Characterization and Uncertainty Reduction on Facies Distribution and Probability Cubes with Ensemble Kalman Filter History Matching

Abstract

Abstract Power of ensemble based data assimilation methods to history match oil and gas reservoirs has been revealed during the past couple of years by different field application works (Oliver and Chen, 2011). Moreover the achievements in advanced parameterization (Lorentzen et al. 2011) and (Hanea et. al. 2014) to deliver posterior models which respect prior geology drastically enhanced the robustness of the procedures. In this study a huge filed case represented by a mega geological model has been the candidate for assisted history matching with ensemble based techniques, ES-MDA (Emrick, Raynold 2013) in particular. To insure the geological realism of history matched models, a parameterization approach similar to (Hanea et. al. 2014) has been adapted and improved, as the initial workflow for lithological modeling has been based on truncated Gaussian simulations (TGS). Adding 3D uncertain facies proportion distributions which has been produced by stochastic inversion of seismic information was the way to enrich and ameliorate the method. These proportions have been used as the base of truncation procedure in TGS. In this work we showed how to use geological and geophysical based uncertainties in reservoir modeling, and with an appropriate parameterization, one could obtain fully realistic models which honors all a priori information and also matches the field measurement data. The main challenge with proposed approach was to insure honoring the well data which is normally constraining data in facies modeling procedure, taking into account that both underlying Gaussian variable and proportion values are modified by data assimilation process. To insure such a conditioning certitude, all well data has been reflected in probability maps by an a priori kriging on well location and closed by areas. This strategy guaranteed unchanged well data conditioning on all final realizations.