C147148-01 Technical Note On Geotechnical Aspects and Runout Modelling Scenarios R02

Note on Geotechnical Aspects and Runout
Modelling Scenarios, Dike 2 Pontal

Technical Note | Minas Gerais, Brazil
C147148-01 02 | 26 April 2020
For Review
VALE S.A.
VALE S.A.
Document Control
Document Information
Project Title Dike 2 Pontal System, Runout Analysis
Document Title Note on Geotechnical Aspects and Runout Modelling Scenarios, Dike 2 Pontal
Fugro Project No. C147148
Fugro Document No. C147148-01
Issue Number 02
Issue Status For Review
Fugro Legal Entity Fugro Belgium
Issuing Office Address Avenue de Broqueville 12 – 1150 Brussels, Belgium
Client Information
Client VALE S.A.
Av. Dr. Marco Paulo Simon Jardim, 3.580 - Prédio 1 - Térreo - Mina de Águas Claras - 34006-
Client Address
270 - Nova Lima - MG - Brazil
Client Contact Willyan G. Debastiani
Client Document No. C147148
Revision History
Prepared Approved
Issue Date Status Comments on Content Checked By
By By
MRA, PTH,
01 23 April 2020 For Review Awaiting client comments RJE BSP
RJE, BSP
MRA, PTH,
02 26 April 2020 For Review Inclusion of client comments RJE BSP
RJE, BSP
Initials Name Role
BSP Benoit Spinewine Principal Engineer
MRA Mikaël Ramos da Silva Geotechnical Group Lead
PTH Pedro Tha Engineering Manager
RJE Richard Jewell Principal Geotechnical Consultant
C147148-01 02 | Note on Geotechnical Aspects and Runout Modelling Scenarios, Dike 2 Pontal
VALE S.A.
FUGRO
Fugro Belgium
Avenue de Broqueville 12
1150 Brussels
Belgium
Willyan G. Debastiani, VALE S.A.

Av. Dr. Marco Paulo Simon Jardim, 3.580
Prédio 1 - Térreo - Mina de Águas Claras
34006-270 - Nova Lima - MG
Brazil
26 April 2020
Dear Willyan Debastiani,
We have pleasure to submit this Technical Note providing geotechnical assessments for the length of
failure at Dike 2, Pontal system. This Technical Note presents interpretations of the collected
geotechnical data at Dike 2 and recommendations for depth and lateral extents of failure to be
considered for runout analysis.
Please do not hesitate to ask for any additional information or clarification.
We trust the Technical Note will provide the information requested by AECOM during the online
meeting of 16th April.
Yours faithfully,
Benoit Spinewine, Richard Jewell, Michael Ramos da Silva and Pedro Thá
Principal and Senior Engineers
VALE S.A.
Frontispiece
VALE S.A.
Contents
1. Introduction 1
2. Empirical Evidence and Basis to Define a Cautious Length for Dam Breach 2
2.1 Theoretical and numerical studies 2
2.2 Case history data for tailings dam Case 2A failures 3
2.3 Proposed criterion by Fugro to bound the observed failure data 3
2.4 Application to Pontal Dike 2 4
3. Empirical Evidence for the Angle to Define the Runout Volume 5
3.1 Comment on the cross section of Dike 2 5
3.2 Comment on continuing movement at Pontal Dike 2 6
4. Geotechnical Evidence of Continuous very Loose Layers in the Tailings near Dike 2 7
5. Presence of Residual Soil at Pontal Dike 2 Limiting the Maximum Possible Breach Dimensions 8
6. In situ Water Content, Rheometer Data and Selection of Cautious Parameters for Run out
Analysis 9
7. Results of Runout Modelling Scenarios 10
8. Conclusions 12
9. References 14
List of Plates 15
Tables in the Main Text

Table 2.1: Definition of Dam Breach Types by CDA (Martin et al., 2019) 3
Table 2.2: Results for Expected Failure Length at Pontal Dike 2 4
Table 7.1: Summary of scenarios results 12
Page i of i
VALE S.A.
1. Introduction
Run out analyses supported by associated geotechnical assessments are being provided to
VALE by Fugro for several tailings storage dikes. Most recently, rheometer testing was
completed on samples from tailings retained by Pontal Dike 2. Updated analyses were
completed using rheological properties for the tailings based on these new rheological test
data. Following presentation of findings and discussion of results between Vale and AECOM
on 16th April, Fugro was requested to provide more detailed justification for the selection of
the dam breach dimensions, initially limited to a central 200 m wide section of Dike 2, and to
consider the likelihood and assess the potential consequences of a wider dam breach at Pontal
Dike 2 considering a complete and instantaneous liquefaction of the entire dike. This Technical
Note addresses the requested clarifications and additional run out results.
1.1 About this technical note

This technical note is one of a series for Vale that reports on analysis to assess the
consequences of potential breach in tailings storage facilities considered to be at risk. The
assessments are based on numerical analysis for the tailings run out following a hypothetical
dike or dam breach. As part of the work, Fugro has identified that the causes and dimensions
of the dike breach are separate to the analysis of the subsequent run out of the tailings retained
that suffer a sudden loss of support. Fugro has not been commissioned to analyse the stability
of the dike or the probability of failure and breach. However, the dimensions of the expected
breach are an important input to the run out analysis, along with other parameters such as the
rheological properties of the tailings. The basis for selection of reasonable dimensions of dam
breach for Dike 2 Pontal is described in this technical note, together with the consequences
arising from the run out of tailings that would follow.
A proposed semi-empirical method is suggested as the basis to determine reasonably

conservative values for the likely scale of dam breach that might be expected for any tailings
dam where failure would be classified as Case 2A of the current Canadian Dam Association
(CDA) guidelines. The selected length 200 m for dam breach analysis at Dike 2, and the most
likely depth at the dike governing run out, is justified also by analysis of the available
geotechnical data. Fugro has also considered a maximum credible length and dimensions of
dam breach as requested by AECOM, to investigate the likely consequences should a larger
breach occur. The rheological test data for tailings at Dike 2 have been assessed following the
current standard methods.
The above work is based on the data made available to Fugro and interpretation of related
analysis and guidance published in the literature. It represents a professional assessment of
the likely dimensions of any failure of the dike, and presents numerical analysis for the
subsequent run out of tailings and impact on structures downstream of the dike. This technical
note does not include an assessment for the probability of failure for Dike 2, which is outside
the scope of work commissioned.
Page 1 of 15
VALE S.A.
2. Empirical Evidence and Basis to Define a Cautious

