Gradient-based production optimization with simulation-based economic constraints

摘要: In reservoir management, production optimization is performed using gradient-based algorithms that commonly rely on an adjoint formulation to efficiently compute control gradients. Often, however, economic constraints are implicitly embedded within the procedure through well performance limits enforced at each simulation time-step. These effectively restrict operational capabilities of wells, e.g., they stop or shut down depending a predetermined profitability threshold for well. Various studies indicate accuracy gradient and, by consequence, algorithm suffer from this type heuristic constraint enforcement. paper, analytical framework developed study effects enforcing simulator-based when performing relies derivatives obtained formulation. The attributes loss in sensitivity non-differentiable unscheduled changes model equations. discontinuous nature these leads inconsistencies inconsistencies, turn, reduce quality and subsequently decrease algorithmic performance. Based framework, we devise efficient mode enforcement yields gradients with fewer consistency errors. implementation, equations violate removed governing system right after violation occurs not reinserted until next status update. modes further coupled strategy improves selection initial controls subsequent iterations procedure. After given simulation, resulting combination open shut-in periods generates update schedule, history. history current optimal solution saved used make part solution. novel simulation-based without history, applied two cases where, large set guesses, different realizations, it retains search compared common during optimization.