Mechanisms of failure in large landslides

1
Mechanisms of failure in large landslides David
Petley International Landslide Centre University
of Durham
2

• Aims
• To gain a better understanding of deformation
  processes in landslides
• To develop a new model for the development of
  failure in first time landslides

3
The Saito technique

• Saito (1965, 1969) noted that during periods of
  acceleration to failure, plotting 1/velocity
  against time produced a straight line
• Conceptualised by Voight (1988, 1989)
• Developed by Fukozono (1990)
• Range of approaches used to predict time of final
  failure (see Federico et al. 2002)

Time of catastrophic failure
1/velocity
0
Time
4
The Vajont example
Failure
5
Developing the Saito approach

• But the Saito approach does not take into
  account the soil mechanical behaviour and its
  influence on the progression of the landslide
  (Federico et al. 2002)
• i.e. no explanation of why the Saito approach
  works
• no systematic analysis of whether the Saito
  approach applies universally
• What can the accelerating behaviour tell us about
  basal deformation processes?

6
Selborne Landslide 1988-89Deliberately induced
failure in Gault Clay
Picture courtesy of Prof E Bromhead
7
Selborne LandslidePost-failure morphology

• Picture courtesy of Prof E Bromhead

8
Selborne Landslide

• Picture courtesy of Prof E Bromhead

9
Selborne landslide failure
10
Selborne linearity
First seen here
Then seen progressively up and down slope
11
Butlinearity is not always seen
Data from A Angeli et al. (1989) B Salt
(1985) C CNR IRPI (2002) D USGS (2002)
12
Findings

• Saito linearity does not always apply!
• A second pattern is observed for many landslides
• asymptotic trend
• This trend is always associated with either
• Sliding along existing shear surfaces
• Deformation in ductile layers
• The asymptotic trend is seen in landslides that
  are undergoing rupture surface formation (crack
  growth) (Petley et al 2002 Geology, 8, 719-22)

13
Laboratory testing

• Verification of the link between monitoring data
  and material deformation can only be achieved
  through laboratory testing
• Use of novel stress path testing approach

14
(No Transcript)
15
? 27
16
(No Transcript)
17
Gault slow test10 kPa per day pore pressure
increase
18
Gault slow test10 kPa per day pore pressure
increase
19
Gault clay instant pore pressure increase
20
Undisturbed Haney Claypo 515 kPa, q 807 kPa
Data from Capanella and Vaid (1974)
21
Undisturbed Haney Claypo 515 kPa, q 807 kPa
Data from Capanella and Vaid (1974)
22
Tessina Landslide (Italy)
23
200 kPa reinflation experiment 1CP 400 kPa,
BPi 200 kPa, BPf 360 kPa
24
310 kPa reinflation experimentCP 450 kPa, BPi
140 kPa, BPf 400 kPa
25
200 kPa reinflation experiment 2CP 400 kPa,
BPi 200 kPa, BPf 390 kPa
26
(No Transcript)
27
Conclusion from laboratory tests

• Linearity is associated with rupture surface
  formation
• Asymptotic trend is associated with ductile
  deformation or sliding on an existing surface
• Therefore we can propose a new conceptual model
  for the development of a first time failure

28
Initiation of rupture surface formation
Normal range of FoS from pore pressure
variations
1.x
Rupture surface formation begins WHY? Shear
strength is exceeded locally
Factor of Safety
1
Time
29
Pore pressure decrease stops rupture surface
development, but overall FoS has been reduced
1.x
Pore pressure increase beyond new threshold
reinitiates rupture surface development
Factor of Safety
1
Time
30
As rupture surface develops, shear stress on
remaining undeformed material increases with
exponential function
1.x

• Therefore
• velocity increases exponentially
• FoS vs time gradient increases
• After a critical point, pore
• pressure no longer controls the
• system
• Failure is probably unavoidable

Factor of Safety
1
Time
31
1.x
When rupture surface fully formed, failure occurs
and FoS 1
Factor of Safety
1
Time
32
In the early phase of the development of failure
the system is stress controlled
1.x
Factor of Safety
1
Time
33
In the early phase of the development of failure
the system is stress controlled
1.x
Factor of Safety
In the latter stages the system is driven by
stress, but final failure occurs at a critical
strain it is strain controlled
1
Time
34
Conclusion I

• Two styles of accelerating behaviour are seen in
  landslides
• Linearity rupture surface development
• Asymptotic sliding on existing surfaces /
  ductile deformation
• Now have a fundamental understanding of when /
  why Saito approach works
• Analysis of movement records and laboratory tests
  allow an understanding of basal deformation
  processes

35
Conclusion II

• A new model for the development of progressive
  failure in brittle landslides is proposed
• System is initially driven and controlled by
  stress
• Once rupture surface reaches unstable crack
  growth stage, system is stress driven but failure
  is strain controlled this is the CRITICAL
  STRAIN
• Period of development of final failure can be
  very long (Vajont 70 days)
• This may be why slopes sometimes fail in the
  summer with no apparent trigger

36
(No Transcript)