ARIMA modeling of wind speed for wind farm reliability analysis

The output power of a wind energy conversion system (WECS) fluctuates with wind speed variations. Accurate wind speed modeling is essential to forecast wind power changes in a site. This paper is intended to develop an autoregressive integrated moving average (ARIMA) time series model for wind speed. In ARIMA time series modeling, it is possible to change a nonstationary time series to a stationary time series by taking differences. A wind speed collection of one year from a 99MW wind farm situated in Theni, Tamil Nadu, India is used in this modeling. The developed model is used to evaluate annual reliability indices like loss of load probability (LOLP), loss of load expectation (LOLE), and loss of energy expectation (LOEE) by convolving wind farm generation model with load model of the grid. Furthermore, variations of LOLE and reliability with changes in peak load are carried out. The study illustrates that system risk index LOLE improves with decrease in peak load and WECS has high reliability to meet the changes in peak load.