Fiber optic distributed acoustic sensors (DAS) provide a new perspective on the subsurface environment and allow observations of previously unrecorded signals, such as the strain field associated with fracture opening during stimulation. Utilizing GEOS, a flexible multi-scale, multi-physics simulation environment, we are developing the capability to extract maximum information from the signals to constrain key subsurface parameters such as fracture length, width, and aperture. By varying the rate at which fluid leaks off into the reservoir during the stimulation of a hydraulic fracture, we test the ability of these sensors to monitor the placement of proppant in a fracture. We found that DAS signals are a strong indicator of proppant placement issues, such as tip screen-out.
The development of fiber optic distributed acoustic sensor (DAS) technology provides an unparalleled avenue for monitoring processes that occur within the subsurface. Because the strain measurements provided by these sensors differ from previous subsurface instrumentation, interpretation is not always straightforward, and we are exploring the use of computational geomechanics to understand the full potential of the sensors in understanding hydraulic fracture processes. An improved understanding of the subsurface fracture process is fundamental to future improvements and increased efficiencies in extraction of unconventional resources. In this paper, we use computational models to assess the potential utility of DAS in mapping the final distribution of proppant in a fracture. The location of proppant is notoriously difficult to estimate using traditional geophysical sensors but may be possible with DAS.
In previous work (Sherman et al., 2017, 2018), we demonstrated an approach for modeling DAS measurements with the 3D thermo-hydro-mechanical (THM) code GEOS. GEOS is a flexible and well-validated code designed to model subsurface fractures. It has been used to solve problems related to unconventional energy production including hydraulic fracture propagation (Settgast et al., 2016) and induced micro-seismic activity (Sherman et al., 2016). Simulated DAS measurements are generated by embedding virtual fiber-optic sensors that are placed throughout the numerical model and tied to an underlying finite element mesh. Both the low-frequency strain as created by an opening (or closing) fracture and the high-frequency dynamic signals produced by microseismic activity can be modeled. We assume that the fiber is perfectly coupled to the surrounding media and that it measures linear strain (or strain rate) along the fiber at a set of discrete points representing the channel spacing. Gauge length can also be implemented as a convolutional operator equivalent to the gauge length.
