Evaluation of liquefaction in

tailings and mine waste:

an update
P.K. Robertson
2018
 Definitions of Liquefaction
Level – gently sloping ground
• Cyclic (seismic)
Liquefaction
– Zero effective stress
(during cyclic loading) Moss Landing, CA

Steeply sloping ground

• Flow (static)
Liquefaction
– Strain softening
response Fundao, Brazil
 Tailings & mine waste

Flow Liquefaction is the one of the main design

issues for most tailings and mine waste structures
• High static shear (“steeply” sloping ground)
• Various trigger mechanisms
– Cyclic loading only one form of trigger
• High risk if significant strength loss possible
 Flow liquefaction - Case histories
• Common soil features:
Stava, Italy, 1985; 268 deaths, 190,000m3
– Very young age
– Non-plastic or low-plastic
– Little or no stress history (Ko ~ 0.5)
– Very loose (contractive)
– Low effective stress (s’vo < 3atm)

• Common instability features:

Fundao, Brazil, 2015 19 deaths
– Some triggered by very minor disturbance
– Failures tend to occur without warning
– Failures tend to be progressive & rapid
– Observation approach not valid
Flow liquefaction – Evaluation Steps
Evaluation Sequence:
1. Evaluate susceptibility for strength loss
2. Evaluate stability using post-earthquake shear
strengths
3. Evaluate trigger for strength loss
If soils are susceptible, and instability possible
(FS<1), it is often prudent to assume trigger
will occur
 Outline
• Trigger events
– Small events can trigger strength loss
– Critical stress paths
• Classify materials susceptible to strength loss
– Application of SCPTu
• Stability
– Influence of high stresses
– Unsaturated tailings/waste
– Progressive failure
 Triggers - Flow Liquefaction
After Olson & Stark, 2003

Undrained loading e.g. rapid construction A-B-C

Undrained cyclic e.g. earthquake A’-E-C
Unloading e.g. increasing GWL A-D-C
 Example - Fundao
http://fundaoinvestigation.com/
Unloading stress
Lab. testing
paths are the
most critical -
can be drained
or undrained

Higher static
shear stress ratio
- smaller the
trigger
Classify susceptibility to strength loss
• Geo-materials must be strain softening in
undrained shear
• Strain softening geo-materials are contractive
at large strains

CPT can be used to identify contractive soil

for young, uncemented soils - Robertson(2010)
Case Histories of Flow Liquefaction
Case Histories
Nerlerk (sand) – 19,20,21
Jamuna (sand) - 34
Fraser River (silty sand) - 27
Sullivan mines (silty tailings) - 35
Northern Canada (silty clay) – 36
L. San Fernado Dam (silt) – 15

CPT data in critical layers +/- 1 sd.

(Average stress level < 200 kPa)

All flow liq. case histories plot in ‘contractive’

portion of CPT SBT chart
Good theoretical support via State Parameter
Case Histories of Flow Liquefaction

Robertson, 2010

34

27
15
35
36 19,20,21

su(liq) / s ’vo = liq. undrained strength ratio (sand-like)

Case Histories of Flow Liquefaction

Robertson, 2010

34

27
15
Fundao 35
36 19,20,21
Fundao

su(liq) / s ’vo = liq. undrained strength ratio (sand-like)

 Soil structure
• Some natural soils and mine tailings/waste have
some form of ‘structure’ that make their behavior
different from ‘ideal’ soil
– Macrostructure (layering, fissuring, etc.)
– Microstructure (particle scale – aging, bonding, etc.)

Canadian Geotechnical Journal, 2016

“CPT-based Soil Behavior Type (SBT) Classification System – an
update” P.K. Robertson
 Seismic CPTu

SCPTu
6 -7 measurements!
qt
fs
u2
Vs (Vp)
t50
Diss.test
uo

i
After Mayne, 2014
 Identification of microstructure
• CPT penetration resistance, qt – controlled by
peak strength
• Shear wave velocity, Vs – controlled by small
strain stiffness
• Potential to identify ‘structured’ soils from
SCPT by measuring both peak strength and
small strain stiffness
 New Go/qn Chart
1000
Robertson, 2016
Small strain
rigidity index:
Cementation/bonding
100 & aging IG = Go/(qt-svo)
Qtn

Normalized
10 rigidity index:

K*G = IG (Qtn)0.75
KG = (Go/qt)(Qtn)0.75
1
1 10 100 1000
IG = G
Goo/q/ tqn
 New Go/qn Chart
1000
Robertson, 2016
Soils with microstructure Average
20 24
(e.g. cementation/bonding normalized
& aging)
100 9 10
32
rigidity index for
28

