Research On Spatial-temporal ARX Modeling And Dynamic Programming Of Polymer Flooding

With the further development of reservoir, oil recovery is gradually reducing. How to enhance oil recovery has been the basis of output stabilization of oil fields. Polymer Flooding is one of the most important technologies of enhanced oil recovery, which has been widely used in China’s eastern oil fields. While the injection process is complex and time-consuming, and the polymer is an expensive chemical material. What’s more, the model of polymer flooding is complicated. In order to improve the crude oil production and help to make more scientific developing schemes, in this paper, the time-space decomposition ARX model of polymer flooding is put forward. In addition, considering the changing oil price and the calculating of relative permeability, the optimal control problem of new polymer flooding model is solved by dynamic programming.A crude oil price predicting method is proposed to solve the problem of oil price changing, which is based on dynamic correcting ε-Support Vector Regression Machine( ε-SVR).The hybrid RNA genetic algorithm(HRGA), with the position displacement idea of bare bones Particle Swarm Optimization(PSO) changing the mutation operator, is presented. The performance is improved compared with standard RNA genetic algorithm(RGA). A dynamic correction factor is used to correct forecasting errors. Then, to make the forecasting result more accurate, the HRGA is applied to the optimize parameters of ε-SVR. The predicting result is very good.As to the hypo-osmoticity and complexity of oil fields, the Darcy formula is further deduced considering the starting pressure gradient, capillary force, gravity and water saturation. For the inaccuracy and tediousness of relative permeability calculation, a novel calculating method on relative permeability curve is proposed based on improved RBF neural network. HRGA is applied to optimize the value of radial basis function centers in the hidden layer of RBF neural network. This method is used in the calculation of relative permeability curve. The experimental result tells us that HRGA-RBF can improve the calculating accuracy obviously. A new empirical equation is proposed to solve the relative permeability of low permeability reservoir. In this equation, an improved item is introduced on the basis of Rose experience formula and Al-Fattah experience formula by comparison and summarizing. Then the parameters of equation are optimized by HRGA. The matlab simulation is carried out on data which is obtained from the typical low permeability reservoir core 27-1 of Gu Dong well 54. The data is processed by new Darcy formula before used. The result shows that the accuracy and rationality of empirical equation is much better than that of others.To deal with the complexity and low computational efficiency of polymer flooding model, spatial-temporal decomposition method and ARX are introduced to model new distributed parameter modeling. Karhunen-Loeve(K-L) decomposition is used to model the state parameters of reservoir(such as water saturation, pressure, grid concentration) during spatial-temporal decomposition. The polymer injection concentration and time coefficient got from the decomposition are taken as the input and output information. After identified by least square method, the time iterative ARX models of all state variables are obtained. Then, we build the ARX model among pressure, water saturation, grid concentration and moisture content of production well, and identify it with recursive least-squares method. After combing the above two models, we get the polymer flooding ARX model with spatial-temporal decomposition. The accuracy of model is proved by simulation, the data is got from the model with four injection wells and nine production wells on CMG platform. In addition, the permeability calculation of CMG platform has been improved by the empirical equation promoted in this paper. In the end, in order to enhance the polymer flooding oil recovery when oil price is changing, we build the dynamic programming model of polymer flooding based on the spatial-temporal ARX model. Iterative dynamic programming is applied to optimize dynamic programming model to get the optimal injection to production scheme.