comptest 2013 - book of abstracts · emission, optical fiber sensors) open a very broad field of...

CompTest 2013 - Book of Abstracts Editors: O.T. Thomsen, B.F. Sørensen, C. Berggreen

CompTest 2013

6th International Conference on Composites Testing and Model Identification

22-24 April 2013

Department of Mechanical and Manufacturing Engineering, Aalborg University, Denmark

ISBN: 87-91464-49-8

6th International Conference on Composites Testing and Model IdentificationO.T.Thomsen, Bent F. Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

CompTest 2013 - Book of Abstracts Editors: O.T. Thomsen, B.F. Sørensen, C. Berggreen

CompTest 2013

6th International Conference on Composites Testing and Model Identification

22-24 April 2013

Department of Mechanical and Manufacturing Engineering, Aalborg University, Denmark


Aalborg, 2013

Copyright © 2013 Department of Mechanical and Manufacturing Engineering, Aalborg University, Denmark

ISBN: 87-91464-49-8


Aalborg, 2013

ACKNOWLEDGEMENTS

The CompTest 2013 organising committees wish to thank the following organisations and companies for their contribution to the success of the conference:

i


Aalborg, 2013

CONFERENCE SCOPE

CompTest 2013 was held 22-24 April 2013 in Aalborg, Denmark. Previous conferences in this series of conferences were held in Châlons en Champagne (France) in January 2003, Bristol (UK) in September 2004, Porto (Portugal) in April 2006, Dayton, Ohio (USA) in October 2008, and lastly Lausanne (Switzerland) in February 2011.

This aim of CompTest 2013 has been to bring together the International scientific community working in the field of testing and modeling of composite materials and structures. It is well accepted that testing such heterogeneous and anisotropic materials and structures raises a number of challenges to researchers, such as the identification of numerous parameters, the development of specific test fixtures (shear, compression, fracture toughness), or the control of parasitic effects. As a consequence, the development of testing and model identification procedures is broadly recognized as an interesting and important area.

Moreover, recent developments in optical whole-field measurement techniques (speckle interferometry, digital image correlation, among others) and in-situ damage monitoring (acoustic emission, optical fiber sensors) open a very broad field of investigation. Testing and identification procedures for composites which have been developed over the last few decades based on limited local strain measurements have to be adapted to make full use of the enormous amount of data that whole-field methods provide. Apart from the large general composites conferences (ICCM, ECCM), there are very few occasions to exchange information on the topic of composites testing and model identification.

The focus of CompTest 2013 encompasses all issues related to identifying parameters for modelling the mechanical and physical behaviour of composite materials. Particular attention will be given to innovative identification procedures, interactions between testing and modelling, in-situ damage monitoring, and the use of whole-field measurements. A multitude of composite materials are addressed including polymers, cements, ceramics, metallic matrices, carbon fibres, glass fibres, natural fibres, and wood.

22 March 2013

Ole Thybo Thomsen

Conference Chair, CompTest 2013

ii


Aalborg, 2013

ORGANISATION OF COMPTEST 2013

Conference Chairs:

Prof. Ole Thybo Thomsen, Chairman Aalborg University, Denmark & University of Southampton, UK

Prof. Bent F. Sørensen, Co-Chairman Technical University of Denmark, Denmark

Assoc. Prof. Christian Berggreen, Co-Chairman

Technical University of Denmark, Denmark

Prof. Fabrice Pierron, Co-Chairman Arts et Métiers ParisTech

Prof. Michael Wisnom, Co-Chairman University of Bristol, UK

International Scientific Committee:

Prof. Janice Barton University of Southampton, United Kingdom

Dr. Bill Broughton NPL, United Kingdom

Dr. Mark Battley University of Auckland, New Zealand

Prof. John Botsis EPFL, Switzerland

Prof. Pedro Camanho University of Porto, Portugal

Prof. Josep Costa Balanzat University of Girona, Spain

Dr. Peter Davies IFREMER, France

Dr. Carlos Davila NASA Langley Research Center, USA

Dr. Endel Iarve AFRL, USA

Prof. Frédéric Jacquemin GeM laboratory, Nantes, France

Dr. Alastair Johnson DLR, Germany

Prof. Masamichi Kawai University of Tsukuba, Japan

Prof. Javier Llorca IDMEA Spain

Prof. Stepan Lomov Katholieke Universiteit Leuven, Belgium

Dr. David Mollenhauer AFRL, USA

Prof. Adrian Mouritz RMIT, Australia

Prof. Ozden Ochoa Texas A&M University, USA

iii


Aalborg, 2013

Prof. Wim van Paepegem Ghent University, Belgium

Prof. Ivana Partridge Cranfield University, United Kingdom

Prof. Marino Quaresimin University of Padua, Italy

Prof. Paul Robinson Imperial College London, United Kingdom

Prof. Janis Varna Luleå University of Technology, Sweden

Prof. Dan Zenkert Royal Institute of Technology, Stockholm, Sweden

Local Organizing Committee:

Dr. Kim Branner Technical University of Denmark, Denmark

Prof. Jesper de Claville Christiansen Aalborg University, Denmark

Assoc. Prof. Lars R Jensen Aalborg University, Denmark

Assoc. Prof. Jørgen Kepler Aalborg University, Denmark

Prof. Ryszard Pyrz Aalborg University, Denmark

Assoc. Prof. Jens Chr. Rauhe Aalborg University, Denmark

Assoc. Prof. Jan Schjødt-Thomsen Aalborg University, Denmark

iv


Aalborg, 2013

COMPTEST PROGRAM – OVERVIEW

v


Aalborg, 2013

vi

6th International Conference on Composites Testing and Model Identification O.T.Thomsen, Bent F. Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

TABLE OF CONTENTS Acknowledgements .................................................................................................................................. i

Conference Scope .................................................................................................................................... ii

Conference Organisation ........................................................................................................................ iii

Conference Program ................................................................................................................................. v

Table of contents ................................................................................................................................... vii

PLENARY LECTURES .................................................................................................................... 1

‘Mechanics of Ultra-thin Ply Laminates’, Pedro Camanho ........................................................................................................................................ 3

‘Issues in Characterising and Modelling the Response of Aerospace Composites to Fire’, Geoff Gibson ............................................................................................................................................ 5

‘Wind Turbine Blade Materials and Structures – State-of-the-art and Future Developments’, Torben K. Jacobsen .................................................................................................................................. 7

ORAL SESSIONS ............................................................................................................................... 9

Oral session 1 – Damage and fracture (5 presentations) .................................................................. 11

‘Validation of FEM Based Damaged Laminate Model Measuring Crack Opening Displacement in Cross-ply Laminate Using Electronic Speckle Pattern Interferometry (ESPI)’, M.S. Loukil, J. Varna, Z. Ayadi ............................................................................................................. 13

‘Influence of Transverse Cracks on the Onset of Delamination, Application to L-angle Composite Specimens’, F. Laurin, A. Mavel, E. Auguste ........................................................................................ 15

‘Modelling the Size-dependent Mode I Translaminar Fracture Toughness of Unidirectional Fibre-reinforced Composites’, S.T. Pinho, S. Pimenta .................................................................................... 17

‘Mixed-mode Translaminar Fracture: Failure Analysis, Fractography and Numerical Modelling’, M.J. Laffan, S.T. Pinho, P. Robinson ..................................................................................................... 19

‘Damage Mechanisms of 3D Woven Hybrid Composites Loaded in Tension. Testing, Inspection and Simulation’, R. Muñoz, C. González, J. Llorca ............................................................................... 21

Oral session 2 – Modelling & constitutive behaviour (5 presentations) .......................................... 23

‘Embedded Element Method in Meso-finite Element Modeling of Textile Composites’, S.A. Tabatabaei, S.V. Lomov, I. Verpoest ............................................................................................. 25

‘Development of Constitutive Material Model for Composite with Nonlinear Fibers and Matrix’, L Pupure, R. Joffe, J. Varna ................................................................................................................... 27

‘Validation of Simulated Unidirectional Composites Microstructure: Statistical Equivalence to Real Fibre Arrangements’, V.S. Romanov, L. Gorbatikh, S.V. Lomov, I. Verpoest ............................. 29

vii

‘Cyclic Deformation of Polyethylene/Clay Nanocomposites: Observations and Constitutive Modeling’, A.D. Drozdov, J.deC. Christiansen, R. Klitkou ................................................................... 31

‘Experimental Characterization and Modeling of Thin Ply-size Effect’, R. Amacher, J. Cugnoni, J. Botsis .......................................................................................................... 33

Oral session 3 – Fracture and fatigue -1 (5 presentations) ............................................................... 35

‘Use of DIC for the Failure Analysis of Complex Composite Structures’, G. Crammond, S.W. Boyd, J.M. Dulieu-Barton .................................................................................... 37

‘Debond Growth Assessment in GFRP-BALSA Sandwich Structures’, E. Farmand-Ashtiani, N. Nasri, J. Cugnoni, J. Botsis ............................................................................ 39

‘Explicit Expressions for the Crack Length Correction Parameters for the DCB, ENF, and MMB Tests on Multidirectional Laminates’, S. Bennati, P. Fisicaro, P.S. Valvo ................................. 41

‘Stress-strain Analysis of a Carbon PPS During and After Fatigue Loading Conditions’, W. van Paepegem, I.. De Baere, S. Daggumati, C. Hochard, J. Xu, S.V. Lomov, I. Verpoest, J. Degrieck ...... 43

‘X-ray Tomography Assessment of Damage During Tensile Deformation of ±45° Carbon Fiber Laminates’, F. Sket, A. Enfedaque, C. Alton, C. González, J.M. Molina-Aldareguia, J. Llorca ........... 45

Oral session 4 – Fracture and fatigue – 2 (4 presentations) ............................................................. 47

‘Probabilistic Anisomorphic Constant Fatigue Life Diagram Approach to Prediction of P-S-N Curves for Composites’, M. Kawai, K.-I. Yano ..................................................................................... 49

‘Automated Delamination Length Video Tracking in Static and Fatigue DCB Test’, F. Lahuerta, S. Raijmaekers, J.J. Kuiken, T. Westphal, R.P.L. Nijssen ................................................. 51

‘Effect of the Crack Length Monitoring Technique During Fatigue Delamination Testing on Crack Growth Data’, D. Sans, J. Renart, J.A. Mayugo, J. Costa ........................................................... 53

‘Biaxial Fatigue Testing of Glass/Epoxy Composite Tubes’, M. Quaresimin, P. Carraro ..................... 55

Oral session 5 – Defects, delamination and debonding (4 presentations) ........................................ 57

‘Tension and Compression Testing of Multi-directional Laminates with Artificial out of Plane Wrinkling Defects’, S.R. Hallett, M.I. Jones, M.R. Wisnom ................................................................. 59

‘Effects of In-plane Waviness on the Properties of Carbon Composites – Experimental and Numerical Analysis’, J.-P. Fuhr. J. Baumann, F. Härtel, P. Middendorf, N. Feindler ........................... 61

‘Experimental Validation of a Matrix Crack Induced Delamination Criteria’, L. Zubillaga, A. Turon, J. Costa, S. Mahdi, P. Linde ............................................................................. 63

‘Determination of fiber/Matrix Interface Debond Growth Parameters from Cyclic Loading of Single Fiber Composites’, A. Pupurs, J. Varna, P. Brøndsted, S. Goutianos ......................................... 65

Oral session 6 – Damage and dynamic properties (4 presentations) ............................................... 67

‘Full-field Curvature Measurements to Assess Impact Damage in Composite Plates Using an Indicator Based on Mechanical Equilibrium’, C. Devivier, F. Pierron, M.R. Wisnom ......................... 69

viii


Aalborg, 2013

‘Calibration of LS-DYNA Strain Rate Dependent Composite Material Models’, R. Eriksen, C. Berggreen, J.M. Dulieu-Barton ....................................................................................... 71

‘Damage Accumulation Investigation in Fiber-reinforced Polymer-matrix Composites: From Test Coupons to Structural Elements’, A.J. Brunner ............................................................................. 73

‘DCB Fracture Specimens with side Notches’, H. Toftegaard, M. Rask, S. Rasmussen, B.F. Sørensen ......................................................................... 75

Oral session 7 – Test methods (3 presentations) ................................................................................ 77

‘Novel Test Setup for Determination of High Temperature Mechanical Properties of Composites’, A. Chripunow, M. Ruder ........................................................................................................................ 79

‘Static Strin and Deformation Controlled Testing of Composite Beams’, J. Høgh, J. Waldbjørn, H. Stang, C. Berggreen, J. Wittrup-Schmidt, K. Branner ................................. 81

‘The Use of Digital Image Correlation for full Field Analysis of Polymer Foams’, R.M. Stubbing, M. Battley ..................................................................................................................... 83

Oral session 8 – Health/condition monitoring (3 presentations) ...................................................... 85

‘Self-sensing of Damage in Carbon Nanotube Vinyl ESTER Composites’, J. de J. Ku-Herrera, A. May-Pat, F. Avilés ............................................................................................ 87

‘Influence of the Protective Coating of Fiber Bragg Grating Sensors on the Structural Distorsion and Sensing Accuracy when Embedded in Fiber Reinforced Polymers’, N. Lammens, G. Chiesura, G. Luyckx, E. Voet, J. Degrieck ................................................................. 89

‘Monitoring Strain Gradients in Adhesive Composite Joints by Embedded Fibre Bragg Grating Sensors’, L.P. Canal, B.D. Manshadi, V. Michaud, J. Botsis, G. Violakis, H.G. Limberger ................ 91

Oral session 9 – Fracture and fatigue – 3 (5 presentations) ............................................................. 93

‘On the Use of In-situ SEM Testing and Simulation to Study Deformation and Failure Mechanisms in Composite Materials’, C. González, L.P. Canal, J. Segurado, J. Llorca ....................... 95

‘Bio-based Composites with Different Moisture Contents Under Static and Dynamic Loading’, N. Doroudgarian, M. Anglada, A. Mestra, R. Joffe ............................................................................... 97

‘Failure Mode Specific Fatigue Testing of Nanoparticle-modified CFRP Under VHCF-loading’, J.B. Knoll, R. Koschichow, I. Koch, K. Schulte, M. Gude .................................................................... 99

‘Experimental and Numerical Analysis of Skin/Stiffener Debonding Under Bending’, I. Urresti, A. Barrio, J. Renart, L. Zubillaga......................................................................................... 101

‘Quantification of Damage Due to Environmental Conditions on Carbon Fibre/Epoxy Composite Samples’, E. Guzman, J. Cugnoni, T. Gmür ........................................................................................ 103

Oral session 10 – Testing, material concepts and joining (4 presentations) .................................. 105

‘Secundary Stress Effects During Load Introduction into Unidirectional Composite Test Coupons’, L.P. Mikkelsen, J.I. Bech ..................................................................................................................... 107

ix

‘Discontinuous-ply Composites for Enhanced Ductility, S. Pimenta, P. Robinson’, G. Czel, M.R. Wisnom, H. Diao, a. Bismarck ........................................... 109

‘The Effect of Pre-bond Moisture and Temperature on the Fracture Toughness of Bonded Joints for Composite Repairs’, S. Budhe, A. Rodríquez-Bellido, J. Renart, J. Costa .......................... 111

‘Extracting the Strain Softening Response of Composites Using a Detailed Finite Element Analysis as a Virtual Digital Image Correlation Technique’, N. Zobeiry, A. Forghani, R. Vaziri, A. Poursartip, X. Xu, S.R. Hallett, M.R. Wisnom ...................... 113

POSTER SESSIONS ....................................................................................................................... 115

Poster session 1 - Damage and Failure (9 presentations) ................................................................ 117

‘Model-based Damage Identification in Composite Structures Using Spatial Wavelet Transform’, A. Katunin ............................................................................................................................................ 119

‘Characterisation of Bridging Mechanisms in a Single Z-pinned Composite Laminate’, M Yasaee, J.K. Lander, G. Allegri, S.R. Hallett .................................................................................. 121

‘Choice of an Analytical Scheme in Correlating Strain Energy Release Rate, Crack Length and Opening of the Faces of an Adhesively Bonded, Thick Composite DCB Specimen’, A. Bernasconi, A. Jamil........................................................................................................................ 123

‘CFRP Fatigue Testing and Issues for Aeronautical Applications’, V. Dattoma, R. Nobile, F.W. Panella ................................................................................................... 125

‘Effect of Ply Thickness on the Fatigue Delamination Growth in Tapered Laminates: Measurements and Analysis’, S. Giannis, C. Jeenjitkaew.................................................................... 127

‘Identification of Damage Modes in Ceramic Matrix Composites by Acoustic Emission Signal Pattern Recognition’, N. Godin, M.R’Mili, P. Reynaud, G. Fantozzi .................................................. 129

‘Lifetime Prediction with Acoustic Emission During Static Fatigue Tests on Ceramic Matrix Composite at High Temperature Under Air’, N. Godin, M.R’Mili, P. Reynaud, G. Fantozzi ............. 131

‘High-speed DIC for Blast Testing of Composite Panels’, S. Giversen, C. Berggreen, B. Riisgaard .............................................................................................. 133

‘The Effect of Various Diffuse Damage Levels on the Transverse Cracking Evolution for T700/M21 Cross-ply Laminated Composites’, H. Nouri, D. Traudes, G. Lubineau ........................... 135

Poster session 2 – Manufacturing/processing & materials characterisation (11 presentations) ............................................................................................................................... 137

‘On the Use of Digital Image Correlation to Determine the Permeability and Compaction Law of Fabrics in Vacuum Infusion Process’, J. Vilà, C. González, J. Llorca ............................................ 139

‘Fabric Permeability Testing and Their Use in Infusion Simulation’, J. Sirtautas, A.K. Pickett ......... 141

‘Shrinkage and Thermal Expansion Model for a Glass/Epoxy Laminate’, J. Jakobsen, J.H. Andreasen, E.A. Jensen, O.T. Thomsen ................................................................... 143

‘Examination of Compression and Shear Properties of Glass/Carbon Hybrid Laminated Composites’, S.A. Oshkovr, M. Rezaei, C.M. Markussen, T.L. Andersen, F. Aviles, O.T. Thomsen ....................................................................................................................................... 145

x


Aalborg, 2013

‘Evaluation of the Mechanical Properties of Polymer Concretes Under Various Conditions’, K.-C. Jung, J.-H. Bae, S.-H. Chang ...................................................................................................... 147

‘Geometrical Characterization & Micro-structural Modelling of Short Steel Fiber Reinforced Composites’, Y. Abdin, S.P. Lomov, A. Jain, H. van Lente, I. Verpoest ............................................ 149

‘Strain Gauge Application in Soft Material Testing’, S. Zike, L.P. Mikkelsen ................................... 151

‘Methodology of Material Parameters Identification in Sandwich Panels Versus Computer Simulation’, M. Chuda-Kowalska, A. Garstecki .................................................................................. 153

‘Validation of Short Dynamic Specimen Geometry for Identification of Rate Dependent Model on a Large Range of Strain Rates’, J. Berthe, M. Brieu, E. Deletombe ............................................... 155

‘Integration of Microstructured Optical Fibres into Carbon Fibre Reinforced Plastic Materials – Determination of the Initial Strain State’, C. Sonnenfeld, G. Luyckx, F. Collombet, Y-H. Grunevald, B. Douchin, L. Crouzeix, M. Torres, S. Sulejmani, T. Geernaert, K. Chah, P. Mergo, H. Thienpont, F. Berghmans ............................................................................................... 157

‘A Micro-computer Tomography Technique to Study the Interaction Between the Composite Material and an Embedded Optical Fiber Sensor’, G. Chiesura, G. Luyckx, N. Lammens, W. van Paepegem, J. Degrieck, M. Dierick, L. van Hoorebeke ........................................................... 159

Poster session 3 – Material concepts, modelling and applications (10 presentations) .................. 161

‘A Study on the Low-velocity Impact Characteristics of Impact Limiter Materials for Nuclear Spent Fuel Transport Cask’, J.H. Kim, K.B. Shin, W.S. Choi ............................................................. 163

‘Post-fire Mechanical Properties of Marine Sandwich Composites’, L. Tranvan, V. Legrand, P. Casari, F. Jacquemin ................................................................................ 165

‘Development of Automated Finite Element Models for Large Wind Turbine Blades’, M. Peeters, W. van Paepegem .............................................................................................................. 167

‘Bend-twist Coupling Identification in Composite Beams’, V. Fedorov, C. Berggreen ...................... 169

‘A New Damage Tolerant Design Approach for Sandwich Panels Loaded in Fatigue’, G. Martakos, J.H. Andreasen, O.T. Thomsen ...................................................................................... 171

‘Changes in Mechanical Behaviour of a Glass Fibre Reinforced Epoxy by Adding Polyamide 6 Nano-fibres’, I. De Baere, B. De Schoenmaker, S. van der Heijden, W. van Paepegem, K. de Clerck .......................................................................................................................................... 173

‘Pseudo-grain Discretization and Full Mori Tanaka Formulation for Random Heterogeneous Media: Predictive Abilities for Stresses in Individual Inclusions and the Matrix’, A. Jain, S.V. Lomov, Y. Abdin, I. Verpoest, W. van Paepegem.......................................................... 175

‘SMA/GFRP Composite Plates: Passive Damping and Interface Strength’, M. Bocciolone, M. Carneale, A. Collina, N. Lecis, A. Lo Conte, B. Previtali, C.A. Biffi, P. Bassani, A. Tuissi ......... 177

‘Mechanical and Morphological Properties of Talc Filled High Density Polyethylene’, A.K. Mehrjerdi, M. Skrifvars ............................................................................................................... 179

‘Mixed Mode Delamination in Hybrid Laminate Under DMMB Test’, J.-W. Kang, O.-H. Kwon, J.-H. Kwak.................................................................................................. 181

xi

AUTHOR INDEX .............................................................................................................................. 183

xii


Aalborg, 2013

PLENARY LECTURES

11


Aalborg, 2013

MECHANICS OF ULTRA-THIN PLY LAMINATES

Pedro P. Camanho1, Albertino Arteiro1, Giuseppe Catalanotti1 and António R. Melro1

1DEMec, Faculdade de Engenharia da Universidade do Porto Rua Dr. Roberto Frias, 4200-465 Porto, Portugal

Email: [email protected], web page: http://www.fe.up.pt

Keywords: Composite materials, thin-ply laminates, notched strength.

ABSTRACT

A combined experimental and numerical investigation of the mechanical response of a new class of advanced composite materials manufactured using thin plies is presented. These materials are manufactured by a process that continuously and stably opens the fibre tows. The manufacturing process is able to produce flat and straight plies with dry ply thicknesses as low as 0.02 mm. Analysis methods based on micromechanical and mesomechanical models are developed to study the effects of the ply thickness on the loads required to start delamination and transverse cracking. The analysis models are based on cohesive elements and on appropriate material models for the fibre and for the polymer resin. Tensile and compressive tests in both unnotched and notched specimens are performed using two different lay-ups [1]. The notched tests are based on specimens with central cracks and with circular holes, loaded in tension and compression. Digital image correlation is used to monitor the onset and propagation of damage on the surface plies. The results show that the lay-up with blocked plies and with higher differences in fibre orientation angles between consecutive plies has lower unnotched strength. However, due to the larger fracture process zone observed in the notched tests, such lay-up has marginally higher notched strengths. A size effect on the strength is observed for both the open-hole tension and compression tests. The size effect and the associated notch sensitivity of thin non-crimp fabrics are similar to those observed in typical aerospace grade unidirectional pre-impregnated composite materials. It is also concluded that the thin laminates exhibit an improved response to bolt-bearing loads over traditional composite materials

REFERENCES [1] A. Arteiro, G. Catalanotti, J. Xavier, P.P. Camanho, Notched response of non-crimp fabric thin-

ply laminates, Composites Science and Technology, in press, 2013 (doi: 10.1016/j.compscitech.2013.02.001).

33


Aalborg, 2013

ISSUES IN CHARACTERISING AND MODELLING THE RESPONSE OF AEROSPACE COMPOSITES TO FIRE

Geoff Gibson

School of Mechanical and Systems Engineering, Newcastle upon Tyne NE1 7RU, UK Email: [email protected]

Keywords: fire behaviour, aerospace, carbon fibre composites, thermal properties

ABSTRACT Their complex behaviour is a significant hindrance to modelling the response of composites to fire. This involves resin decomposition, gas evolution and non-ideal transport properties. Moreover there is a lack of useful information on these properties. This paper discusses modelling issues and describes how high quality information for modelling can be obtained from small-scale, low cost tests.

CHARACTERISING AND MODELLING FIRE BEHAVIOUR UNDER LOAD Fire testing under load can be accomplished with a simple loading frame, using a calibrated heat flux from a propane burner, as in Fig. 1a. Fig. 1b shows the thermal profiles through the thickness of a laminate and Fig. 2a shows a set of test results, expressed as stress-rupture curves.

(a) (b)

Figure 1. (a) compressive test under load and (b) measured and modelled thermal profiles at 116 kWm-2. The most important effect on fire exposure is resin decomposition. This is modelled either by an Arrhenius law or by the assumption that the decomposition state is a simple function of temperature. The decomposition endotherm needs to be taken into account. Also, volatile decomposition products travel towards the hot face, removing heat against the thermal gradient. The relationship that provides basic modelling capability is the Henderson Equation1-3, which, in one-dimensional form is:

GGGCPP hx

MhhQt

MxT

kxt

TC (1)

where: T, t and x are temperature, time and through-thickness coordinates, respectively. , CP and k are the density, specific heat and conductivity of the composite. MG is the mass flux of volatiles. hC

and hG are the respective enthalpies of the composite and evolved gas. QP is the endothermic decomposition energy. The three terms on the right relate respectively to heat conduction, resin decomposition, and convective heat transport by volatiles. The material parameters in Equ. 2 all evolve as functions of temperature and resin decomposition. This model has been integrated into a one-dimensional finite difference package, COM-FIRE, which contains measured or default values of the thermal parameters for a range of common composite materials.

