Electronic Thesis and Dissertation Repository


Doctor of Philosophy


Chemical and Biochemical Engineering


Jesse Zhu


The microscopic flow structure is studied systematically and comprehensively through a visualization system in a narrow rectangular circulating fluidized bed (CFB) transparent Plexiglas riser with a height of 7.6 m and a cross section of 19 mm × 114 mm. A visualization system consisting of a high-speed video camera, a light source, and image processing and analyzing programs is designed and developed to enable the flow structure to be visualized directly and studied quantitatively. FCC particles of 67 μm are used as the bed materials under different operating conditions with superficial gas velocity (Ug) and solids circulation rate (Gs) in the range of 3.0-12.0 m/s and 50-150 kg/m2s respectively.

To study the microscopic flow structure quantitatively, for the first time, a new calibration method is developed to correlate the solids holdup of the FCC particles and the grayscales of the images obtained by the high-speed video camera, based on the light illumination consistency verified by a reference plate. To achieve stable and homogeneous fluidization with uniform solids holdup, the calibration experiment is conducted in a well-designed liquid-solid bed. The obtained calibration curve and equation are used as the basis for the entire study. With the calibration method, cluster can be “peeled-off” by given solids holdup thresholds through transforming the original gray images into binary images. The change in cluster population with operating conditions consistent with previous researchers further proves that the image calibration method developed in this study is effective and very useful.

To further verify the newly developed image calibration method, an optical fiber probe is applied as a reference for the measurements of the solids holdup of the FCC particles. The solids holdup distribution profiles obtained from the two methods under identical operating conditions have good agreement, reflecting the reliability of the image calibration method. Further comparison of the results of image calibration method from the current study with the measurement results of optical fiber probe from other researchers also show good agreement under the same operating condition. These comparisons clearly confirm the feasibility and accuracy of the image calibration method.

Using the image calibration method, the mean solids holdup under different operating conditions can be calculated from the mean grayscale of the images. The results show that the mean solids holdup increases with the increasing Gs and decreases with Ug. The transformation from grayscale images into Hue, Saturation and Value (HSV) images using various solids holdup thresholds allows the dense and dilute phases with obviously different solids holdup to be clearly visualized under different operating conditions in the fully developed region. A term “relative dense phase area” is introduced to quantify the solids phase separation. A critical solids holdup value of εsc = 0.04 is chosen by carefully examining the variation profiles of the relative dense phase area with solids holdup thresholds, to demarcate the dilute and dense cluster phases. The cluster fraction is then obtained through the εsc value and ranges from 1 % to 59 % under the different operating conditions of present research. With images divided into three regions along the lateral direction, it is found that cluster fraction at the wall region is higher than that of the core and the middle regions.

With further examining the solids holdup distribution of the microscopic structures, the dense (or cluster) phase is considered as a “compound” of core clusters and intermediate dispersed particles, which is in the processing of coalescence or breakup, with higher solids holdup than the dilute phase. To identify stable existing core clusters, a systematic cluster identification process is presented by adopting a threshold selection method to obtain the cluster threshold solids holdup (εsct) so that clusters can be identified under different operating conditions. The cluster fraction is calculated by dividing the total number of pixels belonging to the core cluster with the total pixels number of the entire image. Based on the εsct, a cluster equivalent diameter (dc) is determined by the area of the cluster in the binary image. At the same time, the cluster solids holdup can be determined by converting the grayscale of the cluster from the original image to the solids holdup. Moreover, cluster vertical velocity can be determined by the shift of clusters between sequential binary images. Typical dense (Ug = 3.0 m/s; Gs =100 kg/m2s) and dilute (Ug = 9.0 m/s; Gs =50 kg/m2s) operating conditions are selected to compare the variation of the cluster size and velocity.