Long‐Term Earthquake Response in a Midrise Reinforced Concrete Building: Insights From Hybrid Seismic Monitoring
Abstract
We present a multidecadal analysis of the seismic response of a midrise reinforced concrete (RC) building in Taipei, Taiwan, combining legacy strong‐motion accelerograph (SMA‐1) data with recent recordings from a dense array of low‐cost microelectromechanical systems (MEMS)‐based Quake Structural Integrity System (QSIS) sensors. The dataset spans more than 25 years and includes responses from three major earthquakes, the 1999 Chi‐Chi (Mw 7.6), 1999 Chiayi (Mw 6.4), and 2024 Hualien (Mw 7.4) events, along with numerous moderate earthquakes recorded by QSIS. This hybrid SMA‐1 and QSIS archive provides a rare opportunity to evaluate how a single structure's dynamic properties evolve under repeated strong shaking. We integrate several complementary techniques, including frequency response functions (FRFs), recursive AutoRegressive with eXogenous inputs (RARX) model on system identification, singular spectrum analysis (SSA), and unwrapped phase analysis, to track both transient and interevent changes in modal behavior. The results reveal amplitude‐dependent nonlinear softening during intense shaking, with fundamental frequencies decreasing by up to 15% during strong shaking intervals. Across the historical records, the Library section shows interevent differences in frequency, modal energy distribution, and phase patterns. Because excitation, environmental conditions, and modal correspondence are not fully controlled, these differences are interpreted as modal variability rather than conclusive evidence of localized damage or quantified stiffness loss. This work integrates established methods across a rare hybrid archive combining legacy strong‐motion records with recent continuous MEMS observations. The QSIS records extend the legacy SMA‐1 archive by providing continuous observations of seismic events and ambient vibrations, thereby complementing triggered strong‐motion instrumentation and supporting low‐cost, long‐term monitoring of building dynamic properties. This study shows the value of continuous, multidecadal observation for tracking building response, and it delineates what such records can and cannot establish about localized damage.