The geometry of fractures, both natural and hydraulically induced, is commonly represented by a discrete fracture network (DFN) in reservoir simulations. Although microseismic data underestimate the total fracture creation and deformation accompany hydraulic-fracturing, it is considered the best means of representing a DFN used as the initial input framework for modeling hydraulic-fracture stimulations. Extracting a DFN from microseismic data requires both the source location and the corresponding fracture orientations determined from moment tensor inversion (MTI). However, although monitoring from a single array may be the most common geometry used for downhole microseismic, inverting for the full moment tensor is not generally applicable due to the ill-posed geometry. We therefore use a new multi-event MTI approach applicable to single monitoring well geometry. This study presents a workflow to generate the DFN using a probabilistic approach applying the orientation constraints from the fault plane solutions estimated from the single-well microseismic data using a multi-event moment tensor inversion (ME-MTI) method. Firstly, the events are grouped with similar source mechanisms from the P/Sh and P/Sv amplitude ratio patterns and the ME-MTI algorithm is applied to each event group to get the common moment tensor. Secondly, a bi-axial potency decomposition is applied to the common moment tensors to estimate the fault plane solutions. Finally, the DFN is extracted using a modified Hough transform algorithm which searches for the most probable planar patterns from the event locations and the fault plane orientation distributions. We use a dual monitoring well data set, acquired in Cardium sand formation in west Alberta, Canada, to demonstrate the feasibility of the workflow. The two-well monitor data provides a bench mark result by extracting the DFN from the population of individual-event moment tensor solutions results. Using the workflow outlined above, we then compare the single-well DFN results with the benchmark. The results show that,

  1. the multi-event moment tensor inversion from a single monitoring well provides results comparable with the dual-well derived population of individual-event solutions, with reduced the uncertainty, and

  2. the DFN extracted from the single monitoring well sub-set data is comparable with the results from the dual-well data set.


With the development of microseismic monitoring to directly measure hydraulic fracture geometry, it has become evident that a complex fracture network is often activated or created instead of a single bi-wing planar fracture (Cipolla, et al. 2008, Maxwell, et al. 2002) due to the heterogeneity of the formation and the interaction of the existing natural fractures with the hydraulically induced fractures (Cooke and Underwood 2001). Accordingly, the prediction of reservoir performance requires models with more complex fracture geometries. As such, discrete fracture networks (DFN) is designed to represent the complex fracture geometries and flow paths created and affected by hydraulic fracturing and widely used as a framework for reservoir simulation (Cipolla, et al. 2010, Xu, et al. 2009, Williams, et al. 2013, Williams, et al. 2010). A DFN is a collection of connected or disconnected fracture patches with certain orientation and density distribution representing the natural fractures and/or hydraulically induced fractures.

This content is only available via PDF.
You can access this article if you purchase or spend a download.