Comparison of Response Surface Methodologies and Artificial Neural Network Approaches to Predict the Corrosion Rate of Carbon Steel in Soil

Abstract

Soil corrosion is a critical problem that has recently interested many scientists. Several soil factors affect the corrosion rate of carbon steel, and they can all be relevant at the same time, thus making it difficult to maintain conditions across soil corrosion studies. There are currently two potential methods for predicting corrosion rates in a complex environment such as soils: the response surface methodology (RSM) and artificial neural network (ANN). RSM is the method using statistics to design experiments, while ANN predicts the corrosion rate through training based on human brain systems. In this study, these two methods will be implemented to predict the corrosion rate of carbon steel considering three factors: pH, temperature, and chloride. The prediction of corrosion rate is successful in both methods, and they have their own advantages and disadvantages.