You are on page 1of 6

Renewable Energy xxx (2013) 1e6

Contents lists available at SciVerse ScienceDirect

Renewable Energy
journal homepage: www.elsevier.com/locate/renene

Fracture mechanics approach for failure of adhesive joints in wind


turbine blades
Y.M. Ji*, K.S. Han
Graduate School of Wind Energy, Pohang University of Science and Technology, Pohang 790-784, Republic of Korea

a r t i c l e i n f o a b s t r a c t

Article history: Composite components of wind turbine blade are assembled with adhesive. In order to assess structural
Received 25 February 2013 integrity of blades it is needed to investigate fracture of joints. In this study, nite element analysis based
Accepted 3 July 2013 on fracture mechanics was conducted to characterize failure of adhesive joint for wind turbine blade. The
Available online xxx
cohesive zone model as proposed fracture mechanics approach was veried through the comparison of
numerical results with experimental data. Finite element models of wind turbine were developed to
Keywords:
predict damage initiation and propagation. Numerical results based on fracture mechanics showed that
Fracture mechanics
failure was initiated in the edge of the adhesive bond line due to high level of shear stress prior to
Adhesive joints
Wind turbine blades
reaching the extreme design loading and propagated progressively.
Finite element analysis 2013 Elsevier Ltd. All rights reserved.
Cohesive zone model

1. Introduction results showed that the aerodynamic skins debonding of the ad-
hesive joints is the initial failure mechanism causing a progressive
Energy crisis and global warming have led to a higher demand collapse of the blades.
for renewable energy. Wind energy is expected to one of the Thus engineering approaches for adhesive joint have been of
important energy among renewable energy in the future. An ef- great interest. In Ref. [3] a three point bending test for asymmetric
cient way to further improve the performance of wind turbine is to beam was conducted in order to study the adhesive performance
reduce the weight of the blades. Large scale wind turbine blades are between shear web and spar cap of the wind turbine rotor blade,
mainly based on ber reinforced polymer (FRP) composite for light investigating thick bond-lines. In Ref. [4] composite I-beams were
weight. numerically and experimentally investigated to examine the me-
The wind turbine blade is aerodynamic structure which consists chanical behavior of adhesive bond lines. A nite element model
of skin and webs. Adhesive joining method is used to assembly the was developed to predict and simulate the damage initiation and
parts of the blade, as shown Fig. 1. Adhesive joints have fewer evolution. In Ref. [5] cohesive laws were used for prediction the
sources of stress concentrations, higher toughness and more uni- load carrying capacity of medium size adhesive joint specimens
form stresses distribution through the joined area compared with subjected to four point exure. The scaling from small specimens to
mechanically fastened joints, such as rivet and screw. Adhesive medium-size specimens was successfully achieved. Although these
bond lines in large scale turbine blades are thick and much longer studies have made effort to investigate behavior of adhesive joint,
than in other applications. Web to skin bond line continues along there are differences between subcomponents and actual blades.
the entire length of the blade. Adhesive joints of blade do not have regular thickness and undergo
According to the studies concerning failure of wind turbine complex load case. Consideration of the actual geometry and load
blade, debonding is one of the main causes of blade failure. In Ref. condition of blades is required to predict accurate strength of ad-
[1] a post mortem investigation was carried out on the failed sec- hesive joints.
tions of blade. Adhesive joint failure observed between skin and The present study aims to verify the numerical results for ad-
spar. In Ref. [2] an actual collapse testing was conducted under the hesive joint damage compared with the experiments and to pro-
ap-wise loading for a large full-scale wind turbine blade. The pose the fracture mechanics approach to describe the fracture
behavior of adhesive joints for wind turbine blades. Adhesive joint
* Corresponding author. Tel.: 82 54 279 8216; fax: 82 54 279 2845. failure of wind turbine blade was investigated by means of Cohe-
E-mail address: cosmo@postech.ac.kr (Y.M. Ji). sive Zone Model approach.

