Epistemic Uncertainties in Component Fragility Functions

Abstract

This paper is concerned with the inclusion of epistemic uncertainties in component fragility functions used in performance-based earthquake engineering. Conventionally fragility functions, defining the probability of incurring at least a specified level of damage for a given level of seismic demand, are defined by a mean and standard deviation and assumed to have a lognormal distribution. However, there exist many uncertainties in the development of such fragility functions. The sources of epistemic uncertainty in fragility functions, their consideration, combination, and propagation are presented and discussed. Two empirical fragility functions presented in literature are used to illustrate the epistemic uncertainty in the fragility function parameters due to the finite size of the datasets. These examples and the associated discussions illustrate that the magnitude of epistemic uncertainties are significant and there are clear benefits of the consideration of epistemic uncertainties pertaining to the documentation, quality assurance, implementation, and updating of fragility functions. Epistemic uncertainties should therefore always be addressed in future fragility functions developed for use in seismic performance assessment.