Research On Fracturing Process Optimization Based On Data Mining Method

Fracturing is the main stimulation treatment of low permeability reservoir. A reasonable fracturing design provide technical guidance for fracturing construction, and the core of a fracturing design is the optimization of fracturing parameters. Traditional method for fracturing design is a cumbersome process, and that problem is hard to solve in a traditional way. On the other hand, the data which growth in oil and gas development field is booming like a blowout. But the valuable information can't be obtained from vast amounts of data. With the advent of the era of big data and popularization of data mining which application in the industrial field, the optimization method of fracturing design which based on data mining is turning out to be a task of value.This master thesis taking a reservoir of glutenite which in western China as the research object, based on geology of reservoir, petroleum engineering, data mining and MATLAB programming knowledge, and treat production forecast and fracturing parameter optimization as the target of this study, obtain the following results and understanding.?Through data acquisition of geological and fracturing parameters, established a Fracturing Database on account of Access which contain the basic data of 174 wells. This Fracturing Database supports the function of data storage and real-time updates.?Based on the gray correlation analysis method, it come to the conclusion that the correlation degree of influence factors of production in descending order is effective thickness, permeability, pad fluid ratio, average intensity of proppant, pressure coefficient, displacement, porosity, reservoir temperature, average sand ratio.?Based on support vector machine theory, established a model for the production prediction of glutenite reservoir exploration wells after fracturing. Excluding the wells which production is less than lt/d, the average error of the production prediction is 6.44%.?On account of decision function of yield prediction model, established a model for the optimization of fracturing parameters by using ergodic optimization method for solving the optimal production and getting optimal fracturing parameters.?Support vector machine theory and mathematical model of production prediction and fracturing parameters optimization based on the front, on account of Matlab GUI programming platform developed Fracturing Optimization System which specially suit for exploration wells of low permeability glutenite. This system can more accurately predict the production after fracturing, and optimize fracturing parameters. Besides, this system can also realize the function of querying geological and fracturing parameters of wells, recommending sealing mode, analyzing effect of fracturing factor etc. All those conclusions and functions can play core roles of a fracturing design.