White papers
Field-grade technical research on running an oil and gas operation on intent. Written for COOs, VPs of Operations, field superintendents, and the engineers who have to make the math real.
Pump by Priority
Closed-loop, AI-driven operational execution in upstream oil and gas
The successor to Pump by Exception. The signal-based model has hit a ceiling no amount of SCADA or alarm tuning will lift. This paper formalizes Pump by Priority as a closed-loop stochastic optimization problem: per-asset Bayesian anomaly detection against Arps-decline priors, risk-adjusted economic scoring, constraint-aware ALNS routing, and closed-loop learning. Field-validated at a 5,000-well, three-basin deployment: 15% more free cash flow on the same crew.
Optimal Route Planning in Oil and Gas
Multi-class field work and constraint-aware crew dispatch at 125+ assets per operator
Why morning crew dispatch is genuinely intractable by hand, and how a system built for it cuts miles driven by 35% while improving free cash flow 15% on the same crew. Includes the formal MC-VRPTW-S model, the activity-based FP&A scoring substrate, the ALNS solver, and field-validated outcomes from a 5,000-well, three-basin deployment.
Taking Pump by Exception to the Next Level
A practical playbook for evolving exception-based operations into prioritized, economically ranked daily plans
Pump by exception was the right idea with a ceiling: alarm overload, equal-looking exceptions, thresholds that drift off declining wells. This is the operator-level playbook for the next level: prioritized exceptions, economic scoring of alarms, and the ranked daily plan in the cab by 6 AM, plus the operating-model changes (morning meeting, foreman role, KPI swap) and a 90-day path that make it stick. The practical companion to the Volume I Pump by Priority research paper.
How One Operator Automated Their Hydraulic Model Builds
From P&IDs and alignment sheets to a living model: the Golden Record, versioned studies, and simulation-file management
The anonymized story of a midstream operator that stopped rebuilding hydraulic models and started maintaining one. Models built from the P&IDs, alignment sheets, and GIS they already had; a versioned Golden Record as the single substrate; studies under version control; managed simulation artifacts; and continuous reconciliation between systems. Model builds went from weeks of manual effort to minutes of verification across 2,000+ miles of pipeline modeled.
Are your best people working on your most valuable work today?
The research is the why. Your data is the proof.
Every result in these papers was measured on a live deployment. The fastest way to test the math is to run it against your own field.