LICUAFACCIÓN
LICUAFACCIÓN
LICUAFACCIÓN
September 2017
Nico Vermeulen 1
Disclaimer
The principle objective of this presentation is to elicit response, comment & debate
on the topic of static liquefaction in the context of tailings dams.
2
CONTENTS
1. Definitions
2. Static Liquefaction
3. Investigations
4. Analysis
5. Challenges
3
1. DEFINITIONS
4
Types of “Liquefaction”
Static Liquefaction
flow type failure as a result of strain softening accompanied by increased pore
pressure and a consequent reduction of the undrained shear strength
some researchers still adhere to a model that allows for the collapse of a meta-
stable fabric or packing (card house structure) of the soil skeleton (solid particles)
Mike Jefferies brought us the NorSand CSSM that is capable of modeling static
liquefaction within a sound engineering framework.
Many Triggers
Often heavy rainfall
can also be seismic 5
Dynamic / Cyclic Liquefaction
failure as a result of the cyclic build-up (ratchetting) of excess pore pressures in
response to repetitive loading or shaking typically associated with seismic events.
Repeated loading
Often earthquakes
6
‘Quick Sand Conditions’
boiling or quicksand type failure as a result of seepage forces overcoming the self
weight and frictional interaction between non-cohesive grains.
Hydraulic Gradient
Poor seepage control
7
‘Quick Clay Conditions’
Dispersion / Flocculation
Mechanical disturbance
8
2. STATIC LIQUEFACTION
9
Liquefaction
Behaviour of Soils = Composition x State
COMPOSITION STATE
Mineralogy
Size (PSD) Structure Stress-Density
Shape
Texture Fabric Void ratio / Density / State parameter
Critical State / Steady State Bonding Effective stress
Fiction angle Collapsibility Pore pressure
Cohesion Stress history / OCR
Dilatancy (Dilate & Contract)
Earth pressure
Elasticity / Plasticity / Yield / Hardening
Physical properties of particles Packing arrangement & glue History & Starting Point10
Computational Model
Critical State Soil Constitutive Models
Examples:
Granta Gravel (simplified sand)
Cam-clay (Clay)
Modified Cam-clay (Clay, numerical solutions) CSSM Model for Clays - Cambridge
NorSand (Sand & Silts -> Tailings)
11
Static Liquefaction: The Recipe for Disaster
1. Liquefaction SUSCEPTIBILITY
potential for undrained strain softening associated with a contractive soil state or low
density
2. Liquefaction TRIGGER
trigger mechanisms that may initiate strain softening (break the camel’s back):
• static or monotonic stress (static loading from self-weight or surcharge)
• cyclic stress (blast/construction vibrations, seismic/earthquake)
• deformation (undermining collapse, squeeze)
• Piping/boiling, overtopping, progressive sloughing . . .
3. Post-liquefaction STABILITY
extent of strain softening and loss of strength and the resultant impact on slope
stability, settlement, bearing capacity etc.
15
Desk Study
Seismic Data – Ground Acceleration (PPV)
Earthquake (Trigger?)
10% Probability of exceedance in 50 years, or
Maximum Credible Earthquake (Australia vs. Northern SA on the Rift Belt)
Be careful of the questions that you ask, you may not like the answers
Environmental Data
Water Balance
Rainfall
1:100 year 24hr event 16
In situ Investigations
Geotechnical Drilling
SPT – always good as a back-up check for liquefaction potential
Tube sampling (undisturbed, Shelby Tube, Mostap) for laboratory testing
Core extraction (disturbed samples) and visual logging
Piezometer installation
Sonic Drilling
Faster than conventional rotary core drilling
Good core recovery in soft tailings
Continuous casing of the borehole to prevent collapse
Allows SPT, tube sampling & piezometer installation
Can drill without water / drilling mud, i.e. ‘dry-drill’
18
Core recovery by Sonic Drilling
Note no inner sample sleeve used
19
CPT measures Soil STATE & BEHAVIOUR
Penetration Testing - SCPTu
CPT – cone resistance, sleeve friction (now common) & inclination
u – dynamic pore pressure and equilibrium pore pressure
S – shear wave velocity (should become common!!)