5
22 26 30 young,
4 11 31

Qtn 3 6 19
23
25
uncemented
29

15
21
27
silica-based
10
2
13
16 17
18
soils:
7 14
1 8
12
K*G = 215
K*G = (Go/qn)(Qtn)0.75
1
1 10 100 1000
IG = Go/qn
 Updated CPT-based SBT Charts
Ideal soils – Robertson, 2016
no microstructure Microstructure
Microstructure

Ideal

Normalized SCPTu parameters: Qtn, Fr, U2 and IG

Canadian Geotechnical Journal, 2016
“CPT-based Soil Behavior Type (SBT) Classification System – an update” P.K. Robertson
 Updated SBTn Charts
1000 Robertson, 2016

IB = 32 Behavior Descriptions
IB = 22 Soil Behaviour Type
SD 1: CCS Clay-like - Contractive - Sensitive
100 CD = 70 2: CC Clay-like - Contractive
3: CD Clay-like - Dilative
4: TC Transitional - Contractive
5: TD Transitional - Dilative
TD
Qtn

SC CD 6:
7:
SC
SD
Sand-like - Contractive
Sand-like - Dilative
TC
10
CD = (Qtn - 11)(1 + 0.06Fr)17
CCS CC IB = 100(Qtn + 10)/(70 + QtnFr)

1
0.1 1 10
Fr (%) Soils with no microstructure
 Updated SBTn Charts
1000 Robertson, 2016

IB = 32 Behavior Descriptions
Aging, cementation
Density
IB = 22 Soil Behaviour Type
SD 1: CCS Clay-like - Contractive - Sensitive
100 CD = 70 2: CC Clay-like - Contractive
3: CD Clay-like - Dilative
4: TC Transitional - Contractive
5: TD Transitional - Dilative
Young-uncementedTD
Qtn

SC
NC to LOC soil CD 6:
7:
SC
SD
Sand-like - Contractive
Sand-like - Dilative
TC
10 OCR
CD = (Qtn - 11)(1 + 0.06Fr)17
Sensitivity
CCS CC IB = 100(Qtn + 10)/(70 + QtnFr)

1
0.1 1 10
Fr (%) Soils with no microstructure
 Updated SBTn Charts
1000
Saturated soils
IB = 32 Behavior Descriptions
IB = 22 Soil Behaviour Type
SD 1: CCS Clay-like - Contractive - Sensitive
100 CD DRAINED
= 70 2: CC Clay-like - Contractive
CPT 3: CD Clay-like - Dilative
4: TC Transitional - Contractive
5: TD Transitional - Dilative
TD
Qtn

SC CD 6:
7:
SC
SD
Sand-like - Contractive
Sand-like - Dilative
Transition -
TC
Partial drainage
10
CD = (Qtn - 11)(1 + 0.06Fr)17
UNDRAINED CPT
CCS CC IB = 100(Qtn + 10)/(70 + QtnFr)

1
0.1 1 10
Fr (%) Soils with no microstructure
 Updated SBTn chart

Robertson, 2016

Fundao OCR > 4

36 15

Soils with no microstructure

Contours of Qtn,cs on updated SBTn chart
Kc
Robertson & Wride, 1998

60
44
OCR >4

Ic 70
60
Contours of Qtn,cs 44

Based on case histories of cyclic liquefaction

Contours of su(liq) / s’vo based on Qtn,cs

Based on Qtn,cs
(extended to Ic > 2.6)

OCR >4

0.25

0.10
Contours of su(liq)/s’vo 0.05

su(liq) / s ’vo = liq. undrained strength ratio (sand-like)

Contours of su(liq) / s’vo based on Qtn,cs

Based on Qtn,cs
(extended to Ic > 2.6)

OCR >4

to 5
Only applies 0.2
Sand-like (SC)
0.10
drained CPT0.05
Ic < 2.6 or IB > 32

su(liq) / s ’vo = liq. undrained strength ratio (sand-like)

 Clay-like soils (undrained CPT)
remolded strength & sensitivity (St)
Robertson (2016)
Modified for
clay-like soils
(when Ic > 2.6)

OCR >4 0.60

0.01 0.10 0.25
OCR OCR
St
fs ~ su(R)
fs /s’vo = Fr Qtn /100 St
Contours of su(liq/R)/s’vo 0.25
0.10
NC Clay 0.05
St ~ 1

