| الملخص: |
Abstract: Reservoir management requires high resolution numerical geologic models of facies and petrophysical properties. Facies are arguably the most important reservoir heterogeneity. Many geostatistical facies modeling techniques have been proposed during the years of heavily practiced geostatistics in reservoir assessment. Several aspects in current practices, outside the modeling technique itself, induce potential deficiencies in the representation of facies. This thesis develops novel tools, techniques, and understanding that support geostatistical literature and improve reservoir modeling practice. Notable features of this thesis are (1) addressing information loss in facies upscaling process through a proposed measure which captures variability on non-major facies; (2) proposing a novel inverse modeling approach to estimate shale continuity in the form of a probability distribution function; (3) introducing stochastic regridding to correct the conventional approach of nearest neighbor assignment; (4) investigating the construction of high resolution models in MPS aspect; (5) enforcing the connectivity of disparate facies units such as levees by proposing an effective sequence of dilation-erosion approach while preserving the global proportions; and finally (6) proposing multiscale ranking that optimizes the selected realization performance over different recovery settings, and introducing realization clustering as an alternative to ranking when more than one factor is representative of the reservoir performance. |