In this paper, we combine reservoir data; reservoir physics; machine learning algorithms; and the "know-how" of the subsurface team in one efficient workflow to quickly build reservoir models that lead to an improved understanding of the Gjøa field offshore Norway. Key elements of the solution include:
Consistent integration of static and dynamic data – when they arrive – in one automated and repeatable modelling process.
Ability to address uncertainty in all parts of the modelling chain – from G&G data interpretation to flow simulation and dynamic data conditioning.
Single platform for multidisciplinary collaboration between the subsurface disciplines in the modelling and reservoir management efforts.
Scalability and robustness – the workflow is equally applicable for small and large reservoirs.
Additionally, we implement a prescriptive analytics solution to quickly identify and rank robust infill-well targets using data conditioned reservoir models as input.
The amount of data we collect from a reservoir is rapidly increasing. Hence, it is more important than ever to utilize tools that address the shortcomings of the traditional reservoir modelling approaches. Having the ability to capture the information found in the collected static and dynamic data in a consistent manner when they arrive, while embracing the inherent uncertainty of the reservoir modelling process, will lead to improved reservoir management decisions, and ultimately an increase in the recoverable volume.