Early Identification Technology for Overflow Induced by Gas and Oil Invasion During Drilling

ABSTRACT: The deep and ultra-deep strata are characterized by complex lithologies and frequent well control incidents. The economic losses caused by well control accidents worldwide exceed 2 billion US dollars annually. Early detection and intervention can reduce these losses by at least 80%. This study employs machine learning methods, including neural networks, decision trees, and gradient forests, to investigate the surface manifestations of gas and oil invasion during drilling in China's onshore trap oil and gas reservoirs, tight oil and gas reservoirs, and shale oil and gas reservoirs. It focuses on analyzing the changes in real-time engineering data collected at the surface during three stages: encountering gas and oil layers, gas and oil invasion into the well, and return to the surface. The study finds that the micro-drilling time curve for trap oil and gas reservoirs shows a significant upward trend at the moment of encountering gas and oil layers. After gas and oil invade the wellbore, the drilling pump pressure curve exhibits an exponential increase, albeit slowly, with the growth coefficient directly correlating with the invasion volume and indirectly correlating positively with the bottomhole negative pressure difference and reservoir liquid supply capacity index. When gas and oil return to the surface, the outlet flow rate rises rapidly, with the increase pattern strongly correlating with the flow rate of gas and oil invasion into the wellbore, the negative pressure difference between the formation and the wellbore, the gas phase ratio, and the vertical depth of the wellbore. The research indicates that constructing surface engineering parameters for overflow complexity within the late arrival time cycle, tailored to different burial depths and reservoir types, can effectively support engineering technicians in analyzing the instant gas and oil invasion into the wellbore and optimizing pressure control methods based on the fluid and pressure distribution in the wellbore.