Turbine Blades Povl Brndsted Materials Research Division Ris - - PowerPoint PPT Presentation

Composite Materials for Wind Turbine Blades

Povl Brøndsted Materials Research Division Risø National Laboratory for Sustainable Energy Technical University of Denmark

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Application Windmills – Wind turbines

Entertaining Challenging Larger and Larger

2011-05-15 2 MatWind, Keynote

Small

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

5M Wind Power Turbine, Brunsbüttel, Germany, 61.5 m blades

(Courtesy of LM Glasfiber A/S)

2011-05-15 3 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Growth

2011-05-15 MatWind, Keynote 4

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

The Wind Turbine Blade

LM 61.5 m (17.7 tons)

y = 0.0005x2.6589 5 10 15 20 25 30 10 20 30 40 50 60 70

Blade Length (m) Blade Weight (metric ton)

Trendline blades < 40 m 2011-05-15 5 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Blade construction

• an aerodynamic shell and a load-carrying beam

2011-05-15 6 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Blade construction

2011-05-15 MatWind, Keynote 7

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Load Type

2011-05-15 MatWind, Keynote 8

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Blade construction

• an aerodynamic shell and a load-carrying beam

2011-05-15 9 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 10

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Adhesive Joints

2011-05-15 MatWind, Keynote 11

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Material Selection for Blades

• Optimise against Stiffness
• Optimise against fatigue
• Optimise agaist Weight
• Life time 20 Years => >100.000.000 load cycles

2011-05-15 MatWind, Keynote 12

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Selection Tool - Stiffness

2011-05-15 MatWind, Keynote 13

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Mechanical performance

• f composites

Fibre content Fibre orientation Fibre length Porosity Fibre properties Matrix properties Fibre/matrix interface properties Fibre packing ability

Composite Parameters

2011-05-15 14 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Typical properties of fibres and composites

Type Stiffness Ef GPa Tensile Strength f MPa Density, f g/cm3 Vol. Fraction Vf Orientati

• n 

Stiffness Ec GPa Tensile Strength c MPa Density, c g/cm3 Merit Ec

1/2/c

Glass-E 72 3500 2.54 0.5 0º 38.0 1800 1.87 3.3 0.3 Random 9.3 420 1.60 1.9 Carbon 350 4000 1.77 0.5 0º 176.0 2050 1.49 8.9 0.3 Random 37.0 470 1.37 4.4 Aramid 120 3600 1.45 0.5 0º 61.0 1850 1.33 5.9 0.3 Random 14.1 430 1.27 2.9 Polyethylene 117 2600 0.97 0.5 0º 60.0 1350 1.09 7.1 0.3 Random 13.8 330 1.13 3.3 Cellulose 80 1000 1.50 0.5 0º 41.0 550 1.35 4.7 0.3 Random 10.1 170 1.29 2.5 Composites Fibres

2011-05-15 15 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Matrix:

• Thermosetting polymers
• Thermoplastic polymers
• Metals
• Ceramics

Fibres:

• Glass fibres
• Carbon fibres
• Aramid fibres
• Cellulose

fibres

• and more ….

+ = Composite materials Spider silk fibres

Composite Materials

2011-05-15 16 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Composite Architectures

2011-05-15 MatWind, Keynote 17

• Fibre-orientation
• unidirectional
• weaves/patterns
• random
• rientation
• Boundary

fibre/matrix:

• interface
• interphase (zone)

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Manufacturing

• Hand layup
• Vacuum Assisted Resin Transfer Moulduíng

2011-05-15 MatWind, Keynote 18

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 19

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Tvind Møllen

2011-05-15 MatWind, Keynote 20

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 21

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 22

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 23

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 24

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Root End

2011-05-15 MatWind, Keynote 25

Composite Length Scales

26 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Properties

2011-05-15 MatWind, Keynote 27

• Measurements of Mechanical Properties on Different Scales
• Full scale structures
• Wind turbines
• Components
• Blades
• Subcomponents
• construction details from blades, adhesive joints
• Materials performance
• Standards, recommendations, experimental models
• Microstructures
• Single fibre tests, ESEM tests

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 28

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Blade test - quality control LM 61.5 m

2011-05-15 29 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Blade tested to failure – static loading (well above design load)

2011-05-15 30 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Complicated failure - many failure modes

2011-05-15 31 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fracture modes

