ICOLD Guidance for Slope Stability

Analyses of Dams
September 20, 2022
Part 01 – Context

Andy Small, Klohn Crippen Berger

Chair – ICOLD Committee L – Tailings and Waste Lagoons

1
Outline

• Context
• Overview of ICOLD Tailings Dam Safety Bulletin
• ICOLD Slope Stability Guidance
• Post Seismic Stability Analysis

• About 50 slides

2
Fundao Failure - 2015

• Fundao Dam Failure at Samarco Mine in 2015

• Contractive soils in shell of the dam
• Stability analysis assumed contractive behavior would not occur and used
drained parameters for FoS
• Overstated FoS and led to unsafe condition

3
Fundao Stability Analysis

• From Samarco Panel Report

4
Fundao Stability Analysis

• Right Abutment
• Drained parameters c and phi
• FOS = 1.91 – designers before the failure were using drained analyses and
stating the dam was safe. Stated no trigger for liquefaction.

5
Fundao Stability Analysis (cont’d)

• Peak undrained shear strength – Su/sv’ =0.25

• Undrained FOS = 1.00

6
Fundao Stability Analysis (cont’d)

• Post liquefaction undrained shear strength – Su/sv’ =0.07

• FOS = 0.40

7
Feijao Failure - 2019

• Near Brumadinho
• Closed tailings facility
• Rapid failure
• Contractive tailings in the shell
• Peak undrained FOS>1 (bonding)
• Post liq FOS<1

8
Call to Action

• Issue is not limited to Brazil.

• CDA initiated a Slope Stability Working Group in 2016
• Identified many examples in Canada and elsewhere where similar approaches
have been used.
• Concern about level of safety for many dams.
• Analyses by reputable and experienced engineers.
• Drove the need for comprehensive guidance.
• Developed guidance under CDA Banner until 2019.
• Transition to ICOLD.

9
ICOLD Guidance for Slope Stability
Analyses of Dams
September 20, 2022
Part 02 – Overview of ICOLD Tailings Dam Safety Bulletin
Andy Small, Klohn Crippen Berger

10
Tailings Dam Safety Bulletin – Writing Team

• Harvey McLeod (Canada)

• Annika Bjelkevik (Sweden), Rafael Monroy (UK), Paul Ridlen (USA), Duncan
Grant Stewart (South Africa), David Brett (Australia) Jiri Herza (Czech
Republic), Joaquim Pimenta de Avila (Brazil), Andy Small (Canada)
• Imran Gillani (co-opted member ICMM)

11
Tailings Dam Safety Bulletin – Context

• 2018- ICOLD Tailings Committee identified need for consistent international

guidance for tailings dam safety, commenced work in late 2019
• 2020 - Global International Standard for Tailings Management - mainly
governance aspects
• 2021 - ICMM developed a set of conformance protocols for GISTM
• 2021 - ICMM issued a Good Practice Guide for Tailings Management –
Governance and Implementation of Good Engineering Practice for Tailings
Management
• 2022- ICOLD Tailings Dam Safety Bulletin to be issued, supporting GISTM,
ICMM - provides technical guidance for tailings dam safety

12
Tailings Dam Safety Bulletin – Context

Presentation by Michael Davies, Teck,

2019

GISTM

ICMM, MAC
governance
guidance

ICOLD targeting
technical support

13
Bulletin 194 - Tailings Dam Safety– Outline

1. Introduction • Appendix A - Shear Strength Deformation Behaviour

2. Tailings Storage Facility Governance of Soils and Tailings
3. Closure • Appendix B - Stability Analysis Framework for
4. Dam Classification Tailings Dams with Contractive Soils
5. Site Characterization
6. Tailings Characterization
7. Design Overview of Bulletin given at ICOLD 2022
8. Risk Management
9. Dam Failure / Breach Analysis https://youtu.be/4q0ldq5HLxw

10. Emergency Preparedness and Response Planning

11. Construction
12. Operations

14
Tailings Dam Safety Bulletin – Section 7 - Design

7.1. Introduction
7.2. Life Phases and Design Stages of a Tailings Dam
7.3. Design Steps for a New Tailings Dam
7.4. Design of Raises and Ongoing Operations
7.5. Risk Informed Design
7.6. Dam Failure Modes
7.7. Design Basis
7.8. Design Criteria
7.9. Slope Stability Assessment
7.10. Earthquake Assessment (Seismic Stability)
7.11. Seepage Design
7.12. Hydrotechnical Design
7.13. Environmental Design

