated zone above the water table.
rocks and processing fluids, their particle size ranges from The soil below the phreatic level is usually saturated and the
sand to clay/silt size. Surface storage facilities for the pore water pressure is positive, whilst above the phreatic
disposal of wet tailings are known as tailings dams (Fig. surface there is a capillary fringe consisting of saturated
1(a)) and represent some of the largest and complex zone with a negative pore water pressure. An unsaturated
geotechnical structures in the world. These structures region with a negative pore water pressure and a degree of
usually comprise of a raised earth embankment constructed saturation less than 100% lies above the saturated capillary
in stages over the life of the impoundment, and a basin fringe (Fig. 2(c)).
aimed to store billions of tons of tailing materials. The importance of understanding the behaviour of soils
According to the classical hydraulic segregation profile, a in the unsaturated portion of the ground is recognized in
higher permeability zone made up of coarse particles would engineering practice. However, a gap still exists between
exist near the embankment/discharge point, with a lower engineering practice and research in unsaturated soil
permeability zone (pond) made up of fine particles furthest mechanics, mainly because it requires sophisticated
from this point, and between them there is an intermediate apparatus and time-consuming experimental methodologies.
permeability zone (beach). Despite this, the unsaturated conditions of tailings are
The high rate of historical and recent collapses poses fundamental tools to understand the climate effects on
serious environmental threats when the mine closes or is tailing dams. These include the consequences of rainfall,
still active, causing human loss of lives (Fig. 1(a)), snowmelt, or poor management on the stability of an
environmental and economic damages (Fig. 1(b), Fig. 1(c)). embankment of which these are considered some of the
Tailing basins are heterogeneous systems subjected to main worldwide causes of failure (Rico et al. 2008,
many external factors affecting their operation (Alonso and Zanardin et al. 2009).
Gens 2006, Bhanbhro 2014, Zuoan et al. 2019), i.e., Hydraulic and mechanical behaviour in unsaturated
conditions is coupled (Alonso 1990, Tarantino and
Tombolato 2005, Lu and Likos 2006, Estabragh and Javadi
Corresponding author, Ph.D. 2014, Arab et al. 2015, Nicotera et al. 2015, Kim and Jeong
E-mail: [email protected] 2017, Horn et al. 2018, Deng et al. 2020). Therefore, the

Copyright © 2021 Techno-Press, Ltd.

http://www.techno-press.org/?journal=gae&subpage=7 ISSN: 2005-307X (Print), 2092-6219 (Online)
 Gianluca Bella

(b)

(a) (c)
Fig. 1 (a) Numbers of victims associated with the major tailing failures (Santamarina et al. 2019); Examples of
historical and recent failures: (b) Stava tailing dams - Italy - 1985 (Lucchi, 2020) and (c) Feijão tailing dam - Brazil -
2019 (Robertson et al. 2019).

Fig. 2 (a) Representative cross section of a tailing dam, (b) Soil properties with respect to deposition location and (c)
Schematic representation (I and II) of the unsaturated conditions and (III) pore water pressure profile in tailing basins,
(modified from Zanardin et al. 2009)

accuracy in predicting the hydraulic soil response is Gan et al. 1988, Fredlund et al. 1996, Oberg and Sallfors
essential for determining the mechanical soil behaviour in 1997, Khalili and Khabbaz 1998, Rassam and Williams
terms of shear strength. Indeed, the pore-fluid affects the 1999, Rassam and Cook, 2002, Toll and Ong 2003,
shear strength via its suction but also its degree of saturation Tekinsoy et al. 2004, Xu 2004). Other strength criteria were
so, the same suction but with different degrees of saturation proposed within different constitutive models for
will lead to different values of shear strength. A relevant unsaturated soils (Alonso et al. 1990, Sun et al. 2000,
example of stability analysis of tailing dams based on this Sheng et al. 2008, Alonso et al. 2010). Therefore, the
concept was given by the recent collapse of the Feijão degree of saturation that can be obtained from the water
tailing dam (Brazil), where high rainfalls and high retention curve (WRC) for a certain suction has been
infiltrations were proved to decrease the suctions by approx. embedded into strength criteria for unsaturated soils
50 kPa in the unsaturated zone above the water table, which (Fredlund et al. 1996, Vanapalli et al. 1996, Oberg and
in turn reduced the shear strength by 10 kPa to 15 kPa, Sallfors 1997, Toll 1990, Toll and Ong 2003). Although
providing one of the cause that supposedly led to the some other shear strength criteria for unsaturated soils do
instability of the embankment (Robertson et al. 2019). not include the degree of saturation explicitly (Khalili and
Over the years several shear strength criteria for Khabbaz, 1998; Sheng et al. 2008), the key parameters
unsaturated soils have been proposed in the literature. Some involved in these criteria, such as the air-entry value, are
of them were based on regression analyses of experimental embedded in the WRC. Knowledge of the WRC can
data from either triaxial or direct tests (Fredlund et al. 1978, provide other applications with the estimation of the
 Water retention behaviour of tailings in unsaturated conditions

(a) (b)
Fig. 3 Application of unsaturated soil mechanics in the stability analysis of Stava tailing dams: (a) relationship between B-
Skempton parameter and Sr and (b) variation of the residual strength with the degree of saturation (Bella 2017)

(a) (b)
Fig. 4 Stava basins. (a) Geographical localization and (b) sketch: A upper basin, B lower basin, (1) on-dam cyclone for the
extraction of the coarser sand components from the tailing wastes, (2) upper beach, (3) upper pond, (4) drainage service,
(5) emergency drainage, (6) service road, (7) lower beach, (8) lower pond, (9) drainage from the upper dam, (10)
caretaker’s house (modified from Luino and De Graff 2012)

hydraulic conductivity of unsaturated tailings, generally In summary, an in-depth geotechnical characterization

expressed by the combination of the saturated hydraulic of the hydro-mechanical behaviour of tailings represents a
conductivity and the relative hydraulic conductivity, the fundamental tool for reliably assessing the stability of
latter depends upon the degree of saturation (van tailing dams and therefore in improving their design. For
Genuchten, 1980). Finally, recent applications of this reason, this study is aimed at providing a preliminary
unsaturated soil mechanics regarding tailing dams stability insight into the range of hydraulic behaviour predicted in
can be found in Bella (2017) with reference to the tailing materials, by taking into account the variation in the
occurrence of the static liquefaction which is considered density, the particle sorting which could reasonably be
one of the main causes of failure. In undrained conditions, expected in situ, and preparation technique. The soil density
when the degree of saturation slightly decreased below the and grading considered in this study can be related to
unit due to changes in suction, the excess pore pressure of plausible locations and depths within a basin through a geo-
the unsaturated Stava silty tailings was proved to be much statistic approach (Esposito et al. 2010). This study
lower than that in saturated conditions (Fig. 3(a)), leading to provides the basis for defining the spatial distribution of
an increase in the soil’s shear strength and so strongly hydraulic variables from a better understanding of the
influencing the occurrence of the static liquefaction (Fig. unsaturated behaviour of tailing dams, which, when
3(b)). modelled numerically (Tarantino and Mountassir 2013,
 Gianluca Bella

Tarantino and Di Donna 2019) will result in a more accurate

evaluation of the tailing dam safety.

