R.P.L. Nijssen D.R.V. van Delft L.G.J. Janssen UPWIND: Blade Materials and Structures European Wind Energy Conference 2007 Session: Structural Design and

R.P.L. NijssenR.P.L. Nijssen

D.R.V. van DelftD.R.V. van Delft

L.G.J. JanssenL.G.J. Janssen

UPWIND: Blade Materials and Structures

European Wind Energy Conference 2007European Wind Energy Conference 2007

Session: Structural Design and MaterialsSession: Structural Design and Materials

Thursday, May 10Thursday, May 10thth, 2007, 2007

WMC Knowledge Centre

Wind turbine Materials and Constructions

● History● Blade & Material testing for over 20 years

● Part of Delft University of Technology until 2003

● Activities● Full-scale wind turbine structural testing

● Material research

● Software Development

● Facilities● Flexible full-scale test laboratory

● Fatigue test machines

● Workshops

● Specimen production

● Projects (EZ/EU)● OPTIMAT

● INNWIND

● UPWIND

Wind turbine Materials & ConstructionsWind turbine Materials & Constructions



Full Scale TestingFull Scale Testing



Material ResearchEU Project: Optimat Blades


Objectives: Reliable Optimal Use of Materials for Wind Turbine Rotor Blades

Period: 2002-2006Scientific Coordinator: WMC

Financial Coordinator: ECN



Variable Amplitude Loading Establishment of reference S-N curves Comparison between CA and VA

Wisper spectrum Block tests

Establishment of new Wisper spectrum Influence load sequence1 Comparison of lifetime prediction for ECN/TUDT-Data with Wisper

- Mean Value -

0

50

100

150

200

250

300

350

100

Cycles

Str

es

s [

MP

a]

DLR: Linear MinerTUDT: LG log-logTUDT: LG lin-logTUDT: LG lin-log incl. R=0.1TUDT: alt lin-log R=0.1UP: Linear Miner, batch IUP: Non-Linr Miner, batch IUP: Linear Miner, batch IIWisper ECN, IWisper ECN, IIWisper TUDT, I

101 102 103 104 105 106

Results of various Lifetime Prediction Models

Variable Amplitude Loading Establishment of reference S-N curves Comparison between CA and VA

Wisper spectrum Block tests

Establishment of new Wisper spectrum Influence load sequence11 Comparison of lifetime prediction for ECN/TUDT-Data with Wisper

- Mean Value -

0

50

100

150

200

250

300

350

100

Cycles

Str

es

s [

MP

a]


101 102 103 104 105 106

Comparison of lifetime prediction for ECN/TUDT-Data with Wisper- Mean Value -

0

50

100

150

200

250

300

350

100

Cycles

Str

es

s [

MP

a]




101 102 103 104 105 106

Results of various Lifetime Prediction Models





Complex Stress State Checks on uni-axial and bi-axial tests Investigate beam model of rotor blade vs. shell

model for stress components Establish guidelines and safety factors for various

levels of sophistication in the design

2

Complex Stress State Checks on uni-axial and bi-axial tests Investigate beam model of rotor blade vs. shell

model for stress components Establish guidelines and safety factors for various

levels of sophistication in the design

22





Extreme Conditions Check influence of extreme conditions

T -40° and +60° Humidity

Identification of degradation parameters Phenomenological modelling and exp. determination

0.0005x 0.0025x

0xE

xiE 1xiE

x

[ / ]x mm mm

Loading curve of the GF/ EP [+452 -452]s

-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451

[ ]x MPa3

Degradation of Stiffness during Fatigue Life

Extreme Conditions Check influence of extreme conditions

T -40° and +60° Humidity

Identification of degradation parameters Phenomenological modelling and exp. determination

0.0005x 0.0025x

0xE

xiE 1xiE

x

[ / ]x mm mm


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451


-20

0

20

40

60

80

100

120

140

160

0 0,005 0,01 0,015 0,02

x [mm/ mm]

x

[M

Pa]

Specimen CP451

[ ]x MPa33

Degradation of Stiffness during Fatigue Life





Thick Laminates & Repair Difference between thick and thin laminates

studied Material properties in thickness direction Measurements using embedded optical fibres Repair techniques Influence of brick and shell FE models

External patch

Filler

Scarf patch

Smooth or small steps

Parent structure

(A)

(B)

4

Plug/Patch and scarf repair systems FE analysis of the test specimen

Thick Laminates & Repair Difference between thick and thin laminates

studied Material properties in thickness direction Measurements using embedded optical fibres Repair techniques Influence of brick and shell FE models

External patch

Filler

Scarf patch

Smooth or small steps

Parent structure

(A)

(B)

44

Plug/Patch and scarf repair systems FE analysis of the test specimen





Residual Strength & Condition Assessment

Development of predictive model for residual strength reduction

Definition and validation of condition monitoring strategies for laminates in rotor blades

