Back to the programme printer.gif Print




Delegates are invited to meet and discuss with the poster presenters in this topic directly after the session 'Advanced rotor technologies' taking place on Tuesday, 11 March 2014 at 11:15-12:45. The meet-the-authors will take place in the poster area.

Soo-Hyun Kim Korea Institute of Energy Research, Korea, Republic of
Co-authors:
Soo-Hyun Kim (1) F P Hyung-Ki Shin (1) Hyung-Joon Bang (1) Moon-Seok Jang (1) Young-Chul Joo (1)
(1) Korea Institute of Energy Research, Daejeon, Korea, Republic of

Printer friendly version: printer.gif Print

Presenter's biography

Biographies are supplied directly by presenters at EWEA 2014 and are published here unedited

Mr. Joo is currently a technologist at the Korea Institute of Energy Research in Daejeon, Rep. of Korea, and he has been involved in various wind turbines design and blade development projects.
He studied structure and production engineering at the Korea Aerospace University, and was granted a M.S degree in 2010.
His research is on the analysis of the wind turbines design and feasibility study and construction of wind farms.

Abstract

Turbulent wind seed selection using robust statistics for design load analysis of wind blade design

Introduction

Wind energy is one of the fastest growing renewable energy technologies. Due to meet that demand, the commercial wind turbines have been developed consistently in size over the years. Because of the huge size and severe external loading conditions, the load calculation of wind turbine system should be effective as well as conservative. However an excessively conservative load in blade design would lead to an over-sized rotor design and consequently result a higher cost on the entire systems. Therefore an appropriate method for design load calculation of wind turbine has to be selected carefully.

Approach

The IEC 61400-1 standard provides the stochastic turbulence wind models to be generated with the specified mean value (mean wind speed at hub height) and the standard deviation (turbulence intensity). Because of its statistical characteristics, the load simulations using the turbulent wind model should be performed with different initial values (“seeds”) for producing the turbulent wind field. As the difference in seed values of turbulent wind model could lead to considerable variations in the results, the seeds were selected for turbulent wind generation prior to wind turbine analyses.

Main body of abstract

In this paper, a turbulent wind seed selection method is analyzed using robust statistics, which provides an alternative approach to standard statistical methods. Design load simulations of turbulent wind were performed for two different wind turbine system; 2MW and 7MW class models. For the design load simulation of turbulent wind, 15 wind fields with different random seeds were generated for two system models.
To choose the best appropriated turbulent wind seeds for blade design load simulations, the standard score (z-score) of classical and robust statistics were used to the extreme load results. The classical statistic methods rely heavily on assumptions that the data errors are normally distributed, so it often has poor performance when there are outliers in the data or other small departures from model assumptions. In the robust statics, the median and normalized interquartile range (IQR) values are used, and it is known that the robust estimators are less affected by outliers than classical estimators.
Classical and robust z-scores were calculated from the analytical results of the blade bending moment. We could find that differences of 10~25% were observed according to wind seed variations. We also found that classical and robust statistics analyses yielded different results to select the three best wind seeds. This is caused by the influence by some outliers which lead to the difference in the mean value and median value of extreme load results. Therefore it could be considered that the robust z-score suggest more proper representative value.


Conclusion

The 15 wind fields with different random seeds were generated at the rated wind speed and cut-out wind speed for the design load simulation of turbulent wind. To select the best appropriate turbulent wind file for the blade design load, classical and robust z-scores were calculated from the extreme load results. The results of classical and robust statistics analyses show different results to select the three best wind seeds. As the comparison result shows that the classical z-score had more influence by some outliers than robust z-score, we could conclude that the robust z-score suggest more proper representative value.


Learning objectives
To avoid an excessively conservative load in blade design, an appropriate approach for design load calculation with the turbulent wind model is suggested.
The extreme peak load results of 15 wind fields with different random seeds show that we should conduct a proper seeds selection process prior to design load simulation.
A turbulent wind seed selection method was studied using classical and robust statistics, and the two results are compared.