Abstract
This paper compares the simulation results for two stochastic optimization power market models. EMPS uses aggregation and heuristics to calculate the optimal dispatch. SOVN simulates the operation of the power system in one large linear programming problem taking each single plant and reservoir into consideration. The comparison is for a future system in Europe where wind and solar power production supplies 61% of the annual demand. Three different alternatives for the Norwegian hydropower system is studied: present generation capacity (about 30 GW), increased capacity to about 41 GW and further to about 49 GW. The analyses show that SOVN to a larger degree than EMPS manage to increase production in high price periods and pump in low price periods. This particularly applies to the weeks before the change from the depletion (winter) to the filling (summer) period. This better ability to exploit the flexibility of the hydropower system is due to applying a formal optimization in SOVN compared to advanced heuristics in EMPS. For regions without pumping possibility, there is less difference between the models.