An investigation on the effect of various factors affecting the corrosion of metals in soils using statistical analyses

In this article the effects of various factors on steel corrosion in soil has been investigated by statistical analyses. Due to lack of such data in Iran, we used corrosion data from NBS studies, conducted between 1922 to1947. In order to determine the portion of each factor in soil corrosion, a multivariate regression analysis was carried out between each one of the corrosion factors and mean mass loss of the buried steel pipes. Finally the best model has been determined based on the P value criteria shown in F test and T test tables. This analysis repeated for the average time of 1.5, 11.5 years and the average of the entire test (18 years) after burial. Results of correlation matrix show that the highest correlations are among test year and then total acidity respectively, and the rest of variables (chloride and sulphate ion content, moisture, precipitation, resistivity, and the pH) have been rejected. The final model is exponential model and its correctness has been confirmed according to standardized residuals analysis. Since the data are related to long term and real-world conditions, the results are consequential.