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils

su(R) / s ‘ vo = fs / s ‘ vo = (Fr Qtn )/100
Contours of su(liq) / s ’vo & su(R) / s ’vo

Modified for
clay-like soils
(when Ic > 2.6)

0.60
OCR
St
Contours of su(liq/R)/s’vo 0.25
0.10
0.05

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
Contours of su(liq) / s’vo based on Qtn,cs

Based on Qtn,cs
(extended to Ic > 2.6)

OCR >4

0.25

0.10
Contours of su(liq)/s’vo 0.05

su(liq) / s ’vo = liq. undrained strength ratio (sand-like)

 Contours of
Soils with no microstructure
su(liq) / s ’vo
su(R) / s ’vo
IB Based on case histories:
0.60
•Flow liquefaction for sand-
OCR
like soils
•Remolded shear strength for
St clay-like soils
0.25
Contours of su(liq/R)/s’vo
0.10 Both large strain
0.05 undrained strengths

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Stability
• Tailings dams are becoming increasingly
higher (>100m) with much higher overburden
stresses (> 8 atm).
– How do we extrapolate to higher stresses?
• Many tailings dams/waste pads have large
regions that are unsaturated
– How do we evaluate influence of lack of
saturation?
 Questions?

Are all contractive soils strain softening?

Are all strain softening soils brittle?
How does stress level affect these?

PBDIII Earthquake Geot. Eng. , Vancouver - 2017

“Evaluation of Flow Liquefaction: Influence of high stresses” P.K. Robertson
 State parameter in sands

p’cs p’o p’o /p’cs = 10(y/l)

(+) Loose
eo CONTRACTIVE
Y = eo - ecs
ecs

(-) Dense
State Parameter after Jefferies and Been, 1985
DILATIVE Critical State Line
(CSL)

State Parameter after Been and Jefferies, 1985

 Brittleness
Bishop (1967) defined brittleness:
IB = (tp – tr)/ tp

tp = peak strength
tr = residual (large strain) strength at same effective
normal stress
t tp
IB = 1.0 (100% strength loss)
IB = 0 (no strength loss)
tr
g
 CSL for range of sands

Contractive Critical State Lines (CSL)

Dilative

Jefferies and Been, 2016

Limited Linear approximation for CSL

stress range
 CSL over wide stress range

Contractive Critical State Lines (CSL)

Dilative

CSL non-linear
Schnaid et al, 2013
over wide stress range

Verdugo & Ishihara, 1996

 Bolton (1986) Relative Dilatancy Index

Bolton (1986) Relative Dilatancy Index

Increasing
compressibility/crushability

Boulanger 2003 Schnaid et al, 2013

Verdugo & Ishihara, 1996

 Bolton (1986) Relative Dilatancy Index

Bolton (1986) Relative Dilatancy Index

Increasing
compressibility/crushability

This format is very helpful to estimate CSL

Bolton’s empirical relationship is helpful guide

Boulanger 2003 Schnaid et al, 2013

Verdugo & Ishihara, 1996

 Example – Erksak Sand

Y = 0.07
Contractive
p’(y/l ~2.0)
o /p’cs = 100
Dilative
Y = 0.20
p’(y/l
p’o /p’cs = 10(y/l) o /p’cs = 4.5
~0.9)

Mtc = 1.2
Loosest state Y = 0.25
p’o(y/l
/p’cs ~0.6)
= 3.5
 Brittleness (IB) vs (p’o /p’cs)
A
IB
Regardless of fabric and direction of loading

(Ottawa)
(Illinois)
B (Mississippi)

C
IB = 0.9 – 0.84(p’o/p’cs)

Modified from
~NC Clay po’/p’cs Sadrekarimi & Olson, 2011
 Brittleness (IB) vs (p’o /p’cs)
A
IB Can use y but requires slope of CSL l
(y/l) which is changing
(Ottawa)
(Illinois)
B (Mississippi)

C
IB = 0.9 – 0.84(p’o/p’cs)

Modified from
~NC Clay po’/p’cs Sadrekarimi & Olson, 2011
 Brittleness (IB) vs su,cs/s’vo
(Ottawa)
~NC Clay
(Illinois)
C (Mississippi)

B
su,cs/s’vo

Modified from
IB Sadrekarimi & Olson, 2011
 Brittleness (IB) vs su,cs/s’vo
Low Brittleness

