Predicting Cumulative Gas Production in Fractured Wells Using Machine Learning Models

This study investigates the application of machine learning models in predicting accumulated gas production from hydraulically fractured shale gas reservoirs. A dataset, generated through computer simulations of reservoir behavior, was used to train and evaluate several machine learning models, including Linear Regression, Polynomial Regression, Support Vector Regression, and Artificial Neural Networks (ANN). The simulation stage validated the model's consistency with expected reservoir performance, particularly with respect to production rate curves, accumulated production, and pressure profiles. Notably, fracture width was found to have no significant influence on the final production volume under the given petrophysical conditions. A selection of geometric variables related to the fracture was then used as input for the machine learning models. Among the models tested, the Artificial Neural Network (ANN) with a four-layer architecture and fifty neurons demonstrated the highest accuracy, with an absolute relative percentage error of (7.09±9.75)% during validation. This result suggests that machine learning models can provide comparable predictive performance to commercial reservoir simulation software, with significantly reduced computational time. The study highlights the potential of integrating machine learning techniques with hydraulic fracturing processes, offering a promising avenue for optimizing reservoir management in the energy industry.