HYBRID · PRODUCTION FORECASTING AND OPTIMISATION
Forecasts that
keep the decline physics.
Hybrid puts decline curves, time series, machine learning and a hybrid data-physics model in one workspace for wells, fields and operators. Every method is scored on months it did not see, so you know which forecast to take to the economics.
From public data to drilling decisions
ONE WORKFLOW
THE WORKSPACE
Wells, production, completions, prices and operators synced from public regulator and vendor data, with a full report for any well.
– Maps, type curves, benchmarking and outlier detection
– Forecast any well, field or operator four ways
– Well design with Monte Carlo, development plans and sweet-spot maps
Acquire, explore, forecast
WHY PHYSICS INSIDE THE NETWORK
A neural network learns each well's decline parameters; a physical decline model turns them into rates. The last 15% of months were hidden from every model and used only for scoring.
The data runs out, the physics carries on
– Hybrid: 26.6 test RMSE (e³m³/month), Giroux Lake gas
– Random forest: 147.4 on the same held-back months
BETTER TOGETHER
One forecast drives production, risk and chemical spend
→
SmartFlow, date by date
Rates and water, well by well, through field life
Hybrid forecast
All five threats at every pipe and vessel, for every date
→
Chemicals and cleaning
When each threat peaks, and what it will cost
Every forecast valued, and every weak one flagged
AN HONEST LEADERBOARD
SCORED
Same test, same months
Only runs scored the same way are ranked together. Others are listed as not comparable.
FLAGGED
Flags what it can't trust
A forecast that runs flat, starts high or starts low is marked, because its volumes and NPV would be overstated.
VALUED
From forecast to value
Enter prices, royalty, opex and capital once. Every run gets NPV, IRR, payout and breakeven.