EUROMECH Colloquium 464b Wind Energy

EUROMECH Colloquium 464b Wind Energy International Colloquium on Fluid Mechanics & Mechanics of Wind Energy Conversion

October 4 - 7, 2005 University of Oldenburg, Germany

www.forwind.de/euromech • [email protected] • phone: +49-(0)441-36116-720

© 2005 Die Alchemisten GmbH, Oldenburg

• wind climate & wind field • gusts, extreme events & turbulence • rotor aerodynamics & wake effects • power production & fluctuations • sea states & wave loads • materials & fatigue • structural health monitoring

Preamble of the EUROMECH Colloquium No.464b - Wind Energy October 4 – 7, 2005 Carl von Ossietzky University of Oldenburg Oldenburg, Germany Welcome to the EUROMECH Colloquium No.464b – Wind Energy! Chairs of the Organizing Committee are Prof. Dr. Joachim Peinke and Prof. Dr.-Ing. Peter Schaumann. Members of the Scientific Committee are Prof. Dr. Martin K¨ uhn, Prof. Dr. Gijs van Kuik, Prof. Dr. Soeren E. Larsen, Prof. Dr. Eur.-Ing. Ramgopal Puthli and Prof. Dr. Daniel Schertzer. The meeting is being hosted by ForWind – Center for Wind Energy Research at the Carl von Ossietzky University of Oldenburg and the University of Hannover. Local organizers are: Moses Kärn, Elke Seidel, Margret Warns, Martin Grosser, Stephan Barth, Frank B¨ ottcher. The colloquium is supported by: EUROMECH, Federal Ministry for Education and Research, City of Oldenburg, University of Oldenburg. For additional information, please refer to the meeting’s website at http://www.forwind.de/euromech/.

Meeting Site The meeting is held at the Carl von Ossietzky University of Oldenburg, Wechloy campus. The University is located at the west end of Oldenburg with easy access to the downtown area by public transfer. The Wechloy campus can be reached from downtown with the bus line 306 and is the terminal stop. The city of Oldenburg offers a variety of events, galleries and museums at which visitors and residents can experience art, history and culture. And Germany’s first pedestrian area offers many shopping opportunities. We encourage you to discover and explore the vitality that makes Oldenburg a great place to live and visit.

Oral Sessions All oral presentations will be held in the lecture hall (W3-1-161) of the Wechloy campus. Contributed papers are limited to 10 minutes (15 minutes for invited speakers) with an additional 4 minutes for discussion and 1 minute for transition to the next speaker. To avoid disadvantages of following speakers we will use restrictive time control: A warning via a timing monitor will be given at 8 minutes (13 minutes for invited speakers) to indicate that the speaker has to finish in 2 minutes. At 10 minutes (15 minutes for invited speakers), a warning will indicate that the speaker’s presentation time is over. At 14 minutes (19 minutes for invited speakers), a signal informs the speaker and audience that the transition to the next speaker must take place.

Poster Session Posters will be presented in front of the lecture hall. The poster session will be on Thursday, Oct. 6th , but posters will be up during the whole meeting. Poster boards and pushpins will be provided. The poster boards will be labeled with a letter number combination. Authors must hang up their poster at the board labeled with the combination that is given for their poster in this program. All authors of a poster have the opportunity to advertise their poster in a short oral presentation (limited to 2 minutes and 2 slides) at the shotgun-session on Wendesday, Oct. 5th .

Audiovisual Equipment The lecture hall will be equipped with an LCD projector, a computer with PowerPoint and Acrobat Reader installed on it, overhead projector, screen, pointer, and a microphone. Speakers should hand over their presentation to the organizers at the meeting check-in upon arrival and have it tested on the computer used for presentation. Speakers are strongly advised NOT to use their own laptop for presentation. Any loss of time due to technical problems will otherwise be taken off their time to speak.

Conference Dinner A highlight of the meeting will be the Conference Dinner on Thursday evening, October 6th , 2005. The Reception will be held on a boat on Lake Bad Zwischenahn. Moor, heath, woods and meadows, well-tended parks with mighty trees and old farmhouses characterize the wonderful countryside around the lake of Zwischenahn – the “pearl of the Ammerland”. All fully registered attendees receive a ticket to this event which includes free food and one welcome drink. Additional drinks are not included. Shuttle bus will be provided.

Special Events th

On Tuesday evening, October 4 , a guided city tour in English language will be offered. Interested participants should meet at the Acara Hotel at 8.30 pm. On Wednesday evening, October 5th , the mayor of Oldenburg will welcome participants at the “Ehemalige Exerzierhalle”, located downtown within easy walking distance of the Hotel Acara. We will provide guides for the walk from the hotel to the reception. The street address is: “Am Pferdemarkt” – in case one needs to find it on the map. At the reception catering with free snacks and drinks will be available. You also have the chance to see the exhibition “Mehr Licht! – Die Geschichte der Energieversorgung Ems-Weser-Elbe” devoted to the history of electrification of the Oldenburg region (English tour provided).

Registration & Information All participants must be registered, and badges must be worn to gain entrance into all sessions and events. The registration fee includes lunch and coffee breaks during the meeting from Wednesday until Friday. The Thursday evening dinner (without drinks), the special events, as well as bus tickets for inner city transportation are also included. Check-in to the meeting will be possible at Acara Hotel on Tuesday, Oct. 4th from 5 to 7 p.m. From Wednesday to Friday the registration and information desk will be located in room W2 1-143 close to the lecture hall with opening hours starting at 8 a.m.

Campus Plan

20:30 Guided City Tour

17:00 - 19:00 Welcome and Conference Check-in at Acara Hotel

Health Monitoring

F: Materials, Fatigue & Structural

E: Wake effects

D: Rotor Aerodynamics

C: Power Production & Fluctuations

& Turbulence

B: Gusts, Extreme Events

A: Wind Climate & Wind Field

TUE, 10/04/2005

W. Bierbooms J. Fernández Puga J.L. Palma E. Bibor

12:05 - LUNCH

(A) (A) (A) (A)

10:40 - COFFEE

9:00 - Opening (A) B. Lange (A) L. Bergdahl (A) S. Larsen (A) J. Tambke

P. Sorensen M. Greiner A. Rauh K. Kaiser J.J. Trujillo T. Weidinger

B. Hillmer W. Z. Shen H. Yang R. Bourguet

W. Geissler C. Sicot J. Rauch R. Langtry

18:30 - Conference Dinner on a boat on Lake Bad Zwischenahn

19:00 - Reception at "Ehemalige Exerzierhalle"

14:45 - Poster

14:25 - COFFEE

(D) (D) (D) (D)

11:40 - LUNCH

(D) (D) (D) (D)

10:15 - COFFEE

(C) (C) (C) (C) (C) (C)

17:00 - End of day#2

13:20 13:40 13:55 14:10

10:35 10:55 11:10 11:25

08:30 08:50 09:10 09:30 09:45 10:00

THU, 10/06/2005

17:40 - End of day#1

16:40 - Shotgun Session

(B) C. Wagner (B) F.G. Schmitt (B) T. Neumann (B) D. Schertzer (B) K. Leroux 15:25 - COFFEE 15:45 (B) J. Mann 16:05 (B) H. Kantz 16:25 (B) G.C. Larsen

13:50 14:10 14:30 14:50 15:10

11:00 11:20 11:35 11:50

09:30 09:50 10:10 10:25

Conference Check-in at Conference Venue

WED, 10/05/2005 J.N. Soerensen V.L. Okulov A. Wessel S. Ivanell S. Aubrun L. Fuchs

M. Vormwald M. Seidel F. Wilke I. Weich J. Reetz L. Batisti 12:05 - LUNCH

(F) (F) (F) (F) (F) (F)

10:05 - COFFEE

(E) (E) (E) (E) (E) (E)

Departure

13:40 (F) W. Hillger 14:00 (F) B.F. Sorensen 14:20 (F) L. Lohaus 14:35 (F) K. Mittendorf 14:50 (F) Bergers/Huhn 15:05 - Closing Remarks 15:30 - End of day#3

10:25 10:45 11:05 11:20 11:35 11:50

08:30 08:50 09:05 09:20 09:35 09:50

FRI, 10/07/2005

EUROMECH COLL. 464b WIND ENERGY - OCT. 04-07, 2005

EUROMECH COLL. 464b WIND ENERGY, 2005 • Oldenburg, October 04 – 07 •

Tuesday, October 04, 2005 17:00

Welcome and conference check-in at Acara Hotel

20:30

Guided city tour

Wednesday, October 05, 2005 08:00

Conference check-in at conference venue

09:00

Opening

A: Wind Climate & Wind Field Chair:

NN

09:30

Offshore wind power meteorology (Invited Talk) B. Lange ISET e.V., Kassel, Germany

page 13

Wave loads on wind power plants in deep and shallow water (Invited Talk) L. Bergdahl, C. Eskilsson, J. Trumars Chalmers University of Technology, Göteborg, Sweden

page 14

Profiles of mean wind and turbulence in the atmospheric boundary layer above the surface layer S. Larsen, S.E. Gryning, N.O. Jensen, H.E. Jørgensen, J. Mann Risø National Laboratories, Denmark

page 15

Wind speed profiles, convection and turbulent momentum fluxes in the marine boundary layer above the North Sea J. Tambke, J. A. T. Bye, L. Claveri, L. v. Bremen, C. Poppinga, B. Lange, J.-O. Wolff ForWind, University of Oldenburg, Germany

page 16

09:50

10:10

10:25

10:40

Coffee

Euromech Coll. 464b Wind Energy (2005)

11:00

11:20

11:35

11:50

12:05

2

Time domain comparison of simulated and measured wind turbine loads using constrained wind fields (Invited Talk) W. Bierbooms, D. Veldkamp Delft University of Technology, Netherlands

page 17

Fundamental aspects of fluid flow over complex terrain for wind energy applications J. Fern´ andez Puga, M. Fallen, F. Ebert University of Kaiserslautern, Germany

page 18

Models for computer simulation of wind flow over sparsely forested regions J.L. Palma, J. C. Lopes da Costa, F. A. Castro Universidade do Porto, Portugal

page 19

Power performance via nacelle anemometry on complex terrain E. Bibor, C. Masson Ecole de technologie supérieure de Montréal, Canada

page 20

Lunch

B: Gusts, Extreme Events & Turbulence Chair:

NN

13:50

Turbulence modelling and numerical simulation of turbulent flows (Invited Talk) C. Wagner German Aerospace Center (DLR) Göttingen, Germany page 21

14:10

Gusts in intermittent wind turbulence and the dynamics of their return times (Invited Talk) F.G. Schmitt CNRS, University of Lille, France

page 22

14:30

Design parameters/Load assumptions for offshore wind turbines in the German Bight on the basis of FINO1 measurements (Invited Talk) T. Neumann, S. Emeis, C. Illig DEWI, Wilhelmshaven, Germany page 23

14:50

Wind extremes and scales: multifractal insights and empirical evidence D. Schertzer, I. Tchiguirinskaia, S. Lovejoy, J.M. Veysseire 1CEREVE, ENPC, Marne-la-vallee Cedex, France.

page 24


15:10

Boundary layer influence on orographic-flow extreme events K. Leroux, O.S. Eiff Météo France, Toulouse, France

3

page 25

15:25

Coffee

15:45

Simulation of turbulence, gusts and wakes for load calculations (Invited Talk) J. Mann Risø National Laboratory, Roskilde, Denmark

page 26

Short time prediction of wind speeds from local measurements (Invited Talk) H. Kantz MPIKS, Dresden, Germany

page 27

The statistical distribution of turbulence driven velocity extremes in the atmospheric boundary layer G.C. Larsen, K.S. Hansen Risø National Laboratory, Roskilde, Denmark

page 28

16:05

16:25

Shotgun Session Chair:

NN

16:40

Poster presentation (2 min each)

17:40

End of scientific program

Evening Program 19:00

Reception at ’Ehemalige Exerzierhalle’


4

Thursday, October 06, 2005 C: Power Production & Fluctuations Chair:

NN

08:30

Wind farm power fluctuations (Invited Talk) P. Sorensen, J. Mann , U. S. Paulsen, A. Vesth Risø National Laboratory, Roskilde, Denmark

page 29

Network perspective of wind-power production (Invited Talk) M. Greiner Siemens AG, Munich, Germany

page 30

Phenomenological response theory to predict power output (Invited Talk) A. Rauh University of Oldenburg, Germany

page 31

Turbulence correction for power curves K. Kaiser, W. Langreder, H. Hohlen, J. Højstrup Wind Solutions, L¨ ubeck, Germany

page 32

Online modelling of wind farm power for performance surveillance and optimisation J.J. Trujillo, A. Wessel, H.-P. Waldl, B. Lange ForWind, University of Oldenburg, Germany

page 33

Uncertainty of available wind energy estimation based on long term meteorological measurements in Hungary ´ Kiss, A. Z. Gyöngyösi, K. Krassován, B. Papp T. Weidinger, A. E¨ otv¨ os Lor´ and University, Budapest, Hungary

page 34

08:50

09:10

09:30

09:45

10:00

10:15

Coffee

D: Rotor Aerodynamics Chair:

NN

10:35

Modelling of the transition locations on a 30% thick airfoil with surface roughness (Invited Talk) B. Hillmer, Y.S. Chol, A.P. Schaffarczyk University of Applied Sciences Kiel, Germany

page 35

Determination of angle of attack (AOA) for rotating blades W.Z. Shen, M.O.L. Hansen, J.N. Sørensen Technical University of Denmark, Lyngby, Denmark

page 36

10:55


11:10

11:25

5

Unsteady characteristics of flow around an airfoil at high angle of attacks and low Reynolds numbers H. Yang, H. Guo, Y. Zhou Hong Kong Polytechnic University, Hong Kong

page 37

Multi-criteria aerodynamic shape optimization of VAWT blades profile by viscous approach R. Bourguet, G. Martinat, G. Harran, M. Braza Institut de Mécanique des Fluides de Toulouse, France

page 38

11:40

Lunch

13:20

Helicopter aerodynamics with emphasis placed on dynamic stall (Invited Talk) W. Geissler German Aerospace Center (DLR), Göttingen, Germany

page 39

Rotation and turbulence effects on pressure distribution on a wind turbine blade – unsteady experimental approach C. Sicot, P. Devinant, S. Loyer, J. Hureau Ecole Polytechnique de l’Université d’Orléans, France

page 40

Simulation and evaluation of the air flow through wind turbine rotors by comparing 2D and 3D numerical simulations with focus on the hub area J. Rauch, T. Kr¨ amer, B. Heinzelmann, J. Twele, P.U. Thamsen Technische Universit¨ at Berlin, Germany

page 41

Predicting 2D and 3D wind turbine performance using a transition model for general CFD codes R. Langtry, J. Gola, S. V¨ olker, F. R. Menter ANSYS Germany GmbH, Otterfing, Germany

page 42

13:40

13:55

14:10

14:25

Coffee

14:45

Poster session (see pages 8-10)

17:00

End of scientific program

Evening Program 18:30

Departure to Conference dinner on boat on lake Bad Zwischenahn


6

Friday, October 07, 2005 E: Wake Effects Chair:

NN

08:30

Modeling of far wake behind wind turbine (Invited Talk) J.N. Soerensen, V. L. Okulov Technical University of Denmark, Lyngby, Denmark

page 43

Stability of tip vortex equilibrium in far wake behind Wind Turbine V.L. Okulov, J.N. Sørensen Institute of Thermophysics, Novosibirsk, Russia

page 44

Modelling turbulence intensities inside wind farms A. Wessel, J. Peinke, B. Lange ForWind, University of Oldenburg, Germany

page 45

Numerical computations of wind turbine wakes S. Ivanell, D. Henningson Royal Institute of Technology, Stockholm, Sweden

page 46

Modeling wind turbines wakes with a porosity concept S. Aubrun Laboratoire de Mécanique et d’Energétique, Orléans, France

page 47

Acoustic analogy applied on wind turbine noise generation and propagation L. Fuchs, D. Moroianu Lund Institute of Technology, Sweden

page 48

08:50

09:05

09:20

09:35

09:50

10:05

Coffee

F: Materials, Fatigue & Structural Health Monitoring Chair:

NN

10:25

Fatigue assessment of truss joints based on local approaches (Invited Talk) M. Vormwald, H. Th. Beier, J. Lange Technische Universit¨ at Darmstadt, Germany

page 49

Advances in offshore wind turbine technology (Invited Talk) M. Seidel, J. G¨ oßwein REpower Systems AG, Osnabr¨ uck, Germany

page 50

10:45


11:05

11:20

11:35

11:50

7

Benefits of fatigue assessment with local concepts F. Wilke, P. Schaumann Universit¨ at Hannover, Germany

page 51

Extension of life time of welded dynamic loaded structures I. Weich, T. Ummenhofer, T. Nitschke-Pagel Technische Universit¨ at Braunschweig, Germany

page 52

Structural health monitoring of WEC using the multiparameter eigenvalue problem J. Reetz, W.-J. Gerasch, R. Rolfes University of Hannover, Germany

page 53

Influence of wind turbine’s type and size on antiicing thermal power requirement L. Battisti, R. Fedrizzi, S. Dal Savio University of Trento, Italy

page 54

12:05

Lunch

13:40

Structure health monitoring – visions, possibilities and research tasks (Invited Talk) W. Hillger German Aerospace Center (DLR), Braunschweig, Germany

page 55

Performance and qualification of materials used for rotor blades (Invited Talk) B.F. Sorensen, P. Brondsted Risø National Laboratory, Roskilde, Denmark

page 56

High-cycle fatigue of “ultra-high-performance concrete” and “grouted joints” for offshore wind energy turbines L. Lohaus, S. Anders University of Hannover, Germany

page 57

A modular concept for integrated modeling of offshore WEC applied to wave-structure coupling K. Mittendorf, M. Kohlmeier, A. Habbar, W. Zielke University of Hannover, Germany

page 58

14:00

14:20

14:35

14:50

Solutions of details regarding fatigue and the use of high-strength steels for towers of offshore wind energy converters J. Bergers, H. Huhn, R. Puthli, M. Veselcic University of Karlsruhe, Germany page 59

15:05

Closing

15:30

End of colloquium


8

Poster Session A1

A2

A3

A4

B1

B2

B3

B4

B5

On the atmospheric flow modelling over complex relief I. Sl´ adek, K. Kozel, Z. Jaˇ nour Czech Technical University, Prague, Czechia

page 63

Pollutant dispersion in flow around bluff bodies arrangement E. Moryn-Kucharczyk, R. Gnatowska Czestochowa University of Technology, Poland

page 64

The improved mesoscale turbulence model for wind climate and pollutant dispersion in cites A.F. Kurbatskiy, L. I. Kurbatskaya Russian Academy of Sciences, Siberian Branch and Novosibirsk State University, Russia

page 65

Comparison of logarithmic wind profiles and power law wind profiles and their applicability for offshore wind profiles S. Emeis Forschungszentrum Karlsruhe GmbH, Germany

page 66

Superposition model for atmospheric turbulence St. Barth, F. B¨ ottcher, J. Peinke ForWind, University of Oldenburg, Germany

page 67

Extreme events under low-frequency wind speed variability and wind energy generation A.A. Cˆ arsteanu, J.J. Castro Cinvestav, Mexico City, Mexico

page 68

Stochastic modelling and prognosis of turbulent wind time series J. Cleve, M. Greiner Siemens AG, M¨ unchen, Germany

page 69

Quantitative Reconstruction of Drift and Diffusion Functions from Time Series Data D. Kleinhans, R. Friedrich University of M¨ unster, Germany

page 70

Scaling turbulent atmospheric stratification: a turbulence/wave wind model S. Lovejoy, D. Schertzer McGill University, Montreal, Que., Canada

page 71


C1

C2

C3

C4

D1

D2

D3

D4

E1

E2

E3

9

Characterization of the Power Curve for Wind Turbine by Stochastic Modelling E. Anahua, J. Peinke, S. Barth ForWind, University of Oldenburg, Germany

page 72

Handling systems driven by different noise sources F. B¨ ottcher, J. Peinke, D. Kleinhans, R. Friedrich ForWind, University of Oldenburg, Germany

page 73

Experimental researches of characteristics of windrotor models with vertical axis of rotation F. Kayan, S.A. Dovgy, V.A. Kochin Institute of Hydromechanics NASU, Kyiv-57, Ukraine

page 74

Methodical failure detection in grid connected wind parks D. Schulz, K. Knorr, R. Hanitsch Hochschule Bremerhaven, fk-wind: Forschungs- und Koordinierungsstelle Windenergie, Germany

page 75

Performance of the Risø-B1 airfoil family for wind turbines C. Bak, M. Gaunaa, I. Antoniou Risø National Laboratory, Roskilde, Denmark

page 76

Dynamic lift of airfoils T. Bohlen, R. Gr¨ uneberger, J. Peinke ForWind, University of Oldenburg, Germany

page 77

Aerodynamic behaviour of a new type of slow running vertical axis wind turbine J.L. Menet Université de Valenciennes, France

page 78

Simulating dynamic stall using spectral CFD-code B. Stoevesandt, A. Shishkin, C. Wagner, J. Peinke ForWind – Center for Wind Energy Research, Oldenburg, Germany

page 79

Comparing linear and non-linear wind flow models D. Cabez´ on, A. Iniesta, I. Mart´ı National Renewable Energy Centre (CENER), Sarriguren, Navarra, Spain

page 80

Large eddy simulation of an aeroturbine wake A. Jiménez Universidad Politécnica de Madrid, Spain

page 81

Infrasound in wind energy G. Sokol Dniepropetrovsky National University, Dniepropetrovsk, Ukraine

page 82


E4

F1

F2

F3

F4

F5

F6

F7

10

Prediction of wind turbine noise generation and propagation based on an acoustic analogy D. Moroianu, L. Fuchs Lund Institute of Technology, Sweden

page 83

On the influence of low-level jets on energy production and loading of wind turbines N. Cosack, S. Emeis, M. K¨ uhn University of Stuttgart, Germany

page 84

A new damage approach for concrete towers of wind energy converters subjected to multi-stage fatigue loading J. G¨ ohlmann, J. Gr¨ unberg University of Hannover, Germany

page 85

Reliability of wind turbines experiences of 15 years with 1,500 WTs B. Hahn, M. Durstewitz, K. Rohrig ISET e.V., Kassel, Germany

page 86

Systematic modelling of wind turbine dynamics M. H¨ anler, U. Ritschel, J. Kirchner, I. Warnke Windrad Engineering GmbH, Zweedorf, Germany

page 87

Design, fabrication and testing of the wind turbine rotor blades from composite laminated materials B. Raˇsuo University of Belgrade, Serbia and Montenegro

page 88

Multibody system simulation of offshore wind turbines B. Schlecht, T. Schulze, Th. H¨ ahnel, Th. Rosenl¨ ocher Technische Universit¨ at Dresden, Germany

page 89

Integrated monitoring system for offshore wind energy plants R.G. Rohrmann, W. R¨ ucker, S. Said, S. St¨ ohr Federal Institute for Materials Research and Testing (BAM), Berlin, Germany

page 90

TALKS

Session A – Talk 1

13

Offshore Wind Power Meteorology B. Langea a ISET e.V., K¨ onigstor 59, 34119 Kassel, Germany

The wind over the sea is the energy source, which is tapped by offshore wind farm installations. At the same time it is the main source of loads on the wind turbines. Comprehensive knowledge about the wind conditions at offshore sites is therefore crucial for the successful development of offshore wind power. On the other hand, in the past there has been little need to investigate the wind at locations and heights relevant for wind power utilization, i.e. over coastal waters and in heights of 30-200 meters. The knowledge about wind conditions and meteorological effects at these locations is therefore limited. From the point of view of offshore wind power utilization, there are fundamental differences between the wind conditions over land and offshore. Three reasons are most important for these: • the surface roughness of the sea is very low, but also dependent on the wave field • the thermal properties of water are very different to those of land, leading to a different thermal stratification • both the surface roughness and the surface temperature change abruptly at the coastline, which leads to large transition effects for wind blowing from land to the sea For weather prediction purposes, measurements over the sea are made by buoys, lightships, ships, lighthouses, etc. These measure the wind usually at low heights and are not particularly well suited for precise wind speed measurements. Meteorological research focuses mainly on air-sea interaction processes and highly accurate wind speed measurements are often made at low heights and for limited time periods. In recent years, purpose built measurements for offshore wind power utilizations have been erected at a number of locations. The measurements improved the knowledge about the wind conditions relevant for offshore wind farm installations. It has been found that the wave field dependence of the sea surface roughness is of minor importance for resource calculations, but may be crucial for extreme wind conditions. Thermal effects, on the other hand, can have a large influence on the wind conditions and power production, because they can modify the wind profile in the coastal zone significantly when wind blows from land out over the sea. Numerical meteorological models are used to investigate the spatial behaviour of these effects. Their advantage is to provide data for a large area in space. Fed by the output of global models, mesoscale meteorological models are used to determine, investigate and forecast the wind over the sea. While there are still a number of open challenges in modelling wind at offshore locations with these models, the results are very promising. Comparisons between the commonly used wind resource calculation program WAsP and mesoscale models show important differences, which reveal the importance of offshore effects. But resource assessment for planned offshore wind farms is only one of the applications of offshore wind power meteorology. Another important application is the prediction of the wind power output, which is needed for the grid integration of the electricity produced. Because of the large capacity of the planned wind farms a high prediction accuracy is needed. First investigations indicate that the prediction accuracy of numerical weather prediction models is indeed better offshore than for stations on land.

