Transition phenomena are one of the factors that determine the performance of fluid machinery, and are still difficult to predict. Many fluid mechanics researchers have attempted to explain the coherence of flow from the perspective of the stability of the governing equations. Many solutions have also been proposed. However, it can be said that stability analysis textbooks confuse beginners who want to understand and use the flow for higher performance fluid machines. This is because the actual flow is non-uniform and far more complex than what can be captured in the framework of classical stability analysis. There are many factors that can trigger the transition, such as atmospheric turbulence, surface roughness, and noise. Ideally, it is necessary to faithfully reproduce the on-site situation in order to predict transitions precisely, but there are constraints due to the cost of DNS computations and wind tunnel experiments. In recent years, with the improvement of analytical methods, attempts have been made to extend the theories developed so far by making full use of a numerical experimental method called direct stability analysis. In this presentation, I will introduce several examples in which we attempt to explain real flow fields whose transitions are thought to be controlled from the viewpoint of such new analyses.

(This talk will be given in Japanese.)