Permeon is a GPU-native reservoir simulator for the full subsurface — black-oil, geothermal & thermal, CO₂ storage, and coupled geomechanics on one engine. Run validated simulations via cloud API or deploy a direct license — up to 100× faster than traditional CPU-based simulators.
import permeon
# Initialize client with your API key
client = permeon.Client(api_key="pk_live_...")
# Submit a blackoil simulation job
job = client.simulations.create(
model="my_reservoir.DATA",
gpu_tier="h100",
timesteps=365,
)
# Poll until complete (or use webhooks)
result = job.wait()
print(result.summary)
# => { runtime_s: 4.2, cells: 300_000, converged: true }
Production-grade physics, cloud delivery, and developer-friendly APIs — built for the engineer who demands both accuracy and speed.
Our IMPES/fully-implicit solver is written ground-up for CUDA. Every linear algebra operation runs on the GPU — no CPU fallback bottlenecks.
Submit simulation jobs via REST API from any language. Integrate directly into your Python workflows, web apps, or automation pipelines.
Reads standard .DATA file format. Migrate your existing Eclipse models to Permeon with no conversion overhead.
Full three-phase blackoil: oil, gas, water. Dead/live oil, dissolved gas (Rs), wet gas (Rv), Stone I/II three-phase relperm, and WAG gas-trapping hysteresis — all validated against OPM Flow.
Matrix–fracture transfer with pressure, gravity-drainage, and capillary-imbibition drive — all fully implicit. Mass-conserving counter-current flow, validated by physics.
Analytic Fetkovich and Carter-Tracy aquifers with implicit Jacobian coupling, plus numerical (AQUNUM) aquifers. Aquifer and field pressure within ~1% of OPM Flow.
A full energy equation on the flow: multiphase enthalpy transport (water and oil), conduction, and well heat. Low-enthalpy doublets through high-enthalpy water–steam flash (IAPWS-IF97). Thermal front validated against the Ogata-Banks & Avdonin analytics and cross-checked with open-DARTS to ~2%.
Linear poroelasticity coupled to the flow: compaction drive, surface subsidence, and a cell-centered MPSA stress field. Thermo-poroelastic (THM) thermal stress for cold-reinjection. Validated against Terzaghi and Mandel analytic solutions.
CO₂–brine thermodynamics (solubility, density, viscosity) for carbon storage (CO2STORE) and CO₂-plume geothermal — fully-coupled (p,T) energy, with external thermodynamics cross-validated against open-DARTS and MOOSE.
Corner-point geometry (CPG), Cartesian, and radial grids. Handles faults, pinchouts, and non-neighboring connections.
Automatic timestep control based on convergence behavior. Minimize wasted iterations while maintaining accuracy.
Peaceman completions, multi-segment wells (segment-tree hydraulics), VFP/THP control, group controls, and WCONHIST history matching — validated vs OPM to ~1%.
EGRID/UNRST output, summary vectors (WOPR, WBHP, FOPT, etc.), customizable report intervals. VTK export for visualization.
Prefer on-premise? Deploy Permeon on your own GPU hardware with a direct license. Full simulator binary, no cloud dependency.
Run hundreds of realizations in parallel on the cloud. Ideal for ensemble-based history matching and uncertainty quantification.
Every physics capability ships with a reproducible validation report against OPM Flow.
See the validation reports →Permeon's GPU solver doesn't just accelerate existing algorithms — it rethinks the entire solve pipeline for massively parallel hardware. Every matrix factorization, every Newton iteration runs on thousands of GPU cores simultaneously.
The result: multi-million cell models that once took overnight runs now complete in minutes. History matching ensembles that required a cluster now run on a single H100.
Full Performance ReportCPU: 2× Intel Xeon Gold 6348 | GPU: NVIDIA H100
Permeon fits naturally into your existing engineering workflow. No new file formats, no lock-in — just faster answers.
Use your existing Eclipse-compatible .DATA file, or build a new model with any standard reservoir modeling tool. Permeon reads industry-standard input directly.
Submit your model through our REST API or the web dashboard. Specify GPU tier, simulation parameters, and output options in a single JSON payload.
Permeon dispatches your job to dedicated GPU hardware. The solver runs fully on-GPU — no CPU bottlenecks. Most field-scale models finish in minutes.
Retrieve EGRID/UNRST output files and summary vectors via the API. Stream results to your post-processing pipeline, Python workflow, or visualization tool.
Every benchmark is reported on accuracy versus a reference simulator and runtime across single-CPU, multi-CPU, and GPU. Measured on real runs at matched core counts.
The classic Odeh (1981) gas-injection problem — 10×10×3, 300 cells. Matches the SPE reference within ~0.5% on FOPT and FGPT.
Killough (1995) heterogeneous blackoil — 24×25×15, 9,000 cells. Mass conserved to the byte; up to 13.7× faster on CPU, ~1.2× OPM Newton parity.
The Equinor Norne field — 44,431 cells, faulted corner-point grid, 36 wells, WAG, ~9 years of history. Full-history match at ~2× OPM (matched cores).
Christie–Blunt (2001) — 1.12M cells, 8 orders of magnitude permeability contrast. Within ~1% of OPM and faster at matched MPI ranks.
Cloud API plans for teams of any size. Direct license for on-premise deployments. No hidden fees, no lock-in.
Perfect for individual engineers and research teams exploring GPU simulation.
For active reservoir teams running frequent simulations and history matching workflows.
On-premise or private cloud deployment on your own GPU hardware. Full control, no cloud dependency.
All plans include SSL-encrypted transfers, 99.9% uptime SLA, and access to the Python SDK. Have questions? Talk to us.
Join engineers who've replaced overnight CPU runs with minutes-long GPU simulations. Start with a free trial or talk to us about a direct license.