8CUEE CONFERENCE PROCEEDINGS 8th International Conference on Urban Earthquake Engineering March 7-8, 2011, Tokyo Institute of Technology, Tokyo, Japan SEISMIC PERFORMANCE OF A SYSTEM OF INTERDEPENDENT LIFELINE AND INFRASTRUCTURE COMPONENTS Jacopo Selva 1) , Kalliopi Kakderi 2) , Maria Alexoudi 3) , and Kyriazis Pitilakis 4) 1) Istituto Nazionale di Geofisica e Vulcanologia, Sezione di Bologna, Italy 2) Civil Engineer, MSc, Dept. of Civil Engineering, Aristotle University of Thessaloniki, Greece 3) Civil Engineer, MSc, PhD, Dept. of Civil Engineering, Aristotle University of Thessaloniki, Greece 4) Professor, Dept. of Civil Engineering, Aristotle University of Thessaloniki, Greece selva@bo.ingv.it, kakderi@civil.auth.gr, alexoudi@civil.auth.gr, kpitilak@civil.auth.gr Abstract: The aim of this research is to develop a generic procedure for the assessment of the serviceability of a system (single system or system of systems) if one or more interacting components of the system are damaged by an earthquake. The system serviceability (functionality) is evaluated starting from the expected degree of damage of the single components (direct physical damage estimated using appropriate fragility functions) and accounting for their functional interaction (functional system architecture and single or by-directional interactions among components). Through the evaluation of the components’ non-functionality, the overall serviceability of the system is assessed, possibly in the form of a “system serviceability curve”, for different levels of seismic input intensity. Aleatory and epistemic uncertainties are treated using a Bayesian inference. The applicability of the proposed approach is established through an illustrative example. It is shown that the methodology is quite general and applicable to real systems with diverse degrees of complexity and knowledge of system and components details. 1. INTRODUCTION In the framework of a comprehensive risk analysis and management of interacting lifeline and infrastructure systems, their seismic performance should be considered in a rigorous and unified way. This could only be achieved through the assessment of the systems’ functionality, considering the complexity of structures and the interdependencies among systems and their components. Several approaches are available to help describe the relations existing between system’s components. Some of these are: Graph theory, Fault-tree analysis (FTA), Event-tree analysis (ETA), Series system in parallel (SSP), Agent-based models and Complex Adaptive Systems (Amin 2001, Little 2002, Brown et al. 2004, Bernhardt and McNeil 2004, Tolone et al. 2004). The probabilistic evaluation of the performance of the system (PNET) can be carried out employing the methods of System Reliability Analysis. These include expansion methods, such as FORM/SORM or the response surface technique (Ditlevsen and Madsen 1996), as well as the Monte Carlo simulation methods (Rubinstein 1981). Depending on the nature of uncertainty and the aptitude to determine them, the probabilistic approach may be replaced or enhanced by possibilistic approaches based on Fuzzy Logics, so-called Fuzzy Networks. Also, non-simulation methods have recently seen interesting advances, e.g. in the form of Matrix System Reliability Analysis (Song and Der Kiureghian 2003, Der Kiureghian and Song 2008). Moving to a higher level, several researchers have proposed different types of interdependency (interactions between different critical infrastructures) simulation models (Kameda 2000, Giannini and Vanzi 2000, Rinaldi et al. 2001, Peerendoom et al. 2001, Amin 2001, Haimes and Jiang 2001, Little 2002, Li and He 2002, Tang et al. 2004, Yao et al. 2004, Brown et al. 2004, Bernhardt and McNeil 2004, Santos and Haimes 2004). Only few methodologies have incorporated interdependencies in the seismic risk analysis of lifelines (Nojima and Kameda 1991, Scawthorn 1992, Eidinger 1993, Shinozuka et al. 1993, Shinozuka and Tanaka 1996, Menoni 2001, Duenas-Osorio et al. 2007, Tang and Wen 2008). Furthermore, very few studies can be found in the literature dealing with the highest level problem of multiple systems interaction in the case of seismic vulnerability and loss estimates (Duenas-Osorio et al. 2007, Kim et al. 2007). They have still an exploratory character and are based on rather extreme simplifications, being limited to the analysis of at most two systems. Network analysis and graph theory are usually adopted. The Bayesian approach is often used for network analysis. The systems’ serviceability is also analyzed using flow or connectivity analysis. Some recent studies have been focused on the proposal of a methodology to evaluate the associated losses of interacting lifeline elements for various strong motion intensities and the estimation of complex fragility curves of interdependent components (Kakderi et al. 2007, Kakderi et al. 2008, Alexoudi et al. 2008a, b).