Extrapolating short-term corrosion test results to field exposures in different environments

Abstract Corrosion in the field environment is influenced by time-varying humidity, temperature, and pollutants. To predict the corrosion loss, a corrosion rate model is proposed which consists of the link function and the time-varying environmental model. The link function incorporates the effects of multiple environmental factors on the corrosion process. It is suggested that the individual effect of relative humidity, temperature, and pollutants has different nonlinearity, while the combined effect of them is more complicated. The time-varying model describes the actual dynamic environment with seasonal, daily, and random components in good accuracy. It is superior compared to average value method as it includes more valuable information of the environment. Results suggest that the proposed model can obtain accurate predictions of annual corrosion accumulation within 10 % relative error to the actual value. The superiority is generated from the high accuracy of the environmental model and the good adjustment of the link function.