This paper builds on Klenner et al. 2018, which utilized machine learning to understand well-to-well communication ("Frac hits") or fracture-driven interaction (FDI) during hydraulic fracturing operations. This paper introduces an infrastructure that enahances the process for real-time enablement including data acquisition, visualization, and alerting for completion adjustments during operations. A use case is also presented which monitored, characterized, and provided understanding to the root causes of FDI during operations in the field. Machine learning in real-time provided insights and informed decisions regarding treating pressure, slurry rates, or proppant concentrations.

The analysis included real-time streaming of time series completion data of the hydraulically completed infill wells and the time series pressure data of the offset wells. The processing of the data in real-time uses s a 3-step process: 1) Apply analytics to detect and characterize interference events during the active stimulation and offset monitoring 2) Utilize alert mechanisms to notify operators when a significant communication event is occurring and 3) Apply machine learning techniques to determine what completion parameters inform the occurrence of FDI events, and quantify those parameters in order to provide operators with suggestive actions to help control the event in real-time.

The novelty of this approach is the application of analytics and machine learning to multiple streaming data sets in order to alert the operator of events, identify causality and consequently optimize completion design in real-time. The quantification of the completion parameters provides rules or recommendations that enable operators to make informed decisions to mitigate or manage communication. Furthermore, the methodology lends a fundamental analytics framework for visualization, continuous learning to create and optimize other real-time completion and stimulation operations at the asset level.

You can access this article if you purchase or spend a download.