Investigation of Stability of Three-Dimensional Condensation Fronts in Porous Media

Abstract

SPE Members Abstract The stability of three-dimensional condensable gas (steam) displacements in porous media is investigated by precise simulation using a recently developed three-dimensional front tracking model. This model uses a finite element method with an adaptive grid redistribution scheme that accurately represents the front as a moving discontinuity. Frontal displacement is evaluated by precise calculation of the velocity field around the front. Abrupt changes in volumetric flux across the front is a consequence of conservation of mass and energy, which enhances stability. Model results show excellent agreement with data in the literature. Displacement patterns and sweep efficiency correlations are shown for selective flooding patterns using steam. Introduction Injection processes of all kinds, and thermal processes in particular, tend to develop fairly distinct fronts that move through the reservoir. Although diffusion, dispersion, heat conduction, etc. tend to smear these interfaces, they generally propagate in-tact. It is important to describe mass and energy transfer in regions between these fronts; however, production of oil is in large part governed by the location and movement of the fronts themselves (Myhill and Stegemeir, Vogel).Calculation of convection-dominated flow on fixed meshes using conventional simulation techniques tend to yield numerical solutions that oscillate, especially in regions where the solution changes rapidly, especially if the characteristic thickness of the front is much smaller than the mesh size used. Numerical dispersion and oscillations can be decreased if the mesh is sufficiently fine and well graded. However, this means higher cost of computer storage and time such that full-field simulations may not be practical. Therefore, although conventional reservoir simulators can model physical phenomena very accurately, they do not always adequately model hydrodynamic movement of sharp fronts. Even with vector-processing supercomputers, application of thermal simulation models can usually be made only on small fractions of field projects, or must utilize such large grid cells that spatial definition is inadequate. There are computational methods that take into account the highly convective nature of the problem. Some physical processes, such as diffusion occurring around the front, although important to overall behavior of a process, do not greatly affect overall frontal movement. Their effect can be sometimes de-coupled from the hydrodynamic part of the problem. On the other hand, other phenomena such as condensation, combustion, and gravity and mobility changes are actually part of the hydrodynamic problem and must be included in the hydrodynamic solution. For problems in which local physical phenomena can be de-coupled from the hydrodynamic movement of the front, one approach is to model hydrodynamic movement of the front first by only incorporating important physical phenomena that have a strong effect and later superimposing important other physical phenomena over the movement of the front. P. 19^