0.8·N3

N10.8·N1

0.2·N1

0.5·N1 N20.8·N2

0.2·N2

0.5·N2 N3

0.2·N3

0.5·N30

S

Log N 0.8·N3

N10.8·N1

0.2·N1

0.5·N1 N20.8·N2

0.2·N2

0.5·N2 N3

0.2·N3

0.5·N30

S

Log N

Stress levels for testing the residual strength after a percentage of the expected life

5

Residual Strength & Condition Assessment

Development of predictive model for residual strength reduction

Definition and validation of condition monitoring strategies for laminates in rotor blades

0.8·N3

N10.8·N1

0.2·N1

0.5·N1 N20.8·N2

0.2·N2

0.5·N2 N3

0.2·N3

0.5·N30

S

Log N 0.8·N3

N10.8·N1

0.2·N1

0.5·N1 N20.8·N2

0.2·N2

0.5·N2 N3

0.2·N3

0.5·N30

S

Log N

Stress levels for testing the residual strength after a percentage of the expected life

55





Improved Design Recommendations

Task group leaders and certification bodies

State-of-the-art analysis of results using resultsof the task groups 1 to 5

Based on consistent set of tests

Interaction effects also covered

Basis for future design guidelines

6

Improved Design Recommendations

Task group leaders and certification bodies

State-of-the-art analysis of results using resultsof the task groups 1 to 5

Based on consistent set of tests

Interaction effects also covered

Basis for future design guidelines

66





VariableAmplitude

Residual strength and

Condition MonitoringComplex

Stressstates

Extremeconditions

Thick LaminatesandRepair





OPTIDATOPTIDAT

0

5000

10000

15000

20000

25000

30000

35000

40000

45000

prior t

o 02-O

ct-0

3

02-Oct

-03

22-Oct

-03

31-Oct

-03

13-Nov-

03

14-Nov-

03

28-Nov-

03

26-Feb-0

4

12-May-

04

17-May-

04

25-May-

04

27-May-

04

09-Jun-0

4

11-J

un-04

18-Jun-0

4

18-Oct

-04

20-Dec-

04

04-Jan-0

5

14-Jan-0

5

28-Jan-0

5

04-Feb-0

5

12-Apr-0

5

28-Apr-0

5

23-Jun-0

5

01-Jul-0

5

08-Sep-0

5

19-Oct

-05

29-Nov-

05

12-Dec-

05

19-Jan-0

6

09-Feb-0

6

WMC DLR RAL RISØ CRES VUB UP VTT

Totals

18286

51411587 3884

17551673

7028

526

Tes

tin

g t

ime

[h

]

Total: 39880

Available via: www.wmc.eu



Lessons learnt from Optimat BladesLessons learnt from Optimat Blades

● Plate-to-plate and lab-to-lab variations are important in a project of this size● Minor Tg and Vf variationssignificant performance variations?

● Some plates worse in static strength, better in fatigue● Machine-to-machine● Environmental conditions● Sometimes larger than investigated influences ● More realistic assessment of scatter● Preferably: production of all specimens first, then mix and send

out (not possible in practice)

● Establishing an alternative test geometry is difficult● People perceive standards automatically as “better” even when

no background info is provided● Universal geometry is compromise…OK for consistency(?)



Topics for UPWIND WP 3.1Topics for UPWIND WP 3.1

WP 3.1: Tests and phenomenological modeling

WP 3.2: Micro/Meso-mechanical model

WP 3.3: Damage tolerant design

WP 3.1: Tests and phenomenological modeling

WP 3.2: Micro/Meso-mechanical model

WP 3.3: Damage tolerant design



Topics for UPWIND WP 3.1Topics for UPWIND WP 3.1

● Static strength, especially compression

● Constant Life Diagram, especially concentrating on higher number of cycles

● Bi-axial stress states

● Extreme (climatic) conditions as well as influence of the frequency on fatigue test results

● Behaviour of thick and repaired laminates

● Damping

● Life cycle analysis

● Extension with new materials of the public material database OPTIDAT

● www.wmc.eu



Partners WP 3.1Partners WP 3.1

● RISØ

● ECN

● WMC

● CRES

● UP

● GE

● FIBERBLADE

● VTT

● CCLRC

● VUB



Advantages

● Fits measurement equipment

● Eliminate geometry effects

● Flexibility (progressing insights during long-term project…)

● Combined tests, e.g. residual strength

Disadvantages

● Poorer compression characteristics than

ISO/ASTM

● Static● Compression-compression fatigue tests● Buckling (test machine dependent,

elevated temperatures, after fatigue damage)

Universal test geometryUniversal test geometry



Preliminary programmePreliminary programme

● Reference material● Glass Epoxy, delivered by GE

● WMC can also make their own plates

● Layouts

● Embedded sensors

● Study of several rectangular geometries for 4 and 6 layer lay-ups tested in preliminary programme● Free length

● Width

● Tab thickness: from 0 to 2 mm

● Static and zero-mean stress (R= -1) fatigue



First test results (R=-1 fatigue)First test results (R=-1 fatigue)

1.4

1.2

1

0.8

0.6

0.4107106105104103102101100

Cycles to failure

[%

] R08, tab:1 (AK&AQ)

