Pillars and corner points
A corner-point grid describes the reservoir with a set of pillars — coordinate lines, often near-vertical, laid out on a logically Cartesian i, j map. Each cell hangs between four pillars and is pinned by eight corner points, the depths where its top and bottom corners meet those pillars (the COORD and ZCORN arrays in Eclipse-style decks). The result is a structured grid of distorted hexahedra: it keeps a simple i, j, k index, yet each cell can flex to follow dipping horizons and thickness changes.
How it handles faults
Faults are built by letting corner points slip along the pillars so layers are offset — the grid develops throw. Where offset juxtaposes non-adjacent layers, simulators wire them together with non-neighbor connections (NNCs). This works for moderate structure, but the topology never stops being Cartesian, and that is the crux of the problem.
Where it struggles
- Stair-stepping: a fault that cuts diagonally across the i, j grid is approximated by a zig-zag of cell faces, not the true fault plane.
- Distorted cells: cells next to faults and pinch-outs turn non-orthogonal, stretch to high aspect ratios, twist, or collapse to zero thickness.
- Lost orthogonality: those distortions break the K-orthogonality that the two-point flux scheme relies on, concentrating grid-orientation error right where the geology is most interesting.
- Complex faults: intersecting faults, Y-faults and listric geometries are awkward or impossible to represent cleanly.
Because of this, teams working heavily faulted fields increasingly move to unstructured PEBI/Voronoi grids that conform to faults directly.
How AutoMesh-Geo helps
AutoMesh-Geo skips the stair-step compromise: it builds conforming Voronoi (PEBI) grids that place faces on the real fault and horizon surfaces, so cells stay well-shaped and orthogonal. For faulted oil & gas and subsurface models, that means the grid matches the interpreted geology.