Reservoir AI on industrial 3D datasets.
Flow Arena ranks ML surrogates — fast approximations of reservoir simulators — across controlled geology, fault complexity, and well-control regimes. Public academic benchmarks calibrate the methodology; private Flow Arena datasets support extensive evaluation under realistic reservoir conditions.
From public calibration to private hosted evaluation.
Public academic datasets keep the metric definitions, model implementations, and reporting format inspectable.
Private reservoir datasets measure robustness across geology, faults, controls, pressure, and saturation.
Hosted evaluation, model submission, detailed reports, and controlled data access can be handled under agreement.
Flow Arena and academic references.
Flow Arena
Proprietary industrial two-phase flow benchmark spanning property type, fault configuration, and well control.
Academic Benchmarks
Published academic datasets used to calibrate metrics, reproduce paper comparisons, and document methodology.
Training data from OPM Flow, presented safely.
Representative pressure and water-saturation fields show the physical targets behind the public metrics. The raw tensors and complete realizations remain private; the portal publishes controlled summaries, static private-safe views, and aggregate results.
