Abstract— The optimal management of shared transboundary oil fields in the Zagros Basin has been persistently challenging due to the inherent geological complexity of carbonate reservoirs and technical-political hurdles. This paper proposes an Integrated Geology-Engineering AI-Driven Digital Twin Framework for the scientific management of the shared Naft-e-Shahr (Iran) / Naft Khana (Iraq) oil field. This border field features an Asmari Formation carbonate reservoir with a dual-porosity (fracture-vug) system and strong heterogeneity. In this research, by integrating geological (seismic interpretation, facies analysis, diagenesis), petrophysical (core and well log), and reservoir engineering data, a dynamic geology-engineering model (GEM) was constructed. Subsequently, using machine learning algorithms (Graph Neural Networks for fracture modeling and LSTM for reservoir behavior prediction) and reinforcement learning for production optimization, the field\\\'s digital twin was developed. Simulation results indicate this framework could increase ultimate oil recovery by up to 15%, reduce exploratory drilling costs by 25%, and delay water breakthrough by 35%. This study emphasizes the pivotal role of integrating accurate geological models with artificial intelligence to achieve fair and efficient management of shared resources.

Keywords