2. Testing material

The hydraulic behaviour of the Stava tailings was

investigated by means of laboratory tests carried out at the
soil laboratory of Politecnico di Torino (Italy) on
unsaturated samples. Soil tests are performed on the tailing
samples manually collected in 2005 at ground level from
the lower portion of the Stava upper dam, which remained
in place after the collapse (Fig. 1(b)). The two storage
facilities were raised with the upstream technique, built one
above the other on a natural slope near village of Stava,
Italy (Fig. 4). In July 1985, the lower dam failed due to
overtopping caused by the collapse of the upper one and Fig. 5 Grain size distribution of the two fractions of the
triggering a flow-slide of more than 200’000 m3. The causes Stava tailings (modified from Carrera et al. 2011)
of the failure are complex and have been investigated in
detail by others (e.g., Chandler and Tosatti, 1995), but the of clay size material. Results are shown in Table 1, while
unconsolidated state of the tailings, a poor drainage and the Table 2 gives the results of the X-ray diffraction analysis.
high phreatic level were supposed to be the main factors Both fractions were predominantly made up of quartz, with
that caused the static liquefaction, leading the failure of the a significant amount of calcite and fluorite, especially
Stava tailings dam. within the silty fraction (Bella 2017).

3.1 Sample preparation

3. Material characterization
The water retention tests were carried out on silt
Tailings used in the current research are composed of specimens with a different initial void ratio, different
two grain sizes: a sand fraction made up of particles preparation techniques and on mixtures of silt and sand.
retained by sieve n°200, and a silt fraction passing through The axis translation technique was applied in a suction-
a sieve n°200 (Fig. 5). This choice was due to the fact that controlled oedometer on samples having initial size of 50
Stava tailings not passing through 0.074 mm sieve (mesh mm diameter and height of 20 mm. The compacted
sieve n°200) was part of the embankment, while finer specimens (Fig. 6(a)) were prepared mixing a required
particles seemed to be deposited inside the basin. The D10 of amount of dry soil with a certain amount of demineralized,
the sand fraction is equal to 0.08 mm, D50 and D90 are 0.20 de-aired water to obtain the target dry density and water
mm and 0.35 mm, respectively (Carrera et al. 2011). Liquid content. The mixture was then put in a cylindrical mould,
(wL) and plastic limits (wP) show that the silty fraction is and statically compacted by gradually applying an axial
characterized by medium-low plasticity with 8% in weight force with a filter paper on the top and bottom of the
sample. A control axial displacement Wykeham Farrance
Table 1 Geotechnical index properties of the silty and the loading frame was used to compact the sample until the
sandy fraction: liquid limit (wL), plastic limit (wP), plasticity desired volume was reached under controlled water content.
index (PI), specific gravity (GS) and average absolute It is worth noting that during the assembling of the mould
permeability (k). (Carrera et al. 2011) prior the static compaction phase, two filter papers were
placed on the top and bottom face of the sample and grease
Material wL (%) wP (%) PI (%) GS (-) k (m/s) was applied around the mould wherever the water could
Sandy fraction - - - 2.721 9.5∙10-6 drain out preventing water leakage. Furthermore,
Silty fraction 27.4 18.0 9.4 2.828 10-7 deceleration of applying the axial load helped to reduce the
localized excess pore water pressure generated at sample
Table 2 Percentages of minerals obtained from the X-ray ends. The statically compacted specimens were finally
diffraction analyses (Bella 2017) tested by means of suction-controlled oedometer cell by
Sandy applying the axis translation technique (Table 3 and Fig.
Silty fraction
Mineral fraction 6(a)):
(% mass)
(% mass) - pure silt with initial void ratio equal to 0.70 (labelled
Magnetite - 11.0 as SILT-0.70)
Quartz 36.4 70.0 - pure silt with initial void ratio equal to 0.60 (labelled
Feldspars K traces traces
as SILT-0.60)
- pure silt with initial void ratio equal to 0.50 (labelled
Feldspars Na - traces
as SILT-0.50)
Calcite 11.9 6.0 - mixture 70% sand - 30% silt with initial void ratio
Fluorite 47.8 2.0 equal to 0.70 (labelled as 7030-0.70)
 Water retention behaviour of tailings in unsaturated conditions

(a) (b)
Fig. 6 (a) Statically-compacted sample inside the oedometric ring and (b) Remolded samples inside the suction-controlled
oedometer cell

(a) (b)
Fig. 7 (a) Suction controlled oedometer and (b) Axial stress, air pressure and water pressure during time for sample SILT-
0.70 (modified from Bella 2017)

Table 3 Initial state of specimens: void ratio (e0), water

content (w0), degree of saturation (Sr0), dry weight (γd), 20 mm height) was prepared by mixing a required amount
initial net stress (pnet), percentage of sand/silt and of dry soil with a certain amount of demineralized, de-aired
preparation method (RE = remolded, ST = statically water corresponding to the limit liquid, so it assumed the
compacted) consistence of a slurry. Then, it was placed inside the
e0 w0 Sr0 γd pnet Sand Silt Preparatio suction-controlled oedometric cell and tested (axis
Sample
(-) (%) (%) (kN/m3) (kPa) (%) (%) n method translation technique) without any static compaction (Fig.
SLURRY-0.78 0.78 27.4 100.0 15.9 10.0 0 100 RE 6(b) and Table 3):
SILT-0.70 0.70 17.3 70.0 16.6 100.0 0 100 ST - pure slurry silt with initial void ratio equal to 0.78
(labelled as SLURRY-0.78).
SILT-0.60 0.60 15.0 70.0 17.6 100.0 0 100 ST
SILT-0.50 0.50 15.9 90.0 18.9 100.0 0 100 ST 3.2 Experimental techniques
7030-0.70 0.70 17.8 70.0 16.2 100.0 70 30 ST
3070-0.60 0.60 17.2 80.0 17.4 100.0 30 70 ST The hydraulic characterization of Stava tailing samples
was based both on techniques where the suction was
imposed and the water content was measured (axis
- mixture 30% sand - 70% silt with initial void ratio translation and vapour equilibrium technique), and on tests
equal to 0.60 (labelled as 3070-0.60). where the water content was imposed and the suction was
Specimens tested by applying the vapour equilibrium measured with the dew point method.
technique and dew point measurements were prepared in The relationship between the water content θ, or degree
the same way and had a diameter of 20 mm and a height of of saturation Sr and matric suction s was obtained for s ≤
10 mm. 500 kPa in a suction controlled oedometer apparatus (Fig.
The remolded specimen (initial size: 50 mm diameter, 7(a)), by applying the axis translation technique. The axial
 Gianluca Bella

