Study of hydrate generation risk in gas-saturated oil-water emulsion system based on artificial intelligence

• Published in: Chemical engineering research & design Vol. 204; pp. 137 - 146
• Main Authors: Lv, Xiaofang, Chen, Shukai, Zeng, Wenguang, Liu, Yang, Wang, Chuanshuo, Zhou, Shidong, Song, Shangfei, Shi, Bohui
• Format: Journal Article
• Language: English
• Published: Elsevier Ltd 01.04.2024
• Subjects: Blockage risk; CNN; Gas saturation; Hydrate volume fraction; Water-in-oil emulsion; Water-in-oil emulsion; CNN; Blockage risk; Hydrate volume fraction; Gas saturation
• ISSN: 0263-8762
• DOI: 10.1016/j.cherd.2024.02.034

The formation and plugging of hydrates in oil and gas pipelines will bring serious challenges to the safe operation of oil and gas pipelines, but at present, domestic and foreign researchers have not formed a unified understanding of the formation mechanism of hydrates in oil and gas pipelines as well as the plugging mechanism, and its study is still a hot issue in the current domestic and foreign research. For the problem of hydrate plugging in oil and gas pipelines, one of the key parameters is the volume fraction of hydrate in the pipeline, the size of which is directly consistent with the trend of hydrate plugging in the pipeline, and therefore it can be used as a characterisation parameter for the trend of hydrate plugging in oil and gas pipelines. However, the calculation of the volume fraction of hydrate in pipeline is still difficult, especially the quantification of the volume fraction of hydrate in pipeline under the flow regime is affected by multiple parameters, which has become a difficult research point in the industry. To address the above problems, this thesis first carries out flow system experiments to explore the formation, blockage and hydrate volume fraction of hydrates in oil and gas pipelines under different operating parameters, and then establishes a hydrate volume fraction prediction model by using artificial intelligence deep learning methods, trains the model parameters on the above experimental data, and uses the prediction model to predict the other experimental data, and obtains better Effect. It guides the future application of artificial intelligence technology in the prevention and control of natural gas hydrate in oil and gas pipelines. •The study explains the formation process and mechanism of hydrate in water-in-oil system with saturated gas.•The prediction model for hydrate volume fraction is established based on CNN, and the prediction effect is good.•The model predictions under the new experimental conditions matched up well with the actual observations.