| This thesis is divided in two parts. The first part presents the implementation of a three-dimensional semi-analytical simulator based on streamtubes, and the second part presents an upstaging methodology based on the semi-analytical method presented in the first part.;The semi-analytical simulator is a fixed streamtube method that accounts for changes in the mobility during a two-phase flow by varying the flow rate in each streamtube. The method is very fast because it just requires one pressure solution, and it is accurate too for a large number of situations. The results of the method are not very good in displacements with favorable mobility ratios, because for these cases the assumption of small changes in the streamtube geometry along the fluid flow (fixed streamtubes) is not true. Several cross-section, areal and three-dimensional cases were tested and the results compared with the results from Eclipse, a conventional finite difference simulator, and 3DSL, a streamline simulator that periodically updates the streamline geometries.;The second part, and the most significant contribution of this thesis, presents an innovative upstaging methodology based on the semi-analytical simulator developed in the first part. The methodology generates a coarse grid based on streamlines and isobars, whose upscaled properties are accurately calculated using the properties of each streamtube that constitutes the coarse block. Before this curvilinear grid can be used in a conventional finite difference simulator it must be transformed into a Cartesian grid. This kind of grid presents several advantages over conventional coarse grids. It is more defined around wells and in highly permeable regions, it conforms to internal and external boundaries, and as it follows the main flow direction the use of a full tensor for permeability is not necessary. Several example cases were run and the results compared with the results from fine grid simulations, and coarse grids upscaled in a conventional way. Tests showed that the results from the steamtube coarse grid had an excellent agreement with the fine grid solution. Although this upscaling methodology is heavily based on the semi-analytical simulator, it provided good results even for situations in which the semi-analytical simulator is not supposed to work, such as displacements with favorable mobility ratios, and for problems with gravity and compressibility. |
