A Unified Reduced-Order Framework for Adaptive Reservoir Management in Hard-to-Recover Fields

Operational management of hard-to-recover (HTR) reservoirs requires modelling tools that can be repeatedly updated and recalculated as new field data become available. This study presents a reduced-order framework that combines a source-based boundary element formulation for pressure redistribution with streamline-based oil–water transport and sequential history matching. The framework is evaluated using a synthetic benchmark and five field cases with different geological and operational characteristics. Three field cases are additionally compared with existing three-dimensional hydrodynamic models. Under matched simplified physical conditions, the synthetic benchmark reproduced pressure and saturation behaviour with a saturation field MSE below 0.2% and production rate MPE below 3%. In the field cases, retrospective liquid rate forecast MAPE ranged from 3.03% to 8.28%, while oil production accuracy showed greater sensitivity to unresolved displacement complexity. A large-scale test with 2483 wells demonstrated the computational feasibility of the combined workflow. The calibrated model was further applied to injection redistribution and new-well forecasting under uncertainty, with the optimized injection scenario yielding approximately 10% higher predicted cumulative oil production over the first forecast year. Optional components for historical data reconstruction and physics-informed pressure refinement were also considered but were not used in the main field case calculations. The results demonstrate the potential of the framework for repeated operational forecasting and scenario evaluation within the applicability limits of its reduced-order assumptions.