Opaqueness and visual accessibility in turbulent bubbly flows are the main cause of difficulty for conventional experimental techniques to extract properties from the bulk flow. Ultrafast X-ray tomography presents an unique possibility to visualise dense bubbly flows and to provide the ability to characterise the bubbles. With this technique, temporally resolved measurements can be obtained from a scanning plane. The post-processed images are stacked in time, resulting in a three-dimensional matrix with two spatial and one temporal resolutions. The gas flow structures are given straightforward by the image stack, which provides unique insights of the bubbly flow dynamics. Sizing the bubbles requires the velocity to be known, which can be achieved by means of measurement or assumption. Typically, a second measurement plane is needed for the velocity estimation. However, the employment of dual plane measurement data is limited to dilute bubbly flows. To characterise bubbles in the dense regime, a method is presented using only single plane data for size extraction without the need of velocity data. The procedure comprises determinating the Sauter diameter (d₃₂), which depicts the ratio between volume and surface area of the bubble. Contrary to the use of dual plane measurements, here, the bubble diameter is firstly determined and from the bubble size with the assumption of the bubble's shape, the absolute bubble velocity can be roughly estimated. This method is verified with synthetic data of simple ideal-shaped bubbles. Subsequently, real bubbly flow measurements in bubble columns (with and without internals) are applied for the assessment of the resulting bubble sizes and velocity trend.

Ultrafast X-ray tomography is a recently developed imaging technique for multiphase flows. As with conventional X-ray tomography it involves reconstruction of images from X-ray projection data. If used for multiphase flow measurements, it moreover needs to be complemented with automated image processing algorithms for the extraction of flow features, such as gas holdup profiles, bubble/particle size distributions or disperse phase velocities. So far, image reconstruction is carried out with the standard filtered back-projection technique, which is fast but may not be optimal in the presence of noisy or corrupted data. As the latter is a frequent issue, search for optimal image reconstruction and data processing algorithms is continuously ongoing. This paper serves as a foundation of the image reconstruction and processing framework for the application to multiphase flow. A description is given of the procedure from reconstruction to thresholding to property extraction of ultrafast X-ray tomographic images. Two reconstruction techniques, FBP and SART are employed on the basis of phantom measurements. Each technique is evaluated separately, for FBP regarding the choice of filter and for SART regarding the termination criteria. Image reconstruction resolution, computational costs and sensitivity to the threshold value are investigated. Based on the analysis, FBP with the Ram-Lak filter is selected for image processing purposes. Furthermore, it is shown that from experiments with moving objects, there is fair agreement between measurements and the phantom dimensions. The described imaging process can be applied to different attenuation materials, simulating gas-solid and gas-liquid properties.