Investigation of the geometric effects of corrosion pits on the tensile behaviour of structural steel

Abstract

Pitting corrosion removes parts of the structural steel’s cross section in a localised manner, therefore introducing localised stress concentrations which can cause decreased tensile and deformation capacity in a more pronounced effect than uniform cross sectional mass loss.

With aging infrastructures using structural steel used in chloride environments, there is a growing need for reliable assessment of corroded steel. Therefore, a methodology that can provide insights with the linkage of measurable corrosion parameter such as mass loss to mechanical response of the corroded steel material can prove to be a clear contribution for corrosion analysis. The methodology should also consider the localised and stochastic nature of pit formations. This thesis develops and applies a reproducible numerical workflow for this purpose and evaluates it against an experimental tensile programme on low-carbon structural steel coupons immersed in a 5% sodium-chloride solution for up to 42 days.

Flat dog-bone coupons conforming to ISO 6892-1 were corroded for 0, 14, 21, 28, 35 and 42 days, cleaned and weighed gravimetrically, and tested in uniaxial tension. The coupons were protection wrapped for the grip and shoulder regions and only the gauge section were exposed to the chloride immersion. Measured average effective mass losses ranged from 1.29% to 2.49%, with a power-law mass-loss-time relationship indicative of a diffusion-controlled process.

Finite element analyses were conducted with a stochastic pit-generation algorithm developed in MATLAB which was used to create randomised distributions of flat-bottomed cylindrical pits, with the total pit volume matched to the measured mass loss. The resulting pit fields were then converted into discrete cavities in an Abaqus/Explicit coupon model using a Python script that conducts automated pit insertion through reading the MATLAB pit generation output data. The material model was calibrated once using a representative lab-tested experimental uncorroded Day-0 tensile response and was subsequently kept unchanged, thereby isolating the simulated influence of corrosion to geometrical material loss. This calibrated geometry-isolated workflow was subsequently applied to a numerical mass-loss projection study, extending the corrosion severity beyond the experimental range to 12% gauge-section mass loss. At this higher severity, the ultimate tensile strength showed a more pronounced decrease compared to lower mass-loss levels, while the ductility loss obtained from the FE simulations remained strongly influenced by the pit geometry and spatial distribution of each stochastic realisation.

The uncorroded finite-element model reproduced the experimental ultimate tensile strength to within approximately 0.4% and was shown to be mesh-converged and quasi-static through an energy-balance study. Across the corroded cases, the experimental coupons showed no statistically clear monotonic reduction in either ultimate strength or fracture strain at the studied mass-loss levels. It was observed that the corrosion-induced change was smaller than the coupon-to-coupon scatter. On the other hand, geometry-isolated finite-element model, resolved a small reduction in ultimate strength (of order 2% over the exposure range) and a substantially larger reduction in fracture strain and toughness (of order 7-14%). This suggests that ductility is the corrosion-sensitive quantity and that strain localisation at pits are key factors of the loss of deformation capacity. Based on these findings, simple normalised predictive relationships are proposed in which the residual strength, fracture-strain and toughness factors decrease approximately linearly with mass loss, the ductility coefficient being about an order of magnitude larger than the strength coefficient. These relationships come with limitations which are discussed in the later section.

The principal contribution of the thesis is a single, seed-controlled, mass-loss-verified workflow combining stochastic pit generation, idealised-pit finite-element realisation and quasi-static tensile simulation, together with a quantified experimental demonstration which shows that at low-to-moderate mass loss, the geometric effect of pitting on steel coupons is concentrated in ductility rather than strength. The cylindrical pit idealisation, the fixed material card and the coupon scale are acknowledged as part of the limitations of this study.