In-situ and laboratory site specific geotechnical characterization

of hard rock gold mine tailings

Michaël D.Bonin & Marielle Limoges Shaigetz

Golder-WSP, Montréal, Qc, Canada
Ali El Takch
Golder-WSP, Mississauga, On, Canada
Dennis E. Becker
Golder-WSP, Calgary, Ab, Canada
Edouard Masengo
Agnico Eagle Mines Ltd, Toronto, On, Canada

ABSTRACT: The geotechnical characterization of hard rock gold mine tailings was undertaken
to support the geotechnical stability review of an existing Tailings Storage Facility. This paper
focuses on the in-situ and laboratory geotechnical testing programs that were performed to char-
acterize the in-situ state and the stress-strain behaviour of the gold mine tailings.
The laboratory testing program included direct simple shear tests (DSS), isotropic and anisotropic
drained and undrained triaxial compression testing (CID, CIU and CAU), and a Consolidated
Lateral Extension triaxial test (CLEDtx).
Numerical simulations using the NorSand constitutive model and then the NorSand widget were
carried out to determine the material-specific coefficients for both the drained and undrained con-
ditions applied in the CPT interpretation. Electronic field vane tests completed the in-situ testing
program.

1 INTRODUCTION

Due to unfortunate recent tailings dam failures, the geotechnical/mining community was re-
minded of the complex geotechnical behaviour of mine tailings. The increasing numbers of tech-
nical references on topics addressing specific aspects of the stress-strain behaviour of mine tail-
ings reflects the level of attention attributed to the geotechnical behaviour of man-made processed
minerals byproducts. Within the last few years, large mining companies have set internal stand-
ards for good tailings practices and worldwide initiatives (GISTM 2020, ICOLD, in preparation)
were undertaken to promote the application of standard of practice regarding tailings manage-
ment. This was also reflected on the technical side through a few recent state-of-the art papers
that deserve geotechnical engineers and practitioners’ attention (Been, 2016, Schnaid, 2021,
Fourie et al. 2022) to stay connected and up-to-date with the rapid evolution of the mining ge-
otechnical practice. Interestingly, one of the key aspects of contractive mine tailings, i.e. their
potential for undergoing a brittle collapse and liquefaction, is not new, but is now looked at with
a new sight impacted by high economic and social consequences of tailings dam failures. This is
likely the duty of care of related practitioners to make sure that the collective memory and the
technical literacy remain alive and relevant within the universities, the consultants, and the mining
companies.
It is in this important technical context that a geotechnical characterization of hard rock gold
mine tailings was undertaken to support the performance-based stability review of an existing
Tailings Storage facility (TSF) built over a glacio-lacustrine clay deposit in Québec, Canada. This
paper focuses on the advanced laboratory and in-situ geotechnical testing programs that were
performed with a focus on the in-situ state and the stress-strain behaviour. The advanced labora-
tory testing aimed at reproducing different effective stress paths and drainage conditions. Given
the sandy/silty nature of mine tailings, intact or undisturbed specimens could not be practically
collected in engineering practice, and the current practice typically relies on the estimation of
shear strength based on results from laboratory stress-strain testing on reconstituted specimens
complemented by in-situ testing results. As the tailings inferred in-situ state was found contrac-
tive, peak and liquefied undrained shear strength were given more attention.
This case study papers presents the objectives, the methodology and the results of these labor-
atory and in-situ programs conducted in 2020 and 2021. This paper also provides a summary of
some technical challenges faced during the process, such as partial drainage of tailings during
Cone Penetration Test (CPT) and electronic Vane Shear Testing (eVST), tamping-induced densi-
fication and difficulties in preparing fully contractive reconstituted specimens.

2 LABORATORY TESTING PROGRAM

2.1 Material tested
The TSF where the tailings were collected and characterized was commissioned in 2017 when
tailings started being discharged at 64% Solids (S, by mass). Approximately twenty meters of
tailings were discharged over the first 2 years with the deposition points located on the starter
berm. The tailings deposition is made from the free draining (pervious) dikes, through a series of
discharge points, and in a fashion that the pond is maintained far away from the dikes. In the early
stages of the TSF development, the supernatant water was pumped out from the pond that was
forming away from the dikes. An operational spillway was commissioned, which allowed surface
water to drain by gravity. In this context, sub-aerial tailings deposition has mainly occurred, ex-
cept for some localized areas within the pond that was forming in the first two years of its opera-
tion.
The tailings samples were collected using split spoon sampler in a dedicated borehole drilled
from a waste rock pad to allow drilling on the tailings beach. Samples were collected between 0.3
m to 15 m depth from the tailings surface.
The tailings samples were then shipped to a geotechnical laboratory where a composite mix
was prepared for subsequent index and advanced testing. Table 1 provides a summary of the index
tests results. Atterberg limits results showed that the material is classified as non-plastic fines (ML
as per the Unified Soil Classification System (USCS).

Table 1. Index Properties of the Gold Mine Tailings Used in this Study
Material D10 (mm) D50 (mm) D90 (mm) FC (%) Gs USCS
Gold Mine Tailings 0.002 0.025 0.12 76-78 2.78 ML
FC: Fines Content | D10: diameter at 10% finer | D50: diameter at 50% finer | D90: diameter at 90% finer | Gs: Specific Gravity |
USCS: Unified Soil Classification System

2.2 Specimen Preparation

All specimens for advanced laboratory testing were reconstituted by moist tamping using the un-
der compaction method (Ladd, 1978). Moist tamping was used in DSS and CSL triaxial testing to
control the initial void ratio of the reconstituted specimens as suggested by Reid et al. (2022), and
to promote a more brittle fabric resulting in a contractive stress-strain behaviour essential for a
more accurate Critical State Line (CSL) definition. Although, it does not mimic the depositional
environment of slurry tailings, moist tamping typically results in higher brittleness than other re-
constitution methods for an equivalent state parameter. As shown later in this paper, the estimated
state parameters are characteristic of a contractive stress-strain behaviour. For DSS and CLEDtx
testing, the objective was to obtain an initial void ratio and state parameter representative of the
in-situ values, and moist tamping promotes a better control to reach the targeted values. Prepara-
tion water contents were between 6.0% and 9.4% and approximately 9.5%, respectively for the
triaxial and the DSS testing.

