Mechanism of PALU

Liquefaction EQ 2018-
Delayed flow failure
July 24, 2021

FLIP Consortium, Japan

Susumu Iai
e

D (delayed flow)
This study
d* (SOA)
ICL (loose of critical)
d
A (liquefaction) B (flow)
e a b
0 c
C (trampoline)
a* Existing
SSL studies

ICL (dense of critical)

p

2
Delayed flow failure

delayed flow failure

h : Surface crust
s
h : Inflow zone of pore water
in
(Initial sate: q us0 > t st)
h : Outflow zone of pore water
out

3
4
5
 Cocktail glass model (volumetric
mechanism)
lK
dp  p
K L/U   rK K U0  
de '  p0 

e '  e  ed

6
e

ICL
SSL
De QSSL edus: ed at
e0
steady state

ed at phase
transformatio
n
line
pusps0 pus0 p0 p

7
Undrained monotonic shear
(review)

Yoshimine et al (1998), Iai et al (2011)

8
Stress path
undrained monotonic shear

Yoshimine et al (1998), Iai et al (2011)

9
  pus   e  ea  e in
ln      / e mus = 
 pa   1  ea  e mus

FLIP consortium (2011)

10
  qus   e in   e out 
ln       rL  
 qus0   e mus   e mus 

 hout 
rL  rDe  
 hin 
rL

e out  3%

e mus  0.03 (  0.06)

e mus  0.015 (  0.03)

11
Example
 qus   e in   e out 
ln       rL  
q
 us0  e
 mus  e
 mus 
 hout  hin  2 m qus0  50 kPa
rL  rDe  
 hin  hout  8 m hs  2 m
rDe  0.5
rL
8
rL  0.5     2
e out  3% 2
 qus 
   0.02
e mus  0.03 (  0.06)
 qus0 
qus  1.0 kPa
t st   hs sin   1.8  2  sin(2)  1.2 kPa

e mus  0.015 (  0.03)

qus  t st Flow 12
Delayed sand boil –
2011 East Japan Earthquake

13
B

Fs

As1

14
Liquefaction resistance

15
(a)
Mainshock Aftershock

(b) Mainshock
Aftershock

16
Analyses cases
Case1 Case2 Case3 Case4
kBs/kFs 5 0.05 0.05 0.05
Input Main & Main & Main After
motion after after

kFs=1xE-5m/s
kAs1/kFs=10

17
Case1

Case2

18
Case3

Case4

19
 Case1

Case2

Base motion

20
Case3

Case4

21
 (H1=0.0m)

Layer B H2=2.0m
Layer B
H1=5.5m
Layer Fs Layer Fs
H2=1.0m

Layer As1 Layer As1

(a) (b)

22
8
Liquefied and unliquefied
layers in Dagupan City
7 assessed by SPT-N
at damaged ground
at undamaged ground
6
Unliquefied H1

5 Liquefied H2
N<12

2 Case 2
Safety boundary curve
by Ishihara (1985) for
200 gal acceleration
1
Case 1
After Ishihara et al. (1993)

0
0 1 2 3 4 5 6
Thickness of surface unliquefied layer, H1 (m)
23
 Layer B hin=0.2m
Layer B

Layer Fs Layer Fs hout=1.0m

Layer As1 Layer As1

(a) (b)

24
2018 PALU earthquake, Indonesia
(after Irsyam et al, 2019)

25
 Petobo
(Process of soil liquefaction in Petobo Housing Complex
www.Instagram.com/p/BokdLnxDx27/?utm_source=jg_embed)
Average Slope = +2.30

Flow slide and movement

direction (modified from
Mason et al, 2019)
Ground movement
(Bessette-Kirton at al, 2018)
 PETOBO
Tension zone

Compression zone

SILTY SAND
Soil boring
Ground shaking at saturated loose near fan
alluvium Petobo affected
deposit  area
Pore pressure generation 

Redistribution stress due possibility pore pressure dissipation/ water film 

Shear stresses > residual strength  Flow slide

N=4-13
Is there any possibility of breakage of aquifer that contribute to massive ground
displacement?
 Grain Size Distribution of Ejected Soil Samples
100
90
80
Percentage finer by weight

70
60
50
40

30
20
10
0
0.001 0.01 0.1 1
2°
30
Analyses cases
CaseI CaseII CaseIII CaseIV
kS0/kS1 5 0.005 0.05 0.005

S1 kS1 1 1 0.1 1
(1E-4m/s)
qus0(kPa) 20 20 20 50

31
32
Undrained cyclic loading
(a)

(b)

33
Undrained monotonic loading S1
(0m) (a)

(b)

34
Undrained monotonic loading S1
(-6m) (a)

(b)

35
Time histories CaseI
(a)

End of EQ motion
(b)

36
Time histories Cases II,III,& IV

37
(a)

(b)

