Risk management applied to weekly generation scheduling

As energy markets are being restructured all over the world, producers face more uncertainties concerning electricity prices. Optimizing schedules for weekly generation is therefore a stochastic problem. We assume that random occurrences can be represented by a finite set of price scenarios organised in a tree-like structure. In a classical approach, a producer would try to maximize his average profit over this uncertain future. However, such a strategy could lead to an unacceptable solution for a given scenario, especially a low prices scenario, inducing financial losses. In our approach, inspired by financial portfolio risk management methods, each producer reduces his risk by limiting his financial losses on unfavourable scenarios. From a practical point of view, we introduced new constraints in the optimization problem,which is usually solved by conventional operation planning software. This strategy leads to substantial savings on critical scenarios.