Adhesive joint failure Delamination (+/-45 )  Adhesive joint failure Cracks in gelcoat (chanal cracks) Splitting along fibres Skin/adhesive debonding

2011-05-15 32 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fracture modes - example: a wind turbine blade

Sandwich debonding Laminate Foam Delamination Split cracks

2011-05-15 33 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fracture modes - example: a wind turbine blade Sandwich debonding Split cracks Delamination Compression failure

2011-05-15 34 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fracture modes - example: a wind turbine blade

Delamination Delamination Multiple delaminations Split cracks in surface layer Splitting Splitting Buckling-driven delamination Compression failure

2011-05-15 35 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Multiscale modelling

2011-05-15 MatWind, Keynote 36

z

Combine a coarse 3D model with a fine model of each damage mode

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Subcomponent Test – Girder Section

2011-05-15 MatWind, Keynote 37

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Mechanical testing – Test Coupons

Measurement of mechanical properties used in

• qualification of materials
• constitutive models based on material structures and micromechanical

behaviour

• design, reliability and lifetime estimation
• models describing elastic-plastic behaviour and for use in solid mechanics

modelling

2011-05-15 38 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Material Qualification

• Characterisation of Laminates
• Characterisation of Core material and structure
• Characterisation of interface between skin and core
• Characterisation of sandwich

The materials are to be tested in static loading and in fatigue loading

2011-05-15 40 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Laminates

The following static mechanical tests are suggested

• Static tensile testing according to ISO 527/4, ISO 527/5 or ASTM D3039. At least 5 test coupons

are to be tested.

• Static compression testing according to ISO 14126 or ASTM D3410. At least 5 test coupons are to

be tested. In order to obtain the best results it is advised to use combined shear and end loading fixtures.

• V-notch shear testing according to ASTM D 5379. At least 5 specimens are to be tested
• Interlaminar shear strength measurements according to ASTM D 2344. At least 10 test coupons

are to be tested.

2011-05-15 MatWind, Keynote 41

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Laminates

The following fatigue mechanical tests are suggested

• Tension-tension fatigue testing according to ISO 13003. At least 10 test coupons are to be tested.

Range of Number of Cycles: 104 - 107.

• Compression-compression fatigue testing partly according to ISO 13003. No specific standard is
• available. An antibuckling device must be used in order to avoid global buckling, or short gauge

length specimens as in the static compression tests should be used. In order to obtain the best results it is advised to use combined shear and end loading fixtures. At least 10 test coupons are to be tested. Range of Number of Cycles: 104 - 107.

• Tension-Compression Fatigue testing partly according to ISO 13003. No specific standard is
• available. An antibuckling device might be used in order to avoid global buckling, or short gauge

length specimens as in the static compression tests should be used. At least 10 test coupons are to be tested. Range of Number of Cycles: 104 - 107.

2011-05-15 42 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Tensile Tests

2011-05-15 MatWind, Keynote 43

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Compression test

2011-05-15 44 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Test specimen with back-to-back mounted extensometers

2011-05-15 45 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Shear test

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue Test - Universal Testing Machine

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

S-N curves

Stage A Stage B Stage C

Log S Log N

In polymer matrix fibre composites the damage modes and the microstructural damage mechanisms do change basically between the different stages. In constant amplitude fatigue tests stage B can be analytically expressed in a power law:

N S C

m

 

2011-05-15 49 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue model

The number of cycles, NL, to a damage level defined by a stiffness degradation criteria, EL/E, can be calculated from

NL E n E E K C

L

             1

1

Assuming n = m: The constant C in the fatigue power law will depend on the change in stiffness in the material.

2011-05-15 53 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue experiments. Materials

The material is a 4 layers 90:10 fabric with chopped strand mat on both sides, again a nominal symmetric lay-up. The average fibre volume fraction and the porosity content were measured on 3 panels for this material: Vf =38 % and Vp = 7 %.