Bernhard Lange – [email protected]


14

Wave loads on wind power plants in deep and shallow water Lars Bergdahl a , Claes Eskilsson

a

and Jenny Trumars

a

a Water Environment Transport, Department of Civil and Environmental Engineering Chalmers University of Technology

Due to the dynamic behaviour of wind turbines and their foundations, load effects may be quite sensitive to the precise modelling of the wave loading. As the offshore wind turbine technology progresses, larger and larger turbines will be placed in deeper and deeper waters, causing the resonant frequency of the first eigen mode of a traditional bottom- fixed support structure to be typically in the range of 0.25 Hz to 0.35 Hz. The deeper the water and the larger the turbine, the lower the resonance frequency will be. Both in deep and shallow waters spectra from steep waves exhibit a second order peak at a frequency twice the modal peak frequency. This is due to the fact that in steep waves the linear assumption that the wave elevation is symmetrical around the mean water level no longer holds: Wave crests are peaked and wave troughs are shallow. In shallow water the effect is enhanced. As an example, measured wave spectra from the offshore wind farm Bockstigen with modal peak frequencies of 0.12 to 0.15 Hz show second peaks at 0.25 to 0.3 Hz. Such peaks cannot be modelled with a linear wave model, and wave models taking non-linearities into account have to be used. The pressure and velocity attenuation with depth from the from the double frequency components follows the attenuation of the first order ”parent”waves and is thus much smaller than anticipated. The structural response of the turbines may very well be dynamically amplified and more sensitive to such loads at ”unexpected”high frequency. In the paper, it is sketched how second order, irregular, non-linear waves can be realised from spectra in deep and shallow water. The problem of unintended addition of variance in the realisation process, especially for shallow water, is addressed. It is also discussed how the transmission of waves from deep water into shallow water can be modelled by phase-average models and phase-resolving models. The phase-average models calculates the change of the wave spectra as the waves progress over a bottom topography with respect to wind, wave interaction in the frequency domain and dissipation due to bottom friction and breaking waves. A phase resolving Boussinesq type model, on the other hand, simulates the progress of an actual wave train keeping track of the phases between long and short waves as well as locked and free waves, but is more time-consuming and restricted by bottom depth and wave breaking. Examples of phase average and Boussinesq-type modelling are given. For the latter an animation of a wave passing a pile is shown. Also examples of time series of irregular linear and non-linear waves are given and the effect on the fatigue load is presented for an offshore wind turbine with a slender support structure. The forces on the structure are either calculated using Morison’s equation, integrating along the structure and lumping the loads in nodes for the structural calculations or by simulating the pressure distribution in the time domain.

Correspondence: Lars Bergdahl – [email protected] – fax: +46 31 772 2128


15

Profiles of mean wind and turbulence in the atmospheric boundary layer above the surface layer S.E.Larsena S.E. Gryninga , N.O.Jensena , H.E.Jørgensena and J. Manna a Department Of Wind Energy, Risø, DK4000, Roskilde, Denmark

The growing height of wind turbines has increased the need for knowledge about the mean wind and turbulence field at these heights of the atmospheric boundary layer (above, say 100 m). We present a review of the existing knowledge, and compare with one year of data from the recently established site for testing of larger wind turbines at Høvsøre at the North Sea coast of Denmark. The site is equipped with a 110-meter climate mast with six levels of for profiles and turbulence measurements. Additionally two 160-meter masts for warning lights have offered possibility for measurements at that height as well. The position of the site at a coastline allows us to evaluate flow models for both homogeneous terrain, and for a water-land surface change. The overall conditions in the lower atmosphere as well as on the up-stream and down-stream surfaces are integrated into the data analysis from the NCEP data, together with other databases. Most of the micrometeorological models used in wind power studies are based on homogeneous surface layer description, and mostly for thermally neutral conditions, with added models to include the effects in the surface layer of terrain changes. Our studies show that it becomes increasingly difficult to justify the thermally neutral homogeneous description for increasing height, because the importance of heterogeneity and thermal structure increases with height. We illustrate this behaviour from our data. Also we have found that even for thermally neutral conditions the wind profile was less predictable than thought, characteristically showing both situation with enhanced shear and reduced shear relatively to the Logarithmic profile. Although models for both are formulated the predictability is still not satisfactory. These aspects are enhanced when stability is included .The situations are systematised, with respect to both mean flow and turbulence, and the importance for power production is discussed. Additionally to the ”homogeneous situationsthe site has allowed us to compare the behaviour of both mean profiles and turbulence with height at the positions of the different masts, with fetch varying with wind direction. Hereby, extension of the current ”change of surface modelsfor the wind and power production has been possible. We illustrate these features from the data set that allow us as well to discuss their quantitative importance at the Høvsøre site, and thereby, by conjecture, also at other sites.

Correspondence: S.E.Larsen – [email protected] – fax: +45 4677 5970


16

Wind Speed Profiles, Convection and Turbulent Momentum Fluxes in the Marine Boundary Layer above the North Sea J. Tambkea , J. A. T. Byeb , L. Claveria , L. v. Bremena , C. Poppingaa , B. Langec , J.-O. Wolffd a ForWind Center, Institute of Physics, University of Oldenburg, Germany b School of Earth Sciences, The University of Melbourne, Victoria, Australia c ISET e.V., University of Kassel, Germany d Physical Oceanography (Theory), ICBM, University of Oldenburg, Germany

For wind energy resource assessments, 48 hours wind speed forecasts as well as for calculations of loads and wakes, the vertical wind profile above the sea has to be modelled with high accuracy [1]. We analysed profiles of marine wind speeds and momentum fluxes that were measured at the two new met masts Horns Reef (62m high) and FiNO1 (100m high) in the North Sea. In this work we present a detailed description of the observed profile characteristics depending on fetch and thermal stratification. In many situations, the wind shear is significantly higher than expected in commonly used boundary layer models. This discrepancy is shown for different mesoscale model schemes, especially for the model of the German Weather Service DWD. Consequently the standard approaches to calculate wind speeds in higher altitudes from observed lower level values lead to considerable underestimations.

Fig. 1: Scheme for the vertical wind profile used in the ICWP model.

For a better understanding of this newly observed effect, we developed an analytic model of the wind velocity profile in the marine boundary layer (figure 1) that considers the non-linear wind-wave interaction. In particular, the flux of momentum through the Ekman layers of atmosphere and sea is described in each fluid by a common wave boundary layer specified by the inertial coupling relations [2], and two outer constant viscosity layers [3]. The three layers are coupled by matching velocity, shear stress and eddy viscosity at a specific height, which can be related to the peak wavenumber of the wave spectrum. We extend the basic model in [1] to stable and convective thermal conditions of the marine boundary layer and present the inertially coupled wind profiles (ICWP) for this system. The results are expressible in terms of a single parameter, which is predicted for a prescribed wind at an arbitrary height. Finally, we compare the ICWP and the usually applied Monin-Obukhov log-linear profiles with the measurements from Horns Reef and FiNO1. The good agreement between ICWP and observations supports the wave boundary layer approach and our assumption that the Ekman layer begins at 15 to 45 m height above the sea surface.

Literatur [1] J. Tambke, M. Lange, U. Focken, J.-O. Wolff and J.A.T. Bye (2005): Forecasting Offshore Wind Speeds above the North Sea, Wind Energy, vol. 8, pp. 3-16. [2] J.A.T. Bye (1995): Inertial coupling of fluids with large density contrast, Phys. Lett. A, vol. 202, pp. 222-224. [3] J.A.T. Bye (2002): Inertially coupled Ekman layers, Dyn. Atmos. Oceans, vol. 35, pp. 27-39. Correspondence: Jens Tambke – [email protected] – phone: +49 (0) 441-36116736


17

Time domain comparison of simulated and measured wind turbine loads using constrained wind fields Wim Bierbooms

a

and Dick Veldkamp

b

a Wind Energy Research Group, Delft University of Technology, Kluyverweg 1, 2629 HS Delft, The Netherlands b Vestas Wind Energy Systems, Alsvej 21, 8900 Randers, Denmark

Objectives Comparison between wind turbine load measurements and simulations is complicated by the uncertainty about the wind field experienced by the rotor. Usually the wind speed is measured at just one location (hub height) which makes direct comparison between the measured and simulation load time histories impossible. Instead the power spectra of the loads and the equivalent fatigue load ranges are compared. The objective of the present work is to demonstrate a method which makes direct comparison in time domain possible. Scientific innovation and relevance At Delft University a new method (so-called constrained stochastic wind field simulation), has been developed in order to achieve a realistic and verified description of extreme gusts based on the stochastic properties of wind. Until now this method has been used to generate gusts with maximum amplitude to assess the ultimate loading of stall turbines, and to generate gusts with an extreme rise time to assess the ultimate loading of pitch regulated wind turbines. The method of constrained stochastic simulation is general, and it can also be used to generate stochastic 3D wind fields which exactly match measured time series at one (or more) point(s). Description of the work A computer implementation of the method of constrained stochastic simulation is made in order to generate stochastic wind fields (in the rotor plane, all 3 velocity components) in such a way that the measured wind speed time series (at more locations) are exactly reproduced. In the method the correlation of the turbulence components at different locations is taken into account in a consistent way. The generated wind fields may be considered to be wind fields which could have occurred during the measuring period; the actual wind field can never be reproduced in case of just a few measurement locations, since turbulence is a stochastic process. The influence of the number of anemometers as well as their position will be investigated for a common pitch regulated variable speed turbine. Results and conclusions The scatter in the load time series resulting from different stochastic wind field simulations (for given mean wind speed, turbulence intensity and (co) spectrum) will be compared to the reduced scatter in case of constrained stochastic simulation for several configurations of anemometer(s). Furthermore the method will be applied to compare measured and simulated loads on a Vestas 80 m 2.75 MW turbine at Tjæreborg.

Correspondence: Wim Bierbooms – [email protected] – fax: +31 15 278 5347


18

Fundamental aspects of fluid flow over complex terrain for wind energy applications J. Fernández Pugaa M. Fallenb and F. Eberta a Institute for Mechanical Process Engineering, University of Kaiserslautern, b Institute for Fluid Flow and Positive Displacement Pumps, University of Kaiserslautern, Erwin-Schr¨ odinger-Straße, D-67663 Kaiserslautern, Germany

It is well known that the analysis of wind potential of complex inland sites for wind energy purposes is very difficult because of local topographic influences such as hills and vegetation. A reliable forecast can be performed using measurements and computations. Therefore, the flow over steep hills has been investigated experimentally and theoretically at the University of Kaiserslautern. Sinusoidally-shaped hills have been chosen for the investigation which follow the parametric shape equations of the RUSHIL experiments. The velocity and turbulence fields over the hill have been measured with the Laser Doppler Anemometry (LDA) technique in the wind tunnel of the University (see figure 1 left). The measurements have been carried out using a three-component system employing seeding particles with a size of 2µm. These particles are made by spraying a solution of glycerine and water. The atmospheric boundary layer had to be approximated. Since it is a short wind tunnel a sufficiently thick boundary layer must be generated. To this end several devices thickening the boundary layer artificially have been investigated. Designs based on cylindrical rod generators and spires have been analysed.

Fig. 1: Measured and computed profiles of the longitudinal (left) and the vertical (right) velocity in the lee of the 2-d hill The experimental results have been compared with the results of the numerical simulations. Thereby, calculations have been conducted using different turbulence models. The results show clearly that the velocity field can be reproduced satisfactorily (see figure 1), whereas the results for the turbulence parameters differ remarkably from experimental values. This can be observed in the recirculation zone in the lee of the hill where the flow is mostly affected by the hill. However, depending on whether the detachment point or the reattachment point is considered some models perform better than others. The hill size has been varied in order to analyse the influence of the steepness on the flow field. Additionally, a comparison between a two-dimensional and a three-dimensional hill has been carried out. Large eddy simulations show clearly the unsteadiness and the vortex structures of the flow over the hills (see figure 2).

Fig. 2: Isosurfaces of constant vorticity; left: 2d hill, right: 3d hill.

Correspondence: José Fern´ andez Puga – [email protected] – fax: +49 (0) 631-205-3055


19

Models for computer simulation of wind flow over sparsely forested regions J. C. Lopes da Costa, F. A. Castro, J. M. L. M. Palma CEsA - Research Centre for Wind Energy and Atmospheric Flows, Faculdade de Engenharia da Universidade do Porto, Rua Roberto Frias s/n , 4200-465 Porto, Portugal

The wind flow near or within forest patches is strongly affected by the presence of the trees. Because of the flow complexity and wind power reduction, these regions were not considered as candidate for wind parks until recently. However, because of the shortage of clear sites, this situation was reverted, adding new motivation to the study of the wind flow over canopies and forested patches. In the present work, various canopy models [4, 2, 5] are studied over idealised and complex terrains, using a CFD code, based on a k − ε turbulence model [1]. Computer simulations of the flow over canopy on a flat plate and sinusoidal hills configurations are compared with wind tunnel measurements [3]. The emphasis is on the prediction of the flow as it enters and exits the canopy regions. Wind turbines are often located in clearings and these issues are the most important and difficult to master. The study concludes with computer simulations of real forested regions and comparison with field measurements at different distances above the ground level.

Fig. 1: Wind flow (longitudinal velocity) over a canopy on a flat plate, based the wind tunnel measurements [3]

Literatur [1] F. A. Castro, J. M. L. M. Palma, and A. Silva Lopes. Simulation of the Askervein flow. Part 1: Reynolds averaged Navier-Stokes equations (k- turbulence model). Boundary-Layer Meteorology, 107:501–530, 2003. [2] J. Liu, J. M. Chen, T. A. Black, and M. D. Novak. E - ε modelling of turbulent air flow downwind of a model forest edge. Boundary-Layer Meteorology, 77:21–44, 1996. [3] B. Ruck and E. Adams. Fluid mechanical aspects in the pollutants transport to coniferous trees. BoundaryLayer Meteorology, 56:163–195, 1991. [4] C. Sanz. A note on k −ε modelling of vegetation canopy air-flows. Boundary-Layer Meteorology, 108:191–197, 2003. [5] U. Svensson and K. H¨ aggkvist. A two-equation turbulence model for canopy flows. Journal of Wind Engineering and Industrial Aerodynamics, 35:201–211, 1990. Correspondence: J. M. L. M. Palma – [email protected] – fax: +351 22 5082154


20

Power performance via nacelle anemometry on complex terrain E. Bibora , C. Massona a Mechanical Engineering Department, Ecole de technologie superieure, 1100 Notre-Dame Ouest, Montreal, Canada

Performance evaluation of wind turbines (WT) to common standards has been carried out for more than 10 years[1]. Standards regarding performance tests have been proposed by the International Electrotechnical Commission (eg. IEC 61400-12). However, there are still areas of limited knowledge. For example, the performance testing of a park composed of a large number of turbines reveals the difficulties related to nacelle anemometry. The basis of this technique is to establish a relationship between nacelle wind speed (Vnac ) and free-stream wind speed (V∞ ) for the reference WT. The assumption that this correlation can be used for the other WTs of the farm is reasonable over a flat terrain [1]. The purpose of this study is to investigate the precision of this technique when applied to complex terrain. More specifically, the consequences on power performance testing and on the projected annual energy production will be quantified using both experimental and numerical approaches. The experimental installation used for this study is located on a very complex site, in Riviere-au-Renard, in Eastern Canada. First, a single general correlation between Vnac ⇔ V∞ has been established using the measurements from all valid sectors. The influence of the terrain was then quantified by constructing specific correlations for each individual 10o sector. Significant differences are noted between the specific-sector and general correlations, as shown on the left figure.

In the range 4 < Vnac < 16m/s, errors in the V∞ estimates associated with the use of the general correlation, instead of the specific-sector correlation, can reach 5.6%. The consequences of such an error on the free-stream wind speed estimate can be significant on the measured power curve, as illustrated in the centre figure. Furthermore, numerous control parameters, such as the cut-in and cut-out wind speed, are directly related to the Vnac ⇔ V∞ correlations. These aspects also can have important impacts on the annual energy production. A numerical model, based on the commercial code F LU EN T , has been developed in order to have a better understanding of the nacelle anemometry. This model [2] takes into account the wind acceleration (or deceleration) produced by the blades, the nacelle and the anemometer mounting. The rotor is represented by an actuator-disk with the forces calculated by the blade element method. The model has been validated with the experimental correlations obtained previously. Good agreements are obtained only when the mounting system of the anemometer is taken in account (see the right figure).

Literatur [1] R. Hunter et al. (2001): European Wind Turbine Testing Procedure Developments, Riso-R-1209. ˆ C. Masson (2005): On the Rotor Effects upon Nacelle Anemometry for Wind Turbine, Wind [2] A. SmaOli, Engineering, in press. Correspondence: Christian Masson – [email protected] – fax: 1 (514) 396-8530

Session B – Talk 1

21

Turbulence modelling and numerical simulation of turbulent flows C. Wagnera a Institute for Aerodynamics and Flow Technology, DLR G¨ ottingen, Bunsenstr. 10, G¨ ottingen, Germany

Principally, there are three different approaches to predict turbulent flows numerically. Due to the dramatically increased efficiency of the super computers and new numerical methods it is possible to solve the time-dependent Navier-Stokes equations without any turbulence model in a Direct Numerical Simulation (DNS). However, the huge computational resources, which are required to conduct a DNS restricts their application to turbulent flows at lower Reynolds numbers with in most cases just an academic objective. The obtained turbulent flows though are characterized by random three-dimensional fluctuations with a continuous spectrum of length scales ranging down to flow structures which dissipate the excessive energy, the Kolmogorov scales. Turbulent flow calculation with a more applied objective were and are still performed solving the Reynolds averaged Navier-Stokes equations (RANS). In many cases the obtained solution is stationary and, depending on the number of homogeneous directions involved, one or two dimensional. The statistical approach is associated with the highest loss of information and with a closure problem which is not satisfactorily solved. Spectral information is completely lost, since any statistical quantity is an average over all turbulent scales. The obtained flow field describes the mean flow, which is enough for many applied problems, while the turbulent information is described with the Reynolds stress tensor. This tensor has to be modeled with empirical or semi-empirical models. There are a huge number of different turbulence models, because so far there is no generally valid statistical turbulence model. In the last decade the interest in time-dependent, three-dimensional analysis of turbulent flows increased, since many technical problems are associated with large scale motions. To obtain these time-dependent flow structures researchers more and more conducted instationary RANS simulation on three dimensional meshes with increasing number of grid points. Doing this they pushed the RANS technique towards the third approach in numerical turbulence simulation, the Large-Eddy Simulation (LES) technique. This method basically resembles a compromise between RANS and DNS since it allows to predict the dynamics of the large turbulent scales while the effect of the fine scales are modeled with a subgrid-scale model. This will be demonstrated at the colloqium presenting results produced in DNS, RANS and LES of turbulent flows in simple configurations. Further, the popular (k, ω)-RANS turbulence model and the some common subgrid scale models will be discussed.

Correspondence: Claus Wagner – [email protected]


22

Gusts in intermittent wind turbulence and the dynamics of their return times F.G. Schmitt

a

a CNRS, Wimereux Marine Station, University of Lille 1, UMR ELICO, 28 av Foch, 62930 Wimereux, France

Regions interesting for the installation of windmills possess large mean wind velocity, but this is also associated to large Reynolds numbers and huge velocity fluctuations. Such fully developed turbulence is generally characterized by intermittent fluctuations on a wide range of scales (see [1, 2]). These fluctuations may be classically quantified in the inertial range, using the general framework of scaling structure functions. On the other hand, less efforts have been devoted to the return times of gusts – large fluctuations |V (t + τ ) − V (t)| for a small time lag τ (for an hydrological example see [1]). Using several atmospheric wind turbulence databases, we consider the scale (τ ) dependence of the number of gust events |V (t + τ ) − V (t)| for a given time series. We further consider return times of gust intensities larger than a given threshold: we show that successive return times of gust events are generally independent, and possess pdfs close to Lévy-stable laws. This means that the mean return time of gust events diverge with the number of realizations. We quantify this divergence and its limits, and discuss its practical implications. As another application of the same fully developed turbulent framework, we consider also the statistics of power fluctuations, focusing specifically on larger fluctuations.

Literatur [1] F. Schmitt, D. Schertzer, S. Lovejoy, and Y. Brunet (1994): Empirical study of multifractal phase transitions in atmospheric turbulence, Nonlinear Processes in Geophysics, 1, 95-104 [2] U. Frisch (1995): Turbulence, the legacy of A. N. Kolmogorov, Cambridge University Press. [3] F. Schmitt and C. Nicolis (2002): Scaling of return times for a high resolution rainfall time series, Fractals, 10, 285-290.

Correspondence: Francois G. Schmitt


23

Offshore Wind Design Parameters – OWID T. Neumanna , S. Emeisb and C. Illigc a DEWI German Wind Energy Institute, Ebertstr. 96, Wilhelmshaven, Germany b Institute for Meteorology and Climate Research, FZ Karlsruhe (IMK-IfU), Garmisch-Partenkirchen, Germany c DEWI-OCC Offshore and Certification Centre GmbH, Am Seedeich 9, Cuxhaven, Germany

In this contribution we report about the research project ”OWID”, where the project partners DEWI, DEWIOCC and the IMK-IFU are going to evaluate the design conditions for offshore wind turbines and their components on the basis of the FINO1–data*. In co-operation with the manufacturers of offshore wind turbines, certification companies and universities an adaptation for the existing guidelines shall be worked out. The project is supported by the Federal Ministry for the Environment, Nature Conservation and Nuclear Safety and co-financed by the manufacturers involved**. Offshore wind energy can play an important role to support the aim of producing 20% of the total electricity demand by renewable energies by 2020. Most of the projected German areas for the offshore use however lie far away from the coast in water depths up to 40m. For this situation no real experience is given. Even the Danish North Sea wind farm “Horns Rev” erected in 2002 cannot be compared directly as it has a much shorter distance to the coast line of about 14-20km. The FINO1 platform which is installed in 2003 about 45km before the island Borkum is equipped with a 100m met mast and records the long term meteorological and oceanographic conditions in the North Sea. It shall be used to reduce the incomplete knowledge when adapting wind turbines to the maritime conditions. We start with a thoroughly evaluation of the acquired FINO1-data with the focus on the mechanical loads that a future offshore wind turbine is exposed to. In addition to the measured undisturbed wind field, the disturbed wind within a wind farm will be simulated by CFD calculations. Both undisturbed and disturbed wind fields are used to calculate the loads on an idealised wind turbine in order to evaluate the effect on lay out and life time. The project shall result in proposals to improve the applicable guidelines to minimise risks in the planning, building and operating phase of offshore turbines and to set up a reliable basis for financing and insurance of the scheduled projects in the German Bight. * FINO: Research Platform in north and baltic sea, financed by the Federal Ministry for the Environment, Nature Conservation and Nuclear Safety ** Ermittlung designrelevanter Belastungsparameter f¨ ur WEA in der Deutschen Bucht auf Basis der FINOMessdaten (OWID), BMU-Project 0329961A

Literatur [1] Launder, B.E., Spalding, D.B.: Applied Mechanics and Engineering, 3, pp. 269-289, 1974. [2] Freedman, A.F.R., Jacobson, B.M.Z.: Journal of Boundary Layer Meteorology, 102, pp. 117-138, 2002. [3] Sten Frandsen, Risø Report R-1188, Risø National Laboratory, DK-4000 Roskilde, Denmark.