Length for Dam Breach
2.1 Theoretical and numerical studies
Fugro has compared the analytical and numerical work published on the expected length of
failure for long, linear geotechnical structures with relevant empirical data from case histories
of tailings dam failure.
The analytical and numerical work is summarised in a paper by Hicks and Li (2018). The
objective, motivated initially by water retaining dikes in the Netherlands, was to predict the
length of failure that was likely within a long dike embankment system. Vanmarcke and Calle
tackled this question, separately, by analytical means, adopting simplifying assumptions, as
described by Hicks and Li (2018). These latter authors also completed direct numerical
modelling to compare with the results from the analytical solutions.
The premise is that natural variations along any long embankment structure or slope (linear
geotechnical structures) result in localised failure; long structures are not homogenous. The
feature that leads to localised failure may be a continuous zone or layer of weaker/looser soil
(natural or fill), or locally steeper or higher slope conditions, or a zone of higher water pressure
than elsewhere. These potential causes of local slope failure are termed “features” and the
length of such a feature (a slope may have several of them) is termed the “scale of feature” or
SOF by Hicks and Li. This terminology has been maintained in the present work. The objective,
therefore, is to identify the resulting length of slope failure in relation to the SOF; there are
typically many features within a long geotechnical structure.
For Pontal Dike 2, very loose tailings could provide a feature (or weakness) that might trigger
preferential failure. The net height of the dike, the difference in level between the retained and
downstream tailings material, is largely constant at Dike 2. No apparent source or evidence
was found for high localised water pressures along the dike. Therefore the geotechnical data
has been analyzed to see if there are loose and continuous zones of tailings material below
Dike 2 that could trigger either deeper failure or failure over a significant length of the dike.
The results of the analytical and numerical studies summarised by Hicks and Li are shown on
Plate 2.1. The summary is expressed in terms of the failure length within the long embankment
or slope as a function of the length of the SOF, the zone of weakness or higher loading within
the embankment or slope length that triggers failure.
Shown on the plot is the simple but conservative assumption derived by Fugro to apply to
tailings dams. This uses the most conservative result found in each case, and is limited to the
range SOF > 5 m.
The inclined line defined by Fugro simply represents equality between the expected failure
length Lf and the length of the SOF (length of weakness). The simple relation Lf = SOF applies
where SOF > 25 m.
Page 2 of 15
VALE S.A.
As shown below, in practical terms there is little interest in shorter lengths of weakness.
However, if cases with 25 m > SOF > 5 m are to be considered for any reason, then the
assumption SOF = 25 m applies in that range, as shown on Plate 2.1.
2.2 Case history data for tailings dam Case 2A failures

For the analysis of Dike 2, Fugro has compared the case SOF > 25 m with the empicial database
of observed failures for the conditions that apply at Dike 2: these correspond with Case 2A as
defined in the new Canadian Dam Guidelines (Martin et al, 2019), Table 2.1.
Table 2.1: Definition of Dam Breach Types by CDA (Martin et al., 2019)
A total of 14 upstream tailings dam failures of type Case 2A were identified in the literature.
Based on reported details and dimensions, supplemented by data accessible from satellite
imagery of the failed structures, the length of dike failure Lf was determined in each case
permitting the case histories to be plotted as shown on Plate 2.2.
The failure length is expressed as a proportion of the total dam length, Lf/Ld, and is plotted
against the aspect ratio for the dam Ld/Hd .
The limiting case Lf/Ld = 1 represents the total length of the dam participating in failure.
2.3 Proposed criterion by Fugro to bound the observed failure data

For cases with SOF > 25 m, Fugro’s simplified criterion that conservatively bounds the
analytical and numerical results described by Hicks and Li (2018) is Lf = SOF.
This criterion can be plotted on Plate 2.2 by noting that Ld/Hd = (Ld/SOF).(SOF/Hd). Hence a
specific value for the scale of feature as a proportion of the dam height, SOF/Hd , provides a
Page 3 of 15
VALE S.A.
contour line on Plate 2.2, from which the applicable SOF can be found for any specific dike
dimensions Ld and Hd.
The results are illustrated on Plate 2.3 and show that values SOF/Hd equal to 15 or 10 bound
the existing data for upstream tailings dam failures. The case SOF/Hd = 15 bounds all of the
case records identified, while a more realistic case SOF/Hd = 10 excludes only one case history
(Sullivan dam) comfortably bounding the other case record data of tailings failures of CDA type
Case 2A.
Note that for each case history, the implied SOF may be determined from the reported
dimensions Lf , Ld and Hd. In all cases that fall within the locus defined by SOF/Hd = 10, for
example, the applicable value for the case record will be SOF/Hd < 10, equal to a value that
may be determined. Therefore the criterion proposed, for example SOF/Hd = 10, represents
the largest expected value for SOF that might reasonably be expected, based on the available
case history records, not the expected value for SOF should dike breach actually occur.
2.4 Application to Pontal Dike 2

