Rotor-Area Wind Characteristics at the Eolos Wind Research Station in Southeastern Minnesota, USA

The Eolos Wind Research Station in Rosemount, Minnesota, was established in 2011 to support basic research and development of new technologies for wind energy generation. The site includes a 2.5 MW utility-scale wind turbine (80 m hub height) and a 130m meteorological (met) tower. The met tower is instrumented with sonic anemometers at approximately 10, 30, 80, and 128 m AGL mounted on 5.5 m booms extending southwest from the tower (the direction least likely to be affected by the tower; see below). We use six years (2012-2017) of hourly data from the met tower to investigate wind speed and turbulence characteristics across the rotor swept area at diurnal and seasonal time scales, as well as their relationship to broader synoptic patterns.

The Eolos research station (45^oN, 93.5^oW) is located in a predominantly agricultural area of low, rolling topography. Mean wind speeds at 10 and 30 m are fastest in the spring and have a secondary peak in autumn; winds at 80 m have nearly identical peaks in spring and autumn; and winds at 128 m are fastest in autumn with a secondary peak in the spring. Winds at all heights are weakest in the summer. Wind directions are predominantly from the northwest in the winter, are roughly bimodal (south/southeasterly and northwesterly) in spring and autumn, and most often from the south/southeast in the summer with a secondary cluster of northwesterly winds.

The distribution of mean hourly wind speeds is most peaked for low heights and shifts toward higher speeds and wider distributions as height increases. As expected, near-surface wind speeds are strongest during the day and weakest at night, while speeds at higher heights show the opposite pattern. For all heights, the largest diurnal cycle occurs in summer and the smallest in the winter. Wind speed variability (the standard deviation) exhibits a marked diurnal cycle at all heights, with highest variability during the day and lowest at night; standard deviations at 80 and 128 m are nearly identical over the diurnal cycle. Daytime wind speed probability density functions (pdfs) at 10 m conform closely to a Weibull distribution but pdfs for 30, 80, and 128 m show progressively more positive skew than would be represented by a Weibull fit. As has been found in previous work, nighttime pdfs are more highly skewed than in the daytime, confirming that the Weibull distribution may be a poor estimator of wind speeds across the rotor disk, especially at night.

Wind shear exhibits a marked diurnal cycle at this site. For the lower part of the rotor (30-80 m), shear is around 0.40 at night and 0.15 during the day; across the upper part (80-128 m), shear varies from around 0.44 (night) to 0.18 (day). Turbulence intensity (TI: the wind speed standard deviation divided by the mean speed) exhibits a more muted diurnal cycle with higher values during the day and lower values at night, with the diurnal variation decreasing at progressively higher heights. Diurnal variability in turbulent kinetic energy (TKE) is opposite what is observed with shear: TKE is highest during the day and lowest at night. Diurnal TKE is driven largely by convection at this relatively flat site and the larger daytime instability (higher TKE) promotes mixing and thus lowers shear, whereas the more stable nocturnal surface layer (lower TKE) is characterized by much higher shear. This variability in shear is consistent with observations at other tall-tower sites and underscores the need to account for stability effects on the wind profile when estimating the wind resource.

Seasonal variation in shear and turbulence is markedly different than the diurnal pattern. At seasonal time scales, variation in TKE strongly resembles the variation in the frequency and intensity of synoptic-scale systems, suggesting that at these time scales, TKE is dominated by mechanical ("weather"-related) rather than buoyancy effects. Low summer TKE is coincident with the higher shear and TI observed during these months, when there are fewer/less intense cyclones. Higher TKE coincides with lower values of shear and TI, especially in the spring when latitudinal temperature gradients are strongest, supporting stronger cyclonic systems. Curiously, the secondary peak in TKE observed in autumn does not clearly coincide with lower shear and TI, although both shear and TI are smaller in autumn compared to their summertime values.

Wind speed and turbulence characteristics were also grouped by synoptic circulation regime as classified by the Lamb scheme. As expected, anticyclone patterns are associated with slower wind speeds compared to cyclonic and directional-flow regimes. There is no clear relationship between circulation types and shear or TI, which appear to depend primarily on the specific location of synoptic systems in relationship to the tower. Lower values of TKE, however, are clearly associated with anticyclonic types and strong peaks in TKE were associated with specific cyclonic types. Assessment of the impacts of synoptic conditions on variables such as wind speed and TKE may yield important information on the long-term impacts of changes in synoptic systems on the productivity of wind energy facilities, and this is the subject of ongoing research.

Acknowledgement: Our sincere thanks to Chris Milliren for providing the quality-controlled Eolos met tower data set.