Scribd
Upload
55 views

Lecture Probabilistic Approach in Slope Stability Analyses

Probabilistic slope stability analysis recognizes that slopes have some probability of failure, even if small. It defines the failure or safety of a slope using a performance function G(X) that depends on random input parameters X. Failure occurs if G(X)<0 and safety if G(X)>0. The performance function is commonly the factor of safety F(X). The probability of failure is defined as P(F<1). Reliability is defined as 1 - probability of failure. A reliability index β combines the mean and standard deviation of F to quantify uncertainty in the slope's stability.

Uploaded by

Zanele Mbatha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
55 views

Lecture Probabilistic Approach in Slope Stability Analyses

Probabilistic slope stability analysis recognizes that slopes have some probability of failure, even if small. It defines the failure or safety of a slope using a performance function G(X) that depends on random input parameters X. Failure occurs if G(X)<0 and safety if G(X)>0. The performance function is commonly the factor of safety F(X). The probability of failure is defined as P(F<1). Reliability is defined as 1 - probability of failure. A reliability index β combines the mean and standard deviation of F to quantify uncertainty in the slope's stability.

Uploaded by

Zanele Mbatha
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 26

Probabilistic approach in slope

stability analyses
Introduction

• Probabilistic approach in slope stability analysis recognises that any


earth structure has some probability of failure, however small.

• In contrast, a deterministic approach leads to an expectation that failure


of a slope cannot occur under the conditions for F > 1.

• Failures of slopes designed to be safe are not uncommon (Chowdhury,


1984).
Performance function

• In a probabilistic approach, the failure or safety state of a slope can be


defined by the performance function G(X) where X denotes the collection
of random input parameters.

• The performance function is defined in such a way that failure of the slope
is indicated by G(X) < 0 and safety by G(X) > 0 (Mostyn & Fell, 1997).

• G(X) = 0 is called the limit state boundary.


Performance function

• The performance function for a slope is usually taken as one of the


following formats:

G(X) = R(X) – S(X) 1

or G(X) = F(X) – 1 2

where R(X) is the resistance and S(X) is the action and F(X) is the factor of
safety.

• Equation 2 is more common than equation 1 because the safety of slopes is


traditionally characterized by the factor of safety.
Performance function

• The performance function for a slope is usually formulated using the


limit equilibrium analysis.

• Once the performance function is defined, the probability of failure of


a slope can then be determined.

• The probability of failure is defined as the probability that the


performance function has a value below the threshold value (Aleotti &
Chowdhury, 1999).
Performance function
• If (F) is the performance function and the threshold value is 1, then the
probability of failure (Pf) is given by:

Pf = P[F < 1] 3

• The reliability defined as the probability that a slope will be safe, is the
probability of success (Ps) and is the complement of Pf.

• Thus: Ps = 1 – Pf 4
a

Figure 1. Two probability distributions of the factor of safety, F,


frequently used in probabilistic calculations; a) normal distrbution,
b) lognormal distribution (modified from Chowdhury, 1984).
Probabilistic modelling of soil properties

• Properties of soil have generally been modelled using the single-random-


variable (SRV) approach in which the random variation of a soil property
within the soil profile is represented by one single random variable.

• This implies that the realisations of a soil property are the same at every
location within the soil profile.

• Such an approach is erroneous because soil and rock properties do vary


from point to point even within a statistically homogeneous profile.
Probabilistic modelling of soil properties

• Use of the SRV approach results in a gross overestimation of the failure


probability of slopes because it ignores the variance factor due to spatial
averaging.

• The random variable or random field model, which gives a more realistic
prediction of the failure probability, is now well accepted as a useful model
for probabilistic modelling of soil profiles.

• Properties of soil which are uncertain are best defined as random variables
described by a mean and standard deviation.
Probabilistic modelling of soil properties

• The random variable is described by its probability law, which in turn may
be characterised in a number of ways; most common is through the
probability distribution of the random variable.

• Quantities summarizing only the dominant features of the distribution may


suffice.

• The most common such descriptions are the mean, the variance, the
standard deviation and the coefficient of variation (COV) which is the ratio
of the standard deviation to the mean.
Probabilistic modelling of soil properties

• For a continuous random variable, this is represented by the probability


density function (PDF)

• A PDF describes the relative likelihood that a random variable will assume
a particular value in contrast to taking other values.

