Key points are not available for this paper at this time.
The sucker rod is constantly exposed to the risk of fracture failure during the oil production process. Accurate detection and prediction of the working condition and lifespan of sucker rods are crucial steps in mitigating and preventing sucker rod fracture failures. Currently, major domestic and international approaches employ mechanistic and data-driven models for condition monitoring and life prediction of pumping units. To differentiate the application of detection and prediction models under different fracture failure mechanisms, this paper provides a comprehensive summary of related models in terms of sucker rod fracture failure mechanisms, detection research, and life prediction. It also discusses the challenges encountered by sucker rod fracture detection and life prediction models in practical engineering application. The review results indicate that although both the mechanism-driven model and data-driven model can fulfill the accuracy requirements for detecting working conditions and predicting the lifespan of conventional sucker rod systems, the former is limited by certain assumptions, and the latter faces challenges regarding data quality assurance. Therefore, it is recommended to integrate the mechanism-driven model into future research and create an interpretable, extensive dataset based on the mechanism model. Additionally, research should focus on developing data-driven machine learning models that adhere to relevant theoretical constraints. This approach will offer a theoretical foundation for fracture detection and life prediction throughout the entire lifespan of modern sucker rods.
Jiang et al. (Mon,) studied this question.