| dc.description.abstract | The use of fiber-optics in reservoir surveillance can bring valuable insights to fracture geometry and fracture-hit identification, stage communication, and perforation cluster fluid distribution in hydraulic fracturing processes. However, the complexity of multiple information streams associated with realistic field data makes interpretation challenging for engineers and geoscientists. In this work, I generate distributed strain sensing (DSS)/low-frequency distributed acoustic sensing (LF-DAS) synthetic data of a cross-well fiber deployment. This data incorporates the physics governing hydraulic fracturing treatments. Forward modeling streamlines the interpretation task by exploring data richness, which has the potential to improve completion design and optimize production.
Forward modeling relies on analytical and numerical solutions. The analytical solution is developed by coupling two models: a 2D fracture (e.g., Khristianovic-Geertsma-de Klerk [KGD]) and Sneddon’s induced stress. DSS is estimated using the plane strain approach that combines calculated stresses and rock properties (e.g., Young’s modulus and Poisson’s ratio). In turn, the numerical solution is implemented using the displacement discontinuity method (DDM) with net pressure and/or shear stress as the boundary condition. In this case, the fiber gauge length concept is incorporated deriving displacement (i.e., DDM output) in space to obtain DSS values. In both methods, LF-DAS is estimated by the differentiation of DSS in time.
My simulator models classic features present in field data including: the heart-shaped pattern from a fracture approaching the fiber, stress shadow, and fracture hits. Incorporating shear stress in simulation creates strain time histories that entail complexities beyond those observed in cases in which tensile stress is the unique failure mechanism, thus highlighting the significant impact promoted by natural fractures. Moreover, a large gauge length (i.e., popular 10 m size used in the field) can mask strain data richness, distorting intrinsic characteristics of fracture systems. Fracture corridor extension signature, occasionally observed in LF-DAS field data when pumping stops, is verified in synthetic results for small pressure drop gradients, revealing that fractures continue to propagate in this scenario. Quantitatively, fracture geometry characterization is improved by estimating width in multiple locations as time increases, with the support of deep learning (DL) algorithms I developed using data from multiple monitor wells.
The model framework captures a wide range of relevant phenomena and provides a solid foundation for generating accurate and rich synthetic data representing multiple distinct scenarios leading to interpretation optimization. Also, the development of specific packages (commercial or otherwise) that explicitly model DSS/LF-DAS, incorporating the impact of fracture opening and slippage, is still in its infancy. This project is novel in this regard, and it opens new avenues of research and applications of synthetic DSS/LF-DAS in hydraulic fracturing processes. | |