5

A.G. Gibson

The main problem modeling structures is the incompatibility between Equ. 1 and FE packages, due to the decomposition term and the gas flow. Two methods have been proposed to avoid these problems. The first is ‘apparent thermal diffusivity’, where the decomposition endotherm is incorporated into an apparent specific heat. This neglects gas flow, but gives temperature-dependent values of density, specific heat and conductivity, for use in FE. However it is not always possible to ignore gas flow, especially with carbon fibre composites. Here, it can be observed that there is a temperature, usually about 300°C, below which there is no decomposition and the gas flow may be neglected. It is then possible, depending on the complexity of the structure, to use a hybrid approach, with COM-FIRE modelling heat flow near to planar or near-planar heated surfaces and FE for the remainder. Fig. 2b shows an example, with a skin-stringer assembly exposed on the skin side. COM-FIRE was used to model the heat flow up to the 300°C isotherm. This was then used as the input for modelling the remainder of the structure with ANSYS. Equ. 2 is an empirical model for the variation of any property with temperature and decomposition. In fire, temperature is the most important variable, since CFRP in compression, loses most of its strength before the resin starts to decompose. The degree of decomposition, R, (1 for intact resin and 0 for full decomposition) mainly determines properties after fire.

(2) P(T) is the property; PU and PR are the low and high temperature values; T’ is the transition temperature and k the transition breadth. Not much is known about the effect of decomposition, but the results so far correlate with n=1-3 for compressive strength. Having found the modulus of each ply the bending stiffness and hence the buckling strength can be found. To predict strength, an average1 of the strength through the plies may be used, as in Fig 2a, which gives a slight overestimate. The alternative is to model the full stress-strain curve with temperature, which requires a lot of data.

Fig. 2. (a) Measured and modelled stress rupture curves (b) hybrid model for a skin-stringer system after 180s subject to a lower surface heat flux of 116 kWm-2.

This continuing work is supported by the EU 7th Framework FIRE-RESIST Partnership, the aim of which is to develop new concepts for lightweight, fire-resistant composite materials. REFERENCES 1. Mouritz, A.P., Feih, S., Kandare, E., Mathys, Z., Gibson, A.G., DesJardin, P.E., Case, S.W. and

Lattimer, B.Y. (2009) Review of Fire Structural Modelling of Polymer Composites. Composites A, 40: 12, 1800-1814.

2. Gibson, A.G., Browne, T.N.A., Feih, S. and Mouritz, A.P. (2012) Modelling Composite High Temperature Behaviour and Fire Response under Load. J. of Composite Materials. 46: 2005-2022.

nRURU RTTkPPPPTP tanh5.0

6


Aalborg, 2013

WIND TURBINE BLADE MATERIALS AND STRUCTURES – STATE-OF-THE-ART AND FUTURE DEVELOPMENTS

Torben Krogsdal Jacobsen

LM Windpower A/S Østre Allé 1, DK-6640 Lunderskov, Denmark

Email: [email protected], web page: http://www.lmwindpower.com

Keywords: Wind, Rotor Blades, Materials technology, Testing, Design

ABSTRACT

Wind energy is more than ever competing with other sources of energy. The technology has now matured to a stage where on-shore grid parity has been reached. Further optimizations are still emerging and every year the cost of energy reduces. One of the key components for the wind turbine is the rotor blade. Stiffer and lighter blades at a balanced price and performance ratio remain a constant focus area for the industry. Therefore, the choice of materials technology for rotor blades is not only driven by the direct cost of materials, but also the processing cost and product reliability.

Composite materials development is focused on decreasing the cost of specific stiffness and expanding the strength design limits. Strength limits can be expanded by either using more costly material systems or by more advanced testing and design methods that allow the designer to go closer to the strength limits. Probabilistic design methods are starting to emerge and may eventually replace the current partial coefficient design approach.

This paper will discuss and show examples of the current state-of-the-art testing and design methods for composites used in rotor blades. Furthermore, identify some of the research directions in composite materials, testing and design methods that could enable reduction in the cost of energy of future wind turbine power plants.

77


Aalborg, 2013

ORAL SESSIONS

99


Aalborg, 2013

Oral session 1 – Damage and fracture (5 presentations)

1111


Aalborg, 2013

VALIDATION OF FEM BASED DAMAGED LAMINATE MODEL MEASURING CRACK OPENING DISPLACEMENT IN CROSS-PLY LAMINATE USING ELECTRONIC

SPECKLE PATTERN INTERFEROMETRY (ESPI)

M.S. Loukil1,2, J. Varna1 and Z. Ayadi2

1Department of Engineering Sciences and Mathematics, Luleå University of Technology

SE-97187 Luleå, Sweden Email: [email protected], web page: http://www.ltu.se

2Institut Jean Lamour, Université de Lorraine, EEIGM 6 Rue Bastien Lepage, F-54010 Nancy Cedex,

France web page: http://www.ijl.nancy-universite.fr

Keywords: Damaged Laminate, Crack opening displacement (COD), ESPI, FEM

ABSTRACT

Composite laminates during service undergo complex combinations of thermal and mechanical loading leading to microdamage accumulation in the plies. The most common damage mode and the one examined in this work is intralaminar cracking in layers. The crack opening displacement (COD) and the crack sliding displacement (CSD) during loading reduce the average stress in the damaged layer, thus reducing the laminate stiffness. Finite element method (FEM) studies were performed in [1,2] to understand which material and geometry parameters affect the COD and CSD most and simple empirical relationships (power law) were suggested. All these studies and analysis assume a linear elastic material with idealized geometry of cracks. The only correct way to validate these assumptions is through experiments. The effect of material properties on COD was measured experimentally using optical microscopy of loaded damaged specimens in [3,4]. It was shown that the measured average values of COD are affected by the constraining layer orientation and stiffness. The experimental determination of the average COD and CSD needs the measurement of the displacement for all points of the crack surfaces, which justifies the use of full-field measurement technique, Electronic Speckle Pattern Interferometry (ESPI). ESPI is an optical technique that provides the displacement for every point on a surface and offers the possibility to measure both, the in-plane and out-of-plane displacement without surface preparation. This technique was used in [5,6] to measure the COD for inside cracks on the specimen’s edge. It was shown that the profile of the crack on the edge is elliptical. The main objective of this paper is to study cracks in surface layers by measuring the COD along the crack path. For this reason the cross-ply laminate with surface cracks was selected. In particular, the displacement field on the surface of a [903/0]S carbon fiber/epoxy laminate specimen with multiple intralaminar cracks is studied and the relative displacement dependence on the applied mechanical load is measured. By looking to the displacement field the cracks appear as singularities and the corresponding displacement jumps are directly related to COD and CSD. The transverse cracks are parallel to the fiber orientation in the layer, which in our case corresponds to a 90 direction with respect to the tensile axis. Consequently, there is no relative sliding of the crack faces and the only displacement of these crack faces corresponds to COD. In other words, the crack displacement discontinuity measured on the surface is directly the COD.

1313

M.S. Loukil, J. Varna and Z. Ayadi

0.0

0.2

0.4

0.6

0.8

1.0

0.00 0.05 0.10 0.15 0.20

No

rma

lize

d T

hic

kn

ess

Nomalized COD (μm/Mpa)

Experimental resultsFEM results 0.000.020.040.060.080.100.120.140.160.18

0 0.2 0.4 0.6 0.8 1 1.2

No

mal

ized

CO

D (μ

m/M

pa)

Normalized Width

Experimental resultsFEM results

Figure 1. The COD as a function of the normalized thickness

Figure 2 . The COD along the normalized width of the sample

In this work, the crack opening displacement profile along the thickness of the damaged 90° layers is investigated and compared with FEM. A very good agreement is shown in Fig1. To study edge effects on the opening displacement, the COD is measured along the crack path and the results are presented in Fig 2.

REFERENCES

[1] P. Lundmark and J. Varna, Constitutive relationships for laminates with ply cracks in in plane loading, International Journal of Damage Mechanics, 14 (3), 2005, pp. 235-261

[2] P. Lundmark and J. Varna, Crack face sliding effect on stiffness of laminates with ply cracks, Composite Science and Technology, 66, 2006, pp. 1444–54

[3] J. Varna, L. A. Berglund, R. Talreja and A. Jakovics, A study of the crack opening displacement of transverse cracks in cross ply laminates, International Journal of Damage Mechanics, 2, 1993, pp. 272–89

[4] J. Varna, R. Joffe, N. V. Akshantala and R. Talreja, Damage in composite laminates with off-axis plies, Composite Science and Technology, 59, 1999, pp. 2139-2147

[5] L. Farge, Z. Ayadi and J.Varna,Optically measured full-field displacements on the edge of a cracked composite laminate, Composite. Part A, 39, 2008, pp.1245-1252

[6] L. Farge, J. Varna and Z. Ayadi, Damage characterization of a cross-ply carbon fiber/epoxy laminate by an optical measurement of the displacement field, Composite Science and Technology, 70, 2010, pp. 94-101

1414


Aalborg, 2013

INFLUENCE OF TRANSVERSE CRACKS ON THE ONSET OF DELAMINATION, APPLICATION TO L-ANGLE COMPOSITE SPECIMENS

F. Laurin1, A. Mavel1 and E. Auguste2

1Composite materials and structures department, Onera - The French Aerospace Lab 29 av. Division Leclerc, F- 92322 Châtillon Cedex 3

Email: [email protected], web page: http://www.onera.fr

2Airbus Operations SAS, 316 Route de Bayonne, 31060 Toulouse Cedex 9, France

Email: [email protected], web page: http://www.airbus.com

Keywords: Composite L-angle specimens, Transverse cracks, Delamination, Identification

ABSTRACT

The use of composite materials, and especially unidirectional laminated composites, is increasing in civil and military aerospace structures due to their very interesting ratio mass / stiffness / strength compared to more conventional metallic solutions. These materials are now used for the manufacturing of primary structures (such as centre wing box, wings, fuselage...) that ensure the structural integrity of the airplane. Most of these structures are subjected to complex three-dimensional loading essentially carried-out by composite laminated L-angle specimens which permit to joint the different perpendicular composite panels. These L-angle composite specimens are subjected to (i) in-plane loadings (transferred by the jointed different panels) but also to (ii) moments that tend to unfold the specimens leading to delamination in the corner radius of the specimens. A procedure for the identification of out-of-plane tensile and shear strengths, obtained performing tests on laminated composite L-angle specimens, has already been proposed at ONERA [1]. This was done for test cases where no transverse cracking is observed prior the final fracture of the specimen due to delamination. However, for cross-ply laminates subjected to uniaxial tensile loading, it has been shown experimentally that local-delaminations are generated at the tips of transverse cracks [2]. In the case of L-angle laminated specimens subjected to unfolding loading, these micro-delaminations induced by transverse cracking lead to a weakening of the interface and thus decrease the strength of the considered structures (premature failure caused by delamination since the interface has been pre-damaged by the local-delamination located at the tips of the transverse cracks). In order to accurately predict the strength of such composite structures, it is essential to introduce into the proposed modelling the coupling between the transverse cracking and the delamination. This coupling between intralaminar and interlaminar damage is already available in some advanced damage models [2-4]. However, the identification of the parameters of the coupling between inter/intra damages currently remains a scientific challenge. Therefore, the objective of this study consists in proposing an experimental setup which permits to identify the influence of transverse cracking on the onset of delamination, and especially on the out-of-plane tensile strength. Four-point bending tests on laminated L-angle specimens have already been used to identify the out-of-plane tensile strength [1]. The positions of the spans in this four-point bending experimental device have been optimized, through finite element simulations, to maximize the in-plane stresses in the lower part of the radius of the specimen in order to generate transverse cracking before the final failure of the structure due to delamination. These tests were instrumented with acoustic emission (to monitor the evolution of the transverse cracks), digital images correlation applied on one side of the specimens and microscopic observations (while the loading is maintained) on the other side of the specimen to measure the evolution of the transverse cracks density (see Fig. 1.b). The measured evolutions of the crack density for the different stacking sequences were compared with predictions of the multiscale damage and failure approach proposed by ONERA [4]. This

1515

First A. Author, Second B. Author and Third C. Author

advanced damage approach has already been validated through comparisons with unnotched laminated specimens subjected to uniaxial tensile loading [5]. In the previous test campaign [1], the final failure of L-angle specimens subjected to four-point bending loading, in which no in-plane damage was observed prior final failure, was due to delamination observed only on the more loaded interface, located at mid-thickness in the radius. In this study, a coupling between the transverse cracking and the delamination is clearly observed on the failure pattern as reported in Fig. 1.b. The delamination is indeed observed both at the most loaded interface and at the interface located close to the most in-plane damaged ply. The predictions of the advanced damage and failure approach in terms of locations of delamination and failure loads are compared with experimental results to determine the predictive capabilities of such model. Moreover, additional unfolding tests on laminated L-angle specimens have been performed to validate the proposed approach and the associated identification procedure concerning the coupling between the transverse crack density and the onset of delamination.

Figure 1: (a) Presentation of the four-point bending experimental device for composite laminated L-angle specimens and (b) observation of the transverse cracks in L-angle specimens prior the final

failure due to delamination.

REFERENCES

[1] J.-S. Charrier, N. Carrere, F. Laurin, T. Bretheau, E. Goncalves-Novo, S. Mahdi, Proposition of 3D progressive failure approach and validation on tests cases, Proceedings of the 14th European Conference on Composite Materials, Budapest Hongry, 07-10 June 2010.

[2] C. Huchette, Analyse multiéchelle des interactions entre fissurations intralaminaire et interlaminaire dans les matériaux composites stratifiés. Doctorate thesis, University of Paris VI, 2005.

[3] E. Abisset, F. Daghia, P. Ladevèze, On the validation of a damage mesomodel for laminated composites by means of open-hole tensile tests on quasi-isotropic laminates, Composites Part A, 42(10), 2011, pp. 1515-1524.

[4] F. Laurin, N. Carrere, C. Huchette, J.-F. Maire, A multiscale hybrid damage and failure approach for strength predictions of composite structures, Journal of Composite Materials, 2013, accepted in the framework of the World Wide Failure Exercise III-Part A.

[5] F. Laurin, C. Carrere, C. Huchette, J.-F. Maire, A multiscale hybrid damage and failure approach for strength predictions of composite structures, Proceedings of the 15th Conference on Composite Materials, Venise Italy, 24-28 June 2012.

1616


Aalborg, 2013

MODELLING THE SIZE-DEPENDENT MODE I TRANSLAMINAR FRACTURE TOUGHNESS OF UNIDIRECTIONAL FIBRE-REINFORCED COMPOSITES

S T Pinho1, S Pimenta1

1Department of Aeronautics, Imperial College London South Kensington Campus, London SW7 2AZ, United Kingdom

Email: [email protected], web page: http://www.imperial.ac.uk/people/silvestre.pinho

Keywords: Translaminar toughness, mode I, stochastic, fibre failure, fractal, experiments, predictions

ABSTRACT We present a model for the translaminar tensile toughness of UD FRPs, based on fibre and interfacial properties and assuming the formation of statistical fractal fracture surfaces. The translaminar tensile toughness of a UD composite ( ) is the energy required to fracture the material perpendicularly to the fibre direction (per unit nominal [or projected] area fractured). This property governs the damage tolerance of structures with load-aligned fibres, as well as the strength of real components with geometric discontinuities. The translaminar toughening mechanisms of FRPs have been extensively investigated (as reviewed by Kim and Mai [1]), and methods to measure the corresponding fracture toughness have been developed (as recently reviewed by Laffan et al. [2]). All studies concluded that composites are orders of magnitude tougher than their constituents; this is due to the formation of intricate 3D fracture surfaces (Figures 1 and 2), featuring not only mode-I fibre and matrix fracture, but also large interfacial debonds and pulled-out fibres and bundles [1-3].

(a) 0.125 mm thick 0° plies with = 65 kJ/m2. (b) 0.250 mm thick 0° plies with = 132 kJ/m2.

Figure 1: Size effects on the translaminar fracture toughness of UD carbon-epoxy plies [3].

(a) Fibre bundle failure in a recycled CFRP [4]. (b) UD ply failure in a multidirectional CFRP [3] Figure 2: Hierarchical and quasi-fractal features on the fracture surface of UD composites.

1717

S. T. Pinho and S. Pimenta

(a) Fractal failure model [5]. (b) Comparison between predictions [4] and experiments [3,7,8] for the translaminar

fracture toughness of several CFRP. Figure 3: Fractal model the fracture of UD composites.

The model we present is based on a fractal description of the failure process (Figure 3a); at each level, debonding between the bundles is allowed through an idealised fracture mechanics model [5]. The stochastic strength of each bundle is predicted using a hierarchical model with stochastic fibre strength and local load sharing between the tows [6]. Once the failure morphology is predicted, the fracture energy is calculated by adding the contributions due to debonding (of the different tows) and friction (during pullout). The results (Figure 3b) compare favourably with experimental data available in the literature for different carbon-epoxy systems.

REFERENCES [1] J.-K. Kim and Y.-W. Mai, High strength, high fracture toughness fibre composites with

interface control | A review, Composites Science and Technology, 41(4), 1991, pp. 333-378. [2] M. J. Laffan, S. T. Pinho, P. Robinson, and A. J. McMillan, Translaminar fracture toughness

testing of composites: A review, Polymer Testing, 31(3), 2012, pp. 481-489. [3] M. J. Laffan, S. T. Pinho, P. Robinson, and L. Iannucci, Measurement of the in situ ply

fracture toughness associated with mode I fibre tensile failure in FRP. Part II: Size and lay-up effects, Composites Science and Technology, 70(4), 2010, pp. 614-621.

[4] S. Pimenta, S. T. Pinho, P. Robinson, K. H. Wong and S. J. Pickering, Mechanical analysis and toughening mechanisms of a multiphase recycled CFRP, Composites Science and Technology, 70(12), 2010, pp. 1713–1725.

[5] S. Pimenta and S. T. Pinho, An analytical model for the translaminar tensile toughness of fibre composites with statistical fractal fracture surfaces, In preparation for publication, 2012.

[6] S. Pimenta and S. T. Pinho, Hierarchical scaling law for the strength of composite fibre bundles, Submitted to Journal of the Mechanics and Physics of Solids, 2012.

[7] S.T. Pinho, P. Robinson, and L. Iannucci, Fracture toughness of the tensile and compressive fibre failure modes in laminated composites, Composites Science and Technology, 66(13), 2006, pp. 2069–2079.

[8] R. Teixeira and S.T. Pinho, Translaminar fracture toughness of laminates composites. 15th European Conference on Composite Materials. Venice, Italy, 25-28 June, 2012.

1818


Aalborg, 2013

MIXED-MODE TRANSLAMINAR FRACTURE: FAILURE ANALYSIS, FRACTOGRAPHY AND NUMERICAL MODELLING

M. J. Laffan1, S. T. Pinho1, P. Robinson1

1 Department of Aeronautics, Imperial College London

South Kensington, SW7 2AZ, UK Email: [email protected]

Email: [email protected], web page: http://www3.imperial.ac.uk/people/silvestre.pinho Email: [email protected], web page: http://www3.imperial.ac.uk/people/p.robinson

Keywords: Translaminar fracture, Mixed-mode, Fractography, Modelling

ABSTRACT

Experimental investigation of mode I translaminar, fibre-breaking, fracture (see Figure 1) has concluded that the failure mode can be treated as a self-similar crack at the laminate level that can be quantitatively characterised as a fracture process in terms of fracture energies (GIc) [1] or a traction-separation law [2]. This laminate-level material property encompasses the energy dissipated by all micro-level damage mechanisms such as fibre-matrix debonding, fibre fracture and fibre pull-out. Whilst the experimentally obtained GIc has proved useful for numerically simulating translaminar failure [3], pure mode I loading represents a narrow portion of the design spectrum that an engineering component might see during its operational lifetime. A complete understanding of the translaminar fracture behaviour of composite laminates under a range of loading conditions is required for numerical design. Thus far, relatively few studies have experimentally investigated the translaminar fracture behaviour of notched laminates under mixed-mode loading and, to the knowledge of the authors, none has explored the possible routes for numerically simulating this failure mode. This paper will present work that aims to address the questions that arise in both these areas. Firstly, we will present an experimental investigation of mixed-mode translaminar fracture with the aim of identifying the relevant micro-mechanisms of failure through detailed fractographic analysis, the conclusions of which will provide the insight necessary for devising an appropriate modelling approach. A mixed mode compact tension specimen and fixture have been developed, shown in Figure 2, such that fracture tests can be performed on specimens under several mixed mode ratios. An advantage of this new configuration over specimens that have been previously used in the literature is that the crack can be propagated in a stable manner under several mixed-mode loading ratios, therefore allowing for the detailed investigation of damage zone initiation and development. Results will be presented for specimens tested at mixed mode ratios between GII/Gtotal = 0.1 and 1.0.

G /GII total

0.0 0.1 0.2 0.5

Figure 1: Fracture surfaces of failed specimens tested between pure mode I and GII/Gtotal = 0.5.

Increased amounts of splitting and fibre-pull out due to delamination indicate more diffuse damage with increasing proportion of mode II loading.

1919

Matthew J. Laffan, Silvestre T. Pinho and Paul Robinson

Figure 2: Fixture used for mixed-mode translaminar fracture testing.

It will be shown through scanning electron microscopy of the fracture surfaces of failed specimens, Figure 1, that increasing the component of mode II loading increases the amount of delamination and splitting that occurs prior to fibre fracture. This amounts to an increasing amount of fracture energy being dissipated with increasing proportion of mode II loading. While fracture at low proportions of GII can be characterised by a traction-separation law (such as has been done for pure mode I), failure cannot be idealised as a single crack as the proportion of shear loading increases. Secondly, we will present methodologies for numerical simulation of mixed-mode translaminar fracture designed to capture the relevant micro-mechanisms of failure. Results will be presented from simulations using modelling tools already implemented within the commercial finite element analysis package Abaqus and from a newly developed damage model. The model, implemented as a user material within Abaqus, is able to predict the response of the tested specimens and the features of the damage zone using a single layer of 2D elements alone. This new approach significantly reduces the analysis time by eliminating a considerable amount of model pre-processing and by increasing the computational efficiency of the analysis.

REFERENCES [1] M. J. Laffan, S. T. Pinho, P. Robinson and L. Iannucci, Measurement of the in situ ply fracture

toughness associated with mode I fibre tensile failure in FRP. Part II: Size and lay-up effects, Composites Science and Technology, 70, 2010, pp. 614-621

[2] R. Gutkin, M. J. Laffan, S. T. Pinho and P. Robinson, Modelling the R-curve effect and its specimen-dependence, International Journal of Solids and Structures, 48, 2011, pp. 1767-1777

[3] C. G. Dávila, C. A. Rose and P. Camanho, A procedure for superposing linear cohesive laws to represent multiple damage mechanisms in the fracture of composites, International Journal of Fracture, 158, 2009, pp. 211-223

2020


Aalborg, 2013

DAMAGE MECHANISMS OF 3D WOVEN HYBRID COMPOSITES LOADED IN TENSION. TESTING, INSPECTION AND SIMULATION

R.Muñoz1, C.González1,2 and J.Llorca1,2

1IMDEA Materials Institute c/ Eric Kandel, 2, 28906 – Getafe, Madrid, Spain

Email: [email protected], web page: http://www.materials.imdea.org/

2Department of Materials Science. Polytechnique University of Madrid E.T.S de Ingenieros de Caminos 28040, Madrid, Spain Email: [email protected]: www.mater.upm.es

Keywords: Hybrid composites, 3D woven, Finite elements, Unit cell, X-Ray computed

microtomography (XCT)

ABSTRACT Combination of several types of fibres has been shown to be effective in improving the mechanical performance of composite materials, especially against impact loads [1]. Likewise, fibre architecture may have a strong influence on composites response, especially for the case of 3D woven, which typically improve delamination resistance. Previous studies regarding either hybrid composites or 3D composites can be found in literature, but there is lack of data when both factors are combined in one single material. In this work, the mechanical behaviour and failure mechanisms of a 3D woven hybrid orthogonal composite subjected to tensile loads at several angles are discussed. Specimens were inspected at several stages of the stress-strain curve by both X-Ray Computed Tomography and optical microscopy to better understand failure mechanisms. Both geometrical and material properties were used as input in a finite element model of a unit cell, based on a micromechanical approach. As stated above, the material considered is a non-crimp fabric reinforced in the through-thickness direction with a yarn made of Ultra-High-Molecular-Weight Polyethylene (UHMWPE). Layers are laid-up in a cross-ply manner, with a stacking sequence [90,0,90,0,90,0,90], which leads to a laminate 4.1mm thick. The first 4 layers are made of S2-glass fibre, followed by one layer of hybrid Carbon-S2-glass fibres and by other two layers of only carbon, leading to a non-symmetrical laminate. Material was manufactured by VARTM (Vacuum Assisted Resin Transfer Moulding). Before testing the whole composite, bundles of each fibre type were first impregnated and then tested in tension according to ASTM D4018. Different behaviours were observed depending on the fibre type. While highest strains were observed in Dyneema® ( = 4%), carbon fibres showed the strongest and the most brittle behaviour. S2-glass ranked in between in terms of both ductility and strength. Such data was very useful to better understand the behaviour of the 3D woven composite. After that, specimens made of the 3D woven hybrid orthogonal composite were tested in tension at room temperature in an electromechanical machine according to ASTM D3039 at 0º, 45º and 90º. To assess strain field measurement, digital image correlation was used in all cases along with an extensometer. Load was applied quasi-statically under displacement control to specimens of 250x25 mm2. Specimens were tabbed with a plain weave E-glass stacked embedded in an epoxy resin (MTM 28/GF T600) at ±45º. Since the laminate is non-symmetrical, coupling between extension and warping takes place during tensile loading. Specimens tend to bend to the side where glass-fibres are clustered, due to their higher compliance. Two peaks on the stress-strain curve were also observed: one corresponding to the maximum strain of the glass-fibre and another one to the less ductile carbon fibre. Different stress-strain curves were also obtained under longitudinal and transverse loading (Fig.1). A highly non-linear

2121

R. Muñoz, C. González and J. Llorca

stress-strain curve was obtained with specimens tested at 45º, revealing a progressive orientation of fibres in the load direction. Once stress-strain curves were found, a second experimental campaign was carried out. Specimens were impregnated with penetrant liquids for later X-Ray tomography inspection. Loading was interrupted before achieving each peak of strength in the stress-strain curves. After that, several samples were cut from damaged specimens and deeply analysed by means of X-Ray tomography and optical microscopy. Several damage mechanisms were observed, like fiber-matrix debonding, matrix cracking, bundle separation and fibre breakage. Z-yarn seems to act at the same time as stress raiser and crack stopper. Furthermore, it has some influence not only on damage patterns, but also on geometry, by modifying the final cross section of some bundles as well as introducing some crimping.

