Based On Multi-point Geostatistical Facies Random Simulation Study

Traditional two-points geostatistics is based on the variation function as a tool for reservoir stochastic modeling to characterize the spatial structure of the reservoir. However, variation function only reflects the spatial correlation between two points, and can't adequately describe the continuity and variability of the complex spatial structure. In the case of less wells or it's data is little, fewer points are used to calculate the experimental variogram, not accurately reflecting the spatial correlation between two points.Using "training image" instead of the variogram in the two-point method, multi-point geostatistics method applies sequential algorithm to stochastic simulation, combining with a large amount of a priori information to accurately descript the complex reservoir facies distribution. In this paper, based on the comparative analysis of multi-point and two-point geostatistics, presenting the basic principles of multi-point geostatistics and Sensim algorithm, the parameters analysis on the factors is carried out, such as multilevel grid, search radius, search angle, and the simulation results analysis on different parameter combinations is conduct to obtain the optimal combination. The obtained results demonstrate that the mesh size is connected with the size of the work area, when the grid is too small to simulation results scattered, or improvement is not obvious on the simulation results, and the selection the largest condition data effects the reasonability of the simulation results.Finally, the study area information and modern deposition research results, combining with the former parameter sensitivity analysis to set algorithm parameters, were comprehensively used to establish a reasonable training images. On this basis, multi-point simulation and sequential simulation were applied on the S oilfield's main layers of X block in the NWE basin, and the simulation results were analyzed. The results show that multi-point geostatistics method is superior to the conventional two-point method on the reproduction of reservoir space structure characteristic, and it is an important research directions on geological modeling in the future.