Extracting Physical Mechanisms from High Fidelity Simulations

Currently, the understanding and characterisation of unsteady flowfields relies on statistical quantities such as turbulent kinetic energy. Even today, unsteady flows are viewed from a fundamentally steady or time-averaged point of view, with no existing framework to switch to a truly unsteady perspective. However, such averaging can greatly change the appearance of the flowfield. In this paper, a new method is demonstrated by which unsteady flow features can be analysed and understood. Utilising scale-resolving high fidelity simulations of an unshrouded mixed flow impeller, a structure identification method is introduced along with a new formulation for the transport of energy of flow fluctuations. Together, these are used to determine loss generation by different types of flow structures. The method is shown to enable the analysis of flow from a truly unsteady, structure-by-structure perspective, thereby maximising use out of a high fidelity CFD dataset. It is also demonstrated to enable the interrogation of individual unsteady structures, and quantification of their impact on overall machine performance.