Full-field reservoir simulation with fractal multigrid V-cycle solver. O(N) convergence. Lock-free parallel. No GPU required.
Industry-standard simulators (Eclipse, CMG, tNavigator) use flat-grid iterative solvers with O(N²) or O(N¹∙³) complexity. The fractal multigrid V-cycle converges in O(N) — the same physics, fundamentally different math.
| Metric | Industry Standard | ZYZ |
|---|---|---|
| Solver | Conjugate Gradient / GMRES | Fractal Multigrid V-Cycle |
| Complexity | O(N1.3) | O(N) |
| Iterations (1M cells) | 350+ | 20 |
| Parallel Scaling | 60-70% efficiency | 95%+ (lock-free) |
| GPU Required | Yes (tNavigator) | No |
| Scaling Behavior | Degrades with size | Improves with size |
| 10M cell model | 12 hours | 3 minutes |
Multigrid V-cycles solve at every scale simultaneously. The coarse grid corrects the fine grid. Convergence is independent of problem size.
Lock-free architecture means no mutex contention. Every core works at full speed. Efficiency improves as you add cores.
No PCIe bottleneck. No kernel launch overhead. No GPU memory limits. Runs on commodity hardware at GPU-beating speeds.
Run thousands of scenarios in the time it takes competitors to run one. Find optimal well configurations that extract 2-5% more recoverable oil from every field.
Calibrate reservoir models against production data interactively. What takes days of compute becomes a conversation with the model.
Monte Carlo over geological uncertainty — hundreds of full-field runs per hour instead of per week. Know your risk before you drill.
2% improvement in US recovery factor = 13 billion additional barrels. At $70/bbl that's $910B in value. The math isn't incremental — it's civilizational.
Fractal multigrid solver with lock-free parallel V-cycle and LoNC well placement optimization. Black-oil, compositional, EOR, CO₂ storage, unconventional, digital twin. Part of the Origin 22 portfolio: 91 provisional patents, 25,000+ claims.