Machine learning-aided peak and residual displacement-based design method for enhancing seismic performance of steel moment-resisting frames by installing self-centering braces

Shuling Hu, Songye Zhu, M. Shahria Alam, Wei Wang

Research output: Journal article publicationJournal articleAcademic researchpeer-review

2 Citations (Scopus)

Abstract

Conventional steel moment-resisting frames (SMRFs) absorb seismic energy through steel yielding behavior, leading to significant residual displacement. Although steel yielding behavior can ensure the seismic safety of SMRFs under strong earthquakes, excessive residual displacement may lead to post-earthquake demolition decisions, causing a large amount of economic loss. This paper aims to develop a peak and residual displacement-based design (PRDBD) method for controlling the peak and residual inter-story drift responses of SMRFs by installing self-centering braces. The peak and residual displacements are both set as the design targets in the proposed PRDBD method. To this end, the machine learning prediction models of inelastic and residual displacement ratios were first developed based on the median responses of single-degree-of-freedom systems under earthquakes. The detailed design steps of the proposed PRDBD method were subsequently introduced. The three- and nine-story demonstration buildings were retrofitted by using the PRDBD method with two different design targets. Static and dynamic analyses were conducted to validate the effectiveness of the proposed PRDBD method. The static analysis results indicated that the self-centering braces could efficiently enhance the SMRF's stiffness and strength. The retrofitted SMRFs showed no strength deterioration, whereas the original SMRFs showed obvious strength deterioration at the roof drifts of 3.2% and 2.5% in the three- and nine-story buildings, respectively. The dynamic analysis results confirm that the self-centering braces can efficiently reduce the peak and residual inter-story drift responses of the existing SMRFs and the retrofitted SMRFs can achieve the peak and residual inter-story performance objectives under the considered seismic intensity. Moreover, the retrofitted SMRFs can be fully recoverable after maximum considered earthquakes by controlling the maximum residual inter-story drift lower than 0.2%.

Original languageEnglish
Article number114935
JournalEngineering Structures
Volume271
DOIs
Publication statusPublished - 15 Nov 2022

Keywords

  • Machine learning
  • Moment-resisting frames
  • Peak and residual displacement
  • Post-earthquake repairability
  • Residual displacement-based design method
  • Self-centering brace

ASJC Scopus subject areas

  • Civil and Structural Engineering

Fingerprint

Dive into the research topics of 'Machine learning-aided peak and residual displacement-based design method for enhancing seismic performance of steel moment-resisting frames by installing self-centering braces'. Together they form a unique fingerprint.

Cite this