15
ICOLD Guidance for Slope Stability
Analyses of Dams
September 20, 2022
Part 03 – ICOLD Slope Stability Guidance
Andy Small, Klohn Crippen Berger

16
Tailings Dam Safety Bulletin – Section 7.9 - Slope Stability
Assessment
7.9. Slope Stability Assessment
7.9.1. Introduction
7.9.2. Slope Stability Assessment Methods
7.9.3. Limit Equilibrium Method
7.9.4. Target Factors of Safety for Limit Equilibrium Stability
Analyses
7.9.5. Stability Conditions
7.9.6. Post Liquefaction Stability Conditions
7.9.7. Residual Strength in Clay and “Clay Like” Tailings
7.9.8. Additional Stability Conditions
7.9.9. Adjustments to Target FOS
7.9.10. Three-Dimensional Stability L-E Analyses
7.9.11. Limitations of Limit Equilibrium Analyses
7.9.12. Safety Evaluation with Non-Linear Deformation
Analyses
7.9.13. Performance-Based Stability Evaluation

17
Slope Stability Assessment - General

• Assessment, not always analyses

• Limit equilibrium commonly used, but challenges
• Non-linear deformation analyses becoming more common
• Limit states design not in the Bulletin
• Supplement with performance

18
Slope Stability Assessment – Typical case

• Material Zoning & Geometry

• Material Properties
• Boundary Conditions
• Design Loads

19
Slope Stability Assessment - Considerations

• Consequences of failure
• Complexity
• Contractive/dilative
• Variability and uncertainty
• Comprehensiveness of site investigations and geotechnical monitoring
• Strain‐incompatibility of the different materials forming the dam and its
foundation
• Use of observational method

20
Strain Incompatibility

From KCB, 2015. Mount Polley Tailings Dam Failure, Assessment of Failure
Mechanism
21
Strain Incompatibility (cont’d)

• Three-dimensional deformation analysis

• Progressive failure
• Strain-weakening within a thin shear band in the glaciolacustrine unit at the
base of this slide (>5% sear strain)
• Failure extended into the core
• Shear resistance of rockfill delated because of relative ductility

• Limit equilibrium analyses cannot represent this

• Increase target FOS or discount strengths to account for this
From Zabolotinii, et al. 2022. Mechanism of Failure of the Mount Polley
Tailings Storage Facility. Canadian Geotechnical Journal, Volume 59, Number
8, August 2022.
22
Target Factor of Safety

Target Minimum Factor of

Stability Condition
Safety

Static Conditions 1.5

Post-Liquefaction Conditions 1.1

• Leading international practice for site characterization and parameter

selection
• For planned, operating, or closed dams
• For slip surfaces resulting in loss of freeboard, significant portion of dam
crest, or damage to critical elements
23
Slope Stability Assessment - Additional Stability Conditions

• End of construction
• Rapid drawdown
• Seismic and post seismic loading for dams with dilative soils

24
Slip Surfaces
• Target FoS intended for:
• Slip surfaces for a failure that involves the dam slope and potentially results in
uncontrolled release of the contained materials
• Slip surfaces that intercept a core and could destroy the effectiveness of the core
• Slip surfaces that prevent internal drainage elements from operating properly.

25
Rate of Failure

• Important component of guidance

• For saturated soils…
• Drained failure - rate of failure is slow enough that excess pore pressures
generated during the shearing dissipate, then the failure would occur in a
drained manner.
• Undrained failure - rate of failure is faster than the ability of the excess pore
pressures to be dissipated (i.e. rapid shearing)
• Most dams fail in undrained or partially undrained/drained conditions.

26
Slope Stability Calculation

Just before or at point of failure

27
Slope Stability Calculation (cont’d)

Just before or
at point of
failure

Shear Stress at time of failure

Also as: Shear strength at

failure/shear stress mobilized
along slip surface
28
Contractive Soil

• Typically loose
• When sheared slowly, such that there is no change in the pore water
pressure during shearing, the volume of the soil reduces - contracts
• When saturated and sheared quickly, such that the pore water pressures
cannot dissipate, then the tendency to contract causes an increase in pore
water pressures – reduces the strength by over 50%

29
Contractive Soil (cont’d)

• Soil grains are loosely packed

• Contact on edges
• Imagine water in openings (pore spaces)
• If sheared – collapse
• Sheared slowly – water drains, soil contracts
• Sheared quickly
• Water does not drain
• Collapse soil applies a pressure to the
pore water
• Reduces strength of soil by over 50%

