Day-ahead energy and reserve scheduling under correlated wind power production

Empirical reports validate significant correlation among adjacent wind farms. However, in practice, the market operator schedules the amounts of energy and reserve in the day-ahead market without considering the correlation. This paper develops a computational framework, which allows the market operator to clear the day-ahead market under the correlated wind power production, and evaluate how the energy and reserve schedules may change due to ignoring the correlation. To accomplish this task, a scenario generation methodology is proposed to provide correlated scenarios of wind production in the daily horizon. Through the methodology, a multivariate model presents the power production of each wind farm as a nonlinear function of wind speed and autoregressive moving average model. The multivariate models of wind farms are coupled through their crosscorrelation matrix. The methodology is applied to the real-world data of wind farms located at diverse geographic regions in United States. The scenarios are considered as input data for clearing a joint energy and reserve market, which is formulated as a two-stage stochastic model. The potential impact of significant correlation among wind farms is investigated through 6-bus and IEEE 118-bus test systems for different wind penetration levels. The results show that for higher correlation, the policy adopted by the market operator is to schedule less wind energy in the day-ahead market, at the expense of greater energy and reserve capacity allocated to conventional generating units. By ignoring the correlation, the market operator underestimates the reserve capacity that may threaten system security and impose additional cost on the real-time operation.