C
B
su,cs/s’vo
Case histories

A
High Brittleness

Modified from
IB Sadrekarimi & Olson, 2011
 Brittleness (IB) vs su,cs/s’vo
Low Brittleness

C
B
su,cs/s’vo
Case histories
su,cs/s’vo < 0.15
have higher brittleness
IB > 0.4 A
High Brittleness

Modified from
IB Sadrekarimi & Olson, 2011
 Contours of
Soils with no microstructure
su(liq) / s ’vo
su(R) / s ’vo
Based on case histories:
0.60
•Flow liquefaction for sand-
like soils
High IB OCR
at low stress •Remolded shear strength for
clay-like soils
0.15
Contours of su(liq/R)/s’vo
Both large strain
undrained strengths

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Liquefaction Summary Sand-like and Dilative – SD

Soils with no microstructure •Potential for cyclic liquefaction –

depends on size and duration of
cyclic loading
•CRR will increase with
increasing static shear
•Sampling difficult & in-situ
testing preferred
IB •Estimate CRR based on Qtn,cs
0.60
•No strength loss expected unless
OCR
state changes and/or some
microstructure (e.g. cementation)
St
0.25
Contours of su(liq/R)/s’vo
0.10
0.05

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Liquefaction Summary
Contours in SC/TC Sand-like and Contractive –
region based on SC
s‘vo < 2atm) •Potential for cyclic & flow
liquefaction
•CRR will decrease with
increasing static shear
•Sampling difficult & in-situ
testing preferred
IB •Strength loss possible
0.60
•Estimate state (y) and su(liq)/s’vo
based on Qtn,cs
OCR
•Strain to trigger strength loss can
St be small
0.25
Contours of su(liq/R)/s’vo
0.10
0.05

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Clay-like and Contractive –
Liquefaction Summary CC/CCS
•Potential for cyclic softening &
Soils with no microstructure flow liquefaction
•CRR can decrease with increasing
static shear
•Sampling possible and
recommended
•Strength loss possible
•Estimate state (OCR) and su(R)/s’vo
IB
0.60 based on CPT & FVT
•Strain to trigger strength loss can
OCR be large – depends on plasticity
•Evaluate sensitivity using CPT fs
St and FVT
0.25 •Check drainage CPTu dissipation
Contours of su(liq/R)/s’vo tests
0.10
0.05 •Measure wn and Atterberg

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Liquefaction Summary
Clay-like and Dilative – CD
Soils with no microstructure
•Liquefaction unlikely
•Sampling possible, if not too stiff
•No strength loss expected unless
state changes and/or some
microstructure
•Measure wn and Atterberg
IB •Drained strength valid
0.60

OCR

St
0.25
Contours of su(liq/R)/s’vo
0.10
0.05

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Liquefaction Summary
Transitional– TD/TC
Soils with no microstructure
•Evaluate plasticity and drainage
(Atterberg Limits & CPTu
dissipation tests)
•Sampling possible depending on
plasticity and fines
•Evaluate assuming both sand-
IB like and clay-like and compare
0.60
• typically drained
strengths are more
OCR
conservative in SD, SC,
St TD, TC and CD
0.25
Contours of su(liq/R)/s’vo
0.10
0.05

su(liq) / s ’vo = liq. undrained strength ratio for sand-like soils

su(R) / s ’vo = remolded undrained strength ratio for clay-like soils
 Case History
• Filtered mine waste placed via conveyor and radial arm
stacker with surface irrigation
• Placed in uncompacted lifts of about 20 to 30m thickness
• Overall slope of about 4H:1V
• Current max. height of 200m
• Approx. 100 million tons of waste
• Low precipitation & high evaporation region

Canadian Geotechnical Journal, 2017

“Characterization of unsaturated mine waste: case history”
Robertson, P.K., da Fonseca, A.V., Ulrich, B. & Coffin, J.
 Typical index test results
• Uniform grain size distribution – sandy silt
• Fines content ~ 50%
• D50 ~ 0.065mm
• Specific gravity, Gs = 2.73
• Mostly non-plastic
• In-situ water content (w) varies from around
2% to 15% (depending on depth, location and
irrigation) with degree of saturation
5% < S < 50%
 Typical CPTu

High stress
s’vo ~15atm
 Typical Normalized CPTu

High stress
s’vo ~15atm
 Typical Normalized CPTu
Updated SBTn Charts

Mostly Mostly drained Little or no microstructure

Transitional Contractive CPT penetration
 Updated CPT SBTn chart
Robertson, 2016

0 to 15m depth range

s’vo < 300kPa
Qtn-Fr chart indicates
that material is mostly
sand-like and
CONTRACTIVE at
large strains

