An Optimized Proxy Modeling Framework for Simultaneous Well Placement Across Multiple Development Units

This paper presents an accelerated and more granular version of the scalable proxy modeling framework for simultaneous well placement optimization by introducing asynchronous workflows and including multiple development units (DUs) scenarios. Traditional simulation-based well placement is computationally intensive for large reservoirs, motivating the use of AI-based proxy models for rapid optimization. Our methodology integrates DU-specific parameters with parallelized, vectorized workflows to evaluate multiple scenarios simultaneously. The framework was applied to an ADNOC carbonate reservoir which is divided into two major and fifteen minor units, achieving a 70× speed-up over conventional approaches while improving NPV by 4.6%. Enhanced trajectory control, orientation awareness, grid interpolation, and DU-aware zoning ensure operational feasibility and scalability. The novelty of this work lies in simultaneously optimizing across multiple reservoir compartments with a demonstrated multifold computational speed-up, a challenge rarely addressed in prior studies. The framework is scalable to both greenfield developments and brownfield projects, making it widely applicable across diverse carbonate settings. Results validate the framework’s effectiveness for complex, compartmentalized fields. Future work includes integrating multi objective optimization (e.g., production profile plateau constraints) to further broaden its applicability.