Solid particle erosion prediction in elbows based on machine learning and swarm intelligence algorithm

Abstract

Continuous impact of solid particles causes severe pipeline wear, and may result in leakage in directional change areas such as elbows. Accurate prediction of erosion is essential in the petroleum industry. In this study, a new machine learning (ML)-based model for predicting elbow erosion was established. Seven parameters were selected as feature inputs from the fluid characteristics, particle characteristics, and pipe characteristics. The maximum erosion rate was used as the predicted output. Based on the gas–solid flow data, the prediction accuracy of different ML models was compared. The kernel extreme learning machine (KELM) model was considered as the optimal model. For gas–liquid–solid flow, incorporating swarm intelligence (SI) algorithms, the whale optimization algorithm–hybrid kernel extreme learning machine (WOA–HKELM) erosion prediction model was proposed; the predictions were compared with the experimental values. The root mean square error (RMSE) of the prediction was 0.82 × 10−3, which is consistent with the experimental results. It was also demonstrated that the model can capture the trend of the influence of mixed dimensionless inputs on the prediction results in churn and annular flow.