Probabilistic seismic demand model for pounding risk assessment

Enrico Tubaldi, Fabio Freddi, Michele Barbato

Research output: Contribution to journalArticle

18 Citations (Scopus)

Abstract

Earthquake-induced pounding of adjacent structures can cause severe structural damage, and advanced probabilistic approaches are needed to obtain a reliable estimate of the risk of impact. This study aims to develop an efficient and accurate probabilistic seismic demand model (PSDM) for pounding risk assessment between adjacent buildings, which is suitable for use within modern performance-based engineering frameworks. In developing a PSDM, different choices can be made regarding the intensity measures (IMs) to be used, the record selection, the analysis technique applied for estimating the system response at increasing IM levels, and the model to be employed for describing the response statistics given the IM. In the present paper, some of these choices are analyzed and evaluated first by performing an extensive parametric study for the adjacent buildings modeled as linear single-degree-of-freedom systems, and successively by considering more complex nonlinear multi-degree-of-freedom building models. An efficient and accurate PSDM is defined using advanced intensity measures and a bilinear regression model for the response samples obtained by cloud analysis. The results of the study demonstrate that the proposed PSDM allows accurate estimates of the risk of pounding to be obtained while limiting the number of simulations required.

Original languageEnglish
Pages (from-to)1743-1758
Number of pages16
JournalEarthquake Engineering and Structural Dynamics
Volume45
Issue number11
Early online date14 Mar 2016
DOIs
Publication statusPublished - 30 Sep 2016

Keywords

  • intensity measure
  • performance-based design
  • pounding
  • probabilistic seismic demand model

Fingerprint Dive into the research topics of 'Probabilistic seismic demand model for pounding risk assessment'. Together they form a unique fingerprint.

  • Cite this