Predictions of flow transitions in a bubble column by chaotic time series analysis of pressure fluctuation signals

In this study, predictions of flow transitions of a bubble column are carri ed out by chaotic time series analysis of pressure fluctuation signals. Pre ssure signals are measured for four different distributor plates with the s uperficial gas velocity varying from 0 to approximately 0.22 m/s. Methodolo gy of the chaotic time series analysis is presented, including procedures o f the lime-delayed phase-space reconstruction and calculation of the chaoti c invariants. Four chaotic invariants, which include the largest Lyapunov e xponent, metric entropy, correlation dimension and mutual information, of a ll pressure signals are computed. The critical gas velocities are determine d by the sharp increase or decrease of the obtained chaotic invariants at t he flow regime transitions. This simple criterion is consistent with the ex perimental observations. The obtained critical gas Velocities by various ch aotic measures for the four different distributor plates are quantitatively consistent and are in good agreement with the results by the gas holdup da ta. (C) 2001 Elsevier Science Ltd. All rights reserved.