Probabilistic constitutive model for corroded structural steel with stochastic pits under monotonic tension

This study investigates the probabilistic constitutive model of corroded structural steel based on the stochastic characteristics of corrosion pits extracted from real corroded surfaces. Accelerated corrosion tests are conducted on plate specimens in general atmosphere and offshore atmosphere, and pit characteristics like depth, aspect ratio, shape, and density, are studied through surface morphology scanning. The pit depths and aspect ratios follow the logarithmic normal distribution, with greater variability of pit dimensions exhibited in the offshore atmosphere. A stochastic simulation method of pits and a finite element modelling method for randomly pitted plates are established and validated by tensile tests of corroded specimens. Monte Carlo simulations are performed to explore the statistical features of mechanical property indicators and curve shape parameters under different pitting damage. Random pits induce great degradation and variability in the mechanical properties of corroded steel, more pronounced in offshore atmospheres. Yield strength, peak strain and hardening curve shape parameters follow the normal distribution related to pitting damage parameters, and other properties vary linearly with yield strength. A probabilistic constitutive model of corroded structural steel considering the stochastic characteristics of the corrosion pits is established with the pitting damage parameter as the control parameter.