Multivariate weighted recurrence network inference for uncovering oil-water transitional flow behavior in a vertical pipe

Zhong-Ke Gao,Yu-Xuan Yang,Qing Cai,Shan-Shan Zhang,Ning-De Jin

DOI: https://doi.org/10.1063/1.4954271

2016-01-01

Abstract:Exploring the dynamical behaviors of high water cut and low velocity oil-water flows remains a contemporary and challenging problem of significant importance. This challenge stimulates us to design a high-speed cycle motivation conductance sensor to capture spatial local flow information. We systematically carry out experiments and acquire the multi-channel measurements from different oil-water flow patterns. Then we develop a novel multivariate weighted recurrence network for uncovering the flow behaviors from multi-channel measurements. In particular, we exploit graph energy and weighted clustering coefficient in combination with multivariate time-frequency analysis to characterize the derived complex networks. The results indicate that the network measures are very sensitive to the flow transitions and allow uncovering local dynamical behaviors associated with water cut and flow velocity. These properties render our method particularly useful for quantitatively characterizing dynamical behaviors governing the transition and evolution of different oil-water flow patterns. Published by AIP Publishing.