0960-1481/$ e see front matter 2013 Elsevier Ltd. All rights reserved.
http://dx.doi.org/10.1016/j.renene.2013.07.004

Please cite this article in press as: Ji YM, Han KS, Fracture mechanics approach for failure of adhesive joints in wind turbine blades, Renewable
Energy (2013), http://dx.doi.org/10.1016/j.renene.2013.07.004
2 Y.M. Ji, K.S. Han / Renewable Energy xxx (2013) 1e6

Fig. 1. Structural detail of wind turbine blade.


Fig. 3. Geometrical conguration of specimen.
2. Theoretical background
hardened at room temperature (w25  C) for 24 h. The specimen has
In this work, fracture mechanics approach was used to charac- length of 150 mm, width of 13 mm and thickness of adhesive is
terize the initiation and propagation of damage at adhesive joint. 1 mm. Non-bonded region is placed at the one end of the specimen.
The Cohesive Zone Model (CZM) was employed to describe material The dimensions are schematically depicted in Fig. 3.
separation under a static loading. The CZM combines a strength-
based failure criterion to predict the damage initiation and a frac- 3.2. Experimental test procedure
ture mechanics-based criterion to determine the damage propa-
gation [6]. The stress transferred across the crack faces of the The test set-up is shown in Fig. 4. The bending moment was
fracture process zone can be described in terms of cohesive laws. applied by loading on top of the specimen. Constant cross head
Fig. 2 shows that schematic of a CZM for failure prediction of speed was 0.5 mm per minute. The applied load and displacement
adhesively bonded joints. It is assumed that the damage initiation in the mid-span of specimens were recorded up to failure.
criterion will depend on traction stress by current state of stress [7].
The damage initiation is dened by 3.3. Experimental results
 2  2  2
sn st ss The crack initiated and propagated along the interface between
1 (1)
snmax stmax ssmax the adhesive layer and the compressive adherend as shown in
Fig. 5. The crack propagated to the mid-span of the specimen at
where sn is the normal stress to surface of adhesive layer, and st and 263 kN with a drop in the load. In subsequent specimen inspection,
ss are the shear stress components along adhesive layer. it was observed that adherends remained elastic during the entire
The failure criterion is expressed as a power law: test without any plastic deformation. This indicates that all the
 n  n  n energy dissipation during the crack initiation and propagation
GI GII GIII process was contributed by fracture of adhesive.
1 (2)
GIC GIIC GIIIC Typical loadedisplacement curves obtained from the experi-
mental test are shown in Fig. 6. In the current fracture test, a critical
where GIC, GIIC and GIIIC are the critical values of fracture energy in strain energy release rate is dened the GII value when the
the three modes; normal and two tangential, respectively. maximum load is reached. The experimental value of critical strain
energy release rate GIIC can be calculated according to Eq. (3) [8]. It
was obtained 0.8 kJ/m2 of GIIC from test results.
3. Experimental and numerical failure analysis of adhesive
joints
9P 2 Ca2
GIIC  (3)
3.1. Specimen preparation 2b 2L3 3a3

In this study, failure behavior of adhesive joint was investigated where P is critical load at crack propagation, a is non-bounded
by end notched adhesive joint specimen. The adherend was made length measured from support point, b is specimen width, C is
of glass ber reinforced epoxy composite laminates by vacuum specimen compliance and L is one-half support spacing.
infusion and post-cured. The lay-up of the laminates was [03]. The
structural epoxy paste adhesive which is applied in the wind tur- 3.4. Numerical method
bine blade was used to bond together two adherends. It was
A numerical model of specimen was developed using two-
dimensional nite element (FE) method. The material properties

Fig. 2. Schematic of a CZM for failure prediction of adhesively bonded joints. Fig. 4. Set-up of three point bending experiment.

Please cite this article in press as: Ji YM, Han KS, Fracture mechanics approach for failure of adhesive joints in wind turbine blades, Renewable
Energy (2013), http://dx.doi.org/10.1016/j.renene.2013.07.004
Y.M. Ji, K.S. Han / Renewable Energy xxx (2013) 1e6 3

Fig. 5. Crack growth path (a) and fracture surface (b).

