Analysis of Glubam Joint Behavior and Truss Model Updating Based on Optimization Algorithms

Bamboo, as a naturally fast-growing renewable resource, is abundant in China and supported by a wellestablished industrial foundation, making it a crucial material for promoting green building development. Modern engineered bamboo structures are typically fabricated using industrialized processes with engineered bamboo-based panels as raw materials, which reduce environmental impact while enabling standardization and prefabrication of building components. These structures can effectively meet performance requirements in terms of safety, economy, and comfort. Glue-Laminated Bamboo (glubam), a representative type of engineered bamboo, features a high strength-to-weight ratio, low adhesive content, and stable physical and mechanical properties. As an innovative and sustainable construction material, glubam has shown great potential for application in modern structural engineering. Compared to traditional building materials such as timber and steel, glubam offers significant advantages in strength-to-weight performance, renewability, and environmental friendliness. Despite its excellent material properties, research on the hysteretic behavior of glubam structural joints and the seismic performance of glubam composite truss systems remains insufficient. Comprehensive design theories and reliable numerical modeling approaches are still lacking. Moreover, integrating artificial intelligence (AI) optimization algorithms into the seismic performance analysis of glubam structures represents a promising yet underexplored research direction. In response to these challenges, this study systematically investigates the mechanical behavior of glubam joints and their corresponding truss assemblies under cyclic loading through a combination of experimental testing, numerical simulation, and AI-based optimization methods. The main research contents and findings are summarized as follows: First, according to relevant testing standards, two types of glubam joints and their corresponding planar truss and roof truss structures were subjected to quasi-static cyclic loading tests to evaluate their hysteretic behavior and failure modes. Based on the experimental observations, both high-fidelity three-dimensional finite element models and simplified low-fidelity hysteresis models were developed to capture the nonlinear mechanical responses of the two joint types. For the simplified models, two parametric hysteresis constitutive models were proposed to reproduce critical features observed under cyclic loading, such as pinching effects, asymmetry, and strength degradation. Three representative AI optimization algorithms—Genetic Algorithm (GA), Bayesian Inference (BI), and Neural Network (NN)—were introduced to perform parameter identification and model calibration, significantly improving the accuracy and generalizability of the models. Finally, using the calibrated hysteresis models, a macro-scale numerical model of the glubam truss structure was constructed by combining the joint models with beam-column elements. Structural-level model updating was then performed using AI algorithms, and the optimized model was used to analyze the structural response of glubam trusses under cyclic loading. The detailed research tasks and contributions of this study are summarized as follows: This study first conducted axial monotonic and cyclic loading tests on two types of glubam joint connections with distinct configurations: the steel-insert glubam joint and the steel-plate clamped glubam joint. The fasteners used in these joints were designed with varying geometric dimensions. Through systematic experimentation, the mechanical behavior of both joint types under cyclic loading was comprehensively analyzed, including characteristics of their hysteresis curves, stiffness degradation patterns, energy dissipation capacity, and typical failure modes. The test results demonstrated that both types of glubam joints exhibited favorable hysteretic behavior and excellent energy dissipation performance. Their failure processes were primarily ductile in nature, indicating promising seismic resistance potential. In addition, the influence of geometric parameters of the fasteners on the mechanical performance of the joints was further investigated. It was found that these parameters significantly affect the joints' load-bearing capacity, initial stiffness, and energy dissipation efficiency. Building upon the joint performance investigation, planar truss and roof truss systems were designed using the two connection types (steel-insert and steel-plate clamped) and subjected to quasi-static cyclic loading tests. The study systematically evaluated the global hysteretic performance, energy dissipation capacity, and seismic behavior of the two types of truss systems under cyclic loads. Test results indicated that glubam truss systems exhibited good deformation capacity and high energy dissipation efficiency, meeting the basic requirements of seismic design. In the numerical simulation component of this study, high-fidelity three-dimensional finite element (FE) models were developed for both types of glubam joint configurations. A novel modeling approach was proposed by coupling the "element deletion method" with the Hill yield criterion, enabling simultaneous characterization of the orthotropic mechanical properties and crack propagation behavior of glubam. These constitutive mechanisms were implemented via a user-defined material subroutine (UMAT) in Abaqus and successfully applied to the high-fidelity 3D finite element model of the steel-insert glubam joint. The simulated load–displacement curves closely matched the experimental results, validating the model’s accuracy and reliability in capturing the nonlinear hysteretic response of the joints. To enable more efficient simulation at the structural (macro) scale, two sets of low-fidelity simplified hysteretic models were further developed for the aforementioned joint configurations. These models innovatively combined multiple types of spring elements—each representing distinct mechanical behaviors such as ideal elastoplasticity, pinching, and gap characteristics—through series and parallel arrangements. This approach significantly reduced computational cost in structural analysis and facilitated subsequent parameter identification and model updating. The simplified hysteretic models systematically incorporated key nonlinear features observed during cyclic loading, including stiffness degradation, unloading stiffness recovery, strength deterioration, and energy dissipation. Comparison with experimental data demonstrated that the simulated load–displacement curves agreed closely with test results, confirming the proposed hysteretic models’ accuracy and engineering applicability. To ensure that the numerical hysteresis models accurately capture the actual cyclic behavior of glubam joints, it is essential to identify and calibrate multiple key model parameters. However, due to the high dimensionality of these parameter sets, manual tuning is inefficient and often fails to yield stable and reliable results. To address this issue, this study incorporates three mainstream artificial intelligence (AI) optimization algorithms—Genetic Algorithm (GA), Bayesian Inference (BI), and Neural Networks (NN)— into the finite element (FE) simulation workflow, thereby establishing an intelligent parameter identification framework. By conducting a comparative analysis of the three algorithms in terms of accuracy, convergence speed, and robustness, the most suitable optimization strategy was selected. The resulting calibrated numerical hysteresis models exhibit both high accuracy and strong stability, and are capable of faithfully reproducing the cyclic behavior of the joints under repeated loading, providing a reliable basis for subsequent structural-level modeling. Building on this foundation, the calibrated simplified joint models were embedded into macro-scale glubam truss models, enabling simulation of the coupled behavior between the joints and the overall structural system. To further improve the predictive accuracy of the structural mo dels under realistic loading conditions, an advanced model updating procedure was implemented using optimization techniques. The updated models were validated through systematic comparisons between numerical simulations and experimental results, confirming the accuracy and practical value of the proposed model updating methodology. The research findings demonstrate that glubam joints and their corresponding truss systems exhibit excellent energy dissipation capacity and mechanical stability under cyclic loading, highlighting their significant potential in seismic design and sustainable construction. This dissertation not only systematically uncovers the hysteresis evolution characteristics of glubam joints and truss systems but also proposes a comprehensive modeling and optimization framework—from constitutive joint modeling and parameter identification to structural-level model integration and updating. These contributions lay a solid theoretical and technical foundation for the engineering application of glubam-based structural systems in seismic design