Coalbed methane (CBM) reservoirs are typically characterized by ultra-low permeability, pronounced heterogeneity, and strong stress sensitivity. A primary challenge in horizontal-well hydraulic fracturing is to accurately characterize complex fracture morphology and efficiently simulate fracture–matrix flow, which is essential for coordinated optimization of production performance and economics. This study proposes the Fracture Connection Element Method (FCEM), which departs from conventional grid-based discretization by representing both matrix and fractures as nodes. Matrix nodes are generated via Poisson-disk sampling to achieve spatially uniform coverage. The hydraulic-fracture network is converted into a node–edge topology using Dijkstra’s shortest-path algorithm, preserving both primary fractures and branching structures. On this connection-element system, an inter-nodal conductance (transmissibility) model is developed to characterize mass transfer among fractures, matrix, and their interactions. A semi-analytical formulation is further introduced for rapid production forecasting, enabling a favorable balance between numerical stability and computational efficiency. In a conceptual case under constant water-rate production, FCEM achieves an RMSE of 7.34% relative to the commercial simulator ECLIPSE while reducing runtime by approximately 80×. A dual-horizontal-well CBM field case is then validated against dynamic production data. History matching is performed using simultaneous perturbation stochastic approximation (SPSA), yielding match ratios of 84.87% and 80.31% for the two wells, respectively. Overall, FCEM reproduces pressure responses and fracture-network connectivity with high fidelity and offers improved efficiency, stability, and scalability compared with traditional grid-based approaches, providing a robust framework for fracturing-parameter design and production-schedule optimization in unconventional reservoirs.
Wang et al. (Thu,) studied this question.