Condition-Based Maintenance in Drilling Rigs: A Holistic Framework to Address Operational Context, Early Failure Detection, and Predictive Lifecycle Management

Abstract Drilling rig equipment operates under highly variable load conditions, making traditional sensor-only Condition-Based Maintenance (CBM) systems prone to false alarms and limited in predictive capability. This paper presents a holistic, context-aware CBM framework that integrates multisensing technologies - ultrasound, vibration, and temperature - with operational context data such as rig states and Electronic Drilling Recorder (EDR) channels. The goal is to enable both early fault detection and long-term Remaining Useful Life (RUL) estimation through a scalable digital twin architecture. The proposed framework addresses three key challenges: (1) distinguishing between rig states induced signal variations and genuine mechanical anomalies, (2) capturing degradation across different failure stages, and (3) enabling predictive analytics that adapt to rig-specific operational modes. A burst-mode sensing strategy is employed to optimize data volume and battery life, while rig-state tagging ensures that sensor data is interpreted within the correct operational context. Field trials were conducted on a land rig, with a total of twenty multisensing units deployed on one top drive, two mud pumps, and one centrifuge. Strategic sensor placement enabled detection of incipient failures, leak identification during pump-off states, and vibration severity tracking aligned with ISO 10816 standards. The digital twin continuously ingests and synchronizes sensor, EDR, and maintenance data, applying machine learning (ML) models tailored to rig states. This enables real-time anomaly detection, contextualized service hour tracking, and predictive maintenance (PM) scheduling. A novel application of ultrasound sensing during pump-off states proved effective for identifying fluid leaks—an area traditionally difficult to monitor in noisy environments. Additionally, the integration of in-situ maintenance data and drag-and-drop dashboards supports collaborative decision-making across field and remote operations. This work establishes a modular and extensible foundation for intelligent maintenance in drilling operations. By bridging the gap between sensing technologies and operational realities, the framework offers a practical path toward reducing non-productive time (NPT), improving equipment reliability, and enabling predictive lifecycle management.