Professional Documents
Culture Documents
G-2P1J-01-0 ODTSF Parameters Calibration F - AOS - 0-0
G-2P1J-01-0 ODTSF Parameters Calibration F - AOS - 0-0
PO Box 943
West Perth WA 6872, Australia
Project Memo
Client: SRK Australasia Date: 11/08/2020
1 Introduction
1.1 Project background
BHP Olympic Dam Corporation Pty Ltd (BHPOD) operates the Olympic Dam polymetallic mine located
approximately 580 km north of Adelaide in South Australia. The tailings stream is deposited in two
active tailings storage facilities (TSFs), Cell 4 and Cell 5, which are raised by upstream staged lift
construction. BHPOD recently appointed SRK Consulting (Australasia) Pty Ltd (SRK) to design a
stabilizing buttress and drain for the North, East and South walls of TSF 5.
The most recent TSF stability assessment by limit equilibrium analyses predicted that the FoS in some
areas of the TSF5 embankment will be inadequate following the routine FY2021 wall raises. As a
result, the TSF5 North, East and South walls will require reinforcement in FY2021 (North and East)
and FY2022 (South) to achieve the minimum factor of safety stability requirements. The key objective
is to reinforce the TSF5 by building a reinforcing buttress to ensure a compliant FoS following the 2021
regular wall -and subsequent tailings level- raise to an embankment crest height of approximately
RL124.2 m. Given TSF5 is proposed to be raised to a final maximum crest elevation of RL 132 m,
additional buttressing (Phase 2) is likely to be required in the future (Figure 1-1). The current study is
only required to achieve an adequate stability FoS for an embankment crest height up to RL 124.2 m
(Phase 1).
Currently, SRK Australia is undertaking trade-off studies for the buttress and drain configurations.
Several buttress and drain geometries are being analysed by means of static and post-seismic limit
equilibrium analyses. One of these configurations will be selected for deformation analysis.
SRKAU contacted SRK Argentina to undertake a numerical deformation assessment with the aim of
assessing an adequate construction rate for the reinforcement buttress and to estimate the post-
seismic configuration after three Safety Evaluation Earthquakes (SEE) earthquakes defined by
AECOM (2018), such that ANCOLD (2019) guideline is complied. This document presents the
calibrations of the constitutive models to be used in the subsequent deformation modelling.
Can SRK confirm AECOM (2018) complies with the requirements of ANCOLD (2019) Guidelines on Design
of Dams and Appurtenant Structures for Earthquake and ANCOLD (2019) Guidelines on Tailings Dams
Planning, Design, Construction,
Figure Operation
1-1: Conceptual andcross-section.
buttress Closure. A recent review has indicated the Norwest fault (100
km west of Olympic Dam at nearest point) is included in the Australian Neotectonic Features Database and
SOTTILE/SFRISO G-2P1J-01-0 ODTSF Parameters Calibration F_AOS.docx 11 August 2020
is probably capable of producing earthquakes of Mw 7.5 or larger, but was not included in NSHA18 or
AECOM 2018. Norwest fault was not included in NSHA18 but may have been added to the neotectonics
SRK Consulting Page 2
• 𝜓𝑎 : dilatancy angle;
• 𝐾0𝑛𝑐 : stress ratio at oedometer compression in normally consolidated regime;
• 𝜈𝑢𝑟 : elastic unloading / reloading stiffness;
• 𝑂𝐶𝑅: overconsolidation ratio;
• 𝐾0 : Lateral stress coefficient.
Figure 2-1: a) Yield surfaces of HSS model in p-q plane. b) Stiffness parameters 𝑬𝟎 , 𝑬𝒖𝒓 and
𝑬𝟓𝟎 of HSS model in a triaxial test. (Plaxis Manual, 2020).
2.3 PM4Sand model description
PM4Sand is based on the constitutive model for sands developed by Dafalias & Manzari (2004), which
is a bounding surface plasticity model formulated in the stress-ratio controlled framework and is critical
state compatible. It is worth noting that many modifications were incorporated, with the most significant
one being the formulation for just 2D plane strain loading conditions. As it will be shown in next section,
the calibration can be extended for non-plastic/low-plasticity sandy/silty tailings as well, as the model
can reproduce the generation of excess pore pressure during cyclic loading and can nicely capture
the N-CSR curves for different states of the material.
