Seismic risk analysis using Bayesian belief networks - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 178

2002; Mollamahmutoglu et al. 2003), for example, showed that liquefaction
and related phenomena can still be responsible for a tremendous amount
of building damage. The 1964 Alaska earthquake (March 27, 1964, Anchor-
age, Alaska, M w 8.6) and 1964 Niigata earthquake (June 16, 1964, Niigata,
Japan, M w 7.5), for example, have highlighted the dramatic consequence of
liquefaction-induced damage (Bird et al. 2006). Lateral spreading is the
most pervasive type of liquefaction-induced ground failure, which is an
indicator of damage severity that ranges from a few centimeters to several
meters (Abdoun and Dobry 2002). Lateral spreading induced damage to
deep foundation (e.g. piles) is costly and various studies are reported in the
literature (Abdoun and Dobry 2002; Takahashi and Takemura 2005; Finn
and Fujita 2002; Miura et al. 1991). Hence, a reliable liquefaction and lateral
spreading risk assessment tool is of utmost importance for loss assessment
in seismic risk analysis (Bradley et al. 2010). Discernment of liquefaction
and lateral spreading vulnerability, however, is a complex and nonlinear
procedure that is infl uenced by model and parameter uncertainty. With
increasing availability of in situ data, however, soft computing techniques
can be used for identifi cation of sites prone to liquefaction and lateral
spreading risk.
7.3.1 Application of soft computing for liquefaction
Various empirical and semi-empirical formulations of liquefaction quanti-
fi cation are reported in the literature (Youd and Idriss 2001). Hwang et al.
(2004) developed a reliability-based method for assessing soil liquefaction
potential. Liyanapathirana and Poulos (2002) have developed a numerical
model for dynamic soil liquefaction analysis. Idriss and Boulanger (2006)
have developed semi-empirical procedures for evaluating liquefaction
potential during earthquakes. Cabalar et al. (2012) have developed an
Adaptive Neuro-Fuzzy Inference System (ANFIS) for evaluating liquefac-
tion. Juang et al. (2002) have developed probability-based methods for
liquefaction potential evaluation using logistic regression and Bayesian
mapping. ANN techniques have also been used to model the liquefaction
potential (e.g. Baziar and Nilipour 2003; Goh 1994, 1996, 2002; Tung et al.
1993).
7.3.2 Application of soft computing for lateral spreading
To predict the lateral displacement, various empirical (Valsamis et al. 2010;
Finn and Fujita 2002; Tamate and Towhata 1999) and numerical methods
have been proposed. Bartlett and Youd (1992, 1995) have introduced
an empirical model using a multi-linear regression model (MLR). The
empirical models were revised by Youd et al. (2002) due to changes of the