The importance of wind turbine aerodynamics · The importance of wind turbine aerodynamics Illustrated with results from international cooperation projects Gerard Schepers (ECN/NHL)

www.ecn.nl

The importance of wind turbine aerodynamics

Illustrated with results from international cooperation projects

Gerard Schepers (ECN/NHL)

Keynote lecture PhD seminar

European Wind Energy Academy

Stuttgart, 24-9-2015

Content

• Importance of wind turbine aerodynamics

• Aerodynamic design modelling

– EU FP7 project AVATAR

• Aerodynamic wind turbine measurements

– IEA projects on field and wind tunnel measurements

(IEA Tasks 14, 18, 20 and 29)

– New Mexico: Wind tunnel experiments in Large Low Speed Facility of

German Dutch Wind Tunnel

• Conclusions and recommendations

2

IMPORTANCE OF AERODYNAMICS

Prof G. van Kuik, TUDelft“Aerodynamics is the mother of all sciences for wind turbines”

322-9-2015

IMPORTANCE OF AERODYNAMICS

‘The mother of sciences’ for:

• Energy production• Loads (strength, costs)• Stability (failure, damage)• Control

– Stall– (Individual) pitch– Distributed aerodynamic control

• Wind farm effects• Acoustics

422-9-2015

CHALLENGES IN AERODYNAMICS

• Aerodynamics means solving Navier Stokes flow equations

• The exact solution is not known

• Proving Navier Stokes existence and smoothness is one of the 7 Millennium Prize Problems, see also 1) and 2) (http://www.claymath.org/millennium/)

• Simplified aerodynamic calculational models should be applied

1) Werner Heisenberg"When I meet God, I am going to ask him two questions: Why relativity? And why turbulence? I really believe he will have an answer for the first."

2)Horace Lamb"I am an old man now, and when I die and go to heaven there are two matters on which I hope for enlightenment. One is quantum electrodynamics, and the other is the turbulent motion of fluids. And about the former I am rather optimistic."

So Aerodynamics is difficultBUT….…….

• Aerodynamics of wind turbines is even extremelydifficult

1. Rotating

2. In lower part of atmosphere � extremely turbulent

3. Instationary

4. Stall

5. Variation in scales and huge size(Diameter can be twice span of Airbus A380)

6. Constraints and interactions, e.g. – System dynamics

– Thick airfoils

Wind turbine aerodynamic

calculations are extremely time consuming

o Load calculations: Many long time series needed to get ‘statisticsright’ (~106 nr of time steps *)

• Calculational time of design calculations >>>lifetime of the wind turbine!!!!

o Generally calculations need to be done with extremely simplified(and so uncertain) Blade Element Momentum (BEM) Theory

o Impact of aerodynamic modelling uncertainties :

• Underprediction of loads: Design risk and costly safety margins

• Overprediction of loads: Overdimensioned, costly design

• Uncertainties in performance: Economic impact

*) J.G. Schepers Engineering models in aerodynamics, (2012) Technical University of Delft repositoryhttp://repository.tudelft.nl/view/ir/uuid:92123c07-cc12-4945-973f-103bd744ec87/

Content






(IEA Tasks 14, 18, 20 and 29)




8

Wind turbine model types/ characteristics

1. Blade Element Momentum (BEM)

Steady, 2D theory for aligned flow conditions with additional engineering models

– Needs airfoil tables (cl and cd as function of angle of attack)

– Engineering models are added with calibration parameters for e.g. instationary and 3D effects

2. Vortex wake models

Induced velocity from vortex wake Blade loads� vorticity

22-9-2015

Wind turbine model types/ characteristics

3. Coupled viscous-inviscid models

– Inviscid outer flow

– Viscous boundary layer

4. CFD (Computational Fluid Dynamic) models

1022-9-2015

Picture DTU

Computational Fluid Dynamics

(CFD)

(U)RANS (Unsteady Reynolds-Averaged Navier-Stokes)– Decomposition of flow quantities into a time-averaged and a

fluctuating quantity– With the averaging, models for the description of turbulence are

neededLES (Large Eddy Simulation)

– Large turbulence scales are resolved in time and space– Small scales are described by a subgrid scale model– Excessive number of cells needed to properly resolve attached (blade)

boundary layersHybrid RANS / LES or DES (Detached Eddy Simulation)