End of EQ motion
(c) Start of delayed flow
slide (Cases II&IV)
Start of delayed flow
tst: static shear stress slide (Case III)

38
(a)

Start of delayed flow

End of EQ motion
slide (Cases II&IV)
(b) Start of delayed flow
slide (Case III)

(c)

39
Distribution of PP

(a) Layer S0 (b) Layer S0

Initial vertical
Initial vertical
effective stress
effective stress

Layer S1 Layer S1

CaseI CaseII&IV
40
Volumetric strain distribution

(a) Layer S0 (b) Layer S0 hin=0.2m

Layer S1 Layer S1

hout=5.0m

CaseI CaseII&IV

41
 qus distribution

Start of delayed flow slide Start of delayed flow slide

(a) Layer S0 (b) Layer S0
End of EQ motion
End of EQ motion
tst: static shear stress tst: static shear stress
induced by slope Layer S1 induced by slope Layer S1

Case II Case III

42
Volumetric strain and qus
distribution
Start of delayed flow slide
(c) Layer S0
End of EQ motion

tst: static shear stress

induced by slope

Layer S1

Case IV

43
DSP

Case II

44
Inflow zone: Case-II & IV
(a) pst=tst/sinff

Start of
flow slide

psu0=qus0/sinff
End of EQ
motion (Case-II)

End of EQ
motion (Case-IV)
Start of EQ
motion

45
Outflow zone: Cases-I,II,IV
(a)
pst=tst/sinff
psu0=qus0/sinff
Start of EQ motion

Dissipation of
pore pressure

46
Inflow zone: Cases-II & IV

Start of flow slide Start of flow slide

(a) (b)

47
 Outflow zone: Cases-I through IV

Dissipation of pore pressure

(a) (b)

48
Simplified/generalized 2D model
analysis of delayed flow failure

Surface crust layer: undrained condition

49
Global failure mode (at the
instance of 5m slide in mid zone)

50
Failure mode in tension zone

Tension fracture mode of complex random

deformation gradually spreading from the edge
toward the mid zone of slope
51
Failure mode in compression zone

more or less orderly deformation

mode of compressive shear

52
 0.20

Residual strength/vertical effective stress

0.16

0.12

0.08

0.04
This study (2018
Palu earthquake)
0.00
0 2 4 6 8 10 12 14 16
Normalized SPT blow counts (N1)60

53
Remark on “Water film”
• In this study, the effect of water film often
observed beneath the less permeable surface crust
was not explicitly discussed. To quote Whitman
(1985), “If, during or after shaking, the disturbed
sand … leaving a liquid film at the interface, an
unstable situation occurs. Actually, it is only
necessary for a thin layer atop the sand to loosen
enough that its steady state resistance becomes
less than the static shear stress.” The nonlinear
dynamic analysis performed in this study supports
Whitman’s perspective.
54
Summary of the earthquake
response analysis by FLIP
• Delayed failure: Some time after the earthquake motion,
the less permeable capping surface crust layer (2m thick
with 2 degree slope with static shear stress of tst=1.2kPa)
begins to slide downward with a steady motion at the top of
the liquefiable layer having steady state (undrained) shear
strength ranging from qus=20 to 50kPa at the initial state.
• Sliding tends to localize just below the capping surface crust
layer.
• Tension zone shows tension fracture of the capping surface
crust layer
• Compression zone shows deformation of the capping
surface crust layer in compression shear mode
• All the above results are consistent with those observed

55
Mechanism in delayed flow failure
• Pore water migration into the sand just below the
capping clay layer⇒volume expansion of the
sand⇒reduction in qus
• When qus< t, delayed flow failure is triggered.

delayed flow failure

Less permeable surface crust (2m)

Liquefiable sand (10m)

Pore water migration

56
Suggestions for practice
• Permeabilities of surface crust layer and liquefiable
soil layer are the key parameters that govern the
occurrence of delayed flow failure and delay time.
• Permeable surface crust having higher permeability
than that of liquefiable soil does not develop
delayed flow failure. This fact should be beneficial
in engineering practice of risk assessment and
mitigation of delayed flow failure.

57
Imposed inflow analysis (aquifer)
• Excess pore water pressure of 68kPa at a depth of
10m without earthquake shaking

58
Coefficient of permeability (m/s)
qus=20kPa Case-1C Case-2C Case-3C
Layer S0 5E-7 5E-7 5E-5
Layer S1 1E-4 1E-5 1E-5
Layer S2 1E-7 1E-7 1E-7

qus=50kPa Case-1D Case-2D Case-3D

Layer S0 5E-7 5E-7 5E-5
Layer S1 1E-4 1E-5 1E-5
Layer S2 1E-7 1E-7 1E-7

59
Case-1C: less permeable surface crust
(high permeability contrast)
qus=20kPa

Initial vertical
effective stress

Imposed EPWP

60