Correspondence: Your Name – [email protected] – fax: +49 (0) 123 456-7891


24

Wind extremes and scales: multifractal insights and empirical evidence D. Schertzerab , I. Tchiguirinskaiaa , S. Lovejoyc , J.M. Veysseired a CEREVE, ENPC, 6-8, ave. Blaise Pascal, Cité Descartes 77455 Marne-la-vallee Cedex, France b Météo-France, Paris, 1 Quai Branly, 75007 Paris. France c Physics, McGill University, 3600 University st., Montreal, Que., Canada d Direction de la Climatologie, Météo-France, 42 Av. G. Coriolis, 31057 Toulouse Cedex, France

An accurate assessment of wind extremes at various space-time scales (e.g. gusts, tempests, etc.) is of prime importance for a safe and efficient wind energy management. This is particularly true for turbine design and operation, as well as for wind potential estimates and wind farm implementation. We will first show that a theoretical consequence of the multifractal behaviour of the wind field over a wide range of space-time scales is that the probability tail is a power-law and therefore much ”fatterthan usually assumed. In this framework, extremes are much more frequent than anticipated with the help of rather thin tail probabilities. The critical and practical importance of the power law exponent qD of the probability tail could be understood by the facts that the probability of having an extreme 10 times higher decreases only by the factor 10qD and that all statistical moments of order q ≥ qD are divergent (i.e. the theoretical moments are infinite, the empirical ones depend on the sample size). Such a behavior have been often called ”self-organized criticality”. If the wind correlations had only short ranges, then the wind maxima distribution would be a Frechet law (GEV2, in the classical theory of Generalized Extreme Value) instead of the classical Gumbel law (GEV1), and the Frechet law tails will have the same power law exponent qD . We will point out generalizations with long range correlations in a multifractal framework. We will critically examine quantitative empirical evidence of this extreme behavior for the wind field at different scales and in various meteorological and climatological conditions, e.g. from mid-latitude to tropical conditions, including the two tempests over Europe in 1999, cyclones over China sea and some long time series.

Correspondence: Shaun Lovejoy – [email protected] – fax: 1-514 398-8434


25

Boundary layer influence on orographic-flow extreme events K. Lerouxa,b , O. S. Eiffa,b a Institut de Mécanique des Fluides de Toulouse, allée du Professeur Camille Soula, 31400 Toulouse, France b Météo-France, Centre National de Recherches Météorologiques, 42 avenue G. Coriolis, 31057 Toulouse Cedex, France

The effects of stratified flow over mountains need a particular attention due to their extreme events nature. For low Froude-numbers flows, strong downslope winds and lee wave rotors appear, and the wave amplitude is strong enough for the wave to overturn and break. A good knowledge of such orographic flow is necessary to forecast aeronautical hazards and to optimize the implantation of wind-energy turbines. Indeed, a 4 m/s upstream flow can reach a 10 m/s lee side speed on a 600 m mountain in such conditions, but only on a lee side portion after which it reverses, generating a “rotor”. A lack of in situ measurements due to the sporadic nature of this phenomenon leads us to pay a detailed attention to the realism of our models boundary conditions. Some boundary conditions are investigated via a series of experimental studies in hydraulic towing-tanks of different dimensions. The main focus of our study is the boundary layer effect on downslope windstorms and wave breaking dynamics. Previous experiments in our laboratory have shown that in addition to wave-breaking, trapped lee-waves and a rotoräppear even under uniform flow and stratification conditions [1]. Only when the no-slip boundary condition on the obstacle was accounted for in our previous numerical simulations, were these additional phenomena well reproduced, taken into account that the no-slip boundary condition on the obstacle also modifies the severe downslope windstorm velocities as well as the wave-breaking dynamics [2]. Doyle and Durran [3] confirmed via numerical simulations that a slip condition inhibits rotor formation for non-uniform flows. Rotor formation was shown to be due to wave-induced boundary-layer separation. Here we investigate via Particle Image Velocimetry (PIV) the effect of including no-slip conditions downstream of the obstacles, in addition to no-slip conditions on the obstacle and for different Reynolds numbers flows. The experiments, with and without plates attached to the trailing edge of the obstacle, show that the rotoränd the trapped lee-wave formation occur independently of the boundary condition downstream of the obstacle, in contrast to the boundary condition on the obstacle itself. The downstream boundary condition only slighly influences the trapped lee-wave amplitude and wavelength but not the lee-side winds nor the first rotor which extends from the obstacle beyond the trailing edge. Even for uniform conditions, the rotoränd trapped lee-wave formation is also shown to occur independently of the presence of wave-breaking. Here, we show that for very low Reynolds number, the wave breaking event is very sensitive to vertical Froude number for quasi two-dimensional ridges freed from side effects, but depending on obstacle aspect ratio. This highlights the on-obstacle-developed boundary layer effect. The bigger the boundary layer is, the more three-dimensionally dispersive the lee waves are. A comparison with high Reynolds number experiments confirms the strong obstacle-boundary-layer influence on the wave overturning, so its interaction on the wave field.

Literatur [1] O. S. Eiff and P. Bonneton (2000): Lee-wave breaking over obstacles in stratified flow, Phys. Fluid, vol. 12, pp. 1073-1086. [2] F. Gheusi, J. Stein and O. S. Eiff (2000): A numerical study of three-dimensional orographic gravity-wave breaking observed in a hydraulic tank, J. Fluid Mech , vol 410, pp. 67-99. [3] J. D. Doyle and D. R. Durran (2002): The dynamics of mountain-wave induced rotor, J. Atm. Sc., vol. 59, pp.186-201.

Correspondence: Leroux – [email protected] – fax: +33 (0) 05 61 95 66


26

Simulation of turbulence, gusts and wakes for load calculations Jakob Mann Wind Energy Department, Ris¯ National Laboratory Frederiksborgsvej 399, DK-4000 Roskilde, Denmark

In order to simulate dynamic loads on a wind turbine rotor it is necessary to simulate the turbulent inflow to the rotor realistically. In this presentation we review various methods to do so. In many situations, for example for strong winds over flat terrain, turbulence is not far from being Gaussian. In this situations it is enough to know the second-order statistics, e.g. spectra and cross-spectra, to simulate. The second order statistics are either obtain from empirical one-dimensional spectra and coherences as in the Sandia method, or from a semi-empirical three-dimensional spectral tensor as in the Mann model. These methods, which both appear in the third edition of the IEC standard 61400-1, will be compared. The assumptions of the models, stationarity, homogeneity, and gaussianity, can all be questioned, especially when the terrain is not simple or when turbines appear in wind farms. Here a variety of methods are possible and we will present a few practical examples of these. Particular emphasis will be put on “constrained Gaussian simulation”where extreme gusts can be embedded in a turbulent field. This method can also be used simulate entire fields based on measurements in a few points. Practical simulations of wake turbulence based on meandering will also be presented.

Correspondence: Jakob Mann – [email protected] – fax: +45 46775970


27

Short time prediction of wind speeds from local measurements Holger Kantza a Max Planck Institute for the Physics of Complex Systems, N¨ othnitzer Str. 38, 01187 Dresden, Germany.

The prediction of wind speeds with a prediction horizon of a few seconds based on local measurements would be extremely useful for the optimal control of wind turbines. We present an overview of different schemes for such purposes and compare their performance. Special emphasis is laid on the prediction of turbulent gusts, where data driven continuous state Markov chains turn out to be quite successful. Model verification by reliability plots and a test of their performance by ROC statistics is discussed in detail. Taking into account correlations of several measurement positions in space enhances the predictability. As a striking result, stronger wind gusts possess a better predictability.

Literatur [1] M. Ragwitz and H. Kantz (2002): Markov models from data by simple nonlinear time series predictors in delay embedding spaces, Phys. Rev. E 65 056201. [2] H. Kantz, D. Holstein, M. Ragwitz, N.K. Vitanov (2204): Markov chain model for turbulent wind speed data, Physica A 342 315.

Holger Kantz – [email protected] – fax: +49 351 871 1999


28

The statistical distribution of turbulence driven velocity extremes in the atmospheric boundary layer – Cartwright/Longuet-Higgins revised G.C. Larsena and K.S. Hansenb a Wind Energy Department, Risø National Laboratories, P.O. Box 49, DK-4000 Roskilde, Denmark b Department of Mechanical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark

The statistical distribution of extreme wind excursions above a mean level, for a specified recurrence period, is of crucial importance in relation to design of wind sensitive structures. This is particularly true for wind turbine structures. Based on an assumption of a Gaussian ”mother”distribution, Cartwright and Longuet-Higgens [1] derived an asymptotic expression for the distribution of the largest excursion from the mean level during an arbitrary recurrence period. From its inception, this celebrated expression has been widely used in wind engineering (as well as in off-shore engineering) - often through definition of the peak factor, which equate the mean of the Cartwright/Longuet-Higgens asymptotic distribution. However, investigations of full scale wind speed time series, recorded in the atmospheric boundary layer, has revealed that the Gaussian assumption is inadequate for events associated with large excursions from the mean [2]. Thus, the more extreme turbulence excursions (i.e. the upper tail of the turbulence PDF) seem to follow an Exponential-like distribution rather than a Gaussian distribution, and a Gaussian estimate may under-predict the probability of large turbulence excursions by more than one decade, which is obviously unfortunate in relation to modelling of an asymptotic extreme distribution based on a Gaussian “mother” distribution. The potential problems with applying the Cartwright/Longuet-Higgens distribution to describe extreme gust events, related to wind speeds in the atmospheric boundary layer, is further stressed by the fact, that many extreme value investigations of wind speed gusts (e.g. [3, 4, 5]) have shown, that observed extreme values are well/excellently described the Gumbel EV1 distribution, which, on the other hand, differ from the asymptotic Cartwright/Longuet-Higgens expression. The present paper presents an asymptotic expression for the distribution of the largest excursion from the mean level, during an arbitrary recurrence period, based on a “mother” distribution that reflects the Exponential-like distribution behaviour of large wind speed excursions. This is achieved on the expence of an acceptable distribution fit in the data population regime of small to medium excursions which, however, for an extreme investigation is unimportant. Further, the derived asymptotic distribution is shown to equal a Gumbel EV1 distribution, and the two distribution parameters are expressed as simple functions of the basic parameters characterizing the stochastic wind speed process in the atmospheric boundary layer. Finally, model predictions of the derived model are compared to predictions from the Cartwright/Longuet-Higgens model as well as to results derived from full-scale measurements of wind speeds in the atmospheric boundary layer.

Literatur [1] D.E. Cartwright and M. S. Longuet-Higgins (1956). The statistical distribution of the maxima of a random function, Proc. Royal Soc. London Ser. A 237, pp. 212-232. [2] M. Nielsen, G.C. Larsen, J. Mann, S. Ott, K.S. Hansen and B.J. Pedersen (2003). Wind Simulation for Extreme and Fatigue Loads. Risø-R-1437(EN). [3] G.C. Larsen, K.S. Hansen and B.J. Pedersen (2002). Constrained simulation of critical wind speed gusts by means of wavelets, 2002 Global Windpower Conference and Exhibition, France. [4] G.C. Larsen and K.S. Hansen (2001). Statistics of Off Shore Wind Speed Gusts, EWEC’01, Copenhagen, Denmark, 2-6 July. [5] G.C. Larsen and K.S. Hansen (2004). Statistical model of extreme shear. Special topic conference: The science of making torque from wind, Delft (NL), 19-21 April, Delft University of Technology, pp. 433-444. Correspondence: Gunner C. Larsen – [email protected] – fax: +45 4677 5056

Session C – Talk 1

29

Wind farm power fluctuations P. Sørensen, J. Mann , U. S. Paulsen and A. Vestha a Risø National Laboratory, VEA-118, PO Box 49, DK-4000 Roskilde, Denmark

This paper presents the results of a study of horizontal wind speed variations measured on five meteorology masts on Risøs test site for wind turbines in Høvsøre near the west coast in Denmark. The aim of the study is to provide input parameters for a model simulating horizontal wind speeds in a number of points corresponding to the positions of wind turbines in a wind farm [1]. This model is used to simulate power fluctuations in wind farms [2], which is useful for planning of operation of power systems with large wind power penetration, especially when a large part of the wind power is concentrated in a limited geographical area. The wind simulation model is based on power spectral densities for the wind speed measured in a single point, combined with a direction dependent coherence function describing the coherence between the wind speeds in two points. The simulations with the wind speed model presented in [2] have assumed a Kaimal PSD, which is also used for structural design of wind turbines. However, the Kaimal spectrum does not include the large, slow variations of the wind speed, which are very important for power fluctuations from large wind farms because the slow variations are the most coherent, and thus they are smoothed less when the power from many wind turbines is added. One way to include the slow variations in the model is to modify the PSD. Figure 1 shows the PSD of the measured wind speed using segments of 215 = 32768 seconds, i.e. more than 9 hours length. It is seen that the measured spectrum agrees reasonably for frequencies above 0.002 Hz, but for slower variations the agreement is poor. It is proposed to add a low frequency (LF) term to the Kaimal spectrum to fit an analytical spectrum better to the measurements.

Fig. 1: Wind models for simulation of power fluctuations from wind farms. J. Wind Eng. Ind. Aerodyn. (2002) (no.90) , 1381-1402

Literatur [1] P. Sørensen, A.D. Hansen and PAC Rosas (2002): Wind models for simulation of power fluctuations from wind farms, J. Wind Eng. Ind. Aerodyn., vol. 90, pp. 1381-1402. [2] P.A.C. Rosas : Dynamic influence of wind power in power systems. Ph.D. thesis. Technical university of Denmark. Correspondence: P. Soerensen


30

Network perspective of wind-power production M. Greinera a Corporate Technology, Information & Communications, Siemens AG, D-81730 M¨ unchen, Germany

Power production in wind farms faces fluctuations on various levels. On the level of a single turbine it is the temporal longitudinal fluctuation of the wind velocity coupled with its transverse spatial heterogeneity. The intra-farm wind flow introduces more spatial and temporal heterogeneity, supplemented with numerous wakeflow effects. At last, the accumulated power output of a wind farm itself represents a volatile source for the power grid. It is this layer which demands for a control of these source fluctations as well as those resulting from intra-grid power redistribution. New concepts for such a control can be borrowed from modern information and communication technologies. With a distributive routing and congestion control selforganizing communication networks are able to adapt to the volatile traffic sources and the current network-wide congestion state [1, 2]. As a result network operation becomes very robust. The exportation of these ideas into wind-powered systems requires to see wind farms as well as the power grid from the network perspective. Intra-communication is needed for information exchange and artificial intelligence is required for information processing, diagnosis and prognosis. Based on such network-function capabilities, first distributed control schemes will be presented for the selforganized monitoring and controlling of wind farms and the power grid.

Literatur [1] I. Glauche, W. Krause, R. Sollacher and M. Greiner: Distributive routing & congestion control in wireless multihop ad hoc communication networks, Physica A, vol. 341, pp. 677-701 [2] W. Krause, J. Scholz and M. Greiner: Optimized network structure and routing metric in wireless multihop ad hoc communication, arXiv:cs.NI/0503010

Correspondence: Martin Greiner – [email protected] – fax: +49 (0) 89 636-49767


31

Phenomenological Response Theory to Predict Power Output A. Rauha a Institute of Physics, University of Oldenburg, Oldenburg, Germany

When the power output L(ti ) of a wind energy converter (WEC) is plotted versus longitudinal wind velocity u(ti ), a broadly spread cluster of points is obtained, in general, which hides and covers over the power curve. The latter can be considered as an attractor which is reached under, sufficiently long prevailing, constant wind speeds. Apart from the fluctuating wind and of noise, the deviations from the power curve are caused by the fact that the WEC does not respond intantaneously to wind speed changes but with some delay of the order of a couple of seconds up to minutes depending on the size of the WEC. In order to examine the influence of the delayed response on the average power output, we have proposed a phenomenological model [1]: A first order differential equation which contains a time dependent relaxation parameter r(t). The latter, essentially, couples to the wind field via a frequency dependent response function of the WEC. The reasoning for the model and implications are discussed in some detail. As a numerical application, measured data of power output are analysed within the model, to extract the frequency dependent response function. Thereby, a recent method [2] is used the establish the power curve from the power and wind data.

Literatur [1] A. Rauh and J. Peinke, ’ A phenomenological model for the dynamic response of wind turbines to turbulent wind’, J. Wind Eng. Ind. Aerdyn. 92(2003). [2] E. Anahua, F. B¨ ottcher, S. Barth, J. Peinke, M. Lange, ’Stochastic Analysis of the Power Output for a Wind Turbine’, Proceedings of the European Wind Energy Conference-EWEC in London 2004, as CD.

Correspondence: [email protected]


32

Turbulence Correction for Power Curves Klaus Kaisera , Wibke Langrederb , Harald Hohlenc , Jørgen Højstrupd a Wind Solutions, Engelsgrube 25, 23552 L¨ ubeck, Germany b Ingenieurb¨ uro, Gr. Burgstr. 27, 23552 L¨ ubeck, Germany c DeWind GmbH, Seelandstr. 1, 23569 L¨ ubeck, Germany d Suzlon Energy A/S, Kystvejen 29, 8000 Aarhus, Denmark

Summary The awareness that measured power curves depend on more site-specific parameters than covered by international standards is growing. The effect of turbulence intensity is difficult to evaluate because the anemometer as well as the turbine react in specific but different ways to turbulence intensity. Furthermore the process of binning will introduce an inherent numeric error due to the non-linearity of the power curve. Because the physics makes it difficult to determine the effects of turbulence on anemometer, turbine and numerics separately, a method is presented to quantify all three effects in total. The method has been tested for different types of turbines with good results. Hence the uncertainty related to site specific turbulence on the power curve can be reduced. Description The awareness for reliable and precise power curve measurements and verifications is growing within the community of investors, banks and insurances. Most contracts suggest the procedure according to IEC [1] to verify the power curve of the purchased WT either on site or under standard conditions. However the IEC does not cover all effects on the measurement related to site specific wind climates like e.g. turbulence. It is known that power curve measurements depend strongly on turbulence which is influenced by the topography at the wind turbine’s actual position [2]. Both anemometer and wind turbines respond to turbulence in very specific ways. Hence measurements of the power curve will naturally be effected by turbulence. Furthermore the process of binning will introduce an inherent numeric error due to the non-linearity of the power curve. Therefore power curve measurements of identical turbines at different locations will lead to different results. As a consequence measured power curves are neither comparable nor transferable. To decrease the uncertainty related to site specific effects a conversion of power curves from test sites to other sites is needed. As a result the uncertainty of energy calculations would be reduced. Because the physics makes it difficult to determine the effects of turbulence on anemometer, turbine and numerics separately, a method is suggested to quantify all three effects in total. The classical method of bins can be extended by expressing the measured power as the sum of a turbulence independent and a turbulence dependent part [3]. With more than two values in a bin the system of equations is over-determined. The redundant equations can be used to minimise the error of the computation using the least-square method [4]. The presented method has been successfully tested on preliminary data in [5]. The analysis has been extended for different types of turbines with good results. A conversion of power curves from site specific turbulence distributions to different turbulence distributions seems to be possible. Hence besides standardising power curves with respect to air density a standardisation in relation to turbulence intensity can be introduced.

Literatur [1] IEC 61400-12: Wind Turbine Generator Systems – Part 12: Wind Turbine Power Performance Testing. [2] T. F. Pedersen et al: Wind Turbine Power Performance Verification in Complex Terrain and Wind Farms, Risø-R-1330, Roskilde 2002. [3] A. Albers, C. Hinsch: Influence of Different Meteorological Conditions on the Power Performance of Large WECs, DEWI Magazin Nr. 9, Wilhelmshaven 1996. [4] W. Langreder et al: Numerics of Power Curve Measurements, Proceedings Glowec, Paris 2002. [5] K. Kaiser et al: Turbulence Correction for Power Curves, Proceedings EWEC 2003, Madrid 2003. Klaus Kaiser – [email protected]


33

Online modelling of wind farm power for performance surveillance and optimisation J. J. Trujilloa , A. Wessela , H.-P. Waldlb , B. Langec a ForWind Center, Institute of Physics, University of Oldenburg, Germany b Overspeed GmbH & Co. KG, Oldenburg, Germany c ISET e.V., University of Kassel, Germany

For wind farms, the surveillance of the energy yield is a crucial task to ensure the positive commercial performance of the project. For this purpose, the performance of single turbines in a farm must be investigated and optimised. It is known from practical experience that many reasons for an unsatisfying energy yield occur. Wrong adjustment of parameters as the blade pitch angles, or inadequate control parameters of the turbine control system can lead to significant energy losses. The turbines in a wind farm interact via wake effects. So it is hard to tell for a turbine inside a wind farm if the power production meets the expected values or not; even, if an external met mast is installed. The readings from nacelle anemometers which are used in practice to obtain the power curve of an individual turbine are also very uncertain. Because it is influenced by the rotor and the nacelle itself, the measurements are recalibrated by the wind turbine SCADA firm ware which could be faulty. In addition, the measured values depend on the current control strategy of the turbine. We will present a method how to obtain the expected performance for each single turbine in a wind farm for comparison to the real measured power performance. Our approach is based on modelling this value as time series with an advanced wind farm wake flow model (FLaP, [1, 2]). The modelling was tested against measurement data from two farms, a large wind farm with 34 stall regulated turbines, and a 7 turbines farm with pitch control. The current power characteristics and the energy yields are estimated by the model and compared to the real world measurements for short (10 minutes) as well as for long (month, year) time periods. A main focus of the work is to model not only the expected energy yields, but also the uncertainty of these values due to limited modelling accuracy. Furthermore, we will present an on-line application of the method as part of the wind farm management system OptiFarm [3].

Literatur [1] FLaP, http://www.forwind.de/gb/produkte dienstleistungen/flap/index.shtml [2] A. Wessel: Modelling turbulence intensities inside wind farms. This conference. [3] www.optifarm.de

Correspondence: Igor Waldl – [email protected] – fax: +49 441 361163-10


34

Uncertainty of Available Wind Energy Estimation Based on Long Term Meteorological Measurements in Hungary ´ am Kissb , András Zénó Gyöngyösia , Krisztina Krassovánb and Botond Tamás Weidingera , Ad´ Pappb a Department of Meteorology, E¨ otv¨ os Lor´ and University, P´ azm´ any P. st. 1/A, Budapest, Hungary b Department of Atomic Physics, E¨ otv¨ os Lor´ and University, P´ azm´ any P. st. 1/A, Budapest, Hungary

With the development of low cut-in wind speed turbines for continental climatological conditions (3-4 m/s yearly mean wind speed), wind power has become an economical source of energy also in Hungary. The first high power wind turbine of the country is operational since late 2000 (Inota, Balaton Highlands, 250 kW). Currently 6 wind turbines are working at a total power of 3.25 MW. By the end of 2010 3.6% of the total domestic electric power production is planed to be covered by renewable energy. Major portion (>200 MW) of renewable energy is planed to be generated using wind energy. In the estimation of wind energy above various terrain features the joint analysis of long term climatological data series and local wind measurements become more important. Based on Hungarian wind data series, the inter-annual variability of average wind speed and the uncertainty of the power-law wind profile method has been investigated (Figure ). It is can be easily seen that a small deviation in the average wind speed results in a relatively large deviation in the amount of energy production due to the third power dependence of wind energy on wind speed. The joint effect of inter-annual variability and the variation of the power law exponent on (i) wind power estimation and (ii) rate of return at planed wind farm projects has been quantified. We compare energy estimations based on the Central European Wind Atlas to other available wind data. The annual variation of available wind energy has been analyzed, and the uncertainty of the estimation methods has been given.

