Acrylic acid modified indapamide-based polymer as an effective inhibitor against carbon steel corrosion in CO2-saturated NaCl with variable H2S levels: An electrochemical, weight loss and machine learning study

Abstract

In this work, the performance of an acrylic acid-indapamide based polymer, poly(AAI), was investigated as a corrosion inhibitor in mitigating C1018 carbon steel in a 3.5 % NaCl solution saturated with CO₂ in the presence of different H₂S concentrations simulating an oilfield sweet and sour corrosive environment. The effectiveness of the inhibitor was evaluated using weight loss and electrochemical techniques supported by SEM, EDS, and AFM surface techniques. At a concentration of 80 ppm, the polymer showed an inhibition effect of over 94 %. The effect of H₂S concentration on the inhibitor's effectiveness was also investigated. The inhibitor was effective under all conditions tested. The adsorption of poly(AAI) was consistent with the Langmuir adsorption model and poly(AAI) acted mainly as a cathodic-type inhibitor. EIS, SEM/EDS, and AFM studies of the steel surface morphology showed that poly(AAI) forms a protective layer on the steel surface. The corrosive elements were successfully prevented from accessing the steel surface by the protective layer. Machine learning was also performed to predict the %IE of poly(AAI) using four distinct machine learning models: decision tree regression (DTR), artificial neural networks (ANN), linear regression (LR), and Gaussian process regressor (GPR). A 10 k-fold cross-validation in addition to the mean absolute error, mean squared error, root mean square error, mean absolute percentage error, and determination coefficient procedure was used to assess the models' performance accuracy. Overall, the LR performs the best in terms of performance hierarchy robustness, followed by GPR, ANN, and =DTR. Consequently, the LR model was found to be a more reliable choice for estimating the %IE of poly(AAI), providing potentially more accurate information about material performance in practical applications.