Oil and gas companies have, for decades, built reservoir models, aiming at obtaining “digital twins” of the hydrocarbon fields they are exploiting. The traditional approach in our industry would focus on building a single bas-case model, which is calibrated against observed data from the field, before using the model for reservoir management decisions such as studying incremental recovery from additional drillings campaigns, effect of water/gas injection etc.
The traditional approach to modelling is tedious, and, too often, operational outcomes of the industries modelling efforts are sub-optimal, e.g., infill wells are quickly watered out, wells are drilled in poor quality or depleted areas etc. Often, this can be traced back to our ability as an industry to systematically consider subsurface uncertainties in modelling efforts and in the decision-making process, e.g., choice of new drilling targets.
With the advent of ensemble methods, we can take the advantage of the algorithms’ ability to efficiently calibrate an ensemble (75-100 models) to observed data, providing an uncertainty-centric framework that allows for oil & gas companies to harness the power of this new approach to change their decision-making process to the better by making uncertainty an integral part of their modelling efforts.
An ensemble is a collection of equiprobable realizations representing the subsurface, capturing both static and dynamic uncertainties. A Kalman family data assimilation algorithm known as Ensemble-Smoother with Multiple Data Assimilation (ES-MDA) is able to quickly calibrate the ensemble to the production history, reduce static and dynamic uncertainties, thereby improving the reservoir characterization understanding, and assess realistically the risks relevant to potential development scenarios for informed decisions.
This paper highlights the learnings from implementing an integrated uncertainty-centric ensemble modelling approach to hydrocarbon fields as part of a digital transformation strategy and spotlight some of its successful applications in the Asia-Pacific area, and its future use, most notably to model - and derisk - Carbon Capture and Storage potential targets.