Figure 1: Stress-strain curve and the corresponding XCT image of a cross section after breakage A finite element model based on a Micromechanics approach was generated to replicate the stress-strain curves obtained during the experimental tests described above. Since the geometry of the material is repeated periodically following a given pattern, a representative volume element was created. A python script was specifically created to automatically generate unit cells, where bundle and resin pockets could vary in size. Geometrical data were obtained from optical microscopy, whereas material properties and volume fractions were estimated from the abovementioned experimental campaign and X-Ray tomography, respectively. Every single bundle was modelled as transversely isotropic elastic up to failure. A continuum damage model based on the LaRCO4 failure criteria [2], where the components of the stress tensor follow a softening law dominated by the material fracture energy has been used. This model is implemented as a user subroutine VUMAT in Abaqus Explicit. Resin pockets were inserted between them and considered as linear elastic isotropic materials. Good correlation between experiments and simulations was found in terms of stress-strain curves, especially for the elastic part. In addition, combination of experiments, advanced damage inspection techniques and numerical tools led to a deep comprehension not only of the failure mechanisms that take place in 3D woven orthogonal hybrid composites under tensile loads, but also of its time sequence.

REFERENCES [1] A. Enfedaque, J.M. Molina-Aldareguía, F. Gálvez, C. González, J. Llorca,

Effect of Glass Fiber Hybridization on the Behavior Under Impact of Woven Carbon Fiber/Epoxy Laminates, Journal of Composite Materials, 44, 2010, pp. 3051-3068 (http://dx.doi.org/10.1177/0021998310369602).

[2] P. Maimí, P.P Camanho, J.A. Mayugo and C.G. Dávila, A continuum damage model for

composite laminates. Part I: Constitutive model, Mechanics of Materials, 39, 2007, pp. 897-908 (http://dx.doi.org/10.1016/j.mechmat.2007.03.005).

2222


Aalborg, 2013

Oral session 2 – Modelling & constitutive behaviour (5 presentations)

2323


Aalborg, 2013

EMBEDDED ELEMENT METHOD IN MESO-FINITE ELEMENT MODELING OF TEXTILE COMPOSITES

S.A. Tabatabaei, Stepan V. Lomov and Ignaas Verpoest

Department of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven, Kasteelpark Arenberg 44, B-3001 Leuven, Belgium

Email: [email protected]

Keywords: Meso-scale, FEM, Embedded elements, Textile composites

ABSTRACT Since the textile composites are hierarchical structures (fibers, yarns and fabrics), there are different levels/scales in their modelling. Among the micro, meso and macro scales, the meso-scale is the most important level for numerical modelling of textile composites. Lomov et.al [1] proposed a road map for meso-FE modelling of textile composites and investigated different steps in proper modelling of a unit cell or RVE. They created geometrical textile models in WiseTex [2] software and then used the MeshTex software as an intermediate tool for meshing of the constructed yarns and matrix. Finally, they could create a proper RVE. However, using an intermediate software for meshing and importing the meshed parts to FE software (Abaqus) not only is a time consuming process but also may lead to loss of geometrical information while importing the models. Moreover, the quality of the mesh in MeshTex is not high enough. On the other hand, the interpenetration of the yarns was one of the main challenges for proper meshing of the parts. The common method for unit cell modeling is the conventional (full) method in which matrix and yarns are merged as a unit cell. However, there are problems in the full method during the meso-FEM that can be categorized as: a) Interpenetration of the yarns; b) Quality meshing of the matrix volume; c) Local coordinate system assignment in yarns. Besides the full method, ’’M-cube’’ and ‘’domain superposition technique’’ have also been proposed [3], [4]. In this paper, the concept of embedded element method (EEM) in Abaqus is used in meso-FEM of textile composites. In the EEM, the translational degrees of freedom of the embedded element (yarn) are constrained to the interpolated values of the corresponding degrees of freedom (DOFs) of the host element (matrix). The yarn geometry is modelled in WiseTex based on spacing, width and thickness of the yarns. A Python script (D.S. Ivanov) is used for importing the constructed model from WiseTex to Abaqus directly. Then, the matrix‘’ box’’ is modelled separately and assembled with yarns to build up a proper unit cell. Finally, the ‘’embedding equation’’ is defined to link DOFs in the yarns and matrix. To investigate the effectiveness of proposed method, a 5-harness satin composite is considered [5]. The unit cell is modelled according to the ‘’embedding element’’ and full methods and Dirichlet boundary conditions are applied for both models. Figure (1) compares the Von Misses stresses in the EE and full models.

(a) (b)

Figure 1: Unit cell model (a) Embedded element method (b) Full method. In table (1), the homogenised mechanical properties for both solutions are compared.

2525

S.A. Tabatabaei, Stepan V. Lomov and Ignass Verpoest

Table.1 Mechanical properties of the 5-harness satin weave from different methods.

There is an acceptable agreement between the EEM and full method in homogenized properties. The difference between the full method and the EEM for longitudinal and transverse stiffnesses are 0.3% and 7% respectively. For shear moduli, the maximum difference of two methods is less than 9%. In addition, the calculated homogenised properties using the EEM have a good agreement with experimental results. Finally, the local stress variation in the weft yarn for both models is compared in figure (2).

Figure 2: The first principal stress profile for EE and full methods in the weft direction.

The similarities and differences in the full and EE model will be discussed in the conference. Acknowledgement: The work leading to this publication has received funding from the European Union Seventh Framework Programme (FP7/2007-2013) under the topic NMP-2009-2.5-1, as part of the project HIVOCOMP (grant agreement n° 246389).

REFERENCES [1] Stepan V. Lomov, Dmitry S. Ivanov, Ignaas Verpoest, et.al, Meso-FE modelling of textile

composites: Road map, data flow and algorithms, Composites Science and Technology 67 (2007) 1870–1891

[2] Verpoest, I. and S.V. Lomov Virtual textile composites software Wisetex: integration with micro-mechanical, permeability and structural analysis. Composites Science and Technology 65 (2005) (15-16): 2563-2574

[3] S. Honda, M. Zako, T. Kurashiki, H. Nakai, S.V. Lomov, I. Verpoest, A proposal of stress/strain analytical procedure of textile composites with stitch by M3 method. in 13th European Conference on Composite Materials (ECCM-13). 2008. Stockholm. [4] W. G. Jiang, S.R. Hallett, M. Wisnom, Development of domain superposition technique for the modeling of woven fabric composites, Mechanical response of composites, P.P. Camanho, Editor. 2008, Springer. [5] S. Daggumati , W. Van Paepegem , J. Degrieck, J. Xu, S.V. Lomov, I. Verpoest , Local damage in a 5-harness satin weave composite under static tension: Part II – Meso-FE , composite Composites Science and Technology 70 (2010) 1934–1941.

Mechanical coefficients

Full model Embedded element model

Experiment

E11(GPa) 59.5 59.36 57±1 E22(GPa) 59.5 59.36 E33(GPa) 10.55 11.28 ν12 0.057 0.061 0.05±0.02 ν13 0.41 0.54 ν23 0.41 0.54 G12(GPa) 4.305 4.155 4.175 G13(GPa) 3.286 3.573 G23(GPa) 3.286 3.573

2626


Aalborg, 2013

DEVELOPMENT OF CONSTITUTIVE MATERIAL MODEL FOR COMPOSITE WITH NONLINEAR FIBERS AND MATRIX

L. Pupure1, R. Joffe1,2 and J. Varna1

1Department of Engineering Sciences and Mathematics, Luleå University of Technology SE-97187 Luleå, Sweden

Email: [email protected], web page: http://www.ltu.se

2Department of Materials and Manufacturing, Swerea SICOMP PO Box 271, SE-94126, Piteå, Sweden

web page: http://www.swerea.se/sicomp/

Keywords: Natural fibres, bio-based polymers, viscoelasticity, viscoplasticiy, material model

ABSTRACT Due to environmental and economic reasons, industries as well as society demands new non-fossil derived materials; this promotes development of bio-based materials. Natural fibres (e.g. wood, flax, hemp) have been used in composites already for a while but most recently several thermosetting resins, which are also bio-base, became available, allowing to produce whole bio-based composites. It has been successfully demonstrated [1] that high quality bast fibres, such as flax and hemp, compare well with glass fibre properties, especially if specific properties are considered. However one of the main disadvantages of these fibres is variability to properties, due to location of growth, processing conditions etc. Therefore another type of reinforcement with natural origin – regenerated cellulose fibres (RCF) have attracted attention. These fibres are continuous, with constant, reproducible geometry and properties but with one significant disadvantage - they exhibit highly non-linear behaviour (see Fig. 1). In order to analyse and predict mechanical behaviour of composites with such fibres appropriate material models have to be employed.

Figure 1: Stress-strain curves for RCF (on left) and hysteresis loops from loading-unloading experiments (on right).

A large number of experiments need to be carried out for each fibre/matrix composition in order to obtain all parameters in the nonlinear time dependent material model. Since numerous combinations of matrix and fibres are possible, the ambition is to make predictions more convenient by using material model for the composite which is based on performance of constituents characterized separately. Thus, tests on different composites will not be needed, except for limited number of experiments to verify the model.

2727

L. Pupure, R Joffe and J. Varna

The material modelling is based on model developed by Schapery [2,3] to describe time dependent behaviour of materials. His original model for nonlinear viso-elastic behaviour is complemented with Zapas model for viscoplasticity [4] and modified to account also for microdamage [5]:

tddgd

Sgd VP

t

,'0

210max (1)

The objective of this investigation is to develop micromechanics models to predict mechanical behaviour of composites with long and aligned fibres, based on empirical generalized models for constituents capturing their time-dependent behaviour. In order to identify and quantify parameters needed for modelling, multi-step creep and strain recovery experiments for composite constituents were carried out and material models for the resin and the fibres were developed. In this study RCF and EpoBioX (75% bio-based) matrix are used. This resin has very similar behaviour of synthetic polymer (e.g. epoxy). Several step creep tests (Fig. 2) for resin showed negligible amount of irreversible strain, whereas results obtained for RCF composites show very significant amount of irreversible (presumably visco-plastic) strain. Thus, in composites, the nonlinearity is exhibited mostly due to properties of fibre. This is very different from synthetic composites, where only matrix is contributing to the non-linear behaviour of composite. Extensive damage tolerance and creep tests were performed to characterize composites for comparison with results of micromechanical modelling.

Figure 2: Several step creep tests for EpoBioX resin (on left) and RCF/EpoBioX composite (on right)

REFERENCES [1] Wambua P., Ivens J. and Verpoest I., Natural fibres: can they replace glass in fibre reinforced

plastics?, Composite Science and Technology, 63, 2003, pp. 1259-1264. [2] Lou Y.C. and Schapery R.A., Viscoelastic characterization of a nonlinear fibre-reinforced

plastic, Journal of Composite Materials, 5, 1971, pp. 208-234. [3] Schapery R.A., Viscoelastic and viscoplastic constitutive equation based on thermodynamics,

Mechanics of Time-Dependent Materials, 1, 1997, pp. 209-240. [4] Zapas L.J. and Crissman J.M., Creep and recovery behaviour of ultra-high molecular weight

polyethylene in the region of small uniaxial deformation, Polymer, 25, 1984, pp. 57-62. [5] Marklund E., Eitzenberg J. and Varna J., Nonlinear viscoelastic viscoplastic material model

including stiffness degradation for hemp/lignin composites, Composite Science and Technology, 68, 2008, pp. 2156-2162.

2828


Aalborg, 2013

VALIDATION OF SIMULATED UNIDIRECTIONAL COMPOSITES MICROSTRUCTURE: STATISTICAL EQUIVALENCE TO REAL FIBRE ARRANGEMENTS

Valentin S. Romanov1, Larissa Gorbatikh, Stepan V. Lomov and Ignaas Verpoest

1Department of Metallurgy and Materials Engineering (MTM), University of Leuven (KU Leuven) Kasteelpark Arenberg 44 - bus 2450 B-3001 Heverlee, Belgium

Email: [email protected], web page: http://www.mtm.kuleuven.be

Keywords: Carbon fibres, Finite element analysis (FEA), Statistics, Stress concentrations, Simulated fibre arrangements

ABSTRACT

Finite Element (FE) modelling of the onset and propagation of matrix cracks in fiber-reinforced composites on the micro scale requires adequate representation of material’s microstructure. Representative Volume Elements (RVEs) statistically equivalent to the real random fibre placement need to be generated. So called “hard-core” random generation of fibre centres is not able to create fibre distribution with condition of no-overlapping for fibre volume fraction (VF) over 50…55%. Random Microstructure Generation (RMG) method [1] overcomes this difficulty by using a tree-step heuristic algorithm of movement/adding fibres in the RVE till required VF is obtained. The algorithm is able to work with periodic patterns and allows creation RVEs with VF up to 75%, with computational expenses much lower than in more rigorous methods based on molecular dynamics analogues. However, the method was not fully validated: it was not investigated whether RMG algorithm creates fibre arrangements statistically equivalent with real UD FRC microstructures. Furthermore, we introduced condition of random minimum distance between fibre centres: in order to represent the effect of waviness of fibres in out-of-plane direction that leads to presence of a certain gap between fibres. Modified RMG algorithm has to be statistically analysed. In the present work we compare simulated and real fiber arrangements found in unidirectional composites, using statistical descriptors. The comparison is done for geometrical and mechanical parameters such as distributions of the fibre positions and statistical functions of the stress fields. The real fiber arrangements are extracted from microscopy images of a 3D non-crimp woven carbon/epoxy composite [2] with fiber volume fractions of 60-68%. The stress fields are compared for the case of transverse tension of RVE.

a b c

Fig.1 Stages of the statistical validation: (a) RVE with VF=62% created by RMG algorithm, (b) Pair Distribution Function for real and RMG fibre placement, (c) histogram of von Mises

stress distribution in matrix under transverse tension

2929

Valentin S. Romanov, Larissa Gorbatikh, Stepan V. Lomov and Ignaas Verpoest

The results show good correlation between both the geometrical and mechanical statistical parameters of the real fibre placement and the simulated RVEs. We conclude that the RMG algorithm is able to generate random fibre distributions that are statistically equivalent to real fibre distribution found in UD composites. The algorithm can be used as the base for the definition of 3D micromechanical finite element models of UD composites. Acknowledgement. The work has been funded by EC, “IMS&CPS” (FP7) project.

REFERENCES 1. Melro, A.R., P.P. Camanho, and S.T. Pinho, Generation of random distribution of

fibres in long-fibre reinforced composites. Composites Science and Technology, 2008. 68(9): p. 2092-2102.

2. Karahan, M., et al., Internal geometry evaluation of non-crimp 3D orthogonal woven carbon fabric composite. Composites Part A: Applied Science and Manufacturing, 2010. 41(9): p. 1301-1311.

3030


Aalborg, 2013

CYCLIC DEFORMATION OF POLYETHYLENE/CLAY NANOCOMPOSITES: OBSERVATIONS AND CONSTITUTIVE MODELING

Aleksey D. Drozdov1,2, Jesper deC. Christiansen2 and Rasmus Klitrou2

1Department of Plastics Technology, Danish Technological Institute Gregersensvej 7, DK-2630 Taastrup, Denmark

Email: [email protected], web page: http://www.dti.dk

2Department of Mechanical and Manufacturing Engineering, Aalborg University Fibigerstræde 16, DK-9220 Aalborg East, Denmark

Email: [email protected], web page: http://www.m-tech.aau.dk

Keywords: Nanocomposites, Cyclic deformation, Viscoelastoplasticity, Constitutive modelling

ABSTRACT The talk is concerned with experimental investigation and constitutive modelling of the viscoelastic and viscoplastic behaviour of polyethylene/clay nanocomposites under cyclic loading with finite deformations. A particular attention is focused on the fading memory phenomenon: when two specimens are subjected to multi-cycle tensile deformation programs with increasing maximum elongation ratios per cycle that differ along the first n cycles and coincide afterwards, the mechanical responses of these samples become identical starting from the (n+1)th cycle. Low-density polyethylene (LDPE) Riblene FL20 (density 0.921 g/cm3, melt flow rate 2.2 g/10 min, melting temperature 109 ºC) was purchased from Polimeri Europa (Italy). Maleic anhydride-grafted polypropylene (MA-g-PP) Eastman G3015 (acid number 15 mg KOH/g) was supplied by Eastman Chemical Company (USA). Organically modified montmorillonite nanoclay (NC) Delitte 67G was donated by Laviosa Chimica Mineraria S.p.A. (Italy). Polymer/clay nanocomposites were manufactured in a two-step process [1]. The nanocomposite masterbatch with clay/compatibilizer proportion 2:1 was compounded in twin-screw extruder Thermo Scientific PRISM Eurolab 16 (barrel temperature 180 ºC, screw speed 300 rpm, feed rate 2 kg/h). Pellets of the masterbatch were mixed with LDPE in proportion corresponding to NC concentration of 4 wt.%. Dumbbell specimens for tensile tests (ASTM standard D638) with cross-sectional area 9.95 x 3.75 mm were molded by using injection-molding machine Ferromatic Milacron K110. Uniaxial tensile tests were conducted on neat LDPE and LDPE/clay nanocomposite at room temperature by means of universal testing machine Instron-5586 equipped with Instron 2630-113 static axial extensometer for control of longitudinal strains. Tensile force was measured by 50 kN load cell. The engineering stress σ was determined as the ratio of axial force to cross-sectional area of undeformed specimens. The experimental program involved: (i) tensile tests with cross-head speeds 1, 10, and 100 mm/min (which correspond to strain rates e=1.7·10-4, 1.7·10-3, and 1.7·10-2 s-1) up to breakage of specimens, (ii) short-term relaxation tests with elongation rations k=1.2 and 1.4, (iii) cyclic tests (N=20 cycles) with cross-head speed 100 mm/min, fixed maximum elongation ratios kmax=1.1, 1.3, 1.5 and minimum stress σmin=1 MPa, (iv) cyclic tests with cross-head speed 100 mm/min, minimum stress σmin=1 MPa, and increasing maximum elongation ratios kmax(n)=1.1, 1.3, 1.5, 1.7 and kmax(n)=1.2, 1.4, 1.6, (v) cyclic tests with a mixed program: specimens suffered cyclic pre-loading with maximum elongation ratios kmax(n)=1.1, 1.2, 1.3,

3131


1.4 and kmax(n)=1.2, 1.4, were subjected to cyclic deformation (N=20 cycles) with kmax=1.4 and σmin=1 MPa. Stress-strain curves in a tensile test with cross-head speed 100 mm/min and cyclic tests with various programs are depicted in Figure 1. This figure shows a slight improvement of strength of LDPE due to its reinforcement and demonstrates unambiguously fading memory of deformation history in neat and nanoclay-reinforced polyethylene.

Figure 1: Engineering stress σ versus elongation ratio k. Symbols: experimental data on neat LDPE (A) and LDPE/NC composite (B) ( – tensile test; – tests with fixed kmax=1.1, 1.3, 1.5; – test with kmax(n)=1.1, 1.3, 1.5, 1.7; – test with kmax(n)=1.2, 1.4, 1.6; – test with kmax(n)=1.1, 1.2, 1.3, 1.4; – test with kmax(n)=1.2, 1.4). To describe the experimental data, a constitutive model is developed for the viscoelastic and viscoplastic behavior of nanocomposites with semicrystalline matrices under multi-cycle deformation with finite strains. Applying the homogenization concept, we treat a nanocomposite as a one-phase equivalent isotropic and incompressible medium. With reference to the theory of transient networks, we associate its viscoelastic response with separation of active chains from their junctions and attachment of dangling chains to the network. Irreversible deformations of the equivalent network (that reflect sliding of junctions between chains in the amorphous phase and fine and coarse slip of lamellar blocks in the crystalline phase) are described based on the viscoplasticity theory with two plastic deformation gradients. Stress-strain relations are derived by means of the Clausius-Duhem inequality for an arbitrary three-dimensional deformation with finite strains. Adjustable parameters in the constitutive equations are found by fitting the observations. Ability of the model to predict the mechanical response of nanocomposites is confirmed by comparison of results of numerical simulation with observations in independent tests. Financial support by the EU Commission through Project Evolution-314744 is gratefully acknowledged.

REFERENCES [1] A.D. Drozdov, A.-L. Hog Lejre, J. deC. Christiansen, Viscoelasticity, viscoplasticity, and creep

failure of polypropylene/clay nanocomposites. Composites Science and Technology, 69, 2009, pp. 2596-2603.

[2] A.D. Drozdov, R. Klitkou, J.deC. Christiansen, Cyclic viscoplasticity of semicrystalline polymers with finite deformations. Mechanics of Materials, 56, 2013, pp. 53-64.

3232


Aalborg, 2013

EXPERIMENTAL CHARACTERIZATION AND MODELING OF THIN PLY-SIZE EFFECT

R. Amacher, J. Cugnoni and J. Botsis

Laboratory of Applied Mechanics and Reliability Analysis, Ecole Polytechnique Fédérale de Lausanne LMAF-STI-EPFL, Station No 9, CH – 1015 Lausanne, Switzerland

Email: [email protected], web page: http://lmaf.epfl.ch

Keywords: Ultra-thin plies, Size effect, Characterization, Finite elements, Failure mechanisms

ABSTRACT

Ultra-thin plies composites are raising more and more interest since several authors [1-3] have shown their advantages. Today, high-end applications such as formula one begin to use them, and even conservative domains such as aeronautics are considering their use. In order to benefit plainly of their enhanced properties, it is however necessary to understand the ply-size effect and to translate this knowledge into appropriate design rules. In this work, three unidirectional prepreg tapes with different fiber areal weight (2x150, 100 and 30 g/m2) were produced at North-TPT Switzerland from M40JB fibers and ThinPregTM 80EP/CF epoxy resin, using the same batches in order to minimize the scatter factors. For the same reason, all specimens were cured in autoclave following the recommended curing cycle. To study the ply thickness effect and avoid volumetric size effects, the scaling method considered was design-oriented, changing only the ply thickness but not the geometrical dimensions of the specimen. In order to verify the baseline properties of the plies, the first stage consisted in a campaign of standardized experiments – uniaxial tension, four points bending of a sandwich beam and ILSS - on 0° UD specimens with the three ply weights. In the second stage of experiments, uniaxial tension, open-hole tension fatigue, open-hole compression, impact, and bolted assembly single lap bearing tests were conducted on quasi-isotropic [45°/90°/-45°/0°]ns and orthotropic specimens [-35°/0°/35°/0°]ns. The scaling used for those was mainly sublaminate repetition, but ply-level scaling and optimized orientations were also tested. Stress-strain curves were recorded using strain gages and complemented with acoustic emission monitoring, digital image correlation and ultrasonic c-scan on selected specimens.

Figure 1: Ultimate strength and onset of damage in function of the plies areal weight and the stacking. Sublaminate scaling with thin plies provides enhancements in both the ultimate strength and the onset

of damage.

Overall, thin-ply composites (30g/m2) exhibited significantly better mechanical performance than standard “thick” ply composites (300g/m2) in most cases. For example, quasi-isotropic unnotched tensile strength increased by 42% (onset of damage +227%), notched compression strength was

3333


improved by 18%, onset of damage in open-hole tension was increased by 31% and open-hole tension fatigue life was increased dramatically (>50x). However, thin ply composites showed a lower ultimate strength (-30%) in quasi-static open hole tensile conditions. As discussed in [1, 2], this difference could be explained by the absence of damage growth at the apex of the hole in thin-ply laminates which would normally relax the local stress concentration in thick ply composites. In most tests, a clear change in failure mode was observed, starting from a multi-mode failure with strong delamination for the thick plies laminates to a quasi-brittle failure with no delamination for the thin plies laminates. In all cases, the onset of damage and free edge delamination was found to be much delayed with thin plies, to the point of being very close to the ultimate failure. The change in onset of damage is of particular interest since it allows for more load bearing capacity in a first ply failure driven design. Moreover, the absence of early damage is also responsible for a significantly enhanced fatigue life. For example, open-hole thin ply specimens loaded in tension (R = 0.1, =316 MPa) could stand more than one million cycle whereas a thick ply specimen would fail before 20’000 cycles. To better understand the mechanisms involved in those changes, finite element modeling was carried out using Simulia Abaqus Explicit. Meso-scale 3D models of unnotched quasi-isotropic tests were developed using cohesive damage models to account for transverse ply cracking and delamination. A good phenomenological agreement was found with the damage patterns observed using ultrasonic c-scan, which demonstrated the importance of the free-edge delamination in the thin-ply size effects.

Figure 2: Interfacial damage in quasi-isotropic laminate: (a) FE modeling results (b) C-Scan

Besides their overall strength advantage, as thin-ply composites tend to fail without delamination, much simpler mechanisms of failure need to be modeled which potentially will lead to more predictable performance than the usual multi-mode failure seen in thick ply composites. Together with delayed onset of damage, thin-ply composites represent a big opportunity for enhanced load bearing capacity by straightforward ply thickness reduction or more advanced design optimization.

ACKNOWLEDGEMENTS

This work was supported by Swiss Commission for Innovation and Technology CTI grant in partnership with North-TPT, RUAG and Connova AG.

REFERENCES

[1] S. Sihn, R.Y. Kim, K. Kawabe, S. Tsai, Experimental studies of thin-ply laminated composites, Composites Science and Technology, 67, 2007, pp. 996-1008 (http://www.sciencedirect.com/science/article/pii/S0266353806002168).

