Predicting Shale Production Performance Through Machine Learning: The Development and Application of the Sweet Spot Quality Index

Abstract

Abstract The objective of this research is to address the significant challenges in organic shale evaluation and development by developing a graphical evaluation and ranking technique called the Sweet Spot Quality Index (SSQI). This technique combines reservoir characterization, completion strategies, and reserve estimation to identify the most favorable areas for production. The SSQI is then used with preliminary pre-production data to predict the average shale production performance using Machine Learning (ML). The goal of this approach is to optimize production and improve efficiency in the shale industry. This methodology aims to provide an accurate prediction of the average shale production performance, which can help to improve the decision-making process for shale development. The methodology used in this research involves collecting data sets from various shale plays in North America, including reservoir properties, completion parameters, operational processes, and production data. These datasets are used to create the Sweet-spot Quality Index (SSQI) using Machine Learning (ML) algorithms. The first step is to gather data on reservoir properties, completion parameters, operational processes, and production data, including the hydrocarbon in place, geo-mechanical parameters, field operations expenses, equipment and material availability, economic parameters, and actual production rates. Next, the data is pre-processed and cleaned to ensure it can be used to train the ML model. The machine learning algorithm is then trained on this data to calculate the SSQI, by using the data to find patterns and correlations that can be used to predict production performance. The final step is to validate the calculations with actual production data and use the resulting SSQI to predict production performance, this approach allows to identify the most favorable areas for shale development based on the comprehensive evaluation of the potential for shale production. The research results demonstrate the use of machine learning techniques to evaluate organic shale exploration data and identify shale sweet spots using the Sweet Spot Quality Index (SSQI). The ML models predicted the SSQI from input data of reservoir properties, completion efficiency, and operation conditions with high accuracy and reliability. The use of machine learning techniques allows for a more accurate and efficient evaluation of shale formations compared to traditional methods and can handle dynamic parameters such as oil and gas prices and operation expenses, providing a robust and reliable tool for identifying shale sweet spots. The research concludes that the developed ML models have the potential to be used in the industry as a reliable tool for predicting SSQI and identifying the most favorable areas for shale development, and further research should be done to optimize the model and test it under different conditions and with more data. The research introduces a novel and additive approach for predicting production performance in shale development by using the Sweet Spot Quality Index (SSQI) calculated by a machine learning technique controlled by actual production data. This approach provides an accurate prediction of production performance under several operational, economical, and reservoir parameters, which can greatly enhance the efficiency of shale development decisions. The use of machine learning to calculate the SSQI and control it with actual production data is a new and unique approach not used in the industry. This research provides a powerful tool for the industry to optimize production and improve efficiency in shale development, ultimately leading to more cost-effective and profitable projects.