​Estimativa da rigidez axial e capacidade de carga de pilares de madeira parcialmente reforçados por materiais compósitos laminados

Abstract

Wood is a material widely used in construction since antiquity, commonly found in historical buildings around the planet. As wood is a biodegradable material, historical buildings that use it as a structural typology, without the knowledge of current techniques to ensure the durability of this element, require greater care regarding their protection. Several studies have been observed addressing techniques for strengthening and recovering timber structures, with the use of fiber-reinforced polymers (FRP) being a solution of growing interest in the scientific community. A gap was observed in published studies regarding the application of localized FRP reinforcement in timber columns with longitudinal openings. Thus, obtaining an equation to determine the ultimate load and axial stiffness is necessary to make the strengthening method viable. The present study proposed an equation capable of describing the ultimate load of timber columns with longitudinal openings and partial FRP reinforcement, considering multiplying factors related to instability and the confinement effect generated by the reinforcement. Furthermore, a factor for adjusting the axial stiffness was also proposed. For this, a broad parametric study was carried out through finite element method simulations, comprising 1392 models. The following parameters were considered: wood species, column diameter and length, opening proportion, fiber type, number of layers, and FRP spacing. For the model prediction, a 'white-box' evolutionary algorithm was adopted: symbolic regression. This solution was chosen due to its high capacity for data evaluation and physical interpretation, allowing for the topological dynamism of the generated model. The results of the parametric study revealed that the effectiveness of the reinforcement is not universal, being strictly dictated by the dominant collapse mode. In slender members, the application of low FRP rates optimized the structural response by generating local confinement and mitigating early transverse deformations. In contrast, in robust columns, excessive reinforcement demonstrated marginal efficiency due to premature failures induced by stress concentrations at the wood-composite interface. The predictive models generated by symbolic regression show significant statistical robustness (R2 = 0.92 for load capacity) and surpass pre-existing empirical models in the literature by introducing a slightly conservative variance, ensuring that eventual deviations act in favor of structural safety.