Abstract
This paper presents a comparison of two new advanced statistical short-term wind-power forecasting systems developed by two independent research teams. The input variables used in both systems were the same: forecasted meteorological variable values obtained from a numerical weather prediction model: and electric power-generation registers from the SCADA system of the wind farm. Both systems are described in detail and the forecasting results compared, revealing great similarities, although the proposed structures of the two systems are different. The forecast horizon for both systems is 72 h, allowing the use of the forecasted values in electric market operations, as diary and intra-diary power generation bid offers, and in wind-farm maintenance planning.

Abstract
Wind energy experiences in Portugal an increasing interest. Slightly more than 1700 MW were operating by the end of 2006, in a system with a global capacity of about 12 GW (8,5 GW peak demand). Several new wind farms are under construction and a considerable amount of connection points are or will be granted in the coming years. More than 5000 MW are expected to be connected to the grid around 2012, the global generating capacity being then about 16 GW. Clearly, a wind power forecasting system must be implemented that will help to deal with the significant penetration of the technology in the electrical system. A group of wind farm promoters, owning the majority of the capacity installed so far, ordered to a consortium of universities and research institutes the development of a forecasting tool, giving rise of the EPREV project, wholly financed by them. The system will have the following main characteristics: Wind speed and active power forecasting up to 72 hours; Evaluation of the forecasting uncertainty; Possibility of using the predictions of physical models and the information from the wind farm Supervisory Control And Data Acquisition (SCADA); Capacity of predicting only with SCADA information for very short term. The main components of the system are: A human-machine-interface, allowing the control of the system, the selection and aggregation of forecasting models and the visualization of results; A power forecasting model for individual wind turbines and for wind farms. A cascade of models is used, starting in the mesoscale simulation, with up to 2 km resolution. The outputs of the mesoscale models are corrected and statistically adapted to the fine scale conditions. Two models and different boundary conditions are run, in three nested domains (54x54, 18x18 and 6x6 km). The advantage of using a 2x2 km resolution is also tested. The statistical models are fed with recent information from the wind farms, after a learning process that made use of the historical information of its operation. Three different types of statistical models are employed: Power Curve Model (PCM), Auto Regressive (AR) and Neural Network Assembling Model (NNAM). The wind simulation at the wind farm scale is done both by linearized physical models and Computational Fluid Dynamics (CFD) models, namely using VENTOS®, a code developed at the University of Porto. The duration of the project is planned to be 1 year, including off-line tests of the complete system for 3 wind farms, for performance evaluation purposes.

Abstract
The coordination between wind farms and pumping storage units increases the wind farm's controllability and maximizes the profit. In literature, several optimization algorithms were proposed for deriving the optimal coordination between wind farms and storage units. However, no attention has been given to operational management strategies for following the strategy that results from the optimization phase. This paper presents three possible heuristic strategies for managing the wind-hydro system during the operational day according to a day-ahead optimized strategy. Moreover, a chance-constrained based optimization algorithm, that includes wind power uncertainty, is also described. The algorithms are tested in a real case-study.

Abstract
The participation of an EV aggregator in the electricity market for purchasing electrical energy requires an algorithm for managing the EV charging during the operational day. In this paper the coordination of EV for minimizing the deviation between bid and consumed electrical energy is studied and compared with an uncoordinated strategy. Two algorithms are proposed: a heuristic algorithm that dispatches the EV for each time interval separately, and another one, formulated as an optimization problem for dispatching the EV considering all the time intervals. Furthermore, the aggregator architecture is compared with an autonomous architecture where each EV operates and participates in the market individually. The results, for a realistic case-study, show that the aggregator with an optimized coordination strategy achieves the lowest deviation cost and magnitude.

Abstract
It is foreseeable that electricity retailers for electrical mobility will be market agents. These retailers are electric vehicle (EV) aggregation agents, which operate as a commercial middleman between electricity market and EV owners. Furthermore, with the foreseen evolution of the smart-grid concept, these agents will be able to control the EV charging rates and offer several ancillary services. This paper formulates an optimization problem for the EV aggregation agent participation in the day-ahead and secondary reserve market sessions. Forecasting issues are also discussed. The methodology was tested for two years (2009 and 2010) of the Iberian market, considering perfect and naïve forecast for all variables of the problem. © 2011 IEEE.

Abstract
The increasing levels of wind power penetration motivated a revisitation of methods for setting operating reserve requirements for the next and current day. System Operators (SO) are now moving from deterministic intro probabilistic approaches, and including wind power forecasts in their decision-making problems. In this manuscript, a probabilistic approach that evaluates the consequences of setting each possible reserve level through a set of risk indices is compared with frequently used deterministic rules and a probabilistic rule where wind power uncertainty is described by a Gaussian distribution. The comparison is performed over a period of five months for a realistic power system, using real load and wind power generation data. Results highlight the limitations of deterministic rules, challenge the Gaussian assumption and illustrate the usefulness of risk indices derived from the probabilistic forecast and using a full probabilistic methodology.