Structural Reliability of GFRP-RC Slab Column Connections Based on a Data-Driven Robust Algorithm: Punching Shear Limit State

Abstract

This study focuses on the assessment of the structural reliability of the punching shear limit state in five distinct configurations of glass fiber–reinforced polymers (GFRP)–RC slab-column connections, illustrating their diverse applications through a robust data-driven method. The Nelder–Mead simplex minimization algorithm is used to fit an efficient predictive model for the ultimate punching shear capacity. In the context of the reliability analysis, considered variables include the effective depth, GFRP reinforcement area, concrete compressive strength, GFRP modulus, dead loads, live load, and errors in the mechanical model. The data set, forming the basis for the data-driven model, is derived from numerical simulations conducted using an FEM based on a concrete damaged plasticity model, which was first validated with sixteen experimental observations from the literature. Structural reliability is also assessed using the first-order reliability method in accordance with commonly used design provisions for GFRP-reinforced concrete structures in North America. The results indicate that the provisions based on US standards generally lead to reliability indices above 3.0, while those following Canadian standards result in indices below this threshold in a significant proportion (40%) of cases.