Optimal storage sizing for integrating wind and load forecast uncertainties

One of the main characteristics of wind power is the inherent variability and uncontrollability. Even with state-of-the art forecasting techniques, actual wind generation can be substantially deviated from the forecasted values. In addition, the system load is also variable and needs to be forecasted ahead of time for unit commitment and system planning and operation purposes. Although daily loads follow a pattern, every forecast is associated with a certain degree of uncertainty. In a system consisting of only wind generation and load, with minimum or no connection to the grid, the task of energy balancing is extremely difficult. Incorporation of energy storage units is being considered as a possible solution to this problem. However, energy storage units till date are expensive. Hence the question that arises is, given a generation-load system, what is the optimal amount of storage required. This paper investigates the storage size required by a system consisting of a wind farm and a load in meeting certain specified reliability indices. The optimization model is formulated as a stochastic linear programming problem considering two random quantities, load and wind generation. The minimum initial energy required in the storage unit is also computed. The reliability index of the system is evaluated in terms of the LOLP (Loss of Load Probability). Another reliability index, SGP (Spilled Generation Probability), is defined to determine the probability of a wind energy spillage event due to excess wind generation that cannot be accommodated in the system. A realistic example case is presented where the developed methodology is tested with satisfactory results.