Mario Campos and Jess Tarrant - Amplified Industries
Wes Nolen and Bradley Lennert - SNW Operating
Objectives/Scope
Mature sucker-rod-pumped assets often operate with fixed timer schedules and limited visibility into changing inflow conditions. This study evaluates a high-frequency, AI-enabled pump optimization approach designed to continuously align pumping activity with reservoir inflow, improve pump fillage, reduce unnecessary strokes and mechanical stress, and create a repeatable operating model that can be scaled across a field.
Methods, Procedures, Process
A ten-well pilot was implemented using next-generation microcontrollers that combined edge and cloud computing with AI/ML-driven dynamic control. High-frequency load, position, runtime and pump-fillage data were continuously processed to autonomously evaluate well behavior and dynamically adjust operating cycles, allowing each well to adapt to changing reservoir inflow without relying on fixed timer schedules or routine manual setpoint changes. Before implementation, the wells showed variable behavior, incomplete and inconsistent pump fillage, sub-optimal pump efficiency, and unstable cycles. Following implementation on October 23-24, dynamometer-card behavior and pump-fillage trends were reviewed across all nine wells to assess response consistency and repeatability.
Results, Observations, Conclusions
All ten pilot wells demonstrated a consistent post-implementation response, including more stable operating behavior, improved and more complete pump fillage, sustained high fillage with lower cycle-to-cycle variability, and pump operation more closely aligned with reservoir inflow. The project economic model assumed an 80% power-runtime reduction, a 5% production uplift attributable to stress reduction, and a 50% reduction in production failure rate. Under the model assumptions, the ten-well pilot generated approximately $40,000 of incremental cumulative cash flow over the first three months, corresponding to a modeled ROI of approximately 230%, with first-month payout. The twelve-month projection showed approximately $200,000 of incremental cumulative cash flow and a projected ROI greater than 800%.
Based on the pilot, the expected economics for a 131-well expansion were established by normalizing the nine-well results and applying them proportionally across the larger population. This pre-deployment extrapolation indicated approximately $0.58 million of expected incremental cumulative cash flow over three months and $2.91 million over twelve months, assuming comparable well performance, costs, production response, and operating conditions. The technology was subsequently deployed across the full 131-well field. This paper presents the operational and economic results of that full-field implementation and evaluates how the realized field response compared with the expectations established from the pilot.
Novel/Additive Information
The study demonstrates a practical pathway from a controlled multi-well pilot to full-field autonomous AI/ML-driven optimization using next-generation microcontrollers, edge computing and cloud computing. Rather than stopping at pilot-scale validation, the work follows the technology through deployment across 131 wells, allowing the field-wide operational and economic performance to be evaluated against expectations derived from the original ten-well pilot. This provides a direct assessment of the scalability and repeatability of high-frequency, adaptive sucker-rod-pump optimization at field scale.