While-Drilling Pore Pressure Surveillance Using Machine Learning

Abstract

While-drilling pore pressure surveillance enables timely responses to unexpected drilling events, e.g., wellbore instability, or to pressure changes that could impact mud weight requirements or casing depths. The real-time pressure surveillance and analytics (RT-PSA) system described herein aids while-drilling pressure surveillance by highlighting possible pressure trends and detecting pumps-off gas automatically. The system further assists pressure surveillance practitioners by automatically filtering for lithology and providing a visualization dashboard to highlight possible pressure trends. The pressure trending application calculates slopes/trends for LWD and mechanical data and uses these trends to indicate possible pressure trends along the well path using a heat map. A lithology filtering method has been developed using machine learning (ML) clustering algorithms to remove non-shale data, leaving only clay-rich shale lithology for pressure trending. The gas monitoring application aligns the gas curves back to the time and depth at which gas is liberated from the formation by the drill bit, called herein as at-the-bit curves. The application displays modified total gas, gas exponent, and gas ratio curves as at-the-bit curves. The gamma ray and resistivity LWD logs are also shifted back to the time/depth that the bit drilled the measured formations. Aligning the gas and formation log curves to be at-the-bit provides the pressure surveillance personnel with additional context beyond traditional gas surveillance data to classify gas measured at the surface as pumps-off-gas or formation gas. Results demonstrate that the lithology filtering method using machine learning is effective to filter out clay-rich shale. The pressure trending results are consistent with post-drill pore pressure evaluations generated by pressure prediction experts. The shifted total gas and pumps-off gas have been validated versus post-drill pressure analysis. The system is being deployed to mitigate well control events by improving and standardizing pressure surveillance best-practices across a global organization.