Fig. 1: Uncertainty in the estimation of wind energy at the Balaton Highland Area. Left: Estimated energy production of the 2003 year compared to the yearly average of the 2001-2003 period. Right: Estimated energy production of the 2003 year using 1/7 power law comapred to exponential power with a diurnal variation. In the conclusion of the environmental impact assesment of a planed wind farm in the Balaton Highland Area (45 MW total energy production), the energy production was 40-45% more for the windy 2001 year and 40-50% less for the calm 2003 year compared to the production in the 2001-2003 period, respectively. We are currently working on the temporal and spatial extension of our results, i.e., further details are to be presented in the proposed paper for longer time period and more meteorological stations.

Correspondence: Andr´ as Zén´ o Gy¨ ongy¨ osi – [email protected] – fax: +36 1 372 2904

Session D – Talk 1

35

Modelling of the transition locations on a 30% thick airfoil with surface roughness B. Hillmera , Yun Sun Cholb & A.P. Schaffarczykb a Computational Mechanics Laboratory, University of Applied Sciences Kiel, Grenzstr. 3, 24149 Kiel, Germany b Department of Mathematics & Mechanics, Kim Il Sung University, Pyongyang, DPR of Korea

The sustainability of wind energy production is considerably affected by electricity production costs per kWh. Concerning rotor blades a reduction of costs means in the majority of cases a reduction of masses. Therefore at inboard and mid-span locations of multi-megawatt turbines thick airfoils ( > 30%) are increasingly used. However, an unfavourable design of the leading edge area leads to an increased sensitivity to additional surface roughness of these airfoils. For the airfoil DU97-W-300 [1, 2] with a slightly modified shape extensive experimental research in DNW’s cryogenetic wind tunnel at Cologne (KKK) has been done in cooperation with the wind turbine manufacturer DeWind. Great importance was attached to the influence of surface conditions by investigating cylindrical tripping wires, zigzag tapes of different heights and sandpaper consisting of Carborundum with different grain sizes. During the measurement campaign the Reynolds number (Re) was varied from about one to ten million by decreasing the fluid temperature down to 100K. The analyses done so far considered mainly the performance depending on Re and on the surface conditions considered as spread roughness [3, 4]. For advanced basic comprehension of the flow regime at surface roughness and for improved modelling further numerical flow simulations of the airfoil with tripping wire and zigzag tape have been carried out. For this purpose the CFD code FLOWer developed by DLR including the transition module has been used. New results from the measurements regarding the influence of surface roughness on lift and drag are presented and compared with simulation results.

Literatur [1] R.P.J.O.M. van Rooij, W.A. Timmer (2003): “Roughness Considerations for thick Rotor Blade Airfoils”, Journal of Solar Energy Engineering, vol. 125, pp. 468-478. [2] W. A. Timmer, R.P.J.O.M van Rooij (2003): “Summary of the Delft Universtity Wind Turbine Dedicated Airfoils”, AIAA-2003-0352, pp. 11-21. [3] W.A. Timmer, A.P. Schaffarczyk (2004): “The effect of roughness on the performance of a 30% thick wind turbine airfoil at high Reynolds numbers”, WIND ENERGY 7 (4), pp. 295-307. [4] K.Freudenreich, K. Kaiser, A.P. Schaffarczyk, H.Winkler and B.Stahl (2004): “Reynolds Number and Rougness Effects on Thick Airfoils for Wind Turbines”, WIND ENGINEERING, vol. 28, no. 5, pp.529-546.

Benjamin Hillmer – [email protected] – fax: +49 (0) 431 210-62612


36

Determination of Angle of Attack (AOA) for Rotating Blades Wen Zhong Shen, Martin O.L. Hansen and Jens Nørkær Sørensena a Department of Mechanical Engineering, Technical University of Denmark Building 403, 2800 Lyngby, Denmark

For a 2D airfoil the angle of attack (AOA) is defined as the geometrical angle between the flow direction and the chord. The concept of angle of attack is widely used in aero-elastic engineering models (i.e. FLEX and HAWC) as an input to tabulated airfoil data that normally are established through a combination of wind tunnel tests and 3D corrections. For a rotating blade the flow passing by a blade section is bended due to the rotation of the rotor, and the local flow field is influenced by the bound circulation on the blade. As a further complication, 3D effects from tip and root effects render a precise definition of the AOA difficult. Today, there exists a lot of experimental and numerical (CFD) data from which precise airfoil data may be extracted. However, this leaves us with the problem of determining lift and drag polars as function of the local AOA. To determine the angle of attack from e.g. a computed flow field, two techniques have up to now been used. The first technique corresponds to an inverse Blade Element - Momentum (BEM) method in which a measured or computed load distribution is used as input. The technique gives reasonable results and the extracted airfoil data can be tabulated and used for later predictions. Since the theory is one dimensional, the accuracy of the extracted airfoil data is restricted to the 1D limitation. Moreover, a BEM code needs a tip correction; different tip corrections results in different airfoil data. The second technique is the averaging technique employed in [1] and [2]. This method gives good results. Since the method utilizes averaged data, many input points are needed to evaluate the local flow features. Furthermore, it is difficult to be used for general flow conditions (e.g. at operations in yaw). In the present work we have developed a simple and general method for determining angles of attack for rotating blades. The method only needs load and velocity distributions on a limited number of sections and control points along the blades. The method works as follows. First, a guessed angle of attack and the actual loading are used to determine the lift distribution. Next, the circulation is computed using the Joukowski law. The final step is then to determine a new angle of attack by subtracting the induced velocity due to the bound circulation on the blades from the measured or computed velocity field. Only a few iterations are needed to achieve a converged solution. The method can be used for extracting airfoil data from CFD computations or experiments (for example measurements using a pitot-tube or the PIV technique). In the present work the method is applied to extract data from 3D CFD computations of the flow past the Tellus rotor [2]. Compared to the averaging technique [1, 2] the new method shows promising results.

Literatur [1] J. Johansen and N. N. Sørensen (2004): Airfoil characteristics from 3D CFD rotor computations, Wind Energy, vol. 7, pp. 283-294. [2] M. O. L. Hansen and J. Johansen (2004): Tip Studies Using CFD and Comparison with Tip Loss Models, Wind Energy, vol. 7, pp. 343-356.

Correspondence: Wen Zhong Shen – [email protected] – fax: +45 45 93 06 63


37

Unsteady characteristics of flow around an airfoil at high angle of attacks and low Reynolds numbers Hui Guoad , Hongxing Yanga , Yu Zhoub , David Woodc a Department of Building Services Engineering, Hong Kong Polytechnic University, Hong Kong b Department of Mechanical Engineering, Hong Kong Polytechnic University, Hong Kong c School of Engineering, University of Newcastle, Callaghan, Australia d School of Aeronautical Science and Engineering, Beijing University of Aeronautics and Astronautics, Beijing, China

This paper presents the lift and drag test data, together with laser-sheet visualization, PIV and LDV experimental results of an NACA0012 airfoils at angles of attack from 0◦ to 90◦ and Reynolds numbers of 5 · 103 to 1 · 105 , to explore the aerodynamic characteristics of an airfoil at post-stall angles of attack and low Reynolds numbers, which are essential in starting process of a small wind turbine [1, 2]. The reasons for this study are: (1) there lack of lift and drag coefficient test data in this regions of angles of attack and Reynolds numbers in the available documents; (2) the flow physics in this region is too complex to understand enough by now, where the flow is dominated by strong unsteadiness and separation, together with instability of the separated shear layer, flow transition, formation of vortices and their interaction, as well as the interaction of the two shear layers from leading edge and trailing edge respectively [3]. The Experimental studies are conducted in the Low-Speed Water Tunnel in the Hong Kong Polytechnic University, with test section of 0.3m × 0.6m, and the chord length of the NACA0012 test model is 0.1m. In view of the strong unsteadiness of the flow and the low magnitude value of lift and drag acted on the airfoil model at such low Reynolds numbers, the load cell, which is highly sensitive to the force load, was used to obtain not only the mean but also the fluctuating value of the life and drag. The visualization, PIV and LDV experiments aimed to provide detailed information of the flow field to explore the physics of the lift characteristics. From analysis of these experimental results, some main conclusions can be drawn as follows: (1) The curve of mean lift coefficient versus angle of attack is characterized by a second going up after stall and reaching its second maximum at about α ≈ 40◦ , like those at high Reynolds numbers. (2) The flow is strong unsteady but with obvious regularities, demonstrated by all the results from force measurements, laser sheet visualization, PIV and LDV. Laser sheet visualization clearly reveal the process of oscillation of the separated shear layer, rolling up into vortices, the amalgamation of two adjacent vortices into a bigger one, and then amalgamating into to even bigger one. Such kind of formation and amalgamation of the vortices can be observed at both the leading and trailing edge. In these events, the flow behaves in periodic way. The LDV results reveal the spectral features. (3) Lift characteristics versus angle of attack are mainly determined by the vortices evolution, which depends on the angles of attack. Because change of angle of attack cause the change of adverse pressure gradient at lee side of the airfoil, which influences the instability and evolution of the separated shear layer, and so the forming and development of the subsequent vortices. Analysis of vorticity distributions from PIV demonstrates that the magnitude of lift directly depends on the strength of the formed vortices and the time interval of their flowing over the upper surface of the airfoil in a fundamental period, with which the Karman-cortex- street like vortices shed in the wake of the airfoil flow.

Literatur [1] Wright AK, Wood DH (2004): The starting and low wind speed behavior of a small horizontal axis . [2] Tangler JL(2004): Insight into a wind turbine stall and post-stall aerodynamics, Wind Energy, vol.7, Issue 3, pp. 247-260. [3] Wu JZ, Lu XY, Denny AG, Fan M, Wu JM, 1998, Post-stall flow control on an airfoil by local unsteady forcing. J. Fluid Mech. 371: 21-58. Correspondence: Hongxing Yang – [email protected] – fax: +852-2765-7198


38

Multi-criteria aerodynamic shape optimization of VAWT blades profile by viscous approach R. Bourguet, G. Martinat, G. Harran, M. Braza Institut de Mécanique des Fluides de Toulouse, 6 allée du Professeur Camille Soula, Toulouse, France

1. PHYSICAL CONTEXT OF THE STUDY The purpose of this study is to introduce an original blade profile optimization method including CFD calculation in wind-turbine aerodynamic context. This 2D implementation focuses on Darrieus Vertical Axis Wind Turbines (VAWT). This class of VAWT is generally associated with classical aeronautic blade profiles (NACA00XX) whereas Reynolds number and stall conditions are different in aerogeneration case (cf.[1]). The objectives of this multi-criteria optimization are: to increase nominal power production, to widen efficiency range in order to consider wind instability and to reduce blade weight. Performance graphes are obtained by using Templin static method (cf.[2]) in which the aerodynamic coefficients are computed for several azimuthal positions. The flow is simulated by a Reynolds Average Navier-Stokes (RANS) approach with Chen two-equation “near-wall” model and computed by StarCD.

2. OPTIMIZATION METHOD: DOE/RSM Blade profiles are parameterized by a Bézier curve and thus, the shape parameters are the coordinates of the control points. The optimization approach used in Optimus is a parametric Design Of Experiments (DOE)/Response Surface Models (RSM) method. An hybrid cost function surface is computed by a weighted summation of the RSM associated to each initial criterion. On this hyper-surface, classical deterministic optimization algorithms like Sequential Quadratic Programming method or stochastic ones (genetic algorithms, simulated annealing, auto-adaptive evolution...) are applied to reach the global optimum. An optimization computed on a two parameter shape design space produces an original profile (cf. fig.2) thicker than classical ones (+28% nominal power production, +46% efficiency range). The capabilities of the present method will be emphasized by an increased

Fig. 1: Parametrization, example of RSM and optimal blade profil number of control points and by an asymmetric profile optimization process. Therefore, this approach is promising for VAWT optimization. An improvement of the numerical simulation by a more efficient statistical eddy simulation based on phase-average will accurate the physical model and thus, the realism of the optimization. The next step of this study will consist in implementing other shape design technics like topological analysis and level-set method. Furthermore, improved CFD methods will be used to take into account of unsteady effects like dynamic stall. Acknowledgements for financial support: ADEME, EDF R&D, “réseau RNTL”/METISSE.

Literatur [1] I. Paraschivoiu (2002): Wind Turbine Design, with emphasis on Darrieus Concept, Pol. Int. Press [2] R.J. Templin (1974): Aerodynamic Performance Theory for the NRC Vertical Axis Wind Turbine, N.A.E. report, LTR-LA-160 Correspondence: Rémi Bourguet – [email protected] / Guillaume Martinat – [email protected] / fax: 33(0)5 61285992


39

Helicopter Aerodynamics with Emphasis Placed on Dynamic Stall Wolfgang Geissler

a

a DLR - G¨ ottingen, Germany

Aerodynamic problems on helicopters are versatile and mainly present on blades of the main rotor caused by the different flight conditions of the vehicle: The helicopter can lift off the ground vertically, may stand still in the air or fly forward with varying speeds. These different flight conditions are the cause of specific aerodynamic properties. During the ”hovercondition the rotor blade encounters a so called collective pitch, i.e. a constant incidence of each blade to keep the vehicle in the air. The aerodynamics during this flight condition can be assumed as ”steady¨ or time independent. The condition changes completely if the vehicle starts a forward flight. Then the blades have to change their incidence during blade rotation which is realized via the swash plate introducing the so called ”cyclic pitch¨ of the blades. The cyclic pitch is necessary in order to balance the lift of the rotor during its rotation. In this flight condition the flow around the blades is unsteady, i.e. time dependent and therefore of considerable complexity: The forward moving blade (advancing blade) is running into high speeds up to transonic Mach numbers in the blade tip region and the backward moving blade (retreating blade) has to increase incidence to balance lift and may run into flow separation. The latter process is called ”Dynamic Stall”due to the fact that the process of separation and later reattachment is highly unsteady. In addition to the problems described so far tip vortices are created at the different blade tips and at all flight conditions of the helicopter. These tip vortices may interact with the rotor blades under special flight conditions (descent flight) to create noise, the so called ”blade-vortex-interaction (BVI) noise”. In addition to the aerodynamic problems of the main rotor the interaction between rotor and fuselage plays a further important role as well as the interaction between main and tail rotor. In comparison with wind energy converters fundamental differences occur insofar as the helicopter consumes energy with its engine to drive the rotor and keep the vehicle in the air, the wind turbine rotor extracts energy from the wind flow. Nevertheless very similar aerodynamic problems occur in both cases: Due to local gust loads the wind turbine blades run into flow situations corresponding to the dynamic stall condition as mentioned above. Blade tip vortices are present in both cases as well to either interact again with blades or with a wind turbine tower in its neighbourhood. In the present paper the different aspects of unsteady flows during the dynamic stall process are discussed in some detail. Some possibilities are also pointed out to favourably influence dynamic stall by either static or dynamic flow control devices.

Correspondence: – fax:


40

Rotation and Turbulence Effects on Pressure Distribution on a Wind Turbine Blade – Unsteady Experimental Approach C. Sicot, P. Devinant, S. Loyer and J. Hureaua a Laboratoire de Mécanique et d’Energétique, Université d’Orléans 8 rue Léonard de Vinci 45072, Orléans, France

Incident flows on wind turbines are often highly turbulent, because these devices operate in the atmospheric boundary layer and often in the wake of other wind turbines. Wind tunnel tests have been performed earlier in our laboratory on the influence of turbulence level on wind turbine performance [1]. It has been observed little effects of turbulence on the wind turbine power and thrust. In order to complement these results, we performed wind tunnel tests on a model of wind turbine using a blade equipped with 26 pressure taps. The wind turbine diameter is 1.34 m. The blades have no twist, the airfoil is a NACA 654-421, and its chord length is 71 mm. The object was to study the turbulence effect on the blade airfoil aerodynamics. A turbulence with integral length scale of the order of magnitude of the chord length was generated using a grid. Modifying the distance between the grid and the wind turbine generated three turbulence levels (4.4%, 9% and 12%) on the test section. Pressure distributions have been measured at three positions along the blade (26%, 51% and 75%). Pressure taps were uniformly distributed on the chord to get information on how the position of the separation point on the suction face was influenced by the combination of such turbulence levels with the blade rotation. These taps were connected to 26 high frequency pressure scanners (PSI / ESP-32 HD and ESP-16 HD). Snel [2] and Corten [3] expected, in their theoretical work for non-turbulent incident flow, that a negative chordwise pressure gradient in the separated area arise for a rotating airfoil, due to the Coriolis force. This pressure gradient has been observed in our experiment whatever turbulence level. Moreover, the value of it seems to be independent of turbulence level. It has been observed a significant increase of the measurements dispersion and an important variation of the separation point position with turbulence level. These observations allow a better comprehension of the blade airfoil unsteadiness behaviour, which increases with turbulence level. Comparisons between these results and pressure measurements performed earlier in our laboratory on a non-rotating airfoil have been made, specially focusing on the differences in the stall behaviour.

Literatur [1] T. Laverne (2003): Aérodynamique des éoliennes à axe horizontal. Effets de la turbulence de l’écoulement amont et de la rotation sur le comportement aérodynamique des profils constitutifs. Conséquences sur le fonctionnement et les performances, Ph.D. Thesis. [2] H. Snel (2003): Review of Aerodynamics for Wind Turbines, Wind Energy, vol. 6, pp. 203-211. [3] GP. Corten (2001): Flow separation on wind turbine blades, Ph.D. Thesis.

Correspondence: Sicot – [email protected] – fax: +33 (0) 2 38 41 73 83


41

Simulation and Evaluation of the Air Flow through Wind Turbine Rotors by Comparing 2D and 3D Numerical Simulations with Focus on the Hub Area J. Rauch, T. Krämer, B. Heinzelmann, J. Twele, P.U. Thamsen

a

a Fluidsystemdynamik, Technische Universit¨ at Berlin Sekr. K2, Straße des 17. Juni 135 10623 Berlin, Germany

Blade Element and Momentum Methods (BEM) are the traditional design approach to calculate drag and lift forces of wind turbine (WT) rotor blades. The major disadvantage of these theories is that the airflow is reduced to axial and circumferential flow components [1]. A 3D numerical simulation is accomplished to map and analyze the flow field of the WT with a focus on the hub area and regarding radial flow components. The comparison of 2D and 3D simulation results showed differences especially at the innermost sections. The knowledge extracted from visualizations of radial effects on the flow field along the blade can be used as a tool for the design optimization of wind turbine rotor blades. r and the numerical solution is achieved in In this study the simulation grid is generated with Ansys ICEM 5.1 a multigrid consisting of 3 · 106 hexahedron using the computational fluid dynamics (CFD) commercial software r The rotor blade manufacturer EUROS provided the geometric data and the CFD Service Provider CFX 5.7.1 . CFX-Berlin provided the software support.

Fig. 1: Velocity Distribution 1 m behind the Rotor.

Literatur [1] Glauert H. Airplane Propellers, Vol. 4, Div. L in Aerodynamic Theory, edited by Durand W.F., Dover ed. 1943. Barbara Heinzelmann – [email protected] – fax: +49 (0) 30 314 21472


42

Predicting 2D and 3D Wind Turbine Performance using a Transition Model for General CFD Codes R.B. Langtrya , J. Golaa , S. Völkerb , F. R. Mentera a ANSYS CFX Germany GmbH, b General Electric Company

Wind energy is rapidly becoming an economically viable energy source. Like many aerodynamic devices such as turbine blades and wings, a better understanding of the flow field around the wind turbine could result in design changes that might significantly improve the performance. Because of the costs associated with performing wind tunnel experiments, there is a significant amount of interest in predicting the aerodynamic characteristics of a wind turbine using computational fluid dynamics (CFD). However, a survey of the available literature indicates that there are still some significant issues associated with accurately predicting wind turbine performance using CFD. According to Wolfe and Ochs [4] there are two main issues, poor separation prediction and the need to model laminar-turbulent transition. Recently, a new correlation-based transition model has been developed, which is based strictly on local variables. As a result, the transition model is compatible with modern CFD approaches such as unstructured grids and massive parallel execution and can predict transition in 3D flows. The transition model was first detailed in [1] and has been extensively validated for predicting transition in both turbomachinery (Langtry et al, 2004) and aeronautical flows ([2]). The first part of the presentation will compare the predicted 2D aerodynamic coefficients of the S809 wind turbine airfoil [4] to experiments and the well-known X-Foil airfoil design code. The second part of the presentation will detail the results obtained for the NREL Phase IV experiment [5], which was a 3D wind turbine that used the S809 airfoil for the blade profile. It will be shown that the transition model results in a significant improvement of the predicted power output, particularly at the higher angles of attack (see Figure 1). The differences between the 2D and 3D airfoil characteristics will also be discussed, particularly at the higher angles of attack were there is a significant amount of separated flow and the 3D effects are substantial.

Fig. 1: Shaft torque for the NREL Phase IV wind turbine (left) at different wind speeds (i.e. the higher the wind speed, the higher the angle of attack). Also shown is the flow topology on the suction side for fully turbulent and transitional simulations (right).

Literatur [1] Menter, F.R., Langtry, R.B., Likki, S.R., Suzen, Y.B., Huang, P.G., and Völker, S., (2004) (...) [2] Langtry, R.B., Menter, F.R., Likki, S.R., Suzen, Y.B., Huang, P.G., and Völker, S. (2004) (...) [3] Langtry, R.B., Menter, F.R., Likki, S.R., Suzen, Y.B., Huang, P.G., and Völker, S. (2004) (...) [4] Wolfe, W.P. and Ochs, S.S. “CFD Calculations of S809 Aerodynamic Characteristics”, AIAA Paper AIAA97-0972. [5] Simms, D., Schreck, S., Hand, M, and Fingersh, L.J. (2001) (...) Correspondence: R.B. Langtry

Session E – Talk 1

43

Modelling of Far Wake behind Wind Turbine J.N. Sørensena , V.L. Okulovab a Department of Mechanical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark b Institute of Thermophysics, SB RAS, Lavrentyev Ave. 1, Novosibirsk, 630090, Russia

To simulate basic features of wind turbine rotor aerodynamics various vortex models for far wakes are analyzed, resulting in the development of a new analytical wake model. In the model the vortex system is replaced by N helical tip vortices of strength (circulation) embedded in an inner vortex structure representing the vortices emanating from the inner part of the blades and the hub (see figure 1). The mathematical formulation of the model relies on a combination of the solution of Okulov [1] for the velocity field induced by a single helical vortex and the solution from [2] for columnar helical vortices. A simplified model, in which the inner structure of the wake is represented by a single central vortex line of strength, was firstly formulated by Joukowski (1912). This model can be derived by assuming a constant distribution of circulation of the rotor blades and results in a flow field in which the azimuthally averaged axial velocity is constant and thus independent of the radial distance from the rotor axis. This is in contradiction to experimental results where the axial velocity distribution is found to exhibit Gaussian-like behaviour [3]. To ameliorate the model of Joukowski the developed model is generalized to include various vorticity distributions representing the flow field in the inner part of the wake. In fact, it turns out that a simple model consisting of distinct helical tip vortices embedded in a Gaussian vorticity distribution is capable of modeling almost all measured velocity distributions of wakes behind wind turbines (see figure 2). In the presentation the model will be shown and compared to a series of measurements [3]. In particular we will show how some measured time-series can be explained as a combination of meandering and nonlinear dynamics of the tip vortices.

