Geostatistical history matching - A zonation-based approach using direct sequential simulation

Simultaneous integration of production data under geological consistency, as part of the reservoir modelling workflow, still remains a challenge. A geologically consistent approach aims to avoid solutions that are unrealistic under the reservoir's general geological characteristics. Unrealistic history matching solutions would result in poor reservoir response forecasting. It is also essential to include only geologically realistic models for uncertainty assessment based on multiple models that are consistent with the geological data and also in order to match observed production history. Geostatistical history matching can iteratively update static reservoir model properties through conditional assimilation constrained to the production data, using geologically consistent perturbation. Multiple stochastic realizations are assimilated following a zonation approach to account for the local match quality, thus providing a way to integrate a regionalized discretization of parameters with production data and engineering knowledge. The present project proposes a new history matching technique applied in uncertain reservoir conditions, represented by geologically consistent reservoir zonation, based on fault presence and dynamic consistent patterning, by considering fluid production streamlines as auxiliary criteria for the zonation. The proposed methodology is tested in a semi-synthetic case study based on a braided-river depositional environment.