Summary
Low‐frequency distributed acoustic‐sensing (LF‐DAS) data are promising attributes for detecting fracture hits and fracture characterization. However, measured signals from different wells exhibiting various characteristics and mechanisms attributing to the difference are not well understood, which makes the interpretation of field LF‐DAS data most challenging. In this study, our in‐house hydraulic fracturing simulator is used to simulate fracture propagation. The induced rock deformation and corresponding strain‐rate variations along offset monitor wells are analyzed and related to specific fracture features. The mechanisms for LF‐DAS signals are investigated through five synthetic case studies with single fracture propagation. A typical strain‐rate waterfall plot of LF‐DAS measurements during the fluid injection phase of a fracturing treatment can be divided into two distinct regions. A heart‐shaped extending region forms as the fracture approaches to the monitor well, indicating that the magnitude of extension keeps increasing as the fracture tip gets closer to the monitor well. After the fracture hits the monitor well, the extending region shrinks to a line (the field‐measured data may be a wide band, depending on the spatial resolution of the measurement), and a two‐wing compressing zone is observed, illustrating large compressional strain variations on both sides of the fracture. As the fracture continues propagating, the strain rate tends to be stable, the characteristics of which depend on specific fracture geometry and propagation conditions. The size and shape of observable signatures on LF‐DAS data are directly influenced by fracture width, height, and height growth. Larger fracture width results in larger sizes of heart‐shaped extending region and two‐wing compressing region in the strain‐rate waterfall plot. Larger fracture height also induces a larger heart‐shaped extending region before the fracture hits. However, a fracture with larger height could lead to larger extension along the fiber near the fracture, which results in less overall compression and a zone of decreasing compression in the vicinity of the fracture as the fracture propagates away from the fiber after the fracture hit. This signature is more pronounced when the fracture height growth is considered.
The interpretation of a field example with four clusters based on our forward physical modeling results indicates that, although the distinct signatures of field data are not as obvious as the simulation results because of low measurement resolution and unavoidable noise, they do convey valuable information on fracture characteristics. There is a shrinkage of the extending zone from a heart shape to a band at the fracture‐hit time. During simultaneous multifracture propagation, fracture‐hit time of each fracture, which determines the fracture propagation speed and perforation efficiency, can be identified. The discontinuous extending band after fracture hit could be attributed to the intermittent stop and restart of fracture propagation and relative fracture opening/closing. The results of this study help to better interpret the real‐time LF‐DAS data and provide critical insights into hydraulic fracture characterization using LF‐DAS data.
