Dynamic Prediction And Application Research Of Water Flooding Oilfield Based On System Modeling

Reservoir performance prediction plays an important role in realizing the current status of the reservoirs. Accurate and effective forecast of the reservoir performance is beneficial for the decision makers to set out or adjust the exploitation schemas, which could be guidance of the reservoir exploitation. In recent years, many water flooding oil fields in China and abroad have stepped into the mid-late period of exploitation. In this period, the water cut is growing rapidly, the oil production is declining fast, which leads to more difficulties in reservoir exploitation, and the reservoir conditions are more complex. The traditional prediction methods are hard to meet the present requirements of the reservoir performance prediction. Aiming at solving this problem, this dissertation presents the following aspects of research on the reservoir performance prediction for the water flooding oil fields based on the system modelling method.(1) The characters and the primary laws of the dynamical reservoir system are analyzed, and the traditional methods to predict the reservoir performance are reviewed. Secondly, the possible impact factors of the oil production are determined based on the theories of reservoir engineering and oil field exploitation, and the redundant factors are kicked out based on the analysis of equivalence, effect-cause and procedures.(2) The quantitative analysis and qualitative analysis are used to select the proper factors. Thirdly, the linear and nonlinear multivariate prediction models are built based on the philosophy of system modelling and the Arps decline model.(3) Squares method and the genetic algorithm are employed to estimate the parameters of the above models, and the solutions are given. The models have been validated in the numerical experiments. At last, the proposed linear and nonlinear prediction models are used to predict the production of the real world oil fields. In the case studies, the factors selected based on the quantitative analysis and qualitative analysis are used as input variables, the experiments are carried out using the real world production data. The results indicate that the linear model performs better when the relationship between the oil production and the impact factors tends to be linear, and for the nonlinear relationship the nonlinear model performs better. But the nonlinear cases are more common in the real applications, thus the nonlinear model is preferred for the real world oil fields.To sum up, the prediction models proposed in this dissertation can adapt the features of the oil fields in China and overseas, and they are eligible to predict the oil production accurately, which is beneficial for the decision makers to set out or adjust the exploitation schemas.