Seismic risk assessment of water supply systems - Seismic Risk Analysis and Management of Civil Infrastructure Systems - page 662

facilities under seismic loading is assessed primarily by geotechnical engi-
neers, who have developed models for soil-structure interaction, including
empirical models based on observations from the past earthquakes, closed-
form analytical methods, and numerical simulations, such as fi nite element
and fi nite difference analyses (e.g., O'Rourke and Liu 1999; O'Rourke et
al. 2004a, 2004b; Shi 2006; Wang 2006). Structural engineers focus typically
on the performance of above-ground facilities. The principles of structural
dynamics (e.g., Chopra 2001) are routinely applied in the engineering analy-
sis and design of above-ground structures. Theoretical models, as well as
advanced numerical models, have been developed and applied for analyzing
the above-ground structure responses to ground shaking (e.g., Naeim 2003;
Chopra 2005).
To account for uncertainty with respect to component or facility response,
seismic behavior is frequently characterized by fragility curves that provide
the probability of failure as a function of the demand (e.g., PGV or peak
ground acceleration, PGA). Fragility curves can be derived from either
observations of past earthquakes or, more typically, Monte Carlo simula-
tions that have special capability in quantifying uncertainty.
24.5.1 LADWP system component responses
Seismic risk assessment of the LADWP system has been focused on seismic
wave (TGD) interactions with pipelines, which have a primary impact on
system performance (Wang and O'Rourke 2008). Analytical and/or numeri-
cal methods have been developed to model the seismic wave interaction
with pipelines (e.g., O'Rourke et al. 2004b; Shi 2006; Wang 2006), and they
can be utilized to evaluate the seismic performance of pipelines. However,
to apply those models, details are required on seismic wave characteristics
(e.g., predominant period and apparent wave velocity), pipeline properties
(e.g., cross-sectional area and axial deformation stiffness), and ground con-
ditions (e.g., burial depth, unit weight of soils, and pipe-soil interface friction
angle). Such information is quite often not available, particularly for a large
and complex system, such as the LADWP system, which contains about
12000 km pipes covering an area of approximately 1200km 2 . One alter-
native method, which requires minimal input parameters, for estimating
pipeline damage during earthquakes was to use regressions between
observed repair rates and measured seismic parameters during previous
earthquakes.
24.5.2 Regressions for estimating pipeline damage
The damage to cast iron and ductile iron pipelines was estimated using the
regressions developed by Jeon (2002). Empirical data were collected about