Research Of The Dynamic Historical Matching Of Reservoir Based On Super-sphere Transform EnKF Method

The automatic history matching of reservoir is the basis of oilfield prediction and development,and it is also one of the important components of reservoir closed production optimization.The traditional artificial history matching method,which depends on the subjective understanding of reservoir engineers,can not meet the requirements of refinement,complication and quantification of modern reservoir numerical simulation.With the help of modern information technology and the development of intelligent technology,we need to find a robust and fast automatic history matching method to reduce the uncertainty of reservoir geological parameters and improve the understanding of reservoir fluid distribu tion and future dynamic changes of reservoir,to ensure efficient production of oil wel s.Based on the analysis of domestic and international automatic history matching,this paper proposes a dynamic history matching method based on Super-sphere Transform Ensemble Kalman Filter(SEnKF)method.The main contents of this paper include:(1)Research on automatic historical fitting of reservoir based on Normal Score Transformation Ensemble Kalman Filter(NS-EnKF)method.The method is applied to the Gaussian distribution,which is input to the simulator to predict the production state of the reservoir.Then,the state is updated according to the observed production value.Finally,we use the normal score inverse transformation to convert the permeability into the original distribution space to ensure the original characteristics of its distrib ution.Compared with the simulation results of EnKF and NS-EnKF,the inversion effect of NS-EnKF algorithm is better,and the geological parameters can be described more accurately.(2)Study on automatic history fitting of reservoir based on super-sphere transformation normal score Ensemble Kalman Filtering method.Considering the accuracy of the algorithm,the super-sphere transformation of the permeability field after NS transformatio n is carried out by using the advantages of simple structure,small error and easy realization.At the same time,the localization of covariance(LC)method is introduced to avoid the problem of filtering divergence in the process of aggregation update.So as to obtain more statistical information,improve the accuracy of historical fitting,and reduce the uncertainty of reservoir parameters.(3)The classical two-dimensional four production and one injection model is used to simulate and verify the operability and effectiveness of the algorithm.At the same time,the actual geological parameters of Norne field are used to analyze the algorithm.The results show that the introduction of normal score is a good solution to the non-Gaussian problem of reservoir parameters.The coupling of LC technique,Super-sphere transform and EnKF avoids the filtering divergence when the number of small ensembles is obtained and the effect of model parameters is improved.The reservoir engineer provides a more precise reference to the fine description of the geological fluid.