The current net height of Pontal Dike 2 is given by the difference in elevation between the
tailings downstream at an elevation 786 m (defined by the constructed toe weight) and the
elevation of the retained tailings at an elevation 800 m, giving a net height 14 m.
The difference in height between the base of the dike at elevation 780 m and the retained
tailings at elevation 800 m would give a total height of 20 m. This larger height is not
considered to be applicable for the current dike geometry.
The total length of Dike 2 is about 600 m which provides the two cases Ld/Hd = 30 or 43, with
the latter value, derived from the current net height, considered most relevant. The geometry
for Dike 2 is shown on Plate 2.4 from which the expected maximum length of failure Lf may be
directly determined.
The resulting failure length for the four cases is summarised in Table 2.2.
Table 2.2: Results for Expected Failure Length at Pontal Dike 2

Tabulation for Lf/Ld
Hinital = 20 m Hnet = 14 m
SOF/Hd = 15 0.5 0.35
SOF/Hd = 10 0.33 0.24
Tabulation for Lf (m)

Hinital = 20 m Hnet = 14 m
SOF/Hd = 15 300 210
SOF/Hd = 10 198 144
Page 4 of 15
VALE S.A.
Based on the results shown in Table 2.2, and placing greater weight on the results for the
current net height of Dike 2 of 14 m, a conservative failure length Lf = 200 m was selected
from the range of results and used for dam break analysis. This geometry is applicable for the
current net height of Dike 2 following construction of the stabilising toe weight.
It is noteworthy that the CDA Bulletin on Tailings Dam Breach Analysis for Case 2A does not
recommend considering the full length of the dam for breach analysis purposes. Indeed if the
CDA guidance was followed it would suggest much smaller breach dimensions than selected
by Fugro based on the work summarised by Hicks and Li (2018). Adopting an assumed length
of breach 200 m allows for the theoretical and numerical studies in the literature and the
evidence from relevant case histories. In this respect, Fugro has been more conservative than
would be suggested by the CDA guidelines. Fugro nevertheless considers that the proposed
approach is more complete and provides a logical and prudent basis on which to select a value
for the possible dam breach length for the current risk assessment purposes. Of course, if a
much longer SOF can be determined for a particular tailings storage dike than would be
anticipated by the empirical approach, then that larger value should be taken into account and
the conservative assumption would be to assumed Lf = SOW.
3. Empirical Evidence for the Angle to Define the Runout

Volume
The same set of Case 2A upstream tailings dam failures were examined to determine the
furthest distance of runout from the dam axis, following dam breach, and thereby to compute
the average angle between the base of the runout volume at the dam axis and the elevation
of tailings at the furthest point away from the dam where tailings participated in the runout
failure. The case history data (it was not possible to find all the required inputs for all of the
case histories) were then compared with the typical range proposed in the CDA guidelines, a
range 3⁰ to 5⁰ to the horizontal (Martin et al 2019).
The comparison is shown on Plate 3.1. Based on these findings, a representative angle 4⁰ to
the horizontal was adopted to define the volume of tailings that would run out of the dam
breach, starting from the elevation of the breach at a location close to the rear of the dam
section.
3.1 Comment on the cross section of Dike 2

The geometry of Dike 2 is shown on Plate 3.2 in cross section. This cross section represents
relatively robust dike construction reminiscent of centreline dike cases. Only one feature of
upstream construction is noticeable at the dam crest; but this final dam raising does not retain
a significant thickness of tailings but rather must provide margin against flood water
overtopping, or a similar purpose. The Dike 2 structure appears considerably more robust than
is the usual case for dikes classified as upstream construction.
Page 5 of 15
VALE S.A.
3.2 Comment on continuing movement at Pontal Dike 2

The evidence from InSAR on ground movement over time is useful for risk assessment for both
the dam and the tailings retained. In the case histories of tailings dike failure there is always
measurable continuing movement of both the dike and the tailings that are retained. See the
InSAR studies for the Brumadinho and Cadia tailings dike failures, for example.
The question remains whether the measured movement data from InSAR can be used to
predict imminent failure. In the case of Brumadinho, perhaps yes, in hindsight, but practically
no as a reliable forward indicator. In the case of Cadia the answer is almost certainly yes.
It is the change in measured dam and tailings movement, indicative of deformation, strain,
consolidation or creep in tailings materials, and the response of the dam to changing load,
which would signify imminent failure; an acceleration in the rate of movement is what is sought
as an indication of imminent danger of collapse.
If there is no or very little movement in the tailings or dike structure, as is the case indicated
by the InSAR data for Pontal Dike 2, then internal deformation and creep mechanisms that
might be the source of sudden, brittle static liquefaction of the dam system cannot reasonably
be expected to be occurring. Further, if the dike and retained tailings do not move as a result
of changing water pressure conditions between the wet and dry season, this again must be
indicative of an adequately stable structure and system unlikely to be developing internal
mechanisms that could reduce stability and trigger sudden static liquefaction and failure.
While not reliable as a single indicator of stability, this observation of very little to no movement
at Dike 2 Pontal, even between the wet and dry seasons, is supportive of the notion that the
dike currently (following construction of the toe weight) has a reasonable margin of safety.
A reasonable margin of safety would be defined as mobilising a proportion of the available

shearing resistance in the tailings and potentially critical zones of natural soil in the dike such
that it does not cause continuing creep movement, and can support temporary increase in
water pressure (during the rainy seasons) without incremental and localised tailings
deformation or creep resulting.
The InSAR data for Pontal Dike 2 is available to Vale through a protected online portal where
the measured movement of Dike 2 and the retained tailings can be examined over any selected
period of time between May 2015 and January 2020. Some plots of the data for Dike 2 are
shown on Plate 3.3 as an illustration.
Page 6 of 15
VALE S.A.
4. Geotechnical Evidence for Continuous Loose Layers in