– Attached boundary layers (of the rotor) are calculated in (U)RANS mode

– Separated flow, wake and atmospheric boundary layer are calculated in LES mode 11

Picture: T. Lutz

Summary of wind turbine model

types/ characteristics

1222-9-2015

Figure: H. Ozdemir

Content






(IEA Tasks 14, 18, 20 and 29)




13

FP7-ENERGY-2013-1/ n° 608396 14

EU FP7 Project initiated by EERA *)

November 1st 2013-November 1st 2017

1. Energy Research Centre of the Netherlands, ECN (Coordinator)

2. Delft University of Technology, TUDelft

3. Technical University of Denmark, DTU

4. Fraunhofer IWES

5. University of Oldenburg, Forwind

6. University of Stuttgart, USTUTT

7. National Renewable Energy Centre, CENER

8. University of Liverpool, ULIV

9. Centre for Renewable Energy Sources and Saving, CRES

10.National Technical University of Greece, NTUA

11.Politecnico di Milano, Polimi

12.GE Global Research, Zweigniederlassung der General Electric Deutschland Holding GmbH, GE

13.LM Wind Power, LM

22-9-2015*) European Energy Research Alliance

FP7-ENERGY-2013-1/ n° 608396 15

Main motivation

UPSCALING towards 10-20 MW turbines is expected to lead to turbines with:

� Low induction

� Long slender blades

� Thick airfoils

� High tip speeds

� Passive (e.g. vortex generators) and active (e.g. flaps) devices

22-9-2015

FP7-ENERGY-2013-1/ n° 608396 16

Motivation, ctd

•We simply don’t know if present aerodynamic models are good enough to design 10MW+ turbines– “No mature industry will ever design a MEuro machine with

unvalidated tools” (M. Stettner, GE)

• 10MW+ rotors violate assumptions in current aerodynamic tools, e.g.:– Reynolds number effects,

– Compressibility effects

– Flow transition and separation,

– (More) flexible blades

• Hence 10MW+ designs fall outside the validated range of current state of the art tools.

22-9-2015

FP7-ENERGY-2013-1/ n° 608396 17

Avatar: Main objective

To bring the aerodynamic and fluid-structure models to a

next level and calibrate them for all relevant aspects of

large (10MW+) wind turbines

22-9-2015

FP7-ENERGY-2013-1/ n° 608396 18

Avatar: Work procedure

Problem: No 10 MW turbines are on the market yet:

• Validate submodels against experiments• Pressurized DNW HDG wind tunnel

• Airfoil measurements at Reynolds numbers up to 15 M and low Mach (< 0.2)

• LM: Wind tunnel airfoil measurements at dynamic conditions

• Forwind: Wind tunnel airfoil measurements at known turbulence

• DTU : Danaero: Aerodynamic field experiments on a 2.3 MW turbine and supporting 2D wind tunnel measurements

• TUDelft: Wind tunnel experiments on airfoils with vortex generators, flaps, spoilers

• NTUA: Wind tunnel experiments on airfoils with/without vortex generators

• Note: Other experiments are supplied in-kind (e.g. New Mexico)

22-9-2015

FP7-ENERGY-2013-1/ n° 608396

Reynolds number

• Reynolds number: Re = ρ V L/µ – Ratio between the pressure and shear forces: ρ V2/(µ V/L)– High Reynolds number: Less viscous effects (higher lift, lower

drag)– L is a characteristic length scale, i.e. the chord of the blade

• 10 MW wind turbine: Re 15-25 106

•Wind tunnel :– Small model� Re < 6 106

• Results differ from real world– Increase velocity *) or decrease temperature or

increase pressure to reach ‘real’ Reynolds numbers

*) Note: Velocity < ~100 m/s to avoid Mach number effects

FP7-ENERGY-2013-1/ n° 608396 20

Pressurized German Dutch Wind Tunnel DNW-HDG

September 15

HDG 2d Testing

Author: M. Jacobs, DNW-GUKFor informational use only, not to be referenced.

