Numerical Simulation and Validation of In Situ Stress Field in Fractured-Vuggy Carbonate Reservoirs: A Case Study of the Aiding Block, Tahe Oilfield
Abstract
Accurate characterization of the in situ stress field in ultra-deep fractured-vuggy carbonate reservoirs is essential for optimizing hydraulic fracturing design and well trajectory planning. This study presents a numerical investigation of the stress field in the Aiding block of the Tahe Oilfield, a typical ultra-deep (6500–7500 m) carbonate reservoir. A three-dimensional geological model incorporating faults and fractures was constructed based on seismic interpretation and well logging data. A hybrid discrete fracture network–finite element method (DFN-FEM) model was implemented, and the governing equations, DFN-FEM coupling strategy, boundary conditions, and solver settings are documented to ensure reproducibility. Karst caves are represented as equivalent weak materials rather than true voids because of their complex geometry and filling. A local mesh-convergence study and sensitivity analysis were performed for key model parameters, including fault/fracture modulus, Poisson’s ratio, normal and shear stiffness, stress gradients, fault-zone width, fracture density, and boundary conditions. The simulation results were validated against closure pressures derived from G-function analysis of fracturing pressure data from four wells. The results reveal that: (1) vertical stress (120–150 MPa) is the maximum principal stress and primarily depth-dependent, while σH and σh range from 100 to 130 MPa and 80 to 95 MPa, respectively, with fault zones exhibiting 10–20 MPa reduction and compressional fault zones showing localized stress concentration; (2) fault-induced stress disturbances follow characteristic patterns—stress minima at fault cores, gradual recovery away from faults, stress concentration at fault tips, and rotation of σH from NE30° to NE10–50° and σh from NW320° to NW300–350°; (3) intersecting faults exhibit the most complex distributions and lowest stress values at intersections; and (4) the fault influence zone of 10–20 m under the current grid resolution; after local refinement, it should be regarded as an apparent influence zone and used with caution for cluster spacing optimization. Stress reduction in fault zones suggests lower injection pressures for fracture initiation in faulted intervals and higher pressures near compressional faults.