GEOTECHNICAL ANALYSIS OF

WASTE STORAGE FACILITIES

Lecture 6

Liquefaction Hazard

Jonathan D. Bray, PhD, PE, NAE

Faculty Chair in Earthquake Engineering Excellence
University of California, Berkeley
 LIQUEFACTION HAZARD

Lecture Outline

1. Key Concepts

2. Liquefaction Triggering

3. Post-Liquefaction Residual Shear Strength

4. Closing Remarks
MODELING OF TAILINGS/WASTE
STORAGE FACILITIES

1. Key Concepts
 SOILS SUSCEPTIBLE TO LIQUEFACTION

• SAND “classic”; much experience and testing on uniform sand

• SILT nonplastic silt does not know it is not very, very fine sand

• CLAY some clayey soil can liquefy; other clays can undergo fabric
break-down (e.g., quick clay)

• GRAVEL has liquefied in field and in lab; especially if gravel particles

are “floating” in sand matrix or if gravel is “within” a fine-grained matrix

ASSUME ALL SOIL MAY CYCLICLY SOFTEN UNTIL PROVEN OTHERWISE

TEST
Simplified liquefaction
triggering procedures have
been developed for sand and
nonplastic silty sand. Apply
with judgment to other soil
(e.g., low-plastic tailings). PI = 2 PI = 5

Bray & Sanco (2006)

CSS Testing of PI ≤ 12 & wc / LL ≥ 0.85
50

Reconstituted 40
Susceptible to Liquefaction
Moderate Susceptibility
Not Susceptible

Plasticity Index
30

Plastic Silt Specimens 20

PI = 11 10

[Soil G has PI = 10] 0

0.4 0.6 0.8 1.0 1.2 1.4
Donahue et al. 2007 wc /LL

or Perform Lab Tests

PI = 14 PI = 7
TYPES OF UNDRAINED RESPONSE – MONOTONIC LOADING
dependent on volume change tendencies: contractive or dilative

(a) Effective stress path (b) Stress-strain curve

σ 1 − σ3
2

Limited
Shear stress,

flow No-flow No-flow Limited

flow

Flow Flow

Mean effective stress, p' Shear strain, ε 1 − ε 3

 EFFECTS OF DENSITY – CLEAN SAND
Monotonic Undrained Shear Response of Toyoura Sand in 6 Torsional Tests:
- specimens prepared at different densities (Dr = 25, 40, 50, 60, 70 and 80%) using
air-pluviation, and then isotropically consolidated to p' = 100 kPa
- data show the effects of density on undrained stress-strain response of sand

Effective stress path Stress-strain curve

150 150

p' = 100 kPa 80 % p' = 100 kPa

D = 25-80 %
70 % r
(kPa)

60 %

(kPa)
100 100

D =50 %
80 % r
Shear stress, τ

Shear stress, τ
D =50 %
r 70 %
50 40 % 50
60 %
40 %
25 %
25 %
0 0
0 50 100 150 200 0 10 20 30
Mean normal stress, p' (kPa) Shear strain, γ (%)

From Prof. Cubrinovski

EFFECTS OF EFFECTIVE CONFINING STRESS – CLEAN SAND
Effective stress path Stress-strain curve

Dr = 38%

• Toyoura sand specimens prepared at same relative density (Dr = 38 %); but isotropically
consolidated to different mean effective stress; subjected to undrained triaxial compression

• Though density is the same, specimen responses differ markedly

• Specimen at low confining stress (p' = 0.1 MPa) is primarily dilative and strain-hardening

• Specimen at high confining stress (p' = 3.0 MPa) is contractive with strain-softening

From Prof. Cubrinovski

 STEADY STATE LINE (SSL)
Effective stress path Stress-strain curve

Steady state
e = 0.83 (Dr = 38%)

e
0.83 ( e , p' ) • Four specimens at same density (Dr=38%)
Initial state with different initial confining stress

• Depending on the initial confining stress,

specimens are contractive (develop positive
p' EPWP) or dilative (negative EPWP); all
• Initial state: state prior to shearing specimens with same initial density end up
(after consolidation) at same steady state
From Prof. Cubrinovski
 Effective stress path Stress-strain curve

e = 0.735 (Dr = 64%)

e
0.83
Steady State Line
0.735 • Four specimens at another density
(Dr=64%) and different initial
confining stresses; again, all tests
p' end up at same steady state

