Characteristic values of spatially varying material properties in existing structures

The verification of existing civil engineering structures typically requires selection of characteristic values of the material properties. When measurements are available, these characteristic values can be obtained by application of Bayesian analysis employing a conjugate prior approach for the posterior distribution. In EN 1990, this procedure is facilitated using standard cases for which tabulated values are available. However, this approach does not account for the spatial variability of the material parameters and the potential correlation of measurements performed at nearby locations, which can lead to a non-conservative estimate of the material strength and thus the structural reliability. We propose a hierarchical Bayesian approach that extends the standard approach used in EN 1990 to account for spatial effects; it can be implemented through a set of algebraic equations and only requires the measurement locations and the prior correlation model as additional input information. The method is ideally suited for application to hydraulic structures whose massive size leads to significant spatial variability.