Report - PEER - University of California, Berkeley

was developed to accommodate these needs. In this equation, probabilistic functionslink earthquake intensity measures (IMs) with engineering demand parameters(EDPs). This relationship brings the engineer to what was traditionally the end of theanalysis. However, the PEER framework equations also provides a basis for goingbeyond EDPs by employing probabilistic relationships that link EDPs with damagemeasures (DMs) and subsequently DMs with decision variables (DVs). Specifically,in Eq. 1, ν ( DV ) is the mean annual probability that the decision variable DV exceedsa specific value, G DV DM is the conditional probability that DV exceeds aspecific value given a particular value of DM, dG DM EDP is the derivative (withrespect to DM) of the conditional probability that DM exceeds a limit value given aspecific value of EDP, dG EDP IM is the derivative (with respect to EDP) of theconditional probability that EDP exceeds a limit value given a specific value of IM,and dλ ( IM ) is the derivative of the seismic hazard curve, λ ( IM ).Multiple approaches are appropriate for incorporating information about buildingresponse into Eq. 1. For example, a building-specific EDP, such as maximum roofdrift, could be used to predict the damage state of the building, or component-specificEDPs could be used to predict the damage state of individual component, with thedamage state of the building defined by the cumulative damage states of all of thecomponents. It is generally accepted that the latter approach provides an opportunityto introduce more information into the process and thereby reduce uncertainty. Themodels developed here support the latter approach. Specifically, the results ofprevious research and practical experience are used to develop probabilisticrelationships linking EDPs with DMs for one type of structural component, olderbeam-column building joints.2. EXPERIMENTAL DATAA critical phase of this research effort was the identification of experimental datacharacterizing the progression of earthquake damage in older beam-column joints.The criteria used to choose laboratory test specimens, the characteristics of theexperimental test specimens, and variation in test specimens that could be expected toaffect damage progression are discussed in the following sections.2.1 Criteria Used To Identify Appropriate Laboratory Test SpecimensThree criteria were used to identify experimental data sets for use in this study. First,only laboratory specimens representative of pre-1967 construction were used. Areview of construction drawings for buildings designed prior to 1979 for constructionon the West Coast (Mosier 2000) were used as a basis for defining design details forolder joints. Table 1 lists Mosier’s statistics for critical joint design parameters. The210