[2] B.G. Green, M.R. Wisnom, S.R. Hallet, An experimental investigation into the tensile strength scaling of notched composites, Composites: Part A, 38, 2007, pp. 867-878 (http://www.sciencedirect.com/science/article/pii/S0263822307001705).

[3] M.R. Wisnom, B. Kahn, S.R. Hallet, Size effects in unnotched tensile strength of unidirectional and quasi-isotropic carbon/epoxy composites, Composite Structures, 84, 2008, pp. 21-28 (http://www.sciencedirect.com/science/article/pii/S0263822307001705).

3434


Aalborg, 2013

Oral session 3 – Fracture and fatigue -1 (5 presentations)

3535


Aalborg, 2013

USE OF DIC FOR THE FAILURE ANALYSIS OF COMPLEX COMPOSITE STRUCTURES

G. Crammond1, S.W. Boyd1 and J.M. Dulieu-Barton1

1 University of Southampton Universisty Road, Highfield, Southampton, SO17 1BJ, UK

Email: [email protected], web page: www.southampton.ac.uk/damtol

Keywords: Digital image correlation, failure analysis, principal stress fields

ABSTRACT Digital image correlation (DIC) is used to obtain the surface strains developed in a glass fibre double butt strap joint (DBSJ). Magnifying optics are used to provide high spatial resolution images around the geometric discontinuity between the adherends, as this is where failure initiates. The evolution of complex localised peel and shear strain features are identified relating to the development of damage within the joint. The strain data is manipulated to evaluate the principal strain and stress components within the joint, which govern failure in the brittle epoxy matrix [1], providing a better understanding of the load transfer and damage initiation mechanisms within the joint. An area of 3.1 mm x 2.6 mm around the discontinuity between adherends in the DBSJ was imaged with a Canon mp-e65 macro lens connected to a 5MP LaVision E-Lite camera, mounted on and X-Y- adjustable mount. Due to the small field of view and short focal distance, component strains within the DBSJ were evaluated using 2D DIC by recording images from the specimen as it was loaded at 2 mm/min up to failure. The development of complex localised strain distributions were revealed, as shown in Figures 1 and 2. Analysis of the strain fields identifies the evolution of small, yet critical, through-thickness (peel) strains generated in the outer adherend, and high shear strains in the adhesive material between adherends within the joint. The peel and shear strains show a coupled interaction; with concentrations of high peel and shear strains forming adjacent to each other at the interface between the CSM material in the outer adherend and the adhesive. As the joint material is relatively brittle, the DIC component strain data is used to derive the principal strains and stresses to identify the failure mechanisms in the joint. Due to the severe geometric discontinuity between adherends, the principal stress and strain directions change dramatically around the discontinuity. Therefore the principal strains must be calculated on a point by point basis using a ‘Mohr’s circle’ analysis of the direct and shear strains given by the DIC [2]. The principal strains are then transformed into principal stresses using the following constitutive relationship:

(1)

where [Q]12 is the stiffness matrix in the principal stress directions. The elastic properties in the principal material directions of the joint material were obtained experimentally using techniques described in [3]. To carry out the transformation in equation (1) the stiffness matrix must be populated with elastic properties of the material in the principal stress/strain directions. This was done by transforming the values for stiffness in the principal material directions to the principal stress directions using the angle between the principal material axes and principal stress axes established from the DIC component strains.

37

G. Crammond, S.W. Boyd and J.M. Dulieu-Barton

Figure 1: Peel strain distribution at 18 kN Figure 2: Shear strain distribution at 18 kN Figure 3 clearly identifies large principal stress concentrations forming either side of the adhesive interface. It can be seen that the principal stresses in the adhesive are small but finite indicating that the load transfer through the joint is not entirely through shear. High maximal principal stresses in the predominantly shear loaded interface between the adhesive and the outer adherend are observed to result in the growth of cracks at loads above 16 kN. The generation of damage at the high stress interface with the outer adherend interacts with a region of high peel strain at the root of the discontinuity in the inner adherend. Subsequently damage and crack propagation occurs in this region leading to the catastrophic failure of the joint. In the presentation the methodology for obtaining the principal stresses is described along with application to failure of joints under high strain rate loading.

Figure 3: Principal stresses in the a) 1 and b) 2 directions around the discontinuity at 13 kN [1] I. Daniel and O. Ishai, Engineering Mechanics of Composite Materials. Oxford University

Press, 2005. [2] P. Benham, R. Crawford, and C. Armstrong, Mechanics of Engineering Materials. Prentice

Hall, 1996. [3] S. W. Boyd, J. M. Dulieu-Barton, O. T. Thomsen, and S. El-Gazzani, “Through thickness

stress distributions in pultruded GRP materials,” Composite Structures, vol. 92, no. 3, pp. 662–668, Feb. 2010.

1 2

5

0

5

0

5

0

5

0

50

100

150

-100

-50

0

50

Adhesive

CSM CSM

Inneradherend

Outer adherend

MPa

MPa

Adhesive CSM CSM

Inneradherend

Outer adherend

0

0.005

0.01

0.015

0.02

0.025

0.03

0.035

0.04

Shea

rst

rain

Adh

esiv

e

CSM CSM

Inneradherend

Outeradherend

Edge of joint

-5

-4

-3

-2

-1

0

1

2

3

4

5x 10-3

Inneradherend

Outeradherend

Peel

stra

in

Adh

esiv

e

CSM CSM

Edge of joint

38


Aalborg, 2013

DEBOND GROWTH ASSESSMENT IN GFRP-BALSA SANDWICH STRUCTURES

E. Farmand-Ashtiani, N. Nasri, J. Cugnoni and J. Botsis1

Laboratory of Applied Mechanics and Reliability Analysis Swiss Federal Institute of Technology (EPFL), Lausanne, CH-1015, Switzerland

Email: [email protected], web page: http://lmaf.epfl.ch/

Keywords: Sandwich structure, Balsa wood, Interfacial crack, Energy dissipation

ABSTRACT Interfacial debonding is well recognized as a critical damage mode in sandwich structures. This is due to the fact that the structural performance of sandwich composites depends primarily on the stress transfer between the faces and the core. Characterization of the face-core debonding is therefore of particular interest in the context of sandwich composites testing [1]. Sandwich structures consisting of balsa core and glass fibre reinforced plastic (GFRP) skins are widely used in civil and energy infrastructures. In this work, single cantilever beam (SCB) and four point end notched flexure test (4ENF) setups are used to characterize the interfacial fracture mechanism in GFRP-balsa sandwich specimens. The experimental procedure is accompanied by finite element simulation of the load-specimen configuration to analyse the mechanism of the interfacial fracture process. Sandwich panels with glass/polyester face laminate ([0]3, 960 gr/m2), and balsa core (BALTEK® SB.150) are fabricated by hand/wet lay-up process. A non-adhesive insert as debond initiator is introduced in the face-core interface. The sandwich panels are cured at ambient temperature for 7 days and cut into several beams with nominal dimensions of 255 mm × 25 mm × 16 mm and an initial interface crack of 60 mm in length. The skin–core interface zone on a specimen’s side face is painted white and marked at every millimetre to help measuring the crack length during the debonding tests using a microscope attached to the testing system. Two grooves of 3 mm in depth are machined in the core of the SCB specimens so that it is supported by the testing fixture using two L-shaped rails (Fig. 1a). Steel loading blocks are glued on the pre-cracked face-sheet and the specimens are subjected to monotonic (1 mm/min) and fatigue (5 Hz, R=0.1) loading in displacement control. Several 4ENF tests (Fig. 1b) are performed with a rate of 3.6 mm/min, with an inner span length of 76mm and an outer span length of 171mm. The pin-composite friction is minimized using ball bearings to apply the load. A stable crack propagation, without any kinking into the core is observed in both experiments, the energy release rate associated with the interface crack growth is calculated using the compliance calibration (CC) method as well as the area method using experimentally recorded unloading-displacement curves. Typical load displacement curve from the monotonic SCB tests is shown in Fig. 1c. Energy release rate evaluation of the SCB specimen indicates a constant range of interfacial toughness (0.6 ±0.1 kJ/m2) over the crack propagation range. The strain field associated with the interface crack growth in the SCB specimens is monitored using embedded fibre Bragg grating sensors. Numerical analysis of the strain data shows that interfacial fracture process in the SCB specimens is entirely linear elastic and no toughening or dissipative mechanisms are involved [2]. Figure 1d illustrates the load-displacement curves obtained from the 4ENF experiments. Contrary to the SCB testing, the curves in Fig. 1d clearly indicate that dissipative mechanisms accompany the crack propagation. In situ observations suggest a viscoplastic deformation of the core developing during interface crack propagation. Transverse cracks in the core are also visible.

3939

E. Farmand-Ashtiani, N. Nasri, J. Cugnoni and J. Botsis

Almost constant value of energy release (ERR) rate (3.6 kJ/m2), with crack length, is obtained in the 4ENF tests by using the CC method. Nevertheless, this data reduction method assumes linear elastic behaviour and is therefore incorrect to use. To obtain a reasonable estimate of the ERR, the area method was used with the unloading curves of the loading-unloading experiment (Fig. 1d). These calculations result in values of (8.5 kJ/m2) which are more than 2 times larger than those calculated with CC method, confirming the existence of other dissipative mechanisms.

Figure 1: Photograph showing the SCB loaded specimen (a) and 4ENF test setup (b),

Load–displacement curves from SCB test (c) and from 4ENF test (d). Dashed lines indicate unloading. Cyclic loading-unloading is shown in blue curves, the black curve is obtained from another specimen

tested in one cycle. Note the hysteretic behavior and the residual deformation. While the response of the SCB configuration is fully understood and characterized using linear elasticity based numerical modeling, the debonding-deformation response of the 4ENF results for the sandwich system used here are more complicated. Thus they are modeled by accounting for the time dependent response of the core and the skin as well as transverse cracking. The authors acknowledge the financial support from the Swiss National Science Foundation under Grant 200020_137937/1.

REFERENCES [1] Carlsson and G.A. Kardomateas, Structural and Failure Mechanics of Sandwich Composites,

Solid Mechanics and its Applications, Vol. 121, Springer Science+Business Media B.V, USA, 2011.

[2] E. Farmand-Ashtiani, J. Cugnoni, and J. Botsis, Monitoring and characterization of the interfacial fracture in sandwich composites with embedded multiplexed optical sensors, Composite Structures, 96, 2013, pp. 476–483. L.A.

(d)

(a) (b)

4040

6th International Conference on Composites Testing and Model Identification Ole T. Thomsen, Bent F. Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

EXPLICIT EXPRESSIONS FOR THE CRACK LENGTH CORRECTION PARAMETERS FOR THE DCB, ENF, AND MMB TESTS ON MULTIDIRECTIONAL LAMINATES

Stefano Bennati1, Paolo Fisicaro1 and Paolo S. Valvo1

1Department of Civil and Industrial Engineering, University of Pisa Largo Lucio Lazzarino, I-56126 Pisa, Italy

Email: [email protected], [email protected], web page: http://www.unipi.it

Keywords: Multidirectional laminates, Delamination toughness, DCB, ENF, and MMB tests, Crack length correction parameters, Beam-theory model

ABSTRACT

The double cantilever beam (DCB) and end-notched flexure (ENF) tests are the simplest and most commonly used testing methods to determine the delamination toughness of laminated specimens under fracture modes I and II, respectively. For I/II mixed-mode fracture, a widespread testing method is the mixed-mode bending (MMB) test, which can be regarded as the superposition of the DCB and ENF tests. For unidirectional (UD) laminated specimens, American, European and Japanese standards exist for such tests [1]. However, delamination toughness characterization of multidirectional (MD) composite laminates is still an open issue [2]. Several theoretical models are used in the literature to interpret the experimental results of the DCB, ENF, and MMB tests. The simple beam-theory (SBT) model considers the specimen as an assemblage of three rigidly connected Euler-Bernoulli beams [3]. The corrected beam-theory (CBT) model better accounts for the actual deformation of the specimens by considering the transverse shear deformability and the effects of deflections and rotations at the crack tip. This result is accomplished by replacing the actual delamination length, a, by an increased delamination length, a + h (where h is the specimen’s half-thickness and is the so-called crack length correction parameter), in the SBT formulas for the compliance, C, and energy release rate, G [4, 5]. Actually, the current ASTM standard for the MMB test suggests formulas for the mode I and II crack length correction parameters, I and

II, which can be used for UD laminated specimens [6]. For MD laminated specimens, de Morais and Pereira have proposed a modified beam-theory (MBT) model, where the crack length correction parameters are computed by considering the homogenised flexural and shear moduli [7]. The authors have developed an enhanced beam-theory (EBT) model of the MMB test, wherein the laminated specimen is considered as an assemblage of two identical sublaminates partly connected by a deformable interface. The sublaminates are modelled as extensible, flexible, and shear-deformable laminated beams. The interface is regarded as a continuous distribution of linearly elastic–brittle springs. An exact analytical solution for the internal forces, displacements, and interfacial stresses of the MMB test specimen has been deduced [8]. Furthermore, useful approximate expressions have been determined for the compliance and energy release rate of the DCB, ENF, and MMB test specimens. Such quantities can be expressed by introducing the following crack length correction parameters [9]:

EBT EBTI II

1 1

12

1 1

1 1 ,12 ( )

4

21 andz

x

hk hk

h

A

D

D

DC

(1)

where 1A , 1C , and 1D are the sublaminates’ extensional stiffness, shear stiffness, and bending stiffness, respectively, and kx and kz are the elastic constants (per unit area) of the distributed springs in the tangent and normal directions to the interface plane, respectively. Eqs. (1) define crack length

4141

Stefano Bennati, Paolo Fisicaro and Paolo S. Valvo

correction parameters having the same physical meaning that they have in the CBT model. The EBT expressions, however, have been deduced based on a rigorous analytical solution. Furthermore, they of course apply not only to UD, but also to MD laminated specimens. It is worth noting that comparisons between the CBT and EBT models for UD laminated specimens show very good agreement (see, for instance, Fig. 1). In the present work, we demonstrate the application of the EBT model to MD laminated specimens with several stacking sequences and compare our theoretical predictions with experimental results and numerical analyses. The first obtained results look very promising.

0 50 100 150 2000

1

2

3

4

Ex [GPa]

I, II

I

II

CBTEBT

0 5 10 15 200

1

2

3

4

Ez [GPa]

I, II

I

II

CBTEBT

0 5 10 15 200

1

2

3

4

Gzx

[GPa]

I, II

I

II

CBTEBT

Figure 1: Crack length correction parameters as functions of the elastic moduli

of a homogenous orthotropic specimen (from [9]).

REFERENCES [1] D.F. Adams, L.A. Carlsson and R.B. Pipes, Experimental Characterization of Advanced

Composite Materials – 3rd edition, CRC Press, Boca Raton, 2003. [2] J. Andersons and M. König, Dependence of fracture toughness of composite laminates on

interface ply orientations and delamination growth direction, Composites Science and Technology, 64, 2004, pp. 2139-2152 (doi: 10.1016/j.compscitech.2004.03.007).

[3] K. Friedrich (editor), Application of Fracture Mechanics to Composite Materials, Elsevier, Amsterdam, 1989.

[4] J.G. Williams, End corrections for orthotropic DCB specimens, Composites Science and Technology, 35, 1989, pp. 367-376 (doi: 10.1016/0266-3538(89)90058-4).

[5] Y. Wang and J.G. Williams, Corrections for mode II fracture toughness specimens of composites materials, Composites Science and Technology, 43, 1992, pp. 251-256 (doi: 10.1016/0266-3538(92)90096-L).

[6] ASTM, Standard Test Method for Mixed Mode I-Mode II Interlaminar Fracture Toughness of Unidirectional Fiber Reinforced Polymer Matrix Composites, D6671/D6671M-06, American Society for Testing and Materials, West Conshohocken, PA, 2006 (doi: 10.1520/D6671_D6671M-06).

[7] A.B. de Morais and A.B. Pereira, Interlaminar fracture of multidirectional glass/epoxy laminates under mixed-mode I + II loading, Mechanics of Composite Materials, 43, 2007, pp. 233-244 (doi: 10.1007/s11029-007-0023-1).

[8] S. Bennati, P. Fisicaro and P.S. Valvo, An enhanced beam-theory model of the mixed-mode bending (MMB) test – Part I: literature review and mechanical model, Meccanica, 2013 (doi: 10.1007/s11012-012-9686-3).

[9] S. Bennati, P. Fisicaro and P.S. Valvo, An enhanced beam-theory model of the mixed-mode bending (MMB) test – Part II: applications and results, Meccanica, 2013 (doi: 10.1007/s11012-012-9682-7).

4242


Aalborg, 2013

STRESS-STRAIN ANALYSIS OF A CARBON PPS DURING AND AFTER FATIGUE LOADING CONDITIONS

W. Van Paepegem1, I. De Baere1, S. Daggumati1, C. Hochard2, J. Xu3, S.V. Lomov3, I. Verpoest3 and J. Degrieck1

1Department of Materials Science, Ghent University Technologiepark-Zwijnaarde 903, B-9052 Zijwnaarde, Belgium

Email: [email protected], web page: http://www.composites.ugent.be

2Laboratoire de Mécanique et d’Acoustique, CNRS, Université aix Marseille 31 Chemin Joseph Aiguier, F-13402 Marseille 20, France.

3Department of Metallurgy and Materials Engineering, Katholieke Universiteit Leuven Kasteelpark Arenberg 4, B-3001 Leuven, Belgium.

Keywords: Carbon fabric, PPS, Stiffening

ABSTRACT

Although various damage modes which are occurring in woven composites under tension-tension fatigue are reported by several authors, there is a lack of information regarding the correlation between the sequence of the damage initiation and their influence on the macroscopic stiffness / strength of the composite specimen. In this regard, the current study presents the detailed discussion on the sequence of damage events and their influence on the structural response of a carbon fabric satin weave reinforced PPS. First, microscopic damage events are detected on the polished edges of the composite specimen and are correlated with observed fluctuations in the measured macroscopic composite stiffness. Then, residual strength tests were performed and compared to the stiffness and strength of the virgin composite specimens in order to estimate the effect of the micro-scale damage events on the load carrying behaviour as well as the strength of the composite.

Fig. 1 illustrates some of the damage mechanisms during fatigue life. At the end of 1000 cycles (Fig. 1(a)), the only failure observed on the composite polished edge is the weft yarn damage. After 1x105 cycles, i) broken load carrying fibres (Fig. 1 (b)); ii) meta-delaminations (Fig. 1 (b)) and iii) crack conjunction at different levels (Fig. 1(c)) can be seen. The above observed micro-scale damage mechanisms correlate with the reduction in the longitudinal stiffness of the composite.

(a) weft yarn damage at 1000 (b) broken fibers and meta-delaminations at 1x105

(c) crack conjunction at 1x105

Figure 1 Microscopic damage analysis of the composite polished edges at 500 MPa@5Hz

Fig. 2 shows the evolution of the longitudinal strain (Fig. 2(a)) as well as the longitudinal stiffness (Fig. 2(b)) of the composite specimen for a fatigue test between 0 and 600 MPa at2Hz up to 1.1 106

cycles. Initially, the maximum longitudinal strain value increases from 0.98 to 1% around 1 105

cycles (Fig. 2(a)). Thereafter, the strain values stay almost constant till the end of the fatigue test.

4343

W. Van Paepegem, I. De Baere, S. Daggumati, C. Hochard, J. Xu, S.V. Lomov, I. Verpoest and J. Degrieck.

Corresponding to the increase in strain value, the longitudinal stiffness of the composite decreases from 61 GPa to 59 GPa at the end of 1.1 106 cycles. Moreover, at this load level, microscopic analysis of the composite specimen reveals initial fibre breaks as well as meta-delaminations on the composite polished edges at around 1000 cycles. The above specified fatigue test was intentionally terminated at around 1.1 106 cycles to determine the residual stress-strain behaviour of the composite.

(a) Evolution of the strain (b) Stiffness degradation

Figure 2 Measurements of a fatigue test at 600MPa at 2Hz

With respect to the residual stiffness and strength, a number of quasi-static tests till failure were performed on both virgin and fatigued specimens. The fatigue specimens were loaded for 1 million cycles at either 65% or 80% of the quasi-static strength. Fig. 3 illustrates the stress-strain behaviour for both a virgin and a fatigued specimen. When observing the static behaviour of the fatigued specimen (Fig. 3 (a)), the stiffness (e.a. the slope of the tangent line) increases form 52.3 GPa at the start to 63.6 GPa near failure loads, meaning a stiffening of 22%. This is caused by (i) the straightening of load carrying warp yarns, which are allowed to do so because of the matrix cracking and meta-delaminations which arose during the fatigue load and (ii) the local carbon fibre stiffening. This is also confirmed by online video microscopy during several fatigue tests. For comparison purposes, the stress-strain curve of a virgin specimen is illustrated in Fig. 3(b). Here, the stiffness increases from 55.1 GPa to 61.3 GPa near failure, corresponding to a stiffening of only 11%. To determine the elastic modulus, only the part till 0.2% strain is considered as this is the value where damage initiation occurs in the satin weave composite. As the only damage observed during quasi-static testing is weft yarn cracking, the major contribution for the composite stiffening in the virgin specimen is most likely due to local carbon fibre stiffening under the applied load.

(a) post fatigue, 700 000 cycles 625MPa@2Hz (b) virgin specimen

Figure 3 Analysis of the composite stiffening under static tensile load.

The authors would like to acknowledge the Fund of Scientific Research – Flanders (F.W.O.) for sponsoring this research and Ten Cate Advanced Composites for supplying the material.

4444


Aalborg, 2013

X-RAY TOMOGRAPHY ASSESSMENT OF DAMAGE DURING TENSILE DEFORMATION OF ±45° CARBON FIBER LAMINATES

Federico Sket1, Alejandro Enfedaque4, Crystal Alton3,

Carlos González1,2, Jon M. Molina-Aldareguia1, Javier Llorca1,2

1IMDEA Materials Institute c/ Eric Kandel 2, 28906, Getafe, Madrid, Spain

2Departamento de Ciencia de Materiales, Universidad Politécnica de Madrid.

E. T. S. de Ingenieros de Caminos, 28040-Madrid, Spain

3Department of Chemical Engineering & Materials Science, Michigan State University 426 Auditorium Rd, East Lansing, MI 48824-1046, USA

4Departamento de Ingeniería Civil:Construcción, Universidad Politécnica de Madrid.

E. T. S. de Ingenieros de Caminos, 28040-Madrid, Spain

Keywords: Carbon fibre reinforced composites, X-ray tomography, damage evolution

ABSTRACT Carbon fibre reinforced polymers (CFRP) are widely used in structural components due to their high specific mechanical properties. Although there have been extensive experimental and theoretical studies of the material behaviour under different load conditions, improved understanding in regard to damage evolution and eventual failure is still needed. CFRP can present several different damage mechanisms, i.e. fibre fracture, fibre kinking, fibre pull-out, fibres scissoring, matrix cracking, debonding at fibre-matrix interface, delamination, and the dominant one depends on the load and deformation conditions [1]. Prediction of the type and amount of damage in these composites due to mechanical loads is of main importance and the models predicting the effect of damage in the non-linear response of polymeric composites present still discrepancies with respect to experimental data [2,3]. Therefore, a deep understanding of the fracture and deformation mechanisms and correlation to the global loading geometry and local stress field is needed. High resolution X-Ray Computed Tomography (XCT) is a specially well suited technique to visualize the damage mechanisms in fibre reinforced polymers since it provides resolution below the micrometer level, generation of 3D images, deep penetration into either transparent or opaque materials and simple sample preparation. Thus, in this work, XCT was used to assess the sequence of crack propagation into a ±45 epoxy/carbon fibre composite. Matrix cracking is one of the most important damage mechanisms when off-axis load is applied, leading to delamination and fibre breakage. Plain and open hole specimens were sequentially tensile tested up to five different displacements and further examined by XCT. An automatic algorithm was developed for the evaluation of the tomographic data. Damage parameters such as crack number density, crack generation rate, variation of cracks density along laminate width and fibre scissoring evolution were evaluated, providing a better understanding of the material damage behaviour which can be used to feed new models for damage development. A comparison between the plain and open-hole specimens is performed and the effect of the stress concentration around the hole is evaluated in terms of crack density with different loads. The open hole configuration induces stress concentrations in bands at ±45° in all the plies and therefore, the increments of load produces matrix cracking by shear in the direction of the fibres. The shear stresses in the direction of the fibres cracks the matrix relatively easier than the shear induced by the open hole in the direction perpendicular to the fibres. Matrix cracking is restricted to the location of the shear bands at the beginning of the test but it becomes wider with the load increase until the whole laminate

4545


is cracked. The rate of generation of these cracks is larger in the outer plies than in the plies in the interior of the volume.

REFERENCES [1] C.G. Davila, P.P. Camanho , C.A. Rose, Failure criteria for FRP laminates. J Compos Mater,

39, 2005, pp. 323–345. [2] B. Cox , Q. Yang , In Quest of Virtual Tests for Structural Composites, Science, 314, 2006,

1102-1107. [3] J. Llorca, C. González, J. M. Molina-Aldareguía, J. Segurado, R. Seltzer, F. Sket, M. Rodríguez,

S. Sádaba, R. Muñoz, L. P. Canal, Multiscale Modeling of Composite Materials: a Roadmap Towards Virtual Testing, Advance Materials, 23, 2011, 5130-5147.

