Videometry is a useful method to investigate the behaviour of bubbles, particles, and phase boundaries in two-phase flows in (partially) transparent vessels and flow channels. Thereby video sequences of the flow are recorded by means of a fast video camera and subsequently processed by proper image processing and analysis algorithms to extract physical flow parameters, such as geometry and deformation of the phase boundary, velocity fields, bubble size distributions, and coalescence and break-up rates for gas bubbles. By help of an ultra-fast high-speed video camera even fast flows can be analyzed. There are also some degrees of freedom concerning the imaging geometry (mono and stereo imaging, usage of mirrors to image multiple projections) and the illumination scheme (transillumination, reflection, light sheet techniques). Problems occur in flows of liquid-gas type at high gas fractions when multiple reflections and refractions decrease the transparency of the medium. Then phase boundaries being behind one another are no longer discernible in the single images. In order to securely identify single bubbles or parameterize phase boundaries intelligent image processing and analysis methods need to be applied that use all information that is encoded in the sequence of video images and also the information that is know a-priori. Within the department we have two high-speed video imaging systems which are used for investigations on two-phase flows in horizontal and vertical test sections of flow loops. Further, we develop and test image processing algorithms for tracking of bubbles and determination of two-phase flow parameters.
The department currently has two high-speed video cameras with the follwing parameters:
|camera||Redlake Motion Pro||Kodak Motion Corder Analyzer SR-Ultra|
|resolution||1280 x 1024||256 x 240 at 1000 fps|
|frame rate||up to 10 kHz||up to 10 kHz|
|operation mode||controlled by PC (frame rate, exposure time)||image acquisition with recorder
transfer of image data via SCSI interface to PC
The right image shows exemplarily the application of the RedLake video camera at a vertical bubble column. The illumination of the scene is realized by a white screen behind the column that diffusively scatters the light of two 1000 W bulbs. The gas bubbles are recorded in transillumination mode and appear as dark objects with a brighter central area in the images.
After recording of an image sequence the raw images are processed in the computer as required by the application. This includes processing steps such as
The identification of single bubbles is realized by an iterative filling algorithm. This algorithm seeks for isolated bubble clusters, i. e., sets of coherent pixels. The identification of single bubbles within a cluster that might be composed of multiple bubbles is a subsequent step of intelligent image processing. For this purpose information on the central white spot, edge information, and information from changes in the cluster shape within subsequent images of a sequence is formalized and processed. Other major points of interest are the determination of local velocity and dispersion profiles, and coalescence and break-up rates. Therefore we developed special algorithms to extract and analyze bubble trajectories.
With videometry a large number of flow problems can be efficiently investigated. This includes
|bubble trajectory analysis||interaction between bubbles and the wire-mesh sensor (left: lateral observation with high speed camera, right: sensor data visualisation|
(AVI-Video; 0.7 MB alternatively: Indeo 0.9 MB)
analysis of flow patterns in a horizontal channel
- Experimental and Numerical Studies of Flow in Rectangular Bubble Columns.
The 10th International Topical Meeting on Nuclear Reactor Thermal Hydraulics (NURETH-10)
Seoul, Korea, October 5-9, 2003.
Reddy Vanga, B. N.; Krepper, E.; Prasser, H.-M.; Zaruba, A.; Lopez de Bertodano, M. A.