Prediction and Detection of Localised Corrosion Attack of Stainless Steel in Biogas Production: A Machine Learning Classification Approach

Abstract

Biogas contributes to environmental protection by reducing greenhouse gas emissions and promoting the recycling of organic waste. Its utilization plays a crucial role in addressing the challenges of climate change and sustainability. However, the deterioration of process plants involved in biogas production due to corrosion has a critical impact on the safety and durability of their operations. In order to maintain the safety of structures in terms of service life with respect to corrosion, it is essential to develop effective corrosion engineering control methods. Electrochemical techniques have become a useful tool by which to evaluate corrosion resistance. However, these techniques may require microscopic analysis of the material surface and the analysis may be influenced by subjective factors. To solve this drawback, this work proposes the use of SVM models to predict the corrosion status of the material used in biogas production with no need to perform microscopic analysis after the electrochemical test. The obtained results of sensitivity and specificity equal to 0.94 and 0.97, respectively, revealed the utility of the proposed stochastic models to assure the corrosion state of the equipment involved in biogas production. SVM-based models are an effective alternative for accurately evaluating material durability and comparing the corrosion resistance of different materials in biogas environments. This approach facilitates the selection of the most suitable material to achieve greater durability and long-term performance. Synopsis: The results show that the proposed model is a useful tool to predict the behaviour of stainless steel against corrosion according to the environmental conditions to which the material is exposed in biogas production.