论文部分内容阅读
Recent efforts of multi-objective wing shape optimization of large civil transport aircrafts in wing-body-pylon-nacelle (WBNP) configuration, where the interferences between wing and pylon/na celle are taken into account directly, is presented. Massive parallel computation and model fidelity relaxation are adopted to reduce cycle time. An automated optimization framework consisting of CST-based wing parameterization, mesh deformation, flow solving, post processing and genetic-al gorithm-based global optimization is integrated, which allows hundreds of configurations to be ev aluated at the same time on super computer. Using a multi-block Reynolds Averaged Navier-Stoke s (RANS) flow solver and a mesh of 8 million cells, solution can be obtained within 20 minutes f or a WBNP configuration with the help of appropriate fidelity relaxation. In a typical 3-objective optimization application, Computational Fluid Dynamics evaluation of 128 candidates, each with 162 design variables for wing shape, was carried simultaneously on Tianhe-2, the worlds most po werful computer. It took 100 hours to complete more than 50 generations evolution, 3 to 5 count s drag reduction for each objective can be achieved comparing with the baseline case.