Probabilistic seismic demand models developed with two-criteria ground motion scaling for reliable risk assessment

Abstract

The present study aims to identify the optimal criteria for the Normalization Measure (NM) and Intensity Measure (IM) in the two-criteria scaling process of earthquake Ground Motions (GMs) to reduce uncertainty in the development of Probabilistic Seismic Demand Models (PSDMs) for more reliable seismic risk assessment. For this purpose, five 2-, 4-, 8-, 12-, and 20-story steel buildings with a seismic force-resisting system consisting of perimeter Special Moment-resisting Frames (SMFs) located in Los Angeles, California, are selected as the case study. Four sets of GMs including 160 real GMs with mid to large magnitudes at near to moderate distances are selected from Baker’s GM database. The maximum values of transient and residual inter-story drift ratios and peak floor accelerations are considered as Engineering Demand Parameters (EDPs). A total of 85 potential candidates for selecting suitable NM and IM are investigated, categorized into four groups: (I) acceleration-related, (II) velocity-related, (III) displacement-related, and (IV) hybrid criteria. Accordingly, each set of GMs is normalized and scaled with 85 × 85 different combinations for NM and IM. Then, PSDMs for each investigated building are developed using Incremental Dynamic Analysis (IDA) on 2D nonlinear frame models under the scaled GM sets. The developed PSDMs with different NMs are employed for probabilistic seismic risk analysis and the estimation of seismic demand hazard curves. The optimal criteria for developing PSDMs are identified based on efficiency, practicality, proficiency, sufficiency, and reliable seismic risk assessment. The obtained results reveal the high sensitivity of the optimal NM to the building vibration period, the selected set of GMs, and the EDP under study. For instance, while Peak Ground Acceleration (PGA) is the optimal NM for estimating the risk of the acceleration response parameter, a unique criterion cannot be proposed for the transient and residual drift response parameters that would perform optimally under most conditions; however, I_Np is the best NM for most short-period (2-story) SMFs, and Sv_avg and MVSI are the two superior NMs for most long-period (4- to 20-story) ones. This study provides valuable insights into the impact of the mentioned factors on the selection of the optimal criteria.