Tailings at Dike 2
The initial runout studies for Dike 2 considered two scenarios for dam breach:
 A worst-case scenario based on a failure mechanism with a base at 776 m elevation
 A credible scenario based on a failure mechanism with a base at 780 m elevation,

coinciding with the base of the dike.
The shape, depth and extent of the potential failure mechanism at Dike 2 depends on the
strength properties and strength distribution in the surrounding tailings. In particular, the
presence of continuous, weak and potentially liquefiable layers at depth below the dike could
result in a deeper failure surface with a greater volume of tailings being mobilised.
Critical State Soil Mechanics (CSSM) provides a framework to detect the presence of such
layers: weaker layers, looser than critical, show contractile behaviour and generate positive
excess pore pressures upon undrained shear loading. As a result the undrained shear strength
is lower (often substantially) than the drained shear strength.
The state parameter ψ (Jefferies and Been, 2016) is defined as:
𝜓=𝑒−𝑒
where e is the in situ void ratio and ec is the void ratio at critical state at the same effective
mean pressure.
Within a soil or tailings profile, layers with ψ>-0.05 are expected to correspond with contractile
behaviour (Jefferies and Been, 2016), such that liquefied and unfavourable undrained shear
strength properties would apply for stability assessment.
The state parameter can be derived based on the collected CPTu data using screening
procedures described by Plewes (1992), Robertson (2009) and Jefferies and Been (2016). These
different methods were applied to the available CPTu tests for Dike 2. All three methods give
very comparable results for the extent of contracile layers of tailings at each CPTu test location.
The Jefferies and Been (2016) approach was selected for further analysis and interpretation as
it also makes use of the pore water pressure data through the pore pressure parameter Bq. This
additional data on the soil condition is not considered in the other methods
Six cross-sections were defined at Dike 2 as shown on Plate 4.1:
 Three longitudinal cross sections:
o Section L1 (CPTU-02 through CPTU-06) is in the downstream part of the dike
o Section L2 (CPTU-08 to CPTU-10) is at the crest of the dike
o Section L3 (CPTU-11 to CPTU-13) is about 300 m upstream of Section L2
Page 7 of 15
VALE S.A.
 Three perpendicular cross-sections:
o Section P1 (CPTU-01 to CPTU-11) is on the western side of the dike
o Section P2 (SCPTU-03 to CPTU-12) is in the central section of the dike
o Section P3 (CPTU-05 to CPTU-13) is in the eastern side of the dike
The cone resistance and the derived state parameter profiles for each cross-section are
presented on Plates 4.2 to 4.7.
The following conclusions may be drawn from the data on these cross-sections:
 Several CPTs indicate the presence of loose, liquefiable to an elevation 776 m, or even
deeper. The CPTUs tests exhibiting looser tailings are predominantly located in the
South-Western part, towards the left abutment of Dike 2
 However, the zones of loose and liquefiable material do not appear to be continuous
across the dike, neither laterally nor in the perpendicular direction
 There are no significant continuous layers of loose tailings at depth.
Based on the above geotechnical assessment of the available data, and the current geometry
for Dike 2 with a strong layer of base material (Itabrica) to a depth 6 m below the downstream
tailings and toe weight, Fugro considers it is reasonable to assess the most likely dam breach
extending to just below the strong and wide base of Dike 2 at elevation 780 m.
A significant and continuous weak layer would be required at greater depth to compensate for
the extra resistances and energy dissipation that would occur in a larger and deeper failure
mechanism. Fugro considers the evidence for deeper failure as the basis for risk assessment
from dam break is insufficient to justify that as a reasonably expected case.
5. Presence of Residual Soil at Pontal Dike 2 Limiting the

Maximum Possible Breach Dimensions
In all instances, liquefaction of the original natural terrain underlying the tailings seems
extremely unlikely and can be excluded for runout modelling purposes. Therefore the presence
of natural terrain at the base elevation of the anticipated dam break would provide an upper
bound for the maximal breach length.
The maximum thickness of tailings at Dike 2 extends from zero on the left and right abutments
to a maximum of ~55 m near the dike centre, as shown on Plate 5.1. For the deepest anticipated
potential failure elevation 780 m, just below the strong base material forming Dike 2, the width
separating the natural terrain at the left and right abutments is equal to ~318 m. This
dimension is therefore considered as the widest possible breach width for any failure at an
elevation 780 m on the dike axis.
Page 8 of 15
VALE S.A.
The natural terrain has a roughly triangular shape above the base elevation at 780 m and
extending up to the dike crest, Plate 5.2. On each side of the base width of the breach, the
most conservative assumption would be to consider that failure proceeds at the interface
between the retained tailings and the natural terrain, Plate 5.2. However, it is more likely that
the breach sidewalls originating from a dike failure at elevation 780 m would form a much
steeper angle, thereby limiting the breach extension laterally.
Two views of the failure thickness associated with two scenarios are shown on Plate 5.3. To the
left side, the most unfavourable failure with a base elevation 780 m, has a maximum width 318
m at its base, and with sidewalls following the interface with the natural terrain. On the right
side of Plate 5.3, is the more credible but less conservative failure scenario considered by Fugro
as a base case; a breach width of 200 m (at the same elevation of 780 m) and steep side angles
within the remaining, intact dike length.
6. In situ Water Content, Rheometer Data and Selection

