A New Modeling Method And Application For The Sandbody Architecture Of Braided River Reservoirs

Braided river oil and gas reservoirs are widely developed in the Cenozoic terrestrial basins.The frequent migration and diversion of the river has led to complex sand stacking relationships.The reservoir is internally non-homogeneous.The analysis of reservoir development is difficult.There is an urgent need to quantitatively simulate the geometric size of the constituent units of different grades of reservoirs and their interrelationships within the pan-connected giant-thick braided river complex sand bodies.It will be accurately characterized into the model to reveal the distribution pattern of seepage barriers inside the reservoir.And the reservoir configuration modeling methods at the present stage are still inadequate in simultaneously characterizing the spatial spreading of reservoir configuration.As reservoir modeling is continuously and deeply applied in the field of petroleum exploration and development,reservoir modeling algorithms are constantly updated and improved.Among the common modeling methods,the target-based approach takes the reservoir structural unit with complex geometric morphology as the basic modeling unit.The geometrical morphological parameters are determined directly from the geological features of the sediment body.A complete geological body object is simulated at once.However,the method is still inadequate in characterizing the conformational pattern and conditionality of the sand body of braided river reservoir.The simulation process of sequential instruction simulation with human-computer combined post-processing method is controlled by parameters such as variance function variation and azimuth.The spatial geometry and distribution pattern of each geomorphic body are not fully reflected.Moreover,the method varies from person to person and is highly subjective.The multi-point geostatistical modeling method combines the advantages of target-based modeling to carve complex objects and two-point geostatistical modeling for easy conditionality.However,it is difficult to construct 3D training images that conform to the reservoir configuration pattern.And it is also difficult to reasonably model the size of the conformational unit.On the basis of in-depth analysis of the depositional characteristics of ancient and modern braided river reservoir configuration and the current research status of typical geostatistical methods.The difficulties of the existing geological modeling methods in simultaneously characterizing the geometry of the heartland dam and the drop silt layer,the cut-and-stack relationship and the conditionality of well data are analyzed.This paper proposes a new modeling method for braided river reservoir configuration.The geometry,cut-and-stack relationship,and size of the braided river core dam and drop silt layer are well characterized in the model.The well data conditionalization in the case of sparse well network is also realized.The feasibility of the new method for modeling the heartland dam and siltfall layer is first tested by conceptual modeling.Then the modeling effects of different modeling algorithms are analyzed by comparing the modeling effects in actual workings.By comparing the section and plan view of the model,it can be visually seen that the new algorithm builds a more complete model of the heartland dam in terms of geometry.It is more similar to the training image built based on the a priori geological concept.The statistical comparison of the histogram of thickness distribution,variance function,and the number of grids occupied by a single connected sand body for the Heartland Dam It is shown that the new algorithm outperforms the traditional braided river reservoir modeling approach in simultaneously characterizing the geometry of the heartland dam,scale,and well data conditionality.