| This thesis presents the development and application of a three-dimensional, two-phase streamline simulator applied to field scale multiwell problems. The underlying idea of the streamline method is to decouple the full 3D problem into multiple 1D problems along streamlines. Fluids are moved along the natural streamline grid, rather than between discrete gridblocks as in conventional methods. Permeability effects and well conditions dictate the paths that the streamlines take in 3D, while the physics of the displacement is captured by the 1D solutions mapped along streamlines. In this work, the 1D solutions either represent tracer flow, waterflood displacements, or first-contact miscible displacements. Solutions for these mechanisms are obtained either analytically or numerically. If analytical solutions are mapped to streamlines, the final 3D results are free from numerical diffusion, but the method can only be applied to limited situations. By mapping numerical solutions to streamlines the method is extended to changing well conditions, nonuniform initial saturations, and multiphase gravity effects.; The streamline simulator has been applied to field scale infill drilling and well conversion problems. For a 100,000 gridblock problem, the streamline simulator was over 100 times faster than an industry standard simulator. For simple 2D miscible displacements dominated by gravity, the streamline method was almost 1000 times faster than conventional methods. The speed of the streamline method also makes it well suited to the solution of large problems. Examples of 10{dollar}sp6{dollar} gridblock multiwell problems are solved on a conventional workstation and require about 2 CPU days.; The large speedup factors in the streamline method are a result of decoupling fluid transport from the underlying grid. Instead, fluids are moved along the natural streamline paths. Moving fluids between gridblocks in conventional finite-difference models results in grid orientation effects and more importantly, time step limitations due to stability and/or convergence considerations. Transporting fluids along streamlines eliminates stability issues and the method is stable for any size time step. For the same displacement, the streamline simulator requires on average one to three orders of magnitude fewer time steps than a conventional finite-difference simulator. |
