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 'Aerodynamics and rotor design' taking place on Wednesday, 12 March 2014 at 09:00-10:30. The meet-the-authors will take place in the poster area.

Iino Mitsumasa The University of Tokyo, Japan
Co-authors:
Mitsumasa Iino (2) F P Makoto Iida (2)
(1) a, a, Japan (2) The University of Tokyo, Tokyo, Japan

Printer friendly version: printer.gif Print

Presenter's biography

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

Mitsumasa Iino is a Ph.D student at the Research Center of Advanced Science and Technology at The University of Tokyo. He got a master’s degree at department of mechanical engineering in The University of Tokyo. He has been studying small wind turbine aero-elastic modeling and simulation method of dynamic behaviour of horizontal axis small wind turbines. His research focuses on load estimation and design method of small wind turbines.

Abstract

Probabilistic fatigue analysis of small wind turbine blade using aeroelastic analysis

Introduction

Type certification is needed for SWT in many countries for Feed in Tariff. In the type certification, design evaluation is one of the important processes. One of the most recognized criteria for SWT is IEC61400-2 Ed.2.(1) In this Criteria, Fatigue Load calculation without experiment can be done by two options, “Simplified Load Model(SLM)” and ”Aeroelastic Modelling”. Assumption of SLM is very limited and deterministic but easy to apply, and Aeroelastic modelling considers some probabilistic phenomenon but is not as frequently used as SLM. This research compared two method for single turbine and discussed the difference.

Approach

This research aims to establish reasonable and economical load estimation method. In this abstract, fatigue load calculation was done by aeroelastic modelling that considers probabilistic phenomenon from coupling of operating condition and wind speed distribution. Then the fatigue load were compared and discussed with fatigue load calculation by SLM that assumes simple deterministic operating condition.
The target turbine is AURA1000 5-bladed fixed pitch micro wind turbine with 1m rotor diameter. The turbine is designed to IEC61400-2 Class 1. Simplified load model of the standard is used for the structural design.
The overview of fatigue load calculation by SLM and Aeroelastic modelling is explained below.
In the SLM, wind condition is assumed to be around Design wind speed Vdesign. Vdesign is defined as 1.4 Vave which is the annual average wind speed defined in IEC wind turbine class. Turbine classes are defined from Class I to IV, from the highest annual average wind speed to lowest one. From these wind condition corresponding operating condition, the fatigue load amplitude ΔMyb and number of cycle n are defined in below equation
λdesign,Qdesign,ndesign are tip speed ratio, turbine torque and rotational speed at Vdesign. B is number of blades,Td(s) is turbine lifetime that corresponds to 20 years.
Next, fatigue load from aeroelastic model(2)(3) is described below. Before the fatigue analysis, time series aeroelastic calculations are performed. Wind condition is set to be Normal Turbulence Model(NTM) in IEC61400-2 Ed.2. Average wind speed is changed from the cut-in 2m/s to cut-out, 16m/s. wind speed. Wind speed is binned with 1m/s bin width. At each wind speed bin, 10 cases of 10 minutes simulation are performed.
On the fatigue analysis, the rainflow count for load cycle counting is applied for each wind speed bin. Lifetime damage is calculated by weighting the cycle count at each wind speed bin by probability of wind speed. The cycle is extrapolated for 20 years that correspond to turbine's designed lifetime.
The S-N curve is obtained by full blade fatigue test and, parameters of aeroelastic modelings are based on design data from the manufacturer.


