Deepwater drilling often requires simultaneously hole-enlargement- while-drilling to improve project economics and preserve hole size for reaching deep reservoir. One potentially disastrous scenario is when underreamer has been severely damaged, but drill bit is still in good condition. If such situation is not detected promptly and drilling continues, the hole enlargement while drilling operation could potentially create a significantly undergauged hole section whose diameter is similar to drill bit. Such consequence can cause enormous difficulty for running/cementing the casing string, and often result in costly remedial operations, or in severe cases, hole abandonment.

An advanced data analytics method was developed to early detect the failure of underreamer by utilizing the standard drilling mechanics data from both surface and downhole sensors. This patent-pending method includes the development of new performance metrics based on rock cutting physics and wearing mechanism. A warning signal can be displayed immediately to the drilling crew on the rig and the drilling engineers at office when the underlying failure pattern is detected by the new method. Historical drilling data and failure events are used to train the model with the additional inputs from subject matter experts. This presentation will include the development roadmap of this data analytics method and its implementation results at Deepwater wells in Gulf of Mexico.