Edge Optimization Analytics and Control: A Integrated Device of Intelligent Intermittent Pumping for Tight Oil Wells

Intermittent pumping is an effective measure to reduce energy consumption and improve production time rate for tight oil wells with extremely low oil production. However, the intermittent pumping scheme is mainly controlled by manual timing or RTU remote timing in the cloud. The design of intermittent pumping interval lacks reasonable theoretical basis and the control efficiency is relatively low. Therefore, an integrated intelligent intermittent pumping device combining edge optimization analytic and edge control is developed. A closed-loop integrated device containing real-time data acquisition and transmitting module, optimization module and edge control module is realized. The optimization module embeds the pump fullness calculation model and intermittent pumping scheme optimization model. Instead of using dynamometer cards, a deep learning model of transferring power curves to dynamometer cards named PTD (Power to Dynamometer) is established through deep learning in order to lower the oilfield IoT cost. The average single well area error of predicted dynamometer card is less than 3%. To transfer the PTD to untrained new wells, a parallel model is established by parallel connection of PTD and transferring learning model. The parallel model is embedded into the edge computing device to realize the dynamometer card real-time prediction and pump fullness calculation. The calculated data is transmitted to scheme optimization model where a reasonable intermittent pumping scheme can be optimized. Then the optimized pumping scheme is transmitted back to the edge control device to adjust the pumping scheme to realize loop control. This technology was applied to 100 wells in the oilfield, the monthly energy consumption reduced by 30%, and the average pump efficiency increased by about 17%. The application of this technology is insightful to promote the IoT and intelligence of oilfield artificial lifting system.