Analysis of Fractured Sections in Shale Gas Wells Based on PCA - Logistic Regression Model

Abstract

In order to optimize the layout and economic exploitation of horizontal fracturing wells and completion in shale gas reservoirs, we propose a model for evaluating shale gas fractured sections based on an improved principle component analysis (PCA) algorithm with logistic regression. The 229 gas production sections in 22 fractured shale gas wells in the main block of the Fuling Shale Development Demonstration Zone were selected, and PCA is used for dimensionalite reduction. According to the PCA results, 6 key parameters are chosen to determine the productivity of fractured wells. Taking the probability distribution of high production after fracturing as the research objective, a logistic regression discriminant model was constructed using the dichotomy method. The prediction results show that the model has 82.1% accuracy and is reliable. The model can be used to classify and gas wells to be fractured, and it provides guiding significance for reasonable optimization of well sections in the area selected for fracturing.