(a) (b)
Fig. 8 (a) Schematic view of the equipments used to apply the vapour equilibrium technique and (b) Schematic view of the
chilled-mirror psychrometer for dew-point technique (Bulut et al. 2002)

displacement of the oedometer samples were measured by are usually realized into a sealed glass container with a
using a linear variable differential transducer (LVDT) porous support above the chemical solution, which
connected to the top plate. The axial stress (maximum suspends the soil specimens in the vapour environment
allowed 1.6 MPa) was applied by pressurized air acting on (Fig. 8(a)). The total suction within the specimen is
the loading plate. A high Air Entry Value ceramic porous obtained due to a net water exchange between the liquid
stone (AEV = 500 kPa) was located at the bottom of the phase within the unsaturated specimen and the vapour phase
sample, in order to apply the pore water pressure, while the in the headspace of a box, until equilibrium between the
pore air pressure was applied at the top through a coarse two phases was reached. During the transient phase, the soil
porous stone. An air trap was installed in the water system sample was weighted at regular intervals until it reached a
to avoid the creation of air bubbles close the porous stone at constant weight, meaning no further variations in water
the bottom of the sample. A water chamber was installed in content and suction. In order to minimize any changes in
the air system to maintain a high relative humidity of air in moisture due to room temperature fluctuations, during the
order to reduce the evaporation of water vapors. Finally, a experimentation period the temperature and relative
water volume-indicator (capacity 50 cc) was connected to humidity RH of the laboratory were accurately monitored
the bottom AEV porous stone and allowing to quantify the and maintained at a constant (RH = 38.5 %, T = 21°C ±
water volume exchanged by the soil sample during drying 0.5°C). It is important to note that the vapour equilibrium
and wetting processes. technique controls the total suction (ψ), that is defined as
After the unsaturated static-compacted specimen was the sum of two components, matric suction (s) and osmotic
placed inside the cell, suction values and net stress were suction (π). Matric suction is usually assumed to be the
imposed by independent control of the water pressure uw, main component of the total suction in non-plastic soils
air pressure ua and axial stress σa, according to the axis with a pure pore fluid, while osmotic suction is appreciable
translation technique. The drying phase was realized by in high plastic clays or in cases where the pore fluid
maintaining a constant air pressure and decreasing the water contains dissolved salts. According to Carrera et al. (2011),
pressure step-by-step, until equilibrium, in terms of water Stava silt has been defined as an inorganic silt of low
volume exchange, was obtained (Fig. 7(b)). On the opposite plasticity and compressibility, indicating that osmotic
side, the wetting phase was obtained by increasing the water suction represents a small amount of the total suction, so
pressure, until equilibrium conditions were reached. At the that ψ ≈ s. Total suctions imposed on tailing specimens
end of the drying step, the air pressure had increased by 50 were 4.0 MPa, 11.4 MPa, 57 MPa and 70 MPa by K 2SO4,
kPa and the water pressure had decreased by 50 kPa to KNO3 , NaNO2 and Ca(NO3)24H20 saturated solutions.
obtain 400 kPa of suction. Consequently, the axial stress σa Finally, a chilled mirror psychrometer WP4 was used to
was increased by 50 kPa to maintain a constant net stress measure the total suction along the main drying branch of
during the entire test. It is also worth noting that, to soil specimens 20 mm initial diameter, 10 mm height. The
accurately obtain entire WRC for a certain soil sample with dew point technique (range: 0-300 MPa; accuracy: ±0.05
the axis translation technique, the entire test can take up to MPa from 0 to 5 MPa, 1% from 5 to 300 MPa) was applied,
55 days (Fig. 7(b)). together with the vapour equilibrium technique, to obtain
Suction values up to 70 MPa were imposed at soil the water content-suction relationship for high suction
specimens 20 mm diameter and 10 mm height through the values. Immediately after the preparation, samples were
vapour equilibrium technique (VET). The vapour exposed at regular intervals to the laboratory atmosphere for
equilibrium technique was implemented by controlling the a couple of hours. At each step, their total suction was
measured with WP4 (equilibrium time: 15-30 minutes)
relative humidity to impose a desired total suction. In the
together with the gravimetric water content until reaching
current study, this method was used at providing additional
equilibrium when no variation in water content and total
data for the suction levels out of the axis translation suction was recorded. The measured total suction ranged
technique range. The desired total suction was imposed by from 1 MPa to 150 MPa. The measurement of the total
creating a controlled relative humidity condition using the suction by using the dew point technique consists of
osmotic potential of chemical solutions, as proposed by equilibrating the vapour phase of the water in the air space
Romero (1999), and Tang and Cui (2005). These conditions
 Water retention behaviour of tailings in unsaturated conditions

above the sample in a closed chamber with the liquid phase water retention curves of the Stava tailings, the water
of the water within the unsaturated soil specimen (Fig. retention was assumed to be due to the capillary
8(b)). A cooling system is used to cool a mirror until dew mechanism. The pore size density PSD of the soil samples
forms. An optical sensor is used to detect the dew formed was evaluated by defining a proper range of suction levels
on the mirror, while a thermocouple attached to the mirror s, so its corresponding pore radius r was computed by
measures the dew point temperature (± 0.2°C). In the inversion of Washburn-Laplace equation (Eqs. 4a-4b).
meantime, the temperature of the unsaturated sample is Pores with different sizes were assumed to be cylindrical,
measured with an infrared thermometer. A fan is also used contact angle α between water and pore’s wall equal to 0°
to circulate the air in the sensing chamber and speed up and water tension Ts equal to 72 mN/m at temperature of
vapour equilibrium. Both the soil sample temperature and 20°C:
the dew point are employed to obtain the relative humidity
above the soil sample and so its total suction according to 2 ∙ 𝑇𝑠 ∙ cos⁡(𝛼)
(4a)
𝑟=
the psychrometric law. 𝑠