4646


Aalborg, 2013

Oral session 4 – Fracture and fatigue – 2 (4 presentations)

4747


Aalborg, 2013

PROBABILISTIC ANISOMORPHIC CONSTANT FATIGUE LIFE DIAGRAM APPROACH TO PREDICTION OF P-S-N CURVES FOR COMPOSITES

Masamichi Kawai1 and Ken-ichiro Yano2

1Department of Engineering Mechanics and Energy, University of Tsukuba Tsukuba, Ibaraki, 305-8573, Japan Email: [email protected]

2Graduate School of Systems and Information Engineering, University of Tsukuba

Tsukuba, Ibaraki, 305-8573, Japan Email: [email protected]

Keywords: Woven CFRP, Fatigue model, Failure probability, Identification, CFL diagram

ABSTRACT

This study aims to develop a new engineering method for efficiently predicting a family of P-S-N curves for carbon fiber-reinforced composites fatigue loaded at any stress ratio by means of the anisomorphic constant fatigue life (CFL) diagrams [1, 2] constructed in specified constant values of failure probability. The probabilistic anisomorphic CFL diagram approach proposed in this study is applied to prediction of P-S-N curves for a woven fabric carbon/epoxy laminate, and the accuracy of prediction has been evaluated by comparing the predicted P-S-N curves with experimental ones. To this end, static and fatigue tests were first performed on woven CFRP laminate specimens to collect statistical samples of tensile and compressive strengths and fatigue lives for different stress ratios at each of the selected stress levels. Then, statistical distributions suitable for the static strength and fatigue life data were identified. The parameters of Weibull and lognormal distributions that were shown to be acceptable were estimated by fitting them to the experimental results, respectively. The experimental results indicate that the scatter of fatigue life data for the woven CFRP laminate in the fiber direction becomes more significant in order of stress ratio R = 0.1, , 10, where is the critical stress ratio [1, 2] equal to the ratio of compressive strength to tensile strength. For fatigue loading at R = 0.1 and it was observed that the scatter of fatigue life data tends to become larger at a lower stress level. For R = 10, however, this feature was not clearly observed because of a small gradient of S-N relationship as well as a small difference of stress levels selected for fatigue testing. Two-parameter lognormal and Weibull distributions can approximately be fitted to the static tensile and compressive strength data as well as the fatigue data at different stress ratios for the woven CFRP laminate. For all the fatigue data on the woven CFRP laminate, both the Kolmogorov-Smirnov and Anderson-Daring goodness-of-fit tests have suggested that Weibull and lognormal distributions are both acceptable at a significance level of 5%. For constructing the probabilistic anisomorphic CFL diagram for a given probability of failure, we need not only the static tensile and compressive strengths but also the S-N relationship at the critical stress ratio for the given probability of failure. The static tensile and compressive strengths for a given probability of failure P can be identified with the P percentile points of the distributions fitted to the static strength data. It is important to note that the values of critical stress ratio differ depending on the value of failure probability; i.e. (P1) (P2 ) if P1 P2 . This means that a P-S-N curve for the critical stress ratio (P) for any given probability of failure P , which is required to construct the anisomorphic CFL diagram for the given probability of failure, cannot be developed from fatigue data at a constant critical stress ratio (P1) that corresponds to a particular probability of failure P1 . To this

4949

Masamichi Kawai and Ken-ichiro Yano

new problem, we arrive at a solution by taking advantage of applying the modified fatigue strength ratio that was defined in an earlier study [3]. The anisomorphic CFL diagrams for different constant values of failure probability P = 10%, 50% and 90% that were predicted using the proposed procedure by assuming lognormal distributions of static strength and fatigue life agree well with the experimental CFL diagrams for the same levels of failure probability. The comparison is demonstrated in Figure 1 for the case of P = 10%. In Figure 2, the P-S-N curves for different stress ratios R = 0.1 and 10 that were predicted for different constant values of failure probability using the proposed probabilistic anisomorphic CFL diagram approach are shown to agree well with the P-S-N plots of experimental data.

0

200

400

600

800

1000

-1000 -800 -600 -400 -200 0 200 400 600 800 1000Mean stress m , MPa

Stre

ss a

mpl

itude

a ,

MPa Woven CFRP Quasi-isotropic

10Hz RT Experimental[(±45)/(0/90)]4s

Nf = 101

Nf = 102

Nf = 103

Nf = 104

Nf = 105

Nf = 106

R = ( = 0.571 )

R = 10

R = 0.1

kT = 1 , kC = 0.2

R = -1

Figure 1: Anisomorphic CFL diagram for the failure probability of P = 10%.

0

200

400

600

800

1000

100 101 102 103 104 105 106 107

2Nf

max

, M

Pa

Woven CFRP Quasi-isotropic [(±45)/(0/90)]4s

Fatigue 10Hz RT

ExperimentalP = 90%

P = 10%P = 50%

Predicted

Figure 2: Predicted P-S-N curves for different stress ratios in specified values of failure probability.

REFERENCES

[1] M. Kawai, A method for identifying asymmetric dissimilar constant fatigue life diagrams of

CFRP laminates, Proceedings of the Fifth Asian-Australasian Conference on Composite Materials (ACCM-5), Hong Kong, November 27-30, 2006.

[2] M. Kawai and K. Koizumi, Nonlinear constant fatigue life diagrams for carbon/epoxy laminates at room temperature, Composites Part A, 38, 2007, pp. 2342-2353.

[3] M. Kawai, A phenomenological model for off-axis fatigue behavior of unidirectional polymer matrix composites under different stress ratios, Composites Part A, 35, 2004, pp. 955-963.

5050


Aalborg, 2013

AUTOMATED DELAMINATION LENGTH VIDEO TRACKING IN STATIC AND FATIGUE DCB TESTS

F. Lahuerta, S. Raijmaekers, J.J. Kuiken, T. Westphal and R.P.L. Nijssen

Knowledge Centre WMC Kluisgat 5, 1771 MV Wieringerwerf, The Netherlands

Email: [email protected], [email protected] web page: http://www.wmc.eu

Keywords: Test, DCB, Mode I, Video, Crack, Growth, Static, Fatigue, Measure, Composite

ABSTRACT One of the key measurements to carry out during a DCB (Double cantilever beam) tests is the recording of the delamination or crack growth length. There are different types of DCB test methods such as the mode I static test (ASTM D5528), the fatigue delamination growth onset (ASTM D6115) or an extended fatigue test to characterise the crack growth rate until failure [1]. In all three cases a measurement of the delamination length is required in order to calculate the fracture toughness GIC, the strain energy release rate G and the crack growth rate . The delamination length can be measured using a crack gauge (a single-use limited length strain-gauge type sensor which is bonded over the crack area), or manually (the operator documents the crack length either from visual observations (a video can be used), or using an apparatus that indicates a prescribed crack length increment into the measurement). In both cases, the crack length resolution is ca. 0.5 mm. These methods are labour intensive, crack gauges are limited in length and unreliable in fatigue, and in the manual methods, synchronisation of crack growth information with the load and displacement signals is limited in accuracy, the test is adversely influenced by the need for starting and stopping the test (especially in fatigue), and, depending on the exact implementation (use of video or not), the traceability of the measurement is limited, i.e. objectivity is compromised. An automated video method has the advantages of minimising labour as well as the human factor and being re-usable. The aim of this work is to evaluate an automated measurement of the delamination length in static and fatigue tests based on an in-house digital video processing algorithm and to present static and fatigue DCB tests results based on this tool. In the relevant published literature related to visual crack growth tracking via image processing [2–6] , the main issues that arise are related with: image quality, image filtering algorithms, the conversion from pixels to millimetres, the false positives due to the illumination, coupons shadowing, coupon movement during the test and the correlation between the delamination length measurement and the force and displacement measurements.

Figure 1: Output frame image after the image processing. Auxiliary position lines are placed at the

crack tip position, the center of the coupon and the contour of the coupon in order to position the coupon in the space

5151

F. Lahuerta, S. Raijmaekers, J.J. Kuiken, T. Westphal and R.P.L. Nijssen

The present methodology and algorithm developed in-house avoids the issues mentioned before. The methodology is based on the following procedure. The coupons are painted white and a black background is set up in the test frame in order to facilitate image filtering. An HD camera of 1920x1080 pixel resolution is set up perpendicular to the coupon’s observed surface. The image filtering is based on a three step algorithm, initially a band threshold by colour is applied, followed by a Canny filter with a Gaussian kernel of 5x5 pixels. An Otsu threshold finalises the filtering [7]. Based on the filtered image, the coupon is positioned for each frame allowing to measure the coupon inclination and subsequently the crack tip location can be found (see Figure 1). Each measured frame is correlated with the force and displacement signals via the data acquisition software in post-processing. The accuracy of the system has been calibrated with dummy shapes previously measured with the calliper, placed at different extreme positions and camera set ups. It shows accuracies better than ±0.5mm, which is the minimum required by the ASTM D5528 standard. A set of static and fatigue tests carried out with the developed methodology are shown in the present work. The coupons were manufactured with type E glass fibre and a common epoxy resin used in the wind industry. The static fracture toughness GIC (see Figure 2), the G-N fatigue onset curve and the strain energy increase versus the crack growth rate are reported. Additionally the evolution of the loss factor for fatigue tests is also shown.

REFERENCES

[1] A. Argüelles, J. Viña, A. F. Canteli, M. A. Castrillo, and J. Bonhomme, “Interlaminar crack initiation and growth rate in a carbon-fibre epoxy composite under mode-I fatigue loading,” Composites Science and Technology, vol. 68, no. 12, pp. 2325–2331, Sep. 2008.

[2] M. K. Y. S. Yarlagadda, A. AbuObaid and R. D. N. Hager, “An Automated Technique for Measuring Crack Propagation during Mode I DCB Testing,” Conference: 2004 SEM X International Congress & Exposition on Experimental & Applied Mechanics, 2004.

[3] C. Uhlig, O. Kahle, B. Wieneke, and M. Bauer, “Optical crack tracing - A new method for the automatic determination of fracture toughness for crack initiation and propagation,” in MicroMat 2000. Proceedings 3rd International Conference and Exhibition Micro Materials, 2000, pp. 618–629.

[4] V. Richter-Trummer, E. Marques, F. Chaves, J. Tavares, L. da Silva, and P. de Castro, “Analysis of crack growth behavior in a double cantilever beam adhesive fracture test by different digital image processing techniques,” Materialwissenschaft und Werkstofftechnik, vol. 42, no. 5, pp. 452–459, 2011.

[5] P. Dare, H. Hanley, C. Fraser, B. Riedel, and W. Niemeier, “An Operational Application of Automatic Feature Extraction: The Measurement of Cracks in Concrete Structures,” The Photogrammetric Record, vol. 17, no. 99, pp. 453–464, Apr. 2002.

[6] D. H. Ryu and S. H. Nahm, “Image processing techniques applied to automatic measurement of the fatigue-crack,” Key Engineering Materials, vol. 297–300, pp. 34–39, Aug. 2005.

[7] P. Levi, “Course Image Understanding,” University of Stuttgart, Germany.

Figure 2: Static test, coupon VN03I30. Force and

delamination length versus displacement curves

5252


Aalborg, 2013

EFFECT OF THE CRACK LENGTH MONITORING TECHNIQUE DURING FATIGUE DELAMINATION TESTING ON CRACK GROWTH DATA

D. Sans, J. Renart, J. A. Mayugo, J. Costa

AMADE. Escola Politècnica Superior. Universitat de Girona Campus Montilivi s/n. 17071 Girona, Spain


Keywords: Fatigue, crack growth, delamination, Paris Law, Fibre Bragg Grating

ABSTRACT

The interest on highly optimized composite structures for transport vehicles, mainly aircrafts, raises the need to introduce more refined analysis in the mechanical design. Interlaminar fatigue damage is one of the mechanisms that need to be properly incorporated in the design. This is challenging from the side of both the methods and the experimental tests to provide the required data. Interlaminar fatigue tests on composites are being conducted since the early 80’s, but they are so sensitive to subtle variations of experimental methods and to data reduction procedures, that the results are far from being considerable as robust data. The expected outcome from fatigue tests includes the onset data (the number of cycles required, for a certain load level, to initiate damage) and the propagation data (the crack growing rate at each load level). Regarding propagation data, the most desired parameter is the threshold load level (the load level for which crack growth rate is below the measurable rate). Both onset and propagation data rely on the crack length monitoring during the test, which becomes the most crucial operation. Methodological possibilities include, among others, the interruption of the test to measure the crack length, sophisticated image monitoring systems which automatically read the crack length without test interruption, or the indirect assessment of the crack length by means of the real time measurement of the specimen compliance somehow related to crack length by a calibration procedure. To the author’s knowledge, there is not a published work on the accuracy of these methods and the uncertainty induced on the obtained crack growth rate data (Paris law parameters and load threshold). This communication makes use of Fibre Bragg Grating sensors, FBG’s, embedded in the specimens to measure the crack length during a fatigue delamination test with high accuracy. The optical fibres are located in plane just above the crack propagation plane, so the deformation they read can provide a real time direct indication of the position of the crack length. During the tests, two alternative traditional methods are applied: one method based on the image observation of the edge of the specimen and a second method relying on the real-time measurement of the specimen compliance. An initial study related the static fracture toughness data with the crack growth rate (Fig. 1). On the other hand, the comparison between the propagation curves (Fig. 2) showed that the three explored methods provided almost the same exponents of the Paris law. However, the differences in the threshold load were of more than 50%. In addition, the estimated crack growth rate for a given load level could be underestimated by two magnitude orders with the traditional methods with respect to the FBGs results (Fig. 2). The divergence of results should be associated to the inaccuracy of the measurement of the crack length. For instance, it is well known that the crack does not evolve with a straight crack front and, in consequence, there is a shift between the position identified from the edge of the specimen and the position of the centre of the crack. However, Fig. 3, shows that the difference is not constant during the test but evolves with the crack extension. To discuss this effect, the strain fields around the FBG’s sensors at different crack extensions, which reflect the fracture process zone, are analyzed.

5353

D. Sans, J. Renart, J. A. Mayugo and J. Costa

Fig. 1. Crack propagation rates compared with the static fracture toughness.

Fig. 2. Fatigue crack growth curves obtained by monitoring the crack length by means of the image monitoring of the edge of the specimen (VIS) and by the Fibre Bragg Grating sensors (FBG)

Fig. 3. Crack lengths during the fatigue testing according to VIS and FBG method. The difference

between both evolves during the test.

5454


Aalborg, 2013

BIAXIAL FATIGUE TESTING OF GLASS/EPOXY COMPOSITE TUBES

Paolo A. Carraro, Marino Quaresimin

Department of Management and Engineering, University of Padova, Stradella S. Nicola 3, 36100 Vicenza, Italy


Keywords: multiaxial/biaxial fatigue, tubes, damage mechanics

ABSTRACT Thanks to their high specific strength and stiffness, composite materials have undergone an increasing importance in the design of structural parts, in many application fields. Typical components, such as wind turbine blades, cranks and landing gears are subjected to fatigue loading conditions characterized by a local multiaxial stress state and require therefore suitable and reliable design procedures, still missing in the literature [1]. Although the final fatigue failure of a multidirectional laminate is in general controlled by the load bearing plies (parallel to the external loads), it is proved that transverse matrix cracks nucleate and propagate in the off-axis layers since the early stages of fatigue life. The accumulation of off-axis cracks produces stress concentrations in the 0° layers, as well as the degradation of the global elastic properties of the laminate, which are dependent on the density and on the length of these cracks [2]. It is therefore important to predict the cycles spent for the initiation and propagation of matrix cracks in the off-axis layers, in order to predict their effect in the laminate stiffness. Reliable criteria of general validity have to be based on the damage mechanisms occurring at the microscopic scale, which are responsible for the initiation and propagation of ply cracks. However, a comprehensive experimental investigations which correlate the initiation and propagation mechanisms to the multiaxial stress state are not yet available the literature. Aim of this work is to present an overview on the experimental procedure and testing plan developed at DTG-University of Padova for the characterization and understanding of the influence of the multiaxial loading condition on the fatigue behaviour of composite laminates, with particular reference to the dependence of the damage mechanisms and their evolution on the stress state, as a basis for the definition of a predictive multiaxial fatigue criterion. Tubular samples subjected to combined tension/torsion loadings have been identified as the best way to match this requirement, in particular for studying the effect of the shear stress over the longitudinal or transverse stresses. In fact, if unidirectional (UD) tubes are realised with the fibres oriented at 0° (parallel to the tube's axis), a tension/torsion external loading condition will lead to a combination of longitudinal and in-plane shear stresses ( 1 and 6). On the other hand, if the fibres are oriented at 90° (normal to the tube's axis), the presence of the transverse and in-plane shear stress components ( 2 and 6) is achieved by combined tension and torsion loading. The former one is a very interesting condition for analysing the matrix-dominated behaviour of a UD lamina, which controls the damage evolution in the early stages of a laminate's fatigue life. Glass/Epoxy tubes with lay-up [90n] and internal diameter of 19 and 34 mm have been tested as first part of the project. Larger tubes provide a lower variation of the shear stress across the thickness, and therefore a stress state which closer to a plane stress condition. On the other hand they are more complicated to test since they require a dedicated gripping system, which is not worth the effort since the fatigue results for the two kind on tubes have been found to be comparable. When testing UD tubes, only information about the cycles spent for crack initiation can be obtained, since there is not

5555

Paolo A. Carraro, Marino Quaresimin

a visible progressive damage evolution [3]. To investigate the influence of biaxial stress states on crack propagation, the 90° UD layers have to be constrained by stiffer thin plies to make the propagation more stable, but particular attention has to be paid to avoid the introduction of further stress components in other directions. For this purpose, tubes with lay-up [0F/903/0F] (the subscript F meaning Fabric) have been produced by using very thin 0° fabric layers which does not alter the biaxial stress state, providing a very small contribution of 1 (about -7% of 2). Damage evolution was monitored by eye using an in-house developed LED lighting system positioned inside of the tubes. The frequency of the observations was adapted to the expected value of the fatigue life in order to have a reliable measure of the cycles spent for the nucleation of cracks. The damage evolution during high cycle fatigue tests has been monitored also by means of lock-in analysis with an infrared camera FLIR SC7600, which allows the onset of cracks during fatigue life to be detected. The minimum crack length detectable using this technique was about 0.5 mm. The paper illustrates the results of comparative fatigue tests investigating the influence of the tubes geometry (wall thickness to diameter ratio) on the transverse fatigue response, as well as the effects of an increasing shear stress component on the transverse fatigue strength (figure 1) and damage evolution.

5

50

1,E+02 1,E+03 1,E+04 1,E+05 1,E+06 1,E+07

2,m

ax[M

Pa]

Cycles to failure Nf

type Atype Atype A

12 = 012 = 1 12 = 2

10

20

3040 R = 0.05

Figure 1: Comparison of fatigue data for increasing shear stress component

(biaxiality ratio 12= 6/ 2)

REFERENCES

[1] M. Quaresimin, L. Susmel, R. Talreja. Fatigue behaviour and life assessment of composite laminates under multiaxial loadings. International Journal of Fatigue , 32, 2010, pp. 2-16

[2] C. V. Singh, R. Talreja, A synergistic damage mechanics approach for composite laminates with matrix cracks in multiple orientations, Mechanics of Materials, 41, 2009, pp. 954–968

[3] P. A. Carraro, M. Quaresimin, Fatigue damage evolution in [0F/90U,3/0F] composite tubes under multiaxial loading, Proceedings of ICCM18, Jeju Island, Korea, 21-25 August 2011

5656


Aalborg, 2013

Oral session 5 – Defects, delamination and debonding (4 presentations)

5757


Aalborg, 2013

TENSION AND COMPRESSION TESTING OF MULTI-DIRECTIONAL LAMINATES WITH ARTIFICIAL OUT OF PLANE WRINKLING DEFECTS

Stephen R. Hallett, Mike I. Jones and Michael R. Wisnom

Advance Composites Centre for Innovation and Science, University of Bristol, UK Email: [email protected], web page: http://www.bristol.ac.uk/composites/

Keywords: Defects, Compression, Tension, Strength

ABSTRACT

An important part of composites design is the establishment of material allowables for use in stress analysis and prediction of failure for components. This is typically done using laboratory scale specimens cut from flat plates, which give high quality and thus good results for stiffness and strength. This level of quality cannot always be achieved when manufacturing components with complex geometry or significant thickness. It is therefore necessary to establish the strength for “as manufactured” part quality, giving material allowable “knock-downs” to account for the presence of defects. There exists a large range of defect types, arising from a variety of sources [1]. One of the most important defect types is that of out-of-plane wrinkling, since this can significantly reduce compressive strength. In order to characterise the effect of such defects it is most effective to recreate the out-of-plane wrinkle in controlled laboratory conditions. Here an experimental study has been undertaken on carbon/epoxy (IM7/8552) multi-directional laminates, as compared to the majority of other studies in the literature which have examined uni-directional or cross-ply laminates. The manufacturing technique chosen for creating the artificial wrinkle specimens was that of inserting uncured 90 pre-preg strips into the laminate, only adjacent to the existing 90 plies. This was motivated by the desire not to introduce any pre-cured or non-composite inserts which might alter the interface strength or any new interfaces which again might change the interfacial properties between plies. A schematic and image of the layup is shown in Figure 1.

Figure 1 Schematic of out-of-plane wrinkling manufacture with inserted extra plies

and actual specimen cross-section

In order to achieve a significant level of waviness it was necessary to use a relatively thick specimen, 6mm, with three quasi-isotropic sub-laminates, [45/90/-45/0]3. A very thick composite caul plate (10mm pre-cured composite) was used to obtain a constant thickness, despite the additional inserted material. The final specimen gave a maximum wrinkle angle of approximately 10.2 .

-450

+45

-45

-45

+45

+4590

-450

+450

-45

+450

-45

+4590

90

90

90

90

00

3 ply

4 ply

5 ply

5959

Stephen R. Hallett, Mike Jones and Michael R. Wisnom

Specimens were tested in tension and compression for both pristine and defect configurations. During the tests both standard and high speed video was used to monitor the development of the damage. The compression case is shown first in Figure 2. Of most interest is the fact that these images clearly show the onset of damage as being through matrix cracking and delamination. It is only after the delamination has propagated very suddenly between frames 39 and 40, that the onset of fibre compression failure can be observed.

Figure 2 High speed images at 60,000 fps of final failure of the compression test

A similar situation is observed for the tensile test result, shown in Figure 3. Here matrix cracking and delamination is also the first observable failure. In this case delamination occurs at much lower relative loads compared to the compression test, about 50% of ultimate failure. This delamination then grows stably away from the wrinkle region until ultimate failure by fibre tensile rupture. Results are summarized in Table 1 below.

Figure 3 High speed images at 62,500 fps of first delamination in the tensile test and image of failed specimen

Table 1 Summary of test results Tension Compression

Pristine Defect

Pristine Defect First Delam. Ultimate

Mean (MPa) 750.0 379.6 580.2 625.4 438.0

cv (%) 1.8 1.8 5.0 6.0 6.4

REFERENCES

1 Potter KD, Understanding the origins of defects and variability in composites manufacture, 2009, ICCM17, Edinburgh

Frame #31 Frame #39 Frame #40

First failure

Frame #41

Frame #69 Frame #70 Frame #71

6060


Aalborg, 2013

EFFECTS OF IN-PLANE WAVINESS ON THE PROPERTIES OF CARBON COMPOSITES – EXPERIMENTAL AND NUMERICAL ANALYSIS

Jan-Philipp Fuhr1, Jochen Baumann1, Frank Härtel1, Peter Middendorf1 and Nico Feindler2

1Institute of Aircraft Design, University of Stuttgart Pfaffenwaldring 31, D-70569 Stuttgart, Germany

Email: [email protected], web page: http://www.ifb.uni-stuttgart.de

2Audi Lightweight Design Centre, AUDI AG NSU-Straße 1, D-74148 Neckarsulm, Germany

Email: [email protected], web page: http://www.audi.de

Keywords: Composites, Draping effects, Fibre waviness, Mechanical properties, Finite elements

ABSTRACT High volume production of parts made of carbon fibre reinforced plastics for automotive applications requires manufacturing processes that are quick, highly automatable and cost efficient. During the past years, RTM (resin transfer moulding) processes in combination with a preforming step to drape multi-axial carbon fabrics onto the required shape before infusion turned out to be one promising process amongst others to fit these requirements. During the draping and infusion process of dry carbon fabrics, different effects can occur that influence the local laminate architecture of the final part. Fibre shearing and deviation associated with local variations of fibre volume fraction as well as in-plane waviness of fibres are one of the main process influenced effects. Waviness occurs in zones where fibres are exposed to longitudinal compression stresses during draping and has a potentially huge impact on the mechanical properties within the affected zones depending on the wave characteristics [1-3]. To be able to represent the mechanical behaviour of corresponding laminates in a finite element model it is necessary to study the influences based on experimental testing.

0

200

400

600

800

1000

1200

1400

1600

1800

0,000 0,005 0,010 0,015 0,020

stre

ss [M

Pa]

strain [-]

no wavinessL=20; A=0.6L=10; A=0.3L=10; A=1.5L=20; A=2.6specimen failurestrain gauge failure

116,9 114,4

81,9

38,8 36,5

0

20

40

60

80

100

120

140

Youn

g's

mod

ulus

[GPa

]

Figure 1: Tension test results for CFRP specimens with different characteristics of in-plane waviness.

Tests have been carried out under quasi-static tensional and compressive loading with specimens made of carbon fibres and an epoxy resin system. The stress-strain curves and the Young’s modulus of one series of tests with different periodic in-plane waviness are shown in figure 1. As indicated, the results belong to four different amplitudes (0.3 mm, 0.6 mm, 1.5 mm, 2.6 mm) paired with two different wavelengths (10 mm, 20 mm). An increasing wave amplitude leads to a decreasing stiffness and

6161

Jan-Philipp Fuhr, Jochen Baumann, Frank Härtel, Peter Middendorf and Nico Feindler

strength and a distinct non-linear behaviour, whereas an increasing wave length lowers the effect. It depends on the corresponding degree of local fibre deviation, which leads to a highly anisotropic behaviour. One of the key challenges for parameter identification of in-plane waviness is the preparation of reproducible test specimens with specific characteristics of waviness (varying wave length and amplitude). Different approaches are presented and compared to prepare the specimens, including a TFP (tailored fibre placement) process as well as a test device that is developed to treat usual fabrics “off-the-roll” with different grades of waviness in a reproducible manner for testing. The test series include as well results of optical 3D deformation analyses based on digital image correlation to visualize inhomogeneous strain distributions within the specimen due to local stiffness variation as well as edge effects. To estimate the impact on the global structural performance of a composite part due to local appearance of fibre waviness, numerical finite element models are required to predict the mechanical properties of misaligned regions. Studies on micromechanical models are promising, since they represent the local fibre architecture [4]. To meet the requirements even of large explicit crash models the behaviour is adapted to a phenomenological approach using 2D shell material models. Thus, effects like the local change in stiffness and the different failure modes can be analysed within an efficient time frame.

