Research On Stochastic Modeling Under Constraint Of Space Information Field

The effective development of oil reservoir depends on its precise description. And as the core of the reservoir description technique, stochastic modeling runs through the whole life cycle of oil field. It aims at describing the possible distribution characteristics of the complex underground structure. It is based on the information of those known wells and uses geostatistics as a tool to achieve a series of equal probability and high precision of three dimensions reservoir models. Stochastic modeling is divided into two parts: the calculation of variation function, through information between the known wells to speculate unknown geological variable space variation characteristics;stochastic simulation, that is, under the guidance of variational function through spatial interpolation method to simulate the looks of unknown geological model.Traditional stochastic modeling only uses the known well-data, but the well-data are widespread and uneven distribution of well spacing density is small, well pattern distribution range is far less than the question of the research area, for the whole geological body huge, is only in my humble opinion, lack of overall spatial information,thus the modeling result is necessarily one-sided narrow. To solve these problems, this thesis makes full use of seismic data acquisition density of lateral coverage, and it contains a large number of advantages of spatial information, made up for the inadequacy of information between wells, from two aspects of building of variational function and stochastic simulation. This paper proposes a spatial information constraint stochastic modeling method. In this paper, the main contributions are as follows:1. In view of the using of single well-data of the proceeds of the lack of spatial information, variational function cannot effectively reflect the lateral variation of geological variables. This thesis proposes a spatial information field under the restriction of nonparametric anisotropic building variational function method. Integrated use of well-data and seismic data, the method using seismic inversion obtains spatial information field, on the inversion of wave impedance in accordance with the uniform sampling, angle and distance of the section of all levels of variation function in all directions, and then to integration of variation function, and obtains the final parameters of anisotropic variation function. The advantages of this method is that by using the seismic data good continuity, the advantage of rich spatial information, it effectivelyimproves the reliability of the variation function, and simple and convenient calculation process. And using the nonparametric method can avoid the traditional set of steps.2. According to relying solely on data from the stochastic modeling results only reproduce the macroscopic statistical regularity, it will deviate from the geological real results in different degrees. This thesis proposes a stochastic modeling method under the restriction of space information.The main characteristic of this method is to use spatial information to certain constraints. The specific parameters of anisotropic variation function as a guide is put forward in the modeling process for the above article, it fully shows the spatial correlation of geological variables. At the same time in the process of stochastic simulation creatively joined the seismic inversion steps, it is no longer simulated as an unit with horizontal section step by step. But in seismic trace by simulation, simulation and inversion will join the rest of the known data inversion of seismic trace after continuous simulation, inversion, ans repeated use of the space information, establishing a model of a high reliability and resolution.