𝑒(1 − 𝑆𝑟𝑒𝑓𝑓 )
4. Required variables 𝛥[ ]
𝐺𝑠 (4b)
𝑃𝑆𝐷 =
The main variables required for the estimation of the 𝛥log⁡(𝑟)
water retention curve are summarized in the following. where Sreff is the effective degree of saturation evaluated at
4.1 Evaluation of the water retention curves each suction level by using the van Genuchten model (Eq.
(1)) and Gs is the specific weight (Table 1).
The experimental data were fitted using the water
retention model proposed by van Genuchten (1980):
5. Experimental results and interpretation
1
𝑆𝑟 = (1)
(1 + (𝛼𝑠)𝑛 )𝑚 In Section 5.1 and 5.2 the influence of the initial void
where α is associated to the air entry value AEV of the soil ratio, preparation technique and fine content on the
(additional air pressure, with respect to the water pressure, hydraulic behaviour of samples was investigated by means
required for air to fill the biggest soil pores during the of water retention tests. Then, the effect of initial density on
drying process), while dimensionless, dependent parameters the PSD of Stava silt samples was analyzed in Section 5.3.
n and m = 1-(1/n) determine the shape of the curve. It is worth noting that the pore size was not given by
To account for variation of the void ratio on the water mercury intrusion porosimetry tests, but it has been
retention behaviour, the model proposed by Gallipoli et al. obtained from the theoretical water retention curves.
(2003) was applied, where dimensionless parameters n, m,
ψ and Φ (kPa-1) are soil constants, while ν is the specific 5.1 Evidences of the influence of the initial density on
volume (𝜈 = 1 + 𝑒): the water retention behaviour
1 Fig. 9 shows the experimental results (SILT-0.70)
𝑆𝑟 = (2)
(1 + [𝛷(𝜈 − 1)𝜓 𝑠]𝑛 )𝑚 obtained from the three adopted methods: axis translation,
Gallipoli parameters ψ and Φ are calibrated from the dew-point and vapour equilibrium technique. The black
power law 𝛼 = 𝛷(𝜈 − 1)𝜓 by fitting the points in (AEV - triangles are the experimental points obtained from the axis
e) plane. Those points refer from at least two samples translation technique; the unfilled black squares show the
having the same grain size distribution but different void results obtained from the dew-point method, and the
ratios, with the air entry value (AEV = 1/α) given by unfilled circles show the results from the vapour
previous best fitting of van Genuchten model. Then, the n equilibrium technique. The black diamond marked as “si?”
parameter is calibrated, so m = 1-(1/n) is obtained. The represents the assumed initial state of the specimen in terms
detailed procedure is given in Bella (2017). of the degree of saturation and suction value. Indeed, the
Defining the residual degree of saturation Srres as the initial degree of saturation has been imposed (70%) while
state in which the water phase within the soil pores is the initial suction was unknown. Due to the drying response
discontinuous and water can be moved through the soil only of the sample at the imposed suction 50 kPa at the first step
as water vapour (Srres = 0.05), the effective degree of of the water retention test, the initial suction of the sample
saturation Sreff was obtained: was less than that suction value. Furthermore, main drying
and wetting curves define a domain, so the initial suction
𝑆𝑟 − 𝑆𝑟 𝑟𝑒𝑠 ranges between 4 kPa and 30 kPa. Additionally, as shown in
𝑆𝑟𝑒𝑓𝑓 = (3)
1 − 𝑆𝑟 𝑟𝑒𝑠 the same Fig. 9, suction levels obtained by the vapour
where Srres is the irreducible (residual) degree of saturation, equilibrium technique were partially superimposed on those
assumed to be the one determined at the highest suction obtained by the dew point technique (redundancy).
imposed with the vapour equilibrium technique. Experimental points were fitted using the simplified van
Genuchten model to obtain the main drying (solid line) and
4.2 Evaluation of the pore size density main wetting branch (dotted line). The drying curve was
fitted using experimental points obtained during the drying
In order to give a microstructural interpretation of each
 Gianluca Bella

Fig. 9 Sample SILT-0.70: experimental points and fitting curve for the three applied methods

Fig. 10 Void ratio during the water retention tests

Fig. 11 Sample SILT-0.60: experimental points, fitting curve for the three applied methods

process (axis translation technique, WP4C and VET). It is using points at suction values up to 11.4 MPa
important to note that, since the specimen in the suction (VET/WP4C).
controlled oedometer was dried only up to a maximum During the water retention tests, void ratio was shown to
suction s = 400 kPa, and then wetted, the wetting points in be constant at different suction steps for the slurry sample
the range 50 kPa < s < 300 kPa do not belong to the main (SLURRY-0.78) and for the statically compacted samples
wetting but to a scanning curve. For this reason, the main (SILT-0.70 and SILT-0.60) as shown in Fig. 10.
wetting branch was fitted by assuming just the last two Experimental results obtained for sample SILT-0.60
points (s ≤ 50 kPa) belong to the main wetting curve and by from the three performed methods are shown in Fig. 11.
 Water retention behaviour of tailings in unsaturated conditions

Table 4 List of the van Genuchten parameters α, n, m and air entry values
Main drying branch Main drying branch AEV
Sample
α (kPa-1) n (-) m (-) α (kPa-1) n (-) m (-) (kPa)
SLURRY-0.78 0.030 1.630 0.390 0.280 1.480 0.320 33.3
SILT-0.70 0.042 1.620 0.380 0.320 1.500 0.330 23.8
SILT-0.60 0.019 1.650 0.390 0.100 1.520 0.340 52.6
SILT-0.50 0.009 1.670 0.400 0.065 1.520 0.340 111.1
7030-0.70 0.120 1.550 0.350 0.600 1.700 0.410 8.3
3070-0.60 0.021 1.680 0.400 0.150 1.540 0.350 47.2

Fig. 12 Influence of void ratio on the water retention behaviour

Again, the black triangles are the results obtained from the model is related to the inverse of the air entry value, and
axis translation technique; the unfilled squares are the this was confirmed by the specimen with the lower void
experimental points obtained from the dew-point technique ratio (e0 = 0.40) being associated with the smallest value of
and the unfilled circles are the results obtained from the α. This means that the soil is expected to remain saturated
vapour equilibrium technique. The experimental points even for a relatively high suction which, although still being
were then fitted by the simplified van Genuchten model to a modest value, is still significantly higher that found for
obtain the main drying and main wetting branches. the same material with a lower initial density.
The Van Genuchten fitting parameters for the silt A summary of technical literature data depicts the main
samples and mixtures, both for the main drying and main drying branches of the water retention curves for different
wetting branch together with the air entry values, are given types of soils (both tailings and more standard materials)
in Table 4. under different compaction states was compiled and is
The influence of void the ratio on the water retention shown in Fig. 13 which highlights the similarities and
curve of silty samples with the same grain size distribution, differences with the results from the current research.
but different initial density, was studied using the Gallipoli Fig. 13(a) shows the best fitting water retention curves
model (Fig. 12). The calibration of the Gallipoli parameters, obtained from nickel tailing samples, from the Pedro Sotto
based on the experimental results of SILT-0.70, SILT-0.60, Alba tailings impoundment (Cuba) with different
allowed the water retention behavior for a wide range of compaction states (ei = 1.00, ei = 0.75, ei = 0.50). Fig. 13b
void ratios, i.e. SILT-0.50 and SILT-0.40 to be obtained. It shows the WRC for different initial void ratios (ei = 1.18, ei
can be observed that both the main drying and wetting = 1.13, ei = 1.08) on Pearly clay specimens for a suction
branches are strongly influenced by initial density of the range between 10 kPa and 200 kPa. Figure 10c shows the
samples. Different void ratios also imply a different pore WRC obtained from a compacted clayey silt at three
size distribution, and this affects the hydraulic behaviour in different initial void ratios (ei = 1.00, ei = 0.75, ei = 0.50),
terms of the water retention curve. A decrease in the void while Fig. 13(d) gives the main drying water retention
ratio shifts the water retention curve to a higher suction curves at different compaction states (ei = 1.46, ei = 0.93, ei
because as the size of the pores decreases, a higher suction = 0.64, ei = 0.59) for Boom clay specimens. These data
is required to empty the pores. For this reason, the water from literature, for both for tailing materials (Fig. 13(a))
retention curve of SILT-0.40, SILT-0.50 and SILT-0.60 rest and more standard soils (Fig. 13(b)-13(d)) show a good
above the curve of SILT-0.70. The α parameter of Gallipoli agreement with results of the water retention tests obtained
 Gianluca Bella