Fig. 1: Chronological allocation of WEC-outages of eight systems in the wind park over five months

Literatur [1] V.L. Okulov (2004) On the stability of multiple helical vortices. J. Fluid Mech., vol. 521, pp. 319-342. [2] P.A. Kuibin, V.L. Okulov (1996) One-dimensional solutions for a flow with a helical symmetry. Thermophys. & Aeromech., vol. 3(4), pp.335-339. [3] D. Medici, P.H. Alfredsson (2004) Measurements on a wind turbine wake: 3D effects and bluff-body vortex shedding, In The Science of making Torque from Wind (ed. G.A.M. van Kuik) pp. 155-165. DUWIND, Deft University of Technology. Correspondence: Jens N. Sørensen - [email protected] – fax: +45 4593 0663


44

Stability of Tip Vortex Equilibrium in Far Wake behind Wind Turbine V.L. Okulovab and J.N. Sørensena a Department of Mechanical Engineering, Technical University of Denmark, DK-2800 Lyngby, Denmark b Institute of Thermophysics, SB RAS, Lavrentyev Ave. 1, Novosibirsk, 630090, Russia

As a means of analysing the stability of the wake behind a multi-bladed rotor the stability of a multiplicity of helical vortices embedded in an assigned flow field is addressed. In the model the tip vortices are approximated by helical vortices that are spaced equally along the azimuth. The work is a further development of a model developed by one of the authors [1] in which linear stability of N equally azimuthally spaced helical vortices were considered. In the present work the analysis is extended to include an assigned vorticity field due to root vortices and the hub of the rotor. Thus the tip vortices are assumed to be embedded in an axisymmetric helical vortex field formed from the circulation of the inner part of the rotor blades and the hub. As examples of inner vortex fields we consider three generic vorticity distributions (Rankine, Gaussian and Scully vortices) at radial extents ranging from the core radius of a tip vortex to several rotor radii. The analysis shows that the stability of tip vortices to a large extent depends on the radial extent of the hub vorticity as well as of the type of vorticity distribution. As a part of the analysis it is shown that a model of [2] in which the vortex system is replaced by tip vortices of strength Γ and a root vortex of strength −N Γ is unconditional unstable. Stability of far wakes behind wind turbines, assuming uniform distribution of the hub vorticity in the inner cross-section of the wake [3], has been investigated using the new model. The model is capable of reproducing experimental observations about far wake stability/instability with realistic values of tip vortex pitch (figure 1).

Fig. 1: Neutral curves for far-wake model [3] with different numbers of tip vortices (N = 2, 3) as function of various wake parameters (2πl – helical pitch of tip vortices; a – radius of tip vortex system; σ - radius of tip vortex core; and δ - radius of hub vortex core.

Literatur [1] V.L. Okulov (2004): On the stability of multiple helical vortices. J. Fluid Mech., vol. 521, pp. 319-342 [2] W. Margoulis (1922) Propeller theory of Professor Joukowski and his pupils, NACA Memor. No. 79 [3] J.N. Sørensen and V.L. Okulov (2005): Modelling of Far Wake behind Wind Turbine, present symp.

Correspondence: Valery L. Okulov - [email protected] – fax: +45 4593 0663


45

Modelling turbulence intensities inside wind farms Arne Wessela , Joachim Peinkea , Bernhard Langeb a ForWind, University Oldenburg, Marie-Curie-Str.1, Oldenburg, Germany b now at ISET e.V., Koenigstor 59, 34119 Kassel, Germany

As a result from turbulence generated by the rotors of the wind turbines, the turbulence inside wind farms is enhanced. The higher turbulence lead to increased loads and a reduction in lifetime compared to free standing turbines might result. For the optimization of wind farm layouts (on- and offshore) these turbulence effects have to be taken into account and therefore an accurate model is important. Here we present a semi-empirical model for calculating these turbulence intensity inside a wind farm. The calculated turbulence intensity profile in the wake of a wind turbine is based on the wind deficit profile. Which is derived from an axialsymmetric solution of a simplified Reynolds-Equation based on the model from Ainslie [1]. Additionally Meandering effects of the wake, assumed to follow a gaussian distribution of the wind direction, are taken into account. Furthermore, in wind farm situation, the model deals with superposition of the wakes and partial wake covering of the incident rotor. The model has been verified with measurements from different wind farms. A detailed comparison with measurements from the offshore wind farm Vindeby uses turbulence intensity profiles from double and quintable wake measurements. For the Middelgrunden offshore wind farm, mean turbulence intensities at the wind turbines for particular wind directions are compared with our model results. The turbulence intensities at the wind turbines were estimated from power output fluctuations of the turbines by Jørgensen [2]. The model results are also compared to the results obtained with the Frandsen model [3]

Literatur [1] J.F. Ainslie (1988): Calculating the Flowfield in the Wake of Wind Turbines, Journal of Wind Engineering, vol. 27, pp. 213–224 [2] Hans E. Jørgensen, S. F. and P.V. (2003): Wake Effects on Middelgrunden Wind Farm, Risø-R-1415(EN),Risø National Laboratory [3] Sten Frandsen and M.L. Thogersen (1999): Integrated Fatigue Loading for Wind Turbines In Wind Farms By Combining Ambient Turbulence And Wakes , Wind Engineering, vol. 23(6), pp. 327–340

Correspondence: Arne Wessel – [email protected] – fax: +49 (0) 441 36116-735


46

Numerical Computations of Wind Turbine Wakes S. Ivanella , D. Henningsonb a Royal Institute of Technology, Stockholm, Sweden / Gotland University, Visby, Sweden b Royal Institute of Technology, Stockholm, Sweden

Numerical simulations using CFD methods are performed for wind turbine applications. The aim of the project is to get a better understanding of the wake behaviour, which is needed since today’s industrial design codes for wind power applications are based on the BEM (Blade Element Momentum) method. This method has been extended with a number of empirical corrections not based on physical flow features. The importance of accurate design models does also increase as the turbines become larger. Therefore, the research is today shifting toward a more fundamental approach, aiming at understanding basic aerodynamic mechanisms. The result from the CFD simulation is evaluated and special interest is given to the circulation and the position of vortices. From these evaluations, it will hopefully be possible to improve the engineering methods and base them, to a greater extent, on physical features instead of empirical corrections. ¨ The simulations are performed using the program EllipSys3D”developed at DTU (The Technical University of Denmark). The Actuator Line Method is used, where the blade is represented by a line instead of a large number of panels. The forces on that line are introduced by using tabulated aerodynamic coefficients. In this way, the computer resource is used more efficiently since the number of node points locally around the blade is decreased, and they can instead be concentrated in the wake behind the blades. An evaluation method to extract values of the circulation from the wake flow field is developed. The result shows agreement with classical theorems from Helmholtz, from which it follows that the wake tip vortex has the same circulation as the maximum value of the bound circulation on the blade.

Fig. 1: The figure shows an iso surface of the vorticity with a pressure distribution on the iso surface. The x=0-plane also shows the pressure distribution. The y=0-plane shows the streamwise velocity distribution.

Literatur [1] S. Ivanell (2005): Numerical Computations of Wind Turbine Wakes, ISRN/KTH/MEK/TR/–05/10–SE.

Correspondence: Stefan Ivanell – [email protected] – fax: +46 (0) 29 99 63


47

Modelling wind turbines wakes with a porosity concept S. Aubruna a Laboratoire de Mécanique et d’Energétique, 8 rue Léonard de Vinci; F-45072 Orléans cedex 2, France

The wind ”quality” of a site is neither controllable nor improvable by wind energy users. Then it must be correctly assessed before the implantation of a wind farm. Indeed the wakes from turbines impinge other turbines, modifying their inflow conditions and thus, their performance. Some attempts are made to use roughness-based models to calculate the wind field within a wind farm, increasing the roughness length to take the wind turbines into account [1]. Results are not satisfying and it is not really surprising since this concept is based on the global influence of roughness elements on the above boundary layer, whereas wind farms should be considered as spaced single roughness elements locally disturbing the atmospheric boundary layer. Wind turbines are not plain obstacles and are, for particular applications, considered as porous elements, extracting kinetic energy from the flow, distorting streamlines and generating turbulence. The use of a porosity-drag approach, inspired from concepts developed to model the urban of forest canopy flows [2] would be probably well adapted to simulate the influence of wind turbines on the atmospheric boundary layer. It enables to study the flow between the obstacles, which is unfeasible with a roughness-based concept. The porosity-drag approach is generally employed for numerical modelling but could be also used in physical modelling in wind tunnel. Aubrun et al. [3] used it to model the forest canopy. The present project is to model in a wind tunnel at a geometric scale of 1:400 a wind farm with metallic mesh discs to replicate the actuator discs. A parametric study of the flow field downstream of the wind turbine model (porous disc + cylindrical mast) has been performed in a homogeneous approaching flow (freestream turbulence intensity 5%) upon the disc size, the porosity level, the mesh size. It shows that, according to the porosity level of the mesh, the porous disc generates a constant velocity loss in its wake characterised by an axial flow induction factor up to 1/3, associated with a shear-generated turbulence at its circumference (Figure ). Then the interaction between several porous discs will be studied in order to model an off-shore located wind farm. In this goal, the wind tunnel ”Lucien Malavard” of the Laboratoire de Mécanique et d’Energétique (L.M.E.) will be modified to enable the physical modelling of the atmospheric boundary layer above the sea and under neutral conditions

Fig. 1: Streamwise mean velocity and turbulence intensity distribution, 2 diameters downstream of the wind turbine model, for an axial flow induction factor a = 1/4 (in black, porous disc border, porosity 50%).

Literatur [1] G.P. Corten and A. Brand (2004): “Resource decrease by large scale wind farming” EWEC 2004, November 22-25, London ,UK. [2] Dupont, S., Otte, T.L, Ching, J.K.S. (2004): Simulation of Meteorological Fields within and above Urban and Rural Canopies with a Mesoscale Model (MM5). Bound-Layer Met. 113/1, 111-158. [3] S. Aubrun, R. Koppmann, B. Leitl, M. M¨ ollmann-Coers, A. Schaub (2005): Physical Modelling of a complex forest area in a wind tunnel – Comparison with field data. Agr. and Forest Met. 129, 121-135. Correspondence: Sandrine Aubrun – [email protected] – fax: +33 (0) 238417383


48

Acoustic analogy applied on wind turbine noise generation and propagation L. Fuchs

a

and D. Moroianu

a

a Division of Fluid Mechanics, Lund University, P.O. Box 118, S-22100 Lund, Sweden

The aerodynamical noise produced by a wind turbine is computed using a hybrid method: Large Eddy Simulations (LES) and an acoustical analogy known in literature as the Ffowcs Williams and Hawkings equation [1]. The underlying assumption of the approach is that for the far-field the acoustic analogy ([2, 3, 4]) is rather accurate for describing the acoustic wave propagation. The boundary conditions are such that the acoustic waves are reflected from the ground and are allowed to leave the computational domain without being reflected. At the near field, the quality of the approach depends on the quality of the approximation of the acoustical sources. The acoustical sources include the contribution from the spatial variations in the flow field, the pressure that the blades exert on the air and the acoustical source due to the acceleration of the blades. To ensure an adequate quality of the source terms we use LES for the fully 4-D problem. The LES resolves all the scales that are responsible for the coherent structures generated by the rotating blades and the mast. Also, The spatial resolution is such that it resolves a portion of the inertial subrange. The dynamics of tip and hub vortices together with some coherent structures that appear in the downstream side of the wind turbine mast are depicted in Fig. 1a). A typical acoustical far-field is depicted in Fig. 1b.

Fig. 1: a) Tip and hub vortices; b) Cross sectional wave reflections by the ground at y = 2.5D, full geometry The paper will present details of the current computational approach. Different characteristics of the flow (the content of turbulent kinetic energy spectrum in different monitoring points, vortex visualizations) and also different features of the acoustical field (acoustical spectrum and typical frequencies in different monitoring points, sound pressure level, wave interaction among the wind-turbine blades) will be also included.

Literatur [1] Brentner, K. S., Farassat, F., ”Modeling Aerodynamically Generated Sound of Helicopter Rotors”, Progress in Aerospace Sciences 39, pp. 83-120, 2003. [2] Lighthill, M. J., ”On Sound Generated Aerodynamically: I. General Theory”, Proceedings of the Royal Society of London, Series A, Vol. 211, pp. 564-587, 1952. [3] Lighthill, M. J., ”On Sound Generated Aerodynamically: II. Turbulence as a Source of Sound”, Proceedings of the Royal Society of London, Series A, Vol. 222, pp. 1-32, 1954. [4] Lighthill, M. J., ”Waves in Fluids”, Cambridge University Press, ISBN 0-521-29233-6, 1978.

Correspondence: Laszlo Fuchs – [email protected] – fax: +46 - 46 - 222 43 00

Session F – Talk 1

49

Fatigue assessment of truss joints based on local approaches H. Th. Beier, J. Lange, M. Vormwalda a IFSW, Technische Universit¨ at Darmstadt, Petersenstr. 12, 64287 Darmstadt, Germany

Modern steel-constructions try to use optimised member connections. For often used connection-types the engineer must be able to calculate the bearing capacity (static and cyclic loading) in an easy and safe way. A methodology to obtain W¨ ohler-curves is outlined in the present paper by applying it to truss joints with pre-cut gusset plates (PCGP-joint), figure 1. The approach itself is supposed to be generally applicable to new constructional details, of which the fatigue performance is certainly not yet classified. The question of the static bearing capacity of the PCGP-joint has been answered by Adam and Zhang [1] for elastic and by Suppes [2] and Klinkenberg et al [3] for plastic behaviour. In order to obtain the fatigue resistance of the PCGP-joint a modern concept was used [4]: a minimised number of 15 tests were carried out on two types of normal scale specimens: (a) joints with sharp edges (r=2mm) and (b) with radii (r=10mm) at the free gap. Additionally, the fatigue resistance of these tests were calculated by the local strain approach for the crack initiation and by linear elastic fracture mechanics for crack growth. The Wöhler-curves of the tests and the calculations showed very good agreement. The verified calculation-model was then used to perform an extensive parameter study to obtain the fatigue resistance of PCGP-joints typically used in structural steelwork. As a result the PCGP-joints with radii at the free gap can only be classified in FAT 36 according to EC3 part 1-1 [5]. The joint version with the sharp edges can not be classified in a FAT-class and can therefore not be used in situations with cyclic loading.

Fig. 1: PCGP-joint

Literatur [1] V. Adam, X. Zhang (1994): Eine praktische Bemessungsmethode f¨ ur ausgeschnittene Knotenbleche zum Anschluß von I-Profilen, Stahlbau, vol. 63, pp. 49-58. [2] A. Suppes (1998): Tragverhalten und Optimierung von ausgeklinkten Knotenblechen in Fachwerkbindern, Institut f¨ ur Stahlbau und Werkstoffmechanik, TU Darmstadt, vol. 58. [3] A. Klinkenberg, A. Peter, H. Saal (1999): Berechnungsmodell f¨ ur geschweißte Anschl¨ usse in ausgeschnittenen Knotenblechen, Stahlbau, vol. 68, pp. 173-180. [4] H. Th. Beier (2003): Experimentelle und rechnerische Untersuchungen zum Schwingfestigkeitsverhalten von ausgeklinkten Knotenblechen in Fachwerkträgern, Institut f¨ ur Stahlbau und Werkstoffmechanik, TU Darmstadt, vol. 68 [5] Eurocode 3, Part 1-1, DIN V ENV 1993-1-1 (1994): Stahlbau, Stahlhochbau. Beuth Verlag.

Correspondence: M. Vormwald – [email protected]– fax: +49 (0) 6151-163038


50

Advances in Offshore Wind Turbine Technology M. Seidela , J. Gößweina a REpower Systems AG, Franz-Lenz-Str. 1, 49084 Osnabr¨ uck, Germany

The new generation of large wind turbines which is specifically designed for Offshore application is now in the prototyping phase. Late 2004 the first prototype of the REpower 5M has been installed in Brunsbuettel / Germany. With 5 MW rated power and a rotor diameter of 126 m the 5M is currently the largest wind turbine on the market. Since February 2005 the machine is in normal operation. Results with respect to the dynamic behaviour, power output and the technical performance will be presented. The first Offshore project with this turbine will be realized in 2006 – the DOWNVInD Demonstrator in Scotland in 45m water depth. For such a water depth Monopile and Gravity base foundations are no longer feasible. For the DOWNVInD Project a jacket structure is foreseen as substructure. Commercially available load calculation software (e.g. Flex 5 and Bladed) is only able to incorporate Monopiles as substructure below the tower, thus new developments are required for the integrated treatment of wind and wave loading for this type of support structure. REpower has developed a solution that combines the two existing programs Flex 5 and ASAS(NL) into one solution which makes integrated calculation of wind and wave loading possible. The paper will concentrate on the following aspects: • Experiences from erection and operation of the 5M prototype in Brunsbuettel. • Technical approach to combine Flex 5 and ASAS(NL) incl. validation. • Treatment of fatigue and extreme loading conditions.

Correspondence: Marc Seidel – [email protected]


51

Benefits of Fatigue Assessment with Local Concepts P. Schaumann, F. Wilkea a Institute for Steel Construction, University of Hannover, Appelstraße 9A, Hannover, Germany

For the design of structural parts of wind energy converters assessment of fatigue strength is essential because of high dynamic effects from wind loading and operation. For offshore converters additional dynamic effects arise from wave loading. Due to the combination of both actions the fatigue strength of connections often governs the design of structural parts. This contribution compares, based on load calculations in the time domain, some important issues of the fatigue design for recently discussed types of support structures for offshore wind energy converters (OWECS) taking into account the special environmental influences of their location. For the design of braced or lattice structures, which are discussed for offshore wind farms in greater water depths [1], local concepts for fatigue assessment are required. The most popular one, which is basis of all offshore standards, is the structural stress approach, also known as the hot-spot-concept. Besides the stress calculation with finite element models the offshore standards allow simplified fatigue design using parametric equations for the governing hot-spot stresses. It is shown that some of the special structural concepts for OWECS as well as the loading history do not meet the limitations of these parametric equations. In these cases the use of the finite element method is demonstrated to be essential. Moreover it is shown that more detailed guidelines have to be given for extrapolation techniques and definition of boundary conditions to produce reliable results.

Fig. 1: Typical levels of FE-models for the fatigue assessment with local concepts

The design results for typical structures are compared to those of more sophisticated local concepts like the notch stress approach (Figure 1), showing advantages and limits of the different concepts. As for higher water depths fatigue design leads to increased plate thicknesses, special focus is given to the evaluation of size and mean stress effects.

Literatur [1] P. Schaumann, P. Kleineidam and F. Wilke (2004): Fatigue Design of Offshore Wind Energy Conversion Systems, Stahlbau 73, No 9, pp. 716-726.

Correspondence: Fabian Wilke – [email protected] – fax: +49 (0) 511 762-2991


52

Extension of Life time of welded dynamic loaded structures T. Ummenhofera , I. Weicha , T. Nitschke-Pagelb a Institut f¨ ur Bauwerkserhaltung und Tragwerk, Technische Universit¨ at Braunschweig, Pockelsstr.3, Braunschweig, Germany b Institut f¨ ur F¨ uge- und Schweißtechnik, Technische Universit¨ at Braunschweig, Langer Kamp 8, Braunschweig, Germany

Research has been initiated on the application of weld improvement methods to increase the fatigue life of new and existing wind energy plants. The research is particularly relevant because a large number of welded wind energy plants will reach their calculated life time during the next years. Therefore the main investigation involved fatigue tests of specimen which have been treated with weld improvement methods after they reached their calculated life time. The effect of the application of Ultrasonic Peening (UP) has been compared to the effect of shot peening. These enabled the influence of the treatment method on the improvement of the fatigue life of “old” weld details to be investigated. The tests included testing of butt-, fillet- and longitudinal welds as well as double bevel seam. The results obtained out of the running tests show that both improvement methods produce very good results for preloaded specimen as well as for new notch details. The effect for plates with a thickness of 8mm and 30 mm could be analysed and the influence of the thickness could be determined. S-N-curves for treated notch details are the basis for the implementation of the methods in the design for repair of existing or new wind energy plants. Results show that the value of the slope m of the S-N-curves for treated specimen increases, so that the fatigue strength increases respectively the fatigue life will be raised further more. Laser measurements of the weld seam show that the UP increases the overall weld toe radii but also introduces sharp burrs at the edges of the treated zones. Residual stress measurements prove the introduction of inherent compressive stresses. It can be concluded that the positive effect of the treatment methods results from the inherent compressive stresses. The investigations demonstrate that UP is an effective and practical means to improve the service life of wind energy plants and other dynamic loaded structures, to increase their fatigue strength and last but not least for repair procedures.

Fig. 1: S-N-curves for the 95%- survival probability of the different treated notch details.

Literatur [1] T. Ummenhofer, I. Weich and T. Nitschke-Pagel (2005): Lebens- und Restlebensdauerverlängerung geschweißter Windenergieanlagent¨ urme und anderer Stahlkonstruktionen durch Schweißnahtnachbehandlung, Stahlbau, vol. 6, pp 412- 422. Correspondence: Imke Weich – [email protected] – fax: +49 (0) 531 391-2502


53

Structural Health Monitoring of WEC using the Multiparameter Eigenvalue Problem J. Reetza , W.-J. Gerascha and R. Rolfesa a nstitute for Structural Analysis, University of Hannover, Appelstr. 9A, 30167 Hannover, Germany

The potentiality to prolong the life cycle and thus to improve the efficiency of structures of windenergy converters (WEC) lies in the early identification of the occurring damages on the construction and the prevention of the consequential damage. Due to the bad accessibility to these constructions a regular survey is very complex and expensive. In view of this situation the aim of manufacturers, operators and insurers amongst others is the development of a system of early detection of damages. Eigenfrequencies of the structure depend on the current condition of the structure and are the modal quantity that can be measured most exactly. Assumed that damages on elasto-mechanical structures influence the dynamical behavior the approach is to measure current modal quantities and hence to inference the current condition. In the project presented here the problem is solved by means of inverse identification of the stiffness of the structure on the basis of measured current modal quantities. Therefore, parametrised mathematical models reproducing the dynamical behavior are set up and the eigenvalues of the current structure are measured. With the test data and the numeric model the model-parameters are calculated. Transferring the problem into a multiparameter eigenvalue problem only a minimal set of test data is necessary. In order to apply the method in a Condition Monitoring System for WEC-structures a validation of the method is carried out. Therefore numerical simulations, laboratory tests on models and tests on real constructions have to be made. Questions to be answered are about the quality and quantity of diagnostics. Additionally, the measurement procedures have to be further developed in order to ensure the sensitivity required and to provide a simple and robust system, that can resist the rough offshore conditions over many years. The present status of theoretical investigations shows e.g. high sensitivity, a wide working range and a small detection limit. Future work will be the extension of the method for larger models, other structures and further fields of application.

Correspondence: Johannes Reetz – [email protected] – fax: +49 (0) 511/762-8673


54

Influence of wind turbine’s type and size on antiicing thermal power requirement Battisti Lorenzoa , Fedrizzi Robertoa and Dal Savio Stefanoa a DIMS - Dept. Of Mechanical and Structural Engineering - University of Trento Via Mesiano, 77, 38050 POVO - Italy

There is a growing interest in Europe to build wind power plants at inland sites and especially within mountainous regions, and also in the far north. Characteristics of such sites are not only low air temperatures but also icing accretion on blades and other components. Ice accretion on wind turbine components affects system safe operation and performance. Therefore, special design requirements as ice prevention systems are necessary for wind turbines operating in cold climates. The aim of this paper is to present a model for evaluating the anti-icing energy requirements whit respect to wind turbines size and type as three, two and one bladed. A comparison of the results obtained with other models has been carried out. A sensitivity analysis, helpful in the early design phase, showed the effect of the different variables involved in the evaluation of the thermal power and energy for anti-icing purpose. The investigations enlightened the influence of the type and the size of the wind turbine on their anti-icing power requirements. As example, the figure 1 is reported It shows the variability of the specific heat flux qtb [W/m2 ], the anti-icing thermal power requirement for the wind rotor Qt [kW ] and ratio between rotor anti-icing thermal power and rated machines power with respect to the angle of attach configuration of the blade, for a series of rotors. The higher optimal turning speed at nominal conditions due to the different number of blades of the turbines leads to a direct increase in the cooling thermal fluxes. At the same time, however, there is also an increase in the heat flux due to aerodynamic heating, so there is actually no significant difference with a variation of number of blades in terms of specific anti-icing heat flux. However the total anti-icing heat flux requirement Qt, which depends on the number of blades, nonetheless decreases considerably from a three-blade to a single-blade machine, in a manner that is nearly proportional to the number of blades. From this analysis an indication of the anti-icing power requirements can be deduced, like as the optimum feature of wind turbines dedicated to cold climates.