30
Dilative Soil

• Typically dense and hard

• When sheared slowly, no pore pressure change, the grains of the soil shift
and cause an increase in volume – dilates
• When saturated and sheared quickly, pore pressures actually reduce
because of the dilation and strength increases

31
Dilative Soil (cont’d)

• Soil grains are tightly packed

• Imagine water in openings (pore spaces)
• If sheared – soil grains ride up
over each other
• Sheared slowly – water drains, volume increases
soil dilates
• Sheared quickly
• Water does not drain
• As soil grains ride over each other the water goes into tension
• Negative pore water pressure (below atmospheric)
• Increases strength of soil by up to 50%

32
Rate of Failure

• Rate of failure governs analysis

• Should explain the mechanism of failure and context in stability reports
• Clays and silty sands typically fail undrained
• Drained failure in clays possible through:
• Tertiary creep mechanism
• Layer at residual strength (presheared plane) – no change in pore pressure during
shear
• London clay excavations: drained failures (mobilizing drained fully softened shear
strengths) can take years or decades for drainage and failure to occur1

1Terzaghi, Peck, and Mesri, 1996, 3rd Edition of Soil Mechanics in Engineering Practice
33
Back to Dam Stability (cont’d)

• Sands and silts can fail undrained under rapid static loading or dynamic
loading
• Issue is whether they shear in a contractive manner or dilative manner

• If dense sand – same behaviour as dense clay – use the “drained strength”
as a conservatism

34
Slope Stability Assessment - Focus on Undrained Conditions

• Triggers
• Peak undrained strength in LEM for 1.5
• Residual strengths
• Post liquefaction (seismic or static)
• Appendix A - Shear Strength Deformation Behaviour of Soils and Tailings
• Appendix B - Stability Analysis Framework for Tailings Dams with
Contractive Soils
• Discuss later

35
Slope Stability Assessment - Adjustments to Target FOS

• Higher targets:
• Consequences
• Passive Care Closure
• Uncertainty
• Sensitive or strain softening soils
• Changes to soil properties over time
• Lower targets:
• Consequences
• Comprehensive implementation of observational approach
• Reduced uncertainty

36
 Stability Analysis Flow Chart – Static Loading
Condition

No further analysis
Define is required
ICOLD App B
geometry of the
dam, zone of Yes
tailings No
supporting the Yes Use peak Yes Use post peak
Contractive
dam, and the soil during
undrained FOS meets
strength. FOS meets
foundation. strengths and target? target?
undrained Establish target
Establish shearing? establish target FOS.
strength and pore FOS
pressure No
conditions No
ICOLD App B
Establish target FOS,
use peak strengths
 Stability Analysis Flow Chart – Seismic Loading
Condition

Define
geometry of the Conduct
dam, zone of No deformation Yes
tailings Soils that analysis.
Deformation
supporting the could liquefy Establish No further
Yes meets analysis is
dam, and the during seismic deformation target?
foundation. criteria(freeboard, required
loading are
Establish present? internal damage,
strength and pore etc.) No
pressure
conditions Yes No
Modify dam design
Use post liquefaction residual FOS meets
shear strength peak strengths target?
and meet target FOS of 1,1

Yes
No further
analysis is
required
Appendix A – Shear Strength and Deformation Behaviour

• Fundamental concepts elaborated in the appendix:

• Drained versus undrained shearing
• Dilative versus contractive behavior
• Strain-softening (weakening) versus strain-hardening behavior
• Brittle versus ductile behavior.

39
Appendix A – Shear Strength and Deformation Behaviour
(cont’d)
• Additional Guidance and Cautions
• CPT Based measurement of in situ state and soil properties
• Liquefaction and residual (post-liquefied) strength
• Selection of appropriate shear strength parameters for design and analysis
• Level of conservative appropriate to the level of uncertainty
• Pros and cons of laboratory versus field data
• Stress-dependent behavior (e.g., effect of OCR)
• Partial saturation
• Progressive failure
• Strain compatibility of dissimilar materials and other strain-related concerns

40
Appendix A – Shear Strength and Deformation Behaviour
(cont’d)
• Drained vs Undrained
Shearing
• Drained – constant
pore water pressure
shearing and volume
changes during
shearing
• Undrained – pore
pressures are induced
due to shearing,
volume is not
changed
41
Appendix A – Shear Strength and Deformation Behaviour
(cont’d)
• Dilative vs. Contractive behaviour -> density and stress control
strength

