ISSUES IN SEISMIC RISK ASSESSMENT OF TRANSPORTATION NETWORKS - Earthquake Geotechnical Engineering - page 451

CHAPTER 19
ISSUES IN SEISMIC RISK ASSESSMENT OF TRANSPORTATION
NETWORKS

Anne S. Kiremidjian, Evangelos Stergiou, and Renee Lee
Stanford University, Department of Civil and Environmental Engineering, Terman 238,
Stanford, CA 94305-4020, USA
Tel: [phone]
Fax: [phone]
[e-mail]

Abstract. Seismicriskassessmentoftransportationnetworkshasbeenthesubjectofseveralstud-
iesoverthepasttwodecades.Manyadvanceshavebeenmadeduringthistime,however,numerous
issues remain. Recent research addressing some of these issues has demonstrated their importance
for rational decision making. In this paper an overview is presented of the most commonly used
transportation network risk assessment methodology. Important issues that have received limited
attention over the years are identified and recent developments to address these issues are pre-
sented. The paper draws on research conducted under the Pacific Earthquake Engineering Center
with many contributors (Moore et al., 2005; Fan and Nie, 2006; Kiremidjian et al., 2003; Lee and
Kiremidjian 2006; Stergiou and Kiremidjian, 2006).

Formulationsfornetworkfunctionalitylossunderascenarioeventandforasuiteofpossibleearth-
quakes that can affect the system are developed first. It is shown that functionality loss is not only
an important part of the risk assessment, but is of the same order of magnitude and in some cases
is greatly exceeded than the loss from direct physical damage to network components. Liquefac-
tion appears to be the most important hazard in the loss computation for such systems; however,
more robust models are needed to determine the degree to which liquefaction dominates the risk.
Correlationofgroundmotionbetweenbridgesitesandcorrelationofdamagebetweenbridgeswith
comparable designs are often ignored in transportation risk analysis. It is again demonstrated that
thesecorrelationsareimportantcontributorstotheuncertaintyofloss.Whentheriskfromallearth-
quakes events is considered, network functionality analysis becomes computationally intractable.
A simple method based on Monte Carlo simulationwith importance sampling ispresented forthat
purpose.Thevariousconclusionsareillustratedthroughanapplicationtoasamplenetworkwithin
the San Francisco Bay region.

1. Introduction

The 1989 Loma Prieta earthquake closed 142 roads in the San Francisco Bay Area,
several of which remained closed for more than six months. Five years later, the 1994
Northridge earthquake caused approximately the same number of closures. More than a
dozen remained closed for several months after that event as well. A study by ABAG
(1997) reports that if the Peninsula segment of the San Andreas Fault ruptures an esti-
mated 428 roads may be closed. In the same study a rupture on the northern segment of
the Hayward Fault would result in nearly 900 roadways closures. In the worst case