Literatur [1] Seifert, H., Technical requirements for rotor blades operating in cold climate, DEWI, Deutsches WindenergieInstitut GmbH, 2003. [2] Makkonen L., Laakso T., Marjaniemi M., Finstad K. J., Modeling and prevention of ice accretion on wind turbines, Wind Engineering, 2001, 25(1), 3-21. [3] L. Battisti, R.Fedrizzi, M. Rialti, S. Dal Savio, A model for the design of hot-air based wind turbine ice prevention system WREC05 22-27 May 2005 Aberdeen Correspondence: Lorenzo Battisti – [email protected] – fax: +39 (0) 461 882599


55

Structure Health Monitoring – Visions, Possibilities and Research Tasks W. Hillger

a

a Institute of Composite Structures and Adaptronics/DLR Lilienthalplatz 7 D-38108 Braunschweig, Germany

There is a large need for a coast and time effective testing of rotor blades of wind generators for in- service inspections. Conventional NDT-methods such as thermography and ultrasonics are very time consuming and expensive. Structural Health Monitoring (SHM) has been defined as the “acquisition, validation and analysis of technical data to facilitate life-cycle management decisions” [1]. Furthermore, future SHM-systems will detect critical loadings and will give warnings before defects will be generated. Also the component will carry out a self test and automatically call the service if critical defects are detected. These statements sound futuristic, and of cause it will take more than fifteen years of research and development to design such a system. The basic research for this purpose is already described by many authors. Ultrasonic Lamb waves provide propagation over long distances in composites. In most of all application for the generation and receiving of Lamb waves separate piezo patches are used. The piezos can be embedded into the structure are implicated from outside. The complex construction of the rotor blades of wind generators, especially the different materials and thicknesses, meets a challenge for Lamb wave testing. For a given frequency at least two modes a symmetric and a anti-symmetric one are generated. The modes are dispersive, which means that that their velocities are frequency-depending. For the damage detection it is important to select the optimal wave mode and to know the propagation and the interaction with defects. Therefore research fields are: sensors, excitation, bonding/embedding of sensors, excitation, propagation, interaction with defects, signal processing and signal evaluation. Research tasks at DLR are visualisation of the Lamb wave propagation and the interaction between damages and signal correlations between progressing damage types and active Lamb wave signals. The ultrasonic imaging technique has been adapted for the visualisation of Lamb waves. Figure 1a shows a “classic C-scan” of a CFRPcomponent with two drillings (black) and a delamination (red) recorded by air- coupled ultrasonic testing, 1b presents a “Lamb-wave C-scan” of the same component with wave propagation and interaction with defects. One piezo bonded on the component has been used as a transmitter; an air-coupled transducer scanned the component. The diffraction at the defect is clearly indicated.

Fig. 1: Lamb wave interactions between bore holes and a delamination.

Literatur [1] Glauert H. Airplane Propellers, Vol. 4, Div. L in Aerodynamic Theory, edited by Durand W.F., Dover ed. 1943. Wolfgang Hillger – [email protected] – fax: +49 (0) 531 295 22 32


56

Performance and qualification of materials used for rotor blades B. F. Soerensena , Povl Brønsteda a Materials Research Department - Risoe National Laboratory - DK 4000 Roskilde

The optimal design of rotor blades is today a complex and multifaceted task and requires optimization of properties, performance, and economy. These requirements blades lead to that polymer reinforced composites are candidate materials. The fibres and the matrix resins for the composites are described, and their high stiffness, low density, and good fatigue performance are emphasised. The mechanical properties of fibre composite materials are discussed, with focus on fatigue performance. Methods for qualifying the materials are based on measurements of static and fatigue properties. Damage and materials degradation during fatigue is described. The testing procedures for documentation of properties are reviewed. Stiffness, static strength, and fatigue performance are measured by testing test coupons or components. The properties measured are used to qualify the materials and to document that the materials fulfill the design demands originated in aero elastic modeling of the blades under expected wind loads. Uniform static tests are carried out under quasistatic loading, and fatigue tests are carried out under varying loading conditions. Properties are measured under tensile loads, compressive loads, shear loads, or combinations of these in bi- and multiaxial loading modes.

Correspondence: B. F. Soerensen


57

High-cycle fatigue of ”Ultra-High-Performance Concrete¨ and ”Grouted Jointsfor Offshore Wind Energy Turbines L. Lohausa and S. Andersa a Institute of Building Materials, University of Hannover, Appelstraße 9A, 30161 Hannover, Germany

”Grouted Joints”have already been used in Monopile-foundations for offshore wind energy turbines and Tripodconstructions for offshore platforms in the oil and petroleum industry. In offshore wind energy applications ultra-high performance concrete (UHPC) has been solely used up to now. UHPC is known for its outstanding mechanical characteristics in static tests, but only little is known about its dynamic behaviour in terms of fatigue strength, damage development or the interplay of forces in the ”Grouted Joints”. Experiments on UHPC [3] have indicated that attention has to be paid to its fatigue limit, which seems to be significantly lower compared to normal-strength concrete, as illustrated in figure 1. Even the addition of fibres does not enhance the fatigue limit, only the explosive failure of UHPC can be avoided. Comparable results were obtained in an investigation of the University of Kassel [1]. As stated in the existing design rules for ”Grouted Jointsë.g [2] specimens containing shear keys showed an increasing ultimate load with an increasing shear key height. Another aspect about the behaviour of ”Grouted Joints¨ıs the mode of failure. Almost every specimen with shear keys indicates a shear failure along the shear keys as well as an exceeding of the grout matrix capacity. Both modes of failure can be clearly observed in figure 2. From the results obtained so far it seems as if the first linear part of load-displacement curve is related to the failure of the grout matrix capacity and the second linear section to the shear failure. Apart from the height of the shear keys, the ultimate load of the ”Grouted Jointsßpecimens seems to depend on the concrete mix used. Steel-fibre reinforced concrete mixes showed an enhanced first linear part of the curve, whereas the second part (the ultimate load) was not affected significantly.

Comparative calculations have shown, that the Scandinavian regulations [2] overestimate the ultimate load assuming that the end of the first part represents the ultimate value for design purposes. Therefore further tests are needed to adapt the existing rules to the results described. In addition, tests indicating the influence of fatigue loading are being performed and prepared for numerical analyses.

Literatur [1] E. Fehling, M. Schmidt, et al. (2003): Entwicklung, Dauerhaftigkeit und Berehnung Ultra-Hochfester Betone, Forschungsbericht an die DFG, Projekt Nr. FE 497/1-1, Universität Kassel. [2] Det Norske Veritas (1998): Rules for Fixed Offshore Installations, Det Norske Veritas. [3] Lohaus, L.; Anders, S. (2004): Effects of polymer- and fibre modifications on the ductility, fracture properties and micro-crack development of ultra-high performance concrete. In Proceedings “International Symposium on Ultra-High Performance Concrete”, p. 625-636, Kassel, 13.-15.09.04 Correspondence: Prof. Dr.-Ing. Ludger Lohaus – [email protected] – fax: +49 (0) 511-762 4736


58

A modular concept for integrated modeling of offshore WEC applied to wave-structure coupling K. Mittendorfa , M. Kohlmeiera , A. Habbara and W. Zielkea a Institute of Fluid Mechanics and Computer Applications in Civil Engineering University of Hannover, Appelstr. 9A, 30167 Hannover, Germany

Offshore wind energy can only become economical if adequate design methods allow highly optimized and robust structures with a long lifespan. To achieve this there is a need for reliable design methods to perform integrated simulation of offshore wind turbines in time domain. The analysis in time domain is necessary because of the nonlinearity of the structure response as well as the nonlinearity of the wind and wave loads. In the offshore environment, we have to consider a coupled system, consisting of the support structure, the foundation, the aeroelastic system for the wind model and the hydrodynamic loads due to waves (Fig. 1). The achievement of an integral model suffers from the diversity of different processes and process interactions to be taken into account for the analysis of an offshore wind turbine and its associated sub systems. Thus, the models used by research teams and consulting engineers are normally heterogeneous. Therefore, a flexible structure of the integral model is the main target of current research. A well designed object-oriented and easily extendable set of models and interfaces have to be developed in order to fulfill future demands.

Fig. 1: Components of the integrated model, processes, loads and sub systems.

In the first step a model for fluid structure interaction is implemented and will be presented here. The support structure is modeled with a finite element approach. Regular wave kinematics can be determined from different linear or nonlinear wave theories (e.g. Stokes or Stream Function). Subsequently with the known kinematics from the wave theory, the wave loads on the support structure are predicted with the Morison equation. Real sea state (irregular waves) simulations based on a spectrum even with consideration of the directionality are also feasible (Fig. 2).

Fig. 2: 2D irregular wave load simulation from spectrum.

Literatur [1] ForWind (2004): Annual Report 2003/2004, TP IX Correspondence: Kim Mittendorf – [email protected] – fax: +49 (0) 511 762-3777


59

Solutions of details regarding fatigue and the use of high-strength steels for towers of offshore wind energy converters J. Bergers a , H. Huhn b , R. Puthli

a

and M. Veselcic

a

a Research Centre for Steel, Timber and Masonry, University of Karlsruhe, Kaiserstr.12, 76128 Karlsruhe, Germany b IMS Ingenieurgesellschaft mbH, Stadtdeich 5, 20097 Hamburg, Germany

In the context of the support and development of alternative energies, many offshore wind parks are planned in the North Sea, Baltic Sea and Irish sea. Aggravated conditions such as large water depths, increasing heights of towers and hubs, as well as other environmental conditions require new solutions concerning the stability and the durability of offshore wind energy converters. Therefore, a new research project has been brought into life [1]. The aim of this project is to evaluate the individual problems concerning stability and durability of such constructions and develop new solutions. This includes the investigation of the influence of steel grade and wall thickness, the assessment of the remaining life and the questions about cylindrical and conical connections. Another important aspect is the adjustment of the existing fatigue design concept to the special requirements of offshore wind energy converters. Generally, it is assumed that with an increasing wall thickness the fatigue strength decreases. To estimate the influence of wall thickness more exactly, however, there are different approaches depending on the national standards and the type of construction. According to the current standards, the choice of the steel grade does not have an influence on the fatigue resistance. Now, the wall thickness can be reduced by the use of high strength steel. Particularly, constructions with many welds become more economical that way. This has to be taken into account also with regard to different methods of post weld treatment which can extend the remaining fatigue life of a construction considerably. Laboratory tests and model experiments are planned to estimate the influence of the parameters mentioned above on the fatigue strength. Different construction possibilities, varying dimensions, diverse kinds of post weld treatment and the use of high strength steel shall be taken into account. In parallel, the practical parameter studies are supported by extensive numerical calculations. A finite-element analysis will help to determine suitable surrounding conditions for the fatigue tests and to find a suitable test setup. The results of the fatigue tests and the finite-element analyses obtained will be presented. These results will enlarge the existing standardization and simplify the construction of offshore wind energy converters.

Literatur [1] R. Puthli, M. Veselcic, S. Herion and H. Huhn: Detaillösungen bei Erm¨ udungsfragen und dem Einsatz hochfester Stähle bei T¨ urmen von Offshore-Windenergieanlagen, Stahlbau, planend for June 2005, not published yet [2] Design of Offshore Wind Turbine Structures, Offshore Standard DNV-OS-J101, Det Norske Veritas, February 2004 [3] Guideline for the Certification of Offshore Wind Turbines, Germanischer Lloyd, Edition 2004

Correspondence: Jennifer Bergers – [email protected] – fax: +49 (0) 721 608-4078

POSTER

Session A – Poster 1

63

On The Atmospheric Flow Modelling Over Complex Relief I. Sládeka , K. Kozelb , Z. Jaˇ nourc a Czech Technical University in Prague, U12101, Karlovo n´ amˇest´ı 13, ZIP 121 35, Czech Republic b Czech Technical University in Prague, U12101, Karlovo n´ amˇest´ı 13, ZIP 121 35, Czech Republic c Institute of Thermo–mechanics, Czech Academy of Sciences, Dolejˇskova 5, ZIP 182 00, Prague, Czech Republic

The paper deals with a mathematical and numerical investigation of 3D–flow in the atmospheric boundary layer over complex topography. The concept of a wall functions is evaluated and the results are compared with a no–slip wall modelling. The turbulence is simulated using two different models: 1) simple algebraic one and 2) conventional k − ε model. Also the effect of computational grids on numerical results has been tested. The model is based on system of RANS equations which is modified according to the method of artificial compressibility. Then, it can be re-casted in the conservative and vector form as follows             v w u 0 0 0            2 p  ~ t +  u + %  +  2vu p  +  wu  =  Kux  +  Kuy  +  Kuz  W (1)  wv   Kvx   Kvy   uv   v +%   Kvz  w2 + %p Kwx x Kwy y Kwz z uw vw z x y ~ = (p/β 2 , u, v, w)T stands for the vector of unknown variables: the pressure p, the velocity vector where W ~ V = (u, v, w )T , β is related to the artificial sound speed and K is the turbulent diffusion coefficient, see [1]. The system (1) is solved in domain under stationary boundary conditions for the artificial time t → ∞ to obtain the expected steady-state solution. The finite volume method together with (3)–stage explicit Runge– Kutta scheme are applied. The method is also stabilized by the artificial diffusion term of fourth order. Hereafter, the computational domain is 43 km long, 35 km wide and 1 km high. The mean free stream velocity is U0 = 10 m/s, the roughness parameter z0 = 1 m, the power law velocity profile with exponent 2/9 is imposed at the inlet and neutral stratification is supposed as well. The model (1) closed by the algebraic turbulence model has been used to compute the velocity–pressure flow field over the topography, see figure .

Fig. 1: Left part shows the relief colored by geographical altitude in [m], the flow is oriented along the x-axis; Right part represents a near–ground cutplane colored by the u-velocity component in [m/s] together with stream–lines.

Literatur [1] Sládek I., Kozel K., Jaˇ nour Z., Gul´ıkov´ a E. (2004): On the Mathematical and Numerical Investigation of the Atmospheric Boundary Layer Flow With Pollution Dispersion, In: Urban Wind Engineering and Building Aerodynamics, VKI Brussels, May 2004, pp. C.9.1–C.9.10, ISBN 2-930389-11-7.

Ivo Sl´ adek, [email protected], Phone: (+420-2)24357247, Fax: (+420-2)24920677


64

Pollutant dispersion in flow around bluff bodies arrangement E. Moryn-Kucharczyka and R. Gnatowskaa a Institute of Thermal Machinery, Czestochowa University of Technology, 42-200 Czestochowa, Poland

The process of pollutant dispersion in flow around civil engineering structures depends strongly on the velocity field, responsible for convection and diffusion. To understand the phenomena related with the forming of concentration fields it is necessary to recognize the local features of the flow around the objects with the special emphasize for the mean velocity direction random fluctuations and periodical oscillations accompanying the vortex generation in bodies neighbourhood. The aim of the paper is to discuss the relations between the complex velocity field around the bluff-bodies arrangement and the polluted gas concentration in that area. The main attention has been put on the role of oscillating component of velocity as a factor stimulating the pollutant diffusion process. The analysis has been performed for the 2D case of two sharp-edged bluff bodies in tandem arrangement for which the strong level of oscillating component was available as the result of synchronization processes occurring in the flow surrounding the bodies. The mean concentration profiles of tracer gas (CO2) for various inter-obstacle gap were measured in wind tunnel flow. The local characteristics of flow were obtained using a commercial CFD code FLUENT 6.0. The discussion is partially based on the results of the previous data contained in works of Jarza & Gnatowska [1, 2], which revealed different flow regimes and critical body spacing at which dynamic properties of vortex structures around bluff-bodies system alter rapidly reflecting the stability mode change. The specific conditions generated around bluff bodies arrangement mode is possible to study the gas pollutant dispersion for flow field, in which the main features of built environment are present, namely wind vorticity, high level oscillations, separation, reattachment etc.

Literatur [1] A. Jar˙za, R. Gnatowska: Lock-on effect on unsteady loading of rigid bluff – body in tandem arrangement, Proceedings Of International Conference Urban Wind Engineering And Buildings Aerodynamics Cost C14, Rhode-St-Genese Belgium, 5-7 may, pp.c.2.1-10. [2] A. Jar˙za, R. Gnatowska: Simulation of lock-on phenomena in systems of bluff-bodies, Proceedings of 12th international conference on fluid flow technologies, Budapest Hungary, 3-6 september, pp.: 375-380.

Correspondence: Moryn-Kucharczyk – [email protected] – fax: +48 (34) 3250555


65

The Improved Mesoscale Turbulence Model for Wind Climate and Pollutant Dispersion in Cities A. F. Kurbatskiyaa , L. I. Kurbatskayabb a Department of Physics, Novosibirsk State University, Pirogova Str., 2, Novosibirsk, Russia b Institute of Computationsl Mathematics and Mathematical Geophysics SB RAS Lavrentieva Ave., 6, Novosibirsk, Russia

A computational scheme for an improved Mellor-Yamada Level-3 model is proposed and its performance is examined to simulate the wind field structure above the urbanized surface (city) in the atmospheric boundary layer (ABL). In the present model, two new ingredients are employed: 1) an updated expression for the pressurevelocity correlation, 2) an updated expression for the pressure-temperature correlation. The turbulent momentum and scalar fluxes are determined by the full explicit algebraic expressions which are deduced from the closed transport equations for turbulent fluxes and simplified using the weak-equilibrium assumption. Closure is achieved by solving the evolution equations for the turbulent kinetic energy, its dissipation rate and scalar variance (the three-parametric turbulence model). This improved mesoscale model for the turbulent ABL is capable to reproduce the most important features of a wind field above the city. Simulation results show, that the thermal circulation caused by a longitudinal temperature gradient between heated up air above city and less heated up air of its vicinities, increases speed of wind aloft on a leeside, and the term of the longitudinal turbulent diffusion in the potential temperature equation act to reduce the daytime boundary layer height. Introduction It is well-known that the elevated temperatures in cites arise due to anthropogenic heat generation and retardation of the nocturnal cooling of urban surfaces in contrast to rural surfaces cause higher temperatures over the city relative to its vicinity (the Urban Heat Island effect; e. g., [1]). The urban heat island effect may produce major temporal and spatial alterations to the circulation of the urban boundary layer [1]. This study attempts to formulate a numerical model for the simulation of the urban heat island effect and its influence on the modification of background (synoptic and mesoscale) winds by cites. In this model, the transfer of momentum and scalar (temperature, concentration) in the ABL is computed by the improved Mellor-Yamada Level-3 model.

Literatur [1] Bornstein, R. D. and Oke T. R. Influence of pollution on urban climatology, Adv. Environ. Sci. Engrg., vol. 2, pp. 171-202. [2] Kurbatskii A. F., computational modelling of the penetrative convection above the urban heat island in a stably stratified environment, J. Appl. Meteor., vol. 40, No. 10, pp. 1748-1761.

Kurbatskiy A. F. – [email protected] – fax: +7 (383) 330-72-68


66

Comparison of logarithmic wind profiles and power law wind profiles and their applicability for offshore wind profiles S. Emeis

a

a Forschungszentrum Karlsruhe, Kreuzeckbahnstr. 19, 82467 Garmisch-Partenkirchen, Germany

Two types of wind profile laws are frequently used for vertical extrapolation of wind speeds in the atmospheric surface-layer for wind energy purposes: the theoretically derived logarithmic profile with corrections for nonneutral thermal stratification and the empirically derived power law. Due to its mathematical simplicity the power law is widely used. The first part of this study investigates in which situations the power law is a good approximation to the logarithmic profile. In extension to existing studies it is stipulated that not only the slope of the logarithmic and the power law profiles should coincide in a selected height but simultaneously also the curvatures of the two profiles should coincide in this height in order to give a good fit over a wider height range. The surface roughness and thermal stratification conditions for which such a simultaneous coincidence is possible are calculated analytically. For neutral and unstable conditions slope and curvature of a power law profile cannot coincide simultaneously with that of the logarithmic profile. This can only happen under certain circumstances in a stably stratified flow. The practical result of this study is that the power law offers a good fit to the logarithmic profile over a wider height range for slightly stable conditions and for very smooth surfaces only. Thus the power law profile provides a good description of the offshore vertical wind profile but is not suitable for rough terrain.

Fig. 1: Logarithmic wind profiles for non-neutral stratification (thin) and their approximation by power law profiles (dashed). The middle pair of profiles is shifted by 0.5, the rightmost pair by 1.0 to the right. The two numbers at the top give z/z0 (z0 is roughness length in m) and z/L∗ (giving the static stability: negative means unstable, zero means neutral, positive means stable), the number in the middle the exponent n of the power law. Whether these profile laws are suitable at all for offshore application has to be clarified by analysing the data from the FINO1-offshore platform in the OWID project (funded by the German Ministry of the Environment, BMU). Wind and turbulence profiles from this dataset are currently processed at our lab. It is expected - at least for stable and neutral conditions - that the surface layer (for which the logarithmic law and the power law are valid only) is only a few tens of meters deep, due to the smooth sea surface. Therefore most of the offshore wind turbines will operate in the Ekman layer that is situated ontop of the surface layer. Here the vertical wind speed gradient is usually smaller than in the surface layer but the wind direction turns with height.

Correspondence: Stefan Emeis – [email protected] – fax: +49 (0) 8821 73 5 73 – tel: +49 (0) 8821 183 240

Session B – Poster 1

67

Superposition Model for Atmospheric Turbulence S. Bartha , F. Böttchera , J. Peinkea a ForWind - Center for Wind Energy Research, University of Oldenburg, Oldenburg, Germany

In this work we focus on the scale dependent statistics of atmospheric velocity increments measured at different on- and offshore locations and compare them to that of homogeneous, isotropic and stationary turbulence as realized in laboratory experiments. For isotropic turbulence the statistical moments of increments, the so-called structure functions have been intensively studied. Their functional dependence on the scale τ is described by a variety of multifractal models. Alternatively to the analysis of structure functions, probability density functions (PDFs) are often considered. For turbulent laboratory flows they show a change of shapes. For large scales the distributions are Gaussian while for small scales they are found to be intermittent. The here examined atmospheric PDFs differ from those of turbulent laboratory flows. They show a markable feature of a shape which is nearly independent of the scale. This corresponds to the striking feature of intermittent pdf shapes (heavy tailed and not Gaussian) but perform for the structure functions non multifractal K41 scaling. Speaking mathematically such constant shapes for larger and larger scales is expected only for stable distributions such as Gaussian ones or the Lévy stable laws. Although the decay of the tails indicates that distributions should approach Gaussian ones (as for isotropic turbulence) they show a rather robust exponential-like decay. The challenge is to describe and to explain the measured fat-tailed distributions and the corresponding nonconvergence to Gaussian statistics. This is of great interest because increment values in the tails directly correspond to an increased probability (risk) to observe large and very large events (gusts). We introduce a model that interprets atmospheric increment statistics as a large scale mixture of subsets of isotropic statistics [1]. When mixing is weak the same statistics as for isotropic turbulence is recovered while for strong mixing robust intermittency is obtained.

Literatur [1] F. Böttcher, St. Barth and J. Peinke. ’Small and Large Scale Fluctuations in Atmospheric Wind Speeds’, Stoch. Environ. Res. Risk Assess. (accepted), 2005.

Correspondence: Stephan Barth – [email protected] – fax: +49 (0) 441 798-3579


68

Extreme events under low-frequency wind speed variability and wind energy generation A.A. Cârsteanu

a

and J.J. Castro

b

a Mathematics Department, Cinvestav, Av. IPN 2508, México D.F. 07360, México b Physics Department, Cinvestav, Av. IPN 2508, México D.F. 07360, México

Low-frequency wind speed variability represents an important challenge to statistical estimation for atmospheric turbulence and its impact on wind energy generation, given that it does not allow for the usual stationary assumption in time series analysis. This work presents a framework for the parameterization of a state-space representation of turbulent processes, where low-frequency variability would be resolved by computing conditional probabilities of visitation for the different states. The conditional probabilities are extrapolated for the purpose of obtaining the probabilities of extreme events for wind speed variation and its application in wind energy.

