Evaluation of the effects of corrosion of carbon steel in acid mine drainages by using an explainable artificial intelligence model.

Abstract

One of the main problems faced by sulfide mining facilities is the corrosion of steel caused by acidic leachates, a process that is further accelerated by the presence of acidophilic bacteria that act as catalysts in the reaction. This study presents an AI-based approach to assess acid mine drainage as a result of corrosion, adding novelty to the study by using it alongside Fuzzy Logic, obtaining qualitative models based on fuzzy rules that corroborate and expand on the contributions of statistical techniques. In a laboratory experiment, S-235JR (EN 10025) carbon steel samples were immersed for 26 days in four solutions: two with a high Fe2+/Fe3+ ratio, one from a mine with acidophilic bacteria (with an initial Fe2+ concentration of 1390 mg/L) and one synthetic without bacteria (Fe2+ concentration of 1961 mg/L); one from a mine with a high Fe3+/Fe2+ ratio (with an initial Fe2+ concentration of 0.10 mg/L) and distilled water simulating rain effects. Results showed a greater increase in Fe concentration in solutions with a higher initial Fe2+/Fe3+ ratio (with an increase rate higher than 2). All acidic solutions eventually became anoxic, with a slower transition in the bacteria-free solution. Cu concentration dropped sharply in all cases, regardless of bacteria presence or Fe ratios, likely because the net acidity kept Cu in solution for a longer time. In the synthetic solution without bacteria, pH gradually increased as steel oxidation proceeded more slowly. In contrast, solutions with bacteria showed a sharp pH rise in the early hours of the experiment.

Keywords

• Acid mine drainage
• Artificial intelligence
• Carbon steel
• Corrosion
• Fuzzy logic
• Mining facilities