Abstract
Unconventional gas reservoirs, characterized by their complex geologies and challenging extraction conditions, demand innovative approaches to enhance gas production and ensure economic viability. Well stimulation techniques, such as hydraulic fracturing and acidizing, have become indispensable tools in unlocking the potential of these tight formations. However, the effectiveness of these techniques can vary widely depending on the specific characteristics of the reservoir. In this context, a data-driven approach to assess well stimulation practices offers a promising avenue to optimize recovery processes and reduce uncertainties. This paper presents a comprehensive study that leverages the power of big data analytics and machine learning to analyze and improve well stimulation strategies in unconventional gas reservoirs. By systematically gathering and processing vast arrays of geological, operational, and production data, this study aims to identify patterns and correlations that can predict stimulation outcomes more accurately. The ultimate goal is to develop a robust framework that allows for tailored stimulation designs based on the unique properties of each reservoir, thereby maximizing efficiency and minimizing environmental impacts. This study introduces a new procedure for assessing well stimulation performance, which involves analyzing the EUR through Duong's model, calculating the key performance indicator of the treatment, and establishing a data-driven model to predict the treatment KPI.