• Draw curve connecting the steady states,

and define the STEADY STATE LINE

From Prof. Cubrinovski

 SSL of Toyoura Sand from 70 Undrained TX Compression Tests (from Cubrinovski)

Undrained (e=const) CONTRACTIVE

ue > 0
Steady state line
Drained (p'=const)
∆e < 0 It all depends on
∆e > 0
the location of the
initial e-p’ state
ue < 0 relative to SSL
DILATIVE

• SSL of a sand is “uniquely” defined in the e-p' plane in the framework

• Irrespective of its initial state, each test ends up on SSL (when sheared monotonically)

• Position of the initial state relative to the SSL defines the e-p' path
and sand response for a given loading and drainage conditions
From Prof. Cubrinovski
 State Concept Interpretation of Sand Response
Void ratio, e

FLOW (a) Effective stress path

Flow

σ 1 − σ3
Initial

2
dividing line
Limited flow
Limited

Shear stress,
No-flow flow No-flow

NO-FLOW
Flow

Mean effective stress, p' Mean effective stress, p'

Use SSL as a reference - initial dividing line

From Prof. Cubrinovski

 Effects of Nonplastic Fines on Flow Potential
Clean Sand Sand with Fines
(FC < 5%) (FC = 20%)
e -e = 0.35 e -e = 0.60
max min max min
-20 -20

e max e max
0 0
(%)

FLOW
20 20
FLOW
r
Relative density, D

40 40

60 60

80 80

100 e min 100 e min

0 100 200 0 100 200
Mean normal stress, p' (kPa) Mean normal stress, p' (kPa)

Potential for flow response (strain softening under

undrained loading) increases with increasing FC

(Cubrinovski and Ishihara, 2000)

 Effects of Nonplastic Fines on Flow Potential
FBM-FC (Fitzgerald Bridge Mixture – Fines Content, Christchurch)

• Reconstituted specimens
SSL of sand with non-plastic
fines up to 30%
SSL
• Flow potential increases
FBM-30 FBM-10 with increasing fines
content (when Dr is used
SSL as a basis for comparison)

The sand becomes more

contractive with increasing FC
(Cubrinovski and Rees, 2008; Rees, 2010)

From Prof. Cubrinovski

MODELING OF TAILINGS/WASTE
STORAGE FACILITIES

2. Liquefaction Triggering
 Cyclic Resistance Ratio (CRR: Cyclic Strength)

Liquefaction can be triggered by different

combinations of amplitudes (CSR = τ /σ’vo)
and number of uniform stress cycles (Nc)

Liquefaction defined by:

ru = ue/σ’vo ≈ 100%
3% S.A. Axial Strain
5% D.A. Axial Strain
2% D.A. Axial Strain
3.75% Shear Strain
5% Shear Strain
(Cubrinovski, 2011)
5 SOIL FACTORS AFFECTING
LIQUEFACTION RESISTANCE
1. Relative Density
2. Soil Fabric
3. Stress History
4. Age
5. Seismic History

Seed, H.B. (1979) ASCE

Effects of Relative Density on CRR Curve Shape
Slope of curve steepens with
increase in density

Higher position
of CRR with
increase in
density!

(Idriss and Boulanger 2008)

Effects of soil density on CRR:

- Higher position of curve for high densities
- Steeper CRR curve at higher densities
Differences between Liquefaction of Contractive and Dilative Sand

Contractive sand e = 0.809

σ’c = 294 kPa

Dilative sand e = 0.676

Slower development p' is greater than zero

2 of strains with cycles 1
Importance of Specimen Preparation & Fill Placement Method

Mullis et al. 1975

SILT

Donahue, Bray &

Riemer 2007
 SILT

Donahue, Bray &

Riemer 2007
Seismic History (without causing liquefaction!)

(Idriss and Boulanger, 2008)

 Repeated Liquefaction Events

4 Sept
2010

22 Feb 2011

16 April
2011
13 June 2011: Part
1
13 June 2011: Part
2
(Mark Quigley: Avonside)
Age of Christchurch Soils
After Liquefaction

Age data from

M. Cubrinovski
OTHER FACTORS AFFECTING
LIQUEFACTION RESISTANCE
• Loading Frequency
• Confining Stress
• Fines
• Testing Device
• …
 SILT

Donahue, Bray &

Riemer 2007
EFFECT OF CONFINING (OVERBURDEN) STRESS

Liquefaction resistance is strongly affected by the

effective confining stress.

