Methods Of Reservoir Fine Modeling On Meandering River

During the exploration and development period of oil and gas field, the aim of reservoir description is to build3D quantitative geological model, which is the further development of oil and gas exploration and embodiment of the reservoir to a higher stage of development. There are mainly two ways of building a geological model, namely, deterministic modeling and stochastic modeling. Deterministic modeling methods attempt to predict the determined, unique and true parameters of the reservoir on the basis of determined information. However, with imperfect information and more complex reservoir structure configuration, it is difficult to grasp the authenticity of the reservoir. So the stochastic modeling technologies have been applied popularly, using the random function theory and stochastic simulation method,given a variety of possible predictors of outcome on cross-well location. And stochastic modeling methods have great advantages in many aspects such as describing reservoir heterogeneity, integrating various information and evaluating uncertainty.Reservoir stochastic modeling is the hot and difficult for the last20years. Using stochastic modeling methods to establish a continuous and real curve shape reservoir targets is the problem to be solved. Object-based method can’t solve the problem of reproducing continuous and curvilinear reservoir. Pixel-based method can’t solve the problem of conditioning. The recent development of Multi-point geostatistics includes the advantages of target-based and pixel-based methods, and overcome their defects, represents a new direction in the development of stochastic modeling of reservoir. But how to create a smooth training image remains to be studied. The above modeling methods can characterize the space distribution of reservoir, but rarely integrate information related to depositional process. Result is difficult to description the relationship between the deposition process. Because these methods may reproduce spatial statistics inferred from available condition data, but ignore the relationship and depositional process of the objects. Therefore the rusulting models, which lack geologic realism, may not be representative of reservoir heterogeneity.Given the current reservoir stochastic modeling methods lacking the ability to adequately integrate geological information related to depositonal process, process-based stochastic modeling method arises at the historic moment. This method can consider some factors such as terrain, gravity, friction, dynamics, deposition and erosion rate, as well as integrate random processes to consider reservoir uncertainties. However, because the geological process which control sediment formation is very complex, various control factors are constantly changing as time goes, As aresult, it is difficult to assign various simulation parameters reasonably. Therefore, it is just at exploration stage to reproduce sediment distribution purely from various control factors which relate to sedimentary process. Another approach is to try to simulate a single deposition process, simplifying its specific forming process. The typical example is simulating river’s lateral migration.Based on predecessors’achievements, Pyrcz has made a deep research on this method, and has developed a program named Alluvsim to simulate fluvial reservoir. Process-based stochastic modeling method, which can effectively integrate information related to depositonal process and some proir geologic knowledge, quantitatively reproduce the geometry and spatial distribution of geobodies by simulating their depositional process. At present, this kind of method, which is mainly used in the fluvial reservoir, can simulate sedimentary peocesses such as avulsion, meander migration, depicting the spatial changes of fluvial architectural elements over time. Meanwhile, process-based stochastic modeling methods make use of a series of geomtry parameters through prior knowledge which can quantitatively characterize the geometric shape of reservoir architectural elements. So the resulting model could be more realistic. Therefore, process-based stochastic modeling method can build a more realistic3d architectural element model, improves modeling accuracy and reduces the uncertainty of reservoir prediction.This paper systems analysis principles of Alluvsim algorithm, containing the geometry parameters of the fluvial reservoir architecture elements, streamline operations, program parameters, well data condition etc. This algorithm move streamline associations to honor wells in the horizontal and vertical, if the number of well is small can easily realized, otherwise condition can not realized. In fact, if the distance between well point and streamline is too far, it can generates a new streamline through well point, which is similar to the method of moving streamline; if the distance between well point and streamline less than average width of river then change the local river form, so can honor wells easily; if the distance between well point and streamline less than the maximum channel width and greater than the average width, so can use rotation conditioning, which can be subdivided into source rotation conditioning, end rotation conditioning and continuous rotation conditions. Based on the conditions of data and the river center line in the space, three methods are proposed to improve the conditioning, that is, local conditioning, rotation conditioning and overall move conditioning.useing improved methods to simulate, model results indicate that improved methods not only ensure the geometry of streamline but also honor wells, conditions effect is better than the original method. The revised method can directly serve the fine three-dimensional model construction of reservoir.