Feasibilty of hyperspectral sensing for detection of early stages of corrosion in sturcutral steel

The estimated global cost of steel atmospheric corrosion is 3–4 % of the Gross Domestic Product, despite decades of corrosion mitigation practice. Steel, widely used for infrastructure construction, is composed primarily of iron and therefore is extremely vulnerable to atmospheric corrosion. The free iron atoms are highly reactive with water and oxygen molecules abundantly found in the atmosphere. Noncontact sensing for corrosion detectiojn is limited to corrosion with visual manifestation. In this study, diffuse spectroscopy is investigated for the feasibility of corrosion detection before visual manifestation. A36 steel samples were exposed to a corrosive alkaline medium with increasing exposure time. The sample's reflectance spectra were collected in the visual near-infrared (VNIR) and near-infrared (NIR) ranges. The existence of early corrosion products in visual sound samples was verified using X-ray diffraction. Results indicated the presence of corrosion at 600–710 nm in VNIR with a maximum of 97 % change with respect to the uncorroded samples. Hyperspectral images at around 700 nm were analyzed, confirming the presence of corrosion at an incremental rate (2.71–12 %) as exposure increased.