of Cautious Parameters for Run out Analysis
The rheological properties were evaluated based on tests performed on 30 samples, taken from
3 different depths from 2 different points. Each sample was tested at 5 different water contents.
The points where samples were collected were P12 and P15, which are the same points where
CPT tests were performed.
Rheological tests were conducted in a decreasing rate of strain rate. Strain rates used were:
0.3/63.1/55.2/47.3/31.5/23.7/15.8/7.9/0.04 (1/s).
Samples were intended to be tested with water content in the range 14/18/22/26/30%.
However, due to high strength and stiffness of samples at lower water content, the water
content used for rheological testing was larger than initially considered.
Due to sedimentation which occurred very quickly in the samples prepared, at each step of the
test samples were removed from the rheometer and re-homogenised prior to beginning the
test at the next strain rate.
For each strain rate applied, the shear stress was measured with time until stabilization. The
shear stress for each shear strain rate was taken as the minimum value obtained, as indicated
on Plate 6.1.
After all tests were concluded, a plot with all samples tested and in terms of shear stress vs
shear strain rate was prepared, as indicated on Plate 6.2.
For small strain rates, higher shear stress was measured. This behaviour is attributed to the
grain size distribution of the samples, which largely comprise fine sand with silt content in the
range 10 to 40%.
Page 9 of 15
VALE S.A.
As the maximum shear strain rates determined from numerical modelling of run out at Dike 2
are in the range between 10 and 60 s-1, the test data was evaluated for this applicable range,
as shown on Plate 6.3. The measured shear resistance of the tailings is only slightly dependent
on the shear strain rate, which indicates a low viscosity of the order 1 to 2 Pa.s.
As the shear stress is almost constant, the yield stress for each sample was defined as the
average of shear stress value over the range of strain rate from 10 to 100 s-1.
The values of yield stress for each sample were plotted against the water content. The fitted
curve and yield stress as a function of water content are shown on Plate 6.4.
As the relevant yield stress for the tailings depends on the water content in situ, an estimation
of in situ water content was made based on the relative density estimated on CPTu tests.
For 16 CPTu tests performed on Dike 2 reservoir, water content was estimated as follow:
 Relative density was estimated based on the corrected cone resistance qt
 Void ratio was estimated as: 𝑒 = 𝑒 − 𝐷𝑟(𝑒 −𝑒 )

.
 Water content was estimated as: 𝑤 =
 Considering that the material is saturated, S = 1. Specific gravity for Dike 2 was
considered as 3.33, as the average value of all samples tested for points P12, P15 and
P18
 Yield stress was estimated over depth applying the power curve equation to water
content: 𝑦 = 3. 10 . 𝑤 ,
 An average yield stress was defined for each one of the 16 points as the average value
with depth, from the ground surface to 20 m.
The yield stress versus depth estimated for point P3 is shown on Plate 6.5.
The characteristic yield stress for the tailings was taken as the average value with depth for all
16 points evaluated. The average value obtained was 194 Pa, rounded to 200 Pa.
A strength-degradation model is incorporated in the model to account for the expected brittle
behaviour of the tailings material upon liquefaction, whereby the shear resistance reduces from
an initial intact geostatic shear strength, assumed conservatively as 16 kPa from geotechnical
data review, down to the residual yield stress of 200 Pa as soon as the accumulated shear strain
exceeds 1%.
The viscosity was considered as 1 Pa.s.
7. Results of Runout Modelling Scenarios

Runout modelling was performed using Fugro’s two-layer numerical modelling software, which
adopts a different treatment for the tailings, considered as Bingham non-Newtonian material,
Page 10 of 15
VALE S.A.
and Pontal lake water, considered as a classical Newtonian liquid. The simulations were
performed adopting the rheological properties summarized in Section 6.
The results are discussed in terms of the predicted discharges and volumes of water and tailings
predicted to overflow the Pontal dam crest. Results are presented here for a total of 10
simulations. All of these assume a failure base at elevation 780 m as described in Section 4. The
variations consider 5 different lake levels, respectively at 782, 781, 780, 779 and 778 m, and two
different failure scenarios, the first adopting a central, 200 m-wide breach with vertical walls,
with mobilized tailings volume of 1.944 million m³, and the second adopting a full width breach
at the basal elevation of 780 m, with breach sidewalls constrained to follow the level of the
pristine, natural terrain, with mobilized tailings volume of 2.019 million m³.
The volumes of failure under consideration are shown on Plate 5.3. Note that for the 200 m-
wide breach, the runout volume is not constrained by the original natural terrain, not even in
the soils behind the dike; this results in slightly overestimated runout volumes immediately
upstream of the dike, especially toward the western abutment as visible on Plate 5.3, which is
an additional conservatism in the analysis.
Snapshots of the simulations at various instants for a lake level at 780 m are shown on Plate 7.1
for the 200 m and full-width breach cases.
All simulations predict no overflow of tailings material above the Pontal dam crest. Several
simulations predict a degree of overflow of lake water, however. The overflow volumes,
maximum overflow discharges and hydrograms are illustrated on Plate 7.2 for the 200-wide
breach, and on Plate 7.3 for the full-width breach. As can be seen, the overflow of water, when
predicted, is limited, and lasts only a few tens of seconds.
The magnitude of such water overflow depends strongly on the lake level, and is significant for
a lake level 782 or 781 m. For a lake level of 780 m, however, the overflow reduces to a
negligible value or vanishes, both for the partial-width and full-width breaches. The full width
breach results only in a slight increase of water overflow for the highest lake levels. The
summary of simulations results are presented on Table 7.1.
Page 11 of 15
VALE S.A.
Table 7.1: Summary of scenarios results
Volume, Water [m³]

Failed volume [m³]
Discharge, Tailings
Initial Yield Stress
Discharge, Water
95% degradation
Volume, Tailings
Failure base [m]
Shear strain for
Degraded Yield
Total Overflow
Total Overflow
Viscosity [Pa.s]
Lake level [m]
Max Overflow
Max Overflow
Stress [Pa]
Scenario
[m³/s]
[m³/s]
[m³]
[Pa]
16000 200 0.01 1 778 780 1.944e6 7.8 0 2.30 0
16000 200 0.01 1 779 780 1.944e6 9.9 0 2.86 0

200m
16000 200 0.01 1 780 780 1.944e6 15.9 0 3.53 0
breach
16000 200 0.01 1 781 780 1.944e6 463.4 0 55.1 0
16000 200 0.01 1 782 780 1.944e6 999.4 0 85.9 0
16000 200 0.01 1 778 780 2.019e6 0.1 0 0.03 0
16000 200 0.01 1 779 780 2.019e6 0.8 0 0.19 0

