Approaches to estimate global safety factors for reliability assessment of RC structures using non-linear numerical analyses

The study is focused on the comparison and discussion of different approaches within the use of the global resistance method (GRM) for safety assessment of reinforced concrete (RC) systems using non-linear numerical analyses (NLNAs). With this purpose, a benchmark dataset, comprising 56 experimental results obtained from tests on 40 RC columns with variable slenderness and 16 non-slender RC elements including walls, deep beams and shear walls, is considered. The NLN models for all the 56 members adopt solution strategies able to optimize the agreement between numerical predictions and experimental outcomes. Then, probabilistic hypotheses have been defined regarding both aleatory (i.e., materials and geometry) and epistemic uncertainties (i.e., model) associated with all the 56 RC members. These assumptions form the basis for developing a comprehensive set of probabilistic analyses of the global structural resistance for each RC member. The results of these probabilistic analyses offer valuable insights into the impact of the different sources of uncertainties on the global structural response. In detail, three distinct approaches for estimating the global safety factors within the GRM are outlined and compared. The purpose is to address the effectiveness of the different approaches for the reliability evaluation of RC members within the GRM together with the relevance of both aleatory and epistemic uncertainties. Ultimately, recommendations are provided regarding the adoption of the GRM in the upcoming generation of design codes.