Status of data and pictures included January 2012Questions to [email protected]

FP7-ENERGY-2013-1/ n° 608396

AVATAR experiment in pressurized wind

tunnel

• Reynolds number up to 15 Million– Pressures up to 80 Bar

• Pressure distribution measurements• Wake rake for drag determination,• Estimation of transition location using

embedded Kulite pressure sensors • Fluorescent oil flow visualization• Profile: DU00-W-212, c=15 cm• Results are brought into a blind test,

presented at EWEA 2015 (November 2015).– Definition of cases: see http://www.eera-avatar.eu/

FP7-ENERGY-2013-1/ n° 608396 22

Avatar: Work procedure

Use the different models from partners in the project

o It is a cooperation project!

o In the project we have many models which range from computational efficient ‘engineering’ tools to high fidelity but computationally expensive tools

o High fidelity models feed information towards engineering models and intermediate models

22-9-2015

FP7-ENERGY-2013-1/ n° 608396

Comparison airfoil

performance

Courtesy: N Sorensen

Observations:• Good agreement between

incompressible codes• Compressible codes and

Q3UIC differ• Correlation based transition

models need to be improvedfor high Reynolds numbers

Content






(IEA Tasks 14, 18, 20 and 29)




24

Aerodynamics: What do we need most?

•MEASUREMENTS , MEASUREMENTS, MEASUREMENTS

• Detailed aerodynamic measurements along the blades are urgently needed at several conditions

2522-9-2015

IEA TASKS ON AERODYNAMIC

MEASUREMENTS

– 1991-1997: IEA Task 14 (Field Rotor Aerodynamics, Operating Agent: ECN)

– 1997-2001: IEA Task 18 (Field Rotor Aerodynamics, enhanced, Operating Agent: ECN))

– 2001-2007: IEA Task 20: (Analysis of NREL’s NASA-Ames, measurements, Operating Agent: NREL)

– 2008-2017: IEA Task 29: Mexnext (Operating Agent: ECN)

– Mexnext-I Analysis of Mexico measurements

– Mexnext-II New Mexico experiment

– Mexnext-III Analysis of New Mexico measurements

26

Rotor aerodynamic measurements:

IEA Tasks 14/18

– Period: 1991-2001

– Aim: To coordinate aerodynamic test programs on field facilities

– Participants– Netherlands Energy Research Foundation, ECN (NL; Operating Agent)

– Delft University of Technology, DUT (NL)

– Imperial College/Rutherford Appleton Laboratory, IC/RAL (UK, Only Task 14)

– National Renewable Energy Laboratory, NREL (USA)

– RISØ, The Test Station for Wind Turbines (DK)

– Mie University (JP, Only Task 18)

– Centre for Renewable Energy Sources, CRES (Gr, Only Task 18)

2722-9-2015

Aerodynamic measurements

– To develop, validate aerodynamic models

– Conventional measurement programs: Only indirect, global aerodynamic information

– Desired: Direct local aerodynamic properties (I.e. pressure distributions, inflow angles, inflow velocities)

2822-9-2015

Typical scan rate: 15 kHz port to port

~500 Hz for complete scan

Damping due to tubes for f > ~ 20 Hz

Differential pressures are measured

scanner

Lift Normal

Tangential

Drag

V

probe

Reference pressure

IEA Tasks 14/18: Facilities

2922-9-2015

ECN facility NREL facility

IEA Task 14/18 some results 1)

30

•Database of measurements (ask author)•‘Discovery’ of

•Stall delay at inner part of wind turbine blades(underprediction of loads at large angles of attack when using2D airfoil coefficients)

• Overprediction of tip loads when using 2D airfoil coefficients•‘Compensating’ errors when using global measurements

“The recommendation of wind turbine codes solely on the basis of

good agreement with measured power or blade moments is unjustified” 2)

1) J.G. Schepers, A.J. Brand, A. Bruining, R. van Rooij, J.M.R. Graham, R.J.H. Paynter, M.M. Hand, D.A. Simms, D.G. Infield, H.A. Madsen, T. Maeda, Y. Shimizu, N. Stefanatos. (2002). ‘Final Report of IEA Annex XVIII: ‘Enhanced Field Rotor Aerodynamics Database.’

