Abstract
Naturally fractured shale reservoirs are characterized by a complex 3D discrete fracture network (DFN). DFN density in different stratigraphic layers imposes variability in hydraulic fracture geometry as the hydraulic fracture propagates in the laminated shale reservoirs. The study presented in this paper aims to parametrically determine the hydraulic fracture geometry and consequent well production when varying densities of DFN interact with the propagation and growth of hydraulic fractures.
A 3D multilayer DFN (MDFN) is discretized in vertical grid elements distributed in layers of the formation. This is represented in a geocellular grid model with varying DFN sets in multiple layers. A new hydraulic fracture propagation model with stacked height has been utilized to determine the interaction of the hydraulic fractures with the DFN in all different layers. The vertical variation of the DFN causes more complex fracture propagation because it accounts for the fracture crossing criterion apart from the stress profile variation in layers and the stress shadow impact. Comparison of such complex hydraulic fracture geometry with the simplistic average 2D representation of the DFN reveals the geometry difference. This is further coupled to reservoir simulation in numerical engines to determine the difference in production performance. Finite element geomechanical simulations are made to predict microseismic events, which are then compared to the acquired microseismic data to validate the hydraulic fracture geometry.
Hydraulic fracture geometry was found to differ when an MDFN approach was taken as compared to simplistic single-layer DFN (SDFN) interaction. Production performance of the fracture geometry obtained by considering an MDFN was therefore more realistic. The simulated microseismic data also exhibited a much better match to the acquired microseismic data when an MDFN was used for modeling. Visual analytics tools for comparison of observed microseismic event distribution with predicted microseismic data in the model show errors ranging from 16% to 24% for different view representations. As a conclusion of this study, it was evident that wherever applicable, an MDFN approach should be used to determine the well completion performance. However, there is some scope for improving the modeling of hydraulic fracture interaction with natural fractures by incorporating the dip angle of the MDFN for computing the crossing criteria.
This study represents the first time a more-robust modeling of the interaction of hydraulic fractures with a 3D MDFN was achieved using complex fracture models to determine the impact on productivity. Finite-element-computed microseismic data allowed further validation. Operators can use the approach described in this study to predict their well performance more accurately in complex fracture reservoirs.
