Vulnerability Assessment for Earthquake Liquefaction–Induced Settlements of an Embankment Using Gaussian Processes

Abstract

The major cause of earthquake damage to an embankment is the liquefaction of the soil foundation that induces ground level deformations. The aim of this paper is to assess numerically the effect of the liquefaction-induced settlement of the soil foundation on an levee due to real earthquakes. The seismic vulnerability is evaluated in terms of analytical fragility curves constructed on the basis of non-linear dynamic Finite Elements (FE) analysis.However, FE analysis can be expensive due to very large number of simulations needed for an accurate assessment of the system failure behaviour. This problem is addressed by building a Gaussian Process (GP) emulator to represent the output of the expensive FE model. A comparison with the FE reference results suggests that the proposed GP model works well and can be successfully used as a predictive tool to compute the induced damage on the levee. Findings also illustrate clearly the importance and the advantages of an adequate definition of the input parameters to built the GP model.

Survival analysis of a liquefiable embankment subjected to sequential earthquakes

“…where h 0 (t) is the baseline hazard and β is a parameter that represents the 163 effect of covariate on the outcome. Assuming that one event occurs at a time 164 t i , the parameter β can be calculated by solving the partial likelihood: [43], Lopez-Caballero and Khalil [34], and is detailed in Figure 2.…”

“…It is calculated by considering each ground motion as a single event. The percentage relative crest settlement as calculated by Swaisgood[58] is the ratio of the vertical displacement of the crest to the height of the dam with its corresponding foundation: δu z,rel /H, given that H in this study is 19 m. The relative crest settlement is divided into damage levels[58,34]. The limit values of these levels is still debatable but the ones chosen for this study are shown as dashed lines inFigure 6.…”

Empirical damage and liquefaction fragility curves from 2012 Emilia earthquake data

“…Zhang et al (2008), however, derived fragility functions for different classes of typical bridges in California when subjected to seismic shaking or liquefaction-induced lateral spreading. Lopez-Caballero and Khalil (2018) assessed numerically the effect on a levee of the liquefaction-induced settlement of the soil foundation in terms of analytical fragility curves constructed on the basis of a nonlinear dynamic analysis. Furthermore, using a database of buildings affected by liquefaction during the 2010 Mexicali, 2010 New Zealand, and 2011 Japan earthquakes, and with reference to building damage, Cazares et al (2012) proposed vulnerability functions that are obtained as a result of a combination of the functions defined as empirical and those obtained through damage statistics.…”