The model is based on the Critical State Soil Mechanics framework. It uses a relative state parameter
index (𝜉𝑅 ), defined as the difference between the relative density at critical state (𝐷𝑅,𝐶𝑆 ) and the current
relative density (𝐷𝑅 ), for a certain mean effective pressure ratio (𝑝′/𝑝𝐴 )(Figure 2-2). This is analogous
to the definition of the well-known state parameter (𝜓) proposed by Been & Jefferies (1985), with the
difference that the Critical State Line (CSL) is defined in terms of relative density - rather than void
ratio - based on Bolton (1986), as:
𝐷𝑅,𝐶𝑆 = 𝑅/[𝑄 − ln(100 ∙ 𝑝′/𝑝𝐴 )]
In addition, the model defines the behaviour in the 𝑝′ − 𝑞 space by means of four surfaces: bounding,
dilatancy, critical and yield (Figure 2-3). The bounding and dilatancy ratios are related to the critical
stress ratio (𝑀) as:
𝑀 = 2 ∙ sin 𝜙𝑐𝑣
𝑀𝑏 = 𝑀 ∙ exp(−𝑛𝑏 𝜉𝑅 )
𝑀𝑑 = 𝑀 ∙ exp(𝑛𝑑 𝜉𝑅 )
The bounding surface controls the relationship between the peak friction angle and the relative state;
this surface cannot be surpassed, and for a given stress state (𝑞/𝑝′ ) the distance to the critical surface
can be measured with the aim of computing plastic deformations. The dilatancy surface defines the
location where the contractive to dilative behaviour occurs; under cyclic loading, this translates to the
generation of positive/negative pore pressures. During shearing, both surfaces (bounding and
dilatancy) approach the critical surface, as the state parameter approaches the critical state line (𝜉𝑅 →
0). It is worth illustrating that when 𝜉𝑅 is negative (state denser than critical), the bounding and dilatancy
surfaces are located above and below the critical surface, respectively; and the opposite happens for
a positive 𝜉𝑅 (state looser than critical), such that the dilatation surface locates above the critical
surface and only compression occurs.
The yield surface (Figure 2-3) is formulated as a small cone in the stress space, with a back-stress
ratio tensor (𝛼) that defines its position in the deviatoric stress-ratio space, and 𝑚 that defines the
radius of the cone. The deviatoric unit normal to the yield surface (𝑛) is used to define the general
formulation of the bounding and dilatancy surfaces, and to compute the plastic modulus (𝐾𝑝 ) and the
amount of dilatancy and contractancy (𝐷).
The model parameters are:
• 𝐷𝑅 : relative density;
• 𝐺0 : shear modulus coefficient, that defines small-strain shear modulus as 𝐺 = 𝐺0 ∙ 𝑝𝐴 ∙ √𝑝′ /𝑝𝐴
• ℎ𝑝0 : contraction rate parameter, that adjust contractiveness and allows to calibrate 𝐶𝑅𝑅;
• 𝑝𝐴 : atmospheric pressure;
• 𝑒𝑚𝑖𝑛 and 𝑒𝑚𝑎𝑥 : minimum and maximum void ratio;
• 𝑛𝑏 : bounding surface parameter, that controls relative position of bounding to critical state
surface;
• 𝑛𝑑 : dilatancy surface parameter, that controls the stress ratio of contraction-dilation transition;
• 𝜙𝑐𝑣 : critical state friction angle, that defines the critical state surface (i.e. value of 𝑀);
• 𝜈: Poisson’s ratio;
• 𝑄 and 𝑅: Bolton’s parameters used to define the CSL in 𝐷𝑅 − log 𝑝′ space;
• 𝑃𝑜𝑠𝑡𝑆ℎ𝑎𝑘𝑒: used to activate the reduction of elastic stiffness to simulate the post-shaking
reconsolidation.
Figure 2-2: Definition of the critical state line in the 𝑫𝒓 – 𝒍𝒐𝒈[𝒑/𝒑𝒂 ] space; example for 𝑸 = 𝟏𝟎
and 𝑹 = 𝟏. 𝟓 (Plaxis PM4Sand manual, 2018).
Figure 2-3: Yield, critical, dilatancy and bounding surface in 𝒑’ − 𝒒 space (left). Yield,
dilatancy and bounding surface in 𝒓𝒙𝒚 − 𝒓𝒚𝒚 space (Boulanger & Ziotopoulou,
2015).
