Next-Generation Geological Model Updating and Ranking for Improved Oil Recovery
Halliburton, Highlands Ranch, CO.
The conventional oil production practices recover, on average, approximately one third of the original oil in place with estimated remaining mobile oil. To increase the overall production, large investments are made in Improved Oil Recovery (IOR) of which success greatly depends on ability to estimate volumes and locations of bypassed oil from available historical data using History Matching techniques. We present a new approach with the potential to more accurately capture uncertainty of the inherent geological model, facilitate accurate description of reservoir heterogeneities and honor the conceptual depositional model.
The novelty lies in direct interfacing between Next-generation geological modeling and forward simulator. Efficient model parameterization that enables rapid generation of model updates in wave-number domain is used to characterize the main features of geologic uncertainty space: structural, stratigraphic, facies and petrophysical properties. Model inversion workflow is based on multi-step Bayesian Markov chain Monte Carlo (MCMC). Traditional MCMC methods provide most rigorous sampling of posterior distribution but suffer from high computational cost. We implement an approach where proxy model is guided by streamline-based sensitivities, dispensing with the need to run forward simulation for every model realization thus significantly reducing the computation time. An ensemble of sufficiently diverse model realizations is generated at the high-resolution geological scale that secures more accurate result by obeying known geostatistics and well constraints.
The workflow is validated on a case-study combining geological model with ~1M cells, four different depositional environments and 30 wells with 10-year water-flood history. A history match indicates significant reduction in the misfit between observed and simulated water-cut curves, even for producers with difficult non-monotonic behavior.
Finally, the method is described to rank dynamically the reconciled model realizations for identifying the highest potential, to capture bypassed oil and implement IOR solutions. The main features include use of fast streamline simulations to calculate dynamic model responses (e.g. recovery factors), evaluate their dissimilarity with pattern-recognition techniques and assigning of a few realizations, representative for production forecasting, to full-physics simulation.
AAPG Search and Discovery Article #90155©2012 AAPG International Conference & Exhibition, Singapore, 16-19 September 2012