Enhancing Crude Oil Pipeline Design Efficiency Through Explainable AI: A COMSOL Simulation Approach

Assessing the loss of head, power, and heat is vital to ensure the safe, efficient, and economically sustainable operation of crude oil pipelines. These assessments are crucial for operators aiming to optimize transportation processes, uphold product integrity, and mitigate environmental risks. This study focuses on AISI1020 steel pipelines, consisting of three components with varying diameters, each of equal length. A numerical investigation was conducted using COMSOL Multiphysics software to explore heat transfer in the crude oil pipeline system. Subsequently, data were generated for different heat transfer and fluid parameters by simulating the developed numerical model. Machine learning and neural network models were then employed to predict energy losses in this setup. Moreover, quantile loss prediction was performed to determine the most suitable prediction model. Additionally, SHAP analysis was utilized to deepen our understanding of the primary factors influencing energy losses across the three components of the steel pipelines. These findings underscore the critical importance of temperature management in efficiently designing the crude oil transportation pipeline network system.