Chasing Business Critical Insights in a Fast and Continuous Manner Using Ensemble-based Modeling Approach

The work presented here introduces an uncertainty centric ensemble-based approach for reservoir modelling. The approach, as opposed to traditional step-wise siloed approach, promotes dialogue and collaboration between subsurface professionals to gain a better understanding of the uncertainty associated with the reservoir at each modelling step.

The proposed integrated and automated workflows are described which start by building an initial ensemble using all the known uncertainty in the available data. The initial ensemble is then history-matched using an ensemble-based algorithm that makes use of multiple machine-learning techniques. Subsequently, the history-matched ensemble is used to then perform forecasting for the reservoir under all the inherent uncertainty, enabling subsurface teams to truly analyse risks and opportunities associated with crucial business decisions.

The methodology is demonstrated on a carbonate reservoir and the resulting value gained is highlighted. The approach enables teams to perform reservoir modelling in a fraction of time, compared to traditional deterministic approach, and facilitates assimilation of field data as it becomes available, in a robust and seamless manner.