Mechanical data

Test type Modulus E0 (GPa) Max Stress (MPa) Max Strain (%) Longitudinal tension 26.0 ± 1.1 451 ± 110 2.0 ± 0.6 Transverse tension 11.1 ± 0.9 48 ± 2 2.0 ± 0.3 K-value m value R-value CA fatigue test 1.28 1015 10.55 0.1

2011-05-15 54 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Constant load amplitude fatigue curve for glass/polyester based on stiffness degradation

2011-05-15 55 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

0% 20% 40% 60% 80% 100% 120% 40000 80000 120000 160000

Cycles % of maximum load

Level 1: 50000 cycles Level 4: 1000 cycles Level 3: 20000 cycles Level 2: 30000 cycles Level 5: 20000 cycles Level 6: 30000 cycles

Fatigue experiments

Block loading sequence

2011-05-15 56 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue modelling

Block loading stiffness reduction curve

0.900 0.925 0.950 0.975 1.000 1.025 0.0.E+00 5.0.E+06 1.0.E+07 1.5.E+07 2.0.E+07

Relative Stiffness Number of Cycles

190 240 230 180 220 210 200 260 250

• Max. stress levels in MPa

50 100 150 200 250 300 350 1.0.E+04 1.0.E+05 1.0.E+06 1.0.E+07 1.0.E+08

Stress Level (MPa) Number of Cycles

2011-05-15 57 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue following either a regular or a random pattern as multi-level load histories can be simulated in Fatigue under varying stress amplitudes.

 Step tests  Repeated block tests  Stochastic tests

It is the goal to be able to compare variable amplitude fatigue behaviour to constant amplitude fatigue test results to predict the varying amplitude behaviour based on constant amplitude fatigue test results BUT IS IT POSSIBLE?

2011-05-15 58 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Palmgren-Miner’s rule Cumulative damage in fatigue can be expressed in a one parameter sum:

M n N

i i i j







1

Suggested “rule”: Failure takes place when M = 1.

2011-05-15 59 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Equivalent stress parameter

Using N=S ni in N•Sm=C

n S C

i eq m



 

Seq is the equivalent stress or strain parameter

 

S C n

eq m i m

  

 1 1

Palmgren-Miner’s sum:

 

M n N n C S C n S

i i i i m i i m

     

  



1 1 1 C S n

eq m i

 

 

S n S n M

eq i i m i m m

               

 

1 1

1

2011-05-15 60 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Equivalent stress parameter

For any load history an equivalent load value can be calculated. The lifetime of a specimen subjected to a varying fatigue load history is equal to the lifetime of the same specimen subjected to a constant fatigue load history at the equivalent level In order to calculate the equivalent load Palmgren-Miner’s sum must be known. Why does Palmgren-Miner’s rule M=1 not work? Because the results depend on M1/m and not on M. For example allowing a scatter of 10% in equivalent stress based on experimental results means that m = 3, 0.7 < M > 1.3 - metals m =10, 0.3 < M > 2.6 - GFRP m =20, 0.1 < M > 6.7 - CFRP

2011-05-15 61 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue experiments

Modelled S-N curve and experimental results from block load

0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 1000 10000 100000 1000000 10000000

Cycles Normalised stress

• Eq. values
• Max. values

97.5% stiffness line Constant Load line

2011-05-15 62 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

10 20 30 40 50 60 70 2000 4000 6000 8000 10000 12000 14000 16000

# Cycles Load Level

Peaks Troughs Tension Compression

Fatigue experiments

WisperX loading sequence

2011-05-15 63 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue experiments

Modelled S-N curve and experimental results from WisperX load history

0.000 0.002 0.004 0.006 0.008 0.010 0.012 0.014 0.016 1000 10000 100000 1000000 10000000 100000000

Cycles Normalised stress

• Eq. values
• Max. values

97.5% stiffness line Constant Load line

2011-05-15 64 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Damage model

Damage: w = D = 1 - E/E1

E/E1

Stage 1 Stage 2 Stage 3

N

In stage 2:

E E A N B

1

   d dN A

E E

( )

1 

2011-05-15 66 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Assumption: The rate of change in stiffness, is only a function of the stress level .

d E E dN K E n

1

                 

Damage model

d E E K E dN E E K E N

E E

n N n 1 1 1

1

1                        

 

 

2011-05-15 67 MatWind, Keynote

Fracture mechanics approach to fatigue & fracture

crit Ic

a C K   

a C K     

K (MPa√m) da/dN

Kth KIc Stage II Sub-critical crack growth

a0



 K Static: Dynamic:

 

  K 

Fast fracture No crack growth

IF K > KIc ………. Fast fracture IF Keff < Kth …. No crack growth IF Kmax > KIc …. ...Fast fracture IF Keff < Kth & Kmax < KIc ………. Sub-critical crack growth Static Dynamic

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fracture mechanics concepts

M M L

Large-scale bridging

2011-05-15 69 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Fatigue Life-Time Prediction

• As long as the damage modes and the microstructural damage mechanisms do not

change basically we have found that:

 Fatigue design curves (nominally equivalent to traditional S-N-curves) can be

constructed based on measurements of the stiffness reduction in a material under constant amplitude fatigue loading .

