# Horizontal and vertical variability of wind fluctuations in lower atmospheric boundary layer

## Date

## Authors

## Journal Title

## Journal ISSN

## Volume Title

## Publisher

## Abstract

Horizontal and vertical characteristics of wind flows are investigated related to large wind farms and tall wind turbines, up to 300 m tip height. The topic is investigated in three parts. For the first part of the dissertation, spatial and temporal variability of horizontal wind speed fluctuations have been investigated over an existing wind farm in terms of temporal scale. Wind flows are variable spatially and temporally in the atmospheric boundary layer within which wind energy is produced. In temporal variability, intermittent wind fluctuation would be a critical issue as related to wind power ramp events. Also we investigate whether we should consider the spatial variability of wind speed or use averaged wind speed over a whole wind farm for wind farm power production estimates in terms of assumed wind power forecasting. We investigate which time scales are more related to abrupt wind speed changes. Also the spatial variability for this wind farm size scale is investigated in relation to time scales and wind farm power production estimates. In the second part, we investigate the long term spectral structures of horizontal wind speeds at two different heights and the sensitivity of the turbulence intensity with different averaging time scales depending on the existence of a spectral gap. In the research, total 5 year, 1 minute wind speed time series at 10 m and 100 m height above the ground are used and the site is located around the Rocky mountain range. In terms of the spectra of 5 year, 1 year and seasonal time series, it is found that the diurnal cycle dominates horizontal wind speed fluctuations at both heights. Also, we investigate the difference between the spectral contributions to variance of 5 year and seasonal wind speed time series at 10 m and 100 m in terms of frequency range. Spectral contributions of higher frequency range (higher than 〖10〗^(-4) Hz) decrease with increasing heights from 10 m and 100 m above the ground. 10 m is the standard height of the surface wind measurement and 100 m is the typical hub height of the current wind turbines. It indicates that there are larger higher frequency fluctuations relative to its total variance at 10 m than at 100 m and there can be overestimation of high frequency fluctuations if 100 m horizontal wind flows are estimated from wind flows at 10 m height above the ground. To consider the different spectral characteristics of individual cases, we categorize daily timeseries with spectral density at the spectral transition or spectral gap region between 〖10〗^(-3) and 〖10〗^(-4) Hz at 100 m. In the relationship between turbulent intensity and wind speed, the group of daily time series which have highest spectral density in the spectral gap region shows that turbulence intensity changes more sensitively to the averaging scale chosen in the spectral gap range than the group of daily time series which have lowest spectral energy in the gap region does. For the last part, we investigate the wind fluctuation characteristics at the multiple heights from 5 m to 300 m considering influence of temporal average. We discuss about what could be a standard for identifying the turbulence fluctuation in the horizontal wind speed magnitude, when the spectral gap doesn’t exist. Also if the temporal average scale changes, how would it influence on the variance of wind speed fluctuation as well as the probability distributions of horizontal wind speeds at different height levels. At all heights, the spectral peaks of horizontal wind speed fluctuation are located at the diurnal cycle of 24 h showing that the most of the spectral energy of the horizontal winds exist in mesoscale or large scales up to 300 m height above the ground. On the contrary, the most of spectral energy for vertical velocity exist in the turbulence scales smaller than the temporal scale of 1h while the spectral peaks of vertical velocity increase with the heights with the larger segment variations at the higher heights. The vertical velocity spectral peaks (Λ_w) at each height are used as the turbulence characteristic scales to normalize the temporal averaging scales for generalization. The most of vertical velocity variance are removed at all the height levels with temporal averaging scales of 100 Λ_w except at 5 m. However with 100 Λ_w the variances for the horizontal wind fluctuations are not much removed as for the vertical velocity. Also the low-pass filtered contribution to variance reduce more rapidly in higher height levels in the normalized temporal averaging range between 2 to 2000 Λ_w. Also, in terms of the probability distribution of wind speed magnitudes, the temporal averaging lowers the probability of high wind speed especially over about 5 m s^(-1). This effect is larger with larger averaging scale. The effect of lowering the probability in higher wind speed range decreases with height indicating that temporal averaging is more reliable in higher height in terms of more accurate presentation of original wind speed distribution which could be related to more accurate estimation of wind energy resources.