This paper presents a hybrid method based on proper orthogonal decomposition (POD) with a trained radial basis function (RBF) network, on direct simulation monte carlo (DSMC) solutions for aerothermodynamic front surface optimization of Stardust reentry. Gaussian and multiquadric RBFs are implemented for comparison, and multiquadric functions are chosen due to their insensitivity to diverse shape parameters. Cubic uniform B-spline curves are used innovatively for parameterization of the geometry change, instead of curve fitting the geometry itself. This makes possible to reduce the number of design variables. Gradient based optimization strategy is implemented by regarding the distributions of pressure, shear stress and heat flux along the surface of the geometries. G.A. Bird's two dimensional axisymmetric DSMC solver [1] is used as the physics solver, and 11 species air model are chosen with 41 chemical reactions according to atmospheric conditions of the reentry. Different geometries are obtained via deviating the design variables arbitrarily to form a snapshot pool. In this manner, the approximation success of the POD-RBF methodology is tested on highly nonlinear flow conditions with arbitrarily chosen design of experiment. Finally, the optimized geometries are simulated via DSMC code and the solutions are compared with the solutions of POD-RBF network. Method lowered the optimization time extraordinarily and provided satisfactory results.
展开▼
