Automatic Online Classification of Drilling Activities [PDF DOWNLOAD ABOVE]
Recognition of which drilling activity is occurring at the wellsite is important for calculating key performance indicators (KPIs), diagnosing potential issues during a drilling operation, and providing advice to the automated drilling control system (ADCS). The drilling team is aware of the current drilling activity at any given time, and daily drilling reporting systems categorize drilling operations into activities and sub-activities. Recognizing which drilling activity is being executed in real-time by a computer system is, however, a challenging task. A method has been proposed to uniquely characterize the drilling process state at any instant. This method relies on the decomposition of the possible boundary values of the drilling process into so-called micro-states. The combination of the values of all the micro-states constitutes the drilling process state. The process of estimating the value of each micro-state can be computerized using inputs from measurements and estimations from digital twins. The succession of certain patterns in the flow of drilling process states is characteristic of a specific drilling activity. By recognizing these patterns, it is possible to categorize the current drilling activity. The drilling process is a hierarchical construct, and layers of activities with drilling equipment yield layers of drilling operations. Since the construct is hierarchical it is possible to formulate rules about the relationships between drilling process states and how they characterize drilling activities. In the digital space, a similar technique is Backus-Naur Form (BNF), a standard technique used to describe valid syntax of computer programs in a programming language. BNF rules can define what constitutes a low-level drilling activity, and also specify that only certain combinations of low-level drilling activities make sense. This allows defining drilling activities and sub-activities in a computer-interpretable manner. The use of BNF rules simplifies the implementation of a parser that categorizes drilling activities in real-time using a classical pushdown automaton. Any future adjustments to these rules to accommodate new drilling methods and techniques do not require recoding the drilling activity parser. If there are insufficient states to uniquely identify the drilling activity then the parser can also generate the set of possible drilling activities matching the set of identified states, and it can be used to pro-actively trigger a desired activity based on an entry, for example, in the rig action plan. This work is part of the Drilling and Wells Interoperability Standards (D-WIS) sub-committee of the Drilling System Automation Technical Section (DSATS). The D-WIS infrastructure provides shared functionalities to facilitate the collaboration of multiple applications during drilling operations.