SCPTu
Multiple ‘liquefaction parameters’ are measure simultaneously and continuously
Many liquefaction analyses are based primarily on CPTu data
Shear wave velocity adds another level of interpretation (SPT CPT vs)
Reliable measurement of pore pressure regime in a tailings dam (critical)
Best bang for buck by far, should be the primary investigation tool
Window sampling now possible (Mostap)
Cannot penetrate hard or dense layers, or possibly all the way to the base of TSF
Seismic piezocone not as widely available in SA, or measures top-down
May still be perceived as ‘expensive’!?! 20
SCPTu with Dual/Multi Geophones
1 is good!
2 is better!!
21
Three independent shear wave velocity measurements
CPTu / SCPTu (single geophone) / CSW
Remarkable correlation!
Was this pure luck?
22
Piezometers
All TSF’s should have piezometers that are monitored as part of operations
Site records are typically poor, incomplete and not well understood by site staff
Piezometers are often lost or damaged as part of operations
Extensions to piezometers are not always properly recorded
Electronic Piezometers
Real time data to the cloud and available to all for analysis
Multi-level sensors -> pore pressure regime
Auto-trigger warning levels
Communications technologies allow long term remote sensing (10 year battery life)
Theft/vandalism no longer a major concern due to wireless technologies
Perceived as expensive
Limited ‘track record’
Data flooding 23
Laboratory Investigations
Classification
• SG – Specific Gravity
• PSD - Particle Size Distribution and fines content
• Atterberg Limits - influence of the fines
• Moisture content
• Confirm SBT & FC classifications from CPTu interpretation
Commercial Laboratories
Routine tests, reasonable chance of reliable & repeatable results
Can be done on disturbed but representative samples
Best source of composition data
Commercial Laboratories in SA
Specimen preparation and care are critical, lab manager needs to be aware
Special test procedures are often required for CSSM parameters
Regrettably results are not as reliable as they should be in SA
Capacity and testing time are major challenges
Research lab capacity is extremely limited for commercial testing 25
Fabric Effects on Reconstituted Samples
Chang et al. (2011) Hoeg et al. (2000): Silts & Tailings
26
CIU Triaxial Tests on the same tailings material
140
120
f’ = 19° f’ = 39°
Deviator stress (kPa)
100
80
60
40
20
0
0 50 100 150 200 250 300 350 400
• Electron micrographs – still the best way to see what you are dealing with
28
Walking on Water?
30
Slope Stability = Balancing Act
DISTURBING FORCES
FoS > 1
STABILISING FORCES
31
Olson & Stark (2003) – Case based
Based on undrained shear strength to the vertical effective stress ratio with depth
su/s’vo
Susceptibility:
Contractive/dilative boundary (‘simplification’ of Fear & Robertson, 1995 SPT)
′
𝜎𝑣𝑜 𝑏𝑜𝑢𝑛𝑑𝑎𝑟𝑦 = 1.10𝑥10−2 𝑞𝑐1 4.79
32
Triger Analysis:
𝑠𝑢,𝑦𝑖𝑒𝑙𝑑
𝐹𝑜𝑆𝑡𝑟𝑖𝑔𝑔𝑒𝑟 = FoS ≥ 1
𝜏𝑠𝑡𝑎𝑡𝑖𝑐 + 𝜏𝑠𝑒𝑖𝑠𝑚𝑖𝑐 + 𝜏𝑜𝑡ℎ𝑒𝑟
tstatic stress levels associated with the current static slope stability, i.e.
a static slope failure is a trigger for liquefaction
tseismic based on CSR & CRR, Seed, Youd & Idriss, 1971, 1997, 2001
𝛼𝑚𝑎𝑥 CSR = tseismic / s’vo
𝜏𝑠𝑒𝑖𝑠𝑚𝑖𝑐 = 0.65 𝜎𝑣𝑜 𝑟𝑑 𝐶𝑀
𝑔 CRR = su(yield)/s’vo
PGA Depth/Stress Reduction Magnitude Scaling
33
tother could be deformation driven shear, deformation etc.