(a) (b)

(c) (d)
Fig. 13 Water retention curves for different type of soils and initial void ratio e i: (a) Nickel tailings modified from Zanardin
et al. (2009), (b) Pearl clay, modified from Sun et al. (2007), (c) Clayey silt, modified from Vanapalli et al. (1999) and (d)
Boom clay, modified from Romero (1999).

Table 5 List of the Gallipoli retention parameters n, m, Φ

in this research on the Stava tailings at different compaction and ψ for silty samples
states (Fig. 12). Within the suction range investigated for
Main drying branch Main wetting branch
both these cases, the water retention curves were shifted
upward with decreasing the initial void ratio of compacted n (-) m (-) Φ (-) ψ (-) n (-) m (-) Φ (-) ψ (-)
soil. As shown in this research and the literature data for 1.670 0.400 0.281 5.327 1.500 0.330 5.405 7.810
standard soils (Fig. 13), except for the very loose Boom
clay specimen ei = 1.46, variations in void ratio variations
do not affect the shape of the WRC significantly. This is Genuchten model and Gallipoli model, and a relevant
confirmed by n, m parameters for SILT-0.70, SILT-0.60 and difference was observed. The van Genuchten model (solid
SILT-0.50, for the main drying and wetting branches (Table and dotted grey lines) predicts the water retention behaviour
4). of SLURRY-0.78 well. Otherwise, if the same experimental
points are fitted using the Gallipoli model (solid and dotted
5.2 Evidences of the influence of the preparation
black lines) based on the same calibrated parameters used
techniques and fine content on the water retention
for statically compacted SILT-0.70, SILT-0.60 (Fig. 14(a)),
behaviour
a significant difference is appreciated. This means that, also
for tailings, the water retention behaviour is strongly
Table 5 provides the Gallipoli fitting parameters for silty
affected by the soil structure that, in turns, is influenced by
samples, both for the main drying and wetting branches.
the preparation technique. As shown in Fig. 14a, the slurry
A preliminary evaluation of the influence of the
sample shows a greater water retention capability than the
preparation technique was investigated by best fitting the
static compacted sample, and this probably depends on the
experimental points of SLURRY-0.78. If the experimental
different pore size distributions of the samples obtained
points of SLURRY-0.78 were fitted to both the van
 Water retention behaviour of tailings in unsaturated conditions

Fig. 14 Influence of the preparation technique on the water retention behaviour: (a) Stava tailings tested in this research and
(b) standard soils (modified from Fredlund and Pham 2006)

Fig. 15 Influence of the fine content on the water retention behaviour

Fig. 16 Influence of the void ratio and preparation method with air entry value: (a) Stava tailings tested in the current
research and (b) standard soils (modified from Huang et al. 1998)

from the two preparation methods. The outcomes obtained agreement with the literature results of water retention tests
in this research on the Stava tailings are found to be in good carried out on more standard soils. According to Fredlund
 Gianluca Bella

Fig. 17 Influence of the void ratio and preparation technique on the pore size density

Fig. 18 Influence of the fine content on the pore size density (e0 = 0.70)

Fig. 19 Influence of the fine content on the pore size density (e0 = 0.60)

and Pham (2006), and Pham and Fredlund (2008), the soil- unsaturated conditions, the mechanical response in terms of
water characteristic curves differ, depending on the applied liquefaction strength was also proved to be influenced by
historical stresses and hence on the initial state of the the preparation technique, for both tailings (i.e., Chu et al.
specimen (Fig. 14(b)). It is worth noting that, due to the 2003, Bella 2017), and standard soils (i.e., Vaid and
coupling between hydraulic and mechanical behaviour in Sivathayalan 2000).
 Water retention behaviour of tailings in unsaturated conditions