Normalization of CRR = τl/σ’vo is not perfect.

σ’vo increases faster than τl as depth increases.

CRRσ c′
Kσ =
CRRσ c′ =101

CRR curves differ

for different
densities and
confining stresses
(Ishihara 1996 & Idriss and Boulanger 2008)
Effects of Fines on Field-Based Liquefaction Resistance

SPT criteria for CRR Field-based (CPT & SPT)

Youd et al. (2001) charts indicate silty sand
has higher CRR for a given
penetration resistance
Sand with FC = 35%
This does not mean that
Clean sand the addition of fines
increases the liquefaction
resistance of sands

(Cubrinovski et al., 2010)

30
 Effects of Non-Plastic Fines
At a given Dr , penetration resistance decreases
At the same Dr clean sand has with increasing FC
higher liquefaction resistance
than sand with fines Penetration resistance decreases as size of soil
particles decreases relative to size of probe

(Cubrinovski et al., 2010)

Use Dr of silty sand & clean sand instead of (N1)60

SPT Liquefaction Resistance Data Expressed in Terms of Dr
SPT criteria for liquefaction resistance
Youd et al. (1998) expressed in terms of CRR - Dr
For a given Dr ,
CRR decreases with
increasing FC

Consistent with
lab test results
& state concept

(Cubrinovski et al. 2010)

SUMMARY – Effects of Fines on CRR

SPT criteria for liquefaction resistance For a given SPT blow count,
Youd et al. (2001) CRR appears to increase with
increasing FC
Instead,
SPT blow count decreases as
fines increases due to
increased compressibility

Silty sands are not more

resistant to liquefaction
than clean sands!
EFFECTS OF TESTING DEVICE
cr = (1 + 2 Ko) / 3
Bray & Sancio 2006 Ishihara 1996

Different devices
produce different
 τh   σd  CRRs and their
  ss = cr   TX
σ 'o   2σ '1  CRRs differs from
the field-based CRR

 τh   τh   τh   σd 
  field = 0.9  ss   field = 0.9cr  
 σ ' o  Pyke 1974  σ ' o   σ ' o  Seed 1979  2σ '1 
 Lab Testing to Evaluate Liquefaction
CRR depends on Dr, fabric, OCR, ageing, (among other factors), so
you need “undisturbed” samples to preserve all these characteristics.
HOW?
Sample disturbance is primary reason why H.B. Seed moved to using
penetration resistance from SPT & CPT to evaluate liquefaction.

34

(Ishihara, 1996; data from Yoshimi et al., 1983)

Use of Lab Testing to Evaluate Liquefaction

1. Research projects to gain understanding

(e.g., tests in Seed 1979 identified 5 key factors).
2. Fill project where you can build soil in lab the
same way it is being built in the field.
3. Soil with fines that can be sampled effectively
with excellent samplers (e.g., Dames & Moore
hydraulic fixed-piston sampler).
4. Coarse-grained soil that can be sampled
effectively with specialized equipment
(e.g., soil freezing or Gel-Push sampler)
35
UNDRAINED RESPONSE OF SOIL DEPENDS ON:
- Relative Density (Dr) or Void Ratio (e)
- Soil Fabric (specimen preparation method)
- Stress History (OCR)
- Age or Time under Confinement (t)
- Seismic History
AND
- Loading Frequency (f)
- Effective Confining Stress
- Fines Content
- Testing Device (load path)
AND
- Plasticity of Fines
- Static Shear Stress
- Sloping Ground
- Multi-Directional Loading
- …
 Simplified Liquefaction Procedure Factor of Safety (FS)
FS = CRR / CSR
CSR = 0.65 (amax/g) (σv/σv′) rd

Liquefaction amax = PGA at ground surface

Effects Observed at without liquefaction
C Ground Surface
(e.g., from GMPE)
S
R CRRB = rd = depth reduction factor,
because soil column
CRRM=7.5,σ ‘v=1, α=0 is not rigid
No Liquefaction
Effects Observed at CRR = CRRB MSF Kσ Kα
Ground Surface

qc1Ncs CRRB qc1Ncs represents Dr, Fabric, OCR,

effects of fines on penetration, σ’v, …
& can be correlated to CRRB
Boulanger and Idriss (2008)
 Calculation of qc1Ncs
Boulanger and Idriss (2014)

FC = fines content (%)

