Applying Digital Rapid Response Ranking and Event Detection System in Oriente Basin

During times of uncertainty, upstream producers focus on cost optimization and maximizing net profit bound by the highest safety and environmental impact. They seek the ability to decrease system failure rates, consequently minimizing downtime and lifting costs by extending equipment running life and optimizing operating costs. This paper's main objective is to showcase field results for electric submersible pump (ESP) optimization, ranking, and automatic event detection. The Oriente Basin is the largest brownfield in Ecuador, from approximately 100 producer wells. Production is achieved from different reservoirs with a high water cut. The artificial lift method selected for these wells is solely ESP. There are several operational challenges; however, the main challenge is to rank the wells that need attention to act upon, to successfully improve the efficiency of operations to maximize productivity and reduce the ESP failure event. A smart digital system was installed and equipped with trained artificial intelligence and machine learning (AI/ML) to predict a catalog of undesired and critical events and suggest potential actions, to reduce the time of action and avoid production losses and improve the ESPs’ performance indexes; specifically, mean time before failure (MTBF) and failure index (FI). To achieve these goals, it is necessary to create a digital ecosystem that enables the integration of tools, surveillance, and knowledge. This integration must coincide with the process of going through a digital transformation. The framework includes gathering ESP data frequently, creating a fingerprint for the key ESP problems, understanding how operations conditions vary, and automatically updating the threshold. Opportunities are identified by implementing well thresholds, severity ranking systems, and AI/ML with advanced detection of undesired operating conditions. Integrating the digital solution, in addition to continuous well review and diagnostics by specialist staff, detected critical ESP events, generated key alarms, and provided communication with the field in the appropriate time and way. Together these resulted in enhancing the ESP run life from 247 days to almost fourfold which is 950 days (about 2 and a half years). The capacity to combine field knowledge and real-time data enabled with a rapid response AI/ML customized catalog of workflows created the possibility of intelligent actions in the digital field operations by detecting and ranking operational events, creating a focused list of potential failure threats and providing insights of required actions to change course. In conclusion, it reduced the number of ESPs to be shut in every year.