(assumes no suction)
 Updated CPT SBTn chart
Robertson, 2016

15 to 90m depth range Qtn-Fr chart indicates

s’vo > 300kPa
that material is mostly
sand-like/transitional
and CONTRACTIVE
at large strains

(assumes no suction)
Updated CPT SBTn chart

With increasing depth,

Qtn decreases and Fr
increases
Trends toward a more
clay-like contractive
behavior, but often less
brittle

0.25
0.10
0.05
 Seismic Velocity (SCPTu)

Vs

Vp

Vs1 = Vs (pa/s’vo)0.25
 Seismic Velocity (SCPTu)

Contractive Dilative
Vp < 1500m/s
Unsaturated

Vs Vs1

Robertson, 1995
Vp Vs1 ~ 225m/s
 In-situ test interpretation

• CPT (Qtn) based interpretation suggests that

soils are mostly CONTRACTIVE at large
strains
• Shear wave velocity (Vs) based interpretation
suggests that soils are DILATIVE at large
strains
Which one is correct?
Critical State line (saturated)
Approx.
Reconstituted samples e(max)

Loosest state
Contractive Limiting compression
curve (saturated)

Dilative

In-situ stress range

Approx.
e(min)
Vs contours and CSL

Assuming materials are saturated

Contractive
Vs measured in lab.
at end of consolidation
Dilative

Vs(field) > 200m/s

 Vs1 as a function of suction (s)
Normalize Vs1 based on modified s’vo (adjusted for suction)

3 tests 50 < sc < 600 kPa

4 tests 50 < sc < 600 kPa

4 tests 50 < sc < 600 kPa

Increasing suction
(decreasing S)

Sandy silt tailings

Vs1 as a function of suction (s)

3 tests 50 < sc < 600 kPa

?
In-situ range

4 tests 50 < sc < 600 kPa

Increasing suction
(decreasing S)

Sandy silt tailings

CSL as a function of suction (s)
Moist samples prepared to ei ~ 1.1 to 1.2

~loosest compression line after saturation

s = 0 kPa
Sandy silt tailings
CSL as a function of suction (s)
Moist samples prepared to ei ~ 1.1 to 1.2

~loosest compression line moist

saturated samples

s = 0 kPa
Sandy silt tailings
CSL as a function of suction (s)
Moist samples prepared to ei ~ 1.20

~loosest compression line moist

CSL (unsat.)

s = 200 kPa
s = 20 kPa

s = 0 kPa
Sandy silt tailings
CSL as a function of suction (s)

Project tailings - Strain hardening when unsaturated

Limited strain softening if saturated

CSL (unsat.)

s = 200 kPa
s = 20 kPa

s = 0 kPa
Sandy silt tailings
 Case History - Summary
• Shear wave velocity (Vs) appears to be sensitive
to suction hardening, since it is a small strain
measurement
– Vs better indicator of unsaturated behavior
• Cone resistance (qc) appears to destroy
beneficial effects of suction hardening, since it is
a large strain measurement
– qc may better indicate behavior if soil becomes
saturated
Critical regions to investigate?
Simplified piezometric
surface

Tailings/waste

Foundation soils
 Critical regions to investigate?
Simplified piezometric
surface

Tailings/waste

Foundation soils

Toe region:
• Higher static shear stress ratio
• Low effective confining stress
• Higher possibility to be 100% saturated
• Risk of progressive failure
 Conclusion (1)

• Flow liquefaction can be triggered by

relatively minor conditions and ’unloading’
stress paths are the most critical, e.g.
– Earthquakes (even small earthquakes)
– Rising piezometric surface
– Unloading (movements in foundations or tailings)
 Conclusion (2)
• Not all CONTRACTIVE soils are strain
softening in undrained shear
• Not all strain softening soils are brittle
– brittleness varies with state, stress level and PI
• Although soils tend to become more contractive
with increasing stress they also tend to become
more ductile
• Low stress regions maybe most critical
 Conclusion (3)
• SCPTu an excellent tool to evaluate potential
risk of flow liquefaction
– 5 to 6 measurements
– Ability to identify microstructure
– Classify soil behavior
– Vs and Vp powerful additional measurements
• Lab. testing to helpful to support evaluation via
CSL
Questions?
PDF Copy of slides:
www.cpt-robertson.com

Free (recorded) Webinars:

www.greggdrilling.com/webinars