The effects of geographical distribution on the reliability of wind energy

Abstract

We examine the effects of geographic distribution of wind power plants (WPPs) on the reliability of electrical output within the Midwestern United States. North American Regional Reanalysis (NARR) data are extrapolated to 80 m using the power law and used to characterize the wind resource at 108 NARR grid points corresponding to existing WPPs. These sites are then organized, on the basis of nearest neighbors, into networks ranging from single WPPs to the full network of 108 WPPs. For each network, a suite of statistics is computed and used to characterize energy reliability as it relates to the number of WPPs within, and the area enclosed by, the network. The results demonstrate that WPP dispersion reduces variability and thereby improves the reliability of electrical output from WPPs. As scale increases, marginal improvements in reliability diminish, but there is no saturation of benefits on the scales considered here. The results are combined with wind resource information to identify sites that can further improve reliability for aggregated wind power in the study region.

Introduction

Global wind energy resources far surpass current energy demand (Kempton, Pimenta, Veron, & Colle, 2010). Wind power is the fastest growing energy source in the world with an annual growth rate of approximately 35% (Sathyajith & Philip, 2011). However, the variability of wind, and the resulting intermittency of the wind power resource, is frequently cited as an obstacle to provision of baseload power by wind and its further penetration into the electricity market (DeCarolis & Keith, 2005; Sovacool, 2008). As an alternative to siting wind power plants (WPPs) only in regions with low wind variability, interconnection of WPPs through the transmission grid shows great promise for improving the reliability of electricity generated from wind (Archer & Jacobson 2007; Carlin & Haslett, 1982; Kempton et al., 2010; Kahn, 1979; Simonsen & Stevens, 2004). At a single site, or over the area occupied by a typical commercial WPP, wind speeds are highly variable. However, autocorrelation of wind speed decreases with distance (Robeson & Shein, 1997), so that as area increases, average wind speed is less variable. Over a sufficiently large area, meteorological and topographic conditions vary enough to produce a balance between areas with high and low wind speeds, and more importantly, a reduction in the frequency of calm conditions throughout the network.

Kahn (1979) was the first to suggest that geographically dispersed WPPs could improve the reliability of wind power. He analyzed networks of 2–13 WPPs and found that instances of zero power decreased as sites were added to the network. Archer and Jacobson (2003) analyzed surface measurements at 1327 weather stations and sounding measurements from 87 stations from the National Climatic Data Center and found that the standard deviation of wind speed was consistently greater at individual locations than when averaged over multiple locations. They also found that, in an eight-site, 385,000 area stretching across parts of New Mexico, Oklahoma, and Texas, average wind speed at 80 m never fell below 3 m s⁻¹, which is significant because 3 m s⁻¹ is a common cut-in speed for wind turbines (GE Energy, 2010, p6). Simonsen and Stevens (2004) analyzed one year of wind speed data at 28 sites across Iowa, North Dakota, Kansas, and Minnesota, and found that connecting the sites reduced the variability of power output by a factor of 1.75–3.4. Archer and Jacobson (2007) analyzed wind speed data at 19 sites spanning across parts of Kansas, New Mexico, Oklahoma, and Texas to determine if wind could be used as baseload power. They found that, on average, 33% of yearly averaged wind power could be used as baseload and that the standard deviation of wind power produced decreased by 35% from one site to 19 aggregated sites. Kempton et al. (2010) examined the power output of a hypothetical network of 11 offshore WPPs along the Eastern Seaboard. They found that compared to individual sites, hourly fluctuations of capacity factor of the entire network were dramatically reduced.

While the studies cited above have analyzed aggregated wind power over large geographic areas, the effects of aggregated wind power within an area corresponding to an Independent System Operator (ISO; the organization that manages the operation of the electrical power system within a region) have not been considered. Furthermore, existing studies have focused on either the number of aggregated WPPs or the area enclosed by a network of WPPs, but not both, resulting in confusion regarding the source of improvement in reliability. This study addresses these issues by examining the effects of aggregating the energy production of existing WPPs within the area corresponding roughly to the United States component of the Midwest ISO and evaluating the role of the number of WPPs relative to the geographic area covered by the WPPs. We also use our findings in conjunction with wind resource data to identify new areas for wind power development aimed at improving reliability.