Bayesian model selection for the nonlinear hysteretic model of CLT connections

Abstract

Connections are important components to cross-laminated timber (CLT) structures. Although there are some hysteretic models available for CLT connections, the main challenge is how to select a predictable and robust model. In this paper, three popular hysteretic models, including the Bilinear model, SAWS model, and Pinching4 model, are briefly revisited. These hysteretic models exhibit their advantages and disadvantages, making them difficult to select the most suitable model for the simulation of hysteretic response. Based on the test data, the Bayesian approach is used to determine the most plausible model from five candidate models. The identified results show that the symmetric Pinching4 model has the highest evidence in shear loading, while in the tension loading the SAWS model possesses the highest value. For the evaluation of the identified results, both the idealized backbone curve and Park-Ang damage index are compared, showing a good agreement between the test and model results. These mechanical characteristics are crucial to the design and rehabilitation of timber buildings.