Full-
width 16000 200 0.01 1 780 780 2.019e6 4.8 0 0.46 0
breach
16000 200 0.01 1 781 780 2.019e6 215.4 0 37.0 0
16000 200 0.01 1 782 780 2.019e6 1151.2 0 60.7 0
8. Conclusions
Evidence has been provided for the selection of an assumed 200 m dam breach width for
Pontal Dike 2 for the purposes of assessing the consequences of tailings runout. This length
is considered to be reasonable taking account of both the theoretical and numerical analyses
for long slope and embankment structures, and the available database of comparable
upstream tailings dam failures of Case 2A (CDA guidelines).
Examination of the available geotechnical data for Dike 2 does not provide evidence that the
dam breach would extend deeper than to pass beneath the strong base of the cross section at
elevation 780 m, 6 m below the level of the tailings and toe weight downstream.
The available case history data for Case 2A tailings dam failures was examined and compared
with the current CDA guidelines to estimate the volume of tailings that could be expected to
run out from a dam breach. This comparison supports the selection of a plane at 4⁰ to the
horizontal, extending from the base of the dam breach, to define the volume of tailings that
would run out.
As requested most recently, Fugro has also considered the case of a greater dam breach width
extending between the natural ground at the two abutments of Dike 2. This defines a breach
width 318 m at an elevation 780 m. As a further conservatism, side slopes in the breached
tailings were assumed to follow natural terrain at the left and right abutments, although steep
sides in the breached tailings dams are normally observed in practice. As described in Section
Page 12 of 15
VALE S.A.
5, Fugro has considered the widest possible breach that is defined by the limiting natural terrain
extended up to the top of the tailings at both abutments.
The volume of tailings mobilized on both scenarios are quite equivalent.
The results of the run out analyses for both a central 200 m wide breach and for a full-width
breach constrained by natural terrain have been presented in this technical note. A sensitivity
analysis considering different water levels from 778m to 782m in the lake behind the Pontal
Dam has been performed. The simulations predict no overflow of tailings for either assumed
dam breach case and indicate some overflow of lake water that reduces to practically zero for
a lake water level at 780 m or less.
Page 13 of 15
VALE S.A.
9. References
Hicks, M.A. and Li, Y. (2018) Influence of Length Effect on Embankment Slope Reliability in 3D,
International Journal for Analytical and Numerical Methods in Geomechanics, Vol 42, 891-915
Jefferies, M.G. and Been, K. (2016) Soil Liquefaction: a Critical State Approach
Martin, V., Al-Mamun, M. and Small, A. (2020) CDA Technical Bulletin on Tailings Dam Breach
Analyses
Plewes, H.D., Davies, M.P. and Jefferies, M.G. (1992) CPT Based Screening Procedure for
Evaluating Liquefaction Susceptibility, Innovation Conservation and Renovation
Page 14 of 15
VALE S.A.
List of Plates
Title Plate No.
A copy of Figure 13 from Hicks and Li (2018) showing Fugro’s assumption for tailings dams. 2.1
Case records of upstream tailings dam failures of CDA type Case 2A 2.2
Contours of constant SOF/Hd compared with Case 2A failure data. 2.3
Pontal Dike 2 cases with Hnet = 14 m (current conditions) and Hinitial = 20 m 2.4
Empirical data and comparison with CDA recommendations. 3.1
Cross-section for Dike 2 Pontal 3.2
InSAR data of Dikes 2 and 3 3.3
Selected Cross-sections at Pontal Dike 2 4.1
State Parameter and Cone Resistance Profile Per Cross-section 4.2 to 4.7
Thickness of Tailings (m) between Present Day Terrain and the Original 1966 Topography 5.1
Cross-section of Present Day and Original (1966) Terrain across Dike 2 Crest 5.2
Thickness of Failure for a Central 200 m Wide Breach with Vertical Walls (Left) and for a Full Width Brreach
5.3
Constrained by Original Terrain. Failure Base Set at Elevation 780 m in both Cases
Shear Stress Versus Time 6.1
Shear Stress versus Shear Strain Rate 6.2
Shear Stress versus Shear Strain Rate from 10 to 100 s-1 6.3
Yield stress versus Water Content 6.4
Yield stress estimated for P3 6.5
Snapshots of the Tailings Runout at Selected Instants after Initial Failure by Liquefaction (from Top to
Bottom, Respectively after 0 s, 30 s, 60 s, 120 s, 180 s and 240 s). Left: 200 m Wide, Central: Breach with 7.1
Vertical Sidewalls, Right: Full Width
Overflow properties at Pontal dam crest for the 200-m wide breach.
Top: maximum overflow discharge of water as a function of lake water level
7.2
Middle: total overflow volume of water as a function of lake water level.
Bottom: overflow hydrogram versus simulation time for a lake level of 780 m
Overflow properties at Pontal dam crest for the full-width breach.

Top: maximum overflow discharge of water as a function of lake water level
7.3
Middle: total overflow volume of water as a function of lake water level.
Bottom: overflow hydrogram versus simulation time for a lake level of 780 m
Page 15 of 15
VALE S.A.
FUGRO PROPOSED BASIS TO COMPUTE TAILINGS DAM BREACH DIMENSIONS SHOWN ON

FIGURE 13 FROM HICKS AND LI (2018)
Plate 2.1
VALE S.A.
CASE RECORDS OF UPSTREAM TAILINGS DAM FAILURES OF CDA TYPE CASE 2A
Plate 2.2
VALE S.A.
SOF = Scale of Feature causing inhomogeneity such as a weak soil layer,

greater slope steepness or height, larger water pressure, etc.
Hd = Height of Dam
CONTOURS OF SOF/Hd COMPARED WITH CASE 2A FAILURE DATA
Plate 2.3
VALE S.A.
Bounding Limit with SOF/Hd = 15 (to include Sullivan Dam case history)
Suggested Limit with SOF/Hd = 10 bounding all data except Sullivan Dam
For Pontal Dike 2: net height Hnet = 14 m (current elevation difference)

