RECOGNITION OF HIGHLY OVERLAPPING BUBBLES IN THE WAKE OF A VENTILATED HYDROFOIL

SUMMARY A study is being carried out at Saint Anthony Falls Laboratory (SAFL) of the University of Minnesota to develop the necessary algorithms to determine the velocity deficit and bubble size distribution in the bubbly wake behind a ventilated or cavitating hydrofoil. This is done by utilizing background illumination of the bubbly wake. A simplified experiment, to validate the proposed bubble detection algorithm, was carried out in a fish tank. Compressed air was injected into a porous media, located in the bottom of the tank, which broke up the air into bubbles that rose to the surface. Shadows of the bubbles, cast on a bright background, were collected by a high-speed camera. For processing, a reference image, taken when no bubbles were present in the flow, was subtracted from the images to reduce background noise. Once this was done, the image was segmented using an edge detection technique. The Canny algorithm was determined to be the best suited. Applying the proper thresholds ensured that only bubbles in the plane of interest were considered during analysis. Once bubble edges were detected, a curvature profile method was employed to distinguish individual bubbles within a cluster of highly overlapping bubbles. The utilized algorithm was made to detect partly overlapping bubbles and reconstruct the missing parts. The movement of recognized individual bubbles was tracked on a two dimensional plane within a flow volume. In order to obtain quantitative results, the wake of a ventilated NACA0015 hydrofoil was investigated by applying a shadowgraphy technique and the described bubble detection algorithm. There experiments were conducted in the SAFL high speed cavitation tunnel.