In order to optimize gas lift rates and maximize production profitability for gas lift wells, a ‘cloud-edge-end’ intelligent control system is developed. By leveraging an ensemble of artificial intelligence (AI) agent at the edge, the system is able to consume real time data and provide automatic gas lift control.
Flow meters is integrated with machine learning models to ensure measurement accuracy. Machine learning (ML) is used to build the AI agent which embedded in the wellhead edge after fine tuning. Real time data from sensors and meters are sent to the AI agent. It calculates the optimum flow rates with edge analytics capabilities and compares with the flow values detected by flow meters. When the detected value is inconsistent with the calculated one, the AI agent will take actions (rewards or penalties) automatically by regulating the gas lift injection valve and production choke, until meeting the optimization goal. The solution is able to automatically control the gas injection rate and liquid production rate in a close loop system with edge architecture. Compared with manual control, the constant flow rate is maintained and flow fluctuations are avoided.