Geotechnica 𝒌𝒉 𝒌𝒗
l units [m/day] [m/day]
Figure 3-1: TSF4 West section piezometer assessment. Comparison between measured
excess pore pressure and Plaxis results. (SRK, 2018b).
Figure 3-2: TSF4 Rate of rise assessment. Comparison between measured excess pore
pressures and Plaxis results at Pad A – VWP01. (SRK, 2020).
PM4Sand formulates the CSL in terms of relative density rather than void ratio; therefore, an estimation
of minimum and maximum void ratios is necessary. Within this context, 𝑒𝑚𝑎𝑥 and 𝑒𝑚𝑖𝑛 are just model
parameters which, put together with the other parameters, produce a reasonable and realistic
calibration of the material behaviour. The ongoing laboratory testing suggests a void ratio range
between 0.50 and 0.90; the first representative of samples sheared at higher confining stresses and
the second representative of loose samples at their initial configuration. On the other hand,
Cubrinovsky & Ishihara (1999) suggest that the void ratio amplitude (𝑒𝑚𝑎𝑥 − 𝑒𝑚𝑖𝑛 ) for silty soils can
vary between 0.60 and 0.90. A void ratio amplitude of 0.70 is chosen for this case, which is evenly
distributed to achieve 𝑒𝑚𝑖𝑛 = 0.35 and 𝑒𝑚𝑖𝑛 = 1.05.
In addition to the minimum and maximum void ratio, the CSL is calibrated with two additional
parameters, 𝑄 and 𝑅. Figure 4-1 shows the CSL calibrated using 𝑄 = 10, 𝑅 = 2.4, 𝑒𝑚𝑖𝑛 = 0.35 and
𝑒𝑚𝑎𝑥 = 1.05. A comparison is presented with respect to the three draft semi-log CSL obtained from the
ongoing testing program, which is expressed in terms of relative density as:
𝑒𝐶𝑆 = Γ − 𝜆10 ∙ log 𝑝′
𝑒𝑚𝑎𝑥 − 𝑒𝐶𝑆
𝐷𝑅,𝐶𝑆 =
𝑒𝑚𝑎𝑥 − 𝑒𝑚𝑖𝑛
Using these relative densities at critical state (𝐷𝑅,𝐶𝑆 ), the CSL is back computed as:
𝑅
𝑝 ′ = exp (𝑄 − )
𝐷𝑅,𝐶𝑆
In addition to the CSL, a line with a constant state parameter of 𝜓 = 0.10 is plotted using the expression
shown below. This representative state parameter value was chosen from the statistical interpretation
from CPTu data (SRK, 2020a), for which Plewes (1992) and Been & Jefferies (1992) screening
methods were used. This line is important, as it will define the values of initial relative densities for a
given mean effective stress, representative of the in-situ state of the material.
𝑒𝑚𝑎𝑥 − (𝑒𝐶𝑆 + 0.10)
𝐷𝑅,𝜓 =
𝑒𝑚𝑎𝑥 − 𝑒𝑚𝑖𝑛
It must be noted that a good agreement is achieved between the PM4Sand CSL and those obtained
for the PT29 and PT30 semi-log CSL, for the expected mean effective pressure range of 𝑝′ =
50|300𝑘𝑃𝑎.
Figure 4-1: PM4Sand Critical State Line (CSL) calibration. Comparison with linear CSL, from
SRK 2020.
4.2 Calibration of the remaining parameters
The remaining parameters of PM4Sand are calibrated based on the cyclic direct simple shear tests
(CDSS) performed on tailings samples. In addition, a sensitivity analysis is undertaken to assess the
influence of some parameters on the overall model response.
Two testing programs were undertaken by Golder. The first one was conducted in 2014 and entailed
eighteen CDSS tests performed on tailings reconstituted samples that were prepared from slurry in 1D
columns under a load of 75 kPa (Golder, 2014). The second one was conducted in 2017 and entailed
six CDSS tests performed on tailings intact piston tube samples (Golder, 2017). Unfortunately, the
void ratio was not measured with enough accuracy, therefore it is not possible to quantitively define a
state parameter of the material. However, both samples (i.e. slurry-reconstituted and indisturbed)
reasonably represent the range of states expected in situ, and sensitivity analyses will be performed
in the subsequent modelling to account for this uncertainty.
A summary of the tests is shown in Table 4-2, and the full test data is presented in Appendix A.
