A Method for Predicting the Gel Point of Mixed Crude Oil Based on Deep Learning
Bo Ouyang, Guangchao Zhang, Wei Liu, Shujuan Qiu, Wendian Ding, Zihang Zhang, Huai Su
Abstract
Accurately predicting the gel point of mixed crude oil is crucial for efficient pipeline transportation and blockage prevention. Traditional manual methods are inadequate for real-time monitoring. This study introduces an accurate prediction model for the gel point of mixed crude oil using a hybrid deep learning model that combines Graph Connection-Skip Neural Networks (GCSGCN) and Gated Recurrent Units (GRU). Using physical properties data (gel point, viscosity, density) from an oil mixing station in China (2009-2023), the proposed model outperforms empirical formulas and machine learning models like XGBoost.
Topics & Concepts
Crude oilComputer scienceDeep learningPoint (geometry)Artificial intelligencePetroleum engineeringMachine learningEngineeringMathematicsGeometryPetroleum Processing and AnalysisHydrocarbon exploration and reservoir analysis