Resumo em Inglês:

Abstract Dynamic characterization of cement-based composites is crucial for understanding material behavior. When exposed to highly dynamic loading conditions, the strain-rate dependence of material causes the material response to differ significantly from that under quasi-static loading conditions. In this paper, a numerical investigation on the dynamic tensile behavior of cement-based materials. A gravitational split Hopkinson tension bar was used to characterize the dynamic tensile behavior of cement-based at high strain-rates. The commercial finite element software LS-Dyna is adopted to conduct the computations. The material specifications of cement-based are characterized by the Karagozian & Case (K&C) concrete model that accounts for shear dilation, strain-rate dependence, and strain softening. The model accuracy is verified with available experimental results in the form of strain signals, strain-rates, and tensile strengths. It was found that the results computed with the automatic generation version of K&C are slightly different from the experimental ones. Therefore, to achieve better agreement, the model was extended by calibrating a few parameters of the K&C material formulation. Finally, the simulation predictions were found to represent the experimental results with good agreement.

Resumo em Inglês:

Abstract As a primary load-bearing component, accurately predicting the bearing capacity of concrete-filled rectangular steel tube (CFRST) members is an essential prerequisite for ensuring structural safety. Machine learning methods are employed to model and predict the axial load bearing capacity of CFRST columns. A test database containing 1119 members is established, and the input parameters of the machine learning model are determined using a combination of data preprocessing and correlation analysis. Four machine learning algorithms, namely Lasso, ANN, RF, and XGBoost, are selected to build the prediction models for axial load bearing capacity, and a comparative analysis of their predictive performance is conducted. The feature importance analysis is performed using the SHAP method. The results indicate that the model based on the XGBoost algorithm achieves the highest prediction accuracy. Through comparison with six existing calculation methods in domestic and international codes, the reliability of its predictive performance is verified.