Statistical constitutive model of steel in randomly pitted structures

Abstract

Pitting corrosion degrades mechanical properties of steel by inducing thickness reduction and stress concentration on the rough surface of a structural member in ship and marine structures. This paper presents a scheme for ultimate strength assessment of randomly pitted structures to treat such an adverse effect using an equivalent steel of degraded mechanical properties. Tensile tests of steel specimens with/without artificial corrosion pits were carried out to validate their numerical models. Numerical analyses were performed to explore the effects of pitting parameters (angularity, spacing, coalescence, and ratio of diameter, d, to depth, h, of pits) on tensile strength of pitted specimens. Numerous stochastic simulations indicate that random natures of pitting parameters induce a great variation of mechanical properties of a randomly pitted specimen obeying a normal distribution. Meanwhile, tensile strength of randomly pitted specimens is highly depended on equivalent d/h (the ratio of mean diameter to mean depth of all pits) besides degree of volume loss (DOV) of corroded material. A constitutive model of randomly pitted steel was established to be bilinear, whose material parameters were statistically described by the equivalent d/h and DOV. It was verified to be sufficiently precise in predicting the ultimate strength of pitted structures.