Distribution of Defects in Wind Turbine Blades &

Reliability Assessment of Blades Containing Defects

Authors:

Henrik Stensgaard Toft, Aalborg UniversityKim Branner, Risø-DTUPeter Berring, Risø-DTUJohn Dalsgaard Sørensen, Aalborg University / Risø-

DTU

Contents

• Introduction

• Distribution of Defects in Wind Turbine Blades

• Influence of Delaminations and Reliability Assessment

• Non Destructive Inspection

• Conclusion

• Future Work

Introduction

Uncertainties in calculation of the load carrying capacity for wind

turbine blades.

1. Material properties

• Physical uncertainty (Aleatory)• Statistical uncertainty (Epistemic)

2. Finite Element calculation• Model uncertainty (Epistemic)

3. Failure criteria• Model Uncertainty (Epistemic)

Leading edge

M ain spar(load carrying box)

Upw ind side

Downwind side

Towards tip

Tra iling edge

Aerodynam icshell

Introduction

Uncertainties – captured by partial safety factors.

Local production defects are not taken into account !

• Quality control of the production process• Non Destructive Inspection (NDI) of the blades

Stochastic models for the distribution of defects.

The models are set up on an empirical basis and have not yet been

calibrated against observations from blade manufacturing and testing.

Production Defects

Influence parameters:

• Type of defect• Size of defect• Position of defect

Local production defects:

• Delaminations Voids• Wrinkles Defects in glued joints• Matrix cracks

Delaminations:

• Areas of poor or no bonding between adjacent plies.

Distribution of Defects – Model 1

Defects are completely random distributed in the blade.

The blade can be considered as a 2-dimensional planar region A since

defects tend to occur in a layer or in the interface between two layers.

i) The number of defects in the region A follows a Poisson distribution

ii) The distribution of defects is an independent random sample from a uniform distribution

Distribution of Defects – Model 2

Defects occur in clusters which are randomly distributed in the blade.

i) The “parent” defects follow model 1

ii) Each “parent” defect produces a number of “offsprings” following a Poisson distribution

iii) The position of the “offspring” defects relative to their “parents” is independently and identically distributed according to a bivariate probability density function. (Normal distribution)

Distribution of Defects

Model 1Completely Random Distribution

Model 2Random Cluster Distribution

Load Carrying Capacity of Main Spar

Transverse strains in main spar – Failure defined according to:

• Maximum Strain• First Ply Failure

0 0.2 0.4 0.6 0.8 1 1.2-1

0

1

2

3

4

Normalized load [-]

Nor

mal

ized

str

ain

[-]

Transverse xxLongitudinal zzFirst Ply Failure

Production Defects – Delaminations

• Size refers to percent of plate width

• Through thickness position 20%

Defect Size Longitudinal strain

(compression)

Transverse strain

(tension)

No defect - 1.00 1.00

Delamination

30% 0.63 0.82*

- 40% 0.47 0.74*

- 50% 0.32* 0.66** The value is estimated.

Influence of Delaminations

Delamination at the most critical position in the main spar cap.

Size refers to percent of cap width.

Defect Size Load Carrying Capacity

No defect - 1.00

Delamination

30% 0.91

- 40% 0.86

- 50% 0.81

Influenced only by strength reduction in the transverse direction.

Buckling stability increased by the aerodynamic shell.

Reliability Assessment of Blades Containing Defects

Reliability of blade containing defects:

The limit state function cannot be used directly.

factor characteristic load can be increased before blade failure (characteristic material properties)

Influence of specific defects introduced by reduction factor on the load

carrying capacity.

, , ,R Lg X R X L max maxσ ε E D

intact

, ,R Lg X R X L

max maxσ ε E

Reliability Assessment of Blades Containing Defects

Reliability of main spar with delamination at the most critical position.

Target reliability index in IEC 61400-1, = 3.09, PF = 10-3 per year.

Defect Size PF

No defect - 1.910-3 2.90

Delamination 30% 4.010-3 2.65

- 40% 6.210-3 2.50

- 50% 1.010-2 2.33

1FP

Non Destructive Inspection

Updated probability of failure:

Defects are assumed perfect repaired.

0 0.2 0.4 0.6 0.8 10

0.2

0.4

0.6

0.8

1

Delamination size s [m]

Cum

ulat

ive

Val

ue [

-]

Without NDIWith NDI

0 0.2 0.4 0.6 0.8 10

0.2

0.4

0.6

0.8

1

Delamination size s [m]

Cum

ulat

ive

Val

ue [

-]

PoD-curve

Defect Size PF

No defect - 1.910-3 2.90

Delamination | NDI

30% 2.210-3 2.85

- 40% 2.210-3 2.84

- 50% 2.310-3 2.84

Probability of Detection (PoD)

Delamination size without/with NDI

| , intact

defect

0 |

0 1 |

F defect NDIP P g P NDI

P g P NDI

Conclusion

• Probabilistic models for the distribution of defects have been proposed and can be calibrated to observations.

• Probability of failure increased by large delaminations.(strength reduction in transverse direction is estimated)

• Updating of PF by Non Destructive Inspection.

Future work

• Quantify uncertainties in stochastic models for reliability.

• Local defects influence on material properties in fatigue.

Distribution of Defects in Wind Turbine Blades &

Reliability Assessment of Blades Containing Defects

Authors:

Henrik Stensgaard Toft, Aalborg UniversityKim Branner, Risø-DTUPeter Berring, Risø-DTUJohn Dalsgaard Sørensen, Aalborg University / Risø-

DTU