Architectures and algorithms of an autonomous small-scale drilling agent

Abstract

This paper describes the core architectures and algorithms of an autonomous small-scale drilling agent. The agent operates in a laboratory rig, demonstrating drilling scenarios with limited or even no human intervention. The work illustrates its performance through self-coordinating state transition, Rate of Penetration (ROP, drilling speed) optimization capability, formation classification, and drilling incidents management. The agent is an original rule-based system, and its control architecture utilizes finite states automation. The novel ROP optimization strategy employs a gradient search in Weight on Bit (WOB)-rotational speed (RPM, Revolutions per Minute) control parameter space. It generates an increasing ROP trend with time and requires re-iteration at abrupt formation changes. Several drilling incidents are managed using ‘if-then’ logic-based activity decomposition. A key learning outcome from the study is the comprehension of the requirement of standard software architecture and Applications Programming Interfaces (API) for continuous research and development of the agent. Such interfaces enhance interoperability between systems and stimulate innovative thinking among independent developers to produce a better-faster set of algorithms. Laboratory testing and evaluation is an essential part of promoting the adaptation of digital technologies for drilling automation. Such studies are a useful, safe, and cost-effective solution for testing, integrating and improving hardware, software, and data management before expensive full-scale testing and integration.