Development of a Probabilistic Seismic Demand Model for Pounding Risk Assessment of Buildings

Abstract eng:
The seismic-induced pounding between adjacent buildings is an undesirable event that can cause major damage and even structural collapse for structures with inadequate separation distance. This issue is particularly important in metropolitan areas, where the land space is limited and expensive. In order to minimize the pounding risk, existing design codes provide simplified numerical procedures and analytical rules for estimating the minimum separation distance that is needed to avoid pounding under a target seismic hazard scenario. However, these code procedures are characterized by unknown safety levels and, thus, do not permit to control explicitly the risk of pounding or the consequences of the impact. Previous research by two of the authors developed a reliability-based design methodology for the separation distance that corresponds to a target probability of pounding during the design life of adjacent buildings. This methodology was successfully applied to linear elastic structures. Further studies are required to make reliability-based methodologies applicable in an efficient way to more complex nonlinear building models, which require the use of computationally expensive numerical simulations to accurately predict the structural response. This paper illustrates an efficient probabilistic seismic demand model (PSDM) for pounding risk assessment consistent with modern performance-based design frameworks. A PSDM consists in the analytical representation of the relation between a seismic intensity measure (IM) and an engineering demand parameter (EDP). In this specific problem, the EDP of interest is the peak relative displacement between the adjacent buildings at the most likely impact location. The PSDM can be used to estimate the seismic vulnerability and the mean annual frequency of pounding between adjacent buildings via convolution with the site’s hazard curve. First, an extensive parametric study is performed by considering the case of two adjacent buildings modeled as linear singledegree-of-freedom (SDOF) systems. Different IMs are proposed for the problem at hand, whose choice is motivated mainly by efficiency criteria. The parametric study results are utilized to evaluate the efficiency and sufficiency of the proposed IMs employed in conjunction with a PSDM based on the linear regression of the seismic demand variation with respect to the IM in the log-log space. Successively, the case study of two realistic steel buildings modeled as nonlinear hysteretic multi-degree-of-freedom sheartype systems is considered to evaluate the effectiveness and accuracy of the IMs and PSDM introduced for the buildings described as SDOF systems. A bilinear PSDM is proposed to achieve a better fit of the seismic median demand and dispersion over the entire range of seismic excitation levels. Finally, comparisons are made between the risk estimates obtained by using the linear and bilinear PSDMs and the corresponding estimates obtained via incremental dynamic analysis (IDA) in order to evaluate and compare the accuracy of the proposed regression models. It is found that the use of a bilinear PSDM in conjunction with cloud analysis provides seismic pounding risk estimates that are very close to those obtained through IDA at a small fraction of the computational cost and without scaling the records.

• Export as: BibTeX | MARC | MARCXML | DC | EndNote | NLM | RefWorks
• Add to your basket