Spatiotemporal reduced-order model of supersonic exhaust plume based on POD
Large eddy simulation(LES)was used to compute the three-dimensional unsteady plume flow field.Low-frequency and high-energy large-scale coherent structures in the flow field were extracted by using low-pass filters.Spatial reduction was achieved by Fourier transform and proper orthogonal decomposition(POD)in the azimuthal direction of the plume,while temporal reduction was achieved by extracting the Fourier mode from the leading temporal POD mode.Consequently,a spatiotemporal reduced-order model(ROM)for supersonic plume reaction turbulence was established.The results demonstrated that low-pass filters and POD truncation can successfully filter high-frequency and low-energy small-scale structures in the supersonic plume.The first two azimuthal modes accounted for 80.9％of the energy in the plume.In the dominant azimuthal mode,the pressure POD spatial mode showed sharp peaks in the core region of the jet due to the interaction between compression wave and shock wave.Furthermore,POD spatial modes of the component and temperature exhibited sharp disturbances downstream because of afterburning.An alternating wave packet structure with a steady wavelength was displayed in the POD spatial mode under the second-order azimuthal mode.The POD spatial mode of both components and temperature showed a similar wave packet structure.In addition to reducing numerical instability,temporal reduction based on Fourier mode energy selection guaranteed excellent reconstruction accuracy.This research can support target intelligent feature engineering applications by offering theoretical techniques for evolution laws and feature extraction on supersonic rocket exhaust plumes.
NETL
NSTL
万方数据
