Development of a Fast Parallel Hydraulic Fracture Simulator for Shale Reservoirs with Complex Natural Fractures

Abstract Hydraulic fracturing is a cornerstone technology in the development of unconventional oil and gas reservoirs. However, accurately modeling fracture propagation in geologically complex formations—such as those characterized by stress anisotropy, bedding interfaces, and natural fracture networks—remains computationally challenging. This study presents the development and application of a fast, parallel hydraulic fracture simulator designed to address these challenges by combining the Displacement Discontinuity Method (DDM) and Newton-Raphson iteration within an OpenMP-accelerated architecture.The simulator models both fracture propagation and fluid flow using dynamically growing matrix systems, which reflect evolving fracture complexity. A traditional serial LU decomposition solver is first implemented and benchmarked against a novel parallel solver using OpenMP-based multifrontal techniques, symbolic factorization, and optimized BLAS routines. The performance of the two solvers is evaluated using a field-realistic case from the Sichuan Basin, China—a region known for dense natural fracture networks and heterogeneous geomechanical conditions. Inputs include layer-specific properties such as Young's modulus, Poisson's ratio, and in-situ stresses, as well as a fully integrated 3D Discrete Fracture Network (DFN).Simulation results confirm that the parallel solver achieves a 10× speedup compared to the serial solver, without loss of accuracy in pressure propagation or fracture geometry. The simulator is further scaled to a 10-well cube development configuration spanning four vertical layers, demonstrating its capability to handle large, multi-well models in under an hour. This work offers a scalable, accurate, and computationally efficient solution for simulating complex fracture networks, enabling rapid, high-fidelity modeling for field development optimization. The platform lays the groundwork for future integration with AI-based optimization and real-time digital twin applications.