Data Assimilation under Geological Constraints

An important key in reservoirs engineering is the development of reliable reservoir models with high predictive capacity of production behavior in the existing and new drilled wells. The reservoir predictability is the starting point for guiding the production plan to an optimum. In order to achieve this, the internal geological description of the reservoir plays a crucial role. When referring to the internal geological description of the reservoir, one may think of a description of the spatial distribution of the petrophysical properties (permeability and/or porosity) or more natural, on a description of the spatial distribution of the geological deposit formations (facies). The prior modeling of the facies distribution is carried out using geostatistical tools in specialized modules, defining thus, the geological simulation model and its uncertainty. The choice of the geological simulation model for facies field simulation takes into account the geological particularities of the reservoir. Irrespective of the geological simulation model used, the poorly knowledge of geology leads to an initial reservoir model whose parameters are uncertain. This prior uncertainty is usually so large that the predictive capacity is limited. To improve the predictive capacity, the prior uncertainty is reduced in an Assisted History Matching (AHM) process. To be able to couple the geological simulation model with an AHM method, a parameterization of the facies fields is needed. The research objective of this thesis was to elaborate reliable parameterizations of the facies fields in a such manner that, introduced in an AHM process, the updates to be able to: match the past data within the model and measurement uncertainties; incorporate all types of information (different measurement types) and preserve geological realism (consistent with the initial geomodel).Mathematical PhysicsElectrical Engineering, Mathematics and Computer Scienc