Optimizing the flexibility of a portfolio of generating plants to deal with wind generation

The uncertainties resulting from the integration of a large amount of wind and other stochastic forms of renewable generation affect the reliability of the power system. While it is generally agreed that coping with this uncertainty will require a more flexible system, not enough work has been done to provide dependable estimates of the amount of “flexibility” needed. This paper discusses a technique to determine the optimal amount of flexibility that a generation portfolio should provide for different levels of wind penetration. This optimization must bridge the gap between long-term investment decisions on the type of plants to be built and short-term operational decisions on how these plants are scheduled to meet the load and provide sufficient reserve. To achieve this goal, the unit commitment (UC) problem has been extended to consider not only whether a particular generating unit should be committed at a given time but also whether building this unit would reduce the sum of the operational and investment costs. A proper assessment of the balance between these two costs must take into account the seasonal variations in the load profile and must therefore consider an optimization horizon representative of a year. This paper describes the techniques that have been developed to make such calculations both accurate and achievable with a reasonable amount of computing resources. Test results based on the IEEE RTS system are presented and demonstrate how different amounts of wind generation capacity and reserve requirements affect the need for flexibility.