非结构混合网格鲁棒自适应技术

Robust adaptation techniques for unstructured hybrid mesh

数值格式、湍流模型和计算网格是影响CFD数值模拟精度的3个主要因素。结合流场信息的网格自适应技术具备动态优化计算网格的能力,被NASA列为未来CFD发展的一项关键技术。本文针对非结构混合网格,发展了网格单元分布优化、表面网格几何投影和空间网格协调匹配3项关键技术,建立了高鲁棒性几何保真的网格自适应系统。首先,为了提高自适应方法的鲁棒性和通用性,发展了基于标准面网格的多面体网格单元分布优化方法。其次,发展了仅依赖表面网格信息的局部曲面重构技术,采用参数点映射方法实现了新增表面网格点的几何投影,消除了自适应系统对几何CAD系统的依赖。再次,采用改进的距离函数方法实现了空间网格与投影后表面网格的快速匹配。最后,结合基于流场特征的自适应探测器,采用二阶格式的有限体积方法,开展了30P30N三段翼绕流和三角翼大迎角绕流的网格自适应数值模拟。结果表明,通过网格自适应对网格单元的分布进行优化后,流场求解的收敛性和模拟精度都得到了显著提高。

Numerical schemes,turbulence models,and computational mesh are the three main factors that affect the precision of CFD simulation.Mesh adaptation based on the information of flow field has the ability to dynamically optimize the mesh,which is considered by NASA as a vital technique for the development of CFD in the future.Aimed at the unstructured hybrid mesh,three critical techniques,including the distribution optimization of mesh elements,surface mesh projection onto geometry,and volume mesh conformal deformation,are proposed to establish a robust adaptation system with high geometry fidelity.First,the distribution optimization technique based on polyhedral elements with standard faces are developed to improve the robustness and the universality of the adaptation system.Next,the local surface reconstruction method which relies only on the points of surface mesh is developed.The new inserted surface points can be parametrically projected onto the geometry,eliminating the dependency of mesh adaptation on the CAD system.Then,the improved distance function method is adopted to rapidly match the volume mesh elements to the projected surface mesh.Finally,combined with the mesh adaptation based on the flow features detector,the flow simulation with the second order finite volume method is carried out for the 30P30N airfoil and the delta-wing with a large angle of attack.The numerical results show that,after the mesh optimization with mesh adaptation proposed in this paper,both the convergence performance and precision of the flow simulation are significantly improved.