Fracture mechanics techniques for the design of structural components with adhesive joints for wind turbines. Authors: Iñaki Nuin,Carlos Amézqueta, Daniel Trias, Javier Estarriaga, Marcos del Río, Ana Belén Fariñas, EWEC09 Marseille, 18 March 2009
Table of contents • Why dealing with Fracture Mechanics?. • Let’s introduce the problem. • Fracture Mechanics approach. • VCCT approach. • VCCT approach. In-house code. • Application scenario 1. • Application scenario 2. • Conclusions – Future work. • Acknowledgements.
Why dealing with Fracture Mechanics? • Years ago, CENER was involved in a 40 meter length glass fiber epoxy blade. • Guidelines reading. • Design scenarios: Static and fatigue. • For static loads: • Fiber Failure: Common theories. (MAX.STRAIN / TSAI-WU…) • Matrix Failure: General agreement. (PUCK / LARC03-04…) • ¿How to deal with bonding lines?. • For fatigue loads: • Detailed S-N approach for glass and carbon epoxy / polyester composites. (GL guidelines) • ¿How to deal with bonding lines?
Why dealing with Fracture Mechanics? • GL guideline: • Static: • 7 MPa: Limit defined for the characteristic shear stresses. • It covers stress concentration factors up to a factor of 3.0. • Fatigue: • 1 MPa: Limit defined for the equivalent constant-range spectrum for 107 load cycles. • It covers stress concentration factors up to a factor of 3.0. NOTES: The adhesive must be approved by GL. The bonding lines must not include discontinuities (fatigue).
Why dealing with Fracture Mechanics? • DNV guideline: • Difficult to match real local stresses with numerical analyses. • Due to simplifications. • Due to FEM-meshing effects. • It is necessary to combine analytical with testing approach. • Purpose: • Update the predicted resistance of the joint with the results from the tests. • Gain knowledge.
Why dealing with Fracture Mechanics? • Testing and field experience: • Adhesive failure may happen… • Comment from a Blade manufacturer: • The most difficult part of the manufacturing process is trying to bond the two shells together. • Trailing edge defects can grow to full blade failure. • Bonding problem is the biggest issue.
Let's introduce the problem • Stress approach. • Local stress levels dependent on the mesh size. • As element size gets smaller, local stress gets higher. • No reliable method for bonded components design. • If we refine the mesh…..¿when do we stop?.
Fracture Mechanics approach • History: • Theoretical concepts developed at the beginning of 20th century. • First real applications for the industry in the eighties. • From 1995 till today it is commonly used. • Concept: • Specially well-suited for brittle behaviour. • Provides concepts which fill the gap between micro-scale and real component dimensions. • Energy based analysis: Stable solution for local effects. • Based on crack propagation analysis. Combinations mixed modes
Fracture Mechanics approach • Energy release rate (G): Elastic energy released when the defect grows one unit of area. • The critical value for G is a material property. It is common that: • GIc < GIIc < GIIIc : Normalized tests. • The crack grows under a pure mode deformation if: • G > Gic with i=I, II, III. • For mixed modes, there are different approaches which try to deal with an equivalent G value.
Fracture Mechanics approach • How can we measure it? • FCEM: Finite crack extension method. (two analyses) • Based on Griffith balance. • CCT: Crack closure technique. (two analyses) • Energy necessary for the crack to grow = External work needed for the crack to close. • VCCT: Virtual crack closure technique. (one analysis) • Based on the auto-similarity concept.
VCCT approach • Numerical model definition. • Adhesive paste is substituted by linear springs. • The stiffness of each spring considers: • Bonded area. • Elastic modulus of the adhesive (modified by Hooke’s laws). • Thickness of the adhesive layer.
VCCT approach • Stable solution. • As element size gets smaller, G reaches a stable value. ¡¡…a reliable method for bonded components design!!
FEM model with rigid links at the adhesive area (rbe2) • FEM model modification: • Equivalent stiffness • Adhesive properties • Bonding paste thickness • Bonding area. Pre-process(PATRAN) Modified model (including adhesive behaviour) (MATLAB) NASTRAN model NASTRAN model (MATLAB) Stresses Post-process(PATRAN) G calculation (VCCT) Crack initiation criterium RESULTS (Crack stability) Critical areas definition (crack initiation) VCCT approach. In-house Code
VCCT approach. In-house Code In-house developed software. User interface.
VCCT approach. In-house Code • FMAC. • STEP -1- • FEM model definition. Rigid links for bonding areas . • Adhesive elastic properties, critical energy release rate (GIc, GIIc, GIIIc) and thicknesses definition. • Automatic definition of the modified model. NASTRAN analysis. • STEP -2- • Critical areas definition attending to stress criterion or other factors (manufacturing problems…) • Crack definitions. • Automatic definition of the cracked model. NASTRAN analysis. • STEP -3- • GI, GII, GIII calculation by VCCT approach. • Failure indexes definition.
