A Big Data Study: Correlations Between EUR and Petrophysics/Engineering/Production Parameters in Shale Formations by Data Regression and Interpolation Analysis

Abstract

Shale hydrocarbon production has become an increasingly important part of global oil and gas supply during the past decade. The life of projects in unconventional plays, such as shale oil and gas, tight oil and gas, coal bed methane etc., heavily depends on the Estimated Ultimate Recovery (EUR). However, the correlation to predict EUR in conventional plays becomes invalid for unconventional plays, which significantly affects the economics of relevant unconventional projects. The objective of this paper is to investigate the correlations between EUR and petrophysics/engineering/production parameters by data regression and interpolation analysis via big data mining from Eagle Ford. Furthermore, a 4-D interpolated EUR database and EUR prediction models are established based on the relevant regression and interpolation results. This study not only helps us understand the physics behind EUR prediction in unconventional plays, but also facilitates determining the viability of projects in unconventional formations from a big data perspective. In this study, petrophysics/engineering/production data from 4067 wells in Eagle Ford is summarized for analysis. Firstly, a sensitivity analysis is carried out to determine the most sensitive petrophysics and engineering controlling factors. In particular, the physics behind the EUR predictions is discussed in details. Following it, the 2-D nonlinear regression and the multivariate linear regression are applied to evaluate the relationship between EUR and engineering/production data. In addition, a 4-D interpolated EUR database is established to predict EUR based on the petrophysics parameters. The applied nonlinear multivariate interpolation methodology is the Triangulated Irregular Network based Nearest Interpolation Method (3-D). Finally, the 4-D interpolated EUR database are applied to several wells in the Eagle Ford to test its accuracy, confidence and reliability. Based on the sensitivity analysis results, Vitrinite Reflectance Equivalent (VRE), Total Organic Carbon (TOC) and Resource Density (porosity, hydrocarbon saturation and gross formation thickness) are the most sensitive and important parameters in Eagle Ford shale formation. Based on the data-mining results, effective lateral length has a positive monotonic relation with EUR; EUR increases with more proppant weight and higher true vertical depth. Frac stage and perf per cluster do not have a strong correlation with EUR. In addition, azimuth has a vague relation with EUR while drilling along the North-South orientation is the safest approach in Eagle Ford Shale. The physics behind the correlations is analyzed and discussed in detail. Finally, several DCA EURs of wells from Eagle Ford are used to test the established 4-D interpolated EUR database, and the study results show that the relative errors in EUR predictions are within 30%, indicating that the methodology in this study has great potentials for unlocking more reserves economically in shale formations. This study offers an insightful understanding of unconventional hydrocarbon production mechanism from a big data perspective, as well as a feasible and accurate method to predict EUR and evaluate projects economic feasibility in Eagle Ford. This methodology can be also applied to other unconventional fields such as Utica, Permian and Bakken Shale plays, if data is available.