Application Of Stochastic Simulation In 3D Geological Modeling

The key technology of establishing reservoir geologic model is to interpolate and extrapolate the estimates of reservoir parameters in and outside the known control points according to the data of the known control points. That needs to seek and select a numerical calculation model which can most accord with the actual spatial variation of reservoir geological variables,to realize the correct quantitative description of spatial variation of reservoir characteristics. At present,there are many specific modeling methods which can be divided into two kinds. One kind is deterministic modeling method,another kind is the stochastic modeling method. The paper focuses on the research of Stochastic modeling (Stochastic Simulation) method. Combining the domestic research achievements of others,several common modeling methods are compared and evaluated,and application conditions of the methods as well as the advantages and disadvantages of them are also summarized in the paper.Geostatistics,especially Stochastic modeling technology is a key technology of 3D geological modeling. How reasonable and efficient use of the theory,such as geostatistics and stochastic simulation,to build 3D geological modeling is becoming the focus of modeling research.On the basis of deep research on the related theory of geostatistics ,and on the modeling process and key algorithms of reservoir property,3D properties model of XX gas field is built by using mainstream modeling software Petrel. The test results of model analysis show that the porosity attribute key section analysis and the geological understanding are basically same. In addition,on the basis of deep research on property modeling module of mainstream modeling software,we have learned and mastered Gslib (geostatistics package) invented by Stanford University,and have designed and implemented a property modeling module,with Gslib as its core,which interface with domestic reservoir 3D geological modeling software GeoModel,and have tested the model of actual work area. Eventually,compared with modeling results of other mainstream modeling software,the results we obtain are better.