Enabling Autonomous Well Optimization by Applications of Edge Gateway Devices, Automatic Fluid Level Analyzer and Analytical Dashboards

Abstract

The task of monitoring data in a substantial oil field, devoid of a digital platform, is a formidable challenge. However, it is of paramount importance in the context of Artificial Lift System (ALS) monitoring and optimization. In particular, for sucker rod pumping wells, the real-time collection and analysis of dyna cards assume critical significance. This process provides essential insights into downhole pump behavior and the overall system's health. The current practice of manual dyna card collection, occurring twice a week for approximately 47 horizontal wells, is notably infrequent, given the imperative need for real-time dyna card data, which necessitates a minimum frequency of 256 data points per minute. The analysis of such data proves exceptionally effective in the endeavor to optimize well production and enhance the longevity of both pumps and rod runs. The absence of real-time monitoring has, regrettably, led to well downtime and associated production losses. To address this issue, the amalgamation of Internet of Things (IoT), cloud computing, and machine learning has been introduced, thereby transforming our approach from a reactive to a proactive stance. This digital transformation has played a pivotal role in ALS optimization and has contributed significantly to mitigating production losses. The data is seamlessly transmitted to the Se Suite Central, a web-based Decision Support System hosted on the cloud. Given the sheer volume of dyna cards generated daily, the system has been equipped with an algorithm leveraging automated card classification, incorporating computer-driven pattern recognition techniques. The real-time data is harnessed for analysis, encompassing basic statistical methods and machine learning algorithms designed to classify thousands of dyna cards each day. Machine learning libraries have been employed to identify distinct pump signatures, subsequently categorizing them. Multiple informative dashboards have been meticulously developed to facilitate rapid analysis of ALS performance, including, but not limited to: Well Operational StatusDyna Cards Interpretation ModuleSucker Rod Pump (SRP) Parameters VisualizationMachine Learning Model Calibration ModulePump Performance Statistics Accumulating a substantial volume of data and harnessing domain-specific knowledge, these insights have been instrumental in driving ALS optimization efforts. Moreover, intelligent alarm systems have been deployed, drawing on statistical and machine learning settings. These systems promptly issue email alerts when anomalous behavior or erratic dyna cards are identified. This proactive approach has resulted in a reduction in well downtime during select events that were previously addressed reactively. The fusion of domain expertise with digitalization has empowered decision-makers to take informed and efficacious actions. This project has exemplified its capability in remotely managing an asset encompassing over 47 wells, all while operating with limited resources. The implementation has proficiently sustained intermittent operation of low Productivity Index (PI) wells, leading to substantial power savings associated with surface pumping units. This digitalization initiative has, in no uncertain terms, averted numerous pump and rod failures, thereby preserving significant workover jobs and minimizing well downtime.