Assessment of Existing Structures using Probabilistic Analysis Methods in Combination with Nondestructive Testing Methods

Abstract

Increasing daily traffic volume with rising numbers of axles and allowable load per axle demands appropriate load models for structural analysis. Such models have been developed and incorporated according to latest standards. By adopting the new load models, the load-bearing capacity of many in-service structures can be increased. Therefore, a realistic assessment of the structure especially for the existing bridges is needed. Nondestructive testing (NDT) is used to give realistic import values for structural analysis. This article outlines a procedure for combining the results of NDT in civil engineering (NDT-CE) and the stochastic models used for probabilistic analysis to quantify the reliability of a structure given by the reliability index. The idea of implementing the measurement results as statistic variables in stochastic models is not new. This article presents a novel flexible approach to quantify the uncertainty of measurement according to the Guide to the Expression of Uncertainty in Measurement (GUM). The GUM approach allows one to easily update input quantities and quantify their effect on the total uncertainty without repeating the whole process. The reliability of an NDT method can be quantified in advance by conducting carefully designed probability of detection (POD) studies. A case study of an existing bridge is presented here, where it will be shown by means of a sensitivity analysis how detailed knowledge—gained from reliable NDT-CE measurements—of the dead loads or the exact position of tendon ducts can influence the reliability index. Combining reliable NDT-CE measurements and probabilistic analysis allows assessment of the “true” as-built structure with the purpose of proving its stability based on detailed structural and statistical knowledge.