As an increasing number of automated functionalities become available from different providers, it remains challenging to ensure these functionalities work consistently and reliably together at the rig site. SPE Drilling System Automation Technical Section recognizes this difficulty and launched a subcommittee focused on understanding the industry's challenges, current state, and exploring verification and validation solutions to enable multi-providers distributed automated drilling control. This paper will introduce this initiative and point toward a framework that can potentially assist in scaling adoption of distributed multi-vendors drilling automation.
Drilling automation systems need to cope with uncertainty and complex drilling conditions, where unexpected changes can disturb the overall process. Hence, continuous identification of local conditions is essential for a safe and efficient drilling process. When an automation system is fully integrated under one entity, complete knowledge of the different parts may be assumed. The more distributed an automated system is, the more fragmented and challenging to assemble this knowledge is. Thus, information flow across the agents in a multi-agent system is crucial to assess current conditions. A verification and validation process could assess if the different agents can efficiently collaborate at any time and identify potential inconsistencies.
The structure of a multi-agent drilling automation system is not fixed, different agents may change throughout the overall operations, become unavailable or being replaced by others. Therefore, a verification and validation of the interplay between the agents should be designed to be dynamic and account for such changes online. Each agent can be relying not only on certain input data to be available in the system for a proper functioning, but also on features provided by other agents. For instance, an agent providing an optimization of Rate of Penetration can rely on another agent providing values for safe operating envelopes functionalities. While each automation agent may undergo a-priori an independent verification and validation, due to the distributed nature and corresponding dependencies in a multi-agent system, there is a need to continuously assess if required data and features are available at any time. Such a framework can facilitate an automated online consistency check of data and features in a multi-agent automation system.
Addressing a framework that can cope with the on-the-fly structural changes can lay the basis for online automated verification and validation of complex distributed drilling automation systems. This could enhance current capabilities of existing automated functionalities to new performance levels while ensuring stability of the system, reliability, and safety in the overall drilling process. Even though the paper focuses on the use case of drilling, the introduced framework for online verification and validation of a complex multi-agent system can be extended to completion, intervention, and P&A operations.