Determinação da rigidez da ligação aparafusada do compósito madeira-concreto sob flexão: um estudo paramétrico

Abstract

Timber-concrete composite (TCC) structures have been extensively investigated due to the advantages arising from its use. The combination of these two materials provides greater bending stiffness (concrete) and reduced self-weight (timber). A key property in determining the bending stiffness of this type of composite is the connection stiffness, which can be evaluated at the serviceability limit state (SLS) [Kser] and ultimate limit state (ULS) [Ku]. However, due to the complexity of its determination, few accurate estimation models are available. The estimation proposed by Eurocode 5 (2004) for screwed connections does not consider important parameters such as the timber’s elastic modulus and the screw embedment length in the timber. In order to enable a more reliable estimation of connection stiffness in beams with varying connection configurations, screw spacing, and span lengths, this study aimed to propose regression models that are more accurate than the Eurocode 5 (2004) predictions. To address this gap, a parametric numerical study was conducted using the Abaqus software, employing beam models instead of push-out specimens to reflect real structural behavior. The stiffness values were estimated as functions of timber elastic modulus (7,635 MPa and 13,800 MPa), screw diameter (8 mm and 16 mm), screw embedment length (80 mm and 160 mm), timber beam height (175 mm and 300 mm), screw spacing (250 mm and 500 mm), and beam span (2.8 m and 6.8 m), all varied at two levels, resulting in 64 simulations. The goal was to increase the safety and efficiency of stiffness prediction in this structural typology. Results indicated that connection stiffness correlates better with the beam failure load in bending than with the connection's capacity itself. The most influential parameter was the timber’s elastic modulus, followed by screw diameter, embedment length, beam span, span-to-height ratio, and span-to-screw spacing ratio. For Ku,60%, the beam height also proved to be significant. However, screw spacing (for both Kser and Ku,60%) and beam height (for Kser) were not individually significant. The developed regression equations for Kser and Ku,60% yielded R-squared values of 0.842 and 0.91 and Pearson correlation coefficients of 0.92 and 0.954, respectively. ANOVA confirmed the statistical validity of these models. Compared to estimates from Eurocode 5 (2004) and Mirdad & Chui (2020a), the numerical simulations produced significantly more conservative stiffness values, with mean relative errors of 135.79% and 149.13% for Kser, and 131.92% and 147.29% for Ku,60%, respectively. Thus, it is recommended to use these regressions when the values range within the investigated parameters. For push-out tests, however, these equations are not applicable due to the significance of the span length and timber beam height parameters. The application of this equation is highly valuable for the design of TCC structures, as it enhances both the safety and accuracy of the stiffness prediction, as validated by the results.