Reservoir Stochastic Modeling Based On Pattern Reconstruction

Reservoir stochastic modeling is the key technology in fine reservoir characterization. Based on the good parts of object-based modeling, surface-based modeling, process-based modeling, variogram-based modeling and multipoint geostatistical modeling, a new modeling method based on pattern was proposed. This method treats pattern as the geometry of geobody and time-space distribution properties controlled by specified rules, treats reservoir modeling as pattern reconstruction constrained by conditional data.Firstly, this paper described this method in detail, the key algorithm was given. Under the guidance of this method, the study and practice was processed in microscale and macroscale reservoir characterization. In terms of mircoscopic reservoir characterization, the complex geometry of rock grains was constructed by extracting skeleton axis and overlapping multiple spheres along skeleton axis. The distribution of rock grains was controlled by 3D possion process. In the aspect of fluvial reservoir modeling, the bank retreat formulation was used to simulate planform of meandering river. By recording indexes of different types of sediments on 3D grid, the point bar and lateral accretion interlayer were modeled accompanied with evolution of meandering rivers. In the aspect of deep-water reservoir modeling, with the constraint of palaeogeomorphic map, the flow path of sediment gravity flow was determined by slope and the sheet-form lobe was constructed by piling multiple flow event sediments under the control of compensation deposition.