Correspondence: A.A. Cˆ arstenau – [email protected] – fax: +52-55-5061-3876


69

Stochastic modelling and prognosis of turbulent wind time series J. Clevea , M. Greinera a Corporate Technology, Information & Communications, Siemens AG, D-81730 M¨ unchen, Germany

A precise understanding and modelling of small-scale turbulent wind fields is important for the design and operation of modern wind power plants. For example, the design of rotor blades requires detailed knowledge of the wind field and its statistical properties. The control of single wind turbines as well as whole wind farms require the predictive modelling of the wind field. We will present a stochastic model for the small-scale turbulent wind field. The model is based on a multifractal extension of fractional Brownian motion. It provides a consistent description of the statistics of the velocity field as well as the energy dissipation field. Although the modelling of the turbulent wind field statistics already provides some predictive power in itself, superior for on-line control is a direct prognosis of the time series. There are many tools which can perform this task. Among them, artificial neural networks (ANN) are the most promising candidates. Ultra short-term predictions for turbulent wind time series based on state-of-the-art ANN architectures are presented.

Correspondence: Jochen Cleve – jochen [email protected] – fax: +49 (0) 89 636-49767


70

Quantitative Reconstruction of Drift and Diffusion Functions from Time Series Data D. Kleinhansa , R. Friedricha a Westf¨ alische Wilhelms-Universit¨ at, Institut f¨ ur Theoretische Physik, Wilhelm-Klemm-Straße 9, 48149 M¨ unster

Stochastic analysis of wind measurements is of increasing interest in wind energy research for example for the description of the power output of wind turbines [1]. Such processes in many cases can be characterized by their drift and diffusion functions. We discuss the extraction of drift and diffusion functions from measured data. In the common procedure [2] deviations enter when performing the required limit of small time increments. Additionally statistical errors complicate the interpretation of the results. The impact of such sources of error is estimated. Furthermore an extension to this method is presented that circumvents the limiting procedure. It is based on the minimization of the Kullback-Leibler distance between the joined probability density functions of the measured data and simulated stochastic processes. By this means quantitatively accurate results can be obtained even for sparsely sampled measurements.

Literatur [1] E. Anahua, M. Lange, F. B¨ ottcher, S. Barth and J. Peinke. Stochastic Analysis of the Power Output for a Wind Turbine, Proceedings of the European Wind Energy Conference (EWEC), London, UK, 2004. [2] S. Siegert, R. Friedrich and J. Peinke. Analysis of datasets of stochastic systems, Physics Letters A 243, 1998.

Correspondence: D. Kleinhans – email: [email protected]


71

Scaling turbulent atmospheric stratification: a turbulence/wave wind model S. Lovejoya , D. Schertzerb a Physics, McGill University, 3600 University st., Montreal, Que., Canada b CEREVE, ENPC, 6-8, ave. Blaise Pascal, Cité Descartes 77455 Marne-la-vallee Cedex, France

We critically re-examine empirical vertical and horizontal statistics of the horizontal wind and find that the −5/3 balance of evidence is in favour of the Kolmogorov kx scaling in the horizontal, Bolgiano-Obukov scaling −11/5 kz in the vertical corresponding to a Ds = 23/9D stratified atmosphere. This interpretation is particularly compelling once one recognizes that the 23/9D turbulence can lead to long range biases in aircraft trajectories and hence wind, tempertature and other statistics. Indeed, we show quantitatively that one is able to reinterpret the major aircraft-based campaigns (GASP, MOZAIC) in terms of the model. In part I we have seen that this model is compatible with turbulence waves which can be close to classical linear gravity waves in spite of their quite different mechanism. We then use state-of-the-art lidar data of atmospheric aerosols (considered a passive tracer) in order to obtain direct estimates of the effective dimension of the atmosphere, the elliptical dimension, Del = 23/9 = 2.55 ± 0.02. This result essentially rules out the standard 3D or 2D isotropic theories which have D = 3, 2 respectively. In this paper we focus on the multifractal (intermittency) statistics showing that to within experimental uncertainty, the high and low order statistics are indeed stratified in the same way as the statistics near the mean field. We critically re-examine theories of turbulence in a stratified atmosphere arguing that they should be buoyancydriven rather than energy driven and that they should be scaling but anisotropic, not isotropic. We compare the leading statistical theories of atmospheric stratification: the 7/3D linear gravity theories and the classical fractionally integrated flux (FIF) 23/9D unified scaling model in which the horizontal wind has a k −5/3 spectrum as a function of horizontal wavenumber determined by the energy flux and a k –11/5 energy spectrum as a function of vertical wavenumber determined by the buoyancy force variance flux. We argue that although the 23/9D FIF model is more physically satisfying - being based on turbulent fluxes – that the key weakness of the classical version of the model is that it’s structures are too localized, it displays no wave-like phenomenology. We show how to extend the 23/9D FIF model to account for more realistic wave-like structures. The basic idea is that the FIF model – which relates turbulent fluxes to observables such as v, ρ – can be readily generalized to include the wave effects. The key point is that the FIF requires two propagators (space-time Green’s functions). The first determines the space-time structure of the cascade of fluxes, this must be localized in space-time in order to satisfy the usual turbulence phenomenology. In contrast, the second propagator relates the turbulent fluxes to the observables, this propagator can be unlocalized in space-time (although the spatial part is the same as before, it is still localized in space, now in wave packets). We display numerical simulations (fig. 1) which demonstrate the requisite (anisotropic, multifractal) statistical properties as well as wave-like phenomologies.

Fig. 1: A cloud rendition of a simulation of a horizontal cross-section of a component of the horizontal wind (with statistics very close to a passive scalar). The statistics obey anisotropic generalizations of the Kolmogorov law, and have mulitfractal intermittency characterized by universal mulitfractal parameters α = 1.8, C1 = 0.08. Note the wave-like structures. Correspondence: Shaun Lovejoy – [email protected] – fax: 1-514 398-8434

Session C – Poster 1

72

Characterization of the Power Curve for Wind Turbine by Stochastic Modelling E. Anahuaa , J. Peinkea , and S. Bartha a ForWind - Center for Wind Energy Research, University of Oldenburg, Oldenburg, Germany

We investigate how the power curve of a wind turbine is affected by turbulent wind fields. The electric power output can be separated into two parts, namely the relaxation part which describes the dynamic response of the wind turbine on sudden changes in wind velocity and a noise part. We have shown that those two parts describe the power curve properly [1] if calculated from stationary wind measurements. This analysis is very useful to describe power curve characteristics for situations with increased turbulent intensities and it can be easily applied to measured data.

Literatur [1] E.Anahua, F. Boettcher, S. Barth, J. Peinke. ”Stochastic analysis of the Power Output for a Wind Turbine”. Proc. of Wind Energy Conference, European Wind Energy Asociation, London 2004.

Correspondence: J. Peinke – [email protected] – fax: +49 (0) 441 798-3990


73

Handling systems driven by different noise sources F. Böttchera , J. Peinkea , D. Kleinhansb & R. Friedrichb a Institute of Physics, University of Oldenburg, Carl-von-Ossietzky-Str. 9-11, D-26111 Oldenburg, Germany b Institute of Theoretical Physics, University of M¨ unster, Wilhelm-Klemm-Str. 9, D-48149 M¨ unster, Germany

A frequent challenge for wind energy applications is to grasp the impacts of turbulent wind speeds properly. Here we present a general new approach to analyze time series – interpreted as a realization of complex dynamical systems – which are spoiled by the simultaneous presence of dynamical noise and measurement noise. It is shown that such noise implications can be quantified solely on the basis of measured times series. It is argued that a profound understanding of the influence of measurement noise is needed for a proper description of the response of a dynamical system to the noise. To this end we show how the conditional moments M (n) can be defined and how they have to be interpreted in order to reconstruct the system characteristics. On the basis of several examples we show that from our proposed method, the underlying deterministic response characteristics as well as the magnitude of dynamical noise and measurement noise can be extracted. This will be illustrated for a simplified dynamical response model of a wind turbine.

Correspondence: Frank B¨ ottcher – [email protected] – fax: +49 (0)441 798 3990


74

Experimental Researches of Characteristics of Windrotor Models with Vertical Axis of Rotation S.A. Dovgy, V.P. Kayan and V.A. Kochin

a

a Department of the informational systems in fluid mechanics and ecologies Institute of Hydromechanics NASU, 8/4 Zelyabov str., Kyiv-57, Ukraine

Windpowers with vertical axis of rotation (VAR) of windrotor (like the Darye’s rotor) have several advantages in comparison with widespread windrotors with horizontal axis of rotation, and have attracted increasing attention of researchers and designers for last two decades. The basic deficiency of such windrotors is high speed of wind stream at which the rotor starts in rotation. Control of the windrotor blades in a circular trajectory of their movement allows to lower the magnitude of speed, and also to improve performance of the windrotor. In the hydrotray of IHM NASU, experimental researches were performed for several windrotor models with VAR with both a rigid fixing of blades and the mechanism controlling the position of blades. The blade profile was axisymmetric of the type NACA-0015. With the rigid fixing of blades on cross-pieces of windrotor, the slope of the blade profile chord to the tangent to the circle of rotation was +4◦ . The mechanism controlling the position of blades during rotation of the windrotor model allowed varying angles of incidence of blades from −14◦ to +25◦ . Recording and processing of experimental data, their graphical representation and storing, taking of measurements protocols were carried out by the measuring system created on the basis of 12-digit analog-to- digital converter installed as an expansion card to IBM-compatible computer. Experiments were done in the hydrotray with 0.6 × 0.6m2 throat area, flow speeds varied from 0.3 m/s to 0.65 m/s. Three windrotor models (diameter of blades installation D=0.175 and 0.35 m, blade length lb=0.15 and 0.3 m) were investigated. In the experiments were determined the flow speed “V” in front of the windrotor model and the promptness “n”, at given the loading moment “M” on the shaft of model. Results of the research are presented graphically in the form of dependencies of operating ratios of the flow energy Cp and coefficients of the twisting moment Cm of windrotor models on the coefficient of specific speed Z. All the windrotor models with controlled blades were able to self-start at essentially lower flow speed than models with fixed blades. It was revealed, that increased number of blades of the rotor with controlled blades leads to increasing specific speed of the rotor that is opposite to the case of the rotor with rigid fixed blades.

Fig. 1: Characteristics of the windrotor model N2 with controlled (1o - 2o) and rigid fixed (1f - 2f ) blades.

For windrotor models with controlled blades, the magnitude of twisting moment coefficient Cm has appeared 2.5-3 times higher than that for similar models with rigid fixed blades, while the excess in magnitude of operating ratio of flow energy Cp is obvious, but insignificant. Very high twisting moment coefficient Cm and low speed of self-start allow expecting that windpowers with rotors of the described design will be rather effective as pumping units for oil or water production. V.P. Kayan


75

Methodical failure detection in grid connected wind parks D. Schulza , K. Knorrb , R. Hanitschb a University of Applied Sciences/ Competence Center Wind Energy, An der Karlstadt 8, Bremerhaven, Germany b Technical University Berlin, Einsteinufer 11, Berlin, Germany

This contribution deals with the handling of very common failures and the discussion of outages of wind energy converters (WECs) with doubly-fed asynchronous generators located in wind parks. Almost all failure cases were followed by discussions about the error cause and the resulting responsibility due to the high cost of manpower and cost of necessary measurements. One part of the discussion is the sensitivity of the devices against external over- and undervoltage, transients and voltage dips. These tolerances are given in EN standards and IEC-guidelines [1, 2]. In the described wind park with eight WECs a lot of outages cumulated over some months. The results were: loss of energy yield and additional cost for failure identification. Main question was the determination of the failure source as an external or device-internal event. Fig. 1 shows a failure-time-circle with the measured tripping events of the eight devices over a time span of five months. With this long-term failure monitoring a methodical evaluation of the failure sources and a decision about the counteractive measures were possible. The final contribution shows, that also for synchronous tripping of some devices the external effects may only trigger an internal device problem [3]. A detailed description of the measurement devices, the applied systematic measurement plan and detailed results of the monitored electrical values are presented in the full paper. This includes an overview of the possible external and internal distortion sources.

Fig. 1: Chronological allocation of WEC-outages of eight systems in the wind park over five months

Literatur [1] EN 50160: Voltage Characteristics of electricity supplied by public distribution systems. March 2000. [2] IEC 61000-2-12: Electromagnetic compatibility (EMC) - Part 2-12: Environment - Compatibility levels for low-frequency conducted disturbances and signalling in public medium-voltage power supply systems. 04/2003 . [3] Plotkin, Y.; Saniter, C.; Schulz, D.; Hanitsch, R. (2005): Transients in doubly-fed induction machines due to supply voltage sags. PCIM Power Quality Conference, Nuremberg, pp. 342-345. Correspondence: Detlef Schulz – [email protected] – fax: +49 (0471) 4823-341

Session D – Poster 1

76

Performance of the Risø-B1 Airfoil Family For Wind Turbines Christian Baka , Mac Gaunaaa and Ioannis Antonioua a Wind Energy Department, Risoe National Laboratory, P.O.Box 49, DK-4000 Roskilde, Denmark

It has been common to choose airfoil sections for wind turbine blades that originally have been developed for airplanes. These airfoils have shown to perform well and their performances have been known both from wind tunnels measurements and from their use on wind turbines. However, traditional airfoils are not designed for operation with roughness at the leading edge, with high relative thickness and they are not tailored for a particular type of wind turbine power regulation. Risø has developed the Risø-B1 airfoil family with a relative thickness range from 15% to 53%. The family is designed for large variable speed pitch controlled wind turbines . The design was carried out using an in-house method based on numerical optimization. The airfoils are designed for high maximum lift and high driving force in a broad angle-of-attack range before maximum lift. They are also designed to be insensitive to leading edge roughness. Efficient thick airfoils combined with roughness insensitivity can potentially result in more slender blades while maintaining energy capture. This paper presents the verification of the performance of the 15%, 18%, 24% and 27% Risø-B1 airfoils. They were tested in the VELUX wind tunnel, which is of the closed return type with an open quadratic jet of side length 3.4m. The four airfoil-sections of chord length 0.6m were tested at a Reynolds number of 1.6 million with 62 pressure taps along the airfoil surface. In the chord direction the surface pressure was measured at angles of attack between –8◦ and 32◦ . To measure the drag of the airfoil, the velocity deficit was measured in the wake using a wake rake with pressure tubes. Lift, drag and moment were found by integrating the measured pressures. The airfoils were tested both with and without aerodynamic devices typical for wind turbine blades such as stall strips, vortex generators and Gurney flaps. Furthermore, their roughness sensitivity was tested using different kinds of adhesive tape simulating roughness applied around the airfoil leading edge. The measurements showed good agreement with the predicted performance. The airfoils without aerodynamic devices showed the desired maximum lift of 1.74, 1.64, 1.62 and 1.54 for the 15%, 18%, 24 and 27% airfoil, respectively. For a MW-size wind turbine corresponding to a Reynolds number of 6 million, CFD computations showed that the maximum lift would be somewhat higher. Furthermore, the airfoils showed high insensitivity to leading edge roughness. Vortex generators and Gurney flaps on the 24% and 27% airfoils increase the maximum lift significantly. Compared to the commonly used NACA and FFA airfoils the maximum lift-drag ratio for the airfoils was higher.

Correspondence: Christian Bak – [email protected] – phone: +45 4677 5091 – fax: +45 4677 5960


77

Dynamic Lift of Airfoils T. Bohlena , R. Gr¨ unebergerb , J. Peinkec a Faculty of Physics, Carl Ossietzky University of Oldenburg, D-26111 Oldenburg, Germany b Faculty of Physics, Carl Ossietzky University of Oldenburg, D-26111 Oldenburg, Germany c Faculty of Physics, Carl Ossietzky University of Oldenburg, D-26111 Oldenburg, Germany

We present preliminary measurements of dynamical lift by rotating an airfoil in a wind tunnel experiment. In contrast to statical lift values, there is an increase of lift before flow separation on the suction side occurs. The magnitude of lift depends on the speed of changing the pitch angle of the airfoil. The measurements correspond to fluctuations of the wind direction in turbulent winds, which lead to extreme mechanical loads for blades of wind turbines. In order to quantify this effect, the lift is calculated through the integral of pressure distribution at the wind tunnel walls, while rotating the airfoil with defined angular velocity.

Correspondence: Thomas Bohlen – [email protected] – fax: +49 (0)441 798-3579


78

Aerodynamic behaviour of a new type of slow running vertical axis wind turbine J.-L. Menet

a

a Ecole Nationale Supérieure d’Ingénieurs en Informatique Automatique Mécanique Energ´ ´ ´ etique Electronique de Valenciennes (ENSIAME) - Université de Valenciennes - Le Mont Houy F-59313 Valenciennes Cedex 9

The Savonius rotor (Fig. 1) is a slow running vertical axis wind turbine, the advantages of which are numerous [1, 2]; however, it has a poor aerodynamic efficiency. We present a study aiming to raise this efficiency by adjusting several geometrical parameters, in particular the overlap of the paddles and their respective position.

A parametric study [3] has been continued by a bidimensional numerical simulation, using the CFD code Fluent v6.0. First the numerical model is validated on the conventional Savonius rotor. Then the geometry of an optimised Savonius rotor is proposed, the overlap ratio of which is 0.242 [4]. Last a different positioning of the paddles (Fig. 2) leads to an optimal paddle angle β of about 55◦ , corresponding to the maximum of the mean starting torque coefficient. The flow around the rotor can be calculated, so that the pressure distribution (Fig. 3) and consequently the torque coefficients (Fig. 4) and can be predicted on the new rotor.

Literatur [1] Ushiyama I., Nagai H. : Optimum design configurations and performances of Savonius rotors. Wind Eng. 1988 ; 12-1 : 59-75. [2] Menet J.-L., Valdès L.-C., Ménart B. : A comparative calculation of the wind turbines capacities on the basis of the L-σ criterion. Renewable Energy 2001 ; 22 : 491-506. [3] Menet J.-L., Bourabaa N. : Increase in the Savonius rotors efficiency via a parametric investigation. 2004 European Wind Energy Conference, Proceedings, London, 22-24 november 2004. [4] Menet J.-L., Leiper A. : Prévision des performances aérodynamiques d’un nouveau type d’éolienne à axe vertical dérivé du rotor Savonius, 17è Congrès Fran¸cais de Mécanique, Proceedings, Troyes, 2005. Correspondence: Jean-Luc MENET – [email protected] – fax: (33) 03 27 51 12 25


79

Simulating Dynamic Stall using Spectral CFD-Code B. Stoevesandta , A. Shishkinb , C. Wagnerb , J. Peinkea a Forwind, Marie-Curie-Str. 1, Oldenburg, Germany b Institute of Fluid Mechanics, DLR G¨ ottingen, Bunsenstr. 10, 37073 G¨ ottingen, Germany

Reduction of weight of wind turbine blades is of large interest to manufacturers. To do so without losing the necessary stability, loads on blades caused by dynamic stall effects have been a major task in blade-research during the last years. The aim of the project is to calculate loads and drag caused by the effect of dynamic stall as they arise in real turbulent wind fields. This is done by using wind tunnel measurments and CFD simulations. A spectral code has been chosen to achieve high accuracy[1]. Nevertheless the code appears to be very sensitive to the choice of boundary conditions. Due to the dynamic stall the boundary conditions have to be flexible. Thus the main focus of the project lies at the time on the reliability of the CFD code.

Literatur [1] Karniadakis, G. E. and Sherwin, S. J. (1999): Spectral/HP Element Methods for CFD Oxford University Press, Oxford, 1999

Correspondence: Bernhard Stoevesandt – [email protected] – fax: +49 (0) 441 798-3579

Session E – Poster 1

80

Comparing linear and non linear wind flow models A. Iniesta a , D. Cabezón a , I. Mart´ı a a Wind Energy Department, National Renewable Energy Centre (CENER) C/Ciudad de la Innovaci´ on, 31621 Sarriguren, Navarra SPAIN

SUMMARY Traditional linear techniques for wind flow modelling have traditionally been accepted for moderate flat sites with moderate uncertainties for wind field modelling. Nevertheless, as terrain gets more complicated linear models begin to fail, being necessary to increase the accuracy of the wind field simulations. That is why the wind energy market is increasingly demanding more sophisticated methods based on non linear tools which reduce the uncertainties to be avoided by the developers. The results of the comparison between CFD models versus linear models are presented, showing the improvements of the CFD in the wind field simulations. FULL DESCRIPTION This paper compares and validates a group of four linear and non linear wind field simulation models in order to analyse their behaviour in the same wind farm located in complex terrain. Selected linear models – WasP and WasP Engineering – have been widely used for wind resource assessment as traditional tools reliable for particularly flat sites with very satisfactory results on them. In contrast, selected non linear models – WindSim and Fluent – are CFD solvers which take into account effects that linear models are not able to capture,specially in complex terrain. The difference of the CFD models is based on the resolution of the complete Navier Stokes equations. Wind speed fields are be compared for the cases of WAsP, WindSim and Fluent whereas turbulence intensity calculations are compared forWindSim, Fluent and WAsP Engineering. The results are be validated for the prevailing direction sectors with some meteorological masts installed in the site situated in Navarra (north of Spain) with an average elevation of 1050 meters above sea level and an average RIX (ruggedness index) of 16%. Although computing requirements are higher, non linear models are promising tools in complex topography contexts for reducing wind speed and turbulence intensity uncertainties. The increase of the wind field simulations accuracy helps to carry out the wind farm design with lower risk.

Correspondence: A. Iniesta – : [email protected] – fax: +34 948270774


81

Large eddy simulation of an aeroturbine wake ´ Angel Jiminéza , Antonio Crespoa , Emilio Migoyaa and Javier Garc´ıaa a Fluid Mechanics Laboratory, ETSII, UPM

In industrial or enviromental applications, where Reynolds numbers are usually very high, direct-numerical simulations (DNS) of turbulence are generally impossible, because the very wide range that exists between the largest and smallest disipative scales cannot be explicitly simuled even in the largest and most powerful computers. A large-eddy simulation (LES) approach is conceptually more suitable for studying the turbulence evolution in an aeroturbine wake, because the larger scales of turbulent motion, those that control turbulent diffusion of momentum, are computed explicitly, whereas only the effect of the small-scale motion that tends to be more isotropic and disipative has to be modeled. A LES model with simplified wind boundary conditions has been implemented in a CFD code to simulate and characterize the turbulence generated by the presence of an aeroturbine. In this work we present the results obtained by this new aproach as well as a comparison with the results provided by the code UPMWAKE based on an explicit algebraic model for the components of the stress tensor, and with experimental data from the Sexiberium wind farm. Also comparison is made with correlations proposed by the authors in previous works.

Correspondence: – fax:


82

Infrasound in Wind Energy G. Sokola a Department of technical Mechanic, Physical and Technical Institute, Dniepropetrovsky National University Nauchnaya, St., 13, Dniepropetrovsk, 49050, Ukraine

The fact is that acoustic waves are harmful for the service staff in industry, in Wind Energy and in transport some times. The standards for security of people working under the influence of low frequency acoustic waves (LFAW) have been worked out and adopted. In accordance with those standards even at the sound pressure level of 100 dB time of staying people in the low frequency acoustic waves spreading zone should be limited. Review of the scientific works in the inadequate research area of the acoustics in particular: in the infrasound oscillations is presented. It was analyses 803 denominations that have been published for more than 20 years. The review came into being as the monography “The peculiarities of the acoustic processes within the infrasound frequency range”. In ten chapters of the book the author describes the works about infrasound in nature; peculiarities of its spreading; influence of infrasound upon the biological beings; infrasound noise of wind equipment and vibrations; construction of generators and of significance horn in them; the methods and equipment for measurements of infrasound signals; the area of technological application of infrasound generators working in liquids and gases. One of the ways of LFAW generating is by interaction of powerful permanent airflow with surface in the state of oscillatory movement has been already discovered. Another way of LFAW generating is on the basis of vortex sound phenomenon the nature of which is connected with vortexes formation in the air flux during flowing round the obstacles and is already known. There are many works about LFAW produced by different types of machines with rotating parts and units: turbines, derricks, building machines, construction equipment, turbo-prop-airplanes and helicopter screws. In atmosphere low frequency acoustic waves are generated by a flux flowing round a very rough surface, in particular around mountain masses. Powerful single discharges or an explosion can also generate infrasound. In such a case LFAW are generated as one or some harmonics in the common spectrum of noises. LFAW are generated in the atmosphere by the bodies moving with high velocities such as meteors, high-speed trains, underground electric trains. The sound pressure increasing was registered in tunnels. A LFAW of 4 Hz was registered during rocket starts. On reaching supersonic barrier by an aircraft, blasts appear. LFAW heat sources of significant interest are engines of airplanes and rockets. Energy peak is observed at 0.1-2 Hz and 16-18 Hz frequency range. It is possibel the conclusions make, that infrasound is the problem in the acoustic of the Wind Energy. [1] G. Sokol. Some Aspects in Low Frequency Acoustic Processes (200). [2] G. Sokol. Infrasound Review, Annual Scientific Conference GAMM 2001, EHT Z¨ urich February 12-15, 2001. – Z¨ urich:, 2001 - p. 136.

