India, as one of the world's largest economies, faces the challenge of meeting increasing energy demands while reducing greenhouse gas emissions. Utilizing unconventional energy sources like shale, gas hydrates, and geothermal reserves can help achieve this goal. By tapping into these energy sources, we can achieve self-sufficiency in energy for several years and reduce our carbon footprint. However, the efficient energy extraction from these reservoirs poses significant challenges due to their low intrinsic permeability. Creating engineered fracture networks in these low-permeability reservoirs can enhance production. Existing 2D computational models are limited, as fractures are 3D and non-planar, and interactions between fractures and physical processes need to be accurately captured. Thus, a robust model for subsurface fracture propagation is essential.

Embedded finite element methods such as the Generalized/eXtended Finite Element Method (G/X-FEM) use enrichments on a fixed background mesh to represent macroscopic displacement fields, eliminating the need for continuous re-meshing to accommodate crack geometry. However, handling multiple interacting cracks, especially in 3D or unconventional reservoirs with branching, merging, and arrest, remains a significant challenge in G/X-FEM.

The phase-field method has gained popularity for modelling brittle fractures, as it regularizes sharp crack topology with a diffuse damage band. It ensures the irreversibility of the fracture process using a scalar phase-field and an additional partial differential equation to solve the damage field evolution. Unlike previous approaches, the phase-field method doesn't require explicit tracking of cracks, as crack directions are implicitly incorporated in the variational formulation model. It can handle complex crack patterns like branching and merging. However, its main limitation is its high computational cost, requiring a significant number of elements to accurately resolve the smeared damage zone. This makes it currently impractical for field-scale reservoir applications, limiting its usage to 2D and small 3D problems. We aim to develop a computationally efficient phase-field model for simulating subsurface fracture propagation. The developed model will leverage parallel computing on distributed memory environments, and adaptive mesh-refinement to offset the high computational costs. As a first step, we have developed a serial code for phase-field modelling of fracture in 3D. The model has been validated against benchmark results in the literature and the results are discussed herein.