Investigation of Intraclass Correlation of Seismic Capacity for RC Bridge Piers Based on Hierarchical Model

Abstract

The framework for performance-based seismic design and assessment of bridges grapples with the pressing challenge of precisely estimating the seismic capacities correlation amongst bridge piers. To address this issue, a Bayesian semiparametric hierarchical model is proposed in this study to characterize such correlation. This model was applied to a database populated with multiple-column tests conducted by distinct research cohorts. Given that each group performed tests under analogous conditions, the model could be operationalized on the database to derive intraclass correlation coefficients that reflect the statistical correlation among piers tested by individual research groups. It is postulated that these coefficients also typify the capacity correlation across multiple bridge piers within a bridge system. After a thorough examination of the data set, particularly focusing on the nature of missing values, an appropriate multiple imputation technique was selected to address the missing data, aiming to approximate the most accurate intraclass correlation. The analysis revealed that, among the various damage limit states, the correlation coefficients spanned between 0.386 and 0.883. This suggests that the seismic capacity of bridge piers is influenced not only by design parameters but also by the experimental group. Utilizing independent or fully correlated relationships between capacities, rather than true correlation coefficients, can lead to significant deviations in the fragility curve, which increase with the number of bridge piers. Such observations accentuate the necessity of integrating authentic pier capacity correlations in the bridge system fragility analysis.