Estimating corrosion: a statistical approach

This study was an attempt to determine the corrosivity of seawater on pipes made of aluminum-brass and carbon steel by conducting experiments using a 2(7-1) fractional factorial design. Important variables used in assessing seawater corrosion included temperature, oxygen, sulfide, urea, chloride and inhibitor at two levels each. Thirty-two experiments were conducted for each alloy. The study investigated main effects (single variable), and two- and three-way interactions. The general linear model was used to estimate the various effects, and a statistical test of significance was carried out to separate the changes in the corrosion rates attributed to the impact of specific variables. Confidence intervals of the various effects were obtained, and the significance of various interactions were established for each alloy. It was found that oxygen and temperature contribute significantly in enhancing corrosion levels. (C) 2003 Elsevier Science Ltd. All rights reserved.