REFERENCES [1] H. M. Hsiao and I. M. Daniel, Effect of fiber waviness on stiffness and strength reduction of

unidirectional composites under compressive loading, Composites Science and Technology, 56, 1996, pp. 581-593.

[2] K. Potter, B. Khan, M. Wisnom, T. Bell and J. Stevens, Variability, fibre waviness and misalignment in the determination of the properties of composite materials and structures, Composites: Part A, 39, 2008, pp. 1343-1354.

[3] M. R. Piggott, The effect of fibre waviness on the mechanical properties of unidirectional fibre composites: a review, Composites Science and Technology, 53, 1995, pp. 201-205.

[4] G. Karami and M. Garnich, Micromechanical study of thermoelastic behavior of composites with periodic fiber waviness, Composites: Part B, 36, 2005, pp. 241-248.

6262


Aalborg, 2013

EXPERIMENTAL VALIDATION OF A MATRIX CRACK INDUCED DELAMINATION CRITERIA

L. Zubillaga1,4, A. Turon1, J. Costa1, S. Mahdi2 and P. Linde3

1AMADE, University of Girona Av. Montilivi s/n Building P-II, 17071, Girona, Spain

Email: [email protected]@[email protected]

web page: http://amade.udg.edu/ 2 Airbus Site de Saint Martin du Touch, 31060 Toulouse Cedex, France

Email: [email protected], web page: http://www.airbus.com/ 3 Airbus

Kreetslag 10, 21129 Hamburg, Germany Email: [email protected], web page: http://www.airbus.com/

4 IK4-Ikerlan P J. M Arizmendiarrieta 2, 20500 Arrasate-Mondragon, Gipuzkoa, Spain

Keywords: Composite plates, Modal analysis, Identification, Properties, Finite elements

ABSTRACT

The increase of the use of composite materials makes necessary the development of efficient design tools for industrial use. Different failure modes should be taken into account in the design of composites structures. One of those failures modes is delamination. Delamination can be caused mainly because of three different phenomena an impact, free-edges and matrix cracks induced delamination.

In a recent work the authors present a failure criterion for Matrix Crack Induced Delamination (MCID) [1]. However, the failure criterion was not fully validated with experimental data. In this work, the validation of the proposed failure criterion is performed. An experimental campaign has been carried out and the obtained experimental data has been compared to the formulation proposed in [1] and with the predictions obtained using cohesive elements [2].

Two different types of specimens have been tested. On the one hand, specimens without any pre-existing crack. Matrix cracks are generated by the loading and delamination will propagate from some of these cracks. These specimens are called MCID specimens. On the other hand, to minimise the influence of extensive matrix cracking that can arrest interlaminar delaminations, another set of specimens with some plies with a physical discontinuity have been tested. These specimens are denoted as Ply Discontinuity Induced Delamination (PDID) specimens. The test matrix is summarized in Table 1.

MCID Specimens PDID Specimens

[453/-453]s [03/03cracked]s[603/-603]s [03/05cracked]s[604/-604]s [03/09cracked]s

[302/-302/-604/302/-302/604]s [03/01cracked]s[452/-452/-604/452/-452/604]s [05/03cracked]s[452/-452/904/452/-452/904]s [09/03cracked]s

Table 1: Test matrix.

63

L. Zubillaga, A. Turon, J. Costa, S. Mahdi and P. Linde

The specimens in Table 1 have been tested under tensile loading. Two different optical devices have been used depending on the specimen type. A macro lens for MCID specimen where the localization of the delamination initiation was unknown and a QUESTAR(R) for PDID specimen where the location of the initial crack was known in advance have been used.

Figure 1: Damage development in [452/-452/-604/452/-452/604]s.

The experimental results obtained showed damage development on the edge of the specimens, for the case of MCID specimens, matrix cracks appeared in the laminate and delamination grew from them Figure 1. Additionally, for PDID specimens the failure of the specimen was even noticeable in the Force-Displacement curve where a sudden load drop was observed.

Finally, the data obtained from the experimental campaign has been compared to the prediction made by the MCID criteria proposed by [1] and cohesive zone model proposed by [2] in FEM code ABAQUS (R) model [3].

REFERENCES

[1] L. Zubillaga, A. Turon, P. Maimí, J. Costa, S. Mahdi and P. Linde. An energy based failure criterion for matrix crack induced delamination in laminated composite structures. (To be submitted).

[2] A. Turon, P.P. Camanho, J. Costa, C.G. Dávila. A damage model for the simulation of delamination in advanced composites under variable-mode loading. Mechanics of Materials, , 38, 2006, pp. 1072-1089 (

[3] ABAQUS Analysis User’s Manual, Simulia, www.simulia.com, 2011.

Dam

age

deve

lopm

ent

[452/-452/-604/452/-452/604]S

t1

t2

t3

t4

64


Aalborg, 2013

DETERMINATION OF FIBER/MATRIX INTERFACE DEBOND GROWTH PARAMETERS FROM CYCLIC LOADING OF SINGLE FIBER COMPOSITES

A. Pupurs1,3, J. Varna1, P. Brøndsted2 and S. Goutianos2

1Department of Engineering Sciences and Mathematics, Luleå University of Technology SE-97187 Luleå, Sweden

Email: [email protected] , web page: http://www.ltu.se

2Department of Wind Energy, Technical University of Denmark P.O. Box 49, Frederiksborgvevj 399, DK-4000 Roskilde, Denmark

3Swerea SICOMP

P.O. Box 271, SE-94126 Piteå, Sweden

Keywords: Fiber/matrix interface, Debond, Cyclic loading, Power law, Identification

ABSTRACT One of the key mechanisms of fatigue damage in unidirectional composites is fiber/matrix interface debond crack growth from a fiber break. The debond growth analysis in this study was based on fracture mechanics concepts of strain energy release rate. For calculation of the energy release rate analytical calculations in the steady-state growth region (long debonds) and FEM calculations for short debonds were performed. An exact analytical model following Nairn et.al. [1] was used in this study for Mode II strain energy release rate IIG calculations in the steady-state region. To calculate IIG for short debonds FEM calculations were performed using ANSYS software [2]. The FEM model was optimized for use with virtual crack closure technique [3], the size of the model was chosen based on parametric analysis performed in [4] and results were in good agreement with BEM results [5]. Considering debond as an interface crack we assumed that its growth in cyclic loading can be described by a power law, where the debond growth rate is a power function of the change of the strain energy release rate in one cycle. The power law expression was chosen similar to Paris law, an expression widely used in fatigue crack growth characterization in metals. It was assumed that the normalized debond length dnl (debond length divided by fiber radius ) increases with the number of load cycles N by following relation:

m

IIdn

GG

BdNdl m

GB *

* (1)

where IIGG is the energy release rate range in one cycle, *G is the normalizing constant equal to 1

J/m2, *B (where 2* 2/ frBB 22 frfB ) and m are the unknown parameters in the power law. To validate the applicability of a power law (1) and to obtain the values of the power law parameters cyclic loading of fragmented single glass fiber/epoxy specimen was performed. Measurements of the debond length increase with the number of load cycles in tension-tension fatigue were performed using optical microscope. Each sample was first loaded statically to create a fiber break. Then cyclic load with load ratio was R=0.1 and the frequency was 2Hz was applied. Two different values of maximal strain level

maxm were applied: 1) 1.76 % for samples A and B; 2) 1.32 % for sample C. These maximal strain levels correspond to approximately 80% and 60 % respectively of st1s1 , i.e., the average stress at the occurrence of the first fiber break. After certain number of cycles, the specimen was removed from the

6565

A. Pupurs, J. Varna, P. Brøndsted and S. Goutianos

testing machine to perform measurements of the debond length. The validity of the power law was certified by the obtained linear relationships in log-log axes. By fitting the modeling and experimental results the interface failure parameters in fatigue were determined. The determined values of power law parameters for samples A and B are summarized in Table 1.

Sample m ln(B*) A 22.94 -99.72 B 28.80 -118.60

Table 1: Values of power law parameters.

The determined interface fatigue parameters were validated at different stress levels. Fig. 1 shows experimental and modelling results for samples A,B and C. Modelling data for samples A and B are self-predictive, while modeling data for sample C were obtained using parameters from sample A. The good agreement between modelling and experimental results shows that the power law with respect to the strain energy release rate change is applicable for debond growth characterization in tension-tension cyclic loading of composites.

Figure 1: Experimental and modeling results for debond growth in cyclic loading.

REFERENCES

[1] J.A. Nairn, Y.C. Liu, On the use of energy methods for interpretation of results of single-fiber

fragmentation experiments, Composite Interfaces, 4, 1997, pp. 241-267. [2] ANSYS Release 13.0, ANSYS Academic Research, Ansys Inc., Canonsburg, PA, USA, 2011. [3] G. R. Irwin, Fracture, Handbuch der Physik, Vol. 5, Springer Verlag, Berlin, 1958. [4] A. Pupurs, J. Varna, Fracture mechanics analysis of debond growth in single fiber composite

under cyclic loading, Mechanics of Composite Materials, 47, 2011, pp. 109-124. [5] E. Graciani, V. Mantič, F. París, J. Varna , Numerical analysis of debond propagation in the

Single Fibre Fragmentation Test, Composites Science and Technology, 69, 2009, pp. 2514-2520.

6666


Aalborg, 2013

Oral session 6 – Damage and dynamic properties (4 presentations)

6767


Aalborg, 2013

FULL-FIELD CURVATURE MEASUREMENTS TO ASSESS IMPACT DAMAGE IN COMPOSITE PLATES USING AN INDICATOR BASED ON

MECHANICAL EQUILIBRIUM

Cédric Devivier1,3, Fabrice Pierron2 and Michael R. Wisnom3

1Mechanics Surfaces and Materials Processing, Arts et Métiers ParisTech Rue St Dominique, BP 508, 51006 Châlons-en-Champagne, France Email: [email protected], web page: http://www.msmp.eu

2Faculty of Engineering and the Environment, University of Southampton

Highfield, Southampton SO17 1BJ, UK Email: [email protected], web page: http://www.camfit.fr

3Advanced Composites Centre for Innovation and Science, University of Bristol

Queen’s Building, University Walk, Bristol, BS8 1TR, United Kingdom Email: [email protected], web page: http://www.bristol.ac.uk/composites/

Keywords: Composite Plates, Impact Damage, Deflectometry, Damage Detection, Virtual Fields Method

ABSTRACT

Detecting damage in composite materials has always been of great concern because their layered structure makes them very sensitive to transverse impacts. This paper presents a new damage indicator based on the application of the Virtual Fields Method to thin plates in bending. It calculates the local gaps in equilibrium over a sliding window and these gaps are sensitive to stiffness gradients and to violations of thin plate assumptions coming from local discrepancies in the deformation map. The experimental results are obtained using the grid method in deflectometry. This method provides local surface slopes which can be differentiated to obtain curvatures. Using the Love-Kirchhoff theory (thin plate), surface strains are obtained by multiplying the curvatures by half the thickness. However, where the thin plate assumptions do not apply, the obtained result is proportional to the curvatures but does not represent surface strains. A major advantage of using curvatures lies in the fact that they are much more sensitive to impact damage than in-plane strains. Also, thanks to the high sensitivity and low noise level of this measurement technique, curvature fields could be retrieved without smoothing during differentiation, ensuring excellent spatial resolution. The samples were made from IM7-8552 carbon-epoxy pre-preg (quasi-isotropic lay-up with 24 layers), coated with a thin layer of reflective resin, and their dimensions were 190 mm long, 140 mm wide and 3 mm thick. The samples were impacted at different energies (15 J, 20 J, and 25 J) and with different boundary conditions (clamped between two plates with circular cut-outs and simply supported). The plates were then tested in bending. The bottom corners and the top right corner from Fig 1 were simply supported and a 20 N point load perpendicular to the plate was applied near the top right corner of the plate. Fig. 1 presents the strain fields along (a) the horizontal direction, (b) the vertical direction, (c) the shear strain field, (d) the C-scan of the sample and (e) a map of gap in equilibrium for a sliding window of 40-by-40 pixels and a sliding pitch of 3 pixels. As opposed to cantilever beams [1], examining the strain fields directly does not lead to an easy detection of the damage, as illustrated in Figs a to c. This is caused by the fact that the strain maps for the plates have much higher frequency content than the ones from the cantilever beams (linear evolution of strains in the longitudinal direction).

69

Cédric Devivier, Fabrice Pierron and Michael R. Wisnom

The C-scan presented in Fig. 1-d highlights that damage is present in the lower part of the structure. The map of gaps in equilibrium from Fig. 1-e successfully reveals the location and extent of damage. The map is not uniformly zero away from the damage because noise is present in the measurements. Around the load point on the top right corner, there are non-zero values. They are caused by the fact that the underlying assumptions of the indicator are violated there (thin plate theory). This indicator combines the wealth of information contained in the three strain maps obtained using deflectometry to output a map that reveals the location and extent of damage in a nearly binary format.

(a) xx (b) yy (c) ss

(d) C-scan

(e) EG map

Figure 1: (a-c) Equivalent strain maps, (d) c-scan and (e) EG map for the impacted sample loaded at 20 N near the top right corner.

REFERENCES [1] C. Devivier, F. Pierron, and M. R. Wisnom, Damage detection in composite materials using

deflectometry, a full-field slope measurement technique, Composites Part A: Applied Science and Manufacturing, vol. 43, no. 10, pp. 1650–1666, 2012.

70


Aalborg, 2013

ABSTRACT The LS-DYNA strain rate dependent composite material model MAT_158 is calibrated by performing tensile tests on 0° and 90° UD test coupons of Glass/epoxy and Glass/Lpet composite materials. Strain rate effects are captured by testing at different strain rates using a high speed servo hydraulic machine. High-speed Digital Image Correlation is used for better assessment of failure modes in the tests and for assessing the strains. Boundary conditions from the experiments are modeled in the numerical simulation and the results from the simulation are compared to the experimental data. Further, the calibrated material model is used for simulation of tensile testing of a quasi-isotropic layup which are validated experimentally as well. This comparison is carried out to investigate if strain effects obtained for UD testing can be transferred to a different layup and still yield reasonable numerical results. Numerical simulation of blast events on composite panels requires accurate material models that can reproduce the impact behavior of the material. The explicit FEM code for transient analysis, LS-DYNA [1] offers different composite material models with different types of failure modes included for modeling of transient events such as mine blasts. However each model must be carefully calibrated and validated against experimental data to ensure the behavior of the material is estimated correctly by the model. The choice of material model for a particular transient modeling problem is critical to reproduce the correct material response [2]. Increasing strain rates may influence the failure of the material, and it is important that the chosen material model can account for these effects; this is achieved by calibration. However the available composite material models in LS-DYNA [3] require a large number of material parameters and material calibration becomes a cumbersome task. Including strain rate effects increases the number of required tests and each test must provide sufficient data for accurate model calibration. This inevitably results in many tests and very large data sets. Digital Image Correlation is a powerful tool for high speed characterization of composite material to reveal the macroscopic failure pattern at increased strain rates [4]. The full-field nature of the data can support the understanding of the failure modes the material model can simulate. The DIC has been applied by employing high-speed cameras and the displacement and the strain extracted are used here to provide a better understanding of failure modes and to help validate the materials models. A further benefit to the DIC approach is the full field data obtained which allows the immediate visualization of how the load is distributed in the specimen (see Fig. 1). It can be seen if the load is distributed uniformly across the specimen and hence provide an indication of the success of the test. This is particularly important in high speed testing when special grips are often used to clamp the specimens that are not a stable as standard test machine grips. This study examines how well strain effects, seen in a UD layup with respect to failure loads, can be represented by the MAT_158 material model. It is assumed that the quasi isotropic layup will exhibit different failure modes compared to the UD layup and differences will be seen between the simulated and measured response.

CALIBRATION OF LS-DYNA STRAIN RATE DEPENDENT COMPOSITE MATERIAL MODELS

R. Eriksen1*, C. Berggreen1, J.M. Dulieu-Barton2

S. 1Department of Civil Engineering, Technical University of Denmark, Nils Koppels Alle,

2800 Kgs. Lyngby, Denmark Email: [email protected] , webpage: www.dtu.dk

2School of Engineering Sciences, University of Southampton, Highfield, Southampton,

SO17 1BJ United Kingdom Email: [email protected] , webpage: www.southampton.ac.uk/

* Corresponding author([email protected]) Keywords: Composites, Strain rate, Material model, calibration

7171

Gglass/epoxy unidirectional composites are known to be strain rate sensitive for the failure stress [5] and constitute a good base material to assess how well the strain effects are captured in the material model. Fig 1. shows a crossply glass/epoxy specimen failing two places at the same time. Fig 2shows strain rate effects for a glass/epoxy cross ply laminate where strain rate effects are observed at even modest increases in strain rate.

Fig. 1 Cross ply Glass/epoxy specimen failing two places with a growing crack in between.

Images show major strain.

Fig. 2 Strain rate effect for cross ply Glass/epoxy specimens. Each dot is the mean of 5 tests [6].

1. Oasys LS-DYNA Environment . In: . http://www.oasys-software.com/dyna/en/software/ls-dyna.shtml. Accessed November/8 2010

2. Wright A, French M . J Mater Sci 43:6619-6629; 6629(2008)

3. Livermore Software Technology Corporation LS-DYNA Keyword User's Manual. Livermore Software Technology Corporation (LSTC)(2007)

4. Eriksen R, Berggreen C, Boyd SW, Dulieu-Barton JM, . ICEM 14, 14th International Conference on Experimental Mechanics, Poitiers, France, Edited by Fabrice Brémand; EPJ Web of Conferences, Volume 6, id.31013 6:31013(2010)

5. Barré S, Chotard T, Benzeggagh ML . Composites Part A: Applied Science and Manufacturing 27:1169-1181(1996)

6. Eriksen R High speed characterization of composite materials. Master Thesis. Department of Mechanical Engineering, Technical University of Denmark(2010)

0.000

25 0.5 2 30 45400

500

600

700

800

Strainrate ( /s)

Failu

re s

tres

s (M

Pa)

7272


Aalborg, 2013

DAMAGE ACCUMULATION INVESTIGATION IN FIBER-REINFORCED POLYMER-MATRIX COMPOSITES: FROM TEST COUPONS TO STRUCTURAL ELEMENTS

Andreas J. Brunner1

1Laboratory for Mechanical Systems Engineering, Empa, Swiss Federal Laboratories for Materials Science and Technology

Überlandstrase 129, CH-8600 Dübendorf, Switzerland Email: [email protected], web page: http://www.empa.ch

Keywords: damage accumulation, polymer-matrix composites, test coupon and structural scale,

acoustic emission monitoring, failure prediction

ABSTRACT Damage accumulation in fibre-reinforced, polymer-matrix (FRP) composite structures and elements is investigated with experimental methods and modelling for an improved understanding of damage mechanisms and their interaction leading to failure (see, e.g., [1]). The inherently complex damage behaviour of FRP composites yields stochastically varying behaviour ([2]) even in relatively simple experiments on test coupons. In FRP structural elements, additional complexity may arise from manufacturing defects which are not taken into account in design, damage models and simulations under service loads ([3]). Acoustic emission (AE) is a non-destructive test method [4] that yields information on the occurrence of damage in FRP composites and other materials with high time resolution (milliseconds and lower). This allows exactly identifying the loads at which damage occurs and for discriminating different stages of damage accumulation. Examples of the application of AE monitoring to damage in FRP composites have recently been reviewed in [5] and cover length scales from single fibre tests [6] to engineering structures such as, e.g., bridge decks [7]. Analysing AE activity, AE intensity, AE Felicity-ratio, and AE source location as a function of load [8] is proving useful for failure prediction in various FRP composites from test coupons to structural scale.

Figure 1: Quasi-static, step-wise tensile load test on GFRP (left) and CFRP laminate (right) showing AE signal amplitude distribution and load versus time (note difference in amplitude scales).

Figure 1 shows AE signal amplitudes from tensile strength tests on glass and carbon fibre reinforced (GFRP and CFRP, respectively) composite test coupons. Clear differences in number of AE signals and their amplitudes are seen between GFRP and CFRP. Repeating tests on nominally identical laminate coupons results in some variability in strength which clearly affects damage accumulation behaviour and hence AE activity and AE intensity as a function of load. For a model description and comparison between different FRP composite materials, the challenge then is to suitably average the individual single specimen data. AE monitoring allows detecting differences in damage accumulation including “anomalous” specimen or element behaviour that may have to be excluded from the ensemble to be averaged. The Felicity-ratio is defined as ratio between the load at which AE is again observed upon reloading and the previous maximum load. In FRP and other composites, AE does

7373

Andreas J. Brunner

usually not set-in when passing the previous maximum load level, but at somewhat higher or lower loads depending on the amount of damage accumulation. Figure 2 shows the AE activity (cumulative number of AE signals per channel) and the Felicity-ratio for two channel groups as a function of load level for a sandwich T-joint with balsa-wood core and CFRP-facings under three-point bending. AE activity shows that the most active channel is changing from the one side (channel 5) to the other side (channel 9) of the T-joint. Similarly, the minimum value of the Felicity-ratio is changing from one channel group (1-5) to the other (6-10) at increasing loads (crossover between 100 and 120 kN). This is interpreted as indicating a stress-redistribution when reloading to the level of 120 kN due to significant damage accumulation in one side of the T-joint which, however, was not sufficient to induce failure. Without AE monitoring, this phenomenon could not have been detected.

Figure 2: AE activity (cumulative number of AE signals (left) and Felicity-ratio versus load step for one balsa-wood core-CFRP sandwich T-joints under tensile three-point bending (failing at 160 kN).

Contributions from and discussions with Dr R.A. Nordstrom (now at Portland State University, USA) and technical test support by M. Heusser, B. Jähne, and D. Völki are gratefully acknowledged.

REFERENCES [1] A. Muc, P. K dziora, Z. Krawiec, Damage analysis and monitoring of composite materials and

structures under cyclic loads, Procedia Engineering, 10, 2011, pp. 1315-1320 (doi:10.1016/j.proeng.2011.04.219)

[2] S. Sriramula, M.K. Chryssanthopoulos, Quantification of uncertainty modeling in stochastic analysis of FRP composites, Composites Part A, 40, 2009, pp. 1673-1684 (doi:10.1016/j.compositesa.2009.08.020)

[3] B. Hayman, Approaches to damage assessment and damage tolerance for FRP sandwich structures, Journal of Composite Materials, 9, 2007, pp. 571-596 (doi: 10.1177/1099636207070853)

[4] .C.R. Rios-Soberanis, Acoustic emission technique, an overview as a characterization tool in materials science, Journal of Applied Research and Technology, 9, 2011, pp. 367-379.

[5] K. Ono, A. Gallego, Research and application of AE on advanced composites, Proceedings 30th

European Conference on Acoustic Emission 2012 (Eds. A. Gallego, K. Ono) ISBN 978-84-615-9941-7, Granada, Spain, September 12-15, 2012, EWGAE, Paper 34, pp. 1-44.

[6] R.A. Nordstrom, M.B. Sayir, P. Flüeler, Comparison of AE from glass fiber breaks in bundles and in the single fiber fragmentation test, Proceedings 22nd European Conference on Acoustic Emission 1996 (Eds. L.M. Rodgers, P. Tscheliesnig), Aberdeen, UK, May 29-31, EWGAE, pp. 93-98.

[7] E. Velazquez, D.J. Klein, M.J Robinson, J.B. Kosmatka, Acoustic emissions (AE) monitoring of large-scale composite bridge components, Proceedings SPIE Nondestructive characterization for composite materials, aerospace engineering, civil infrastructure, and homeland security, 2008 (Eds. P.J. Shull, H.F. Wu, A.A. Diaz, D.W. Vogel), 6934, 2008, pp. H9340-H9340.

[8] A.J. Brunner, R.A. Nordstrom, P. Flüeler, A study of acoustic emission-rate behaviour in glass-fiber-reinforced plastics, Journal of Acoustic Emission, 13, 1995, pp. 67-77.

7474


Aalborg, 2013

DCB FRACTURE SPECIMENS WITH SIDE NOTCHES

H. Toftegaard1, M. Rask1, S. Rasmussen1 and B.F. Sørensen1

1Department of Wind Energy, Section of Composites and Materials Mechanics, Technical University of Denmark

P.O. Box 49, Frederiksborgvej 399, DK-4000 Roskilde, Denmark Email: [email protected], web page: http://www.dtu.dk

Keywords: Composite laminates, Fracture resistance, Finite elements, Digital Image Correlation

ABSTRACT

The J integral is an important concept for determination of fracture energy (critical energy release rate), fracture resistance and cohesive laws. The double cantilever beam loaded with uneven bending moments (DCB-UBM) [1] is a comprehensive technique to measure the fracture properties in pure opening mode (I), pure sliding mode (II) and combinations of these (mixed mode). For the DCB specimen loaded in pure bending the J integral value is independent of crack length. This makes it possible to obtain stable crack growth for many materials. The J integral is applicable both for linear elastic fracture mechanics (LEFM) with a small fracture process zone and for laminates that exhibit large scale fibre bridging. For DCB specimens with constant width, the J integral can be expressed in terms of the applied moments, specimen geometry and specimen elastic constants for a number of different configurations. These configurations include a crack in the symmetry plane of a specimen made from a single material [2] as well as a crack in the symmetry plane of a sandwich specimen made from two materials [3]. For some materials and specimen configurations, however, it would be advantageous to use a layered DCB specimen, where some layers are wider than others. That would be the case for a thin ceramic layer glued to two (wider) steel beams [4]; and for a tough fibre reinforced laminate with thermoplastic matrix, where side notches could enable crack growth before occurrence of beam bending failure. This study will focus on the J integral analysis of DCB specimens with side notches. The J integral is defined as a planar contour integral in a length-height plane of the DCB specimen. The properties in the width direction are assumed to be constant. For a DCB specimen with side notches, however, the width is not constant. The two inner layers adjacent to the crack have a smaller width than the two outer layers. In the present study we transform the 3D geometry of DCB specimens with various layers (different heights , different widths and different Young’s moduli ) to a plane (2D) problem, where the widths of all layers are equal to the original crack width ( ). It can be shown [4] that for LEFM conditions the J integral result is equal to the energy release rate, providing that Young’s moduli are transformed as (an equivalent cross-section, preserving the products of width and Young’s modulus for each layer). For a DCB made from a single material and with side notches (Fig. 1b) this concept of an equivalent cross-section leads to the following equations for J assuming plane stress:

, (1) We wish to investigate whether this approach also holds for DCB specimens with side notches in case of materials exhibiting large scale bridging. For that purpose a combined experimental and numerical study is conducted.