2.3 Triaxial Testing

The critical state parameters of the tailings were determined following the methodology presented
in Jefferies & Been (2016). Triaxial specimens tested to determine the CSL were reconstituted
using the moist tamping method to different densities and consolidated isotopically or
anisotropically to a range of confining (consolidation) stresses (50 kPa to 400 kPa). Once consol-
idated, the specimens were sheared under both drained and undrained conditions. The critical state
of a soil is the state at which the soil continues to deform at a constant effective stress and at
constant void ratio (i.e., further shearing does not change the soil volume). After reaching the
critical state, the water contents was determined by first freezing the samples (Jefferies & Been
2016). The critical state locus was defined by equation 1.

ec=Γ-λ10∙log10(p') (1)

Where ec is the critical void ratio, Γ is the void ratio at a reference mean effective stress of 1
kPa, p’ is the mean effective stress, and λ 10 is the slope of the CSL in a semi-logarithm space [e
vs. log (p’)]. Table 2 shows a summary of the triaxial testing results.
Figure 1a shows the effective stress path, while Figure 1b shows the deviator stress versus axial
strain. CIU-01 to CIU-03 results show a typical effective stress path for hard rock mine tailings
reconstituted to a loose state and then isotropically consolidated with a first contractive stress path
followed by a phase transformation point and then dilation up to the end of the test. This type of
results is not ideal for the determination of the CSL as strain localization might occur during
dilation with the global void ratio not being representative of the local void ratio. For this reason,
CAU-04 and CAU-05 included anisotropic consolidation (Kc=σ3/σ1 of about 0.6) to promote brit-
tleness.
Figure 2 shows the CSL defined for the studied tailings. The CSL was determined using five
out of the nine drained and undrained triaxial compression tests (excluding CLEDtx). The triaxial
tests CIU-01, CIU-02, CID-06 and CID-07 were not used to define the CSL because they did not
reach critical state as shown in Figure 1 (continuous increase in deviator stress and/or decrease in
shear induced pore pressure for undrained testing or a decrease in the volumetric strain at 30 %
axial strain for drained testing). Therefore, the final tested state of these four tests does not lie on
the interpreted CSL shown in Figure 2. A comparison with Smith et al. (2019), Torres-Cruz &
Santamarina (2020), Jefferies & Been (2016) and Reid (2015) show that the estimated CSLs of
the studied tailings generally are generally consistent with CSLs reported in the technical literature
as shown in Table 3. The CSL is used for the CPT interpretation as described in Section 3.0. The
resulting CSL is associated to a Normalized Standard Error (NSE) of 1.48% and 0.35%, respec-
tively for NSEλ10 and NSEΓ100 which means a relatively precise CSL acccording to the database
of Torres-Cruz & Santamarina (2020).

Table 2. Summary of Triaxial Testing Program

Test ID ei ec ef ᴪ pi’ p f’ Mtc ηmax ηIL ηPT EoT
(-) (-) (-) (-) (kPa) (kPa) (-) (-) (-) (-) cond.
CIU-01 1.13 0.64 0.64 -0.03 70 111 1.20 1.20 0.48 0.65 Δu/Δεa<0
CIU-02 1.13 0.59 0.59 -0.04 350 580 1.39 1.39 - 1.06 Δu/Δεa<0
CIU-03 1.11 0.66 0.66 0.02 200 121 1.20 1.20 0.66 1.03 c.s.
CAU-04 1.33 0.74 0.74 0.07 65 5 2.20 4.60 0.78 - c.s.
CAU-05 1.34 0.70 0.70 0.05 134 34 1.54 1.54 0.87 - c.s.
CID-06 1.11 0.70 0.65 0.03 70 129 1.36 1.43 - - Δεv/Δεa<0
CID-07 1.14 0.65 0.61 0.005 200 367 1.37 1.42 - - Δεv/Δεa<0
CID-08 0.66 0.63 0.63 -0.04 80 147 1.37 1.65 - - c.s.
CID-09 0.66 0.61 0.61 -0.03 300 552 1.37 1.45 - - c.s.
CLEDtx 0.77 0.71 0.67 0.05 120 116 1.37 1.39 - - c.s.
ei: As prepared void ratio | ec: after consolidation void ratio | ef: Final void ratio | ᴪ : State parameter after consolidation, before shearing
Mtc: q/p’ at critical state for triaxial compression | ηmax : maximum principal stress ratio | ηIL : principal stress ratio at instability limit
ηPT : principal stress ratio at phase transformation | CIU/D: Consolidated Isotropically, sheared undrained/drained | CAU: Consolidated
anisotropically, sheared undrained | CLEDtx: Consolidated Lateral Extension triaxial | EoT cond. :End of test condition (e.g., cs:
critical state, Δu/Δεa ≠ 0, Δεv/Δεa≠ 0)
 a) b)

Figure 1. a) Effective stress path and b) deviator stress versus axial strain in the triaxial testing for the de-
termination of the CSL

0.90
Curved
a = 0.860 Semi-Log

b = 0.192 λ10= 0.066

0.80
Void Ratio, e: (-)

c = 0.166 Γ= 0.796

0.70

0.60
CSL Semi-log (modified) CIU-01 CIU-02
CIU-03 CAU-04 CAU-05
CID-06 CID-07 CID-08
CID-09 CSL Curved Final state
CLEDtx
0.50
1 10 100 1000
Mean Effective Stress, p': kPa

Figure 2. Critical State Line

For the Consolidated Lateral Extension Drained triaxial testing (CLEDtx), the specimen was
first consolidated isotropically in stages to the desired confining (consolidation) stress of 120 kPa.
Then the effective stress path follows a strain-controlled drained shearing at a axial strain rate of
0.5% per hour. The drained shear was stopped when the stress ratio (q/p’) reached the desired
stress ratio. In this case, the targeted stress ratio was selected to be representative of an in-situ at-
rest earth coefficient (K0) of 0.7. After reaching the desired stress ratio, the effective confining
stress (σ’3) was decreased at a rate of about 3 kPa per hour under a stress-controlled condition
until the axial and volumetric strain rates increase rapidly and reached maximum capacity of the
equipment. The reduction in confining stress was carried out under drained conditions.
It is important to note that the CLEDtx specimen reached the critical state as it is forced to the
failure line under stress-controlled conditions. The objective of the test is to see whether the tested
specimen would undergo a brittle collapse (liquefaction) when approaching the instability line,
when there may be a sudden increase in axial strain after only small displacement (strain) in the
specimen. The identification of instability is often made qualitatively, but Ahmed et al. (2022)
observes that cohesionless soils will generally undergo a significant increase in both axial and
volumetric strain at the onset of instability, with a change in deviator stress often becoming neg-
ative (∆q < 0). Jefferies et al. (2019) reports that some contractive soils, when tested in the triaxial
apparatus with drainage valve opened, may not undergo static liquefaction (such as observed in
an undrained test) when they reach the instability line depending on the soil characteristics and
the loading rate.

Table 3. Comparison of the Estimated CSL and Parameters with the Technical Literature
Test ID λ10 Γ
(-) (-)
This study 0.066 0.80
Smith et al. (2019)-based on FC 0.14 1.00
Torres-Cruz and Santamarina (2020)-based on emin 0.07-0.08 0.93-0.98
Jefferies & Been (2016)-based on FC 0.05-0.15 -
Reid (2015)-from Fr 0.05-0.13
FC : Fines Content | emin determined from the standard Proctor curve | Fr: Normalized friction ratio from the adjacent CPT

Examples of similar laboratory unloading effective stress paths conducted on tailings include
Fundao tailings specimens (Morgenstern et al. 2016) and Cadia tailings (Jefferies et al. 2019). For
Fundao tailings a sudden increase in axial strain after less than 0.5% axial strain was observed at
the onset of instability, while the Cadia tailings showed that a sudden increase in axial strain
occurred between 3% to 4% axial strain. Detailed results including the effective stress paths and
stress-strain behaviour observed in the CLEDtx testing are provided in Figures 3 and 4. Table 2
presents a summary of the results.
The initial achieved state parameter of 0.05 of the tested CLEDtx specimen was slightly higher
than the interpreted in-situ characteristic state parameter (0.03), but this value remains representa-
tive of loose contractive tailings as targeted. Once the effective stress path crosses the estimated
instability line, the axial strain underwent a more significant increase as shown in Figure 4. The
instability limit (ηIL) of the anisotropic triaxial compression test (Kc=σ3/σ1 of about 0.6) was as-
sessed applicable for the CLEDtx as well as suggested by Fotovvat et al. (2022) for constant
drained shear testing on gold mine tailings. Once the effective stress path of the tested specimen
reached the CSL, it underwent a steady increase in axial strain at constant deviator stress. In terms
of axial strain, the tested specimen shows a small increase during the drained shearing phase fol-
lowed by a progressive increase during the Lateral Extension Drained (LED) phase. The increase
in axial strain is consistent with the contractive state of the tested specimen and appears to show
characteristics of a ductile soil in the tested conditions. In terms of volumetric strain (not shown
herein), the tested specimen showed a continuous decrease in volumetric strain (contraction) until
its effective stress path reached the CSL at q/p’=1.37 and then showed a small dilation of 0.25%.
At this point, the state parameter of the specimen had undergone a decrease from 0.05 at the end
of the isotropic consolidation phase to 0.01 at the end of the LED phase.
No triggering of static liquefaction in an undrained manner is observed when the effective stress
path of the tested CLEDtx specimen reached the instability line. Although the tailings tested spec-
imen reached the CSL under stress-controlled conditions, it did not undergo a brittle collapse
(liquefaction) for an initial state parameter characteristic of the in-situ conditions, because a
drained shearing condition prevailed beyond the instability line. It is inferred that, under similar
conditions of lateral extension in the foundation clay, if the field rate of loading associated with
tailings deposition and rate of rise results in drained shear conditions in the in-situ tailings, static
liquefaction of the tailings would not take place. In order for static liquefaction to occur in the
tailings deposit, the rate of loading would need to induce sufficient excess pore water pressure
(i.e., no longer a fully drained condition), potentially leading to a brittle collapse (liquefaction) as
the effective stress path approaches the instability line.
 300 25
Drained Shear
250 Lateral Extension
Mtc 20
ηIL

Axial Strain, εa (%)

q = σ1-σ3 (kPa) 200
15
150

10
100

5
50

0 0
0 50 100 150 200 0 50 100 150 200
p' = (σ'1 + 2 σ'3) / 3 (kPa) p' = (σ'1 + 2 σ'3) ) / 3 (kPa)
Figure 3. Stress path in CLEDtx with Mtc=1.37 and Figure 4. Axial strain versus mean effective stress
ηIL = 0.82 from anisotropic triaxial compression in CLEDtx

2.4 Direct Simple Shear Testing

Reconstituted tailings samples were tested using a constant-volume DSS device. The testing de-
vice was an Electromechanical Dynamic Cyclic Simple Shear (EMDCSS) manufactured by GDS.
The tailings specimens were enclosed in a rubber membrane and radially confined by steel stacked
rings. Each moist-tamped specimen was initially loaded under a low vertical seating stress, while
simultaneously being flushed with deaired water to remove matrix suction and its effect on the
effective vertical stress. The specimens were prepared at initial densities to reach targeted post-
consolidation state parameter at the desired applied vertical effective stress (σ'vc = 120 kPa). “Un-
drained” shearing of each sample was simulated by maintaining a constant-volume condition in
the DSS tests, where the specimen diameter was constrained by the stack of steel rings and the
vertical deformation of the specimen was restricted by holding the position of the upper loading
actuator so that there was no change in specimen height. The strain-controlled shearing load was
applied by the horizontal actuator to a limiting shear strain of 20% to avoid the added shear
strength provided by the confinement method (Baxter et al. 2010). As drainage is allowed during
testing, the change in total vertical stress (σv) required to maintain a constant height was taken as
the equivalent pore water pressure that would have developed in a truly undrained test on a satu-
rated sample.
Table 4 presents a summary of the DSS testing conditions and the results from the completed
five (5) DSS tests. Figure 5 shows the normalized undrained shear strength ratio versus shear
strain. Figure 6 provides the normalized effective stress path for the five DSS tests.
The undrained shear strength ratio at phase transformation was used for the two densest speci-
mens (DSS-2 and DSS-2 redo) for two reasons. First, it was estimated that the densest specimens
show what Reid et al. (2022b) call tamping-induced stresses during the specimen preparation.
This effect could cause an overestimation of the undrained shear strength for the corresponding
state parameter. Second, the use of the undrained shear strength at the phase transformation re-
sulted in a more consistent and defensible relationship when compared to the peak undrained
strengths inferred from the other methods (NorSand and eVST-see next section). Overall, the
obtained results on the loosest specimens showed a reasonable range of peak undrained shear
strength ratio against state parameter, based on Golder’s experience and the technical literature
such as Riveros & Sadrekarimi (2021). For a characteristic state parameter of 0.03, the laboratory-
based DSS relationship yielded a Su(peak)/σ’v of 0.27.

Table 4. Summary of DSS Testing Results.

__________________________________________________________________________________________________________
Parameter DSS-1 DSS-2 DSS-2-r DSS-3 DSS-4
__________________________________________________________________________________________________________
Vertical Effective Confining Stress, σ’vc (kPa) 120 120 120 120 120
Initial Dry Unit Weight (kN/m³) 11.7 15.8 15.8 13.5 15.3
Initial Void ratio 1.33 0.72 0.72 1.02 0.78
Post-Consolidation Saturation (%) 96.4 91.9 88.8 84.8 96.9
Post-Consolidation Void Ratio 0.75 0.69 0.70 0.81 0.76
Estimated State Parameter, ψ 1 0.09 0.02 0.04 0.15 0.10
Peak Undrained Shear Strength (kPa) 16.7 54.3 37.3 16.4 27.3
Peak Undrained Shear Strength Ratio 0.14 0.45 0.31 0.14 0.23
PT Undrained Shear Strength (kPa) N/A 33.6 30.0 N/A N/A
PT Undrained Shear Strength ratio N/A 0.28 0.25 N/A N/A
Post-Peak Undrained Shear Strength (kPa) 9.9 51.2 N/A 10.2 20.5
Post-Peak Undrained Shear Strength Ratio 0.08 0.43 N/A 0.09 0.17
__________________________________________________________________________________________________________
1
: CSL shown in Figure 2 and a K0 of 0.7 | PT : Phase Transformation | N/A: Not applicable

0.50
DSS-1 0.50
DSS-1
0.45 DSS-2 0.45 DSS-2
0.40 DSS-2 (redo)
0.40 DSS-2 (redo)
DSS-3
0.35 0.35
DSS-3
DSS-4 DSS-4
0.30 0.30
Peak/PT
τ/σ'vc (-)

Peak/PT

τ/σ'vc (-)
0.25 0.25
0.20 0.20
0.15 0.15
0.10 0.10
0.05 0.05
0.00 0.00
0 5 10 15 20 25 0.0 0.4 0.8 1.2
Shear Strain (%) σ'v / σ'v0 (-)
Figure 5. Normalized Undrained Shear Strength Ra- Figure 6. Normalized Undrained Shear Strength
tio versus shear strain Ratio versus normalized effective stress