• Most significant change in ∆qc1N

occurs for FC = 10 % - 40 %
• FC > 40%  fines-controlled matrix

Calculation of Overburden-Corrected tip resistance qc1N 10% 40%

 Estimating FC
1. Use site-specific sampling and index testing

2. When such data are not available, use

correlation to estimate FC with Ic

Ic = Soil behaviour type index

10% 40%

Large sensitivity of FC on Ic

Large sensitivity of CRR on Ic

Boulanger and Idriss (2014)
 MSF - Boulanger & Idriss (2014)

Shape of CRR curve is density dependent Therefore, MSF curve is density dependent

Consistent pattern with state concept

interpretation of soil response
40

(from Boulanger and Idriss 2014)

 Effective Overburden Stress Factor, Kσ
For each depth and effective overburden stress σ’v

Kσ is a function of qc1Ncs !

Consistent pattern with state concept

interpretation of soil response

Also BI2014 link Kσ & CN

Boulanger and Idriss (2014)

Static Shear Stress Correction, Kα

For each depth and static shear stress ratio, α

Medium Dense
to Dense Soils:
Kα > 1.0

Loose Soils:
Kα = 0.5 to 1.0

Idriss and Boulanger (2008)

 Empirical Database
1. Depth of Assessment
• Most data for depths < 9 m
• Liquefaction can occur at greater depths
• Some factors are depth-dependent:
rd , CN, Kσ
• Uncertainties associated with
extrapolation outside database range

2. Triggering vs. Manifestation

• Database from case histories
focused on liquefaction manifestation
• Effects of liquefaction depend on the
site and structure considered

3. Few Data at High FC

Few data on soils
with FC > 40%!
 LIQUEFACTION EFFECTS

Flow Liquefaction Cyclic Mobility

(strain-softening large strain) (strain-hardening limited strain)
LIQUEFACTION EFFECTS

Cyclic Mobility

Liquefaction
C Effects Observed at
Ground Surface

S
R CRR
Flow
Liquefaction No Liquefaction
Effects Observed at
FS =1.2 Ground Surface

FS =1.2

85

FS = CRR / CSR Idriss & Boulanger 2008

 MODELING OF TAILINGS/WASTE
STORAGE FACILITIES

3. Post-Liquefaction Residual Shear Strength

Saturated, Contractive Sand under Sloping Ground

su(liq) = liquefied shear strength

su(liq)/σ'vo = liquefied strength ratio

Flow Slide Condition

From S. Olson, U. of Illinois
Post-Liquefaction Residual Strength is a System Property

Lower San Fernando Dam Instability in 1971 EQ (H. B. Seed)

 VOID REDISTRIBUTION

Porewater accumulated
beneath silt layer
Idriss and Boulanger (2007)
Failure occurred on
discrete surface

Slope failed after

shaking ended

Porewater pressure redistribution following shaking

Porewater pressures increase beneath low k soil layer
Local void ratio increases & strength decreases
Can lead to delayed failure
Density governing strength is not pre-EQ density
Post-Liquefaction Residual Strength

Olson & Stark 2002

Post-Liquefaction Residual Strength

Flow Slides when

(N1)60cs-Sr < 15

Idriss & Boulanger 2008

Post-Liquefaction Residual Strength

Idriss & Boulanger 2008

Post-Liquefaction Residual Strength

Sr ≠ constant
&
Sr / σ’v ≠ constant

Instead

Sr ≈ constant (σ’v)#
 Post-Liquefaction Residual Strength

Sr(atm) = exp[−8.444 + 0.109N + 5.379S0.1 − 0.253σm ]

where N = average value of (N1)60, S = mean vertical effective stress (atm), and
σm = [1.627 + 0.000796N2 + 0.0194N − 0.027NS0.1 − 3.099S0.1 + 1.635S0.2]0.5
Kramer & Wang (1985) Method (in Kramer 2008; WA-RD 668.1)
Post-Liquefaction Residual Strength

Weber 2015
 Closing Remarks
Silt and some clay soils can liquefy. Test to evaluate monotonic
and cyclic response. Gravels can liquefy too.
Steady state concept (Dr & σ’c) is useful to understand liquefaction.
Key soil factors include Dr, Fabric, OCR, age, σ’c, PI, fines, test, etc.
Steady state is rarely reached in field. Difficult to measure in situ Dr
and small change in Dr leads to large change in Sr because SSL is
relatively flat.
Must consider void redistribution; it’s a system.
Estimate Sr from back-analyses of flow slide case histories.