Search for a command to run...
Abstract Drilling automation can significantly enhance process and equipment safety during well construction. Real-time estimation and automated monitoring of operational limits for equipment and processes enables early detection and response to potential hazards, such as stuck pipe. This paper discusses the integration of drilling engineering models with data-driven approaches to determine limits for hookload (HKLD) and standpipe pressure (SPP), and how these limits are used to automatically prevent abnormal conditions. Physics-based models deployed in real-time define limits for surface parameters, such as HKLD and SPP to ensure wellbore and equipment integrity. Within those limits, data-driven algorithms use statistical analysis of sensor data in real-time to define process operating envelopes at the current operating point. For HKLD, these limits are derived by analyzing block movement, rig activity, and profile of the sensor signal over different time intervals. To include the drag in the system, pick-up (PU) and slack-off (SO) measurements are overlaid with the limits obtained. As soon as the HKLD exceeds a limit, a mitigation process is automatically triggered. This approach implements the concept of fault detection, isolation and recovery (FDIR) put forth by (Cayeaux, et al. 2023). The calculation of SPP limits follows a similar approach but is more challenging than for HKLD, since the detection needs to account for rapid changes in the flow rate and its effect on the behavior of SPP. Limits provided by physics-based models define an absolute space to define integrity. Within this space, data-driven algorithms adapt to changing behavior of HKLD and SPP for all kinds of operations to create a corridor around the current operating point and enable early detection of potential hazards. The integration of both types of limits provides a more holistic protection schema for well construction operations. This approach has been verified and validated in simulated environments and in the field. Field tests have shown that limits reliably adapt to real-time signals. As a result, the monitoring algorithms consistently detect overpulls and set-downs as soon as HKLD transitions outside the operating envelope. SPP limits are carefully determined in scenarios with consistent flow rates and are used to reliably monitor for potential overpressures. The limits quickly adapt to steep changes in SPP, which is critical when automating the circulation system. In deployments where an automated drilling control system (ADCS) is present, abnormal behavior of HKLD and SPP—such as overpulls or pack-offs—was successfully mitigated, thereby increasing safety for personnel, wellbore, and equipment. The presented approach outperforms systems that disregard process dynamics and rely solely on static equipment specifications. Furthermore, it yields better results than physics-based or machine learning models when deployed in isolation. This approach provides dynamic limits derived directly from monitored signals, reducing the need for intermediate processing.