On the possible use of probabilistic techniques for purposes of short-term load dispatching

The security assessment of the working point set by the energy markets is one of the main tasks of any System Operator (SO). The experience of vertically-integrated utilities in the use of probabilistic methods has proved to be very useful also in the new deregulated environment: an increasing number of SOs is presently enforcing with probabilistic tools the deterministic criteria traditionally employed to validate the ex-ante dispatching. Very challenging, promising and discussed, but not yet well-established, is conversely the use of probabilistic tools also for real-time decision-support and very short-term dispatching. The computational complexity of many time-consuming algorithms has constituted for years a crucial barrier; nowadays, the availability of cheap and fast computers discloses new opportunities for probabilistic techniques also for purposes of on-line security assessment. In this paper a probabilistic technique for real-time security assessment and operational decisionmaking is proposed and described. The use of a novel power system simulator, based on a Sequential Monte Carlo technique, is applied to very short-term dispatching, in order to alert the SO if the system reliability is about to decrease in the next few hours, thus anticipating critical contingencies and supporting the control room to find the most cost-effective corrective action. A case study relevant to the IEEE Reliability Test System RTS-96 is shown and discussed.