Flow Structure Analysis in a Rotating Centrifugal Impeller under Gassy Conditions Using Dynamic Mode Decomposition

Electric submersible pumps (ESPs) represent a widely adopted artificial lift technology deployed across diverse oil production settings. Under gas–liquid flow conditions, the internal flow within the impeller exhibits pronounced unsteadiness driven by complex phenomena including gas accumulation and boundary layer detachment. These effects significantly deteriorate pump performance. Conventional approaches for analyzing two-phase flow in ESPs show limitations in identifying dominant flow structures, characterizing instability mechanisms, and predicting transient flow behavior. To overcome these challenges, this study employs Dynamic Mode Decomposition (DMD), i.e., a data-driven modal analysis framework, to systematically characterize unsteady gas–liquid flow dynamics within ESP impellers. DMD is used to decompose spatiotemporal phase and velocity fields, extract dominant modes, and identify characteristic frequencies associated with flow disturbances. The method further enables reconstruction and short-term prediction of the flow field, providing a foundation for assessing flow stability and understanding gas–liquid interaction mechanisms in high-shear rotating environments. Results indicate that DMD effectively isolates coherent structures that dominate the two-phase flow field and reveals key features linked to instability and gas entrainment. Compared with traditional analytical and numerical techniques, DMD offers distinct advantages in dynamic feature extraction, modal interpretation, and computational efficiency. Through extraction of the dominant spatiotemporal features of the flow field, DMD enables enhanced understanding of flow development and transient phenomena. Overall, the findings provide new insight into the mechanisms governing ESP performance under gas-containing conditions and contribute to strategies for enhancing operational stability and reliability.