Reconstruction and Prediction of Multiparametric Steady Flow Field Based on POD

The technology for fast flow field reconstruction and prediction with high accuracy and efficiency is more attractive than computational fluid dynamics (CFD) from the perspective of time and computing consumption. The proper orthogonal decomposition (POD) technology has shown its capacity to reconstruct and predict various physical fields. The present study proposes a method based on POD for the reconstruction and prediction of the multiparametric flow field using a uniform POD mode set. In the traditional POD procedure, a few POD modes with maximal eigenvalues (ME) are selected to represent the original system so that the global system error is minimized. To improve the accuracy of reconstruction or prediction of the flow field under a certain condition, the POD mode selection method is modified. It’s proposed to select the POD modes with the maximum magnitudes of POD coefficient (MC) to reconstruct or predict the flow field instead of the POD modes with the maximum eigenvalues. To validate the feasibility of this POD modes selection method, the velocity and pressure distributions around a 2-dimensional aerofoil NACA0012 are adopted as the steady flow field to be reconstructed and predicted. Flight speed and angle of attack of the aerofoil are used as two independent parameters that determine the flow field. The Lagrange multivariate function interpolation is applied in this paper to calculate the POD coefficients in terms of any combination of both independent parameters. Compared with the traditional ME, a significant accuracy improvement is obtained by MC not only in the reconstruction of an individual flow field included in the original data set but also in the prediction of the flow field with a specified condition.