3 IN-SITU FIELD VANE TESTING

Electronic Shear Vane Testing (eVST) was carried out in two boreholes next to a CPT sounding
to assess the in-situ peak and remolded vane shear strength of tailings. Two different rotation rates
were used namely: 0.1 deg/s (as per the ASTM 2573) in the first borehole, and 0.4 deg/s in the
second borehole. The tests were conducted using two different vane sizes (60x120 mm and
75x150 mm) to assess the blade size effect on the measurements. The procedure for the testing
was the following: a 360-degree rotation was first recorded at the selected rate (0.1 or 0.4 deg/s)
to measure the peak vane shear strength, followed by 10 rotations completed at a fast rate, which
was followed by a second recording until a residual vane shear strength could be interpreted. A
total of 18 tests were conducted up to 360 degrees, while 9 tests were rotated further to measure
the remoulded vane shear strength.
The peak vane shear strength (Su(peak)) was obtained by subtracting the friction rod shear stress
from the peak shear stress. The resulting average peak vane shear strength is 54 kPa, with an
average ratio of Su(peak)/σ’v of 0.29. The tests conducted at a faster rotation rate of 0.4 deg/s show
a lower average ratio of Su(peak)/σ’v of 0.21. The results differ depending on the blade size and the
rotation rate utilized. The use of vane shear testing data is routinely used in industry practice, and
currently is an appropriate method to calibrate the cone factor (Nkt) for CPT interpretation of un-
drained shear strength in fine-grained cohesive soils. The values of Nkt was taken as 17.
The remoulded shear stress (Su(res.)) was obtained after 10 rotations by subtracting the friction
rod shear stress after 10 rotations. The Su(res.) estimated from the tests varied from 3 kPa to 13 kPa
with a mean of 9.5 kPa, resulting in a Su(res.)/σ’v varying from 0.03 to 0.09, with a mean value of
0.06. All the remoulded tests were conducted at a rotation rate of 0.1 deg/s. Figure 8 shows the
peak and residual vane shear strength (see yellow squares on the corresponding profile).
The vane shear test is recognized to provide reliable estimates of undrained shear strength in
cohesive soils; however, ensuring that the tests are performed in an undrained condition in non-
cohesive silts, such as tailings, is a challenge. As reported by Fourie et al. (2022), there are nu-
merous cases where the use of vane shear tests in tailings was applied without sufficient recogni-
tion and consideration of partial drainage. In this case, the measured vane shear strength is that of
a partially drained condition – not undrained. Undrained shear strength of tailings, which often
consist of a large proportion of silt sized particles that can frequently exhibit significant layering
resulting in hydraulic anisotropy, are particularly affected by drainage during vane shear testing
(Reid 2016). As peak undrained shear strength is important to the design of tailings facilities, the
assessment of the drainage condition during vane shear testing is needed so that potential overes-
timation of undrained strength through partial drainage is suitably taken into account (Reid 2016).
To assess the drainage conditions during the eVST tests, the use of a dimensionless time factor
was calculated, using the equation presented in Blight (1968) and updated by Chandler (1988),
which assumes the primary excess pore pressure generation relevant to drainage during shear oc-
curs during the vane rotation. Thus, it is assumed that the drainage path is equal to the diameter
of the vane and the time factor, T, is given by T = (cv*tf)/D2. cv is the coefficient of vertical con-
solidation in m2 per second, obtained from CPT dissipation tests, and interpreted using two dif-
ferent methods namely: Jones & Van Zyl (1981), and Houlsby & Teh (1988). A vertical anisot-
ropy cv/ch = 0.1 was applied. tf is the time to failure as measured from the time when vane rotation
begins to when the maximum torque applied to the rods was measured (seconds). D is the vane
diameter (m).
As outlined by Chandler (1988), for practical purposes, undrained conditions occur when the
degree of consolidation is below 10%, or when the time factor, T<0.05. The peak undrained shear
strength ratio was plotted against the calculated time factor (not shown herein), with the reference
drainage time factor of T=0.05. When using Jones & Van Zyl (1981), most of the test results were
undrained, with a time factor varying from 0.002 to 0.05, except for 4 tests at the standard rotation
rate (0.1 deg/s) that had a time factor varying from 0.05 to 0.12. When using Houlsby & Teh
(1988), most of the tests had a value of T>0.05, with values ranging from 0.02 to 0.97, suggesting
that the tests would not have been carried out in undrained conditions.
Some inherent limitations of using the time factor to determine the drainage condition should
be highlighted. As the vane diameter is known for each test, the uncertainty lies primarily with
the coefficient of consolidation, though there is also uncertainty in the actual diameter of the ver-
tical shearing surface (i.e., it may be larger than that of the vane diameter). The calculated time
factors in this analysis use the nearest CPT Pore Pressure Dissipation (PPD) tests. Dejong & Ran-
dolph (2012) demonstrate that partial consolidation conditions during CPT may introduce errors
in the interpretation of PPD tests. These errors are pronounced when the measured t50 values are
less than 50 seconds. Most of the t50 values obtained are less than 50 s. Thus, these tests may not
be reliable or representative of the in-situ condition.

4 IN-SITU PIEZOCONE TESTING PROGRAM

4.1 Methodology
The CPT were interpreted using empirical methods as well as others embedded in the critical state
soil mechanics (CSSM) framework with a focus on the in-situ state parameter, the peak undrained
shear strength ratio Su(peak)/σ’v and the liquefied undrained shear strength ratio Su(liq.)/σ’v. A distinc-
tion is made herein between the liquefied undrained shear strength estimated from the CPT Su(liq.)
and Su(res.) measured from the eVST as reported in Section 3 above.
The state-of-practice recognizes the CPT as the most reliable in-situ tests to define the in-situ
state of tailings through empirical screening-level methods (Been & Jefferies 1992, Plewes et al.
1992) or material-specific methods (Shuttle & Jefferies 2016). The screening-level methods were
adjusted for material-specfic Mtc (1.37) but not for the corresponding CSL compressibility. The
corresponding empirical relationships were used to obtain the CSL compressibility to account for
spatial variability. Although its full validation in silty tailings is still needed (Fourie et al. 2022),
a numerical analysis such as the material-specific method (Shuttle and Jefferies 2016) is probably
one of the most advanced methods to conduct the CPT interpretation of tailings as it accounts for
specific material behaviour including strain weakening when applicable and large deformations.
The material-specific method relies on the calibration of the NorSand constitutive model (Jefferies
& Shuttle 2002) and the NorSand widget (Shuttle & Jefferies 1998, 2016). The r.18 version of the
NorSand VBA spreadsheet was used for the calibration of the NorSand parameters based on the
CID/CIU/CAU, while the simulation of the CLEDtx was made with the r.23 version including
minor modifications by Mike Jefferies (pers. comm.) to account for a generalized unloading ef-
fective stress path. The NorSand widget v2.5 was used in the current study.
 Figure 7 shows the NorSand simulation of CAU-04, CID-06 and CLED triaxial testing. All the
tests were simulated with the same set of NorSand parameters shown in Table 5, with the excep-
tion of ψ, p’0, Kc, OCR and Gmax. The first 4 parameters were adapted for each triaxial testing
initial stress state and Gmax was estimated from the shear wave velocity (Vs) measurements col-
lected from the seismic CPT with the best fit equation shown in Table 5. It is noteworthy that
NorSand predicts the stress strain behaviour of the CLEDtx with a similar level of adequacy as
for CAU-04 and CID-06 even though the latter was shown not to have reached critical state. The
NorSand widget was then run for both drained and undrained conditions to obtain the material
specific coefficients k/k’ and m/m’ that link between the normalized cone tip resistance (Qp(1-
Bq)+1) and ψ. A minor rigidity dependency was noted in the drained NorSand widget simulations
while no dependency was noted in the undrained widget simulations. The material specific coef-
ficients were then integrated in an in-house spreadsheet. The drained coefficients (k, m) were
applied to infer the in-situ ψ when -0.02<Bq<+0.02 while undrained coefficients (k’, m’) were
applied otherwise.
The peak undrained shear strength ratio Su(peak)/σ’v was defined using two different methodolo-
gies; Nkt cone factor of 17 as explained in Section 3 and NorSand simulations of DSS testing for
the corresponding characteristic state parameter using the model parameters shown in Table 5.
Upon shearing, a contractive material eventually reaches its CSL at large strain and the lique-
fied undrained shear strength (Su(liq.)) controls the resistance of the material to flow failure. Three
methods were applied to the CPT data to estimate this ratio. The first one is Jefferies & Been
(2016) based on the in-situ ψ from the three different methods listed above. This method was
developed from case histories along with the CSSM framework. The second method is Robertson
(2010). The update of Robertson (2022) was not included in the in-house spreadsheet at the time
of the analysis. It is important to emphasize some distinctions between these two methods. Rob-
ertson (2010) was developed on the experience of case histories afterwards characterized by the
variable Qtn,cs, which is empirically normalized to represent the behaviour of clean sands. The
method of Jefferies & Been (2016) was developed on the basis of the CSSM and the same database
of case histories accounting for field-scale effects. The latter method implicitly considers material-
specific parameters such as compressibility (λ10) and the critical state ratio (Mtc). In the case of
mine tailings, these two parameters tend to be different from natural soils, and therefore, a mate-
rial-specific analysis is generally preferable. Finally, the Su(res,) from the eVST was also compared
to the CPT-based values.

