Challenges and Opportunities Related to Use of Models in Drilling Operations

Automation of managed pressure drilling (MPD) has a big potential for improving consistency, efficiency, and safety of operations, and is therefore pursued by many actors. While this development mitigates many risk elements, it also adds some related to for example mathematical algorithms and remote access. This work is based on document reviews, interviews, and working sessions with the industry, and adds insight on how risks associated with moving to higher levels of automation of MPD can be mitigated to a level where benefits are significantly larger than the sum of added risks. The work has resulted in many recommendations for the industry, where most were related to testing, verification, and validation of the models and data inputs, as well as meaningful human control and employing a holistic approach when introducing new models. Recommendations were also given to the Petroleum Safety Authority Norway. These were related to missing or inadequate use of standards by the industry, lack of ICT knowledge, and encouraging increased experience sharing. Future work should address how to enable meaningful human control as models become more complex or to a larger extent is based on empirical data and artificial intelligence as opposed to models based on first principles. Human control is important in unexpected situations in which the system fails to act safely. There is also a need to address ICT security issues arising when remote operation becomes more common.