• Most common distribution encountered in geotechnical engineering is the


normal distribution which is a bell-shaped curve.

• Other distributions are sometimes used to match closely sampled data.


Figure 2. a) Deterministic and statistical distribution of soil
property; b) typical probability distribution functions used in
geotechnical problems (after Lacasse & Nadim, 1996).
Probabilistic modelling of soil properties

• Distributions of soil properties must be determined on the basis of available


data and it can be verified whether a particular empirical distribution
follows any well known mathematical PDF such as normal or Gaussian,
lognormal, exponential or beta.

• Where observations are scarce or absent, parametric distributions can be


assumed from the literature.
Table 1. Typical probability density functions and their
properties (after Lee et al., 1983).
Probabilistic modelling of soil properties

• Care should be exercised however, to ensure that the minimum and


maximum values of the selected distribution are consistent with the
physical limits of the parameter being modelled.

• For example, shear strength parameters should not take negative values.

• If the selected distribution implies negative values, then the distribution is


truncated at a practical minimum threshold.
Probability of failure
• Each parameter affecting slope stability may be regarded as a random
variable with an associated probability density function (PDF) rather than a
constant.

• Variability of some parameters such as unit weight has an insignificant


influence on stability analysis.

• Such parameters may be regarded as constants whereas the random


variables are typically the shear strength parameters c and φ for the soil
materials, and the pore pressure.
Probability of failure
• Results from slope stability calculations using random variables as soil
parameters is a distribution of the calculated factor of safety rather than a
single value, whose PDF may be characterised by means of at least two
statistical parameters:

1. The mean or central value σF, and

2. The standard deviation µF.

• If the probability distribution for each random variable in a slope stability


analysis could be defined, then, in principle, the probability distribution of
the factor of safety could also be constructed.
Probability of failure
Reliability Index
• In practice, there will rarely be sufficient data to justify a full probabilistic
analysis but it would be possible to estimate the mean value of the safety
factor and its standard deviation.

• From these it is possible to determine a standardised reliability index


β,which combines the mean and standard deviation of the performance
function, and is more sensible than the conventional safety factor.

• The reliability index is given by:


 F  1
F
Probability of failure
Reliability Index (β)
• When the shape of the PDF of F is known, the reliability index can be
related to the probability of failure.

• The reliability index is essentially a normalization of the conservatism in


the value of F with respect to the uncertainty in F.

• It may be regarded to as an index of uncertainty and it can be related to the


probability of failure if the frequency distribution is known.

• It incorporates more information on the uncertainty in the values of F.

• Slopes with larger values of β are farther from failure than slope with small
values of β regardless of the best estimate of F.
Figure 3. Probability of failure for normally distributed
factor of safety (after Christian et al., 1994).
Figure 4. Comparisons of two slopes with
different means and distributions of factor of safety
(modified from Christian et al., 1994).
Probability of failure

Reliability Index

• Wolf (1996) suggested that a reliability index of 3 for routine slopes


and 4 for critical slopes is acceptable.

• The U.S. Army Corps of Engineers (1999) recommended the


characterization of safety based on reliability index for implementation
in geotechnical practice.
Expected levels of performance in terms of Pf and β (after U.S.
Army Corps of Engineers, 1999)
Expected performance Reliability index (β) Probability of failure
level (Pf)
High 5.0 0.0000003
Good 4.0 0.00003
Above average 3.0 0.001
Below average 2.5 0.006
Poor 2.0 0.023
Unsatisfactory 1.5 0.07
Hazardous 1.0 0.016
Probability of failure

Reliability Index
• The definition of β does not require that the probability distribution
of F be known but it is assumed that F is distributed in such a way
that can be described meaningfully by its mean and standard
deviation.

• β expresses uncertainty in the stability of the slope.

• Probabilistic approach in slope stability analyses incorporates the


uncertainty in the geotechnical parameters and permits the
quantification and visualization of the effects of uncertainty.
Probability of failure

Reliability Index
• Probabilistic approach in slope stability analyses incorporates the
uncertainty in the geotechnical parameters and permits the quantification
and visualization of the effects of uncertainty.

• F is probabilistically quantified as a function of the uncertainty in the


parameters c and φ, and the result expressed as the reliability index β, is
more robust than the factor of safety alone.
Probability of failure

• What is the accepted value of Pf in slope stability analyses?

You might also like