Predictive Modeling And Application Research Of Monitoring Indicators And Development Indicators In Oil Block

It is discussed in this dissertation how to analyze and predict the development indices of the oilfield through the monitoring indicators of the oil wells, which focuses on the high water cut period of domestic oil fields, the instability of formation pressure and the imbalance of producing degree of oil reservoir and other issues. At first, the dynamic monitoring data is processed, which is extracted from the monitoring instrument using the single well dynamic monitoring technology. Then the block monitoring dynamic characterization is studied through the single well process data, and block monitoring index prediction method is studied, too. Finally, the relationship between block monitoring dynamic and developing dynamic is set up, and the prediction of the block development index based on the monitoring index is implemented according to this association. At present, although the development of the technology of single well monitoring is of rapid growth, these techniques are more used in the evaluation of single well interpretation, and can not guid the dynamic analysis and mining measures of the whole district.To forecast the development situation of a district of the oilfield through monitoring indicators, we construct a method to the represent characters of the monitoring indicators of a district by the data of the indicators of the single wells in this district. The correlative relation between the monitoring indicators and the development indices of the district is discussed and the input-output relationship is simulated numerically. The simulation model is applied in the prediction of the development indices, which is the foundation of the implement of the measures to optimize the production of the oilfields. In this dissertation, there are several aspects studied as follows.(1) The problem of predicting the oil block development index based on single well monitoring index is proposed, which is not only relied on the improvement of dynamic monitoring technology, but also relied on the application of monitoring index, especially its application in the block development and dynamic prediction. These related theories are further studied and improved, and how to effectively select the prediction model is investigated, as well as effectiveness, feasibility and limitation of these methods to improve the prediction accuracy.(2) A data analysis and processing method is proposed in this dissertation based on the monitoring data of the single wells. As the monitoring data of the single wells is disturbed and incomplete in the database, the combining interpolation method is utilized to complete the data series.(3) The monitoring indicators of the district are represented by the indicators of the single wells. The representation of the water intake capacity, oil deliverability, liquid-producing capacity, seepage capacity and formation pressure maintenance of a district are constructed through the data of single wells, which are obtained from the injection profiles, production profile, well test representation and pressure monitoring. (4) The correlation between the monitoring indicators and the development indices is analyzed to get the input-output relationship between them. The weight of the monitoring indicators can be adjusted according to the correlation coefficient when making the prediction.(5) Several methods are employed to predict the monitoring indicators such as time series prediction, grey time series prediction, random time series prediction and fuzzy time series prediction. These methods are combined in Chapter4to compensate the one-sidedness of the methods aforementioned and achieve higher accuracy.(6) Three types of function simulation models are constructed in Chapter5to get the input-output relationship between the monitoring indicators and the development indices, which includes the systematic simulation models, the stochastic simulation model and the fuzzy neural networks. The predicted monitoring indicators are input of these models and the output is the development index to be forecasted. Based on the practical application, we can see that the model based on a dynamic fuzzy neural network is self adaptive and time varying, the prediction results of which are the best.Above all, the monitoring data of the single wells can be utilized in wider areas, which is helpful for us to hold the trend of the future of the development performance of the oilfields. The results of this dissertation are of valuable references for the implementation of the measures which are carried out to improve the production of the oilfield districts.