Statistical appraisal of the atmospheric corrosion of mild steel

Abstract

A linear regression model has been proposed for correlating the rate of atmospheric corrosion with a number of atmospheric parameters, namely temperature, relative humidity, rainfall, number of rainy days, sulphur dioxide and sodium chloride concentrations. Goodness of the fit and conventional F- and t- tests have been carried out to identify the statistically significant parameters. Sulphur dioxide concentration has been identified to be statistically most significant in industrial areas whereas sodium chloride was most relevant in coastal areas. The linear regression model has a fairly good predictive capability. Nearly 76% of the calculated corrosion rates fall within ±40% deviation from the experimental values.