Significant world oil and gas reserves occur in deltaic reservoirs. Characterization of deltaic reservoirs requires understanding sedimentary and diagenetic heterogeneity at the submeter scale in three dimensions. However, deltaic facies architecture is complex and poorly understood. Moreover, precipitation of extensive calcite cement during diagenesis can modify the depositional permeability of sandstone reservoir and affect fluid flow. Heterogeneity contributes to trapping a significant portion of mobile oil in deltaic reservoirs analogous of Cretaceous Frontier Formation, Powder River Basin, Wyoming.
This dissertation focuses on 3D characterization of an ancient deltaic lobe. The Turonian Wall Creek Member in central Wyoming has been selected for the present study, which integrates outcrop digitized image analysis, 2D and 3D interpreted ground penetrating radar surveys, outcrop gamma ray measurements, well logs, permeameter logs and transects, and other data for 3D reservoir characterization and flow modeling. Well log data are used to predict the geological facies using beta-Bayes method and classic multivariate statistic methods, and predictions are compared with the outcrop description. Geostatistical models are constructed for the size, orientation, and shape of the concretions using interpreted GPR, well, and outcrop data. The spatial continuity of concretions is quantified using photomosaic derived variogram analysis.
Relationships among GRP attributes, well data, and outcrop data are investigated, including calcite concretion occurrence and permeability measurements from outcrop. A combination of truncated Gaussian simulation and Bayes rule predicts 3D concretion distributions. Comparisons between 2D flow simulations based on outcrop observations and an ensemble of geostatistical models indicates that the proposed approach can reproduce essential aspects of flow behavior in this system.
Experimental design, analysis of variance, and flow simulations examine the effects of geological variability on breakthrough time, sweep efficiency and upscaled permeability. The proposed geostatistical and statistical methods can improve prediction of flow behavior even if conditioning data are sparse and radar data are noisy. The derived geostatistical models of stratigraphy, facies and diagenesis are appropriate for analogous deltaic reservoirs. Furthermore, the results can guide data acquisition, improve performance prediction, and help to upscale models.