Complex network-based framework for flow pattern identification in vertical upward oil–water two-phase flow

Abstract

The investigation of oil–water two-phase flow in vertical pipelines holds significant research implications for a multitude of industrial applications, including oil production, chemical processing, and wastewater treatment. This research introduces a complex network-based framework for analyzing multi-node measurement signals from an eight-electrode cyclic excitation conductivity sensor, aimed at recognizing intricate flow patterns in vertical upward oil–water two-phase flow. Initially, experiments on vertical upward oil–water two-phase flow were conducted in a 20 mm diameter pipeline, where flow dynamics were recorded using the aforementioned sensor. During the experiments, flow patterns captured by a high-speed camera included dispersed oil-in-water slug flow (D OS/W), dispersed oil-in-water flow (D O/W), and very fine dispersed oil-in-water flow (VFD O/W). Subsequently, the multivariate pseudo-Wigner–Ville distribution time–frequency representation (PWVD TFR) was employed to characterize the flow behavior from both energy and frequency perspectives. Finally, the sensor’s measurement nodes were treated as nodes in a network, and the mutual information between each time series was calculated to construct a complex network; network metrics were then computed to quantitatively characterize the network topology. The findings indicate that our method can effectively integrate multi-channel measurement signals and reveal the evolution of complex flow behaviors.