No abstract

The velocity field around a zigzagging ascent single bubble has a wide dynamic range. It is hard to precisely measure such velocity field via conventional PIV. In this study, to overcome this difficulty the authors employed recursive cross-correlation algorithm and a high-speed video camera. The analysis code of a newly developed PIV system showed a high performance in measuring velocities of a wide dynamic range. At first, the authors demonstrate the performance of the recursive cross-correlation PIV algorithm as compared with the results obtained by conventional PIV algorithm using the standard PIV images with a wide dynamic range presented by The Visualization Society of Japan. At second, the system was applied to the simultaneous measurement of the surrounding liquid motion (i.e. liquid-phase velocities) and surface motion of a single bubble zigzagging upward in purified water and contaminated water. At third, surfactant effects on the surrounding liquid motion (i.e. intensity of the vorticity, size of the vorticity and the structure of a hairpin-like vortex) were discussed as well as those on centre-of-gravity and surface motions of the bubble. In this study, a very small amount of surfactant (1-pentanol) solutions were examined. Although the bulk surface tension was almost the same as that of purified water, the bubble motion and the surrounding liquid motion in the solutions (i.e. contaminated system) were very different from those in purified water (i.e. purified system). A critical concentration of a surfactant was found out. At the critical concentration, the intensity and size of the vorticity become the largest. From the consideration of coupling the asymmetric surface motion of the bubble and the surrounding liquid motion, it was found out that the Marangoni stress owing to a concentration gradient of the adsorbed surfactant on the bubble surface plays a great role in both bubble and surrounding liquid motion.

No abstract

The interaction between bubble motion and its surrounding liquid motion through a collision of a pair of zigzagging bubbles in a rest water column was experimentally investigated. A pair of hypodermic needles and a bubble generator utilizing pressure oscillation were employed in order to exactly extract and highly reproduce the interaction between liquid-phase motion and bubble motion. The recursive cross-correlation PIV technique made it possible to obtain the accurate velocity field of surrounding liquid motion around a pair of bubbles. The vorticity field around a pair of bubbles was calculated from the results of velocity field. First, various kinds of interactions (e.g. bouncing and coalescence) were found out from the results of bubble motion (e.g. velocity) after a collision. We classified such interactions of a pair of bubbles using the dimensionless number l/d. Second, we investigated surrounding liquid motion about the two cases of interactions after the bouncing. One is that only horizontal velocity of bubble decreased after the collision, the other is that both horizontal and vertical velocity decreased. At the former case, the vortical region generated at the rear of a pair of bubbles. Therefore as the reason for the decrease of bubble velocity, the shedding area of hairpin-like vortex is restricted. At the latter case, bubble wake flows into the area between bubbles and vortical region doesn't generate under bubbles after the collision. Hence, bubble wake overtakes a pair of bubbles after the collision, then, bubble velocity decreased.

Two main problems are associated with conventional numerical methods for simulating turbulent flows in high-reaction-type supersonic turbine cascades near the tip of the last stage blade in a steam turbine: the large skewness of computational grids and treatments of boundary conditions when the shock waves hit boundaries. This paper presents a numerical method to deal with these issues. A grid generation technique which uses five-block structured grids has been developed. The orthogonality of the grid is good even for highly staggered and low solidity cascades. In addition, the grids are completely continuous at the boundary between the blocks and at the periodic boundaries. Both the gradient of the grid lines and the change rate of the grid widths connected smoothly. As a result, shock waves can be captured accurately and stably. The inflow and outflow boundary conditions based on the two-dimensional characteristic theory have been applied and diminished the spurious reflections and fluctuations of shock waves at both the inlet and outlet boundaries. Therefore the non-physical reflection does not affect the flow in the cascades. A low Reynolds number k-ε turbulent model has been proposed. Distance from a wall is not used as the characteristic length of turbulent flows so that the turbulent model can be applied to a wake and a separation flow. The validity of the numerical method was verified by comparisons of the pressure distributions on the blade, the loss coefficients, and flow angles with linear cascade experiments of transonic compressor cascades.