A POD-DMD augmented procedure to isolating dominant flow field features in a street canyon

Abstract

This work develops a data analysis procedure, namely, proper orthogonal decomposition (POD)-dynamic mode decomposition (DMD) augmented analysis, to isolate the energy- and evolution-wise dominant features of flow field in a street canyon. This combination aims to extract modes imposing critical influence on pollutant dispersion from both energetic and dynamic perspectives. The two techniques were first conducted based on large-eddy simulation results. Subsequently, based on the POD and DMD ranking, the extracted modes were classified into three types: (1) type 1: energetically and dynamically significant mode; (2) type 2: energetically significant and dynamically insignificant mode; and (3) type 3: energetically insignificant and dynamically significant mode. Results show that mode type 1 contributes to the mainstream flow and the main vortex structures, which can be observed near the stagnation point, the separating point, and the fluid reattachment area. Mode type 2 throws light on where the turbulent kinetic energy is the largest, leading to periodically sudden pollutants increase on the building roof and the wake region. Mode type 3 contributes to the long-term reversed flow structures occurring near the stagnation point, inside the street canyon, and in the wake region. This technique can provide a systematic analysis of the flow field within a street canyon, and it also provides help for potential applications at a city scale, such as solving pollutant dispersion issues in urban areas.