History Matching and Production Optimization Under Geologic Uncertainty
Dr. Ngoc Nguyen
Advisors: Professor Zhangxing Chen, Dr. Long X. Nghiem, Dr. Chaodong Yang
Login to View Article


Abstract
Brugge field, which is a complete synthetic oil field, is studied in this paper. Three case studies are examined: (1) run history matching using 104 realizations from tno; (2) run history matching using the pre-simulation command: the pre-simulation run will create a new realization, and cmost run optimization with this new realization; (3) test different methods of history matching: hybrid particle swam and simulated annealing, Latin hypercube plus proxy optimization, and DECE.
The goal for Case 1 is to maximize NPV via the controlling rate constraints and water cut value. The objective for Case 2 is to maximize NPV as in Case1, while water is injected only into layers 2 and 3. Hence, oil is produced only from layers 1 and 2, with linking GOCAD. Simulation results show that GOCAD does generate new facies and reservoir properties. The results of CMOST runs with linking GOCAD as a whole are better than the results of CMOST run with 104 realizations. Geologic uncertainty is accounted in production optimization by using automatic generated realizations with the pre-simulation command.