Pump by Priority
Closed-loop, AI-driven operational execution in upstream oil and gas.
For three decades, upstream operations marched away from the truck and toward the signal: manual loops to SCADA, SCADA to Exception-Based Surveillance, EBS to Pump by Exception. Each step compressed the time between event and intervention. But the curve is flattening, and the operators with the cleanest Pump-by-Exception deployments are the first to feel the ceiling. This paper formalizes the successor operating model, Pump by Priority, as a closed-loop stochastic optimization problem, and gives the eight subsystems required to run it at scale.
What is inside
- Why the signal-based model (SCADA, EBS, Pump by Exception) has hit a ceiling no further alarm tuning will lift.
- Per-asset Bayesian anomaly detection against Arps-decline priors.
- Risk-adjusted economic scoring that turns every asset into a dollar-weighted decision.
- Constraint-aware ALNS routing across crews, qualifications, and time windows.
- Contextual-bandit closed-loop learning, with a Loop Contraction theorem and regret bound.
- Robust Kalman sensor validation and multi-source observation fusion.
- Value-of-information active sensing (Oilfield Sudoku) and confidence-aware honest handoff.
- Field-validated results from a 5,000-well, three-basin reference deployment.
Field-validated outcomes
Measured at a top-25 private producer across three basins (Western Anadarko, Permian, and Wyoming): 5,000+ wells, same crews, measured against pre-deployment baselines.
Outcomes are specific to this deployment and calibration. New basins, equipment classes, or operating philosophies require a calibration period before results match these benchmarks.
From signal to priority, in every truck cab
The research is the why. WellOPS is the how: the ranked, closed-loop plan that scores every well on cash flow and risk and puts it in the cab by 6 AM.