LOSS ASSESSMENT OF LIFELINE NETWORKS CONSIDERING THE EFFECT OF DAMAGE SPATIAL CORRELATION Alireza GARAKANINEZHAD 1 , Morteza BASTAMI 2 ABSTRACT Seismic loss assessment for a spatially distributed system, such as lifeline networks, depends on the distribution of ground-motion intensity measures (IMs) over region and the vulnerability of structure in the given region. The proper treatment of variability of both ground motion and structural vulnerability is essential. In particular, it has been shown that spatial correlation of both these parameters has an important impact. In recent years, some models have been proposed to quantify the spatial correlation of IMs; however, there is little attention to structure-to-structure damage correlation. In this paper, a methodology based on spatial statistics for considering this correlation in the network risk assessment is proposed. In this methodology, direct loss of structures imposed an earthquake event is considered as a random field whose cross-covariance function is presented using semivariogram functions. The parameters of the model can be estimated from spatial characterization of the expected loss of structures in each damage state. This framework is shown on a transportation network located in Tehran and the effect of damage correlation is investigated. The results indicate that spatial correlation of damage has a significant effect on risk estimate, especially may lead underestimation of risk assessment in rare event cases. Keywords: Lifeline systems; Spatial correlation; Damage correlation; Reliability & Risk Analysis; Transportation networks 1. INTRODUCTION Lifeline networks such as transportation networks, electrical networks, and water distribution networks play a significant role in the capability of a society to come back after a natural hazard like earthquake. In fact, the recovery activities are highly dependent on the performance of these networks during hazards. For this reason, estimation of the performance of lifelines received much attention. However, this estimation has accompanied highly uncertain. In general, spatial correlation of ground- motion intensity measure (IM) and the structural damage result in these uncertainties. These years, some researchers have proposed frameworks to incorporate correlation of both IMs and structures. Several research works studied the effect of spatial correlation of ground-motion IMs on the loss estimation of lifeline networks. It is indicated that for rare events, ignoring spatial correlation of IMs may result in underestimation of the risk assessment, and overestimation in frequent events. In this way, models for different IMs (peak ground acceleration [PGA], peak ground velocity [PGV], peak ground displacement [PGD], and spectral accelerations [SAs]) are presented based on records from different regions. Boore et al. (2003) developed a spatial correlation model of PGA using 1994 Northridge earthquake observations. Wang and Takada (2005) presented a model of PGV using several earthquakes occurred in Japan and the 1999 Chi-Chi. Other studies, such as Goda and Hong (2008), Jayaram and Baker (2009), and Hong et al. (2009) used some well-recorded earthquakes to develop the spatial correlation model for different IMs. In these works, records of each event are 1 Ph. D. student, International Institute of Earthquake Engineering and Seismology (IIEES), Tehran, Iran, a.garkani@iieess.ac.ir 2 Associate Professor, International Institute of Earthquake Engineering and Seismology (IIEES), Tehran, Iran, m.bastami@iiees.ac.ir