QUANTIFYING UNCERTAINTIES IN THE PERFORMANCE PREDICTION OF EXISTING CONCRETE STRUCTURES USING AN EXTENDED DIRECT STIFFNESS METHOD APPROACH

Abstract

Reinforced concrete structures are widely used in many industrialized countries. Currently, significant parts of the patrimony in these regions are reaching the end of their lifetime. As a result, the through-life analysis of existing structures based on quantitative data is rapidly gaining attention. An essential part of this analysis is the performance prediction of existing structures taking time-dependent degradation effects into account, which can be modelled by extending the Direct Stiffness Method (DSM). However, a probabilistic approach has not been performed yet using this extended DSM model. However, such an analysis is crucial, as there are typically many uncertainties involved in the assessment of existing structures, including uncertainties regarding the design, the damaging processes, etc.

Therefore, in this contribution, the abovementioned uncertainties are accounted for in the structural performance prediction using an extended Direct Stiffness Method approach. It is investigated how the uncertainties on several types of input parameters influence the uncertainty on the outputs, i.e. the deformations, deflections, internal forces etc., of a deteriorated structure. Furthermore, the spatial distribution of these parameters is modelled using random fields. It is then investigated how the correlation lengths of several parameters influence the outputs of the DSM-tool. Finally, a sensitivity analysis is performed such that the parameters having the highest influence can be identified. Knowing these parameters is of importance to determine which measurements will be the most effective in reducing the overall uncertainty on the performance prediction. This plays an important role in the decision making process regarding inspection strategies, leading to a more efficient life-cycle methodology of existing concrete structures in everyday engineering practice.