Reservoir closed-loop optimization method based on connection elements and data space inversion with variable controls

Abstract

Reservoir simulation faces challenges in computational efficiency and uncertainty management for large-scale assets. This study presents an integrated framework combining the connection element method (CEM) and data space inversion with variable controls (DSIVC) for efficient history matching and optimized forecasting of reservoir performance. CEM reduces the computational cost of numerical simulation while retaining accuracy. DSIVC enables direct production forecasting after history matching without repeated model inversion. The CEM–DSIVC approach is applied to two reservoir cases. CEM efficiently constructs reservoir models honoring complex geology. DSIVC mathematically integrates production data to reduce uncertainty and parameter space. Without repeated forward simulation, optimized forecasts are obtained under different control strategies. Compared to conventional methods, CEM–DSIVC achieves reliable uncertainty quantification and optimized forecasting with significantly improved efficiency. This provides an effective solution to overcome limitations in simulating and managing uncertainty for large-scale reservoirs. The proposed approach leverages the complementary strengths of CEM and DSIVC, synergistically improving reservoir modeling, management, and decision-making. This integrated data-driven framework demonstrates strong potential as an advanced tool for efficient field development planning and optimization.