Assessing and managing the risk of earthquake-induced liquefaction to civil infrastructure - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 117

update available correlations and/or have revisited some of the established
corrections and procedures used in the liquefaction evaluation, e.g. Cetin
et al. (2004), Seed et al. (2003), Moss et al. (2006) and Idriss & Boulanger
(2006, 2008).
5.2.2 Deterministic vs probabilistic approaches
Liquefaction potential is related to a number of uncertain variables, includ-
ing in-situ resistance, ground water conditions, the composition of the soil,
as well as the loading factors (amplitude and duration of the ground
shaking). For any site, particularly a linear site extending over long dis-
tances, there are signifi cant degrees of uncertainty associated with all these
variables, which mean that to present liquefaction potential as a simple
binary case of 'liquefaction'/ 'no liquefaction' is an over-simplifi cation. The
majority of methodologies do adopt this type of deterministic approach to
assess liquefaction potential, based on empirical evidence, numerical mod-
elling or fi rst principles. However, a number of researchers (e.g. Juang et al. ,
2002; Cetin et al. , 2004; Moss et al. , 2006) have developed probabilistic
approaches based on the statistical scatter of the empirical data. They
provide a very useful framework for taking into consideration the uncer-
tainty in input parameters and presenting the results in the form of a prob-
ability of liquefaction ( P L ). These methods express the probability of
liquefaction as a function of both loading terms (peak ground acceleration
and moment magnitude M w ) and resistance terms (e.g. SPT blow counts,
fi nes content and vertical effective stress). However, these reliability-based
probabilistic methods are not fully probabilistic methods as the ground
motion hazard is not considered in a probabilistic manner. Since in most
cases ground motion hazard is assessed and defi ned probabilistically using
probabilistic seismic hazard analysis (PSHA), this leads to some inconsis-
tency and confusion.
As design procedures move towards a performance-based approach
(Kramer & Mayfi eld, 2007), it may be appropriate to determine the prob-
ability of liquefaction at a given site subjected to ground motions at a range
of hazard levels, and to understand facility performance if these levels are
exceeded. There are several methods that combine probabilistic ground
motions with conventional liquefaction potential procedures (Kramer &
Mayfi eld, 2007; Juang et al. , 2008). In these approaches the standard lique-
faction assessment follows a PSHA in which the joint probability distribu-
tion of the two key input parameters of the conventional liquefaction
assessment, PGA and earthquake moment magnitude for earthquake sce-
narios, is determined. Fully probabilistic methods for liquefaction potential
evaluation (Goda et al. , 2011) should take into account uncertainties in both
ground motion (i.e. earthquake occurrence and ground-motion intensity)