The influence of the fine content on the water retention “standard soils” has been quantified in a number of
behaviour was also studied. Soil-water characteristic curves relatively recent publications (Huang et al. 1998; Vanapalli
of tailing samples prepared with the same initial density, but et al. 1999; Ng and Pang, 2000; Cabarkapa and Cuccovillo,
with different fine content were compared as shown in Fig. 2006) showing good agreement with the experimental
15a (SILT-0.70 with 7030-0.70) and Fig. 15b (SILT-0.60 results obtained in this research on the Stava tailings. Some
with 3070-0.60). In both the pure silt (SILT-0-70 and SILT- literature data depicting the evolution of the air entry value
0.60) and mixtures samples (3070-60 and 7030-0.70), the with void ratio for different type of soils, loam and sand
main drying and wetting branches of the WRC were under different compaction states was shown in Fig. 16b.
affected by changes in grain size distribution. At an increase As for Stava tailings, it was noticed that, for a given
of fine content, correspond a movement of the water material, the AEV increases with decreasing voids, and at a
retention curve to high suction values. Indeed, when given void ratio the air entry value was more than ten times
increasing the fine content, the pores between the sandy higher for the loams than that for the sands.
particles get filled with the small silty particles resulting an
overall lower permeability, and so a higher retention 5.3 Evidences of the influence of the initial
behaviour. A comparison of the air entry values of the two density, preparation techniques and fine content on
samples having the same initial void ratio (e0 = 0.70) but the pore size density
high difference in fine content show that the AEV of SILT-
0.70 (23.8 kPa) is higher than that of sample 7030-0.70 (8.3 A preliminary microstructural interpretation of the
kPa). This means that, despite the same initial compaction different water retention curves of the Stava tailings at
state, the tailing mixture can be desaturated by applying a different initial densities and grading was performed by
smaller suction than that required for the pure silt as, the assuming that the water retention was due to the capillary
latter one is expected to have small pores. Similar mechanism described by the Washburn-Laplace equation.
considerations were done by comparing the water retention The effect of void ratio on the pore size density was
behaviour of sample SILT-0.60 with mixture 3070-0.60. investigated by comparing the pore size distributions of silt
Due to the small difference in fine content, both the main samples with different initial density, i.e. SILT-0.70, SILT-
drying and wetting branches were only slightly affected by 0.60 and SILT-0.50 (Fig. 17). The pore size density peak of
changes in the grain size distribution. The values of the α SILT-0.70 occurs at a larger pore radius than those of SILT-
parameter are 0.019 kPa-1 for the pure silt and 0.021 kPa-1 0.60 and SILT-0.50. This is consistent with their estimated
for the silt-sand mixture. This corresponds to the AEV of air entry values given in Fig 13. Moreover, the pore size
52.6 kPa and 47.2 kPa respectively, indicating that the pure density curve of SILT-0.70 was wider than those of samples
silt fraction can be desaturated by imposing a higher suction SILT-0.60 and SILT-0.50. This means that looser sample
than the sand-silt mixture. Indeed, at the same void ratio the SILT-0.70 has a larger range of pores size than SILT-0.60
sand-silt mixture is expected to have fewer but larger pores and SILT-0.50, so that SILT-0.70 sample has a lower water
than the silt material. retention capability than the other two samples. These
The influence of the initial density on the water considerations were not identified for SLURRY-0.78,
retention behaviour was also observed by the decrease in suggesting the influence that the preparation technique has
the air entry value with the increase in void ratio for the on the water retention behaviour.
statically compacted silt samples (Fig. 16(a), dotted line). The effect of grain size distribution on the pore size
Indeed, a lower pressure (suction) is required to fill the density was observed from the comparison of the tailing
bigger empty pores of the looser samples with air. The samples having the same initial density but different sand-
importance of the pore size distribution on the water silt percentages, i.e. SILT-0.70 and 7030-0.70 (Fig. 18).
retention behaviour was also investigated by comparing the Due to the higher fraction of sand in sample 7030-0.70 than
air entry value of the silt sample SILT-0.70 with the air in SILT-0.70, it is expected that the mixture was more
entry value of the mixture 7030-70 both with an initial void permeable than the latter. The air entry value of 7030-0.70
ratio e0 = 0.70 (Fig. 16, black triangle at e0 = 0.70 and was lower than that of SILT-0.70 (Tab. 4) and this implies
unfilled diamond). Again, a lower pressure is required to fill that the lower the air entry value is, the higher the
the bigger empty pores of the mixtures with air. Similar permeability is. As expected, the highly permeable sample
considerations were done by analyzing the air entry value of 7030-0.70 having many pores, also of large size, required a
the silt sample SILT-0.60 with the air entry value of the lower pressure to fill them. As shown in Fig. 18 the pore
mixture 3070-60, both with an initial void ratio e0 = 0.60 radius peak of 7030-0.70 occurred at a higher pore radius
(Fig. 16(a): black triangle and unfilled square). Moreover, than that of SILT-0.70 which is consistent with their
the influence of the preparation technique on the hydraulic respective AEVs.
behaviour was observed by comparing the dotted line Similar considerations were accomplished by comparing
another pair of samples having the same initial density but
showing the trend of the air entry value of statically
different grain size distribution, i.e. SILT-0.60 with 3070-
compacted samples (Fig. 16a, dotted line) with the air entry
0.70 (Fig. 19). Again, an increase in the fine content shifted
value of SLURRY-0.78 (Fig. 16a, empty circle). Due to the the pore size density curve toward the smaller pore radius.
different preparation technique, the soil structure was
different and the air entry value of SLURRY-0.78 was not
aligned with those of the statically compacted samples. 6. Conclusions
The dependency of the air entry value on the porosity of
Tailing dams, because of their large spatial extension,
 Gianluca Bella

long operational-life, raising methods, depositional water retention tests were successfully compared with the
techniques, that have resulted in a high number of results described in literature on standard soils.
unacceptable failure (both socially and environmentally • A microstructural interpretation of the water retention
detrimental), need detailed investigations during the design capability of the Stava tailings was shown in terms of pore
phases and regular monitoring during the operational size distribution by assuming that the water retention was
phases. Due to the hydro-mechanical coupling in due to the capillary mechanism described by the Washburn-
unsaturated soils, the hydraulic response of tailings affects Laplace equation.
the stability of tailing dams in terms of resistance Following the results obtained in this study, further
properties. Indeed, shear strength changes with suction, that research could cover a larger domain of void ratios as well
in turn is related to the degree of saturation by means of the as other mixtures made up of sandy and silty tailings to dif-
WRC. So, an accurate prediction of the hydraulic behaviour ferent percentages to simulate the heterogeneity of in situ
within the unsaturated layers above the phreatic surface has tailings. Finally, it was observed that the real environment is
a significant effect on the mechanical response, particularly more complex than the conditions reproduced during the
regarding the consequences of the stability of tailing dams. experimentation. Experimentation was adopted to simplify
This research has provided, in unsaturated conditions, a the experimental program, as at the site the tailings are
preliminary characterization of tailing materials collected mixed with water which contains dissolved salts, heavy
from the Stava basins, outlining the effects of void ratio, metals, contaminants, and residual chemicals from the
particle sorting and preparation technique on their hydraulic mineralogical processes. To simulate the real environment
behaviour. Several experimental techniques were adopted to more accurately, more retention tests could be carried out
characterise the dependency of the water retention by using processing water or by considering the aging
behaviour within a wide suction level as follows: effects related to the particle rearrangements that result in
• The water retention behaviour of Stava silty samples the macro-interlocking or the cementation of the particles.
with the same grain size distribution but different initial
density was studied by comparing their WRC. This
condition has a practical relevance in modelling the hydro- Acknowledgments
mechanic behaviour of tailings because it could represent
different depths of a unique vertical cross section at the The Author wishes to thank Prof. G. Musso, M. Barbero
same distance from the embankment. The increase in void and F. Barpi (Politecnico di Torino) for their valuable
ratio was proved to shift the water retention curve to a inputs, suggestions and support while analyzing the
lower suction range because the size of the pores increases. experimental data. The Author also wishes to acknowledge
A lower suction is therefore required to empty the pores, Dr. A. Azizi, Dr. O. Pallara and Mr. G. Bianchi for their
leading to a lower water retention capability in loose help during the laboratory tests presented in this work.
unsaturated tailing than denser ones.
• The hydraulic response of unsaturated tailing mixtures
with the same initial void ratio but different fine content References
was investigated by analysing their water retention curves.
This condition has a practical relevance since it could Alonso, E.E., Gens, A. and Josa, A. (1990), “A constitutive model
represent the same depth for several vertical cross sections for partially saturated soils”, Géotechnique, 40(3), 405-430.
located at the different distance from the embankment. Also https://doi.org/10.1680/geot.1990.40.3.405.
Alonso, E.E. and Gens, A. (2006), “Aznalcollar dam failure. Part1:
for Stava tailings, an increase of the fine content was Field observations and material properties”, Geotechnique,
proved to move the water retention to higher suction values 56(3), 165-183. https://doi.org/10.1680/geot.2006.56.3.165.
as the big pores between sandy particles get filled with Alonso, E.E., Pereira, J.M., Vaunat, J. and Olivella, S. (2010), “A
small silty particles, leading to a higher water retention microstructurally based effective stress for unsaturated soils”,
capability of pure silt samples than that of silt-sand Geotechnique”, 60(12), 913-925.
mixtures. http://doi.org/10.1680/geot.8.P.002.
• The influence of the preparation technique was Arab, A., Belkhatir M. and Sadek, M. (2015), “Saturation effect on
preliminary studied by comparing the WRC of a slurry behaviour of sandy soil under monotonic and cyclic loading: A
sample with those that obtained from a statically compacted laboratory investigation”, Geotech. Geol. Eng., 34(1), 347-358.
https://doi.org/10.1007/s10706-015-9949-6.
specimen. The slurry sample shows a greater water Bella, G. (2017), “Hydro-mechanical behaviour of tailings in
retention capability than that of the same sample created unsaturated conditions”, Ph.D. Dissertation, Politecnico di
from static compaction method due to the different pore Torino, Torino, Italy.
size distribution of the two preparation techniques. Bhanbhro, R. (2014), “Mechanical properties of tailings - basic
• The use of several types of non-standard, sophisticated description of a tailings material from Sweden”, Ph.D.
equipment, the time-consuming experimental Dissertation, Luleå University of Technology, Luleå, Sweden.
methodologies, as well as the use of the axis translation, the Bulut, R., Hineidi, S.M. and Bailey, B. (2002), “Suction
vapour equilibrium method and the chilled-mirror measurements - filter paper and chilled mirror psychrometer”,
psychrometer, were shown to be the best way to investigate Proceedings of the Texas Section American Society of Civil
Engineers, Waco, Texas, U.S.A., October.
the hydraulic behaviour of unsaturated tailings in order to Cabarkapa, Z. and Cuccovillo, T. (2006), “Automated triaxial
perform reliable stability analysis. apparatus for testing unsaturated soils”, Geotechnical Testing
• The experimental results on Stava tailings provided by Journal, 29(1), 21-9. https://doi.org/10.1520/GTJ12310.
 Water retention behaviour of tailings in unsaturated conditions

