| In this research we study the sensitivity of water-alternating-gas (WAG) flows to a number of such parameters through numerical simulation. The results show that the performance of WAG displacements is strongly influenced by the characteristics of the permeability field used in the simulation when the fluid properties are fixed.;A number of procedures are available for generating stochastic random permeability fields, such as the matrix decomposition method (MDM), turning bands method (TBM), and simulated annealing (SA). Unlike MDM and TBM, SA is capable of incorporating information from a variety of sources, like cores, well logs, well test etc. and hence, generates geologically realistic fields. By a series of numerical experiments we show that, even though the SA produces realistic permeability fields, being primarily a Monte Carlo method it is computationally very expensive. A solution to this dilemma is the use of parallel computers. We present two parallelizing schemes that generate stochastic (spatially correlated) permeability fields using the so called 'heat bath algorithm' on Intel iPSC 860 Hypercube.;This research also introduces the application of recently discovered wavelet transform (WT) method to reservoir characterization. Wavelets are orthogonal families of basis functions that are compactly supported in time-scale space. The WT is used to analyze statistically non-stationary permeability fields. The application of WT presents a convenient way (an improvement over the traditional methods) of scaling-up and down of data, such as permeability, capillary pressure etc., where certain mathematical properties of the data are always preserved.;We also propose an analytical model for relating microscopic or grain level properties of a permeable medium, such as the particle size distribution (psd), amount and type of cementing materials present in the pore, to the macroscopic or representative-elementary-volume (REV) permeability. This model is developed by modifying the familiar Carman-Kozeny (CK) equation. However, unlike the CK equation, the proposed model is capable of predicting the permeability of consolidated clastic rocks. When used to predict the permeability of Garn Sandstone cores (a Norwegian oil reservoir) the proposed model performs remarkably well in a large number of cases.;Finally, we study the primary sedimentation processes of eolian deposits to understand the spatial variability of petrophysical properties in interwell regions. In an eolian deposit the sediment flux, mass of sand being transported over a unit surface area, depends on the airflow parameters, such as the shear stress, and the dune geometry and dimension. These parameters, along with the amount of sand supply from external sources, control the speed of migration of dunes. We present a procedure to determine the distribution of airflow parameters over dune-shaped objects. This procedure also solves the sand continuity equation using a method of characteristics approach. The solutions are presented as time-distance diagrams. These solutions determine the long-term configuration of a dune. Such time-varying solutions are useful in understanding the characteristics of buried eolian deposits. (Abstract shortened by UMI.)... |
