Research On Optimizing Wells Or Layers And Productivity Predication For Fracturing In SaErtu Oilfield

There are some problems in the process of select fracture wells and productivity predication, such as subjectivity, inefficiency etc. This paper studied on methods of select fracture wells and productivity predication by using fuzzy theory mathematics and neural networks.Using mathematics and statistics, collected and calculated 15 potential factors which affect fracturing. Appling grey relational analysis and single factor analysis methods, I definited the main factors which influence the effect of fracture, they are water content, producing pressure, oil production, porosity, permeability, effective thickness of small layer and quantity of sand. Using the AHP methods, I definited the weight of the main factors which influence the effect of fracture, the results were confirmed to be reasonable. Established models of optimizing fracture wells by using fuzzy-diagnosis method, grey relational analysis, fuzzy recognition and fuzzy synthetical analysis. And showed how to use these models by calculating examples which has seven fractured wells data, the example's results proved that these models were practical. Acdroding to the field data, I established appropriate network structure for experimental blocks, it can predicate the productivity of fracture,and the compliance rate of relative error which less than 20% is 71.4%,achieved good results.Throgh this paper, we can enhancing the understanding of the geological characteristics of fractured blocks, achieving the goals of improve the level of the reservoir fracturing subdivision; Improving the direction of program design by optimizing reasonable fracturing, and achieving the goals of optimizing fracturing construction. Guided us to reducing the proportion of inefficient wells, improving the usage of remaining reserves and fracturing effect.