Stimulated shale wells show sharp decline rates after several months of production imposing the need for rejuvenating production via re-fracturing or other methods. The sharp decline in production from shale gas/oil wells is attributed to full or partial closure and damage of the induced fracture network within the rock. The poor economic conditions stirred by low oil prices has persuaded operators to consider re-fracturing as an affordable option but the mechanisms behind this technique is not fully understood. In this paper, we utilize a fully coupled model to understand stress re-distribution and identify different modes of fracture closure during production from initial hydraulic fractures. Rock creep, proppant failure and differential fluid depletion are identified as primary causes of closure within shale fracture networks. We incorporate these phenomena into a poroelaso-plastic model that simulates fracture conductivity evolution during production, and re-fracturing treatments illustrated by re-fracture propagation, fracture coalesce or fracture extension along pre-existing fractures. This methodology provides a realistic initial condition to simulate varying re-fracture designs - different treatment schedules and fracturing fluids. The Influence of degradable mechanical and non-mechanical diverters is also incorporated in this model.
Our results reveal major stress redistribution in the fracture network especially at the intersections due to depletion. The results show how fracture closure at the fracture intersections causes abrupt or time dependent shrinkage of the drainage area. Under certain conditions, re-fracturing may effectively open up these bottlenecks while in extreme situations, it creates new fractures which reorient obliquely. Results also reveal the role of in-situ stress anisotropy, magnitude of depletion and complexity of fracture networks in the successful re-fracturing of lower- clay content naturally fractured formations such as the Barnett shale. The results do not suggest strong dependency on stress anisotropy and natural fractures orientation in high clay content reservoir rocks like Haynesville or Marcellus. It shows the quantifiable effect of creep on closure rate and conductivity loss. Results show diverters have a profound effect on the expansion of the re-fracture network, although it hinders reactivation of some clogged fractures. Using adaptive cohesive elements provides the opportunity to model longitudinal, transverse or oblique re-fracture propagation.
The Proposed model introduces a realistic fracture closure mechanism and stress re- distribution during drainage, prior to re-fracturing. This plays a vital role in explaining initial fracture extension, re-frac propagation, fracture coalesce, re-fracture re-orientation, and therefore the effectiveness of re-fracturing treatments. This model offers a practical tool for identifying potentially successful re-fracturing candidates.