A Data-Driven Agile Approach for Faster Response to Lost Circulation

Abstract In this paper we introduce a data-driven agile method specifically designed, implemented, and validated to yield actionable insights for more efficient and accurate responses to lost circulation events. Lost circulation is a significant challenge in drilling operations, leading to non-productive time and financial losses. It is also frequently observed that despite similar field conditions, the success rates of various solutions vary widely due to a lack of insights about the loss zone during planning and execution. Traditionally, response plans and decision trees are not dynamic and rely on categorizing loss rates as seepage, partial, or total losses. However, these plans often overlook critical factors such as the types of losses (e.g., natural fractures, induced fractures, vugs) and dynamic changes over time in loss rates, wellbore stresses, and pressures. Thus, developing a solution that processes these changes in near real-time to provide automated insights to the person monitoring the event can improve the success rates of our loss event response. Additionally, there exists a disconnect between the multiple tools available for addressing lost circulation. These tools, when used together, can help in driving actionable insights and accurate detection of losses; i.e., active pit tracking by mud logging and mud logger's real-time comments, annular pressure while drilling (APWD) bottomhole pressure anomaly detection, geomechanics pressure wellbore limits, etc. Finding a way to bring all these tools and technologies to one platform so they can collaborate and complement each other can be a game changer in the precise detection of loss events and rates. While developing the solution, numerous complexities were observed in pursuing a fully automatic method to put these multiple data sources together and drive a comprehensive outcome with the required precision, thus the concept of keeping a human in the loop was adopted. In this paper, we delve into the details and uniqueness of the lost circulation monitoring workflows. We discuss the methodology used during workflow development and the establishment of a well-defined system for workflow validation using a labeled database. An overview of the uniqueness of the workflow computation, workflow verification process, and tiering of a workflow based on the quality of inputs and outputs will also be provided, along with the results of the real-time job applications and their successes.