75

H. Toftegaard, M. Rask, S. Rasmussen and B.F. Sørensen

DCB-UBM fracture resistance tests on specimens without and with side notches (Fig. 1) are performed for a material exhibiting large scale bridging. For the specimens with side notches the J integral values are obtained using the equivalent cross-section approach (Eq. 1). Acoustic emission (AE) is used to determine initial fracture (Fig. 2a). (a) (b)

Figure 1: Schematics of a DCB specimen of constant width and a symmetrical crack (a), and a DCB specimen with side notches and a symmetrical crack (b).

A 3D finite element (FE) analysis is used to calculate the fracture energy for the DCB specimen with side notches (Fig. 2b). The strains from the FE analysis are assessed using digital image correlation (Fig. 2a), and the cohesive laws of the FE analysis are determined from the curves of fracture resistance versus normal and tangential end-openings from the test of the DCB specimen without side notches (Fig. 1a). The fracture energy from the FE analysis is used to assess the J integral of Eq. 1. (a) (b)

Figure 2: The J integral study combines experiments on specimens equipped with extensometer, LVDT’s, AE transducers and speckle pattern for DIC measurements (a) with a 3D FE model (b).

The authors would like to acknowledge the financial support from the Danish Defence Acquisition and Logistics Organization (DALO) for the RESIST project.

REFERENCES [1] B.F. Sørensen, K. Jørgensen, T.K. Jacobsen and R.C. Østergaard, DCB-specimen loaded with

uneven bending moments, Int. J. Fract, 141, 2006, pp. 163-176. [2] B.F. Sørensen and T.K. Jacobsen, Characterizing delamination of fibre composites by mixed

mode cohesive laws, Composites Science and Technology, 69, 2009, pp. 445-456. [3] G. Bao, S. Ho, Z. Suo and B. Fan, The role of material orthotropy in fracture specimens for

composites, Int. J. Solids Structures, 29, 9, 1992, 1105-1116. [4] S. Goutianos, H.L. Frandsen and B.F. Sørensen, Fracture properties of nickel-based anodes for

solid oxide fuel cells, J. Eur. Ceram. Soc., 30, 2010, pp. 3173-3179.

76


Aalborg, 2013

Oral session 7 – Test methods (3 presentations)

7777


Aalborg, 2013

NOVEL TEST SETUP FOR DETERMINATION OF HIGH TEMPERATURE MECHANICAL PROPERTIES OF COMPOSITES

A.Chripunow1, M. Ruder2

1 Bundeswehr Research Institute for Materials, Fuels and Lubricants (WIWeB)Institutsweg 1, 85435 Erding, Germany

Email: [email protected], web page: http:// www.baain.de/wiweb

2Regensburg University of Applied SciencesPruefeninger Str. 58, 93049 Regensburg, Germany

Keywords: Carbon Fibre Reinforced Plastics, Elevated Temperatures, Testing Device

ABSTRACT

Compared to metallic materials, carbon fibre reinforced plastics (CFRP) show a much faster degradation of their mechanical properties under fire conditions. In the case of structural materials in aerospace applications it is very important to know and understand their mechanical behaviour under these conditions. Since the occurring temperatures are much higher than the usual operating temperatures of CFRP these materials show complete different properties regarding their strength and stiffness and their time-dependence.Extensive research has been done to evaluate the relationship of the mechanical properties and thesurrounding temperature of fibre reinforced plastics, especially for naval used glass fibre/vinyl ester systems [1, 2]. Less work has been done on carbon/epoxy systems [3], especially on aerospace applied prepreg based material [4]. The hitherto existing test apparatus utilized to obtain strength and moduli values at elevated temperatures usually did not reach temperatures above 250°C or lacked in proper strain measurement. However, this temperature range is of importance if the endurance of composite structures under fire loads or elevated temperature needs to be calculated, as in the case of in-flight or post-crash fires.The usual equipment needed to determine high temperature mechanical properties is quite diverse and expensive. There are two general possibilities to introduce heat to the specimen; either the whole experimental setup including the specimen, grips and strain measurement device is heated [4], which is a common practice, or only the specimen itself is heated [2, 3]. The main issue of the first method, if the moduli are to be determined as well, is the strain measurement at temperatures above 250°C. Since the measurement device, usually strain gauges glued or extensometer clipped on to the test specimen, are not resistant to such conditions, only optical measuring techniques are suitable. Furthermore, most commercially available furnaces for use in universal testing machines do not reach temperatures above 300°C. Most importantly, the heating of the entire test setup is more time consuming and also implies the risk of failure of the specimen within or close to the grips leading to invalid strength values. These shortcomings can be overcome when a smaller furnace is used, heating only the specimen between the grips. The heat-up process is accelerated and the risk of invalid rupture nearby the grips is minimised due to the thermally weakened middle section of the specimen.The test setup presented in this work has many other advantages, in addition to the ones described in [2, 3]. A custom-built furnace was utilised keeping the heated part of the test specimen to a minimum and using a heat gun as the heat source adapting a principle shown in [5]. The setup is modular; the temperature cell can be exchanged depending on the mechanical tests and specimen sizes in order to ensure always optimal test conditions. For the time being cells for tensile and bending had been developed (see Figs. 1 and 2). With the latter one it is possible to perform three-point-bending tests, interlaminar-shear-strength tests as well as tests to determine the interlaminar fracture toughness (mode II). The strain and the deflection respectively can for both setups be measured within the heated zone of the test specimen and the harmful pyrolysis gases are extracted. It is possible to test at temperatures of up to 700°C and to test under an inert gas atmosphere by insertion of nitrogen, not to mention that it is much cheaper than commercially available furnaces. A special flow deflection setup

7979

A. Chripunow, M. Ruder

inside the furnaces ensures a uniform heat-up of the specimen and thereby prevents it from being hit directly by the hot air flow. The data gained by these high temperature tests give reliable results with a noticeable lower standard deviation than using a standard furnace and are used as an input for finite element analysis of one-sided heated structures under mechanical load and as a reference for future creep and relaxation tests, where the quasi-static strength of the material at high temperature is of interest.

a) b)Figure 1 – a) Tensile, b) Bending configuration

Figure 2 – Tensile configuration implemented in universal testing machine: 1), 2) upper and lower grips with temperature cell in between; 3) heat gun; 4) extensometer; 5) 50kN load cell; not directly

visible: extraction tube

REFERENCES

[1] A. P. Mouritz, A. G. Gibson, Fire properties of polymer composite materials, Springer, 2006.[2] R. Chowdhury, R. Eedson, L. A. Bisby, M. F. Green, N. Benichou, Mechanical characterisation

of fibre reinforced polymer materials at high temperature, Fire Technology, 47, 2009, pp. 1063-1080 (doi: 10.1007/s10694-009-0116-6).

[3] S. Cao, Z. Wu, F. Li, Effects of temperature on tensile strength of carbon fiber and carbon/epoxy composite sheets, Advanced Materials Research, 2012, pp. 778-784 (doi:10.4028/www.scientific.net/AMR.476-478.778).

[4] P. Seggewiß, Methods to evaluate the fire resistance of carbon fiber reinforced plastics, Proceedings of the 3rd Composites in Fire Conference (ed. A. G. Gibson), Newcastle, UK,September 9-10, 2003, CompositeLink.

[5] DIN 65466, Aerospace - Fibre reinforced plastics - Testing of unidirectional laminates -Determination of shear strength and shear modulus in tension, 1996

8080

6th International Conference on Composites Testing and Model Identification O. T. Thomsen, Bent F. Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

STATIC STRAIN AND DEFORMATION CONTROLLED TESTING OF COMPOSITE BEAMS

Jacob Høgh1, Jacob Waldbjørn1, Henrik Stang1, Christian Berggreen1, Jacob Wittrup-Schmidt1, Kim Branner2

1Department of Civil Engineering, Technical University of Denmark

Brovej 118, DK-2800 Kgs. Lyngby, Denmark Email: [email protected], [email protected], [email protected], [email protected], [email protected], web page:

www.byg.dtu.dk

2Department of Wind Energy, Technical University of Denmark Frederiksborgvej 399, Building 118, DK-4000 Roskilde, Denmark

Email: [email protected], web page: www.vindenergi.dtu.dk

Keywords: test control, fibre Bragg grating, digital image correlation, glass fibre reinforced plastic

ABSTRACT Test control is commonly performed by a feedback signal from a gauge in an actuator. However, if a certain strain level is sought, it is more accurate to operate the test by measurements acquired directly on the specimen [1]. This is valid when testing a structure of complex geometry and/or under complex loading. The work presented in this paper, documents the use of fibre Bragg grating (FBG) [2] and digital image correlation (DIC) [3] for strain and displacement control, respectively. The test control was performed by software capable of: acquiring data from the FBG and DIC systems and operate the actuator accordingly. The technique was verified in a quasi-static three point bending test with a GFRP beam. The GFRP specimen constitutes 22 plies of E-glass fibre mats oriented in the longitudinally direction and Epoxy resin. The optical fibres are located between the 1st-2nd, and the 21st-22nd plies each containing three sensors. The DIC system includes three measurement points. The test rig with gauges is seen in Figure 1.

19

3,T

1,L

125150

FBG sensorPly 1-2

Center line

Ply 21-22

Strain gauge sensor

5002550 100

P/2P/2

P

Measurement point (DIC)r = 12.5mm

r = 20.0mm

Figure 1: Schematic illustration of the 3-point bending test, the depth is 45mm

The force P is applied by a deformation controlled hydraulic actuator operated by a feedback signal acquired by DIC and FBG measurements. This configuration is handled by LabVIEW 8.6 and is executed in the state-machine framework presented by Figure 2.

81

81

Jacob Høgh, Jacob Waldbjørn, Henrik Stang, Christian Berggreen and Jacob Wittrup-Schmidt

5:Deviation acceptable?

False

True4:DAC from ex-ternal measure-ments: DIC and

FBG

2:Apply actuator response

1:Define position

input

3:DAC from controller: LVDT

and load cell

6:Proceed to next load step

Figure 2: Operating and acquiring data from: controller, FBG interrogator and DIC system

As shown in Figure 2 the position input is defined in (1) after which a deformation is applied by (2). The deformation of the specimen is derived from the external measurement in (4) by an Euler-Bernoulli assumption and compared with the position input. If the deviation exceeds a certain tolerance the actuator is moved in the direction necessary to reduce the inaccuracy with a magnitude equal to the deviation. This process is repeated in a loop from (2) – (5) until a deviation below the acceptance tolerance is obtained. Five GFRP beams were loaded to 3kN in two separate tests A) deformation control by DIC with an acceptance tolerance of 0.01mm and B) strain control by FBG with an acceptance tolerance of 5με. The deviation between position input and external input is presented in Figure 3.

Figure 3: Deviation between position input (cmd) and external measurements: A) DIC and B) FBG

Figure 3 verifies the control loop with a reasonable amount of correlation loops. When the deviation is larger than the acceptance tolerance the program corrects the displacement/strain and proceeds to the next load step as described in Figure 2. The acceptance tolerance is set with respect to the precision and accuracy offered by the DIC and FBG system.

REFERENCES [1] U. C. Mueller, T. Zeh, A. W. Koch, H. Baier, FBG Sensors for High-Precision Structural

Deformation Control in Optical Systems, SPIE, 6167, 2006, pp. 1-12[2] E. Udd, Fiber Optic Smart Structures, IEEE, 84, 1996, pp. 60-67 [3] M. A. Sutton, J.-J. Orteu, and H. W. Schreier, Image Correlation for shape, motion and

deformation measurements, Springer, 2009

0 0.5 1 1.5 2 2.5 3-0.025

-0.02

-0.015

-0.01

-0.005

0

0.005

0.01

0.015

0.02

0.025A

Load [kN]

Dis

plac

emen

t [m

m]

cmd - DICdef. patternerr. tol.

0 0.5 1 1.5 2 2.5 3-1.5

-1

-0.5

0

0.5

1

1.5x 10

-5 B

Load [kN]

Stra

in [-

]

cmd - FBGdef. patternerr. tol.

82

82


Aalborg, 2013

THE USE OF DIGITAL IMAGE CORRELATION FOR FULL FIELD ANALYSIS OF POLYMERIC FOAMS

Richard M. Stubbing1 and Mark Battley1

1 Centre for Advanced Composite Materials Department of Mechanical Engineering, University of Auckland

Private Bag 92019, Auckland, New Zealand Email: [email protected], [email protected],

web page: http://www.cacm.auckland.ac.nz/uoa/

Keywords: Digital image correlation, polymeric foam, full field, strain periodicity

ABSTRACT Digital image correlation (DIC) is a useful and increasingly employed tool for full field deformation analysis. It enables displacement fields to be characterised through identification and tracking of features on the material surface, and strain fields to be calculated from the displacements. The relative performance of different DIC tracking algorithms is however dependant on the nature of the deformation field being observed and consideration of the materials being studied is required when selecting a tracking implementation method. DIC based analysis of polymeric foam core deformation presents unique challenges due to the significant local deformation that they can exhibit due to their cellular structure, visco-elastic materials and complex failure mechanics. Understanding the implications that the foam’s behaviour has for the tracking implementation is vital for ensuring that reliable full field information can be obtained in a computationally efficient manner. The general consensus regarding DIC tracking algorithms is that bicubic or even biquintic interpolation schemes should be utilised for subpixel intensity interpolation due to the increased tracking accuracy that they provide in comparison to bilinear interpolation [1,2]. A smoothing function such as a Gaussian filter is commonly applied to the images prior to tracking both to reduce digital noise and because it has been shown have a positive impact on the systematic errors that can result from subpixel intensity interpolation [2]. The use of an affine or quadratic transformation function to allow for deformation of the tracking subset is also often recommended, particularly for cases where significant deformation is expected [3]. The purpose of this study is to investigate whether these conclusions hold true for the analysis of polymeric foam cores. Consideration is given to the robustness and efficiency of different tracking approaches, as well the impact that variations in tracking accuracy have on the resultant strain fields. Full field displacement tracking is performed using images taken during compressive and block shear characterisation tests of low density PVC and SAN foam cores. For the cases studied the improved tracking accuracy achieved using higher order interpolation schemes has a negligible impact on the reported strain fields. Figure 1 shows the strain field (top) and resulting features successfully tracked using a rigid tracking subset (middle) vs. a quadratic subset transformation (bottom) for a through-thickness compression test of M80 foam. In this case there is significant local cell collapse both at the top face of the specimen and at the approximate centre of the specimen face. The use of a subset transformation function offers no noticeable improvement while at the same time causing convergence issues, particularly in regions where fracture or cell buckling give rise to discontinuities in the real strain fields. It can be seen in Figure 1 that some features have been discarded in the regions exhibiting cell collapse for the case where the tracking subset is considered to be rigid. Due to poor convergence behaviour far more features are discarded however when a quadratic shape function is utilised. Higher order intensity interpolation and subset transformation both increase the complexity of the tracking algorithm and come with an associated increase in computational expense.

8383

Richard M. Stubbing and Mark Battley

Figure 1. Features successfully tracked during compression (strain field shown above) using a rigid tracking subset (middle) vs. a quadratic

subset transformation (bottom).

Figure 2. Features successfully tracked during block shear with a slow fracture (shown above) using a quadratic subset transformation without (middle) and with Gaussian filtering (bottom).

Figure 2 shows results for the case of a block shear test of an M80 core. In this case the specimen undergoes a slow fracture at the top edge of the core adjacent to the steel loading plates. Gaussian filtering of the images is shown to have a positive impact on tracking robustness and is computationally inexpensive. The results illustrate that, although convergence issues are still apparent where fracture occurs when using a transformation function, Gaussian filtering of the images dramatically improves the algorithm’s convergence behaviour. When Gaussian filtering is used in conjunction with a rigid tracking subset (not shown) a significant loss of features does not occur at any location on the specimen. Observations are also made regarding the strain field periodicity that can arise due to systematic errors in the reported subpixel displacements, including a discussion of strategies for dealing with the issue.

REFERENCES [1] H. W. Schreier, J. R. Braasch and M. A. Sutton, Systematic errors in digital image correlation

caused by intensity interpolation, Optical Engineering, 39(11), 2000, pp. 2915–2921. [2] P. Lava, S. Cooreman, S. Coppieters, M. De Strycker and D. Debruyne, Assessment of

measuring errors in DIC using deformation fields generated by plastic FEA, Optics and Lasers in Engineering, 47, 2009, pp. 747–753.

[3] H. Lu and P. D. Cary, Deformation measurement by digital image correlation: implementation of a second-order displacement gradient, Experimental Mechanics, 40, 2000, pp. 393–400.

8484


Aalborg, 2013

Oral session 8 – Health/condition monitoring (3 presentations)

8585


Aalborg, 2013

SELF-SENSING OF DAMAGE IN CARBON NANOTUBE VINYL ESTER COMPOSITES

Jose de J. Ku-Herrera, Alejandro May-Pat, Francis Avilés

Unidad de Materiales, Centro de Investigación Científica de Yucatán A.C Calle 43 No. 130, Colonia Chuburná de Hidalgo, C.P. 97200 Mérida, Yucatán, México.

Email: [email protected] (Francis Aviles)

Keywords: Self-sensing, Structural health monitoring, Carbon nanotube composites.

ABSTRACT

The incorporation of carbon nanotubes (CNTs) within a polymer matrix can render electrical conductivity to the composite through the formation of an electrically percolated network [1]. This electrical network can be altered due to the application of mechanical loading, yielding changes in the electrical resistance of the composite which can be correlated to the applied load [2,3]. This electro-mechanical coupling allows to self-monitor the deformation and damage of the nanocomposite, and this property could be extended to self-monitor damage in multi-scale fibre reinforced polymer composites. Although this strategy has been investigated [4,5], the phenomena underlying this effect are not complete understood, and most of the investigations have been carried out for tension loadings only, while a complete characterization of the electro-mechanical response must include all loading scenarios. Given this motivation, this work investigates the capability a multiwall CNT/vinyl ester composite to self-monitor its reversible and irreversible deformation and damage, correlating the change in electrical resistance due to the application of quasi-static and incremental low cyclic compression loadings to its damage initiation and evolution. The composite material was manufactured using a conventional mechanical-ultrasonic method to disperse the multiwall carbon nanotubes (MWCNTs) within the vinyl ester resin. The process consist on dispersing 0.3 wt% MWCNTs within the vinyl ester resin first by mechanical stirring and then by an ultrasonic horn. Cobalt naphthenate and methyl ethyl ketone peroxide are used as promoter and initiator, respectively, before casting into silicon moulds for crosslinking. The compression specimens were 25.4 mm long blocks of 12.5 mm square cross section. Copper wires separated 8 mm were bonded as electrodes to measure the electrical resistance during the test, as depicted in Fig. 1a. Figure 1 shows the mechanical ( vs. , red curve) and electro-mechanical ( R/R0 vs. , blue curve) response of the composite when is subjected to a quasi-static compression up to failure.

a)

0 2 4 6 8 10 12 14 16 18 20 22 24

0

20

40

60

80

100

120

140

160d

dc'

c c'

c

R/R0

(%)

(MPa

)

a

b

a bo 0

20

40

60

80

100

120

140

160

R/R

0 (%

)

b)

Fig. 1. Electro-mechanical compression testing. a) Specimen instrumentation, b) mechanical ( vs. ) and electro-mechanical ( R/R0 vs. ) response of a MWCNT/vinyl ester composite under quasi-static

compression loading.

8787

Jose de J. Ku-Herrera, Alejandro May-Pat, Francis Avilés

The stress-strain curve in Fig. 1b shows an initial elastic region followed by yielding and then large plastic deformation as the applied load increases. The normalized change in electrical resistance ( R/R0) initially decrease linearly while the specimen undergoes elastic deformation (region o-a in Fig. 1b), then shows a plateau region that coincide with the yielding region (region a-b), and finally increases as the plastic deformation increases (region b-d). For the specimen show, before failure at the point “d”, small matrix cracking in the region c-c’ is only weakly perceived by the vs. curve, but it is clearly identified by the R/R0 vs. curve as a sudden change in electrical resistance ( R/R0 increases from ~60 to ~120 %), as shown in Fig. 1b. Results of incremental compressive cyclic loadings are shown in Fig. 2. In this experiment an incremental deformation was applied to the specimen at each cycle, including deformations in the elastic, yielding and plastic regions (see Fig. 1).

0 5 10 15 20 25

0

50

100

150

200

(%)

max (%) 0.5 1.0 1.5 2.0 4.0 7.0 10 15 20

R/R 0 (%

)

Fig. 2. Electrical resistance changes as a function of incremental cycle applied strain For the cycles where the applied deformation does not reach the elastic limit ( ≤ 2%), the composite does not undergo permanent changes of electrical resistance, i.e. R/R0 returns to zero upon unloading. As the applied deformation reachs yielding ( > 2%), a permanent chance in electrical resistance is detected ( R/R0 ≠0 after unloading), and its permanent magnitude (corresponding to =0) increases after each cycle according to the accumulated plastic deformation. Therefore, it is concluded that the changes in electrical resistance of a MWCNT/thermosetting composite upon loading can identify its elastic, yielding and plastic zones. Upon cycling loading, the permanent increase in electrical resistance can be correlated to irreversible accumulation of damage in the composite. These concepts can be extended to monitor the structural health of advanced fibre reinforced polymer composites, especially if they are coupled with full field techniques such as digital image correlation. [1] B. Fiedler, F.H. Gojny, M.H.G. Wichmann, W. Bauhofer, K. Schulte, Can Carbon Nanotubes Be

Used to Sense Damage in Composites? Annales de Chimie Science des Matériaux, 29, 2004 pp. 81-94.

[2] S.M. Vemuru, R. Whai, S. Nagarajaiah P.M. Ajayan, Strain sensing using a multiwalled carbon nanotube film. The Journal of Strain Analysis for Engineering Design, 44, 2009, pp. 555-562 (doi: 10.1243/03093247JSA535).

[3] Alamusi, N. Hu, H. Fukunaga, S. Atobe, Y. Liu, J. Li, Piezoresistive Strain Sensors Made from Carbon Nanotubes Based Polymer Nanocomposites, Sensors, 11, 2011, 2011, pp. 10691-1723 (doi:10.3390/s111110691).

[4] E.T. Thostenson, T.W. Chou, Carbon Nanotube Networks: Sensing of Distributed Strain and Damage for Life Prediction Self Healing. Advanced Materials, 18, 2006, pp.2837-2841 (doi: 10.1002/adma.200600977).

[5] M.H.G. Wichmann, S.T. Buschhorn, J. Gehrmann, K. Schulte, Piezoresistive response of epoxy composites with carbon nanoparticles under tensile load, Physical Review B, 80, 2009, pp. 2454371-2454378 (doi: 10.1103/PhysRevB.80.245437).

8888


Aalborg, 2013

INFLUENCE OF THE PROTECTIVE COATING OF FIBER BRAGG GRATING SENSORS ON THE STRUCTURAL DISTORTION AND SENSING ACCURACY WHEN EMBEDDED

IN FIBER REINFORCED POLYMERS

Nicolas Lammens1, Gabriele Chiesura1, Geert Luyckx1, Eli Voet2 and Joris Degrieck1

1Department of Materials Science & Engineering, Ghent University Technologiepark-Zwijnaarde 903, BE-9052 Zwijnaarde, Belgium

2FBGS International, Bell Telephonelaan 2H, BE-2440 Geel, Belgium Email: [email protected], web page: http://www.composites.ugent.be

Keywords: Finite elements, Smart materials, Optical fibre sensors, non-destructive testing

ABSTRACT

Optical fibre Bragg grating (FBG) sensors are ideal for structural health monitoring purposes in fibre reinforced polymers (FRPs). Owing to their small size and fibrous nature they can fairly easily be embedded during production, providing important information on both production process as well as in-service behaviour. The multi-axial strain sensitivity of such FBGs provides valuable information on the structural health, which would go undetected using surface-mounted sensors. Finally, their chemical inertness and wavelength encoded data makes them perfectly suited for long-term applications in harsh chemical or electrical environments where traditional foil strain gauges would quickly fail. While much research is available on the general topic of embedded strain sensing using fibre Bragg grating sensors, only a small amount of research has focussed on the influence of coating parameters on both host and sensor. Most often, the default coating provided by the manufacturer is used, which is usually selected to provide adequate protection during handling in telecommunication applications. In general, a more optimal set of coating properties will exist for embedded sensing purposes. This work will present research on the influence of coating parameters on both (1) the presence of stress concentrations in the surrounding host material, as well as on (2) sensing accuracy of the FBG sensor. The most common coating materials have been investigated. The material properties for these materials are stated in Table 1. Although it would be possible to tune the coating material properties to provide optimum performance, this would inevitably increase the manufacturing costs and therefor reduce the industrial value of the results.

