Statistical characterization of surfaces of corroded steel plates

The statistical characteristics of corroded steel plate surfaces exposed to marine environments are of interest for assessing longer term structural safety and integrity using probabilistic methods. This requires information about the variability of corrosion loss and pitting over surfaces. The present paper reports on the observed statistical character of the surfaces of 10 large (1.2 m × 0.8 m × 3 mm thick) steel plates exposed in temperate climate marine immersion, tidal and splash zones for 2.5 years. For the analysis the plates were cut into smaller segments that were mechanically scanned to obtain digitised surface topographies. These were then analysed to estimate the correlation structure and the standard deviation of the surface topography. Considerable differences were found for these and for the mean corrosion loss between different exposure zones. For any one segment the surface topography was found to be highly statistically dependent, implying that smaller coupon sizes can provide adequate estimates of corrosion loss. From this it may be inferred that the deepest pits are not statistically independent as commonly assumed in extreme value statistical representations.