Reliability based robustness of timber structures through NDT data updating

This work presents a framework for reliability-based assessment of timber structures / members using data gathered from non-destructive test results. These results are used for modeling an update of the mechanical characteristics of timber, using Bayesian methods. These methods are suitable to be used for parameter estimation and also allow updating model uncertainties. From the updated model, decisions upon the life-cycle reliability of existing structures may be taken and maintenance or strengthening actions may be considered. In this work, results gathered from ultrasound testing, Resistograph® and Pilodyn® conducted on chestnut wood specimens were used, as well as correlations between those results and compression strength parallel to the grain tests' results. The resistant characteristics are also updated assuming deterioration models applied to specific key elements of the structure, thus, being possible to evaluate reliability based in time dependent factors, as well to categorize that structure in terms of robustness. Timber is a rather complex construction material, due to its anisotropic behavior. Moreover, its properties also vary on space and time. For instance, the material properties of a timber element vary both in different parts of the same cross-section as well as along the element itself. Differences are also visible when comparing different timber elements even if they are from the same species. Therefore, timber structures are better analyzed using probabilistic models rather than deterministically when considering a structural safety evaluation. Reliability methods are prone to be used in this assessment as they allow describing the properties of timber elements by random variables. The probability density functions of timber's mechanical properties are defined in various codes and guidelines and may be updated with the results of mechanical tests. A grading methodology based on visual inspection associated with NDT results is also a suitable source of information for an updating data model. When dealing with existing structures, data updating may be regarded as an important tool in the assessment of its reliability