High-Frequency Artificial Intelligence Control of Variable Frequency Drives for Horizontal Rod Pump Optimization
TL;DR
This paper demonstrates how an AI-driven controller enabling multiple dynamic intra-stroke motor speed adjustments that take advantage of the wave-like nature of stress propagation in the rod allows the well to run almost any user-desired pump plunger velocity profile, which can reduce rod stress or increase production.
Abstract
This paper demonstrates how an AI-driven controller enables multiple dynamic intra-stroke motor speed adjustments that take advantage of the wave-like nature of stress propagation in the rod. This allows the well to run almost any user-desired pump plunger velocity profile, which can reduce rod stress or increase production. The artificial intelligence (AI) controller continuously analyzes well conditions and automatically adjusts the velocity profile several times a second within each stroke, responding to changes in liquid availability, gas slugs, and loading conditions. Unlike conventional controllers with fixed parameters, the AI system recalculates optimal motor speed profiles in real time, enabling dynamic adaptation to any well condition. The mathematical model computes first the optimal downhole pump profile, then the corresponding surface velocity profile using the wave equation and the surface unit kinematic properties. Field examples from over a dozen wells across the Permian and the Eagle Ford basins validate the system's autonomous performance. Horizontal wells show the most significant benefits, with a 20% increase in inferred production at constant rod loading or a 15% reduction in peak-to-peak rod stress achieved. Simultaneously, the average pump fillage is increased by 12% by obtaining a smooth plunger upward velocity profile, which increases the pressure inside the barrel during the upstroke phase of the motion and prevents gas from bubbling out of solution.