Research Of The Ensemble Kalman Filter For Reservoir Parameter Inversion Method

Reservoir numerical simulation is to use numerical method to solve the seepage equation,with the purpose of simulating the underground flow of oil,gas and water.History matching as the basis of predicting the reservoir development is a very important part of reservoir numerical simulation.History matching which is based on static model,uses production data such as oil production rate and bottom hole pressure to correct model parameters,such as permeability and wellbore parameter.It is a rather complex inversion problem.Due to the limitation of people's understanding about geological conditions,model parameters which is used in the calculation may not reflect the real situation of the reservoir accurately.Therefore,there is a certain difference between the simulation results and the actual reservoir dynamic situation.The traditional history matching is lack of the systematicness and standardization,and exists some problems such as high cost and low efficiency.Therefore,a series of automatic history matching methods are developed to realize automatic adjustment of the reservoir model.With the development of digital computing technology and modern optimization algorithm,ensemble Kalman filter(EnKF)as a new automatic history matching method,has been widely applied in recent years.In this dissertation,the ensemble Kalman filter method is combined with layered history matching to carry out the application research on layered history matching for the first time.The state vector of the initial ensemble are normalized,layered data and well data are taken as constraint data respectively to estimate the permeability field of the reservoir.The results show that it can reduce the simulation time and improve the accuracy of history matching.In the end,a new inversion method of large-scale fracturing models is proposed.We preset the half length of fracture and inversion parameters.Within the numerical range of parameter setting,the initial implementation models obeying parameter distribution characteristic of reservoir are generated by using randomized algorithms.The results show that the ensemble Kalman filter can inverse the parameter of large-scale fracturing with effect,and it would be effective in the application.The main contents are as follows:In the first part,we illustrate numerical simulation and well test analysis,and several methods of parameter inversion are introduced in detail.In the second part,we give a detailed introduction about the development and theory of history matching in numerical simulation.In the third part,we start with the background and theory of Kalman filter,then we present the theory of ensemble Kalman filter.In the fourth part,considering the automatic history matching,we research the application of ensemble Kalman filter in reservoir.In the fifth part,a brief description of large-scale fracturing is given firstly.Then,based on ensemble Kalman filter,the parameter automatic inversion for large-scale fracturing with regard to the horizontal well is studied in detail.The method we proposed in this dissertation can enhance the automaticity of parameter interpretation,improve the efficiency of history matching.