外文翻譯--基于特定動態(tài)方法的海上風力發(fā)電轉換器的優(yōu)化設計（英文）

1、1999 European Wind Energy Conference and Exhibition (EWEC 99), March 1 - 5, 1999, Nice, France 1 / 4DESIGN OPTIMISATION OF AN OFFSHORE WIND ENERGY CONVERTER BY MEANS OF TAILORED DYNAMICSM. Kühn Institute fo

2、r Wind Energy, Delft University of Technology Stevinweg 1, 2628 CN Delft, The Netherlands Tel. +31 15 278 51 70 Fax +31 15 278 53 47 Email: m.kuehn@ct.tudelft.nl www.ct.tudelft.nl/windenergy/ivwhome.htmABSTRACT: Ta

3、iloring the dynamics of an offshore wind energy converter can offer an effective design optimisation during the successive stages of the design process. Concerning the particular problem of fatigue due to combined wind a

4、nd wave loading two simplified approaches are proposed and demonstrated which are well suited for the early design stages when integrated, non-linear time domain simulations are too cumbersome. This enables the use of st

5、andard design tools from the wind energy and offshore technology communities by superposition of separate analyses of hydrodynamic fatigue in the frequency domain and aerodynamic fatigue in the time domain. Keywords: Off

6、-Shore, Optimization, Dynamic Models, Fatigue1. INTRODUCTIONAn important step towards cost-competitive large-scale offshore wind farms lies in the application of an integrated design approach considering the entire syste

7、m, including the individual offshore wind energy converters (OWEC), grid connection and operation and maintenance aspects. Within such a methodology the design process is controlled by major aspects of the total system,

8、e.g. energy costs, power quality and persistence, system reliability. With regard to the offshore wind energy converter, i.e. support structure and wind turbine, the overall dynamic behaviour is an important design crite

9、rion. During the Opti-OWECS project [1], an integrated design approach was demonstrated and design was optimised by tailoring OWEC dynamics. Table I illustrates how the economic performance, but also the effort involved

10、in the design calculations, increased during the successive development of three OWEC concepts. During the project only the powerful but expensive method of the integrated, non-linear time domain simulations was availabl

11、e and it became apparent that analytical tools for the entire OWEC with different levels of complexity are required. Concerning the particular problem of fatigue due to combined wind and waves loading, this paper propose

12、s two approaches and illustrates each of them by an example. For a firm background and full details the reader is referred to [2]. Table I: Comparison of OWEC concepts2. ESSENTIALS OF OWEC DYNAMICSOverall dynamics shou

13、ld be considered from the outset of the design work. Obviously the extension of the applied models and investigations will be limited at the beginning and can become sophisticated near the end. In each case, one has to b

14、e aware of a number of aspects particular to OWEC. Due to lack of space these issues can be described here only briefly while the rest of the paper is dealing exclusively with fatigue due to combined wind and wave loadin

15、g. Nonetheless it is important to keep in mind that a successful design process requires a balanced treatment of all aspects mentioned. ? Environmental descriptionIn general, the environmental conditions of offshore sit

16、es differ more significantly than for onshore sites. Directional characteristics of wind and wave conditions (Figure 1) are pronounced due to changes in fetch, roughness and water depth. Thus for each considered site a p

17、articular set of environmental conditions must be established and refined during the design process. High quality databases of correlated environmental conditions are extremely useful [3].GBS soft-stiff monotowersoft-sti

18、ff monopile Soft-soft monopiledesign driver extreme wave aerodyn. fatigue aerodyn. & hydrodyn. fatigue, stiffnesssupport structure cost 100 % 50 % 40 %energy cost 100 % 85 % 80 %01234560 5 10 15 20 25 30hourly wind s

19、peed at 60 m height [m/s]si gn ifi ca nt wa ve he ig ht [m ] direction 1direction 2lumped load casesFigure 1: Scatter of environmental state parameters for two directional sectors at a Dutch North Sea site and ‘lumping

20、’ of load cases for fatigue analysisM. Kühn., Design Optimisation of OWEC by Means of Tailored Dynamics, EWEC ’99 3 / 4domain and the wave response computed in the frequency domain is possible. This is compatible wi

21、th the distinct design tools available in the wind energy and offshore technology communities.3.3 Simplified treatment of short-term fatigue due to combined wind and waves The proposed superposition of the damage equival

22、ent stress ranges of two independent Gaussian processes is based on the linear superposition of the associated variances. The application of a frequency domain expression for the equivalent stress range and the assumptio

