Abstract
Abstract
A practical ensemble-based reservoir modelling workflow has been implemented for the Ichthys gas-condensate field, located in the Browse Basin, North West Shelf of Australia, where the field production commenced in July 2018. Ensemble-based modelling methods are attractive tools for reservoir uncertainty quantification; however, for some development planning such as the well count optimisation, it is challenging to use a full history-matched ensemble because it requires an impractical number of forecast simulation runs. An objective of this work was to establish an ensemble-based workflow that is practically applicable for development planning.
An ensemble of 100 full field reservoir models is generated with sampling from key reservoir uncertainty parameters selected by a multi-disciplinary team considering the impact on the production forecast. Any deterministic cases can also be inserted in the initial ensemble. The Ensemble Smoother with Multiple Data Assimilation (ES-MDA) is applied as an assisted history matching method in the workflow to minimise an objective function that includes shut-in bottomhole pressure and formation pressure measured at development wells. After running the workflow and production forecast with the history-matched ensemble, a smaller set of deterministic cases (sub-ensemble) is generated as a representative of the history-matched full ensemble to practically perform development planning.
Two deterministic cases, a base case and an alternative geological scenario case, were inserted in the initial ensemble. The objective function and uncertainty range of the reservoir properties were effectively reduced by running the workflow in a short time, retaining geological constraints of the reservoir properties. The history-matched full ensemble was used for the production forecast to estimate the uncertainty range of the future production profile and to utilize the result for development planning. Some dynamic property maps, such as permeability thickness, pressure depletion and remaining gas in place maps, were used in addition to geological interpretation and seismic data for the future candidate well location selection. A sub-ensemble was generated with four deterministic cases, the two cases inserted and history-matched in the ensemble, and the two cases generated based on the range of key forecast results from the full ensemble. A well count optimisation study was efficiently performed by the multi-disciplinary team with the sub-ensemble capturing key uncertainties in the production forecast. A development strategy adaptive to multiple reservoir scenarios was then created based on the optimisation result.
We established and demonstrated a practical ensemble-based reservoir modelling workflow that enables us to perform history matching, uncertainty quantification and development planning in a short time, resulting in more time available for comprehensive planning of the development strategy and decision making.