Real-Time Evaluation of the Lost Circulation Zone Based on a Changepoint Analysis of the Mechanical Specific Energy and Standpipe Pressure

Abstract

Abstract There are multiple mechanisms that may drive a lost circulation event while drilling a well. Efficiently dealing with those events requires that the operations team quickly get an understanding of subsurface conditions that caused the event. Globally, some lost circulation events are cured by basic lost circulation materials (LCM) / bridging, while others fail to be cured even after days of attempts with lost circulation cement plugs (LCP) and novel materials. The aperture of the lost circulation zone (LCZ) remains unknow in most of the cases unless open hole logs ran to identify it. The standard approach to cure the losses is to start with less aggressive materials followed by more aggressive, and the curing based on the field practices, rather being linked to the potential opening of LCZs. The manuscript will investigate the hypothesis that the lost circulation events related to penetration of large, connected, open cavities (karsts) can be characterized, and those large open cavities can be identified in near-real time by analyzing the dynamic drilling parameters with help of Machine Learning. There are different mechanisms that are driving the lost circulation events. Within the same formation some of lost circulation zones can be easily cured with LCM/LCP, others require more effort and time and may be uncurable. The invention proposes a workflow and an algorithm to detect from dynamic drilling parameters what is the likely mechanism at play, and whether the lost zone is curable. Large fracture and dissolution cavities will have different mechanical properties, and drilling through those features will require significantly less energy than through a competent rock formation. They are also discrete events within a geological formation, and therefore will have outlier mechanical properties within a formation. The test was performed in over 300 wells, across different lost circulation zones (some wells had several zones). The developed algorithm was incorporated into the software in Real-time Monitoring center, allowing near real-time estimation of the aperture to decide regarding the LCM or LCP to be used. The results showed that in upper sections the multiple lost circulations zones presented with different thickness. The majority of the identified karsts are within range of 2-to-8 ft, with some going over 10 ft. In the deeper formations LCZ with the aperture of 4-to-9 ft were identified. The interesting part was related to the significant difference between MSE while drilling competent formation and the lost circulation zone, when MSE values were dropping almost to zero. The manuscript provides the novel approach allowing to use machine learning to identify the aperture of the lost circulation based on the real time parameters. The proposed approach can be used at any drilling project worldwide.