Main body of abstract

This abstract describes the difference between fatigue load from aeroelastic modelling and SLM.
First, the fatigue load from aeroelastic modelling is described. An aeroelastic simulation code for 5 bladed wind turbines was used. This code is based on NREL's FAST(4) and modified by authors. The calculation is based on blade element momentum theory(5) and multibody dynamics model(6) with the flexible blade and tower by mode summation method.
Prior to lifetime fatigue load estimation results, damage equivalent load(DEL) at each wind speed bin are presented in Figure1. In the latter contents of this abstract, DEL is equivalent to 1Hz cyclic load and normalized by flapwise ultimate strength of the blade. From the figure flapwise DELincreases as wind speed increases, unlike edgewise DEL. The slope of flapwise moment is not proportional to the wind speed and gradually become gentle over 10m/s.
Next, the lifetime DELs for all IEC wind turbine class obtained by aeroelastic modeling and SLM described in Approach section are presented in the Figure 2. The wind speed distribution used in aeroelastic modelling and Vdesign used in SLM are presented in the Figure 3. The evaluation was applied on the flapwise direction. This is because it is superior to the edgewise one except in the low wind speed range, thus it is a main cause of fatigue load when wind speed distribution is considered.
Other than the Class I, DEL from aeroelastic modeling is superior to the one from SLM. The DEL from SLM is at most about 60 percent less than the DEL of aeroelastic modeling.
In order to clarify the reason of the difference between the DEL by two methods. The fatigue equivalent cycles NDELl in equation (2) is introduced for the evaluation result by the aeroelastic model.
NDELl means how wind speed bin l contributes to the total fatigue load by the representation of cycles. 1Hz Lifetime equivalent cycle TLife can be obtained by taking the summation of NDELl for all wind speed bin. In the equation DELLife is lifetime damage equivalent load shown in Figure2 and NDELl is DEL in each bin shown in Figure1, TLife l is the period of wind speed bin l during lifetime. Figure 4 shows NDELl for each wind speed bin and class. From Figure 4, high wind speed region upper 10m/s is dominant for DEL in spite of the different annual average wind speed for each turbine class. So it is important to consider all the operating wind speed range and distribution on the fatigue load estimation although the probability of high wind speed is relatively low. On the other hand, SLM assumes the wind condition only around Vdesign. Because Vdesign becomes lower as class number increases the consideration of high wind speed may not be sufficient with SLM. This is one of the reasons of the lower DEL from SLM than that from aeroelastic modeling.
Another reason is that the operating condition assumed in SLM is different from the condition of aeroelastic modelling that considers turbine control and structural response. SLM considers operation around Vdesign. So the resultant DEL is considered to be the DEL around Vdesign. In this abstract aeroelastic modelling around Vdesign can be obtained by extracting the result for each wind speed bin. Figure 5shows the two results. One is DEL from SLM for each wind turbine class that assumes operation around 14m/s 11.9m/s, 10.5m/s, 8.4m/s and the other is DEL from aeroelastic modelling at 14m/s 11m/s 10m/s 8m/s that is nearby the assumption of SLM. From Figure 5 SLM shows lower DEL around Vdesign, this is because the total response of target turbine, including speed control or turbine inertia, for wind speed change is not fit to the assumption of SLM. This is another reason that DEL from SLM is lower than DEL from aeroleastic modelling.


Conclusion

In this abstract, fatigue load estimation is done by aeroelastic modelling that consider the uncertainty of control tracking delay, wind speed distribution, turbulence etc.. Then the DEL from SLM and aeroelastic modelling are compared in the manner of IEC61400-2 Ed.2.
As a result, SLM showed up to about 60% lower DEL than the DEL from aeroelastic modelling. To clarify the reason of the difference, the contribution of wind speed bin is first considered. Then it is confirmed that for the target turbine the high wind speed region is dominant for DEL in Aeroelastic modelling. And because SLM does not consider the wind speed range, consideration of higher wind speed region is suggested to be one of the reasons of difference. Next the difference of the operating condition around Vdesign is compared. Although SLM assumes operation around Vdesign, the DEL around Vdesign is different between SLM and aeroelastic modelling. This is because overall response of the target turbine for wind speed change is considered in aeroelastic modelling based on design parameters including control or structural properties. On the other hand, control or structural response is not considered in the SLM. So as a whole, the DEL from SLM and the DEL from aeroelastic modelling are different because the assumptions of wind speed distribution, control and structural response is different in SLM and aeroelastic modelling. And the assumptions are not compatible in most cases.
In future, We aim to improve the accuracy of aeroelastic modelling that can reasonably represent the experimental value and consider the elements that lack in SLM, and establish more economic, reasonable load evaluation method.



Learning objectives
This research objectives are to understand the difference of fatigue load on the small wind turbine blade based on Simplified load model and the one based on aeroelastic model quantitatively in the manner of IEC61400-2 Ed.2 and to understand what makes the difference between two methods.




References
(1) International electrotechnical commission ,Wind turbines. Part 2: Design requirements for small wind turbines(2006), International electrotechnical commission.
(2) International electrotechnical commission ,Wind turbines. Part 1: Design requirements (2005), International electrotechnical commission.
(3) Moriarty P., Holley W., and Butterfield S., Extrapolation of extreme and fatigue loads using probabilistic methods(2004), National Renewable Energy Laboratory.
(4) National Renewable Energy Laboratory, “NWTC Design Codes (FAST by Jason Jonkman, Ph.D.).” ,National Renewable Energy Laboratory, http://wind.nrel.gov/designcodes/simulators/fast/
(5) Moriarty, P. J., & Hansen, A. C.. AeroDyn theory manual (2005), National Renewable Energy Laboratory.
(6) Kane, T. R., and Levinson, D. A., Dynamics: Theory and applications(1986), McGraw-Hill.