200 Table 5. Calibrated NorSand Parameters

NorSand Parameters Value
Mtc (-) 1.37
150 N (-) 0.0
Deviator Stress, q (kPa)

χtc (-) 5.0

H (-) 30-200 ψ
100
υ (-) 0.2
Gref (Mpa) 1 51
NorSand
Gexp (-) 0.69
CAU-04 1
: Gmax = Gref(p’/p’ref)Gexp ,with p’ref =100 kPa
50 CID-06
CLEDtx
A reduction factor of 35% was applied to this relationship for
fitting CIU/CAU resulting in Ir between 120 and 195

0
0 5 10 15 20
Axial strain (%)

Figure 7. Deviator Stress as a function of Axial

Strain for CAU-04, CID-06 and CLEDtx and cor-
responding NorSand Simulations

4.2 Results
Figure 8 show the profile of typical CPT interpretation obtained. In tailings, a coefficient of hor-
izontal stress σh over vertical stress σv (K0) of 0.7 was assumed for the CPT interpretation and was
also supported by a finite difference stress-strain analysis. The equilibrium pore water pressure
was defined from the CPT PPD tests and readings from adjacent vibrating wire piezometer for the
corresponding period. As shown in Figure 8, a glaciolacustrine soft to firm clay deposit is found
beneath the TSF. This clay layer was addressed in the design and warranted the realization of the
CLED triaxial testing. This section focuses on the in-situ stress-strain tailings behaviour. On av-
erage, normalized CPT parameters of pore pressure Bq, cone tip resistance Qp, and sleeve friction
Fr of 0.11 ± 0.09, 13.3 ± 14.5 and 0.86 ± 0.36, respectively were found in tailings. Figures 8 and
9 shows the profile of the estimated in-situ state parameter and the statistical distribution, respec-
tively. The box plot provides a helpful overview of the statistical distribution of relevant results
such as state parameter. In Figure 9, the “x” and the horizonal solid line within each box plot
represent the average and the median (50th percentile), respectively. A quick visual examination
of Figure 9 reveals that the three methods show a log-normal distribution as the median is closer
to the 75th percentile (uppermost solid line) than the 25th percentile (lowermost solid line). The
material-specific method of Shuttle & Jefferies (2016) yields the lowest values and the narrowest
distribution of ψ with a characteristic value (between 80th and 90th percentile) between 0.02 and
0.03. Plewes et al. (1992) and Been & Jefferies (1992) showed higher and wider distribution with
characteristic values between 0.12 and 0.13 and between 0.08 and 0.09, respectively. The use of
the material-specific compressibility (λ10=0.066) in the screening-level methods yields the same
characateristics values. Figure 8 shows that the material-specific method typically aligns well with
the two screening-level methods when Bq approaches 0. As Bq is greater than +0.02 over most of
the tailings profile, undrained k’/m’ were mostly applied, except for some thin layers where
drained k/m coefficients were assigned by the in-house spreadsheet. However, a check was made
with drained k/m coefficients solely (not shown herein) and similar to Reid & Smith (2021), the
material-specific method resulted in the upper bound (drained k/m coefficients) and lower bound
(undrained k’/m’ coefficients) of the state parameter values with the two screening-level methods
lying in between. As outlined above, the t50 measured likely reflect a partial drainage condition in
the tailings during the CPT sounding. That point and considering that recent findings suggest that
the material-specific method tend to underestimate the in-situ state parameter in silty materials
(Fourie et al. 2022, Ayala et al. 2021) support the use of the 90th percentile over other characteristic
values. An attempt was also made to compare CPT-based state parameters and in-situ void ratio
through water content, but erratic and non-conclusive results were obtained and could not support
specifically any of the applied CPT-based methods (not shown herein).
Figure 8 also shows the profile of Su(peak)/σ'vo estimated from two methods. The DSS NorSand
relationship yields an average value of 0.23 ± 0.04 and a characteristic value (10th percentile) of
0.24. The cone factor Nkt of 17 yields a wider distribution with an average Su(peak)/σ'vo of 0.63 ±
0.68. This widely distributed range likely reflects partial drainage effects during CPT and/or field
vane testing, especially in the upper part of the tailings deposit. However, a characteristic value
Su(peak)/σ'vo of 0.24 was estimated which is consistent with the other method. A good agreement
was obtained between the characteristic (10th percentile) Su(peak)/σ'vo estimated from laboratory DSS
relationship (0.27), the NorSand DSS relationship (0.23) ant the Nkt inferred values (0.24).
Figure 8 shows the profile of Su(liq)/σ'vo while Figure 10 shows the statistical distribution of the
four CPT-based methods applied to SPCT-20-15. The first three are essentially based on the same
equation grounded in the CSSM framework, but with the state parameter obtained from three
different methods, namely Been & Jefferies (1992), Plewes et al. (1992) and Shuttle & Jefferies
(2016). In addition, Robertson (2010) is also shown on both figures. As for the in-situ state pa-
rameter distribution, Figure 10 shows that the CSSM-based screening-level methods (Been &
Jefferies 1992 and Plewes et al. 1992) show more widely distributed values than the material-
specific method. Robertson (2010) yields the narrowest distribution and the lowest average
(0.05±0.03) and median (0.04) values. The equation of Jefferies & Been (2016) applied to the
state parameter distribution estimated from Been and Jefferies (1992), Plewes et al. (1992) and
Shuttle & Jefferies (2016) results in average values of 0.12 ± 0.04, 0.08 ± 0.07 and 0.10 ± 0.02,
respectively. Characteristic values (10th percentile) of 0.09, 0.01 and 0.09, respectively were ob-
tained for the same three methods. The mode of the material-specific method is 0.10, slightly
above the 90th percentile. The very low Su(liq.)/σ'vo inferred by Plewes et al. (1992) is a combination
of the low compressibility inferred by the method as outlined by Torres-Cruz (2021) and the fact
that ψ>λ10 in the equation of Jefferies & Been (2016). The Su(res.)/σ'v from the eVST are shown in
Figure 8 for comparison with the CPT-based values. A Su(res.)/σ'v between 0.04 and 0.07, with an
average value of 0.05 was measured which is the same as Robertson (2010) but slightly lower
than the characteristic values of Been & Jefferies (1992) and Shuttle & Jefferies (2016).

