Gas Jets in Granular Matter: Observed by a High Speed X-ray Tomography System

Abstract

High speed X-ray tomography is a promising tool to visualize the time-resolved gas distribution for fluidized beds. The tomography unit at Delft University of Technology is composed of three X-ray tubes and three double-layer detector arrays. The X-ray tubes are places at 120? around the fluidized bed. The detector arrays are placed opposite to the X-ray tubes. The full angle scanning can be done simultaneously, without mechanically rotating the X-ray tube as in traditional medical CT. The temporal resolution can be up to 2500 fps. By using this high speed X-ray tomographic system in my dissertation, we study tomographic reconstruction algorithms and measure different gas jets in the fluidized beds. To evaluate the performance of the novel reconstruction algorithms for high speed X-ray tomography, Simultaneous Algebraic Reconstruction Technique (SART) and Adaptive Genetic Algorithm (AGA) are compared in Chapter 3. The reconstruction quality of AGA is better than SART at low resolutions, but SART performs better at high resolution in finding the right shape. The system noise influences AGA less than SART. SART is much faster and does not have reproducibility problems; poor reproducibility influences the reliability of AGA. The features of the reconstruction quality for SART and AGA are further discussed in Chapter 4. AGA is better than SART in distinguishing small phantoms and small distance between phantoms. SART, on the other hand, is better at reconstructing the shapes than AGA. We developed a hybrid approach to combine the advantages of AGA and SART. The result of AGA is used as initial guess for SART in this hybrid algorithm, by which the reconstruction accuracy is indeed improved. A flat-base spouted bed is studied with the high speed X-ray tomography system in Chapter 5. We focus on the spout diameter and position. The hybrid algorithm is used to visualize the spout in the cross sectional area of the spouted bed. For a more accurate quantitative study, we developed a method to process the raw data directly. The time-averaged results of the spout diameters are validated with the literature. A simple model for the particle circulation is developed based on the time-averaged spout diameter. The model is also validated by measurement of the fountain height. The time-resolved results of the spout diameter are analyzed by the Power Spectral Density (PSD). The stability of the spout diameter is discussed for different measurement heights and gas flowrates by comparing the average PSD. We find that the most stable spout diameter happens at U/Ums ? 1.3 in the middle part of the spout. The time-resolved spout position is plotted by polar coordinates. We found that the spout position is more stable when the diameter is less stable, and vice versa. Another gas jet, i.e. the gas distribution below a downward facing micro-nozzle in a fluidized bed, is also measured using the high speed X-ray tomography in Chapter 6. The improved SART method is used for the reconstruction because we need to consider the transition area between the gas voids and bulk phase. Time-averaged 3-D images of the gas distribution below the nozzle are obtained. The results are compared with direct analysis of the raw data. A bubbling area, a diluted area, and a compacted area are found. We analyzed the dynamics of the gas voids by employing a cross-correlation technique and inspecting the reconstructed pseudo 3-D image. The cross-correlation estimates the direction of bubble motion, and also helps to calculate the bubble velocity and bubble size. The pseudo 3-D image shows the pattern of the gas voids. We found that single bubbles are regularly formed by the nozzle, and move upwards. The expansion, splitting and coalescence of the bubbles were also observed. The gas injected from the nozzle mainly ends up in bubbles. We compared the fluidized bed results with those obtained for a similar nozzle in a gas-water system, and found a similar flow pattern and penetration depth. In summary, we improved the accuracy of high speed X-ray tomography by developing novel data processing approaches, such as the hybrid algorithm and raw data processing. We evaluated the dynamics of the spout in a flat-base spouted bed from the tomographic measurements, and developed a model for particle circulation based on these measurements. We also measured the time-resolved gas distribution below the downward facing micro-nozzle using the high speed X-ray tomography. The obtained results illustrate that X-ray tomography is a valuable tool to study gas-solids distribution – both time-averaged and time-resolved in fluidized beds.