maximum height Hinitial = 20 m (dike base to top of tailings)
PONTAL DIKE 2 CASES WITH Hnet = 14 m (CURRENT CONDITIONS) AND Hinitial = 20 m
Plate 2.4
VALE S.A.
EMPIRICAL DATA AND COMPARISON WITH CDA RECOMMENDATIONS
Plate 3.1
VALE S.A.
CROSS-SECTION FOR DIKE 2 PONTAL
Plate 3.2
VALE S.A.
Pontal Dikes 2 and 3 – movement between 01/06/2015 and 24/01/2020.
Pontal Dikes 2 and 3 – movement between 05/01/2019 and 24/01/2020.
Pontal Dike 2 – detailed view of movement between 05/01/2019 and 24/01/2020.
EXAMPLES OF INSAR DATA FOR PONTAL DIKE 2
Plate 3.3
VALE S.A.
SELECTED CROSS-SECTIONS AT PONTAL DIKE
Plate 4.1
Distance (m)
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340 360 380 400 420
800
780
Elevation [m]
760
740
State Parameter
Cone Resistance
CPTU-2a CPTU-03a SCPTU-03 CPTU-05_2 CPTU-06
720
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
State Parameter, [-] State Parameter, [-] State Parameter, [-] State Parameter, [-] State Parameter, [-]
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Cone Resistance [MPa] Cone Resistance (MPa) Cone Resistance (MPa) Cone Resistance (MPa) Cone Resistance [MPa]
Fugro Report No. 147148-01
Plate No. 4.2
Distance (m)
0 50 100 150 200 250 300 350 400
800 800 800
790 790 790
780 780 780
Elevation [m]
Elevation [m]
Elevation [m]
770 770 770
760 760 760

CPTU-09
SCPTU-02
State Parameter
CPTU-09 and SCPTU-02 CPTU-10
CPTU-08a Cone Resistance
750 750 750

-0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5 -0.5 -0.25 0 0.25 0.5
State Parameter, [-] State Parameter, [-] State Parameter, [-]
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Cone Resistance [MPa] Cone Resistance (MPa) Cone Resistance [MPa]
Plate No. 4.3
Distance (m)
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300 320 340
800 800
790 790
Elevation [m]
Elevation [m]
780 780
770 770
State Parameter
Cone Resistance
CPTU-12a CPTU-13
CPTU-11a
760 760
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
State Parameter,[-] State Parameter,[-] State Parameter, [-]
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Plate No. 4.4
Distance (m)
0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300
800 800
790 790
780 780
Elevation [m]
Elevation [m]
770 770
760 760
State Parameter
Cone Resistance
CPTU-1_2 CPTU-02a CPTU-8a CPTU-11a
750 750
-1 -0.5 0 0.5 1-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
State Parameter, [-] State Parameter, [-] State Parameter,[-] State Parameter, [-]
0 10 20 30 400 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Cone Resistance [MPa] Cone Resistance (MPa) Cone Resistance [MPa] Cone Resistance (MPa)
Plate No. 4.5
Distance (m)
0 20 40 60 80 100 120 140 160 180 200
800 800
780 780
Elevation [m]
Elevation [m]
760 760
740 740
State Parameter
Cone Resistance
SCPTU-03 CPTU-07 CPTU-09 CPTU-12a
720 720
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
State Parameter, [-] State Parameter, [-] State Parameter, [-] State Parameter, [-]
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Cone Resistance [MPa] Cone Resistance (MPa) Cone Resistance (MPa) Cone Resistance [MPa]
Plate No. 4.6
Distance (m)
0 20 40 60 80 100 120 140 160 180 200
800 800
790 790
780 780
Elevation [m]
Elevation [m]
770 770
760 760
State Parameter
Cone Resistance
CPTU-05_2 CPTU-10 CPTU-13
750 750
-1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1 -1 -0.5 0 0.5 1
State Parameter, [-] State Parameter, [-] State Parameter, [-]
0 10 20 30 40 0 10 20 30 40 0 10 20 30 40
Plate No. 4.7
VALE S.A.
THICKNESS OF TAILINGS (M) BETWEEN PRESENT DAY AND THE ORIGINAL 1966 TOPOGRAPHY
Plate 5.1
VALE S.A.
CROSS-SECTION OF PRESENT DAY AND ORIGINAL (1966) TERRAIN ACROSS DIKE 2 CREST
Plate 5.2
VALE S.A.
1.944 million m³ 2.019 million m³
THICKNESS OF FAILURE FOR A CENTRAL 200 M WIDE BREACH WITH VERTICAL WALLS (LEFT)
AND FOR A FULL WIDTH BREACH CONSTRAINED BY ORIGINAL TERRAIN. FAILURE BASE SET AT
ELEVATION 780 M IN BOTH CASES
Plate 5.3
VALE S.A.
SHEAR STRESS VERSUS TIME
Plate 6.1
VALE S.A.
1000 D21.00
D20.51
D20.94
D21.46
D22.00
A17.26
A18.51
A22.11
A22.44
Shear stress (Pa)
A22.78
B21.34
B22.1
100
B23.1
B27.27
B29.06
E18.73
E20.01
E22.41
E23.84
E24.87
E25.00
F20.30
F22.29
10 F22.63
0.01 0.10 1.00 10.00 100.00 F23.00
Shear strain rate (1/s) F25.00
SHEAR STRESS VERSUS SHEAR STRAIN RATE
Plate 6.2
VALE S.A.
1000 D21.00
D20.51
D20.94
D21.46
D22.00
A17.26
A18.51
A22.11
A22.44
Shear stress (Pa)
A22.78
B21.34
B22.1
100
B23.1
B27.27
B29.06
E18.73
E20.01
E22.41
E23.84
E24.87
E25.00
F20.30
F22.29
10 F22.63
10 100 F23.00
Shear strain rate (1/s) F25.00
SHEAR STRESS VERSUS SHEAR STRAIN RATE FROM 10 to 100 s-1
Plate 6.3
VALE S.A.
800
700
600
Yield stress (Pa)
500
400
300
200
100
0
10 15 20 25 30 35
Water content (%)
YIELD STRESS VERSUS WATER CONTENT
Plate 6.4
VALE S.A.
P3
Yield stress (Pa)
50 150 250 350 450 550 650 750
0
8
Depth (m)
10
12
14
16
18
20
YIELD STRESS ESTIMATED FOR P3
Plate 6.5
VALE S.A.
SNAPSHOTS OF THE TAILINGS RUNOUT AT SELECTED INSTANTS AFTER INITIAL FAILURE BY

