Report - PEER - University of California, Berkeley

y an earthquake, of which bridges are an integral part. Today, answers to thesequestions are more qualitative than quantitative, mostly based on experience andengineering intuition rather than results of analyses and engineering evaluations.Furthermore, after an earthquake decisions must be made quickly: there is often notime to perform extensive engineering investigations. The goal of our research is toprovide an engineering basis for evaluating the ability of a typical highway bridge tofunction after an earthquake. We address three limit states:1. Repair limit state: to assess how much it may cost to repair a bridge;2. Traffic function limit state: to assess the magnitude of traffic load that can besafely carried by a damaged bridge; and3. Collapse limit state: to asses if the bridge is passable or not.The highway overpass bridges under consideration in this study were chosenbecause they represent close to 90% of all bridges in typical regional highwaynetworks in the U.S. (Basöz 1997). The particular bridges, typical for California, aredetailed in (Mackie 2003). In summary, these reinforced concrete highway overpassbridges have two equal spans, a single bent with a single column, a pile shaftfoundations, and roller abutment supports. Variations of a number of the bridgedesign parameters were studies, but are not the subject of this paper.2. PROBABILISTIC FRAMEWORKThe Pacific Earthquake Engineering Research Center’s (PEER) probabilisticperformance-based design and evaluation approach provides the framework forbridge function evaluation. Data from seismology studies was used to assess theground motion intensity measures (IM). Structural analysis using finite elementmodels was performed to links engineering demand parameters (EDP), for a family oftypical U.S. highway overpass bridges, to ground motion intensity measures usingOpenSees software for non-linear dynamic seismic structural response simulation.The PEER structural element performance database was used to link engineeringdemand parameters to damage measures (DM) in typical bridge structural elementssuch as columns. A combination of finite element simulations and reliability analyseswere employed to develop damage measures pertinent to bridge function. Finally, anumber of decision variables were developed that describe the considered limit statesin terms of measures of induced damage.Previous research (Mackie 2003) has produced a sizeable collection ofinformation regarding Probabilistic Seismic Demand Models (PSDM) that relateEDPs to IMs. PSDMs are generated using Probabilistic Seismic Demand Analysis(PSDA). Two approaches to PSDA include the cloud approach to vary the seismicdemand (IM), and a scaling approach to reach prescribed intensity levels. Theresulting PSDMs may assume any mathematical form; however, erudite choicessimplify the evaluation using the PEER framework. Selections of PSDMs that areoptimal in this regard are detailed elsewhere (Mackie 2003). One studious PSDMform is linear in log space.54