Machine Learning Approach to Accurately Model Corrosion Rates As a Function of Time

Internal and external corrosion in the oil and gas industry is a major concern. A widely-used method to lower the internal corrosion of oil pipelines is injecting corrosion inhibitors (CI) into the oil stream. Monitoring the corrosion rate as a function of time in the presence and absence of corrosion inhibitors is imperative to ensure that failures are timely detected. In the absence of a robust mathematical model to predict the corrosion rate as a function of time, frequent measurements of corrosion rates are performed, which are both expensive and time-consuming. In this study, we have employed different machine learning (ML) approaches to model the corrosion rate as a function of time for any different CI concentration and dose sequences based on the experimental data. We show that a Random Forest (RF) based ML model is accurately able to model the corrosion rate as a function of time. In this presentation, we will discuss the methodology of developing the ML model and the results obtained.