Fig. 6. Boundary condition and meshed geometry of the specimen.

Table 1
Fig. 8. Loadedisplacement curves for experimental test and FE analysis.
Composite properties.

Properties Composite

Elasticity modulus in ber direction, E11 (MPa) 41,400


of the adherend and adhesive are demonstrated in Tables 1 and 2.
Elasticity modulus transverse to ber, E22 (MPa) 13,560
In-plane tangential elasticity modulus, G12 (MPa) 3500
4-Node bilinear with reduced integration with hourglass control
In-plane Poisson ratio, n12 0.311 element (CPS4R) and 3-node linear element (CPS3) were applied to
the FE-model.
CZM approach was used to consider damage initiation and
propagation. Cohesive properties were placed at adherende
Table 2
Adhesive properties for the cohesive zone model.

Properties Adhesive

Elasticity modulus, E (MPa) 4200


Poisson ratio, n 0.45
Tensile strength, snmax (MPa) 12
Shear strength, stmax ssmax (MPa) 48
GIC (kJ/m2) 0.4
GIIC GIIIC (kJ/m2) 0.8

Fig. 7. Numerical predictions for deformation and the shear stress distribution in
specimen at the point at which the damage has just begun to grow. Fig. 9. Global model and submodels of blade.

Please cite this article in press as: Ji YM, Han KS, Fracture mechanics approach for failure of adhesive joints in wind turbine blades, Renewable
Energy (2013), http://dx.doi.org/10.1016/j.renene.2013.07.004
4 Y.M. Ji, K.S. Han / Renewable Energy xxx (2013) 1e6

Fig. 10. Global model and submodel 1 under ap-wise loading.

adhesive interfaces. For damage initiation, the shear strength is 3.5. Numerical analysis results
stmax ssmax 48 MPa and the tensile strength is snmax 12 MPa
[5]. The fracture energies of eq (2) are equal to GIC 0.4 J/m2 [9] and Results obtained from the FE model analysis showed that
GIIC GIIIC 0.8 J/m2. The power n takes the value 1. cohesive damage initiation occurred at 18% of maximum load. Fig. 7
Fine mesh was used near the contact area because disconti- shows the shear stress on adherendeadhesive interface at the point
nuities in the material properties cause high stress-gradients in at which the damage begun to grow. It was observed that high
this region as shown Fig. 7. The numerical analysis was con- shear stress concentration occurred in the adherendeadhesive
ducted by ABAQUS 6.12-1 including geometrical non-linearity. interface at crack location. Therefore, the interface at which stress

Fig. 11. Stresses on adhesive interfaces (at 27% of the extreme design load).

Please cite this article in press as: Ji YM, Han KS, Fracture mechanics approach for failure of adhesive joints in wind turbine blades, Renewable
Energy (2013), http://dx.doi.org/10.1016/j.renene.2013.07.004
Y.M. Ji, K.S. Han / Renewable Energy xxx (2013) 1e6 5

Fig. 12. Damage propagation and stresses of adhesive layer along Path A and Path B.

concentrated was expected to occur crack initiation and propaga- relative dimensions where the highest mechanical values are reached
tion. The location of max shear stress moved along crack-front. in the global analysis. The main purpose of a global model is to provide
Numerical results are compared with experimental results in representative boundaries for a local model.
Fig. 8. The FE model showed good agreement with experimental Fig. 9 shows FE models developed in this study. The global model
results within 2% difference. The FE model predicted adequately the of full scale blade consists of skins and webs. S4R shell elements
stiffness degradation of the specimen and provided details about were used for the global model. The S4R element is a 4-node rst
the damage initiation and propagation. The CZM approach can be order element of reduced integration scheme. Laminated composite
used to predict failure of adhesive joint for wind blade. material properties were assigned to the model according to the
blade lay-up. The blade root was restricted all degrees of freedom as
4. Numerical modeling of wind turbine blade joint boundary conditions. The distributed load was applied on the
pressure side of blade FE model in the ap-wise direction to assign
Finite element analysis of wind turbine blade has been carried out ap-wise extreme design load. The static nite element analysis was
in order to study failure of the skineweb adhesive joint. The meth- conducted including geometrical nonlinear effects.
odology is based on the globalelocal FE approach [10]. In globalelocal Two local models were developed after identifying critical re-
FE approaches, the size of the local model is chosen by considering the gion from global model results. First model which is denoted as

