Prescribing net demand for two-stage electricity generation scheduling

We consider a two-stage generation scheduling problem comprising forward dispatch and real-time re-dispatch. The former must be conducted facing an uncertain net demand that includes non-dispatchable electricity consumption renewable power generation. latter copes with the plausible deviations respect to schedule by making use of balancing during actual operation system. Standard industry practice deals in stage replacing it good estimate its conditional expectation (usually referred as point forecast), so minimize need for real time. However, is well known cost structure system highly asymmetric dependent on operating point, result minimizing amount imbalances not necessarily aligned costs. In this paper, we propose bilevel program construct, from available historical data, prescription does account system’s asymmetry. Furthermore, accommodate strong dependence clustering tailor proposed foreseen net-demand regime. By way illustrative example more realistic case study based European system, show our approach leads substantial savings compared customary doing.