Multifractal magnitude and sign correlations in multiphase flows

Abstract

It is important to reveal the complex phenomenon of bubble coalescence or breakup in gas-liquid two-phase flow to understand the dynamic mechanism of flow pattern transition. Magnitude and sign decomposition pave a way to study the linear and nonlinear fractal characteristic separately. In this work, we extend this method to multifractal analysis, namely the multifractal magnitude and sign correlations. Test was conducted and an eight-electrode rotating conductivity sensor is utilized to collect the signals of different flow patterns in the gas-liquid two phase flow. The signal is initially decomposed into magnitude time series and sign time series. The width and difference of multifractal spectrums are introduced to characterize the nonlinear dynamics in the flow pattern evolution that gas-liquid two-phase slug flow breaks into the bubble flow or loses stability transforms into churn flow. And then these two parameters of multifractal magnitude and sign decomposition are fed into k-Nearest Neighbor for flow pattern identification. The results suggest that multifractal magnitude and sign decomposition has higher prediction accuracy (93%) than traditional multifractal analysis. The findings show that multifractal magnitude and sign correlations analysis is a powerful technique for investigating the flow dynamic of gas-liquid two phase flow.