Please cite this article in press as: Ji YM, Han KS, Fracture mechanics approach for failure of adhesive joints in wind turbine blades, Renewable
Energy (2013), http://dx.doi.org/10.1016/j.renene.2013.07.004
6 Y.M. Ji, K.S. Han / Renewable Energy xxx (2013) 1e6

Submodel 1 was developed to obtain a three dimensional stress adhesive-joint location inside the suction shell of the blade and
solution for adhesive joint. It represents blade root region which is debonding from the adhesive joints is the initial failure mechanism.
from 3% to 7% of full length of the blade. The adhesive layer was Even though, debonding propagation was limited in the end
included in the model as an independent part to consider detail of area of adhesive joint, it was clearly demonstrated that stress was
the skineweb joint. The adhesive layer has a nominal width of redistributed on adhesive bond line during debonding propagation.
150 mm and a nominal thickness of 10 mm. S4R shell elements Stress redistribution could offer actual stress state to predict failure
were used for skin and C3D8R solid/continuum elements were used which can result in structural instability.
for webs and adhesive. C3D8R element typology consists of an 8-
node reduced integration linear element. The material data for the 6. Conclusion
spar cap and the adhesive line was assumed to be the same as used
in specimen analysis. Boundary conditions are transmitted from the The present work has been concerned with adhesive joint fail-
global model to the local model. ure under the ap-wise loading for a wind turbine blade. The
Second local model which is denoted as Submodel 2 was cohesive zone model approach is adopted to predict damage
developed to describe failure of adhesive joint. The C3D8R elements initiation and propagation in FE model of skineweb adhesive joint.
were used for Submodel 2. The CZM parameters for the adhesive The results show that stress concentration occurs in adhesive joint
were taken from the previous specimen analysis and applied on and shear stress is mainly responsible for debonding.
both interfaces of adhesive. It was observed that debonding was initiated in the edge of the
adhesive bond line prior to reaching the extreme design load.
5. Wind turbine blade results and discussion Damage was propagated in the blade tip direction along adhesivee
web interface with the load increase. Damage propagation led to
Fig. 10 shows deformed global model and stress state of Sub- stress redistribution.
model 1 under ap-wise extreme design load. Blade tip deected in It is concluded that skin debonding from adhesive joint is initial
ap-wise direction. Suction side of blade was compressed and failure mechanism which can lead to earlier failure of blade before
pressure side was tensioned. Detailed three dimensional stress in reaching ultimate design load. Therefore, it is needed to consider
the skineweb adhesive joint is shown in Submodel 1. Stress con- realistic adhesive geometry and load condition to verify adhesive
centration occurred at the end of webs and adhesive bond line. joint performance and identify possible debonding region is recom-
The stresses in the adhesive prior to failure initiation are shown mended in blade design. Proposed fracture mechanics approach can
in Fig. 11. The stresses have higher values at the edges of joint. The be used to study behavior of adhesive joints for wind turbine blade.
magnitude of stresses increases moving in the root direction.
Magnitude of peel stress is low due to compression of web while Acknowledgment
shear stresses are high. At adhesiveeweb interface maximum shear
stress occurred along edge of length direction whereas at adhe- This work was supported by the project Development of the
siveeskin interface maximum shear stress was observed at the web weight optimized blade design technology by the Korea Govern-
laminate location along the width and higher than at adhesivee ment Ministry of Knowledge Economy and by the Korea Institute of
web interfaces. These results are due to non-linear behavior of web Energy Technology Evaluation and Planning grant funded by Korea