+ve pore water pressures

Effective stress path

Total stress path

-ve pore water pressures

42
Appendix A – Shear Strength and Deformation Behaviour
(cont’d)
• Brittleness during undrained shearing
• Bishop (1973) proposed the definition of a
brittleness index that, for an undrained case,
reads:

• Robertson (2020) suggests renaming as

“strength loss index” to recognize the
usefulness of IB as a measure of the
magnitude of strain softening.
43
Appendix B – Analysis Framework for Contractive Soils

• Presents a logical framework for selection of analysis techniques and

selection of an appropriate factor of safety when the minimum targets
cannot be met.
• For existing facilities, it may be necessary to accommodate existing
contractive and/or brittle materials in the structural zones. The
treatment of tailings dams when these materials are present should
consider the downstream consequences.
• Emphasizes the importance of the FOS calculated using post-liquefied
(residual) shear strengths over the static FOS.
Appendix B – Analysis Framework for Contractive Soils
(cont’d)
Appendix B – Analysis Framework for Contractive Soils
(cont’d)
• Cases 1 and 2 can support reducing target FOS with peak strengths to
1.3
• FOS >1.1 for post liquefaction, but strength loss not significant
Revised Guidance for Slope
Stability Analyses of Dams
September 20, 2022
Part 4 – Seismic Stability Analysis
Andy Small, Klohn Crippen Berger

47
Post Seismic and Post Peak

• Post Seismic Stability

• Dams that are built of compacted fill over competent foundation
• Non liquefiable zones
• No significant strain softening
• Use deformation analyses

• Post Peak Stability:

• Liquefiable zones – require use of post liquefaction residual shear strengths
• Clay – require consideration of post peak shear strengths

48
Common Practice

• Perform pseudo-static stability analysis as most regulatory requirements require

a factor of safety of one or higher for dams under the design seismic loading.
• If the calculated pseudo-static analysis factor of safety is less than one, some
engineers modify the dam design or design stabilization measures.
• Instead, the next step should be calculating deformations using simplified
methods to understand the expected deformations.
• If the calculated deformation using simplified methods is also not acceptable for
the dam, then advanced deformation modelling would be carried out.

49
Hynes-Griffin and Franklin (1984)

• Hynes-Griffin and Franklin (1984) proposed the following approach to

estimate the k value:
• k = 0.5 * PGA (where PGA is Peak Ground Acceleration)
• Bedrock acceleration – method uses PGA at bedrock
• The discount to 0.5 accounts for amplification and an upper bound displacement of 1
m
• Key point – 1 m!
• Many tailings dams
can take more than 1 m

• Coefficient of 0.5
can be modified
to be deformation specific (https://ascelibrary.org/doi/abs/10.1061/%28ASCE%29GT.1943-
5606.0001833)
50
Key Limitation of Pseudo-static Analysis

• Pseudo-static analysis and simplified deformation calculations are only

applicable for soils that do not show significant strength loss under the
design seismic event.

• Liquefaction susceptibility and cyclic softening potential of the soils should

be assessed before performing the pseudo-static and simplified
deformation analyses.

51
Pseudo-Static Analysis

• Movement away from this analysis as misunderstood, misused, and limited

value.
• ANCOLD and ICOLD are not referring to Pseudo-static analyses

52
Liquefaction

• Most common type of assessment for liquefaction potential under a seismic event is
based on the procedure developed by Youd et al. (2001) and Boulanger and Idriss
(2014).
• This method compares the Cyclic Stress Ratio (CSR) that is induced by the seismic event
to the seismic shear stress that is required to cause liquefaction (i.e. Cyclic Resistance
Ratio or CRR).
• If CSR is greater than CRR, then the soil could liquefy under that design seismic event.

• If liquefaction is possible, consider Appendix B cases.

53
Liquefaction Susceptibility – Example (cont’d)

• Calculate resistance during an earthquake – Cyclic resistance ratio (CRR)

54
Liquefaction Susceptibility - Example (cont’d)

• Calculate stresses in the tailings due to an earthquake – Cyclic stress ratio (CSR)

55
Liquefaction Susceptibility - Example (cont’d)

• Calculate factor of safety against liquefaction = CRR/CSR

56
Summary

• ICOLD Slope Stability Guidance is leading international practice

• Comply with local regulations
• But, check against ICOLD as well

57