Abstract:
Corrosion and fatigue have been dominant degradation mechanisms in offshore
structures, with the combination of the two, known as corrosion fatigue, having
amplified effects in structures in the harsh marine environments. Newer types of
structure are now being developed for use in highly dynamic, harsh marine
environments, particularly for renewable energy applications. However, they
have significantly different structural details and design requirements compared
to oil and gas structures, due to the magnitude and frequency of operational
and environmental loadings acting on the support structures. Therefore, the
extent of corrosion assisted fatigue crack growth in these structures needs to be
better understood.
In this research, fatigue crack growth in S355J2+N steel used for offshore wind
monopile fabrications was investigated in air and free corrosion conditions.
Tests were conducted on parent, HAZ and weld materials at cyclic load
frequencies similar to what is experienced by offshore wind monopile support
structures. The seawater used for testing was prepared according to ASTM
D1141 specifications and was circulated past the specimens through a purpose
designed and built corrosion rig at a rate of 3 l/min, at a temperature of 8-100C
and at a pH of 7.78-8.1. A new crack propagation method accompanied by
constant amplitude loading was used. Crack growth rates in parent, HAZ and
weld materials were significantly accelerated under free corrosion conditions, at
all the stress ratios used compared to in air environment. However, in free
corrosion conditions, crack growth rates in the parent, HAZ and weld materials
were similar, particularly at a lower stress ratio. The results are explained with
respect to the interaction of the loading condition, environment and the rate of
material removal by corrosion in the weldments. A new model was developed to
account for mean stress effects on crack growth rates in air and in seawater,
and was found to correlate well with experimental data as well as with the other
mean stress models tested.