LIQUEFACTION (FROM TOP TO BOTTOM, RESPECTIVELY AFTER 0 s, 30 s, 60 s, 120 s, 180 s and
240 s). LEFT: 200 m WIDE. CENTRAL: BREACH WITH VERTICAL SIDEWALLS. RIGHT: FULL WIDTH
BREACH WITH SIDEWALLS CONSTRAINED BY ORIGINAL TERRAIN. TAILINGS COLOUR, DARK
TO LIGHT, IS ACCORDING TO LOCAL FLOW VELOCITY
Plate 7.1
VALE S.A.
OVERFLOW PROPERTIES AT PONTAL DAM CREST FOR THE 200 m WIDE BREACH.
TOP: MAXIMUM OVERFLOW DISCHARGE OF WATER AS A FUNCTION OF LAKE WATER

LEVELMIDDLE; TOTAL OVERFLOW VOLUME OF WATER AS A FUNCTION OF LAKE WATER LEVEL
BOTTOM: OVERFLOW HYDROGRAM VERSUS SIMULATION TIME FOR A LAKE LEVEL OF 780 m
Plate 7.2
VALE S.A.
OVERFLOW PROPERTIES AT PONTAL DAM CREST FOR THE FULL-WIDTH BREACH.
TOP: MAXIMUM OVERFLOW DISCHARGE OF WATER AS A FUNCTION OF LAKE WATER LEVEL

MIDDLE: TOTAL OVERFLOW VOLUME OF WATER AS A FUNCTION OF LAKE WATER LEVEL.
BOTTOM: OVERFLOW HYDROGRAM VERSUS SIMULATION TIME FOR A LAKE LEVEL OF 780 m
Plate 7.3

C147148-01 Technical Note On Geotechnical Aspects and Runout Modelling Scenarios R02

Uploaded by

Copyright:

Available Formats

C147148-01 Technical Note On Geotechnical Aspects and Runout Modelling Scenarios R02

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

C147148-01 Technical Note On Geotechnical Aspects and Runout Modelling Scenarios R02

Uploaded by

Copyright:

Available Formats

Note on Geotechnical Aspects and Runout

Modelling Scenarios, Dike 2 Pontal

Fugro Project No. C147148

Fugro Document No. C147148-01

Issue Status For Review

Fugro Legal Entity Fugro Belgium

Issuing Office Address Avenue de Broqueville 12 – 1150 Brussels, Belgium

Client Contact Willyan G. Debastiani

Client Document No. C147148

Initials Name Role

BSP Benoit Spinewine Principal Engineer

MRA Mikaël Ramos da Silva Geotechnical Group Lead

PTH Pedro Tha Engineering Manager

RJE Richard Jewell Principal Geotechnical Consultant

Willyan G. Debastiani, VALE S.A.

Dear Willyan Debastiani,

Please do not hesitate to ask for any additional information or clarification.

7. Results of Runout Modelling Scenarios 10

Tables in the Main Text

1.1 About this technical note

A proposed semi-empirical method is suggested as the basis to determine reasonably

2. Empirical Evidence and Basis to Define a Cautious

2.2 Case history data for tailings dam Case 2A failures

2.3 Proposed criterion by Fugro to bound the observed failure data

2.4 Application to Pontal Dike 2

Table 2.2: Results for Expected Failure Length at Pontal Dike 2

Tabulation for Lf (m)

3. Empirical Evidence for the Angle to Define the Runout

3.1 Comment on the cross section of Dike 2

3.2 Comment on continuing movement at Pontal Dike 2

A reasonable margin of safety would be defined as mobilising a proportion of the available

4. Geotechnical Evidence for Continuous Loose Layers in

 A worst-case scenario based on a failure mechanism with a base at 776 m elevation

 A credible scenario based on a failure mechanism with a base at 780 m elevation,

The state parameter ψ (Jefferies and Been, 2016) is defined as:

Six cross-sections were defined at Dike 2 as shown on Plate 4.1:

 Three longitudinal cross sections:

o Section L1 (CPTU-02 through CPTU-06) is in the downstream part of the dike

o Section L2 (CPTU-08 to CPTU-10) is at the crest of the dike

o Section L3 (CPTU-11 to CPTU-13) is about 300 m upstream of Section L2

 Three perpendicular cross-sections:

o Section P1 (CPTU-01 to CPTU-11) is on the western side of the dike

o Section P2 (SCPTU-03 to CPTU-12) is in the central section of the dike

o Section P3 (CPTU-05 to CPTU-13) is in the eastern side of the dike

 There are no significant continuous layers of loose tailings at depth.

5. Presence of Residual Soil at Pontal Dike 2 Limiting the

6. In situ Water Content, Rheometer Data and Selection

 Relative density was estimated based on the corrected cone resistance qt

 Void ratio was estimated as: 𝑒 = 𝑒 − 𝐷𝑟(𝑒 −𝑒 )

The viscosity was considered as 1 Pa.s.

7. Results of Runout Modelling Scenarios

Table 7.1: Summary of scenarios results

Volume, Water [m³]

Lake level [m]

16000 200 0.01 1 778 780 1.944e6 7.8 0 2.30 0

16000 200 0.01 1 779 780 1.944e6 9.9 0 2.86 0

16000 200 0.01 1 782 780 1.944e6 999.4 0 85.9 0

16000 200 0.01 1 778 780 2.019e6 0.1 0 0.03 0

16000 200 0.01 1 779 780 2.019e6 0.8 0 0.19 0