Hydraulic Fracture Geometry Characterization Using Low-Frequency Distributed Acoustic Sensing Data: Forward Modeling, Inverse Modeling, and Field Applications

Abstract

Low-Frequency Distributed Acoustic Sensing (LF-DAS) is a promising fracture diagnostic technique for detecting fracture hits and characterizing fracture geometry. However, measured signals exhibiting various characteristics and mechanisms are not well understood, which makes the interpretation and application of LF-DAS data for hydraulic fracture monitoring and characterization much challenging. In this dissertation, a forward geomechanical model was developed based on the three-dimensional displacement discontinuity method (3D DDM) to simulate the LF-DAS strains and strain rates along horizontal wells during multistage/multicluster hydraulic fracturing treatments. The main applications of the forward model include investigating the observed strain/strain-rate responses and their corresponding fracture geometries for better understandings of LF-DAS signals and proposing guidelines for fracture-hit detection during multifracture propagation. More importantly, a Green-function-based inversion algorithm was proposed to estimate fracture geometry by direct inversion of LF-DAS strain data. The stability, accuracy, and efficiency of the proposed algorithm were tested through synthetic cases of both single fracture and multiple fractures. A few field case studies were performed to demonstrate the capability of the proposed workflow. Lastly, a two-dimensional thermoelastic model was presented to quantify the thermal effects on LF-DAS measurements. A heart-shaped extending region forms before the fracture hit on the waterfall plot of LF-DAS data. After the fracture hits the monitoring well, the extending region shrinks to a wide band, the size of which depends on the spatial resolution of field DAS measurements, and a two-wing compressing zone is observed. The size and shape of the aforementioned signatures are directly influenced by fracture geometries and fracture interactions. General guidelines for accurate fracture-hit detection were proposed based on detailed characterization of LF-DAS measurements around fracture-hit locations. The inverse modeling indicates that LF-DAS data are only sensitive to the fracture segments near the monitoring well. The developed inversion algorithm can provide the dynamic fracture widths and heights near the monitoring wells during hydraulic fracturing treatments. For the field cases from an unconventional shale oil formation, 4-5 fracture hits out of 8 perforation clusters per stage were detected. Fracturing fluid leakage into the previous fracture stage was observed in all studied stages. Fracture geometries near the monitoring well were characterized. This dissertation provides a novel workflow for quantitative hydraulic fracture geometry characterization and detection of fracture hits, which has been successfully applied to field cases. The developed workflow can play an important role in optimizing completion and fracturing designs and maximizing well performance in unconventional reservoirs.