Paste drying control in a rotating-distributor fluidized bed

Resumen

Fluidized beds are used for a wide variety of processes due to its high rates of heat and mass transfer. Due to that, fluidized beds are used for a variety of applications ranging from the gasification or combustion of biomass, to coating and drying processes. All these industrial applications show the main influence of the zone close to the distributor on the bed dynamics. In this way, a non-homogeneous mixing in this zone between dense and gas phases leads to fluidization problems such as segregation of particles, formation of dead zones, heterogeneous distribution of temperatures inside the bed and appearance of channels or preference paths for the flowing fluidization gas. These phenomena usually cause the process stop or the bed defluidization, which increase operational costs. Therefore, it is necessary to study the bed dynamics in order to be able to detect these fluidization problems, and modify somehow the hydrodynamic structure of the bed. This PhD thesis develops the methods needed to study and control the behaviour of a bed equipped with a rotating distributor as a tool to change the bed dynamics. These approaches were developed using a lab-scale facility with an electrical motor coupled to the distributor, capable of imposing its rotation in the horizontal plane. The experimental conditions of poor fluidization quality typically shown in industrial applications are studied to assess the potential use of the rotating distributor on the bed dynamics. Punctual injections of water over the bed surface were carried out to replicate the experimental conditions reported in literature. This water injection improved the development of dead zones over the distributor plate, facilitating the bed defluidization. The defluidization and re-fluidization processes were characterized using pressure fluctuation signals recorded during the experiments. Methods of analysis in the time domain, standard deviation, and in the frequency domain, Welch0s power spectrum, transient spectrogram and wide band energy, were employed to describe the defluidization and re-fluidization processes. By using the rotating distributor, the fluidization quality was improved due to its breakage effect on the agglomerates and channels settled over of the distributor. Such an improvement was shown for both shallow and deep beds. Therefore, the distributor rotation introduces a dynamic structure into the gas-solid fluidized bed. Moreover, concerning the methods of signal analysis, the channels appearance is related to an energy increase of the high frequencies within the pressure signal. The approach initially employed for the computation of the wide band energy was based on the visual division of the frequency domain. However, the results obtained using this method are influenced by the observer, specially if the quality of the measured signal is low. Therefore, an unbiased methodology is proposed for the systematic division of the frequency domain. The approach is based on the statistic distribution fitting of the power spectrum cumulative energy. Hence, the Kolmogorov-Smirnov test is used to compare the cumulative energy with the statistic distribution. The differences that appear at the distribution tails when comparing the cumulative energy with the statistic distribution discriminate the cut-off frequencies that separate the three frequency regions of the frequency domain. For the operation conditions considered, the Student0s t-distribution was used to fit the cumulative energy distribution. The reliability of the method to divide the frequency domain was shown for different fluidization velocities, changing the bed aspect ratio and using different pressure probes. Water-induced defluidization tests were conducted to illustrate the use of the wide band energy as a monitoring tool. Both visual and statistical approaches were compared. The results showed that the energy contained within the frequency regions obtained by the visual method is not able to detect changes in the bed aspect ratio or the beginning of the rotating distributor. On the other hand, the sensitivity exhibited by the proposed frequency division approach, for the range of fluidization conditions tested, encouraged the use of the energy contained in these regions as a diagnostic tool in fluidized bed processes. In order to determine whether or not the bed behavior has changed, the implementation of a monitoring system capable of ensuring the process operation at a certain level of efficiency is required. Therefore, an on-line monitoring methodology is proposed. This approach is based on the Statistical Process Control theory and it is applied to different variables, which were obtained through the analysis of the measured pressure signals. The statistical treatment of the variables allowed the definition of the control state through the study of the sample size and its underlying distribution. The good results showed by the different control variables encouraged the use of the proposed control scheme to any monitoring variable. Finally, once the high influence of the rotating distributor on the bed dynamics was proved, and the analysis and monitoring tools needed were developed, a practical application is presented. The paste drying process with inert particles as a support medium is analyzed recording humidity and temperature measurements together with pressure fluctuation signals. The experiments consisted on the paste-drop, which was composed of a mixture of silica sand and water, to a bed of dry silica sand particles. In this way, the paste tends to be homogeneously dispersed throughout the bed, beginning the drying process. The multi-resolution approach of the pressure fluctuation signals showed the effect of the paste-drop, relating the drying periods to the bed dynamics. To compare the results obtained from the humidity and temperature measurements with those of the pressure signals, a Statistical Process Control approach is applied to define the time needed to recover the bed dynamic. For the static distributor, similar values of the drying and recuperation times are obtained for shallow beds, while for deeper beds more time is needed to complete the drying than to recover the fluidization quality. For the rotating distributor, lower drying and dynamic times are needed for shallow and deep beds.