A Framework for Predictive Modeling of Hydraulic Fracture Propagation in Tight Reservoirs

Thesis Abstract

Hydraulic fracturing in tight reservoirs presents significant challenges in predicting fracture propagation due to the complex interplay of geological, mechanical, and fluid dynamic factors. Accurate modeling of fracture propagation is essential for optimizing stimulation strategies, enhancing hydrocarbon recovery, and minimizing environmental impact. Nevertheless, existing predictive models often lack comprehensive integration of in-situ stress variations, rock heterogeneity, and fluid flow dynamics, resulting in limited reliability and application scope. This study aims to develop a robust, integrated framework for predictive modeling of hydraulic fracture propagation specifically tailored for tight reservoir conditions. The primary objectives include identifying key parameters influencing fracture behavior, formulating a conceptual model incorporating rock mechanics and fluid flow principles, and validating this model through empirical data. The research adopts a mixed-methods approach, combining quantitative modeling with qualitative validation. The quantitative component involves developing a numerical simulation framework utilizing finite element and discrete element analysis, integrated with geomechanical and fluid flow equations grounded in the linear elastic fracture mechanics (LEFM) and poroelasticity theories. The qualitative component comprises expert validation through structured interviews with 15 geomechanical engineers and reservoir engineers. The study's population includes data obtained from ten tight reservoir formations across North America with similar geological characteristics, with a sample of 50 core samples and well logs procured from industry partners. Samples are analyzed through laboratory-based dynamic mechanical testing and well log interpretations to characterize in-situ stress regimes, rock heterogeneity, and pore pressure distributions. Data collection instruments include high-resolution acoustic and optical imaging of core samples, digital strain gauges for laboratory testing, and digital logging tools. The validity and reliability of these instruments are ensured through calibration against industry-standard benchmarks, repeat measurements, and triangulation of data sources. Analytical techniques encompass regression analysis to quantify the influence of individual parameters, sensitivity analysis to identify dominant factors affecting fracture propagation, and finite element modeling employing COMSOL Multiphysics and ABAQUS for simulation. Model calibration utilizes field data, with model parameters iteratively adjusted to match observed fracture geometries from post-stimulation imaging. Expected findings include the identification of key controlling factors such as in-situ stress anisotropy, rock brittleness, fluid viscosity, and injection rate, along with the development of a computationally efficient hybrid model capable of predicting fracture length, height, and complexity across varying reservoir conditions. The model's predictive capability is anticipated to outperform existing models by incorporating heterogeneity and stress interactions, providing a valuable tool for optimizing hydraulic fracturing operations. This research contributes to knowledge by integrating principles of geomechanics, fluid dynamics, and fracture mechanics into a comprehensive predictive framework specific to tight reservoirs, addressing gaps related to heterogeneity and stress interactions often overlooked in prior models. The framework offers a systematic approach for operators to simulate and plan fracture treatments with higher precision, thereby improving recovery efficiency and reducing environmental risks. The study concludes that an integrated modeling framework enhances fracture prediction accuracy and operational decision-making in tight reservoirs. It recommends further validation across diverse geological settings, the incorporation of real-time monitoring data for dynamic model updating, and the development of user-friendly software tools based on the framework. Future research should explore the application of machine learning techniques to refine parameter estimation and adapt models to evolving reservoir conditions, ensuring continuous improvement of predictive capabilities in hydraulic fracturing operations.