Post-trigger Stability:
Assign liquefied strength to soils along failure planes
Carry out a limit equilibrium stability check
FoS ≤ 1 Failure
1 < FoS ≤ 1.1 Progressive failure
FoS > 1.1 No failure
36
∞ NCLs as ƒ() i.e. ƒ(density)
NCLs on both sides of the CSL
Solution to the Problem that:
“Sands always behave OC”
37
Decoupling of the Yields Surface from the CSL
38
8 Model Parameters (dimensionless)
M = Slope of the CSL in stress space, (q/p’)
l = Slope of the CSL in void ratio/stress space, (e/ln(p’))
G = Altitude of the CSL in void ratio/stress space, (e/ln(p’))
H = Plastic Hardening Coefficient, is ƒ() and modelled as H = Ho + H()
Hardening Law determined by Iterative Forward Modelling (IFM)
Can approximate to 1/(l–k) in silty clayey soils, H often taken as zero
Hr = Plastic Softening Coefficient, not yet fully understood
= State-Dilatancy Coefficient, i.e. relationship between & Dmin
External Yield Surface & Internal Yield Surface
Ir = Shear rigidity, G/p’
n = Poisson’s ratio
Helper Parameters
N = Stress-Dilatancy Coefficient, i.e. relationship between hmax, Dmin & M
Volumetric Coupling to capture dilatancy in the Flow Rule, N = 0 in CamClay
Dmin = max dilatancy, where D = ev/eq & e = ep + ee (plastic & elastic components)
hmax = max stress ratio, where h = q/p’
G = Shear modulus, used to model elastic behaviour
Determined from shear wave velocity from geophysics, or bender element tests
Can be determined by load/unload cycles during triaxial shear but requires care
Dependent on initial stress level po
39
With both Ir & n assumed constant, G becomes stress dependent
NorSand – Parameters and Ranges
40
NorSand - Typical Soil Parameters Sets
41
Jefferies & Been (2006, 2016)
Liquefaction explained by way of NorSand CSSM framework
Allows undrained simultaneous & equal plastic/elastic volumetric strains
No requirements for collapse type mechanism to generate pore pressures
Liquefaction Susceptibility:
Based on state parameter: 𝜓 = 𝑒 − 𝑒𝑐𝑠 “How far from CS/SS & on which side”
Simplified: + indicates contractive response, i.e. liquefiable
- indicates dilative response
Dilative/contractive boundary varies from > -0.05 (strong) to > -0.08 (mild)
The in situ state parameter is determined from CPTu test results
Q = ƒ(initial state & compressibility k, m)
Qp = normalised cone resistance
k & m rigidity functions (compressibility) linked to cavity expansion theory
Plewes et al. (1991) screening method, assumes F(cone friction ratio) & M = 1.25
Shuttle & Jefferies (1998), k & m ito l for drained CPTs (F < 1.5% & -0.02 < Bq < 0.02)
Shuttle & Cunning (2007), 𝑘 & 𝑚 ito l & M for undrained CPTs (F > 1%)
Full-monty NorSand: M, N, H, l, n, Gref & n (Gexp)
42
Screening Level
Susceptibility
Dilative
Contractive &
Liquefiable
43
Triger Analysis:
J&B use the CRR vs. CSR concept to check the FoS against a Seismic Trigger
CRR is determined from
Estimates based on for example:
NCEER Robertson & Wride (1998): Q, Ic & modifications factors (Kc, Km, Ks & Ka)
CSR based on
The work of Seed, Youd, Idriss etc. as before,
Typically Seed Simplified as described in Youd et al. (2001)
Post-trigger Stability:
su(yield)/s’vo from CPTu as per usual using
f’ for Ic ≤ 2.5 for silts and sands, where f’ can be derived from
Nkt for Ic > 2.5 for clayey soils and clays, Nkt = 15 typically used
45
Clean SAND
Silty SAND
Sandy SILT
Liquefied strength
based on case histories -0.05 46
Been & Jefferies Testing Recipe
General Approach
Define the soil Profile
Define the Boundary Conditions
Define the Water Table and Pore Pressure Regime
Define variability and identify Anomalies
Determine the in situ state from CPTu testing
Determine the required Soil Parameters from lab testing
Predict liquefaction
Scope of Field and Lab Investigations
CPTu Penetration Testing: In situ state, profile, pore pressures, water table etc.
