Glubam roof trusses: Uncertainty quantification and partial safety factors calibration based on Bayesian methods

This study develops a Load and Resistance Factor Design (LRFD) framework for Glubam (Glue Laminated Bamboo) roof trusses under vertical loads. It addresses the challenges posed by bamboo’s inherent property variability by establishing calibrated partial safety factors for Ultimate Limit State (ULS) and Serviceability Limit State (SLS) conditions under vertical loads, which are crucial for roof truss design focusing on vertical load capacity and mid-span deflection. The study utilized a combination of experimental testing and advanced numerical simulations using finite element (FE) models. Monte Carlo simulations on a 3D continuum FE model of truss joints in Abaqus generated a computational test dataset. This synthetic data was then used to train simplified lumped OpenSees FE models through Bayesian inference. A parallel genetic algorithm was employed for initial parameter identification and to establish priors for the Bayesian inference. The Bayesian inference yielded a posterior distribution for the model parameters, enabling further Monte Carlo simulations on a simplified global roof truss model in OpenSees. The reliability of the simplified model was empirically validated against experimental tests on a full-scale truss. These refined safety factors enhance design precision for structural engineers working with Glubam, supporting a range of reliability targets and promoting the sustainable use of bamboo in construction.