Mo, Jung Youn; Jeon, Wooyoung; Mount, Timothy D. (Charles H. Dyson School of Applied Economics and Management, Cornell University, 2013-10-01)
The purpose of this report is to describe the econometric models used to generate the input data needed for analyzing the effects of renewable energy on operations of an electric grid. The input variables are the hourly wind speed and the hourly electricity load at different locations in New York State and New England. The models estimated are time-series ARMAX models, and for both variables, the hourly temperature is an explanatory variable. The equations for wind speed, temperature and electricity load are estimated individually in two steps, and the computed white-noise residuals are used to estimate the variance/covariance matrix for all equations. For forecasting purposes, the complete set of estimated equations represents a multivariate time-series model. The equations for electricity load fit well and account for roughly 99% of the variability when the estimated residual structure is included, and 90% using just least squares. The equations for wind speed do not fit nearly as well, and the corresponding measures are 80% with the residual structure, and only 10% with least squares.
Make a deposit on eCommmons
Please sign in with your Cornell NetID to continue.