Numerical Image Processing in Centrifuge Testing

1998 
Assessing the performance of body forces accurately in the field of geotechnical model testing requires the use of centrifugation. Models are tested under accelerations 100 to 200 times the early gravity. These models are usually instrumented with transducers to measure displacements, strains, pressures, forces or accelerations in the small-scale structures. When displacements within the soil samples are to be investigated through the transparent wall of the containers, new techniques must be used. The present paper describes the numerical image processing method developed at the Laboratoire Central des Ponts et Chaussees and illustrated with a few examples of the application. A CCD camera is placed close to the model in the swinging basket of the centrifuge. The lenses are chosen according to the desired size of the shot and to the accuracy required. A special device has been designed to solidly fasten both the camera and the lamps in the right direction. The video signals are first transmitted through slip rings to a video recorder and then directly to a computer if real time processing is required. A piece of software has been specially developed for automatically tracking particular spots or lines, which were previously in the model. To enhance the accuracy of displacement measurement, the piece of software also positions the geometrical and optical axis of the camera automatically and corrects the distortions due to the optical imperfections of the transparent wall of the container. The mean accuracy is then 1/5000th the size of the observed scope. This technique is developed and used not only for centrifuge model testing (cyclic loading of foundations, buckling of piles, stability of tunnels, soil reinforcement, flows within hoppers and silos), but also for site investigations (boat berthing, study of the swell).
    • Correction
    • Source
    • Cite
    • Save
    • Machine Reading By IdeaReader
    0
    References
    1
    Citations
    NaN
    KQI
    []