Correspondence: G. Sokol


83

Prediction of wind turbine noise generation and propagation based on an acoustic analogy - Abstract D. Moroianua , L. Fuchsb a Division of Fluid Mechanics, Department of Heat and Power Engineering, Lund Institute of Technology, Ole R¨ omersvag 1 P.O. Box 118, 22100 Lund, Sweden b Division of Fluid Mechanics, Department of Heat and Power Engineering, Lund Institute of Technology, Ole R¨ omersvag 1 P.O. Box 118, 22100 Lund, Sweden

The aerodynamical noise produced by a wind turbine is computed using a hybrid method: LES and an acoustical analogy known in literature as the Ffowcs Williams and Hawkings equation[1]. We assume that the acoustical field can be separated from the flow and the method involves two steps[2, 3, 4]. First is the computation of the flow field from which the acoustical sources are extracted. The next step comprises the propagation of the noise. The flow around of a three dimensional, three blade wind turbine is computed by means of Large Eddy Simulation. The dynamics of tip and hub vortices together with some other coherent structures that appear in the back of the wind turbine mast (see Fig. a) are accounted as potential noise sources.

a)

b)

Fig. 1: a) Tip and hub vortices; b) Cross sectional wave reflections by the ground at y = 2.5D, full geometry

Once the acoustical sources are extracted from the flow field, the propagation of noise is computed with an acoustic analogy. Preliminary results of the acoustical density fluctuation are plotted in Fig. b. In order to study the influence of neighboring turbines, the noise sources can be duplicated in the acoustical domain. The paper will present details of the current approach of the problem. Different characteristics of the flow (the content of turbulent kinetic energy spectrum in different monitoring points, vortex visualizations) and also different features of the acoustical field (acoustical spectrum in different monitoring points, sound pressure level, wave interaction between two wind turbines) will be included too.

Literatur [1] Brentner, K. S., Farassat, F., Modeling Aerodynamically Generated Sound of Helicopter Rotors”, Progress in Aerospace Sciences 39, pp. 83-120, 2003. ¨ Sound Generated Aerodynamically: I. General Theory”, Proceedings of the Royal [2] Lighthill, M. J., On Society of London, Series A, Vol. 211, pp. 564-587, 1952. ¨ Sound Generated Aerodynamically: II. Turbulence as a Source of Sound”, Proceedings [3] Lighthill, M. J., On of the Royal Society of London, Series A, Vol. 222, pp. 1-32, 1954. [4] Lighthill, M. J., ”Waves in Fluids”, Cambridge University Press, ISBN 0-521-29233-6, 1978. Correspondence: Dragos Moroianu – [email protected] – fax: +46 - 46 - 222 47 17

Session F – Poster 1

84

On the Influence of Low-Level Jets on Energy Production and Loading of Wind Turbines N. Cosack a , S. Emeis

a

and Martin K¨ uhn

a

a Endowed Chair of Wind Energy, Institute of Aircraft Design, University of Stuttgart, Allmandring 5b, 70569 Stuttgart, Germany b Institute of Meteorology and Climatic Research, Forschungszentrum Karlsruhe GmbH, Kreuzeckbahnstr. 19 , 82467 Garmisch-Partenkirchen, Germany

Size and height of modern wind turbines have increased rapidly during the last years, resulting in recently erected prototypes of 5MW power output and hub heights of more than 120m. This raises the question, whether specific meteorological situations, that are usually neglected for load and power prediction are of more importance for these large multi-megawatt turbines. The presented work investigates the influence of low-level jets on energy production and fatigue loading of wind turbines. Following measurements in the northern part of Germany[1], analytical equations for approximating wind conditions in the case of a stable boundary layer have been derived. The equations account for the height dependency of wind speed and of wind direction. Based on [1], the input to the industrial wind turbine simulation software Flex5 has been modified in order to reflect the measured conditions. Simulations have been carried out to estimate turbine fatigue loading and performance using deterministic as well as turbulent wind fields. To capture the influence of turbine size and height, results have been derived for two different turbine models: a 1.5MW turbine with hub height of 85m and a rotor diameter of 70m and a scaled version of 5MW with 130m hub height and rotor diameter of 128m. The results show that low-level jets can have a significant short-term effect on wind turbines behaviour, when compared to ”normal” situations. This holds true, even in the case of the smaller 1.5MW turbine. As expected, the consequences are larger for the 5MW turbine. The influence on life-time fatigue loading and annual energy yield depends directly on the site-specific frequency of occurrence and severity of such events. Therefore it is much more difficult to provide an overall judgement. Consequently the investigations have been concentrated on comparison of short-term statistics.

Fig. 1: Time-height-section of mean wind speed (black: 0 m/s, magenta: 10 m/s) showing a low-level-jet in the night May 28/29, 1997 over Essen, Germany, measured by a SODAR. From [2].

Literatur [1] Chr. Kottmeier, D. Lege und R. Roth (1983): Ein Beitrag zur Klimatologie und Synoptik der GrenzschichtStrahlströme u ¨ber der norddeutschen Tiefebene. Ann. Meteorol. N.F., 20, 18-19. [2] O. Reitebuch, A. Straßburger, S. Emeis, W. Kuttler (2000): Nocturnal secondary ozone concentration maxima analysed by SODAR observations and surface measurements. Atm. Environ., 34, 4315-4329. Correspondence: Nicolai Cosack – [email protected] – fax: +49 (0) 711 685-8293


85

A new damage approach for concrete towers of Wind Energy Converters subjected to multi-stage fatigue loading J. Gr¨ unberg, J. Göhlmann

a

a Institute for Concrete Structures , University of Hanover, Appelstr. 9A, 30167 Hannover, Germany.

The development of more efficient Wind Energy Converters (WEC) as well as the planning of offshore windfarms leads to strongly increasing requirements on the construction. WEC are subjected to high cycle loading with number of cycles up to 109. Thus, the fatigue design becomes more important and could be essential for dimensioning concrete structures and their several components [1]. The fatigue design codes for concrete are based on a linear cumulative damage hypothesis. But this simplified damage model is not able to evaluate the effective non-linear damage process in concrete. Therefore, the application of such rules could leads to an unsafe or uneconomical structure. The aim of the research is to develop a modify damage model, which could take into account the non-linear fatigue process sufficiently. Several models for describing the fatigue behavior of concrete were proved and an energy damage model for constant amplitude loading was considered and modified for twostage fatigue loading. First numerical results indicated a good conformity with measurement results available in literature [2]. Hence, the energy damage approach will be further developed and extended for multi-stage loading. At the time, an external algorithm for the introduced damage approach is associated with the finite element program ABAQUS. The calculated current fatigue damage state will be transferred to the concrete plasticity damage model of the FEM-Program in each time step during the nonlinear processing. Therefore, the damage evolution under multi-stage fatigue loading will be computed [3] and applied to analyzing concrete structures for WEC [4]. Also, measurements have been started at a concrete tower for a 5-MW-WEC to observe the fatigue process under real condition [5]. Furthermore, test on cylindrical specimen are performed to investigate the damage increasing under constant and two-stage fatigue loading. The obtained measurement data will be used for validating numerical models.

Literatur [1] Gr¨ unberg, J. and G¨ ohlmann, J. (2004): Zum Erm¨ udungsnachweis von Windenergieanlagen in Spannbetonbauweise, VDI Tagung: Windkraftanlagen – Sicherheit und Zuverlässigkeit. [2] Gr¨ unberg, J. and G¨ ohlmann, J. (2004): Schädigungen im Beton unter Erm¨ udungsbeanspruchungen, Erneuerbare Energien 14. [3] Gr¨ unberg, J. and G¨ ohlmann, J. (2005): Concrete Shell Structures for Wind Energy Converters subjected to multiaxial fatigue loading, 5th International Conference on Computation of Shell and Spatial Structures. [4] Gr¨ unberg, J. and G¨ ohlmann, J. (2005): Concrete Foundations for Offshore Wind Energy Converters subjected to fatigue loading, Copenhagen Offshore Wind (accepted). [5] Lierse, J. and G¨ ohlmann, J. (2004): Experimentelle Untersuchungen an Windenergieanlagen in Spannbetonbauweise, 44. Forschungskolloquium des Deutschen Ausschusses f¨ ur Stahlbeton.

Joachim G¨ ohlmann – [email protected] – fax: +49 (0) 511762-2175


86

Reliability of Wind Turbines Experiences of 15 years with 1,500 WTs B. Hahn, M. Durstewitz, K. Rohrig

a

a Division Information and Energy Economy Institut fuer Solare Energie Versorgungstechnik (ISET) Koenigstor 59, D-34119 Kassel, Germany

Wind energy is going to play an important role contributing to the electricity supply. However, the requirements concerning availability and reliability of wind turbine technology will increase as future plans are going to utilise wind energy in distances of 50 km offshore. The enormous development of wind technology in the recent decades lead to more sophisticated turbines, but still this technology has to be proven concerning it’s readiness for offshore use. Land based wind turbines (WT) achieve an availability of about 98%, meaning that each unit is shut down for about 150 to 200 hours a year. To investigate this more detailed, a broad study on wind turbine failures, their causes, and the resulting downtimes was started at ISET using data collected by the ‘Scientific Measurement and Evaluation Programme’ (WMEP). The WMEP is continuously supervising 1,500 WT in Germany over a period of at least ten years each. More than 62,000 maintenance and repair reports have been collected and evaluated up to now. Concerning reliability, the evaluations show e. g. that in the first four years of operation the failure frequency of 500/600 kW turbines was twice as high as that of smaller turbines whereas megawatt turbines again show twice as high failures rates compared to the 500/600 kW WTs. But overall the average duration of downtimes and the failure frequency are decreasing significantly during the first years of operation. This development is well known from other technical systems. Failures usually occur more often in the first phase of operation and - after a time of constant failure rate - in the last phase of operation. Out of the experience with this large number of wind turbines, the conference paper will give statistically proven figures for typical downtimes, mean time between failures (MTBF) and failure rates sorted by main components and size of turbine. Thus, it gives an overview of the status of reliability and availability of the current wind turbine technology and it allows an assumption on necessary improvements of reliability for future offshore wind turbines.

Berthold Hahn – [email protected] – fax: +49 (0) 561 7294-260


87

Systematic modelling of wind turbine dynamics M. Hänler, U. Ritschel, J. Kirchner and I. Warnke

a

a Windrad Engineering GmbH, Querstr. 7, 18230 Zweedorf, Germany

We present results on an ongoing project to develop a new computer simulation code for wind turbine dynamics and structural loads. This fully independent, self contained program constitutes a multi body system with a strictly modular structure. All flexible components of the turbine are equipped with a variable number of free modes which can be adjusted by the user to his needs. Special emphasis has been laid on the interaction of foundation and ground. The ground response model according to Hsieh [1] and Lysmer [2] has been implemented. This offers users the possibility to get useful information like the bearing pressure distribution directly from the simulation. Validations with measured data have been carried out. As an application we present results on earthquake loads on a wind turbine. Further efforts are made to extend the model to offshore-foundations and wave loads with the final aim to provide a full-fledged commercially available design code for on- and offshore wind turbines.

Literatur [1] T.K. Hsieh (1962): Foundation vibrations, Proc. Institution of Civil Engineers, vol. 22, pp. 211-226. [2] J. Lysmer (1965): Vertical motion of rigid footings, Dept. of civil eng, Univ of Michigan Report, PhD Dissertation. [3] J. A. Studer and M.G. Koller (1997): Bodendynamik, Springer-Verlag Berlin.

Dr. Michael H¨ anler – [email protected] – fax: +49 (0) 382 94/142 99


88

Design, Fabrication and Testing of the Wind Turbine Rotor Blades from Composite Laminated Materials B. Raˇsuoa a Aeronautical Department, University of Belgrade/Faculty of Mechanical Engineering, Kraljice Marije 16, Belgrade, Serbia and Montenegro

The importance of full-scale testing in the development process for fiber-reinforced composite wind turbine rotor blades is discussed, and illustrated by means of an example drawn from Faculty of Mechanical Engineering (University of Belgrade) experience in the use of composites in a wide variety of structural applications. The laboratory investigations of the structural properties are conducted at Aeronautical Department on all flightcritical dynamic components in order to determine structural adequacy in designing process [1]. In this paper the design, fabrication and analysis of behavior by full-scale verification testing for a wind turbine rotor blades of composite laminated materials is given. A development of a wind turbine rotor blades was performed in four phases: (1) the blades design on the working station using designing system Howard-Hughes, (2) preparation and cutting of blade components on the Gerber-Garment cutting system (Figure 1), (3) blade manufacturing in a two-section die (Figure 2), and (4) final verification testing. In the blade manufacturing procedure the conventional composite materials with epoxy resin matrix, a fiberglass filament spar, a ten-section skin of laminated fabrics, some carbon filament embedded along the trailing edge and core were used. All the used materials are standard products fabricated at Ciba-Geigy, Interglas GmbH, Torayca and others.

The verification test program for a wind turbine rotor blades encompassed static and dynamic testing. The static tests of the blade involved experimental evaluation of torsional and flexional blade stiffness and its elastic axis position. Dynamic tests involved testing of vibratory characteristics and verification testing of blade fatigue properties. The aim of the rotor blade vibratory testing program was to determine the blade main aeroelastic properties. The program included determination of the natural oscillation modes and the structure’s natural frequency and also evaluation of blade structural damping for foam core (Rohacell). The logarithmic decrement of the free vibrations was utilized to characterize the structural damping of blades. Q-factor is also usually used to define the structural damping and gives relative energy reduce in successive oscillations. The fatigue test program of the blade included: interlaminar separation (delamination) testing and geometric deformation of the blade cross-sections following the fatigue test program during which real rotor blade loads where simulated - the same loads to which blade is exposed under extrime conditions. The applied test loads include simulated steady centrifugal, vibratory cordwise bending and vibratory torsional pitch motion. The homological fatigue testing program for the blade root involved, conforming to the standards, fatigue testing of six identical blades. This testing includes a program with simulated centrifugal force-relaxing loading [1].

Literatur [1] B. Raˇsuo, Verification Testing of Aeronautical Constructions from Composite Laminated Materials in Designing Process, Journal of Aerospace, SAE 1998 Transactions, V-107-1,Warrendale, USA, 1999. Correspondence: Boˇsko Raˇsuo – [email protected] – fax: +381 11 3370-364


89

Multibody System Simulation of Offshore Wind Turbines B. Schlechta , T. Schulzea , Th. Hähnela and Th. Rosenlöchera a IMM - Institut f¨ ur Maschinenelemente und Maschinenkonstruktion - Technische Universit¨ at Dresden, Mommsenstrasse 13, D-01062 Dresden

During the last years a multitude of wind turbines has been put into operation with a continuously increasing power output. Although in the meantime wind turbines with 6 MW output are in the stage of development, a ”simpleëxtrapolation to larger dimensions of wind turbines on the basis of existing plants and operational experiences is questionable. The cases of damage of drive train components (especially gear boxes) and complete plants – in some cases total losses – are definite hints. Regarding planned service lives of up to 20 years the safe and load adequate dimensioning of the drive train requires the exact knowledge of the acting dynamic loads. This is relevant to both, the development and design stadium and the operation, especially under rough OffShore-Conditions. Actual Simulation Programs to determine the load situation of wind turbines focus mainly on aerodynamic problems and use a very simple 3-Mass-Model for the complete drive train. This allows only a very rough calculation of the outer loads of the main components. It is not possible to analyze dynamical effects in the drive train and especially in the gear box. The paper deals with the simulation of the dynamic behaviour of the complete drive train of a Multi-Megawatt-Wind-Turbine by using a detailed Multibody-SystemModel (MBS) with special respect to the internals of the planetary gear box and the drive train. This procedure allows additionally to the analysis of the torsional vibrations the determination of radial and axial vibrations of the main components (rotor, gear box, generator) and the gear box internals. The rotor blades as well as the nacelle frame and the tower structure will be modelled elastically by using Finite-Element-Structures. Parameter excitation out of the tooth-contact in the gear box will also be taken into account. Therefore one gets much more information (e.g. radial and axial dynamic bearing loads) that could be calculated neither from a wind load simulation program nor a torsional vibration simulation. After the model creation process the analysis of the natural frequencies shows in particular the advantages of the MBS-Simulation compared to the torsional simulation, because the MBS-model contains much more natural frequencies which are excitable by low frequent rotor or tower oscillations. In the time domain various load cases (normal operation, overload, run out, emergency stop and run off) will be discussed. Especially for the load adequate dimensioning of Multi-Megawatt-Wind Turbines for Off-Shore- Applications the paper shows the advantages of the MBS-Simulation in combination with FEM-structures because the testing of a real wind turbine under extreme operating conditions is much more cost-intensive and risky than the virtual testing with a good validated simulation model.

Fig. 1: MBS-Models of complete Turbine with detailed MBS-Model of Drive Train and MBS-Drive-Train on FEM-Structured Nacelle

Correspondence: Prof. Dr.-Ing. B. Schlecht – [email protected] – fax: +49 (0) 351-463-37137


90

Integrated Monitoring Systems for Offshore Wind Energy Plants uckera , S. Saida and S. Stöhra R.G. Rohrmanna , W. R¨ a Division Buildings and Structures (VII.2), Federal Institute for Materials Research and Testing (BAM), Unter den Eichen 87, Berlin, Germany

To ensure a high operational reliability of future offshore wind conversion systems (OWEC) with economically acceptable repair and maintenance efforts, comprehensive diagnosis and supervision concepts are required. Automatic monitoring systems will be an essential part of such concepts. Because of the fact, that during operation there will be static and dynamic interaction between the components ’structure’, ’machinery’ and ’blades’ it is necessary to develop the monitoring techniques in an overall concept. These monitoring systems are supposed to be applied for the design stage as well as for the operation and maintenance of wind energy plants. Within the scope of the IMO-WIND research project BAM, the University of Siegen and six industrial companies are involved in developing an integrated monitoring system for all components of OWEC. Our contribution to the EUROMECH colloquium describes the approach and the emphases of the research project. Furthermore the developed approaches for the measurement of actions and the condition monitoring of all components of an offshore wind energy plant are introduced. This includes the description of the sensors, the data acquisition and the management and the analysis of the data. Based on numerical and experimental results a first concept for damage detection and condition assessment of all components of the system will be described.

Fig. 1: Projected sensor distribution at OWEC with tripod foundation

Correspondence: Rolf G. Rohrmann – [email protected] – fax: +49 (0) 30 8104-1727

Author Index

A Anahua Aubrun Anders Antoniou

Edgar Sandrine S. Ionannis

72 47 57 76

B Bak Barth Battisti Beier Bergdahl Bergers Bibor Bierbooms Böttcher Bohlen Bourguet Braza Brønstedt Bye

Christian Stephan Lorenzo H. Th. Lars Jennifer Etienne Wim Frank Thomas Rémi M. Povl J.A.T.

76 67, 72 54 49 14 59 20 17 67, 73 77 38 38 56 16

C Cabezón Cârsteanu Castro Castro Chol Cleve Cosack Crespo

Daniel Alin F.A. J.J. Yun Sun Jochen Nicolai Antonio

80 68 19 68 35 69 84 81

D Devinant Dovgy Durstewitz E Ebert Eiff Emeis Eskilsson

P. S.A. M.

F. O.S. Stefan Claes

40 74 86

18 25 23, 66, 84 14

F Fallen Fernández Puga Friedrich Fuchs

M. José Rudolf Lazlo

18 18 70, 73 48, 83

G Garcia Gaunaa

Javier Mac

81 76

Geissler Gerasch Gnatowska Göhlmann Gößwein Gola Greiner Gr¨ unberg Gr¨ uneberger Gryning Guo Gyongyösi

Wolfgang W.-J. R. Joachim J. J. Martin J. Réné S.E. Hui András Zénó

39 53 64 85 50 42 30, 69 85 77 15 37 34

H Habbar Hähnel Hahn Hänler Hanitsch Hansen Hansen Harran Heinzelmann Henningson Hillger Hillmer Hohlen Højstrup Huhn Hureau

A. Th. Berthold Michael R. K.S. Martin O.L. G. Barbara D. Wolfgang Benjamin Harald Jørgen Holger J.

58 89 86 87 75 21 36 38 41 46 55 35 32 32 59 40

I Illig Iniesta Ivanell

C. A. Stefan

23 80 46

J Jaˇ nour Jensen Jiminéz Jørgensen

Z. N.O. ´ Angel Hans E.

63 15 81 15

K Kaiser Kantz Kayan Kirchner Kiss Kleinhans Knorr Kochin Kohlmeier

Klaus Holger Vladimir P. J. ´ am Ad´ David K. V.A. M.

32 27 74 87 34 70, 73 75 74 58

Author Index

Kozel Krämer Krassován K¨ uhn Kurbatskiya Kurbatskyab

K. T. Krisztina Martin Albert F. L.I.

63 41 34 84 65 65

L Lange Lange Langrder Langtry Larsen Larsen Leroux Lohaus Lopes da Costa Loyer Lovejoy

Bernahrd J. Wibke Robin B. Soeren E. Gunnar C. Karine Ludger J.C. S. Shaun

13, 16, 33, 45 49 32 42 15 28 25 57 19 40 24, 71

M Mann Mart´ı Martinat Masson Menet Menter Migoya Mittendorf Moroianu Moryn-K.

Jakob I. G. C. Jean-Luc F.R. Emilio Kim Dragos Elzbieta

15, 26, 29 80 38 20 78 42 81 58 48, 83 64

N Neumann Nitschke-Pagel

Tom T.

23 52

O Okulov

Ritschel Roberto Rohrmann Rolfes Rosenlöcher R¨ ucker

U. Fedrizzi Rolf G. R. Th. W.

87 54 90 53 89 90

S Said Schaffarczyk Schaumann Schlecht Schmitt Schulz Schulze Seidel Shen Shishkin Sicot Sládek Soerensen Soerensen Soerensen Stefano Stöhr Stoevesandt

S. A.P. Peter Bertholt Francois G. Detlev T. Marc Wen Z. A. Christophe Ivo Bent F. Jens N. Poul Dal Savio S. Bernhard

90 35 51 89 22 75 89 50 36 79 40 63 56 43, 44 29 54 90 79

T Tambke Tchiguirinskaia Thamsen Trujillo Trumars Twele

Jens Ioulia P.U. Juan J. Jenny J.

16 24 41 33 14 41

U Ummenhofer

T.

52

Dick M. A. J.M. S. L¨ uder Michael

17 59 29 24 42 16 49

Claus Hans-Peter I. Imke

21, 79 33 87 52

Valery L.

43, 44

P Palma Papp Paulsen Peinke Poppinga Puthli

Jose L. Botond U.S. Joachim Carsten Ramgopal

V Veldkamp 19 Veselcic 34 Vesth 29 Veysseire 45, 67, 72, 73, 77, 79 Völker 16 von Bremen 59 Vormwald

R Rasuo Rauch Rauh Reetz

Bosko J. Alexander Johannes

88 41 31 53

W Wagner Waldl Warnke Weich

Author Index

Weidinger Wessel Wilke Wolff Wood

Tam´ as Arne Fabian J.-O. David

34 33, 45 51 16 37

Y Yang

Hongxing

37

Z Zhou Zielke

Yu W.

37 58