Real-Time Well Monitoring and Engineering Analysis of Drilling Activities: Intelligent Rig State Detection and Prediction With Uncertainty

Abstract

Drilling activities are risky and costly, especially when performed offshore. Careful monitoring and real time data analysis are required for safe and efficient operations with minimized down-time. Drilling operations, being fast-paced and not visible, often lead to transient and unforeseen issues. The synchronous assessment and prediction of drilling quality has historically been a challenge. It relies on a prompt collection, analysis and prediction of the multiple sensors data, as well as an immediate comparison to the original drilling plan. Another challenge is achieving real-time well engineering, and automatically and instantaneously providing valuable insights to the engineering and operations teams. A system was successfully developed to tackle these challenges. It is a cloud-based application, made with an event-driven streaming architecture to automatically retrieve real-time drilling data and compare it with planned data. The real-time data is automatically made available to determine the current well operation or rig state, and trigger the subsequent engineering analysis. Next, a forecast model is trained with the engineering calculation outputs and it returns predictions on these outputs while considering their inherent uncertainty. As a result, these predictions enable alerts to be sent when the system detects approaching anomalous conditions. The proposed system is a DecisionSpace® 365 cloud-native application on an open architecture. It is flexible, accessible from anywhere, can be automatically updated for continuous improvement, and can be deployed easily and quickly. It can also be extended to further applications.