Report - PEER - University of California, Berkeley

2.2 Simulation of Engineering Demand Parameters (EDP)For buildings, the most common EDPs are interstory drift ratios, inelastic componentdeformations (e.g., plastic hinge rotations), and floor accelerations. Both peak andresidual deformations are of interest, as the latter impact the post-earthquake repairand safety. Selection of EDPs is largely driven by one’s ability to reliably calculatethe EDPs, coupled with how well they correlate with relevant damage predictions.Inelastic time history analyses are emphasized for accurate EDP response predictionsover the full range of behavior, up through collapse. However, it is envisioned thatstatic inelastic pushover analysis methods will continue to be a viable option fordesign. Any type of inelastic simulation should be as realistic as possible, whereappropriate, taking into account soil-foundation-structure interaction and participationof “non-structural” components (cladding, masonry partitions, etc.).Prospects for accurate computation of the EDP relations vary with the targetEDP. For example, procedures for calculation of nonlinear dynamic response ofductile frames are increasingly becoming routine with validated analytical models andcomputational procedures. Simulation of structural collapse, especially for lessductile systems, remains problematic because of the lack of validated models to trackthe response of softening systems with large deformations. Nevertheless, progress isbeing made — evident, for example, in the collapse predictions in Figure 2.To accelerate the development and implementation of robust numerical models tosimulate inelastic structural response, PEER has embarked on the development of anopen-source, object-oriented software framework. OpenSees (Open System forEarthquake Engineering Simulation; http://opensees.berkeley.edu) is a collection ofmodules to facilitate the implementation of models and simulation procedures forstructural and geotechnical earthquake engineering. An emphasis within this effort isthe development, implementation, and validation of models to simulate collapse ofexisting non-ductile reinforced-concrete buildings, which due to inadequate seismicdesign, may experience severe strength and stiffness degradation under largeearthquakes (e.g., Elwood and Moehle 2003, Kaul 2004, Ibarra and Krawinkler2004).Approaches, such as the IDA technique described previously, permit one tosystematically characterize relationships between the EDP response quantities and theground motion IM. Mathematically, these relationships can be described by aconditional probability, P(EDP|IM), which captures the variability in the prediction ofresponse. In the example of Figure 2, the probability distribution, P(EDP|IM), woulddescribe the peak interstory drift ratios, conditioned on hazard intensity, Sa, where thevariability is solely the result of the ground motion characteristics (so-called “recordto-recordvariability”). While the ground motion and hazard characterization areknown to be a primary source of uncertainties, the simulations and resultingprobability distributions should account for other significant uncertainties in thestructural model itself, e.g., variation of material properties, modeling uncertaintiesassociated with the strength and deformation characteristics of structural components,variations in dead loads and seismic mass, etc.19