Characterization of fluidization regimes by time-series analysis of pressure fluctuations

This work compares lime, frequency and state-space analyses of pressure mea surements from fluidized beds. The experiments were carried out in a circul ating fluidized bed, operated under ambient conditions and under different fluidization regimes. Interpretation of results in time domain, such as sta ndard deviation of the pressure fluctuations, may lead to erroneous conclus ions about the flow regime. The results from the frequency domain (power sp ectral and state-space analyses (correlation dimension, D-ML, and Kolmogoro v entropy, K-ML, together with a non-linearity test) of the pressure fluctu ations are generally in agreement and can be used complementary to each oth er. The power spectra can be divided into three regions, a region correspon ding to the macro-structure (due to the bubble flow) and, at higher frequen cies, two regions representing finer structures that are not predominantly governed by the macro structure of the flow. In all fluidization regimes, t he measured pressure fluctuations exhibited an intermittent structure, whic h is not revealed by power spectral analysis of the original signals. Fluct uations with pronounced peaks in the power spectrum and in the auto-correla tion function, corresponding to passage of single bubbles through the bed, are non-linear with a low dimension (D-ML < 5.5). For D-ML < 5.5, the Kolmo gorov entropy is proportional to the amount of energy in the spectral range of the intermittent structures observed, whereas for D-ML > 5.5 both K-ML (bits/cycle) and D-ML are insensitive to changes in the distribution of ene rgy in power spectra. Thus, the state-space analysis reflects that non-line arity is mostly found in the macro-structure of the flow. Fluidized bed tim e series treated in this work are available at http://www.entek.chalmers.se / similar to fijo (C) 2000 Elsevier Science Ltd. All rights reserved.