Carrera, A., Coop, M. and Lancellotta, R. (2011), “Influence of Luino, F. and De Graff, J.V. (2012), “The Stava mudflow of 19
grading on the mechanical behaviour of Stava tailings”, July 1985 (Northern Italy): A disaster that effective regulation
Geotechnique, 61(11), 935-946. might have prevented”, Nat. Hazards Earth Syst. Sci., 12, 1029-
https://doi.org/10.1680/geot.9.P.009. 1044. https://doi.org/10.5194/nhess-12-1029-2012.
Chandler, R.J. and Tosatti, G. (1995), “The Stava tailings dams Ng, C.W.W. and Pang, Y.W. (2000), “Influence of stress state on
failure, Italy, July 1985”, Proc. Inst. Civ. Eng. Geotech. Eng., soil-water characteristics and slope stability”, J. Geotech.
113(2), 67-79. https://doi.org/10.1680/igeng.1995.27586. Geoenviron. Eng., 126(2), 157-66.
Chu J., Leong, W.K. and Loke, W.L. (2003), “Discussion of https://doi.org/10.1061/(ASCE)1090-0241(2000)126:2(157).
“defining an appropriate steady state line for Marriespruit gold Nicotera, M.V., Papa, R. and Urciuoli, G., (2015), “The hydro-
tailings”, Can. Geotech. J., 40(2), 484-486. mechanical behaviour of unsaturated pyroclastic soils: An
https://doi.org/10.1139/t02-118. experimental investigation”, Eng. Geol., 195, 70-84.
Deng, D., Wen, S., Lu, K. and Li, L. (2020), “Calculation model https://doi.org/10.1016/j.enggeo.2015.05.023.
for the shear strength of unsaturated soil under nonlinear Öberg A.L. and Sallfors, G. (1997), “Determination of shear
strength theory”, Geomech. Eng., 21(3), 247-258. strength parameters of unsaturated silts and sands based on the
https://doi.org/10.12989/gae.2020.21.3.247. water retention curve”, Geotech. Test. J., 20(1), 40-48.
Esposito, T., Assis, A. and Giovannini, M. (2010), “Influence of https://doi.org/10.1520/GTJ11419J.
the variability of geotechnical parameters on the liquefaction Pham H.Q. and Fredlund, D.G. (2008), “Equations for the entire
potential of tailing dams”, Int. J. Surface Min. Reclam. Environ., soil-water characteristic curve of a volume change soil”, Can.
16(4), 304-316. https://doi.org/10.1076/ijsm.16.4.304.8639. Geotech. J., 45(4), 443-453. https://doi.org/10.1139/T07-117.
Estabragh, A.R. and Javadi, A.A. (2014), “Roscoe and Hvorslev Rassam, D.W. and Cook, F.J. (2002), “Predicting the shear
Surfaces for Unsaturated Silty Soil”, Int. J. Geomech., 14(2), strength envelope of unsaturated soil”, Geotech. Test. J., 25(2),
230-238. 215-220. https://doi.org/10.1520/GTJ11365J.
https://doi.org/10.1061/(ASCE)GM.1943-5622.0000306. Rassam, D.W. and Williams, D.J. (1999), “A relationship
Fredlund, D.G., Morgenstern, N.R., and Widger, R.A. (1978), describing the shear strength of unsaturated soils”, Can.
“The shear strength of unsaturated soils”, Can. Geotech. J., Geotech. J., 36(2), 363-368. https://doi/org/10.1139/t98-102.
15(3), 313-321. https://doi.org/10.1139/t78-029. Rico, M., Benito, G., Salgueiro, A.R., Dıez-Herrero, A. and
Fredlund, D.G., Xing, A., Fredlund, M.D. and Barbour, S.L. Pereira, H.G. (2008), “Reported tailings dam failures - A review
(1996), “The relationship of the unsaturated soil shear strength of the European incidents in the worldwide context”, J. Hazard.
to the soil-water characteristic curve”, Can. Geotech. J., 33(3), Mater., 152(2), 846-852.
440-448. https://doi.org/10.1139/t96-065. https://doi.org/10.1016/j.jhazmat.2007.07.050.
Fredlund, D.G., and Pham, Q.H. (2006), “A volume-mass Robertson, P.K., de Melo, L., Williams D.J. and Wilson, G.W
constitutive model for unsaturated soils in terms of two (2019), Report of the Expert Panel on the Technical Causes of
independent stress state variables”, Proceedings of the 4th the Failure of Feijão Dam I.
International Conference on Unsaturated Soil. Carefree, Romero, E. (1999), “Thermo hydro-mechanical behaviour of
Arizona, U.S.A., April. unsaturated Boom Clay: An experimental study”, Ph.D.
Gallipoli D., Wheeler, S.J. and Karstunen M. (2003), “Modelling Dissertation, Universidad Politecnica de Catalunya, Barcelona,
the variation of degree of saturation in a deformable unsaturated Spain.
soil”, Geotechnique, 53(1), 105-112. Santamarina, J.C., Torres-Cruz, L.A. and Bachus, R.C. (2019),
https://doi.org/10.1680/geot.2003.53.1.105. “Why coal ash and tailings dam disasters occur”, Science,
Gan, J.K.M., Fredlund, D.G., and Rahardjo, H. (1988), 364(6440), 526-528. https://doi.org/10.1126/science.aax1927.
“Determination of the shear strength parameters of an Sheng, D., Fredlund, D.G., and Gens, A. (2008), “A new
unsaturated soil using the direct shear test”, Can. Geotech. J., modelling approach for unsaturated soils using independent
25(3), 500-510. https://doi/org/10.1139/t88-055. stress variables”, Can. Geotech. J., 45(4), 511-534.
Horn-Da, L., Chien-Chih, W. and Xu-Hui, W. (2018), “A http://doi.org/10.1139/T07-112.
simplified method to estimate the total cohesion Sun, D.A., Matsuoka, H., Yao, Y.P. and Ichihara, W. (2000), “An
of unsaturated soil using an UC test”, Geomech. Eng., 16(6), elasto-plastic model for unsaturated soil in three-dimensional
599-608. https://doi.org/10.12989/gae.2018.16.6.599. stresses”, Soils Found., 40(3), 17-28.
Huang, S.Y., Barbour, S.L., Fredlund, D.G. (1998), “Development http://doi.org/10.3208/sandf.40.3_17.
and verification of a coefficient of permeability function for a Sun, D., Sheng, D. and Xu, Y. (2007), “Collapse behaviour of
deformable unsaturated soil”, Can. Geotech. J., 35(3), 411-425. unsaturated compacted soil with different initial densities”, Can.
https://doi.org/10.1139/t98-010. Geotech. J., 44(6), 673-686. https://doi.org/10.1139/t07-023.
Khalili, N. and Khabbaz, M.H. (1998), “A unique relationship for Tang, A.M. and Cui, Y.J. (2005), “Controlling suction by the
χ for the determination of the shear strength of unsaturated vapour equilibrium technique at different temperatures and its
soils”, Géotechnique, 48(5), 681-687. application in determining the water retention properties of
https://doi.org/10.1680/geot.1998.48.5.681. MX80 clay”, Can. Geotech. J., 42(1), 287-296.
Kim, Y., and Jeong, S. (2017), “Modeling of shallow landslides in https://doi.org/10.1139/t04-082.
an unsaturated soil slope using a coupled model”, Geomech. Tarantino, A. and Tombolato, S. (2005), “Coupling of hydraulic
Eng., 13(2),353-370. and mechanical behaviour in unsaturated compacted clay”,
https://doi.org/10.12989/gae.2017.13.2.353. Géotechnique, 55(4), 307-317.
Lu, N. and Likos, W.J. (2006), “Suction stress characteristic curve https://doi.org/10.1680/geot.2005.55.4.307.
for unsaturated soil”, J. Geotech. Geoenviron. Eng., 132(2), Tarantino, A. and Mountassir, G. (2013), “Making unsaturated soil
131-142. mechanics accessible for engineers: Preliminary hydraulic–
https://doi.org/10.1061/(ASCE)1090-0241(2006)132:2(131). mechanical characterisation & stability assessment”, Eng. Geol.,
Lucchi, G. (2021), “Tailing Dams: lezioni dal passato e dal 165, 89-104. https://doi.org/10.1016/j.enggeo.2013.05.025.
presente. Stava: Cause e responsabilità[Tailing Dams: lessons Tarantino, A. and Di Donna, A. (2019), “Mechanics of unsaturated
learnt from the past and present: Causes and responsabilities"]”, soils: simple approaches for routine engineering practice”,
Online Conference, GEAM. Rivista Italiana Geotec., 53(4), 5-46.
 Gianluca Bella

