Paper Details

Name: Kapil Sharma, Lalit Chhaunkar, Shriyansh
Country: India

Year: 2026
Volume: Volume-13, Issue-1 (January-June)
Page Number: 450-457
DOI: https://doi.org/10.5281/zenodo.21217842

Mesh generation remains one of the most decisive stages in computational analysis
because the quality of the discretized domain directly influences numerical stability,
accuracy, and computational cost. In thermal and fluid engineering, the challenge is not simply to create cells that fill a geometry, but to generate a mesh that captures sharp gradients, wall-bounded behavior, curved boundaries, and localized physics without making the model prohibitively expensive. This review reorganizes the subject in a clear progression from mesh classification to quality assessment, core generation algorithms, adaptive strategies, and emerging data-driven workflows. Structured, unstructured, hybrid, and polyhedral meshes are compared in terms of flexibility, solver efficiency, and suitability for realistic engineering geometries. The discussion then links commonly used quality metrics such as skewness, orthogonality, aspect ratio, Jacobian quality, and growth rate to actual solver behavior in CFD and FEA. Classical methods including algebraic mapping, PDE-based grid generation, advancing front meshing, Delaunay refinement, and octree approaches are reviewed with emphasis on their practical strengths and limitations. The paper further examines adaptive meshing, boundary-layer treatment, immersed and overset formulations, isogeometric concepts, and the early influence of artificial intelligence on mesh decision-making. The overall argument is that robust meshing is best understood as a balance among geometry, physics, and algorithmic control rather than as a purely automated preprocessing task.