Multi-Point Geostatistical Sedimentary Facies Modeling Based on Three-Dimensional Training Images

As an important modeling parameter in multi-point geostatistics, training images determine the modeling effect to a great extent. It is necessary to evaluate and optimize the applicability of candidate training images before modeling by multi-point geostatistics. Conventional two-dimensional training images can’t describe the overlapping relation of sedimentary facies in space, and there is a deficiency in describing the event relation of single data. This paper puts forward a new training image optimization method. The basic idea is to arrange and analyze sand bodies filled with sedimentary sand bodies in point dams and river channels in different periods. The method of obtaining three-dimensional training images is to use sand thickness maps and sedimentary facies maps for spatial constraints. The simulation test shows that compared with the sedimentary facies model obtained from two-dimensional training images, the sedimentary facies model obtained from three-dimensional training images through multi-point geostatistics has high compatibility and is more in line with geological understanding. On the basis of fully understanding the development characteristics of the sedimentary system and quantitative geometry of sedimentary body in the study area, sedimentary microfacies models based on sequential indicator simulation method and target simulation method are established, respectively. By comparing and analyzing the sedimentary microfacies models established by three different methods, the results show that the multi-point geostatistics can stably present the planar distribution characteristics and overlapping spatial relationship of sedimentary microfacies and reproduce the complex spatial structure and geometric shape of fluvial facies. The model established by this method is more in line with the geological sedimentary model.