5 CONCLUSION

This case study paper presents the methodology and the results of a detailed geotechnical charac-
terization of gold mine tailings including advanced laboratory and in-situ testing. Overall, a good
consistency was observed between the laboratory-inferred results and the in-situ results which in
this case, gives confidence for the use of the interpreted geotechnical tailings properties for anal-
ysis and design verifications for existing containment structures. The effect of partial drainage in
the CPT and or eVST was given consideration, when possible, but this topic warrants further
research especially when applying the material-specific method of Shuttle & Jefferies (2016) and
the NorSand widget with either perfectly drained or undrained condition. It is in the authors’
opinion that anisotropic consolidation be integrated more frequently in advanced laboratory test-
ing as it mimics more closely the in-situ stress state and promotes a higher brittleness which in
turns enhance the accuracy of the estimated CSL.
This paper demonstrates that a good soil mechanics bases fosters a good understanding of the
geotechnical behaviour of mine tailings that is essential for safe and sustained TSF design and
operation.

Cone Tip Resistance, qt Normalised Cone Normalised Friction Pore Pressure Ratio, Peak Shear Resistance Ratio, Post-liquefaction Undrained
State Parameter, ψ (-)
(MPa) Resistance, Q (-) Ratio, F (%) Bq (-) Su(peak)/σ'v (-) Shear Strength Ratio, Sr/σ'v
0 2 4 6 8 10 0 20 40 60 80 100 120 0 1 2 3 4 0.0 0.2 0.4 0.6 0.8 1.0 or Su(liq.)/σ'v (-)
0.0 0.2 0.4 0.6 0.8 1.0 -0.2 -0.1 0.0 0.1 0.2 0.3 0.0 0.1 0.2 0.3 0.4
0 0 0 0
0 0 0
Qtn
2 Predrilled unit (Waste Rock) 2 Qtn-cs 2 2 2 2 2
Qp
4 4 4 4 4 4 4
6 6 6 6 6 6 6
8 8 8 8 8 8 8
10 10 10 10 10 10 10
12 12 12 12 12 12 12
14 14 14 14 14 14 14
Tailings
16 16 16 16 16 16 16
18 18 18 18 18 18 18
20 20 20 20 20 20 20
22 22 22 22 22 22 22
24 24 24 24 24 24 24
26 26 26 26 26 26 26
Depth (m)

​
28 Silt (surface) 28 28 28 28 28 28
30 30 30 30 30 30 30
32 32 32 32 32 32 32

34 34 34 34 34 34 34

36 36 36 36 36 36 36

38 38 38 38 38 38 38
Cohesive Soils
40 40 40 40 40 40 40

42 42 42 42 42 42 42

44 44 44 44 44 44 44

46 46 46 46 46 46 46

48 48 48 48 48 48 48

50 50 50 50 50 50 50
Till
Contractive
Dilative

52 52 52 52 52 52 52
Fig.6.47 J & B (2016) - Ψ from Shuttle and
54 54 54 54 54 54 54 Jefferies (2016)
56 56 Been and Jefferies (1992) Fig.6.47 J & B (2016) - Ψ from Been and
56 Jefferies (1992)
56 56 56 56 Undrained and Drained
Fig.6.47 J & B (2016) - Ψ from Plewes (1992)
Plewes (1992), Reid (2015)
NKT - FVT VST- 20-15 SS-Norsand Undrained and Drained Robertson (2010)

Shuttle and Jefferies (2016) VST- 20-15

Undrained and Drained

Figure 8. Profile of SCPT-20-15

 Figure 9. Box Plots of the distribution ᴪ from Figure 10. Box Plots of the Distribution
three applied methods to SCPT-20-15 Su(liq.)/σ'vo from four CPT-based methods on
SCPT-20-15