23、ns of a SN curve with constant slope µ rewrite this relation. Both the closed-form expression for the equivalent stress range of a process with narrow bandwidth and the semi-empirical expression for a wide bandwidth

24、 proposed by Hancock produce the same result if the equivalent stress range is related to a constant, but arbitrary, reference number of cycles. Therefore the ‘equivalent reference fatigue stress’ ?σeq.R is used. The equ

25、ivalent reference stress ranges of the two responses are weighted by µth root the ratio between the zero-crossing periods Tz of the component process (index a = aerodynamic, h = hydrodynamic) and the combined respon

26、se (index ah). Successively the contributions are added as the root of the sum of the squares. The weighting factors are rewritten in terms of the zeroth and second spectral moment, m0 and m2, since the zero-crossing per

27、iod of the combined response is not directly available.All quantities can be obtained from either time or frequency domain representations of the component responses. For instance, the equivalent reference stress range c

28、an be calculated by Dirlik’s method in the frequency domain while Rainflow Counting is applied in the time domain. The spectral moments of the wind response can be derived from the standard deviation and the counting of

29、the mean zero-crossing period after normalising to zero mean. Expression (1) was investigated by an extensive numerical study considering different types of response spectra, including bimodal block spectra, simplified s

30、pectra for typical wind and wave responses and response spectra from actual simulations of an OWEC. The bandwidth parameter β (the ratio between number of zero-crossings and peaks) of the combined response was varied fro

31、m 0.15 to 1 and the ratio between the variance of the two processes ranged between 0.1 and 0.9. Figure 3 shows the error of Eqn. (1) relative to the result of Rainflow Counting of the superposition of the component time

32、series. The scatter of the error increases for lowers β (i.e. greater bandwidth). It is surprising that for the realistic OWEC spectra the error is in most cases smaller than 5% while the results for some simplified spec

33、tra are poorer. There is a tendency towards conservative predictions, however, slightly unconservative results can not be excluded.3.4 Simplified treatment of long-term fatigue due to combined wind and waves No theoretic

34、al evidence exists that Eqn. (1) can be applied directly for the entire lifetime when statistical properties are varying in time. However from analysis of the long- term fatigue loading of the soft-soft monopile of the O

35、pti- OWECS design solution it became apparent that the equivalent reference stress of the combined long-term response can be approximated by superposition of the equivalent reference stresses according to the law of Phyt

36、hagoras (Eqn. 2).So far, it was not possible to perform a comprehensive simulation study on the superposition of long-term fatigue since this involves several times more computational effort than the previously described

37、 investigation for stationary processes. Instead, expression (2) was checked with the results from the short-term fatigue simulations. In general, the scatter of the error is lower for Eqn. (2) than for Eqn. (1). Howeve

38、r, Eqn. (2) showed a tendency to underprediction of the short-term fatigue loading up to 10 %. This behaviour is undesirable but the magnitude is much lower than other uncertainties in the early design stage, e.g.

39、due to the approximation of the aerodynamic damping. Consequently, use of Eqn. (2) can be of great value in such situations, as long as it is used with caution4. EXAMPLE APPLICATIONS4.1 Lumping of fatigue load cases For

40、an OWEC at least an order of magnitude more fatigue load cases exist than for an onshore wind energy converter. For instance, evaluation of the NESS database [3] for a Dutch North Sea site with class widths of 1 m/s for

41、the wind speed in the production range, 0.5 m for the significant wave height and 0.5 s for the wave period results in 141 load cases with a minimum probability 0.5 ‰. Consideration of 12 directional sectors increases th

42、is number to about 500. Evidently both numbers are too high for integrated time domain simulations and establishment of a lower number of characteristics load cases is required. This so-called called ‘lumping’ of load ca

43、ses is well known in the offshore technology and assumes quasi-static response on wave loading only. In present case the problem is more involved since OWEC suffer both significant wind and wave fatigue and dynamic respo

44、nse is important.-10%-5%0%5%10%15%20%0.0 0.2 0.4 0.6 0.8 1.0ß : bandwidth parameter of combined response [-]error of equiv. stress range [%]simplified OWEC spectra bimodal spectra (low bandwidth)realistic OWEC spect

45、ra bimodal spectra (great bandwidth)Figure 3: Error of Eqn. (1) with respect to Rainflow Counting for different spectra and bandwidth?? ??? ? + ? ++ =? ? ?? ??? ? + ? ? ?? ??? ? = ?2 , ., 2, 0 2 , ., 2, 0, 0 , 0, 2 , 22