 Fatigue lifetime for materials subjected to varying load amplitudes can be predicted

using an equivalent stress or strain parameter weighted by the value of Palmgren- Miner’s sum. However, this value can only be determined experimentally. A prediction can be made based on constant amplitude results if it can be assumed that M1/m equals 1.

 Alternatively fatigue lifetime for materials subjected to varying load amplitudes can

be predicted using the stiffness (or damage) model. The stiffness reduction can be calculated on a cycle to cycle basis using the constant amplitude fatigue curve.

 This approach is based on the assumption that the sequential effect is small (the

damage level of the next cycle does not depend on the preceding load history).

2011-05-15 MatWind, Keynote 70

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Interface testing

• Interlaminar fracture toughness tests in 4 different crack opening modes. No specific standard is
• available. Double Cantilever Beam test specimens are manufactures by cutting a crack starter

notch in a manufactured test beam. The tests are performed in Mode I, Mixed Mode 1:2, Mixed Mode 2:1, and in Mode II. 5 specimens are required for each test mode. Test fixture to be used is special mixed mode test fixture developed by Risø DTU.

2011-05-15 71 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Loading Principle

2011-05-15 72 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Interface strength - single fibre tests

2011-05-15 MatWind, Keynote 73

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Saturated specimen: glass fibre embedded in a polymer matrix Counting number of fibre breaks: a) Optical observation b) Acoustic emission

Fragmentation test

2011-05-15 74 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Single fibre fragmentation set-up Specimen’s geometry

2011-05-15 75 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

a) Fragment l=lc b) Fragment l>lc c) Fragment l<lc

Single fibre fragmentation test

2011-05-15 76 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Polarised-light micrograph showing fibre/matrix debonding debonding yielding

2011-05-15 77 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Weak interface: debonding at the fibre/matrix interface Strong interface: crack propagates into the matrix

2011-05-15 78 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Epoxy specimen tested at 80% of the fibre’s tensile strength a. 1 cycle. b.

• b. 1000 cycles.

c.

• c. 5000 cycles

d.

• d. 10000 cycles

Single Fibre Fragmentation Test - FATIGUE

2011-05-15 79 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

10 20 30 40 50 60 70 80 1 10 100 1000 10000 100000 1000000 Cycles Debonding Length (1E-6m) 80% - d1 80% - d2 80% - d3 80% - d4 60% - d2 60% - d3 60% - d4

Fatigue of epoxy at 60% and 80% of the fibre’s tensile strength

2

Single Fibre Fragmentation Test – FATIGUE

2011-05-15 80 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Concluding remarks

Material mechanics and the continuum damage mechanics approach is used to account for the basic behaviour of the materials and to characterize the damage evolution in long fiber laminated composites. Fracture mechanics and cohesive laws are used for analysing construction details and joints The experimentally determined behaviour are based on both component and full scale tests Larger structures necessitates analyses and tests of scaled components. The relationships between component behaviour and full scale can be analysed using models and mechanical laws based on the fundamentals in materials mechanics

2011-05-15 81 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Challenges

Celluloses fibres

• wood, bamboo, sisal, coco, flax, hemp, jute, straw

Sandwich and core

• Layered constructions, core materials
• Holllow components
• Energy absorbing

Interfaces/interphases

• Chemestry, physics and mechanics in fibre/matrix interfases
• Relation to manufacturing technology
• Controlling of interfase design of composite properties

Sustainability/Recycling

• Life cycle analyses
• Craddle to Craddle
• Environmantal friendly resources
• CO2 balance

2011-05-15 82 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Wind Rotor Blade

2011-05-15 83 MatWind, Keynote

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark

Material Needs

2011-05-15 MatWind, Keynote 84

Risø DTU, Technical University of Denmark Risø DTU, Technical University of Denmark 2011-05-15 MatWind, Keynote 85