A Research Of Automatic Reservoir History Matching Based On EnKF

According to reservoir geology and development condition,reservoir numerical simulation method uses mathematical model to show real reservoir dynamics and study its movement change law.History matching is an important part of reservoir numerical simulation.Due to the lack of systematic and low efficiency of manual history matching,automatic history matching has been widely used.Reservoir automatic history matching can reduce the uncertainty of reservoir geological parameters and lay a good foundation for the formulation of reservoir production optimization scheme.Ensemble Kalman Filter(En KF)is one of the most widely used automatic history matching methods.However,if the number of sets generated by Ensemble Kalman Filter method is small,there will be some problems such as inaccurate gradient calculation and pseudo-correlation,which will lead to incorrect parameter correction and inaccurate geological model inversion.An Ensemble Kalman Filter automatic history matching method based on shrinkage covariance matrix estimation is proposed for this problem.The method uses target matrix and sample covariance matrix to do weighted average,and then contraction of the covariance matrix method are used to get the optimal weights,and then the obtained new matrix replaces the initial sample covariance matrix in the ENKF.It weakens the pseudocorrelation by modifying the gradient of data assimilation,gradually updates reservoir parameters,fits production dynamics,and obtains the optimal reservoir model.The Ensemble Kalman filter and the Ensemble Kalman filter based on shrinkage covariance matrix estimation are compared by an example.The results show that the Ensemble Kalman Filter method based on the shrinkage covariance matrix estimation has certain advantages in the model geological parameters and single well production dynamic fitting accuracy,calculation efficiency and so on.