Sampling through Core Drilling or using MOSTAP window sampler on CPTu
Shear wave velocities for G via SCPTu or Bender Element Tests
Lab Tests on Reconstituted Samples: Composition & CSSM parameters
Indicator Tests including emin and emax
Critical State
Stress-Dilatancy & State-Dilatancy
Plastic Shear Hardening (requires some modelling to compute)
47
Elastic properties
Laboratory Triaxial Testing Programme
You need 10kg of each material for 10 to 15 Triaxial tests
Test specimens are prepared by moist tamping
5xTest A & 2xTest B - Determine the CSL: G, l & M
3xTest C - Determine the Stress-Dilatancy: M & N
Determine State-Dilatancy:
Gmax as ƒ(Gref & Gexp) by guess & fit or bender element & assume n (0.2)
Determine the hardening parameter by IFM: H (best fit model)
Hr is not required but can be determined via cyclic simple shear
48
Seismic Analysis Only
49
Commercial Laboratories
Efforts should be made to:
‘Educate’ Laboratories/Consultants/Clients on the requirements for liquefaction
testing and appropriate methods/procedures
Develop/Agree on a Standard Set of tests for liquefaction parameters
Convince the labs to offer this as a Lump Sum Suite
50
6. CHALLENGES
51
Major Uncertainties
CPTu
when is the probe response drained / undrained / de-saturated [F & Bq]
when is an ANOMALY ( ‘blip’ in cone data) considered ANOMALOUS
how do I interpret the CPTu data above the water table
Material Types
do I have 2 types, i.e. fine & coarse tailings e.g. cyclone over/under flow
when can I test 1 mixed sample vs. a bandwidth of PSDs
Trigger Mechanisms
have I considered all possible or plausible mechanisms
Strength
when and where do I use the undrained shear strength
can I mix D and UD strength models in the same stability calculation
52
Major Uncertainties
Liquefaction Risk
is there a limiting depth to liquefaction, say 20m deep
can liquefaction occur above the water table
and in sub-hydrostatic pore pressure regimes
does increased fines content (silt) reduce liquefaction potential
57
Mount Polley, British Columbia
BIG volume & BIG mud-flow
58
Baffokeng, South Africa
Massive volume discharged
59
Main Liquefaction Risks / Drivers for a TSF
Overtopping – surface water management
Do not use TSF to store excess water
Collect, manage and discharge surface water quickly and safely
Maintain adequate freeboard, check water balance
Build-in redundancy, what if the penstock fails?
Piping & Backwards Erosion – day wall construction
Tailings can be susceptible to suffusion, internal and back erosion
Weak spots in the day wall can trigger piping
Check for boils at the toe of the embankment slope
Upstream Day Wall Construction – the sins of the past
Don’t build a starter wall dam
Don’t burry weak soft tailings below the future embankment wall
Environmental & Economic Pressures
Go-higher & Go-faster may lead to Go-no-more or Go-ne
Upstream construction => short term gain for long term pain
60
Construction over liner systems still a big ?
Water Management