Improvement of Reliability of Two-phase Flow Characterization by Point Correlation Dimension Analysis and Verification Using Submerged Nozzle Pressure Fluctuation Time Series

In order to improve the reliability of two-phase flow characterization by correlation dimension, several estimation methods are tested and their reliabilities are evaluated. The estimation procedure of the correlation dimension involves following steps : attractor reconstruction and fractal dimension estimation. There are many improvements for each step. However, they have been individually proposed and have not used simultaneously. Using numerically obtained quasiperiodic time series, we evaluate their combinations. Analysis results show that a point correlation dimension method with constant time window reconstruction is more reliable than the widely used Grassberger-Proccacia method with constant delay time reconstruction for ill-conditioned time series : data size is small and dimension is high. The estimation using experimentally obtained noisy chaotic pressure fluctuation time series also shows similar results, so we believe that the proposed procedure is reliable for various kind of time series.