| Reservoir model generation for numerical simulation requires data such as porosity, permeability, initial water saturation, residual oil saturation, capillary pressure functions, relative permeability functions, and other petrophysical information. These functions and parameters are necessary to estimate production rate and ultimate recovery and thereby optimize reservoir development.; The available data are varied including well logs, production data, core-scale measurement of permeability and porosity, as well as pore size and type proportions. Measurements are made at different length scales and the correlation among them is generally not known. The varied length scales associated with measured data causes inconsistency during construction of reservoir models and, thus interpretation of reservoir performance. Consequently, we must integrate all the petrophysical data available at appropriate length scale when formulating a reservoir model. In this thesis, a methodology is developed to better link pore size, shape, and type, with core scale permeability and relative permeability measurements as well as well log data. A producing diatomite reservoir is chosen for study because of the large volumes of oil in place. The method described is equally applicable to other reservoir types such as sandstones, chalks, and other carbonates.; A unique 3D-pore network model of two-phase flow was developed to compute permeability, relative permeability, and capillary pressure curves for both drainage and imbibition cases. First network model predictions are verified with known sandstone data. Once the model's behavior is verified in a well documented pore network system, the model is applied to diatomite reservoirs. The diatomite model is constructed using pore-type proportions and throat-type proportions obtained from image analysis of petrographic thin sections and capillary pressure curves for diatomite cores. Multiple pore types are used and each pore type has a unique pore size and throat size distribution. Our network results present good matches for all properties when compared to experimental measurements.; Correlation of network model results to well log data is discussed, thereby interpolating limited experimental results across the entire reservoir column. Importantly, our method has potential to predict petrophysical properties for reservoir rocks with either limited core material or for which conventional experimental measurements are difficult, unsuitable, or expensive. |
