| The oil industry makes decisions and formulates development plans using forecasts obtained from numerical simulation of fluid movements in deep geological formations. Building accurate and predictive reservoir models is a difficult task, requiring the representation of complex fluids, complicated rock properties and structures, as well as complex numerical treatment. History-matching in petroleum engineering is a model calibration exercise using observed field data (fluid rates and pressure) to tune a model such that the numerical simulation and the observations match.; Very often, this inversion is unconstrained and proceeds by trial-and-error thereby rendering geological models that are geologically inconsistent. That is, geological spatial correlations as well as the relation between parameters are not respected. To honor the relations imposed by geology, it is necessary to use statistics, geostatistics, and a simulation tool that provides local, flow-based information. Streamline simulation, for instance, allows one to compute the fluid paths in the subsurface and track zones where the model needs calibration. Once the zones creating mismatch are identified, a correction factor is computed based on the distance between observations and simulation results, and this correction is applied to the model and all correlated parameters. The correction is, nevertheless, not applied directly to avoid non-geological artifacts: it is used in a geostatistical algorithm that insures proper variogram reproduction and accounts for possible local hard data.; This document presents the most recent advancements in streamline-based history-matching while emphasizing the possibility of multiple scale corrections and the achievement of geological consistency in space and across parameters. Synthetic examples are shown for validation, and real field applications are presented to see the benefits and limitations of the method. The framework developed is robust, adaptable, computationally efficient, and may serve as the basis for further studies on geologically constrained history-matching. |
