ABSTRACT
For the assessment of the safe and useful remaining life of existing infrastructure and structural systems estimates are required of the likely future deterioration given that some degree of corrosion may already have occurred. The assessment of the present structural condition and the prediction of the future condition both involve uncertainties. As a result probabilistic models for corrosion loss are appropriate and consistent with modern economic and structural safety decision tools. For corrosion modelling it has been demonstrated recently that for corrosion in a marine environment the deterioration process is a complex non-linear function initially involving oxidation and, in the longer term, anaerobic bacterial activity. The resulting losses of structural material are not consistent with a constant ‘corrosion rate’ as commonly assumed in the structural engineering literature, nor with a power law or similar monotonic function as often assumed in the corrosion literature. This has important implications for the planning and design of experiments to estimate future likely corrosion. It is argued that the data from such experiments can be interpreted properly only with the aid of a properly calibrated corrosion loss model that is consistent with chemical, physical and bacteriological fundamentals and with environmental influences.
