Three-Dimensional Unstructured Grid Geological Modeling Method and Application of Fractured Oil and Gas Reservoirs

Traditional geological modeling methods mainly rely on structured grids. Structured grids are easy to handle and have high computational efficiency, but it is difficult to accurately describe fractured oil and gas reservoirs with highly discrete and irregular natural fractures. Due to the strong randomness of fracture development in fractured reservoirs, based on the geostatistical method, a probability model of parameters such as three-dimensional fracture spatial coordinates, lengths, azimuth angles, and dip angles was established, and a random fracture generation algorithm based on the probability model was developed, achieving the random generation of the three-dimensional spatial geometric shape of fractures. To enhance the accuracy of three-dimensional geological modeling with natural fractures, a kind of quasi-triangular prism grid discretization method was established. This grid has an unstructured feature in the horizontal direction, which can accurately match the irregular shape of the fracture distribution, and retains the structured characteristics in the vertical direction, meeting the demand for fine vertical division of the reservoir. In response to the difficulty in establishing a corresponding relationship in the vertical grid division when the fractures have an inclination angle, the Floater conformal parameter transformation was adopted, maintaining the consistency of the topological structure of the grids at each formation surface, solving the problem of complex cutting calculations required for inclined fractures in traditional unstructured geological modeling, and significantly improving the computational efficiency. The fracture mesh and volume mesh in the constructed three-dimensional geological model are fully compatible and can be used as the mesh input for high-precision numerical simulation. Finally, taking the modeling of the actual shale oil platform in Longdong, Changqing as an example for verification, it was confirmed that the modeling method was accurate and reliable.