REFERENCES

Ahmed, S., Vinod, J.S., Neaz Sheikh, M., Fourie, A., Reid, D. 2022. The εv/εa – p’ method for the deter-
mination of instability of granular soils under constant shear drained stress path, Canadian Geotechnical
Journal, Accepted manuscript.
Ayala, J., Fourie, A., Reid, D. 2021. Improved cone penetration test predictions of the state parameter of
loose mine tailings, Canadian Geotechnical Journal, Accepted manuscript.
Baxter, C.D.P., Bradshaw, A.S., Ochoa-Lavergne, M., Hankour, R., 2010. DSS test results using wire-
reinforced membranes and stacked-rings, Geotechnical Special Publication, GeoFlorida 2010, 8..
Been, K. et Jefferies M.G. 1992. Towards systematic CPT interpretation. Wroth Memorial Symposium,
Thomas Telford, Londres. 121-134.
Been, K. 2016. Characterizing mine tailings for geotechnical design. Australia Geomechanics 51(3).
Blight, G.E. 1968. A note on field vane testing of silty soils. Canadian Geotechnical Journal 5(3): 142–
149.
Chandler, R.J. 1988. The in-situ measurement of the undrained shear strength of clays using the field vane.
Vane Shear Strength Testing in Soils: Field and Laboratory Studies, ASTM STP 1014, A.F. Richards
(ed), ASTM, 13-44.
Dejong, J.T. and Randolp, M. 2012. Influence of partial consolidation during cone penetration on estimated
soil behavior type and pore pressure dissipation measurements. Journal of Geotechnical and Geoenvi-
ronmental Engineering 138: 777-788.
Fourie, A., Verdugo, R., Bjelkevik, A., Torres-Cruz, L.A., Znidarcic, D. 2022. Geotechnics of mine tail-
ings : a 2022 State of the Art, Proceedings of the 20th ICSMGE-State of Art and Invited Lectures,
Rahman and Jaksa (Eds), 2022 Australian Geomechanics Society, Sydney, Australia.
Fotovvat, A., Sadrekarimi, A., Etezad, E. 2022. Instability of Gold Mine Tailings subjected to Undrained
and Drained Unloading Stress Paths, Géotechnique, ahead of print.
GISTM Global industry standard on tailings management. 2020. ICMM (International Council on Mining
& Metals), UNEP (United Nations Environment Program) and PRI (Principles for Responsible Invest-
ment).
Houlsby, G.T. and Teh, C.I., 1988. Analysis of the piezocone in clay, Penetration testing 1988 2,
(Proc ISOPT, Orlando), Balkema, Rotterdam: 777-783.
ICOLD, in preparation. Committee L Tailings Dams and Waste Lagoons, Tailings Dam Safety.
Jefferies, M., Shuttle, D.A. 2002. Dilatancy in general Cambridge-type models, Géotechnique 52(9): 625-
637.
Jefferies, M. G. and Been, K. 2016. Soil liquefaction, a critical state approach. 2nd ed. CRC Press, Boca
Raton, FL, 690.
Jefferies, M., Morgenstern, N.R., Van Zyl, D.V., Wates, J. 2019. Report on the NTSF Embankment Failure,
Cadia Valley Operations, for Ashurst Australia.
Ladd, R. 1978. Preparing test specimens using undercompaction. Geotechnical Testing Journal 1(1): 16-
23.
Morgenstern, N.R., Vick, S.G., Viotti., C.B., Watts, B.D. 2016. Fundao Tailings Dam Review Panel: Report
on the immediate causes of the failure of the Fundao Dam.
Plewes, H.D., Davies, M.P. and Jefferies, M.G. 1992. CPT based screening procedure for evaluating
 liquefaction susceptibility. 45th Canadian Geotechnical Conference, Toronto, 26-28 October 1992. Ca-
nadian Geotechnical Socitety: 4-1 – 4-9.
Reid, D. 2015. Estimating slope of critical state line from cone penetration test – an update. Canadian
Geotechnical Journal52:46-57.
Reid, D. 2016. Effect of rotation rate on shear vane results in silty clay. In Proceedings of Geotechnical and
Geophysical Site Characterization 5. Sydney. Australian Geomechanics Society.
Reid, D. Fanni, R., Fourie, A., 2022a. Slurry deposition preparation method for tailings characterisation -
history, debates, techniques and benefits, Tailings 2022, online conference, July 6-8, 2022.
Reid, D., Fanni, R., DiDonna, P. 2022b. The effect of tamping conditions on undrained shear strengths of
a non-plastic sandy silt tailings. Canadian Geotechnical Journal 59:783-795.
Reid, D., Smith, K. 2021. Interpretation of state parameter in partially drained tailings – a case history
examination, Géotechnique Letters 11(4).
Riveros, G.A. and Sadrekarimi, A. 2021. Static liquefaction behaviour of gold mine tailings, Canadian
Geotechnical Journal 58:889-901.
Robertson, P.K. 2010. Evaluation of flow liquefaction and liquefied strength using the cone penetration
test, Journal of Geotechnical and Geoenvironmental Engineering 136(6): 842-853.
Robertson, P.K. 2022. Evaluation of flow liquefaction and liquefied strength using the cone penetration test
– an update, Canadian Geotechnical Journal:7.
Schnaid, F. 2021. On the geomechanics and geocharacterization of tailings, Proceedings of the 6th Interna-
tional Conference on Geotechnical and Geophysical Site Investigation.
Smith, K., Fanni, R., Chapman, P., Reid, D. 2019. Critical State Testing of Tailings: Comparison between
various tailings and implications for design, Proceeding of Tailings and Mine Waste 2019:1183-1195.
Shuttle, D.A., Jefferies, M. 1998. Dimensionless and unbiased CPT interpretation in sand, International
Journal of Numerical and Analytical Methods in Geomechanics 22: 351–391.
Shuttle, D.A., Jefferies, M. 2016. Determining silt state from CPTu. Geotechnical Research 3(3): 90-118.
Torres-Cruz, L.-A., Santamarina, J.C. 2020. The critical state line of nonplastic tailings, Canadian Geotech-
nical Journal 57(10).
Torres-Cruz, L.-A. 2021. The Plewes Method: A Word of Caution, Mining, Metallurgy & Exploration
38(3):1329-1338.