Traditional geostatistical estimation and simulation use static spatial data indiscriminatingly when integrating well and seismic data without considering spatial sedimentary variation across sampling points and the points to be simulated and the influ-ence of lithofacies change on sampling values,particularly when there are strong spatial heterogeneity of reservoirs and uneven dis-tribution of wells.For tight clastic reservoirs,we propose a facies-controlled simulation method using space-variable data based on a Bayesian-sequential Gaussian algorithm.A 3D spherical model is defined to vary dynamically with the position of the point to be simulated,which changes spatial data from a 1D column to the 3D sphere.Based on the direction of sedimentary source,the data in the 3D spherical model are screened using a variogram function;meanwhile,sedimentary facies are used as the constraints to im-prove the accuracy of sampling values.The weights at the sampling points are determined using inverse distance weighting.Our method was applied to the east slope of the west Sichuan Depression and yields improved-resolution images of continuous channels which are consistent with seismic attributes.The coincidence rate between the simulation results of the blind well and the logging data was 92％,which verifies the accuracy of the method.In addition,the method is demonstrated to be practical because it applies to different hardware configurations.
维普
万方数据
