Probabilistic seismic demand analysis of nonlinear structures

Recent earthquakes in California have initiated improvement in current design philosophy and at present the civil engineering community is working towards development of performance-based earthquake engineering of structures. The objective of this study is to develop efficient, but accurate procedures for probabilistic analysis of nonlinear seismic behavior of structures. The proposed procedures help the near-term development of seismic-building assessments which require an estimation of seismic demand at a given intensity level. We also develop procedures to estimate the probability of exceedance of any specified nonlinear response level due to future ground motions at a specific site. This is referred as Probabilistic Seismic Demand Analysis (PSDA). The latter procedure prepares the way for the next stage development of seismic assessment that consider the uncertainties in nonlinear response and capacity. The proposed procedures require structure-specific nonlinear analyses for a relatively small set of recorded accelerograms and (site-specific or USGS-map-like) seismic hazard analyses. We have addressed some of the important issues of nonlinear seismic demand analysis, which are selection of records for structural analysis, the number of records to be used, scaling of records, etc. Initially these issues are studied through nonlinear analysis of structures for a number of magnitude-distance bins of records. Subsequently we introduce regression analysis of response results against spectral acceleration, magnitude, duration, etc., which helps to resolve these issues more systematically. We illustrate the demand-hazard calculations through two major example problems: a 5story and a 20-story SMRF building. Several simple, but quite accurate closed-form solutions have also been proposed to expedite the demand-hazard calculations. We find that vector-valued (e.g., 2-D) PSDA estimates demand hazard more accurately. This procedure, however, requires information about 2-D seismic hazard which is a relatively new tool. We determine also the required number of analyses to estimate the seismic demand and demand-hazard with a certain accuracy. By considering different sources of uncertainties in those estimations we recommend a number of analyses so that the change in the total uncertainty due to limited number of analyses is within an acceptable limit. ed number of analyses is within an acceptable limit.