ECN-C--02-016, Energy Research Center of the Netherlands, ECN. 2) J.G. Schepers ‘Engineering models in wind energy aerodynamics,’, (2012). TUDelft repository

http://repository.tudelft.nl/view/ir/uuid:92123c07-cc12-4945-973f-103bd744ec87/

Validation measurements, status at end of 90’s

1990:

Measurements of power and loads showed differences but they were too global to form a basis for improvement of aerodynamic models

Desired:

• Local aerodynamic loads (pressure distribution) in field conditions (IEA Tasks 14/18)

• But also: Constant, uniform and controlled conditions (�Windtunnel)

3122-9-2015

3222-9-2015

Measurements in NASA-Ames wind tunnel

•Carried out by NREL (National Renewable Energy Laboratory), USA•Spring 2000 •24m x 36m NASA-Ames wind tunnel.•10 m rotor•Measurement of pressure distributions at 5 locationsalong rotor blade•Analysed in IEA Task 20 (Operating Agent: NREL)•Participants:

–ETS (Canada)–RISO/DTU (Denmark)–CRES/NTUA (Greece)–ECN/TUDelft (Holland)–IFE (Norway)–CENER (Spain)–HGO (Sweden)

Results from IEA Task 20, References

• S. Schreck. (2008). IEA Wind Annex XX: HAWT Aerodynamics and Models

from Wind Tunnel Measurements. TP-500-43508, National Renewable Energy Laboratory, Golden, Colorado. http://www.nrel.gov/docs/gen/fy09/43508.pdf

• J.G. Schepers and R. van Rooij (2008) Analysis of aerodynamic

measurements on a model wind turbine placed in the NASA-Ames tunnel.

Contribution of ECN and TUD to IEA Wind Task XX, Energy Research Center of the Netherlands, ECN-E--08-052

33

Validation measurements, status at ~2005

1990:

Measurements of power and loads showed differences but they were too global to form a basis for improvement of aerodynamic models

Desired:

• Local aerodynamic loads (pressure distribution) in field conditions (IEA Tasks 14/18)

• Constant, uniform and controlled conditions (�NASA-Ames windtunnel measurements from IEA Task 20 )

• But also: Induced velocities and wake velocities (� Detailed flow field measurements from Mexico project) 34

22-9-2015

Mexico and New-MexicoMexico = Model rotor EXperiments In COntrolled conditions

• Collaborative projects, coordinated by ECN

– Mexico: December 2006

– New Mexico: July 2014

• Measurements in Large Low Speed

Facility (LLF) of German Dutch

Wind tunnel (DNW)

– North East Polder (Netherlands)

– Open test section: 9.5 x 9.5 m2

– Diameter of rotor: 4.5 m

o Pressure and load measurements

o Velocity measurements (PIV)

o Acoustic measurements (only New Mexico) 35

3622-9-2015

Flow field measurements with stereo PIV, done by DNW

PIV traverse tower with two camera’s, aimed at (horizontal) PIV sheet:

(38*61 cm (New Mexico) and 35x42 cm(Mexico))in symmetry plane of rotor (‘9 o-clock’).

Traversing in axial+radial directionAxial range: -1.0 D to 1.3 DRadial range: r/R 0.17�1.14 (New Mexico)

r/R 0.53�1.22 (Mexico)

Seeding (tiny bubbles) are brought in air. PIV sheet is illuminated with laser flash, and two digital photographs are taken with a delay of 15-100 nanoseconds;Sheet is subdivided into small ‘interrogation windows’ Velocity vector is the one resulting in maximum cross

correlation between the two shots.

Analysis of (New) Mexico results in IEA Task 29

MexNext-I, II and III

– Netherlands(Energy Research Center of the Netherlands (ECN, Operating Agent), University of Delft, Technical University of Twente, Suzlon Blade Technology, DNV-GL (Only in Mexnext-II/III))

– Canada (École de technologie supérieur Montreal, University of Victoria ) (Only in Mexnext-I)

– China (CWEA) (Only in Mexnext-II/III)

– Denmark(DTU)

– France (Onera and IFPEN) (Only in Mexnext-III)

– Germany(University of Stuttgart, University of Applied Sciences Kiel, Forwind, Fraunhofer-IWES, Windnovation, Enercon, DLR)

– Israel (Israel Institute of Technology (Technion)

– Japan (Mie University/National Institute of Advanced Industrial Science)

– Korea((Korea Institute of Energy Research and Korea Aerospace Research Institute) (Only in Mexnext-I)

– Norway (Institute for Energy Technology/Norwegian University of Science and Technology (IFE/NTNU)

– Spain(Renewable Energy National Centre of Spain (CENER) and National Institute for Aerospace Technology, INTA) (Only in Mexnext-I/II)

– Sweden(Royal Institute of Technology/University of Gotland (KTH/HGO))

– USA (National Renewable Energy Laboratory (NREL))

3722-9-2015

Results from Mexico (Mexnext-I)

399/22/2015

One of the lessons learned:Reflection from laser light on nacelle disturbmeasurements

Content






(IEA Tasks 14, 18, 20 and 29)




40

From Mexico to New Mexico..

