EFFECTS OF NATURAL FRACTURE REACTIVATION DURING HYDRAULIC FRACTURING OF THE BARNETT SHALE, FORT WORTH BASIN, TX

Hydraulic fracture networks develop by sequential reactivation of natural fractures and by new propagation, which are determined by the state of stress and the injection qualities. Pressurizing the natural fracture network also induces evolving stress perturbations that can affect fluid flow and change the stimulation pattern. Previous efforts in the Barnett Shale considered hydrofracturing independent of dynamic stress changes caused by injection. The aim here is to understand the interaction of the reactivating fracture network within the host rock. Two geomechanical models are applied.

First, an analytical model is used to determine the conditions for shear-slip and dilation of fractures. The model uses slip-tendency concepts and solves for failure limits under tectonic stress and internal fluid pressure. Input is from stress data determined in 13 hydrofrac stages in three horizontal wells and in 22 pump-in intervals in four vertical wells of the Barnett Shale. The results are compared to 3D finite element (FE) simulations that solve for stress distribution and volume change during injection. The FE model is 2m x 2m x 2m and is based on general Barnett Shale properties. Four elastic-plastic layers with varied thicknesses each contain two sub-vertical, layer bound fracture sets with mean orientation of 45° and 135°. Effective tectonic stresses are initially applied: overburden σ₁ = 60 MPa (~8,000 ft depth), horizontal stresses σ₂ > σ₃ = 30-55 MPa, and pore pressure P_p = 28.7 MPa (~0.52 psi/ft). In a second step pressure inside the fractures P_f is raised from 0-50 MPa.

In both models shear-slip occurs on a wide range of fracture orientations at low internal pressure P_f < σ_3. Low differential stress (σ₂ ≈ σ₃) and increasing normalized fracture pressure R= (P_f - σ₃)/(σ₁ - σ₃) promotes dilation outside of the σ₁-σ₂ plane. FE simulations show that R as low as 0.05 induces non-uniform stress perturbation and local stress rotation. Rock volume V is constant until R > 0 and then V expands nonlinearly with rapid increase at R > 0.1. Under conditions σ₂ >> σ₃, V expands in the direction of σ₃ and contracts in σ₁ and σ₂, such that dV₃ > 0 > dV₂ > dV₁. When σ₂ ≈ σ₃ growth is oblate spherical: dV₃ ≈ dV₂ > 0 > dV₁.

The results show good agreement with microseismic patterns in the Barnett Shale and have applicability to multi-stage completion design.