Rapid Digital Oilfield Transformation Using Edge Computing and AI

Abstract

Abstract The timing and quality of operational decisions are critical factor in preventing production shortfalls and ensure sustained high level of operational efficiency. This impact of these factors amplifies multiple folds if the operations are in remote fields. Edge computing applications have surfaced as a powerful tool to ensure the right decisions are taken at the right time and by the right people. The paper describes multiple innovative and novel applications of edge computing and machine learning deployed in a remote field with no connectivity to enable peak well and network performance by efficiency improvement of artificial lift system, accurate virtual flow metering, minimize HSE exposure and simultaneously reduce the carbon footprint of the operation. New wells have been drilled in the green field located in western part of Libya. The field is a remote area with no digital connectivity yet and some of the wells require ESP to sustain production. The remoteness of the field also caused potential asset security concerns for the generators and ESP surface system (VSD). A holistic production performance enhancement digital solution package including sensors, solar power and satellite connectivity was deployed in the field centered around an edge computing platform to develop a digital twin of the green field with multiple wells. Virtual flow metering was done for all the wells using different techniques depending on the well architecture. Well models were run on the edge for wells with ESPs to estimate flow rates in realtime, while robust choke models based on well head and line pressure data were used to derive highly accurate well flowrates. The edge computing device was connected to ESP and the data gathered was analyzed in real time at the edge to identify ESP underperformance and detect any anomalies. At the same time machine learned algorithm continuously analyzed the video captured by two strategically positioned cameras to detect any potential intrusion by person or vehicles. It is estimated that apart from the significant production performance gain the solution has reduced 16 metric ton of carbon footprint on an annual basis associated with this operation. The solution allowed for rapid digitization of the field and early realization of value at a relatively much lower capex investment than traditional digital oil field solutions with an opex based model.