41

• One week of tunnel time in Mexico has led to many lessons learned!

– Improved models and understanding of the load driving 3D flowfield

From Mexico to New Mexico

42

• Lessons learned gave possibility for high quality data at relatively low cost *)

� Model still in a good shape

� Answer unanswered questions

� Fill missing gaps from first campaign

� Improved measurement capabilitiesat DNW (PIV, acoustics)

� Repeat less succesfull measurements

� Aerodynamic devices and pitch misalignment

*) Measurements were always supposed to be carried out in 2 slotsbut budgetary constraints prevented us from doing so

New Mexico: One of the lessons learned:

Rhodamine on nacelle and spinner to avoid

reflections

Mexico New Mexico

Results from New-Mexico compared

with Mexico

• Pressure distribution at V = 15 m/s• r/R=0.82• Mexico and New Mexico compared

with Ellipsys3D (courtesy N Sorensen DTU)

• Good repeatability after 8 years!

New Mexico: PIV-animation

Similar flow field behind different blades

Courtesy: Koen Boorsma

Results from New-Mexico compared

with Mexico

• Axial PIV traverse at V = 15 m/s carried out by DNW

• Two radial positions• Mexico and New Mexico

compared with CFD results from DTU’sEllipsys3D

• Who do Mexico and New Mexico results differ??

ECN asked DNW to repeat the tunnel

calibration from 2002…..

Difference in tunnel calibrations from2002 – 2014: 0.24 m/s at V=15 m/s!

Good agreement between 2014tunnel velocity and PIV velocityin empty tunnel gives confidence!!

“MEASUREMENTS WITHOUT INFORMATIONON QUALITY HAVE NO VALUE “

K. Boorsma, Researcher ECN

Content


• Aerodynamic design modelling– EU FP7 project AVATAR

• Aerodynamic wind turbine measurements– IEA projects on field and wind tunnel measurements

(IEA Tasks 14, 18, 20 and 29)

– New Mexico: Wind tunnel experiments in Large Low Speed Facility of German Dutch Wind Tunnel


48

Conclusions

– Since the 80´s detailed measurements have helped to improve and validate aerodynamic models as implemented in wind turbine design codes

– The accuracy of wind turbine design codes has improved enormously • More reliable and cost effective wind turbines

‘Nowadays there isn’t a single designer to find who would dare to

design a wind turbine with the aerodynamic modelling from the 80’s` 1)

1) J.G. Schepers Engineering models in aerodynamics, (2012) TUDelft repositoryhttp://repository.tudelft.nl/view/ir/uuid:92123c07-cc12-4945-973f-103bd744ec87/

4922-9-2015

Recommendations on

aerodynamic modelling

– Work should continue on improving aerodynamic models of all categories ranging from engineering models to CFD *)• Continuous improvement of aerodynamic models is current

practice in comparable disciplines(e.g. aerospace)!

• This is needed for current state of the art turbines as well as for future large wind turbines with unconventionalities

– Model improvements are impossible without GOOD measurements

5022-9-2015

*) Unless someone wins the millenium prize………..

Recommendations on

aerodynamic measurements

– Wind tunnel measurements and field measurements are COMPLEMENTARYo Wind tunnel measurements are done at well known (but less representative)

conditions

o Field measurements are done at representative (but less known) conditions

– We see initiatives on wind tunnel but not on field measurements:o The time is ripe for a PUBLIC aerodynamic field experiment on a representative

wind turbine using the most innovative measurement techniques

o Need for large scale public experiment identified by EERA-Aerodynamics

o Design of experiment included in AVATAR

• Turbine?

• Financing?51

22-9-2015

52Derek Bok, President of Harvard University

Financing wind turbine aerodynamics….

Thank you for your attention

53

Documents

The importance of wind turbine aerodynamics · The importance of wind turbine aerodynamics Illustrated with results from international cooperation projects Gerard Schepers (ECN/NHL)