under ap-wise loading and shear stress difference along thickness government Ministry of Knowledge Economy.
direction. The skin side interface has the high possibility of
debonding according stress distributions.
References
Fig. 12 shows damage propagation with increase of load. Skin
debonding started from adhesive joint corner at 27% of design load. [1] Jensen FM, Falzon BG, Ankersen J, Stang H. Structural testing and numerical
Damage propagated to tip direction gradually. Fig. 12a shows damage simulation of a 34 m composite wind turbine blade. Composite Structures
2006;76(1e2):52e61.
state at 100% design load. Red area is corresponding to damaged
[2] Yang JS, Peng CY, Xiao JY, Zeng JC, Xing SL, Jin JT, et al. Structural investigation
region. Adhesiveeskin interface was almost debonded at 140% load of composite wind turbine blade considering structural collapse in full-scale
while adhesiveeweb interface was not failed and laminate did not static tests. Composite Structures 2013;97:15e29.
reach failure stress until debonding was fully developed. [3] Sayer F, Antoniou A, van Wingerde A. Investigation of structural bond lines in
wind turbine blades by sub-component tests. International Journal of Adhe-
Fig. 12c and d shows stresses at damage state 30%, 70% and 100% sion and Adhesives 2012;37:129e35.
of design load along Path A and Path B. Compressive stress was [4] Zarouchas DS, Makris AA, Sayer F, Van Hemelrijck D, Van Wingerde AM. In-
dominant on two paths in normal direction. Thus, peel stress was vestigations on the mechanical behavior of a wind rotor blade subcomponent.
Composites Part B e Engineering 2012;43(2):647e54.
not observed. The paths were remained under compressive stress [5] Sorensen BF, Goutianos S, Jacobsen TK. Strength scaling of adhesive joints in
due to behavior of web after debonding whereas shear stresses polymerematrix composites. International Journal of Solids and Structures
were concentrated at corner of bond line. Shear stresses reduced in 2009;46(3e4):741e61.
[6] Li S, Thouless MD, Waas AM, Schroeder JA, Zavattieri PD. Use of a cohesive-
debonding region and maximum shear stress was moved to zone model to analyze the fracture of a ber-reinforced polymerematrix
boundary undamaged region. composite. Composites Science and Technology 2005;65(3e4):537e49.
In this study, FE-models captured blade ap-wise bending and [7] Sadowski T, Knec M, Golewski P. Experimental investigations and numerical
modelling of steel adhesive joints reinforced by rivets. International Journal of
web compression as shown in other studies [1,2]. This behavior
Adhesion and Adhesives 2010;30(5):338e46.
effects on stress concentration at web adhesive joint. Shear stress [8] Mandell JF, Cairns DS, Samborsky DD, Morehead RB, Haugen DJ. Prediction of
occurred signicant at end of joint. Non uniform deformation of delamination in wind turbine blade structural details. Journal of Solar Energy
Engineering e Transactions of the ASME 2003;125(4):522e30.
thick adhesive layer induces different shear stress distribution at
[9] Nuin I, Amzqueta C, Trias D, Estarriaga J, Ro M, Farias AB. Fracture mechanics
interfaces. techniques for the design of structural components with adhesive joints for
The results of CZM approach showed that debonding was prior wind turbines. Proceedings European wind energy conference, Marseille; 2009.
to laminate failure. It can be seen that debonding could lead to [10] Reinoso J, Blazquez A, Estefani A, Paris F, Canas J, Arevalo E, et al. Exper-
imental and three-dimensional globalelocal nite element analysis of a
earlier failure before reaching design load. It is accordance with the composite component including degradation process at the interfaces.
results of Ref. [2] that debonding trace can be observed at the Composites Part B e Engineering 2012;43(4):1929e42.

Please cite this article in press as: Ji YM, Han KS, Fracture mechanics approach for failure of adhesive joints in wind turbine blades, Renewable
Energy (2013), http://dx.doi.org/10.1016/j.renene.2013.07.004

You might also like