摘要

An improper well pattern will have considerable adverse effects on the ultimate recovery of oil and gas considering the geological complexities usually associated with reservoirs. Designing an optimal well pattern for a given reservoir is often challenging because of following two categories of reasons: static factors including strong heterogeneities of reservoirs, the existence of outer boundaries, faults and pinch-out belts, variations of sedimentary facie physical properties; dynamic factors including the producers and injectors drilled previously, the multitudes of well patterns and the transformation among them. To overcome the difficulties of designing well patterns under complex conditions, a new method of constructing triangular adaptive well pattern is proposed in this paper. This new triangular adaptive well pattern can adjust the locations of wells spontaneously according to the conditions of reservoirs, achieving optimal effects using fewest wells. Inspired by the similarities between triangular well pattern often encountered in the industry and the triangulation of domains in computational geometry, the well-known Delaunay triangulation is employed to determine the locations of wells. By taking full advantage of the properties of Delaunay triangulation, the construction of triangular adaptive well pattern on the basis of boundaries, faults, and existing wells can be easily obtained and the number of control variables is greatly decreased in the optimization process. Therefore, a gradient-based algorithm coupled with reservoir numerical simulator is used to optimize the well pattern. Compared with conventional regular well patterns, the well pattern proposed here differs in that the scale and orientation of local flooding units are not the same in different parts of the reservoir depending on the geological conditions and the distribution of oil and water in the reservoir. Additionally, the heterogeneity of permeability is taken into account and a uniform displacement of oil in each flooding unit is realized by adjusting the locations of injectors. Detailed results are present ed for two different examples. The results show that the method proposed here can be successfully applied to the construction and optimization of well pattern for large-scale reservoirs.