On predicting long-term corrosion behaviour from short-term tests

It is common in corrosion practice to use relatively short-term observations to make inferences about the longer term reliability of infrastructure components or mechanical equipment. Short-term accelerated tests are widely used, even though they are known to produce only indicative results, not necessarily related to long-term field experience. A significant challenge remains in extrapolating short-term results to the longer term. It has been common, in corrosion science to use a simple model to extrapolate on the assumption that the underlying corrosion process does not change with time. However, as has now been shown for a number of realistic scenarios such as immersion, splash, atmospheric marine environments, this appears not to be the case and that the underlying mechanics are more complex. Some aspects of this were investigated by exposing identical steel coupons to natural seawater and to identical seawater but sterilized so as to remove all living material. This produced differing mass losses as a function of time. The results from this are used in comparison with other data, reported in the literature, to indicate that the controlling, underlying reactions change as the corrosion process proceeds, even without the influence of microbiological activity. It is proposed that these findings have major implications for estimating the long-term reliability of infrastructure.