Application Scenario 1 • Let’s imagine we must estimate the ultimate static load for a metallic component which is bonded to a composite panel: • Load direction; 45º Tests performed at CENER. • How can we proceed?....Let’s go step by step.
Application Scenario 1 • STEP -1-: Material Characterization (elastic properties). • Steel: • Mechanical elastic properties are well known. • Young modulus: 210000MPa • Poisson ratio: 0.3 • Composite panel: • 3 point bending test to obtain the flexural modulus. • Biaxial strain gauge to define Poisson ratio. • Flexural modulus: 7972MPa • Poisson ratio: 0.088 • Adhesive (BETAMATE 7014/7065H) • Universal traction tests. • Elastic modulus: 3.1MPa • Poisson ratio: 0.45 Tests performed at CENER.
Application Scenario 1 • STEP -2-: Gc testing for the bonding interfaces. • Steel-adhesive interface: • ASTM D3433 standard. Tests performed at CENER.
Application Scenario 1 • STEP -2-: Gc testing for the bonding interfaces. • Steel-adhesive interface: • Huge dispersion for Maximum load results (4787N – 5411N). • Different values of G depending on the standard approach: • Considering the DCB specimen FEM model and FCEM, CCT & VCCT approaches:
Application Scenario 1 • STEP -2-: Gc testing for the bonding interfaces . • Composite-adhesive interface: • ASTM D3433 standard. Tests performed at CENER.
Application Scenario 1 • STEP -2-: Gc testing for the bonding interfaces. • Composite-adhesive interface: • Huge dispersion for Maximum load results (276.9N – 466.7N). • Different values of G depending on the standard approach: • Considering the DCB specimen FEM model and FCEM, CCT & VCCT approaches:
Application Scenario 1 • STEP -2-: Gc testing for the bonding interfaces. • Depending on the standard, the values of G are quite scattered: • ASTM D3433 and “Classical Beam Theory” approaches do not consider adhesive paste stiffness. • Rigid adhesives (epoxy). • Small thickness of the bonding layer. • “Orthotropic Theory” and “Modified Classical Beam Theory” take into account shear in plane effects of the adherents. • “Adhesive Theory” considers the adhesive layer stiffness. • FCEM, CCT and VCCT theories are based on FEM models. As a consequence the values for Gc, are supposed to consider all these global effects. • When designing a real bonded component, it is necessary to compare the values of G in between analogous approaches.
Application Scenario 1 • STEP -3-: Ultimate load estimation. • The lowest value of Gc defines the de-bonding interface. • A FEM model is defined considering real test scenario. Linear analyses are performed under different load magnitudes.
Application Scenario 1 • STEP -4-: Test Correlation. • Two tests were performed. • Problems with adhesive cure cycle for one component. • So… only one test result available for comparison. Test failure load is 11400N, 21% higher than predicted value (9428N)
Application Scenario 2 • Let’s compare VCCT approach and Cohesive elements technique against a 3 point bending test of an I-Beam: • Tests performed at WMC facilities. UPWIND project. • …Let’s go step by step.
Application Scenario 2 • STEP -1-: Material Characterization. UD Reinforcement (Flanges) MD Reinforcement (Web) Adhesive
Application Scenario 2 • STEP -2-: FEM models definition. • MSC.MARC. • Linear material behaviour. • Large displacements assumption. • Cohesive elements to simulate the adhesive interface with glass fiber laminates (UD &MD). • 3D laminate properties (out of plane characterization).
Application Scenario 2 • STEP -2-: FEM models definition. • MSC.NASTRAN. • Linear material behaviour. • Small displacements assumption. • VCCT technique defined via in house developed software (FMAC). • 3D orthotropic properties (calculated from laminate properties).
Application Scenario 2 • STEP -3-: Failure prediction – Correlation with test. MSC.MARC (Cohesive Elements) MSC.NASTRAN (VCCT) • Critical local points for both models are located at the same area. • MSC.MARC: First bonding failure under 40.6KN load. • MSC.NASTRAN: First bonding failure under 48.1KN load. • Test Failure 47.6KN……just a coincidence!!
Conclusions – Future work • Conclusions. • Fracture mechanics approach is confirmed as a reliable method when designing bonded components. • VCCT approach predicts the possibility of one defect to start growing… nothing about how it grows (cohesive elements). • Nevertheless, due to bonding process complexity and uncertainties, it is difficult to estimate accurately bonded joints capacity. • Ignorance factors must be considered. • Future work. • In-house code development: • Spring model development (coupled behaviour). • Non-linear behaviour implementation. • Validation test plans: • ENF specimen tests performance. • Mixed mode tests performance. • Subcomponent tests.
Acknowledgements • UPWIND WP3 partners. • ALSTOM-ECOTECNIA wind power department.