- |
Accuracy of load assessments based on modelled turbulence - The German example
![]() |
Lasse Svenningsen EMD International A/S, Denmark ACCURACY OF LOAD ASSESSMENTS BASED ON MODELLED TURBULENCE - THE GERMAN EXAMPLE Abstract ID: 309 Poster code: PO.253 | Download poster: ![]() ![]() |
Presenter's biography
Biographies are supplied directly by presenters at WindEurope 2016 and are published here uneditedLasse Svenningsen has been employed at EMD International A/S since 2008 and now holds the position of R&D Manager. In his work, Lasse mainly focuses on EMD’s external research and innovation projects as well as internal development projects related to EMD’s software products and services (e.g. windPRO and windPROSPECTING). In recent years, Lasses R&D work has focused on turbine load estimation in site suitability assessments as well as mesoscale modelling and downscaling using linear and non-linear microscale models. Lasse holds a PhD in Geophysics from the University of Aarhus in Denmark.
Abstract
Accuracy of load assessments based on modelled turbulence - The German example
Introduction
The objective of this study is to investigate the accuracy of wind turbine fatigue load estimates based on modelled ambient turbulence for sites typical of Germany.
Approach
We apply two commonly used microscale models, a linear and a non-linear model, to predict on-site ambient turbulence levels. The microscale models are applied to 23 locations with high quality wind measuring masts, mostly 100 m or taller. For each mast position, fatigue loads are estimated for the main turbine components, first using the measured turbulence data from the top level anemometer and second using the modelled turbulence of the different microscale models. The results allow a direct assessment of the error introduced by using the modelled turbulence.
Main body of abstract
In mature wind energy markets such as Germany and Denmark with thousands of operating wind turbines, it is not common practice to install wind measurements prior to developing a new wind power project. Instead, production data from surrounding wind farms is used to calibrate the wind flow model. This is a well-established approach resulting in sufficiently accurate production estimates for investment and financial decisions.
However, for turbine site suitability evaluation according to the IEC61400-1 or DIBt standards, additional wind climate parameters are required: turbulence, wind shear, vertical inflow angle and air density. As the main driver of fatigue loads, turbulence is the most critical of the required site parameters, in particular in projects with many existing or planned turbines and increased turbulence due to wake effects. When no measurements are available, suitability decisions are based solely on modelled values of the ambient turbulence.
Previous studies on the accuracy of modelled ambient turbulence in a German context focused on the detailed aspects of the accuracy of modelled versus measured turbulence. Aspects such as accuracy of capturing the directional variation of turbulence, accuracy of capturing the turbulence variation with height and accuracy of the turbulence variation with wind speed. These analyses are relevant and interesting for evaluating the accuracy of the microscale flow models to predict details of the turbulent wind field. Due to the highly non-linear and complex relationship between turbulence and fatigues loads, it is very difficult to conclude directly from these analyses to the consequence for the accuracy of the resulting fatigue loads based on the modelled turbulence. It is the aim of this study to contribute to the insight into the topic by focusing the analysis directly on the fatigue load parameters, which determine the wind turbine site suitability decisions.
Conclusion
The results of this study show that turbulence from the non-linear microscale model leads to the most accurate fatigue load estimates for the 23 investigated mast locations. The mean bias is just +2% for blade flap-wise loads and -1% for tower for-aft loads, both with a standard deviation of 5%. This result suggests that load estimates based on the non-linear model are virtually bias free. Results for the linear microscale model show slightly larger mean biases and a lower precision as the standard deviation is 7% for both blade and tower loads.
Learning objectives
What is the consequence of basing load assessments on modelled instead of measured turbulence?
Which microscale model’s turbulence prediction will result in the most accurate load estimates?
Does use of modelled turbulence lead to reliable site suitability decisions?
What is the sensitivity of modelled turbulence and fatigue loads to the roughness assumptions in the microscale model?
Which additional assumptions are required when using modelled turbulence in suitability assessments?
Follow EWEA on: