Experimental: particle location and interpolation read precomputed face planes instead of rebuilding each cell's planes from its vertices on every test, and the three particle fields are interpolated in one pass, roughly halving particle step time in a debug build. The planes are one per face, shared by the neighbouring cells, built at grid setup for particle runs and catalogued as PlaneNormal*, PlaneCentroid*, SearchCellSize and DualPlane*. Results agree with the per-cell planes to rounding, apart from non-planar k-faces on curvilinear grids, where neighbouring cells now share one plane.