Fracture Conductivity Prediction Based on Machine Learning in Shale – Decision Tree and Random Forest Regression

Abstract

Abstract: Hydraulic fracturing extracts oil and gas from deep underground, with fracture conductivity being crucial for efficient production. Traditional lab techniques for measuring conductivity are costly and time-consuming. This paper explores using machine learning, specifically decision tree and random forest regression, to predict fracture conductivity based on experimental data like Poisson’s ratio and proppant size. Optimizing these models can enhance hydraulic fracturing efficiency in shale formations.