Methods for constraining discrete fracture network (DFN) models have historically relied on two very different scales for data: large-scale sources from which attributes can be coarsely defined for volume elements of 1000s of cubic meters such as seismic data, or small-scale sources where attributes of individual fractures are measured on a meter scale such as wellbores. Populating reservoir models with wellbore data requires upscaling the measured parameters, and the use of the large scale data types is accompanied by assumptions that can have significant uncertainties. A source of data that fills the gap intermediate to the large and small scale fracture parameters is microseismic data. During reservoir stimulation or production, acquisition of microseismic data with a surface array of geophones laid out in multiple azimuths and offsets (e.g., a star-like pattern above the well, or shallowly buried geophones in a grid like pattern), provides a broad sampling of the focal sphere that can be used to invert microseismic events for the source mechanism. This paper presents examples of DFN models constrained with source mechanisms and the implications for reservoir modeling of these more-highly constrained fracture network models.
Induced seismicity can be caused by various reservoir activities such as hydraulic fracturing, water injection or fluid extraction. Tight gas and oil shales have increase in importance as reservoir rocks, and hydraulic fracturing stimulation is required for their economic production. Monitoring the induced seismicity from the stimulations has been increasingly used to optimize hydraulic fracturing design and optimize oil and gas field development and production. This optimization is usually derived from the geometrical distribution of the located microseismic events. However, seismic waveforms recorded by various monitoring systems carry additional information on the mechanism of failure for each of these events. The event source mechanisms can be used to directly quantify and qualify stress changes instead of inferring these changes from the spatial distribution of the located microseismic events. Locations are derived mainly from observed arrival times while source mechanisms are inverted from relative amplitudes of either P or S waves (or both). Arrival times are less sensitive to small perturbations resulting from medium heterogeneity in a reservoir so locations are possible to achieve with a small aperture monitoring array, such as those placed down hole in a well near the well being stimulated.. Given that most of the early monitoring studies were carried out only with this type of very limited aperture array, source mechanism inversions from the uncertainty due to limited observation points produced unreliable results. The source mechanism inversion became more stable with larger numbers of monitoring receivers, such as can be deployed with multiple monitoring boreholes, or as a network of receivers distributed on the surface, or in the shallow subsurface. With larger numbers of monitoring receivers over a larger area there are geophones at multiple offsets and azimuths. The broader aerial coverage of such a receiver network can compensate for lower signal-to-noise ratio and provide more precise and accurate locations and source mechanisms.
