Some aspects of production data integration in reservoir modeling

The realistic description of reservoirs for the purpose of flow simulation is a longstanding problem. The true reservoir description will never be known. The integration of dynamic production data reduces uncertainty; the reduction in uncertainty is proportional to the amount of historical data and the interference between the wells. This thesis develops methods to incorporate historical production data into numerical reservoir models.; The first problem tackled is the simultaneous inversion of porosity and permeability using a geostatistics-based integration algorithm. The algorithm is developed and a numerical code is implemented. Some illustrative examples are presented and sensitivity issues are analysed.; Production data informs the nature of heterogeneity in the reservoir. The inversion algorithm is modified to permit inversion of variograms. The predictive capability of the reservoir models is improved with the updating of the variogram through inversion. The algorithm was also extended to invert for fault zone properties. The fault locations are assumed known through seismic and the properties are inferred with the available production data. The advantages and the limitations are identified.