Evaluating Peak Structural Demands in Steel Frames Using Arias Intensity-based Record Truncation

Aims: To evaluate the computational efficiency and accuracy of Arias Intensity (AI)-based seismic record truncation for optimizing Nonlinear Time-History Analysis (NLTHA), specifically assessing if reduced-duration records accurately predict peak deformations and internal forces compared to full-duration benchmarks. Study Design: 3D Finite Element (FE) steel building models underwent NLTHA to capture material inelasticity. Dynamic equations were solved via step-by-step direct integration using the Hilber-Hughes-Taylor (HHT) alpha method. Place and Duration of Study: Faculty of Engineering (postgraduate division), Universidad Autónoma de Querétaro (UAQ), June 2024 - November 2025. Methodology: Six 3D steel moment-resisting frames (4–12 stories, regular/irregular, 5x3 bays, 6.00 m spans, 3.00 m inter-story heights) were subjected to 30 crustal earthquakes analyzed via NLTHA. Records were truncated using accumulative AI thresholds to isolate strong-motion phases: 5-95% (D5-95), 5-85% (D5-85), and 5-75% (D5-75). Accuracy was assessed against full-duration benchmarks. Results: D5-95 optimally reduced computation time by 35% while maintaining near-perfect accuracy (<2% error) across all Engineering Demand Parameters (EDPs). The D5-85 window saved 55% runtime while maintaining >95% accuracy for stiffer structures, but lost precision for flexible frames. The aggressive D5-75 maximized efficiency (63.1% runtime and 70% storage reduction) while retaining >94% precision for strength demands. However, for flexible structures, D5-75 severely underestimated kinematic parameters, suppressing peak roof displacements and rotations by up to 20.9% and 11% (P < .01), as well as critical inter-story drifts. Conclusion: Universal truncation thresholds are unsafe; accuracy depends on the fundamental period and the evaluated EDP. D5-75 is recommended strictly for assessing strength/force demands. D5-85 suits rigid structures where slight drift underestimation is tolerable. For rigorous Performance-Based Earthquake Engineering (PBEE), the conservative D5-95 is strictly required for flexible systems to capture the long-period cycles driving maximum resonant deformations.