Research On Automatic History Matching Under Strong Heterogeneity Reservoirs

The oil horizons of sandstone reservoirs are in a long span,and the distributions of the oil-bearing areas are wide and scattered,strong heterogeneity between reservoir stratums.The geological situation of this kind of reservoirs has became more complicated because of long-term waterflooding,accurate identification of the characteristics of reservoirs has became the premise of designing reasonable oil field development plan and improving the recovery efficiency,reservoir numerical simulation and automatic history matching technology has became one of common methods to describe reservoirs today.Therefore,reasonable choice for reservoir automatic history method can acquire better the information of reservoir characteristics.Based on the bayesian theory this paper has established reservoir history matching mathematical model,and reduced the parameters by using the method of PCA to improve the computation speed.For the reservoir inversion of high permeability channel model,this paper has selected PCA feature extraction method to deal with the fitting parameters,and solved the objective function of history matching by using the SPSA and SGFD method respectively.Sedimentary facies,used as a kind of attribute parameters to describe reservoir geological lithology characteristics,can effectively reflect heterogeneity characteristic of reservoirs.Therefore,this paper has introduced the discrete sedimentary facies reservoir models based on the level set theory for automatic history matching.Firstly,inverse the sedimentary facies,then inverse the continuity parameters.For complex heterogeneous reservoirs,this paper present the method of PCA combined with DCT for automatic history matching,which has guaranteed the main features information of the heterogeneity of reservoir parameters and enhanced the robustness of inversion.In addition,this paper has also used this method for the test of automatic history matching of large actual reservoirs.The results show that: in the process of the sedimentary facies inversion by using PCA dimension reduction method,the sedimentary facies boundaries extend gradually with the increase of the iteration step until connected,eventually be abe to realize two phase and multiphase inversion well;in the early stages,SGFD method has a faster convergence speed and a more accurate result of inversion than SPSA;by using PCA and DCT method can achieve better parameters field than SVD method;The method applied for large actual reservoir history matching can obtain more accurate inversion results.The results of the study provide the theoretical support for the identification of large complex heterogeneous reservoirs.