In this study, a numerical analysis on bi-directional impulse turbine with spiral flow collector for tidal power system is conducted. The spiral flow collector can improve the turbine performance such as an increase in velocity ratio and a decrease in axial force coefficient. Numerical calculations show that a flow collector with smaller separation region is desired, although it is advantageous to have a large collector opening angle in order to enhance the collecting effect. Numerical calculations also show that the relative flow angle at rotor inlet for spiral flow collector is so small that the predicted full-scale power is smaller than the case of the flow collector without spiral vanes.

Blade shape optimization of marine propellers is generally formulated as an optimization problem to maximize efficiency in uniform flow. For practical design, it is essential to optimize not only propeller efficiency, but also cavitation performance and thrust difference from the target required for propulsion. In addition, because propeller blade geometry is complex, the design variables used to describe the geometry of it are high-dimensional. The search for a reasonable solution is time-consuming. In this study, the blade shape optimization problem of marine propellers is formulated as a multi-objective optimization problem. The optimal propeller blade shape is searched for using multi-objective JADE which is a Differential Evolution algorithm with high search performance. In numerical simulations, authors have shown that the developed method can search for a better blade shape compared to the original propeller and verify the validity of the blade shape obtained in model experiment.

Development of silent products is one of the most important requirements in recent life-styles. Therefore, the development of silent fans would contribute to reduce the noise of products. The fans have been mainly developed by using experimental parameters. However, it is difficult to develop radically the silent fans by using the traditional methods. High performance computing are powerful tools for solving these needs. In this study, work flow system to achieve design optimization with objective functions of static pressure rise and noise by using supercomputer Fugaku was developed. The developed work flow system could run process of blade design based on design variables, grid generation, calculation of the static pressure rise and the noise, and determination of next new design variables by genetic algorism. Then, selected optimized blades from results of 411 cases in 20 generations were types of aft loading. The fans with their blades had performance of high static pressure and low noise.

In a cross-flow turbine downstream side in the flow channel between the lower nozzle wall and the guide vane is a divergent nozzle under small guide vane opening condition. In a divergent nozzle, energy loss happens due to both the occurrence of flow separation and an increase in turbulence in the case of a large expanding ratio. In addition, the turbine cannot be applied in a high head because an increase in the flow velocity causes cavitation at the narrow section in the divergent nozzle. Therefore, the authors propose an energy loss suppression method using air sucked naturally through a hole drilled on the wall upstream of the divergent nozzle. This study confirmed by using CFD that the air sucked causes water flow separation from the wall, thereby reducing the ratio of water power loss to the power at the inlet, and pressure drop decreasing at narrow section. Also, it was confirmed that the total pressure in the jet downstream of the nozzle almost corresponds to the results obtained using the experiment.