Characterization and Learnings of Far Field Fracture Propagation in the Marcellus Shale Using Aggregated LF-DAS Datasets

Abstract

Abstract Economic optimal recovery of hydrocarbons from unconventional reservoirs requires consistent and sufficient stimulation energy across an entire drill spacing unit (DSU), both along laterals and between them. This necessitates an understanding of how and why fractures propagate away from the treatment well. While legacy methods such as Distributed Acoustic and Temperature Surveys (DAS/DTS), pressure responses, logging/imaging has allowed us to investigate the point of stimulation origination, analyzing Low Frequency Distributed Acoustic Sensing (LF-DAS) gives us unique insights into far field fracture behavior. This paper aggregates several existing LF-DAS datasets in the Marcellus as well as the most representative log data available to draw conclusions on the impacts of geology, completion test parameters, stage and well frac order, frac design and materials, lateral placement, and many other influence factors on far field fracture behavior. Both statistical analysis and data visualization of these datasets independently and in aggregate was used to identify numeric as well as engineer-based observations and conclusions. By using these two methods, correlations between influence factors and far field fracture behavior are identified and quantified, and a level of variability is demonstrated for the Marcellus as well as some specific sub-plays. The analysis presented identifies controllable parameters to adjust as well as uncontrollable parameters that must be designed around to achieve stimulation goals. Also included is a description of where additional data collection is required to improve certainty and robustness of the analysis. Aggregated analysis on cross well far field fracture behavior has not previously been performed for Marcellus. This paper displays multi-project trends that were previously unidentified allowing the opportunity to design future tests around and confirm or challenge existing theories and conclusions. Previously non discussed fracture behaviors in the dataset are also described and addressed.