https://doi.org/10.19199/2019.4.0557-1405.005.
Tekinsoy, M.A., Kayadelen, C., Keskin, M.S. and Soylemez, M.
(2004), “An equation for predicting shear strength envelope
with respect to matric suction”, Comput. Geotech., 31(7), 589-
593. https://doi.org/10.1016/j.compgeo.2004.08.001.
Toll, D.G. (1990), “A framework for unsaturated soil behaviour”,
Géotechnique, 40(1), 31-44.
https://doi.org/10.1680/geot.1990.40.1.31.
Toll, D.G. and Ong, B.H. (2003), “Critical-state parameters for an
unsaturated residual sandy clay”, Geotechnique, 53(1), 93-103.
https://doi.org/10.1680/geot.2003.53.1.93.
Vaid, Y.P. and Sivathayalan, S. (2000), “Fundamental factors
affecting liquefaction susceptibility of sands”, Can. Geotech. J.,
37(3), 592-606. https://doi.org/10.1139/t00-040.
Vanapalli, S.K., Fredlund, D.G., Pufahl D.E. and Clifton, A.W.
(1996), “Model for the prediction of shear strength with respect
to soil suction”, Can. Geotech. J., 33(3), 379-392.
https://doi.org/10.1139/t96-060.
Vanapalli, S.K., Fredlund D. and Pufahl, D.E. (1999), “The
influence of soil structure and stress history on the soil-water
characteristics of a compacted till”, Geotechnique, 49(2), 143-
59. https://doi.org/10.1680/geot.1999.49.2.143.
van Genuchten, M.T. (1980), “A closed-form equation for
predicting the hydraulic conductivity of unsaturated soil”, Soil
Sci. Soc. Amer. J., 44, 892-898.
https://doi.org/10.2136/sssaj1980.03615995004400050002x.
Xu, Y.F. (2004), “Fractal approach to unsaturated shear strength”,
J. Geotech. Geoenviron. Eng., 130(3), 264-273.
http://doi.org/10.1061/(ASCE)1090-0241(2004)130:3(264).
Zanardin, M.T., Oldecop, L.A., Rodriguez, R. and Zabala, F.
(2009), “The role of capillary water in the stability of tailing
dams”, Eng. Geol., 105(1-2), 108-118.
https://doi.org/10.1016/j.enggeo.2008.12.003.
Zuoan, W., Yulong C., Guangzhi, Y., Yonghao, Y. and Weimin, S.
(2019), “An alternative upstream method for the
Zhelamuqing tailings impoundment construction of a Copper
Mine in China”, Geomech. Eng., 19(5), 383-392.
https://doi.org/10.12989/gae.2019.19.5.383.

View publication stats