Synthesis of a vector-valued intensity measure for improved prediction of seismic demands in Inter-Story-Isolated (ISI) buildings subjected to near fault ground motions

Abstract

An optimal Intensity Measure (IM) is key to accurate prediction of the Engineering Demand Parameters (EDPs) in structures subjected to seismic excitations. This study synthesizes a structure-specific, vector-valued IM for assessing the EDPs in Inter-Story-Isolated (ISI) buildings under near-fault (pulse and non-pulse type) ground motions. The synthesis is inspired by the comparable modal masses in the fundamental and another higher (generally second) mode in ISI buildings. The synthesis accommodates the dual mode participation by employing the respective pseudo-spectral accelerations Spa and the spectral shape (‘epsilon’ε) parameters in the proposed IM as components. A suite of pulse and non-pulse type motions are employed for the synthesis and subsequently fitting an IM vs. EDP regression model. Thereafter, the “Efficiency” and “Sufficiency” analysis are conducted for validation. The EDPs are obtained by non-linear dynamic analyses on alternative building frames with High Damping Rubber Bearing (HDRB) based ISI systems. A Principal Component Regression (PCR) is employed for accommodating the multi-collinearity in the IM. The choice of the Principal Components (PCs) adopts a leave-one-out cross-validation approach. The proposed IM is shown to offer higher Efficiency and Sufficiency compared with the existing non-structure-specific (Peak Ground Acceleration, Velocity, Displacement, Arias Intensity, Cumulative Absolute Velocity, Characteristic Intensity, Specific Energy Density etc.) and structure-specific (Acceleration/Velocity Spectrum Intensity, Housner Intensity, Pseudo-spectral acceleration at fundamental time period etc.) IMs. Higher Efficiency is also noted under the pulse type motions comparing the non-pulse ones. The IM is more efficient in predicting the Inter-Story Drift Ratios (IDRs), Total Drift Ratios (TDRs) than the Peak Floor Accelerations (PFA) and Bearing Displacements (PBD). The proposed IM is shown to be unbiased and robust.