Buckling behavior of steel bars with different strength grades subjected to random corrosion: An experimental and multi-method predictive study

The random morphology caused by corrosion significantly damages the compressive buckling performance of steel bars. Differences in chemical composition across strength grades lead to distinct corrosion morphologies, affecting mechanical properties. Buckling behavior under compression governs bearing capacity degradation and failure of structural components. To investigate corrosion characteristics, deterioration mechanisms, and quantitative prediction methods, 54 steel bar specimens with varying corrosion degree, yield strength (HRB630E, HRB500, HRB400), slenderness ratio, and diameter were subjected to dry-wet accelerated corrosion, 3D laser scanning, 3D digital image correlation (3D-DIC), and compression tests. Probability analysis of residual cross-sectional area was conducted using 3D scanning data and a two-component Gaussian mixture model. Lateral deformation development during loading was analyzed via 3D-DIC. Stress-strain curves were obtained from compression tests, and the influence of key parameters on buckling mechanical performance was assessed. Based on experimental data, compressive stress-strain curves were simulated using refined finite element modeling, modified explicit constitutive formulas, and fully connected neural networks. A multi-method prediction approach was established to evaluate the compressive buckling performance of corroded steel bars, which suitable for diverse engineering scenarios and data conditions. The experimental data enrich the relevant database, and the proposed method enables prediction across various scenarios.