Well digital twins, flow assurance, transient analysis, material balance and reserves — one unified workspace that runs in a browser and proves its own physics on your history before it forecasts.
One platform
The workflows you license from six vendors today — decline, transient, material balance, PVT, nodal, reserves — sharing one project, one data layer, one forecast.
A physics-first twin per well — steady-state wellbore flow with a full P/T-vs-depth profile, continuously assimilating live measurements and flagging natural-flow integrity risk before it becomes deferment.
Hydrate, wax, scale and asphaltene envelopes computed from composition — not lookup tables.
Bourdet derivative, Horner build-up, deconvolution and numerical type-curves for k and skin.
Blasingame / Agarwal-Gardner RTA and a true IPR×VLP operating point — not a fixed drawdown.
Havlena-Odeh and gas p/z with drive-index diagnostics, backed by a Peng-Robinson PVT table with bubble-point and gas-Z cross-checked against thermo.
Probabilistic decline to P90/P50/P10 EUR, classified to SPE-PRMS 1P/2P/3P, straight into a DCF with NPV, IRR and payout.
A background loop watches the whole fleet, persists alerts and samples measurements to a time-series store.
Every model backtests on your history — precision, recall and lead-time — then re-trains on drift. No unvalidated proxies.
Assemble a black-oil Eclipse deck, run it, and carry uncertainty through to probabilistic forecasts.
The workflow
Import well headers, production history and logs from CSV, historian or an OSDU source. Everything lands in one versioned, provenance-tracked project.
Run decline, transient, material-balance, PVT and nodal studies side by side — each reading the same shared field and well context.
Backtest each model against the real history. The platform reports lead-time and error, and only trusts what verifies.
Carry the validated models into probabilistic EUR, PRMS reserves and a full DCF — with the assumptions cited back to source.
Why ReservoirAI
Desktop suites are powerful and fragmented. We traded none of the rigor and dropped the install, the per-module licensing and the black-box proxy.
Validated physics
The engine is built from published methods and checked against lab and synthetic ground truth — not fitted black boxes.
Open-core pricing
The full engine is open-source and free to self-host. Pay only when you want it hosted, supported and multi-tenant.
Request a demo
Tell us about your field and the workflows you care about. We'll walk you through the platform on data like yours — and show the validation step that sets it apart.
Import a history, run a study, and watch the platform prove its own forecast before you trust it. No install required.