Table 4-2: Summary of Golder CDSS testing.
The shear modulus coefficient (𝐺0 ) is calibrated for a reference pressure of 100 kPa, where 𝐺 [𝑘𝑃𝑎] =
100 ∙ 𝐺0 ∙ √𝑝′ /100 𝑘𝑃𝑎. A value of 𝐺0 = 110 was calibrated using the CDSS data; however, a much
higher value (about 𝐺0 = 550) is expected for the field conditions, as previously shown for TSF4 where
SCPTu data was used (SRK, 2018a). A sensitivity analysis will be done on 𝐺0 to account for this
variation.
The friction angle at critical state was already calibrated for HSS model, for which a value of 𝜙𝑐𝑣 = 33°
was chosen. Default values 𝑛𝑏 = 0.50 and 𝑛𝑑 = 0.20 are adopted for these parameters.
With all parameters already calibrated, only the contraction rate parameter (ℎ𝑝0 ) is left, which defines
the rate of plastic volumetric strains in contraction and allows to calibrate 𝐶𝑆𝑅 − 𝑁𝑓 curves. This
parameter is calibrated by means elemental test simulations in Plaxis with the initial test conditions
specified in Table 4-3. The following comments are made:
i. an initial vertical effective stress of 200 kPa was chosen for the simulations; sensitivity
analyses will be later shown for a different value;
ii. a K0 value of 0.60 was chosen, assuming that samples are normally consolidated;
iii. two values of state parameters were chosen -0.00 and 0.10- representing the expected range
of variation in the field (SRK,2020a), for which the correspondent relative densities were
computed.
A comparison in terms 𝐶𝑆𝑅 − 𝑁𝑓 curves between CDSS elemental test simulations and laboratory
testing results is shown in Figure 4-2. The following observations are made:
i. the 2014 testing represent a lower bound of cyclic strength, which is nicely represented by the
simulations with the loosest configuration (i.e. 𝜓 = 0.10);
ii. the 2017 testing represent an upper bound of cyclic strength, that is well captured by the
densest configuration (i.e. 𝜓 = 0.00);
iii. six subsets were chosen to show a detailed stress-strain comparison in the next section.
Table 4-3: Initial conditions for CDSS simulations in Plaxis.
Simulation 𝝈′𝒗 [𝒌𝑷𝒂] 𝒑′𝟎 [𝒌𝑷𝒂] 𝒉𝒑𝟎 [−] 𝝍 [−] 𝑫𝒓 [−] 𝑪𝑺𝑹 [−] 𝑵𝒇 [−]
A 0.125 32
B 0.140 19
C 200 146.6 15 0.10 0.336 0.160 11
D 0.175 7
E 0.200 3
F 0.165 27
G 0.175 20
0.00 0.479
H 0.200 10
I 0.225 5
Figure 4-2: CSR – Nf curves. Comparison between CDSS elemental test simulations in Plaxis
using PM4Sand and Golder laboratory testing results.
4.3 Comparison at elemental level
A detailed comparison between CDSS elemental test results and laboratory results is presented for
the six subsets defined in Figure 4-2; subsets 1 to 4 correspond to the loosest material state (2014
testing), while subsets 5 and 6 correspond to the densest material state (2017 testing). The
comparisons are shown from Figure 4-3 to Figure 4-8; the following observations can be made:
• the comparison is made in terms of cyclic stress ratio (𝐶𝑆𝑅), cyclic shear strains (𝛾𝑥𝑦 ), number of
′
cycles (𝑁𝑐𝑦𝑐 ) and pore pressure ratio (𝑅𝑢 = 𝑝𝑒𝑥𝑐𝑒𝑠𝑠 /𝜎𝑣0 );
• for subsets 1, 2, 3 and 4, two values of ℎ𝑝0 were chosen: 10 and 15 to evaluate how this impact on
the pore pressure built up. From Figure 4-3 to Figure 4-6, it is observed that: i) the PM4Sand
calibrations nicely captures the 𝛾𝑥𝑦 − 𝐶𝑆𝑅 and the 𝛾𝑥𝑦 − 𝑁𝑐𝑦𝑐 response; ii) the pore pressure built
up is in good agreement with the data, with a slight underestimation at early stages for the ℎ𝑝0 =
15 case; this is improved with a lower ℎ𝑝0 = 10, but also reduces the number of cycles to failure.
