Portal de Periódicos da CAPES

Evaluating Different Approaches for the Hazard-Consistent Assessment of the Seismic Performance of Dams

2022; Seismological Society of America; Volume: 112; Issue: 3 Linguagem: Inglês

10.1785/0120210181

1943-3573

Jorge Macedo, Venkataraman Ramesh, Chenying Liu, Albert Kottke,

Landslides and related hazards

ABSTRACT The current state of practice in the seismic assessment of dams relies on pseudoprobabilistic approaches in which the estimation of ground-motion intensity measures (IMs) is decoupled from the estimation of seismically induced displacements (D). In contrast, in a performance-based approach, which is gaining more popularity in practice, the entire IM hazard curve is used to provide displacement hazard curves (DHCs). Thus, engineers can directly estimate the displacement associated with a selected design hazard level, which is more consistent with performance-based design concepts. This study evaluates different approaches to estimate D in performance-based assessments that use D predictive equations, analytical D models, and advanced numerical modeling. To enable a comparison of different approaches under a hazard-consistent D assessment, we propose using the conditional scenario spectra (CSS) framework to select ground motions that directly reproduce the IM hazard. We perform the evaluations by comparing DHCs for a rockfill dam using (1) the convolution of D predictive models and the IM hazard, (2) the CSS approach combined with dynamic analyses performed with analytical models (i.e., stick-slip and transfer function models), and (3) the CSS approach combined with advance numerical modeling. The results show that the DHCs estimated from advanced numerical modeling are more conservative, which is associated with the inherent assumptions in D predictive equations or analytical D models formulated based on “Newmark-type” calculations. Finally, we discuss the differences between the different approaches to estimate D and provide insights in the context of pseudoprobabilistic-based estimates.