An Application of the Response Surface Metamodel in Building Seismic Fragility Estimation

Abstract eng:
Seismic fragility plays an essential role in the estimation of potential earthquake losses to a building. Its probabilistic framework allows the damage assessment to take into account uncertainties in earthquake hazard as well as in building properties. However, the relationship between building damage, earthquake loads, and structural characteristics is complex and can hardly be represented by a closed-form expression. Therefore, propagation of the uncertainties to formulate damage probability distributions has to rely upon numerical techniques such as Monte Carlo simulation. This approach requires a large number of trials in order to obtain a reliable probabilistic distribution of the damage. Computational demand becomes impractical when thousands of dynamic nonlinear time-history analyses of building models must be performed for that purpose. This work explores an alternative approach for carrying out the intensive structural simulations. Response Surface Methodology in conjunction with Monte Carlo simulation achieves an efficient compromise in building fragility computation. In particular, a response surface is systematically developed to predict the structural response based upon a few nonlinear finite element dynamic analyses. Computational costs imposed by the Monte Carlo simulation are significantly reduced because the propagation of uncertainties can now be performed on a polynomial response surface function, rather than on a complex dynamic finite element model. An example application of the methodology is presented for the seismic fragility assessment of a 5-story steel moment resisting frame (SMRF) building. The building is assumed to be located in Memphis, Tennessee, USA, and is designed in accordance with the seismic provisions commonly used in the region. The computed fragility curves of the building reveal potential damage to the building from future earthquakes, and provide the basis for prioritizing mitigation actions.

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