AI-Enabled Unconfined Compressive Strength (UCS) Real-Time Prediction Utilizing Logging-While-Drilling Measurements

The quality and applicability of the training data used to create predictive models have a major impact on how well data-driven drilling decisions work. The oil and gas sector has been using machine learning (ML) techniques more and more in recent years to glean actionable insights from massive databases of historical and real-time drilling data. Drilling operations planning and execution are being redefined by the capacity to train intelligent systems that can recognize patterns and predict subsurface behavior. The use of logging-while-drilling (LWD) data streams is essential to this evolution because they offer high-frequency information that, when properly utilized, can be crucial in lowering non-productive time (NPT), improving drilling efficiency, and boosting operational safety. The Unconfined Compressive Strength (UCS), which measures a formation's resistance to mechanical failure, is one of the most important rock characteristics affecting drilling performance. Important drilling choices like the ideal Weight on Bit (WOB), rotary speed, and bit type are all directly influenced by UCS. Engineers can react dynamically to changes in the formation by adjusting parameters to either drill more aggressively through soft zones or reduce mechanical stress in harder, abrasive intervals when they have a solid understanding of UCS in real time. Continuous UCS insight becomes not only valuable but crucial in abrasive, heterogeneous sandstone formations, where sudden variations in rock strength can result in excessive bit wear or ineffective drilling. However, real-time access to UCS data has historically been limited. Conventional methods rely on core samples or post-drilling analysis of wireline logs, both of which suffer from delayed availability and limited spatial resolution. These gaps prevent timely optimization and expose operations to avoidable risks. By contrast, a system capable of continuously estimating UCS from surface-acquired or downhole LWD data represents a step-change in drilling intelligence.