To accurately simulate engineering service conditions and evaluate the impact of corrosive environments on crack propagation in a steel box girder structure, this study conducted material- and structural-scale corrosion-environment and mechanical performance tests.

Material-level compact tension (CT) specimens and a steel box girder model were fabricated, with initial cracks introduced via artificial defects and fatigue loading. Corrosion tests were then conducted separately: CT specimens were exposed to a coastal field environment for three months to simulate marine atmospheric corrosion, whereas the box girder model underwent accelerated corrosion in a laboratory chamber for one month. Fracture toughness tests were performed on CT specimens before and after corrosion, and crack propagation tests were conducted on the box girder under cyclic fatigue loading.

Key findings include the following: (1) At the material scale, fracture toughness of the uncorroded welded CT specimens was $237.4\mathrm{M}\mathrm{P}\mathrm{a}\cdot {\mathrm{m}}^{\frac{1}{2}}$, which increased to $380.0\mathrm{M}\mathrm{P}\mathrm{a}\cdot {\mathrm{m}}^{\frac{1}{2}}$ (~60% increase) after three months of coastal exposure. This is attributed to corrosion-induced blunting of the sharp crack tip, which raised the energy required for crack re-initiation. (2) At the structural scale, after accelerated corrosion, only two of three pre-existing cracks on the box girder (at the #4 support point) propagated under cyclic loading, with a significant delay in crack initiation that further supports the crack-tip-blunting effect. Propagation paths traversed corrosion pits, where the crack rate slowed or briefly halted until a new sharp tip formed on the opposite side, rendering the paths more complex and discontinuous. (3) The two propagating cracks at the #4 support point grew essentially perpendicular to the primary load-bearing direction, predominantly as Mode Ⅰ (opening mode); the fastest crack (#4-V-b) propagated at 2.06×10^–3 mm/cycle along a largely linear a–N curve, indicating stable fatigue crack growth.

This study reveals dual effects of corrosive environments on crack propagation in steel box girder structures. Corrosion enhances fracture toughness by blunting the crack tip, inhibiting crack initiation. However, pre-existing corrosion pits can act as preferential propagation paths and cause growth fluctuations, increasing the risk of local failure.

A grid-shaped steel box girder structure serves as the core load-bearing component of ground support systems in a specific engineering project. It bears static support, fuel filling, and gas flow impact loads during service. Owing to the extensive number of structural welds and their concealed locations, weld cracks are inevitable during manufacturing, assembly, and usage. During actual inspection and maintenance, cracks are detected in various weld locations within the structural load-bearing core area. This core area comprises a two-way grid pattern of girders with complex stress conditions, making it difficult to assess crack propagation under high-intensity service conditions. Existing studies primarily employ an analytic hierarchy process or a finite element method to analyze the impact of cracks on structural reliability and service life. However, these studies did not investigate the propagation behavior of cracks under repeated operational loads, and their findings were not sufficiently validated by experimental data.

To assess the crack propagation behavior at critical weld joints under high-intensity service demands and ensure structural service safety, an experimental research approach was adopted. A 1:10 scaled-down model of a load-bearing core structure was designed based on similitude relationships. Six artificial defects were prefabricated at different weld locations within the load-bearing core area by introducing artificially created flaws, with two and four located in the compression and tension zones, respectively. Then, fatigue loading was applied by progressively increasing the load magnitude, resulting in the initiation of four initial cracks. Subsequently, crack propagation tests were conducted on the scaled-down model with these cracks under equivalent service loads.

Test results indicate the following. (1) When the fatigue peak load was increased to three times the equivalent service load, fatigue cracks initiated from all prefabricated defects in the tension zone, whereas no cracks were observed in the compression zone throughout the process. (2) With twice the equivalent service fatigue loading, the cracks in base and vertical plates at support points #2 and #4 of the scaled-down model exhibited similar evolution characteristics. The crack growth rates in the lower sections of the vertical plates were consistently higher than those in the base plates, which aligns with the strain patterns measured during the prefabricated crack tests. (3) The propagation direction of all cracks was essentially perpendicular to the beam's longitudinal direction (i.e., the tensile stress direction), indicating that the cracks were primarily in the opening mode. (4) The a–N curves of all propagating cracks demonstrated relatively stable linear characteristics, with the maximum growth rate recorded at 0.945 mm/10³ cycles.

We proposed and validated the feasibility of using artificially prefabricated defects combined with stepwise increased fatigue loading to generate initial cracks in structural testing. The test results elucidated the propagation behavior of cracked structures at critical weld locations under equivalent service loads. This provides an experimental data reference for technicians to assess the service condition of the load-bearing core structure and forms a basis for optimizing structural inspection and maintenance strategies.