Characterizing Drilling Vibrations by Interlinking Surface Data, Drillstring Design and Lithology of Rock in Utah FORGE Deep Test Well 58-32

Abstract

Dealing with vibrations is inevitable during drilling operations. Monitoring these vibrations is crucial for safe and efficient drilling. While surface vibrations are a good indicator of drilling conditions, the severity of downhole issues cannot be easily discerned from surface data. Although utilizing downhole sensors is a reliable solution to the issue at hand, obtaining real-time downhole data is both complex and expensive. This brings us back to the question of whether surface vibrations can help us to identify drilling dysfunctions? This paper aims to provide an in-depth review of the surface data obtained in Utah FORGE deep test well. Going a step further, the surface vibration data is combined with mud logs and daily drilling reports to build an algorithm to first differentiate and then identify harmful drilling vibrations in isolation to other vibration patterns. Understanding lithology changes is a key step in this analysis. While characterizing surface vibrations help in recognizing drilling dysfunctions, not every vibration pattern can be easily distinguished as harmful or within operational limits of the drill-string. The paper discusses the use of a classification model, which is a machine learning algorithm that classifies vibrations based on its level of severity. The paper thus presents a solution to the complex nature of surface data measurements by selecting inputs such as surface torque, rpm, rate of penetration and weight on bit as major indicators of vibrations. By examining these parameters, the paper successfully classifies the data into regions of severe torsional and axial vibrations leading to a potential stick-slip and bit bouncing phenomenon.