Robust seismic fragility curves for concrete gravity dams incorporating aleatory, epistemic, and fuzzy uncertainties

Abstract

Concrete gravity dams are critical infrastructure assets, and their failure in seismically active regions poses significant risks. The primary problem addressed in this research is the need for a reliable assessment of the seismic fragility of these dams to ensure safety in such areas. The research aims to develop a comprehensive methodology for seismic fragility analysis that accounts for various uncertainties affecting dam performance, including uncertainties in material properties and seismic characteristics and the fuzziness of damage state thresholds through the Latin Hypercube Sampling method. The methodology involves a three-stage approach: first, a finite element model is created using Abaqus software, incorporating uncertainties in material properties and earthquake characteristics using the Latin Hypercube Sampling method (LHS). Second, incremental dynamic analysis (IDA) defines damage states based on concrete cracking severity and overall stability, utilizing four performance categories of limit states (minor, moderate, extensive, and heavy). Finally, a mathematical model is introduced to characterize the ambiguity in damage state thresholds, recognizing the gradual nature of damage evolution. The primary findings demonstrate that fuzziness substantially influences the probability of exceedance across all damage states, with varying effects on different damage indices because of their differing frequency distributions. This research enhances the reliability and accuracy of seismic fragility assessments, contributing to more robust design guidelines and effective risk mitigation strategies for concrete gravity dams.