A Multi-Faceted, Multi-Disciplinary, Machine Learning-Driven Dashboard for Safe and Efficient Drilling

In this webinar, the speaker presents a next generation vision for a state-of-the-art dashboard (QDashRT) driven by machine learning that provides drillers and operations geoscientists with real-time guidance for several mission-critical parameters and measurements. These include a sub-seismic resolution earth model, parameterized with pore pressure and geomechanical properties that are updated in real-time with new LWD data and enables ahead-of-the-bit and away-from-the bit visual awareness of potential drilling hazards; also included is a drilling parameter advisory system driven by an ROP simulator that considers transitions in rock properties; a synthetic logging and seismic-to-well tie product that, in the first instance, simulates DTC, DTS and RHOB for use in real-time pore pressure and geomechanical analysis, and in the second instance, facilitates the most accurate location of the well relative to geologic structure available in real-time. Finally, a novel implementation of a recurrent neural network is used to detect, in advance, the onset of a potentially hazardous formation, such as highly fractured carbonates or depleted sands.

The individual components of the workflow will be described, along with the underlying data science framework and the Midland Basin geology targeted by the real-time earth model. All content contained within this webinar is copyrighted by Gareth Taylor and its use and/or reproduction outside the portal requires express permission from Gareth Taylor.

SPE Webinars are FREE to members courtesy of the