R08, tab:0.5mm (AP)

R07, tab:1mm (AS)

R07, tab:0.5mm (AM)

R17, tab:0mm (AL)

R03, tab:1mm (AN)

R03 (OPTIMAT UD)

R06, tab:2 (OB)mm (AK)



CompressionCompression

● Approach Compression

● Risø set-up (RISØ)

● ISO (WMC)

● Combined loading (WMC)

● Test short OPTIMAT (WMC/UP/CRES)

● Standard OPTIMAT as reference (WMC/UP/CRES)

● Compression test University of Dresden (GE)

● VUB will look at the full optical field for compression

● Objectives

● New geometry for static compression

● Fatigue geometries for R=-1 and R=10



Ageing & FatigueAgeing & Fatigue

● Problem: initial fatigue results within OPTIMAT BLADES ● Lower than expected

● Required lower frequencies

● Is it only temperature or other effects as well?● Check at various frequencies and temperatures

● Check fatigue behaviour at elevated temperatures and high humidity levels



100

150

200

250

300

350

400

1 10 100 1000 10000 100000

Sequences to failure

Sm

ax

WISPER data

Multiple R-ratio (6 R-ratios)Single R; R=0.1

Linear GoodmanShifted Goodman

0

100

200

300

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [

MP

a]

103 Cycles

106 Cycles

109 Cycles

0

100

200

300

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [

MP

a]

103 Cycles

106 Cycles

109 Cycles

0

100

200

300

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [

MP

a]

103 Cycles

106 Cycles

109 Cycles

0

100

200

300

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [

MP

a]

103 Cycles

106 Cycles

109 Cycles

30% unconservative

2000% unconservative

Constant Life Diagram (CLD)Constant Life Diagram (CLD)



0

50

100

150

200

250

300

350

400

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [M

Pa]

0

50

100

150

200

250

300

350

400

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [M

Pa]

R=-0.4

R=0.5

R=0.1R=10

R=-2.5

R=-1● Linear Goodman Diagram: poor performance

● Development of comprehensive CLD formulation (based on regression through multiple R-values)

● Ideally: Static strength + few tests determine fatigue behaviour

0.0

50.0

100.0

150.0

200.0

250.0

300.0

350.0

400.0

-300.0 -200.0 -100.0 0.0 100.0 200.0 300.0 400.0m [M Pa]

a [M

Pa]

100

1000

1E+04

1E+05

1E+06

1E+07

1E+08

Static limit

CLD…modellingCLD…modelling



CLD…Long life tests (>106)CLD…Long life tests (>106)

● Verify S-N curve extrapolations

● Duration typically 1 month….+ scatter!

0

50

100

150

200

250

300

350

400

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [M

Pa]

0

50

100

150

200

250

300

350

400

-600 -400 -200 0 200 400 600

Smean [MPa]

Sam

p [M

Pa]

Most data available here

Most ‘real-life’ loads here

R=-0.4

R=0.5

R=0.1

R=10

R=-2.5R=-1



Thick LaminatesThick Laminates

Static and Fatigue results

100

1000

1 10 100 1000 10000 100000 1000000 10000000 100000000

No. of cycles

max

Thin specimens

Thick specimens

Series3

Series4

Power (Thinspecimens)Power (Thickspecimens)



Circle of lifeCircle of life



Missing link?Missing link?



Subcomponent TestingSubcomponent Testing

●Geometry used up to now: essentially 1D●No sidewise support

●Conservative estimation

●Not actually repaired but scarf connection●Same quality?

●New geometry for 2D tests ●Axial or 4-point Bending

●Repairs and/or buckling

●Glue and glue repairs

●Or double C instead of I-beam

●Or Glue connection

●Sandwiches

●Blade part: UD + web



EndEnd

● Thank you!

● Questions/comments/discussion…?

0.0

50.0

100.0

150.0

200.0

250.0

300.0

350.0

400.0

-300.0 -200.0 -100.0 0.0 100.0 200.0 300.0 400.0m [M Pa]

a [M

Pa]

100

1000

1E+04

1E+05

1E+06

1E+07

1E+08

Static limit

0

5000

10000

15000

20000

25000

30000

35000

40000

45000WMC DLR RAL RISØ CRES VUB UP VTT

Totals

18286

51411587 3884

17551673

7028

526

Testi

ng

tim

e [

h] Total: 39880



Session QuestionsSession Questions

● What advances in methods for the design analysis of wind turbine structural components will be adopted by the industry over the next 5-10 years? What benefits will these advances bring?

● Is the full scale testing of blades and other wind turbine components likely to become more or less important in the future?

● Does the introduction of probabilistic approaches to wind turbine design calculations result in more optimised structural components, more conservatism, or more uncertainty?

● Is it likely that adaptive blades will ever replace pitch regulated blades?

Documents

R.P.L. Nijssen D.R.V. van Delft L.G.J. Janssen UPWIND: Blade Materials and Structures European Wind Energy Conference 2007 Session: Structural Design and