A statistical characterization of wind generation and development of a stochastic LP model for the solution of the unit commitment problem in power systems with high RES penetration

In the last decade national Power Systems faced a large increase in penetration of Renewable Energy Sources (RES) generation. In particular, the largest increase was met by Photovoltaic generation (PV) and Wind Generation (WG), both affected by substantial uncontrollability and high levels of uncertainty and volatility. As a consequence, high levels of uncertainty are transferred to Power System, thus new tools are needed, capable of analyzing RES uncertainty and use the resulting characterization in helping Power System operation. In the present PhD work of thesis both of these topics are treated. Firstly, statistical and clustering techniques are considered in order to characterize actual WG data, coming from Spanish Power System, in particular with respect to the criticalities of the Demand of power. In a second part, Stochastic Programming techniques are introduced in an existing Medium Term Power System Simulator. The resulting model, named sMTSIM, is able solve the large scale Unit Commitment (UC) Problem for Zonal Power Systems with high level of RES penetration. The high levels of uncertainty and volatility characterizing RES generation are taken into account by means of their statistical characterization; then the UC problem is solved by coupling Stochastic Linear Programming techniques and an heuristic technique able to reintroduce the intrinsic Mixed Integer nature of UC problem.