Particulate Science and Technology, Vol.31, No.1, 51-57, 2013
Hydrodynamic Characterization of Bubbling Fluidization by Principal Component Analysis of Pressure Fluctuations
Original pressure signals of a gas-solid fluidized bed were decomposed by principal component analysis into four independent signals (D 1 to D 4), with respect to their structures. The Hurst exponent was used to investigate the decomposed signals. The decomposed signals D 1 and D 2, with the Hurst exponents considerably greater than 0.5, were attributed to the macrostructures (e.g., bubbles). The bi-fractal decomposed signal D 3 with two entirely different Hurst exponents was ascribed to the mesostructures. While the mesostructures of the dilute phase with a higher Hurst exponent are similar to the macrostructures including deterministic behavior (e.g., small bubbles), the mesostructures of the dense phase with a lower Hurst exponent act as microstructures of the bed, with a stochastic behavior (e.g., clusters). The mono-fractal decomposed signal D 4 with the Hurst exponent significantly smaller than 0.5 was regarded as the microstructures of the bed (e.g., particles interaction). Behavior of macrostructures does not change noticeably with growing the gas velocity owing to small change of the Hurst exponent. However, dynamics behavior of the dilute phase of mesostructures develops into more persistent and deterministic. Behavior of the dense phase of the mesostructures and microstructures are inclined to random behavior in the same situation.
Keywords:Cluster;fluidization;Hurst exponent;nonlinear dynamics;pressure;principal component analysis