Breaking Down Complexity: Using Multiple Point Statistics to Improve Fracture Network Modeling in Carbonate Reservoirs

Abstract This study aims to advance carbonate reservoir modeling by employing Multiple Point Statistics (MPS), a geostatistical method designed to capture complex spatial correlations for simulating realistic fracture networks. Traditional Discrete Fracture Network (DFN) models often oversimplify fracture systems, leading to inaccuracies in predicting reservoir performance. By integrating high-fidelity digital outcrop data, seismic attributes, and the Direct Sampling (DS) method (Mariethoz et al., 2010), this research develops a robust MPS framework. Validation using the Teapot Dome dataset demonstrates that MPS reduces prediction errors in permeability, pressure profiles, and production rates by approximately 50% compared to DFN models. Reservoir simulations were conducted with the state-of-the-art reservoir simulator to further cement MPS's superiority, showing sustained pressure maintenance due to effective communication within hydrocarbon, hydrocarbon-water and within water-leg leading to 15–20% higher cumulative oil production due to enhanced fracture connectivity. These findings suggest that MPS offers a powerful tool offering multiple development scenarios for optimizing production strategies and improving reservoir management in carbonate reservoirs, potentially transforming field development planning.