The latter aspect will be further evaluated in the sensitivity analysis;
• for subsets 5 and 6, only the case of ℎ𝑝0 = 15 is analyzed. Figure 4-7 and Figure 4-8 show that the
model reasonably represents the measured behavior, with a slight underestimation of pore
pressures and strains prior to failure.
Figure 4-3: Comparison between CDSS simulations and lab results; Subset 1.
Figure 4-4: Comparison between CDSS simulations and lab results; Subset 2.
Figure 4-5: Comparison between CDSS simulations and lab results; Subset 3.
Figure 4-6: Comparison between CDSS simulations and lab results; Subset 4.
Figure 4-7: Comparison between CDSS simulations and lab results; Subset 5.
Figure 4-8: Comparison between CDSS simulations and lab results; Subset 6.
4.4 Sensitivity analyses
Sensitivity analyses were performed on the loosest sample (i.e. 𝜓 = 0.10) to evaluate how ℎ𝑝0 , 𝐺0
′
and 𝜎𝑣0 affect the result in terms of 𝐶𝑆𝑅 − 𝑁𝑓 response. The following cases were analysed:
• ℎ𝑝0 = 10: the full response was shown in the previous section, where it was anticipated that a lower
value entails a better prediction of pore pressures at early stages but at the cost of decreasing the
number of cycles to failure; this is illustrated in Figure 4-9, where the whole curve is shifted towards
the left. It is recommended to use both values in the deformation modelling, to evaluate a range of
expected deformations;
• 𝐺0 = 550: this is a likely value expected in-situ. The response in terms of 𝐶𝑆𝑅 − 𝑁𝑓 remains the
same when compared to the base case with 𝐺0 = 110 (Figure 4-9). Nevertheless, a higher value
of 𝐺0 will entail lower shear strains prior to failure and also will improve the capacity of the model
to propagate higher shear wave velocities. Therefore, both values will be used in the deformation
modelling;
• ′
𝜎𝑉0 = 400 𝑘𝑃𝑎: this represents a value closer to the maximum expected in-situ. The state
parameter is the same as the base case (𝜓 = 0.10), therefore, the initial relative density is updated
to 0.413 because the confining stress is higher. The response in terms of 𝐶𝑆𝑅 − 𝑁𝑓 remains the
′
same when compared to the base case with 𝜎𝑉0 = 200 𝑘𝑃𝑎 and 𝐷𝑟 = 0.336. This is expected
because the PM4Sand response depends on the (relative) state parameter, which is the same for
′
both combinations of 𝜎𝑉0 − 𝐷𝑟.
Figure 4-9: Sensitivity analyses results in terms of CSR – Ncyc for the loosest sample
configuration.
6 References
ANCOLD (2012) Australian National committee on Large Dams. Guidelines on Tailings Dams.
Planning, Design, Construction, Operation and Closure.
ANCOLD (2019) Australian National committee on Large Dams. Guidelines on Tailings Dams.
Planning, Design, Construction, Operation and Closure. Addendum.
Been, K., Jefferies, M. G. (1985). “A state parameter for sands”. Geotechnique, 35(2), 99-112.
Bolton, M. (1986). “Strength and dilatancy of sands”. Geotechnique, 36(1), 65-78.
Boulanger, R.W., Ziotopoulou, K. (2015). “PM4Sand (version 3): A sand plasticity model for
earthquake engineering applications”. Report No. UCD/CGM-15/01, March.
Brinkgreve, R., B. J., Zampich, L. M., Ragi Manoj, N. (2020). “Plaxis Connect Edition V20 Manual”.
Netherlands.
Cubrinovski, M., Ishihara, K. (1999). “Empirical correlation between SPT N-value and relative density
for sandy soils”. Soils and Foundations, 39(5): 61-71.
Dafalias, Y.F., Manzari, M.T. (2004). “Simple plasticity sand model accounting for fabric change
effects”. Journal of Engineering Mechanics, 130(6), 622–634.
Vilhar, G., Brinkgreve, R., B. J., Zampich, L. M. (2018). “PLAXIS The PM4Sand model 2018”.
Netherlands.
7 Closure
This memorandum presents the calibrations of the constitutive models that will be used to simulate
the staged construction of the TSF5 buttress and its subsequent dynamic deformation modelling.
Please contact the undersigned if you have any queries.
Yours faithfully
SRK Consulting
Appendix A
CDSS Raw Data