This paper considers a wind farm with an obli- gation to purchase at a fixed feed-in tariff. The wind farm production must always stay below an upper limit existing for the obligation to purchase energy. An energy storage sys- tem associated to the studied wind farm allows to increase the installed capacity while maintaining the same upper limit for the wind farm production power. This operation may en- hance the wind farm availability and present an added value to the project by increasing its energy sales. The net present value of the project is calculated as a function of the energy storage system characteristics and the wind farm installed capacity.

This paper presents a fault detector for power distribution systems based on the use of feedforward neural networks. The described method is successfully tested through several simulations. The efficiency of the algorithm to recognize faulty feeders without measuring any voltage in the network and without any threshold is emphasized. Moreover, the sampling frequency of signals and the errors that measuring instruments may introduce do not interfere with the right functionning of the detector.

This paper presents a method to determine the optimal capacities in power and energy and control strategy of a battery storage system installed in a 30 MVA, HV/MV substation. It consists of a practical method for load management in power distribution system. This approach seeks the minimization of annual invoice paid by the MV network operator to the trans- mission grid operator. Genetic algorithms method is applied to solve this optimization problem in order to present an economic assessment of energy storage systems associated to distribution substations.

This paper presents the implementation of a hys- teretic reactor model in a transient analysis package. The model is available in a currently developed EMTP (DCG- EPRI) version. It is demonstrated how the hysteretic reactor equations can be solved to eliminate topological limitations and allow solving large scale networks efficiently. Numerical robustness aspects and the representation of minor loops are emphasized. The model demonstrates new and not previously available computational capabilities. II. PREVIOUS MODELS The original EMTP Type-96 model is based on the the- ory developed in (4). The hysteresis loops are modeled with piecewise linear approximations. Downward and up- ward trajectories are related by a translation on the x-axis and the minor loops are scaled in shape from the major loops. Some modifications have been proposed (5) to match experimental results. The model remained however unable to reproduce the experimental results presented in (3).