Twenty Five Years of Mesh Generation: A Personal Perspective

Since starting research in CFD the early 1980’s as a PhD Student I have worked with a variety of mesh types and topologies. The main application areas have been in aerospace, mostly in turbomachinery, but also in racing car aerodynamics and in the oil & gas process industries. The first meshes I used were simple, sheared H meshes for blade-to-blade flows; these were — and still are — perfectly adequate for this class of simulation and are in use to this day in routine design all around the world. However, most application areas are fundamentally more complex, both in terms of geometry and of flow physics, and necessitate more complex mesh systems. Current simulation trends are relentlessly towards using fully featured, fully 3D geometry and effective, high-productivity mesh generation systems have become of central significance. Over the past twenty five years I have worked with simple structured meshes, multi-block meshes, unstructured meshes based both on Delaunay and advancing front paradigms and, more recently, on octree-based cut-Cartesian meshes and associated body-conformal, hybrid meshes. This has gone hand in hand with increasing interest in computational geometry and geometry editing and parameterisation to support automated design optimisation. This paper represents my personal perspective on these experiences.