Compared to triangular or mixed meshes, pure quadrilateral meshes generally offer higher accuracy and computational efficiency in numerical simulations. Recently, the main obstacles in the generation of quadrilateral meshes have revolved around the identification of mesh singularities and the subsequent decomposition of the domain. In this study, a novel quadrilateral mesh generation method based on domain decomposition and templates is presented. The basic idea is to decompose the domain by separating the mesh singularities rather than identifying their precise locations. Surfaces composed of a set of surface patches are meshed independently. For each surface patch, the problem domain is divided into several simple polygonal subregions, including quadrangles, triangles, and pentagons, using constrained Delaunay triangulation (CDT) and recursive bisections. Subsequently, mesh templates with a limited number of singularities are employed to rapidly generate high-quality quadrilateral meshes over those subregions. By constraining the boundary discretization of each surface patch, mesh conformability is automatically achieved. An integer linear programming equation is presented to ensure that the total number of boundary nodes on each surface is an even value. Several experimental results are presented and discussed, demonstrating the reliability and efficiency of the proposed method.

与三角形或混合网格相比，纯四边形网格通常在数值模拟中提供更高的精度和计算效率。最近，四边形网格生成的主要障碍围绕着网格奇点的识别和随后的域分解。在本研究中，提出了一种基于域分解和模板的新型四边形网格生成方法。基本思想是通过分离网格奇点来分解域，而不是识别它们的精确位置。由一组曲面片组成的曲面独立地进行网格划分。对于每个曲面片，使用约束 Delaunay 三角剖分 (CDT) 和递归二分法将问题域分为几个简单的多边形子区域，包括四边形、三角形和五边形。随后，采用具有有限数量奇点的网格模板在这些子区域上快速生成高质量的四边形网格。通过约束每个曲面片的边界离散化，自动实现网格一致性。提出了整数线性规划方程，以确保每个表面上的边界节点总数为偶数。提出并讨论了几个实验结果，证明了所提出方法的可靠性和效率。