하이퍼루프 차량 공력 해석을 위한 적합직교분해 기법 연구

Abstract

The Proper Orthogonal Decomposition (POD) method was applied for enhancing the computational efficiency of aerodynamic simulations of hyperloop vehicle. At first, two dimensional axisymmetric computations of hyperloop vehicle were performed according to the vehicle speed and pressure inside tube in order to construct a snapshot dataset. Then, a reduced order model (ROM) was constructed through the POD method. For improvement of the accuracy of reconstructed dataset from ROM, POD basis weight coefficients were calculated by the artificial neural network. (ANN) By the comparison of original CFD data and reconstructed POD data, it was confirmed that the POD data follow the features of CFD data; the flow contours and pressure distributions of the POD data showed good agreement with CFD data. After ROM and POD basis weight coefficients by ANN are obtained, it can reconstruct the flow field data with new set of flow conditions quickly. Therefore, the POD method can be sufficiently used for the aerodynamic computations of hyperloop vehicle and ultimately design optimization problem of hyperloop system.