A new visualization technique was developed for a three-dimensional numerical flow simulation. This method is a kind of tracer visualizations, and based on a principle where we can actually perceive a fluid motion through smoke. In this method, several millions of transformable surface elements are released into a computational domain instead of particle tracers, and lighting and transparency effects of each surface element are considered in the visualization. By means of this visualization method, we managed to give three-dimensional effects to tracing visualization methods. In this paper, we tried to visualize some numerical flows using this method. Numerical simulations were carried out by DNS, and a few representative flow phenomena were employed. In particular, we could observe existence of complicated three-dimensional flow structures and turbulence transitions through natural convection and channel turbulence problems. As a result, we confirmed that this visualization method is quite effective for understanding three-dimensional flow structures.

View full abstract

Two-fluid and three-fluid models are used for safety analysis and design cords of light water reactors (LWRs). Interfacial area concentration, interfacial heat transfer and interfacial shear stress have never been understood enough for these numerical codes. Since it is very difficult to measure these phenomena, there is a shortage of the correct experimental data. In the present work, a digital image processing method, consisted of a digital video camera, a stroboscope, a personal computer and data processing software, was developed and applied to subcooled boiling flows. As a result, bubble condensation characteristics under a subcooled boiling condition is clarified.

View full abstract

A parallel computation technique has been proposed for the mixing and segregation of granular mixture with multi-block grid system based on 3D body-fitted coordinates. Unsteady particle motions in the multiple blocks are calculated simultaneously with the distinct element method (DEM). Since a graphic process is also treated as one of the parallel jobs, the particle distributions can be visualized during the computations. The computational method was applied to the granular mixture in horizontal and inclined rotating cylinders. As a result, it was confirmed that the computational time is decreased by the parallel computation and that the mixing and segregation of the particles are successfully predicted in all cases. The characteristic patterns of the predicted particle motions are visualized with the particle pathlines.

View full abstract