PhD Thesis
Measuring and understanding site-specific wind and turbulence characteristics for wind energy applications
Lukas Pauscher (01/2013-08/2017)
Support: Thomas Foken
Onshore wind energy has become the most important source of renewable energy in Germany. This success also lead to rapid developments in turbine technology. The strongly increased turbine size creates a need for the application of new measurement technologies to replace tall, expensive and in exible measurement masts. For this reason, doppler-lidar measurements have become increasingly popular in the wind energy community. Yet, for measurements in complex terrain and for turbulence measurements, conically scanning lidars still su er from systematic errors. However, especially in complex terrain accurate measurements of the mean wind speed and turbulence intensity are key for resource site suitability assessment, as modelling is often associated with high uncertainties. Also, the relation of observed turbulence quantities to surface characteristics and the atmospheric stability regime is often dicult for the experimentalist. This dissertation focuses on the aforementioned problems and presents di erent approaches to resolve them.
The complex terrain error of conically scanning lidars is analysed experimentally and using ow modelling. The simulations revealed a high sensitivity to land cover, especially forest generally reduces the error. Among the investigated models, the linear ow model showed the worst performance in prediction the observed error. The RANSmodels could reproduce the right pattern and magnitude of the lidar error.
As an alternative the multi-lidar (ML) approach investigated. The value of the ML-approach, when compared to a single conically scanning lidar, is demonstrated experimentally in complex terrain for the rst time. Given an appropriate setup, measuring with two or three lidars in the same point signi cantly improved the agreement with a high quality reference measurement. This can greatly reduce the uncertainties of lidar measurements in complex ow regimens.
The multi-lidar (ML) approach was also experimentally demonstrated to be a promising approach to measure turbulence statistics in complex terrain. In contrast, the conically scanning lidar showed a strong overestimation, when compared to the reference measurements. One of the problems which remains unsolved in the ML-approach is the attenuation of small scale turbulence by the spatial averaging of the lidar. This was also clearly visible in the spectral analysis of the ML-experiment. This dissertation approaches the problem by investigating the capability of a pulsed lidar to measure the dissipation rate of turbulent kinetic energy (TKE) using three di erent methods: A previously described method based on short term variances is corrected to remove signi cant systematic errors which were previously present. Moreover, a theoretically suggested method based on the structure function of the radial velocity uctuations is experimentally evaluated for the rst time. The third approach used the power-spectral density in the inertial sub-range. It is shown that, given the knowledge about the spatial averaging function of the lidar and a careful removal of the noise, the dissipation rate of TKE can be estimated with a reasonable accuracy. However, the experimentally determined form of the spatial averaging function and the one derived from theoretical considerations showed signi cant di erences. The di erences between the investigated methods to derive the dissipation rate of TKE are mainly found in their applicability to di erent experimental setups with the structure-function approach providing the most exible option.
Finally, the dissertation investigates observed turbulence quantities and their relation to surface characteristics and stability at a 200-m-mast at a forested hilltop site. A simple approach based on footprint modelling is developed to characterise the surface ruggedness and roughness in the area of e ect of the measurement. It is shown that especially the normalised standard deviation of the wind along the stream lines exhibits a high correlation to the surface characteristics within the footprint. Atmospheric stability also had a strong in uence on the representative turbulence intensity at the investigated hilltop site. The prevalence of stable conditions for wind speeds between 6- 12 m s-1 lead to a signi cantly reduced turbulence intensity in this wind speed range, which is in the order of the di erence between standard turbulence classes for wind turbines.