Prescriptive-Based Decision Support System for Online Real-Time Electrical Submersible Pump Operations Management

Abstract

Abstract Economic development and operations of new discoveries in frontier environments require advancements in production enhancement technologies, which take advantage of existing infrastructures and reduce overall operating expenses. These production enhancement technologies such as downhole pumping and gas lift (artificial lift technologies); injection of diluents, foamers and chemicals; water and/or fluid injection; often represent the only viable option to enhance well productivity, maximize production and improve recovery of hydrocarbon resources. Electrical submersible pump (ESP) systems are becoming the fastest growing form of artificial lift pumping technology, with more than 25% of the wells worldwide deploying ESP to surface their hydrocarbons and with more ESP systems being deployed to harsh and remote environments. With the rapid growth of ESP technologies, an appropriate prescriptive-based decision support system is needed to prevent or avoid critical system failures by analyzing the large amount of (observations, algorithms and calculated) information for extending run-time, increasing production and keeping well operation safe. The majority of the existing solutions provide real time data trending, alarming, ESP analysis and diagnostics. In this work, an efficient and reliable automated system is proposed for change point and anomaly detection, event or fault diagnosis and decision support in ESP operations. Our workflow consists of a quantitative method which combines streaming sensor data, data-rich detection and diagnosis model, fuzzy rule-based expert system and an influence diagram technique to perform prescriptive analysis and provide online real-time recommended action and/or corrective measures concerning the optimal alternative actions to be executed during undesirable ESP incidents or events. System features include a real-time data acquisition device, a real-time modeling engine, a real-time model calibration engine, fault detector program and a fuzzy logic rule-based system that uses an influence diagram technique to advice on ESP operating conditions. The proposed procedure is tested in multiple fields to verify the functionalities of the system. The results show that the proposed methodology can be efficiently used in practice. The methodology was used in several case studies and produced superior results in comparison to conventional methods.