Constructing of 3D Fluvial Reservoir Model Based on 2D Training Images

Training images are important input parameters for multipoint geostatistical modeling, and training that can portray 3D spatial correlations required to construct models. The usually obtained by unconditional simulation using algorithms such as object-based algorithms, in some cases, it is difficult obtain the directly, so a series of modeling methods based on 2D constructing models has been formed. In this paper, new method proposed synthesizing advantages previous methods. Taking fluvial reservoir P oilfield Bohai area an example, comparative study was carried out. By comparing variance function, horizontal vertical connectivity x-, y-, z-directions, style similarity, shows several mutually perpendicular images, paper achieve effect similar directly. case suitable application prospects.