Acrylate Polyimide Ormocer E-modulus [GPa] 0,75 – 1,35 1,50 0,80

Poisson ratio [-] 0,36 – 0,47 0,45 0,32

Table 1: Investigated coating materials

With the coating material properties fixed, only the coating thickness and host material are left as degrees-of-freedom. Since homogenized host properties are assumed in the analysis, the absolute values of optical fibre diameter and coating thickness are not important, and only the relative values are of influence. Therefor, this study focuses on the ratio of coating outer diameter (b) to inner diameter (a). The analysis has been carried out on a carbon fibre reinforced polymer (CFRP) and a glass fibre reinforced polymer (GFRP) host. When optimizing the coating for minimal impact on host strength, the goal is to find a b/a ratio that minimizes the presence of additional stresses in the composite host under loading. However, this optimal b/a ratio is dependent on the type of loading exerted on the structure. Therefore both axial

8989

Nicolas Lammens, Gabriele Chiesura, Geert Luyckx, Eli Voet and Joris Degrieck

loading (along the direction of the reinforcements) and transverse (perpendicular to the reinforcements) are investigated. The results of this analysis on CFRP are shown in Table 2.

Acrylate Polyimide Ormocer Axial loading 2,40 – 1,40 1,50 1,00

Transverse loading 1,66 – 1,67 1,89 1,41

Table 2: Optimal b/a ratios in CFRP The results from Table 2 clearly show large variations in optimal coating thickness depending on coating material and loading circumstances. A limit b/a ratio of 1 is predicted for Ormocer coatings under axial loading. In this specific combination of host and coating material under axial loading, decreasing the coating thickness continuously decreases the host stresses although they are never fully reduced to zero. For reference it should be noted that standard telecom fibre usually has a fibre diameter of 125 micron and a coating diameter of 250 micron, leading to a b/a ratio of 2. Besides influencing the host stresses, the coating also influences the FBG sensor accuracy. Figure 1 shows the theoretical sensor error when an Ormocer coated FBG is embedded (in both UD and cross-ply CFRP) and exposed to a combination of axial and transverse loading, while originally having been calibrated for pure axial loading for different b/a ratios.

Figure 1: Sensor error of an Ormocer coated embedded FBG in (left) UD CFRP and (right) [02/902]2s

cross-ply for different b/a ratios

Figure 1 shows that increasing the b/a ratio is beneficial to sensor accuracy, since the coating acts as a buffer to shield the FBG from transverse strains. It is also shown that the stacking plays an important role. The predicted errors in the cross-ply laminate are much more severe than in the UD laminate. Furthermore the improvement in sensor accuracy with increasing coating thickness is much more pronounced for low b/a ratios than higher ratios. While the default b/a ratio in telecom fibre of 2 provides good sensor accuracy, the sensor error of a b/a ratio equal to 1,60 is very comparable while having a ratio more closely to the optimal value of 1,41 stated in Table 2 for Ormocer coated fibres under transverse loading. These results show that an optimal coating is strongly dependent on a number of factors such as host material and ply lay-up, expected loading conditions, desired sensor accuracy… In general, a trade-off will have to be made between the reduction of host stresses and acceptable sensor errors. The authors gratefully acknowledge the funding by the European Union within the FP7 – SmartFiber project.

9090


Aalborg, 2013

MONITORING STRAIN GRADIENTS IN ADHESIVE COMPOSITE JOINTS BY EMBEDDED FIBRE BRAGG GRATING SENSORS

Luis P. Canal1, Behzad D. Manshadi2, Véronique Michaud1, John Botsis2, Georgios Violakis3 and

Hans G. Limberger3 1Laboratory for Polymer and Composite Technology (LTC), Email: [email protected]

2Laboratory of Applied Mechanics and Reliability Analysis, Department of Mechanical Engineering, 3School of Engineering,

Ecole Polytechnique Fédérale de Lausanne (EPFL), CH-1015 Lausanne, Switzerland.

Keywords: Adhesive bonding of composites, single lap shear test, fibre Bragg gratings, finite elements simulations

ABSTRACT

The adhesive bonding of composite materials is a very adequate technique to assemble light-weight structures for aeronautical applications. Compared with mechanical fasteners, adhesive joints provide more homogeneous stress distributions and noticeable weight reductions [1]. However, the application of adhesive bonds in aeronautic primary load-bearing structures is limited by the lack of adequate testing techniques to monitor the quality of the bond during service [2]. This work evaluates the feasibility of using fibre Bragg grating (FBG) sensors to measure strains developed in adhesive bonds during Single Lap Shear (SLS). SLS is a widely used testing method to characterize and control the quality of adhesive joints [3]. Due to its simplicity, this bonding configuration is also present in many structural applications in industries including automotive, aerospace or sports [4]. The actual distribution of stresses along the adhesive is non-uniform and it depends on the geometry and the mechanical properties of the bond. In addition, the eccentric loading path causes out-of-plane bending moments and peel stresses, reducing significantly the strength of the joint. Thus, the characterization of the deformations is a key requirement to ensure the reliability of the lap joint.

Figure 1: Schematic of the single lap shear specimen. Single Lap Shear specimens were manufactured according to the dimensions depicted in Fig. 1. Unidirectional CFRP panels were manufactured first by stacking 4 plies of NCT 301/HR40 prepreg and consolidated in the autoclave under 7 bars pressure. The different components for the SLS samples were prepared by cutting the composite panel in the adequate dimensions. Finally, the specimens were assembled using Araldite 420A/B and cured in the oven during 1 hour at 160 °C. The strains developed in the lap region were monitored by the integration of an optical fibre with an array of FBGs. The FBG arrays were fabricated in photosensitive fibre using Ar+-244 nm laser and phase mask technique. Each array contained 7 very short gratings (0.4 mm) with a wavelength spacing of about 4 nm, distributed over a total length of 15 mm. Low coherence reflectometry was used to precisely determine the location of the FBGs [5]. Three different positions for the optical fibres were studied (see Fig. 2). The SLS specimens were loaded in uniaxial tension using an electromechanical testing machine at a constant cross-head speed of 0.2 mm/min. The applied load and displacement were continuously recorded by a 100 KN load cell and the cross-head position, respectively. Additionally, the local strain detected by each individual FBG was obtained from the shift in the reflected wavelength. The local strains obtained from the FBGs were compared with numerical simulations. To this end, a three dimensional finite element model of the SLS specimen was developed for each position of the optical fibre in the joint. The main elastic properties of the composite and the glue were obtained from four-point bending and tensile tests, respectively.

9191

Luis P. Canal, Behzad D. Manshadi, Véronique Michaud, John Botsis, Georgios Violakis, and Hans G. Limberger

Figure 2: Micrographs of the specimen’s cross-section showing the studied positions for the optical

sensor: (a) in the 1st ply of the composite (b) in the glue (c) at composite-glue interface.

The strains measured with the FBGs are plotted in Fig. 3 together with the numerical simulation results for an applied load of 4 kN. For the optical fibre located in the 1st composite ply (Fig. 3 a), the longitudinal strains along the fibre showed a smooth evolution along the overlap region reaching a maximum of ≈1800 με at the end of the joint. The numerical model shows strains similar to those experimentally measured. Their maximum difference is always below ≈150 με. The strains at the composite-glue interface (Fig. 3 b) are almost constant (≈500 με) over about 50% of the overlap region (6 mm), according to the simulations. Strains increase in the second half of the region significantly up to a maximum of ≈4500 με. The FBG array placed in the second part of the overlap zone validates the simulations. For the fibre located in the glue, the strain predicted by the simulations shows a symmetric distribution with a plateau in the central part of the overlap region. The experimental measurements show qualitatively the same trend (Fig. 3 c). The discrepancies in the values of the strains were possibly due to inaccuracies in the position of the optical fibre, the background noise generated by viscoelastic behaviour of the glue and the shear efforts in the optical fibre.

0 2 4 6 8 10 120

500

1000

1500

2000 Numerical simulations FBG measurements

S

train

()

Overlap position (mm)

0 2 4 6 8 10 120

1000

2000

3000

4000

5000 Numerical simulations FBG measurements


Stra

in( 3

)

0 2 4 6 8 10 120

500

1000

1500

Stra

in( 1

)


Numerical simulations FBG measurements

Fig. 3. Numerical predictions and experimental measurements of longitudinal strains along the optical fibre: (a) in the 1st ply of the composite (b) at interface composite -glue (c) in the glue.

ACKNOWLEDGEMENTS Funding from the European Union’s Seventh Framework Programme (FP7/2007-2013) under grant agreement nº ACP0-GA-2010-266226 (ENCOMB, Extended Non-Destructive Testing of Composite Bonds) and from SNSF project 200020-138012 (G. Violakis) is acknowledged. Thanks area also due to Dr. Marco Lai and Dr. Claire Bonnafous for useful discussions.

REFERENCES [1] G. Savage, Failure prevention in bonded joints on primary load bearing structures, Engineering Failure

Analysis, 14, 2007, pp. 321–348. [2] 1st Periodic Report Publishable. www.encomb.eu [3] A. J. Comer, K. B. Katman, W. F. Stanley, T. M. Young, Characterising the behaviour of composite

single lap bonded joints using digital image correlation, International Journal of Adhesion and Adhesives, 40, 2013, pp. 215-223.

[4] M. Y. Tsai, J. Morton, An experimental investigation of nonlinear deformations in single-lap joints, Mechanics of Materials, 20, 1995, pp. 183-194.

[5] P. Lambelet, P. Y. Fonjallaz, H. G. Limberger, R. P. Salathé, C. Zimmer, and H. H. Gilgen, “Bragg Grating Characterization by Optical Low-Coherence Reflectometry,” IEEE Photonics Technology Letters 5, 1993, pp. 565-567.

9292


Aalborg, 2013

Oral session 9 – Fracture and fatigue – 3 (5 presentations)

9393


Aalborg, 2013

ON THE USE OF IN-SITU SEM TESTING AND SIMULATION TO STUDY DEFORMATION AND FAILURE MECHANISMS IN COMPOSITE MATERIALS

C. González1,2, L.P. Canal1, J. Segurado1,2, J. LLorca1,2

1Department of Materials Science, Polytechnic University of Madrid and CISDEM, UPM-CSIC, E.T.S. de Ingenieros de Caminos, 28040 Madrid, Spain


IMDEA Materials Institute, C/Eric Kandel, 2 Parque Científico y Tecnológico – Tecnogetafe 28906 Getafe (Madrid)


Keywords: Computational Micromechanics, Fracture mechanisms, In-situ testing, SEM Scanning Electron Microscopy

ABSTRACT

The fracture behavior parallel to the fibers of an E-glass/epoxy unidirectional laminate was studied by means of three-point tests on notched beams. Selected tests were carried out within a scanning electron microscope to ascertain the damage and fracture micromechanisms upon loading. The mechanical behavior of the notched beam was simulated within the framework of the embedded cell model, in which the actual composite microstructure was resolved in front of the notch tip. In addition, matrix and interface properties were independently measured by in situ nanoindentation. The numerical simulations reproduced the macroscopic response of the composite as well as the damage development and crack growth in front of the notch tip, demonstrating the ability of the embedded cell approach to simulate the fracture behavior of heterogeneous materials [1]. Beams with a rectangular cross-section of 2.8mm (depth, D) and 2mm (thickness, t) were machined from the panel with the fibers perpendicular to the beam axis, Figure 1a). A notch was introduced in the central section of the beam with a thin diamond wire. The initial notch depth (a0) was 0.5D and the notch tip radius was around 130 m. The fracture behavior of the composite panel was determined by means of three-point bend tests on notched beams with 11.2mm of loading span S. The three-point bend tests of the notched beams were carried out inside a scanning electron microscope (Zeiss EVO MA-15) to ascertain the dominant deformation and damage mechanisms during fracture. To this end, the notched beams were first sputtered with Au–Pd, and the bending test fixture was set up in a micro-electromechanical testing machine (Kammarth & Weiss Tensile/Compression Stage) which can be fitted in the scanning electron microscope. The actuator of the testing machine was stopped at regular intervals of displacement of 5 m and micrographs of the microstructure in front of the notch tip and along the crack path were obtained at different magnifications levels (50×, 85×, 250× and 500×).

Figure 1: a) Notched three point bending beam geometry, b) and c) In-situ SEM micrographs of the notch root and crack tip indicating the different deformation and failure mechanisms

9595

C.González, L.P. Canal, J.Segurado, J.LLorca

Crack propagation from the notch tip was discontinuous and it was formed by a number of decohered interfaces connected by matrix ligaments, Figure 1b) and c). As the applied displacement increased, matrix ligaments were broken and a continuous crack grew from the notch. The crack path was not straight at the microscopic level but presented some tortuosity as a result of the random fiber distribution in the composite. Observation of the microscopic damage process at higher magnification showed that interface decohesion started at the equator of the fibers and propagated towards the poles. The propagation of the interface crack was accompanied by the growth of an interfacial void and by the shear failure of the matrix between debonded fibers, indicating that the microscopic fracture processed involved significant non-linear deformation of the epoxy matrix. The virtual test to compute the toughness of the composite lamina parallel to fibers was carried out by simulating the three-point bend test of the notched beam by means of an embedded cell model, Figure 2a). The whole notched beam is included in the numerical model but it is divided in two regions. The actual microstructure was represented in front of the notch tip, where all the fracture processes occur, Figure 2a), while the remainder of the beam was assumed to be a homogenous solid. The displacement field was continuous across the interface between both regions. E-glass fibers were assumed to behave as linear elastic, isotropic solids. The epoxy matrix was modeled with the continuum plasticity-damage model developed by Lubliner et al. [2], which takes into account the pressure-sensitivity of the epoxy flow stress under compression and its brittle behavior in tension. The fiber–matrix interface was modeled as a cohesive crack, whose mechanical behavior was expressed in terms of a bilinear traction–separation law which relates the displacement jump across the interface. Figure 2b) shows the snap shot corresponding to the propagation of the crack through the microstucture of the ligament. The deformation and failure mechanisms as well as the macroscopic response of the beam were accurately obtained by the embedded cell model [3].

Figure 2: a) Embedded cell model used for the simulation of the three point bending fracture tests, b)

Snap shot of the crack tip showing the failure modes of the composite material.

REFERENCES [1] L.P. Canal, C. González, J. Segurado, J. Llorca, Intraply fracture of fiber-reinforced composites: Microscopic mechanisms and modelling, Composites Science and Technology, 72, 2012, pp. 1223-1232 (doi: 10.1016/j.compscitech.2012.04.008) [2] A. Matzenmiller, J. Lubliner, R.L. Taylor. A constitutive model for anisotropic damage in fiber composites, Mech Mater, 20 (1995), pp. 125–152 [3] J.Llorca, C. González, J.M. Molina, J. Segurado, R. Setzer, M. Rodriguez, S. Sádaba, R. Muñoz, L.P. Canal, Multiscale Modeling of Composite Materials: a Roadmap Towards Virtual Testing, Advanced Materials, 23, 2011, pp. 5130-5147 (doi: 10.1002/adma.201101683)

9696


Aalborg, 2013

BIO-BASED COMPOSITES WITH DIFFERENT MOISTURE CONTENTSUNDER STATIC AND DYNAMIC LOADING

N. Doroudgarian1, M. Anglada2, A. Mestra2 and R. Joffe1

1 Division of Materials Science, Department of Engineering Sciences and MathematicsLuleå University of Technology, SE-97187 Luleå, SwedenEmail: [email protected], web page: http://www.ltu.se

2 Department of Materials Science and Engineering, Polytechnic University of CataloniaDiagonal 647, SP-08028 Barcelona, Spain

Email: [email protected], web page: http://www.upc.edu

Keywords: Bio-based resins, Cellulosic fibers, Natural composite, Water absorption, Mechanical and environmental durability, Fatigue

ABSTRACT

Natural cellulosic fibers, bio-resins and their composites exhibit fairly high mechanical properties in static loading at ambient temperature and humidity [1, 2]. However, moisture is the Achilles heel of natural composites - it can drastically deteriorate their mechanical properties, e.g. strength and stiffness. Therefore, behavior of bio-based composites at different humidity levels should be studied and mechanisms occurring in material are to be identified and clearly understood. This study is an initial step in development of high performance bio-based composites with improved mechanical and environmental durability (i.e. exposure to fatigue loading and elevated humidity).

Purely bio-based composites with different combinations of bio-based resins and natural cellulosic fibers as well as man-made regenerated cellulose fibers (RCF) were studied. The focus is on composites with sufficiently high (over 40%) volume fraction of fibers which are well oriented. Even though RCFs have lower stiffness than natural fibers like hemp or flax, they possess certain advantages by being continuous with controlled geometry and properties. This allows creatingcomposites with compact and well-defined microstructure which is alike synthetic composites (see Fig. 1). Moreover, because of this resemblance, there are also similarities in the failure initiation and damage accumulation (e.g. fiber/matrix debonding, transverse cracking etc.).

Figure 1: Transversal cross-section of the bio-based unidirectional composite laminate with distinct bundle structure (left) and RCF bundles with high packing density of fibers (right).

However, stress-strain curves obtained from these composite laminates show that material is highly non-linear and has a significant accumulation of residual strains (see Fig. 2). Non-linear shape of

9797

Newsha Doroudgarian, Marc Anglada, Alvaro Mestra and Roberts Joffe

stress-strain curve (Fig. 2a) might be an indication of visco-elastic effects whereas large irreversible residual strains observed after unloading (Fig. 2b) could arise due to damage accumulation or/and plasticity due to applied load. Such behavior complicates analysis of material performance, especially with regards to failure and fatigue where miscellaneous mechanisms of energy dissipation are developing.

Figure 2: Simple tension (a) and loading unloading of the RCF/epoxy composite (b).

Main objective of the current paper is to identify and classify damage mechanisms which occur in natural fiber composites under simple tension as well as tension-tension fatigue. Behavior of materials is studied with respect to parameters of environment (namely, variation of humidity) and properties of reinforcement (natural and RCF with and without surface treatment). This information is essential indevelopment of models which can be used to predict durability, life time limits and degradation of mechanical properties of bio-based materials under service conditions.

Experimental results have been compared with data for glass fiber/epoxy composite at ambient humidity as a reference. The comparison of tensile properties showed that some of the bio-based composites performed on the par with reference material.

REFERENCES

[1] R. Joffe, B. Nyström, L. Rozite and J. Varna, Mechanical performance of polymer composites reinforced with nonlinear cellulosic fibers, Conference of Annual Bioenvironmental Polymer Society, BEPS-2011, Vienna, Austria, September 27-30 2011.

[2] J. Andersons and R. Joffe, Estimation of the tensile strength of an oriented flax fiber-reinforced polymer composite, Composites Part A, 42 (9), 2011, pp. 1229–1235.

9898


Aalborg, 2013

FAILURE MODE SPECIFIC FATIGUE TESTING OF NANOPARTICLE-MODIFIED CFRP UNDER VHCF-LOADING

Julia B. Knoll1, Roman Koschichow2, Ilja Koch2, Karl Schulte1 and Maik Gude2

1 Institute of Polymers and Composites, Technische Universität Hamburg-Harburg, Denickestraße 15, D-21073 Hamburg, Germany

Email: [email protected], web page: http://cgi.tu-harburg.de/~kvwww

2 Institut für Leichtbau und Kunststofftechnik, Technische Universität Dresden, Holbeinstr. 3, D-01307 Dresden, Germany

Keywords: Carbon fibre reinforced polymers, Very high cycle fatigue, Testing, Nanoparticles

ABSTRACT

Continuous fibre reinforced polymers (FRP) are often used for structural applications where they are subjected to cyclic loading. Inverstigations of fatigue life and damage mechanisms were so far limited to the low and high cycle fatigue (LCF/HCF) regime. The development of reliable fatigue life evaluation methods for FRP at extremely high load cycles (N > 108) requires a deep understanding of the successive damage behaviour and the occurring damage mechanisms. For that purpose a new shaker based fatigue test rig as well as adapted test specimens were developed. Common uniaxial fatigue tests for FRP are limited in test frequency due to internal specimen heating. In the very high cycle fatigue (VHCF) test rig (Fig. 1) the specimen is loaded in flexure which minimises internal shear stresses and with that the polymer specific heating during high frequency fatigue loading. This way, loading frequencies of up to 150 Hz can be performed which is of importance to decrease the extremely long testing time of VHCF tests. Due to the stress distribution under flexural loading it is possible to analyse tensile and compressive fatigue behaviour of the composite simultaneously. Depending on the specimen lay-up different failure modes can be tested separately. In this work carbon fibre reinforced epoxy resins are tested parallel and transvers to the fibre direction.

Figure 1: VHCF bending test rig for FRPs

A further focus of research in the field of fatigue of FRP is the modification of the matrix in order to increase the fatigue life. Especially for the matrix dominated failure modes a modification of the matrix material may lead to a better fatigue performance. When loading FRP first damage occurs in the matrix in form of cracks. The crack growth plays an important role and affects the fatigue life of the whole composite. In recent investigations the influence of nanoparticles on the fatigue behaviour of FRP up to the HCF regime was studied and showed a significant improvement in fatigue life [1, 2, 3, 4, 5]. Even the addition of small amounts of carbon nanotubes (CNTs) into the epoxy resin matrix improved the LCF and HCF life of fibre reinforced composites by a factor of six to ten. The transfer of

9999

Julia B. Knoll, Roman Koschichow, Ilja Koch, Karl Schulte and Maik Gude

these findings to VHCF-loading of nanoparticle-modified carbon fibre reinforced composites are validated with the developed testing technique.

REFERENCES [1] L. Böger, J. Sumfleth, H. Hedemann and K. Schulte, Improvement of fatigue life by incor-

poration of nanoparticles in glass fibre reinforced epoxy, Composites: Part A, 41, 2010, pp. 1419-1424 (doi: http://dx.doi.org/10.1016/j.compositesa.2010.06.002).

[2] C.S. Grimmer and C.K.H. Dharan, Enhancement of delamination fatigue resistance in carbon nanotube reinforced glass fiber/polymer composites, Composites Science and Technology, 70, 2010, pp. 901-908 (doi: http://dx.doi.org/10.1016/j.compscitech.2010.02.001).

[3] C.M. Manjunatha, A.C. Taylor, A.J. Kinloch and S. Sprenger, The tensile fatigue behaviour of a silica nanoparticle-modified glass fibre reinforced epoxy composite, Composites Science and Technology, 70, 2010, pp. 193-199 (doi: http://dx.doi.org/10.1016/j.compscitech.2009.10.012).

[4] D.C. Davis, J.W. Wilkerson, J. Zhu and V.G. Hadjiev, A strategy for improving mechani-cal properties of a fiber reinforced epoxy composite using functionalized carbon nanotubes, Composites Science and Technology, 71, 2011, pp. 1089-1097 (doi: http://dx.doi.org/10.1016/j.compscitech.2011.03.014,).

[5] N.d. Greef, L. Gorbatikh, A. Godara, L. Mezzo, S.V. Lomov and I. Verpoest, The effect of carbon nanotubes on the damage development in carbon fiber/epoxy composites, Carbon, 49, 2011, pp. 4650-4664 (doi: http://dx.doi.org/10.1016/j.carbon.2011.06.047,).

100100


Aalborg, 2013

EXPERIMENTAL AND NUMERICAL ANALYSIS OF SKIN/STIFFENER DEBONDING UNDER BENDING

I. Urresti1, A. Barrio1, J. Renart2 and L. Zubillaga1

1Department of Mechanical Engineering, Ikerlan S. Coop

M. Arizmendiarrieta 2, 20500 Arrasate, Spain Email: [email protected], web page: http://www.ikerlan.es

2AMADE, Polytechnic School, University of Girona

Campus Montilivi s/n, E-17071 Girona, Spain Email: [email protected], web page: http://www.udg.edu/

Keywords: Delamination, Damage Mechanics, Bending Test, Crack Growing

ABSTRACT

In this paper, a numerical and experimental study of the debonding process of a skin/stiffener has been performed. The analysis consisted of a validation of the numerical tools to simulate the debonding process, and their implementation in a subcomponent structure in order to predict the debonding process under a multiaxial loading state: a 7 point bending test in a reinforced plate (Figure 1). The results of these simulations were compared to experimental tests performed in a real component. The preliminary validation of the numerical tools was done with the Finite Element Method (FEM) software ANSYS® 13.0. The study consisted of a comparison of the simulations with the results of analytical equations obtained from the Linear Elastic Fracture Mechanics. The classical fracture tests DCB, ENF and MMB were modeled in 2D and 3D. Once the tool had been validated, a 3 point bending test of a simplified skin stiffener specimen was performed for further validation of the numerical tools.

Figure 1 Multi-level analysis of skin/stiffener debonding. For the numerical analysis, the formulation implemented in ANSYS® 13.0 [2] includes the cohesive elements adapted from [3]. As it was previously shown [4], accurate results can be obtained even with the power law formulation for the mixed mode debonding. More precisely, contact elements with Cohesive Zone Model properties [5] have been used for representing the adhesive film, and Solid Shell elements [2] for the composite material plates in the 3D models.

For the experimental campaign the test specimens were made of AS4/8552. The subcomponents (panel and stiffener) were cured separately. The bonding of the two parts was done with FM300 adhesive film which was cured in a secondary cycle. The specimen geometry and test configuration are shown in Figure 2. Two types of specimens were analyzed: a simplified skin stiffener tested in a 3 point bending, and a reinforced panel subjected to a 7 point bending test.

101101

I. Urresti1, A. Barrio1, J. Renart2 and L. Zubillaga1

Figure 2: 3 point (a) and 7 point (b) bending test specimen geometry (c) Support distribution for the 7

point bending test. From the test results of the 7 point bending test it has been observed that the debonding process occurred suddenly having a discontinuity in the force-displacement curve, and thus crack propagated abruptly. The displacement was recorded using both an inductive sensor and a laser displacement measuring device, and the force with a load cell located at each of the 7 points. Strain gauges were also glued on the specimens in order to compare local deformations with the numerical predictions. According to the “no-crack growth” principle used in aeronautical composite structures, the criterion to compare numerical analysis and experimental testing is the load corresponding to the onset of delamination. Thus, for the 7 point bending tests, the reactions in each support were compared with those predicted by the FEM. In order to check the adhesive film status, as well as, to be able of seeing the crack propagation, a C-Scan inspection was performed after the test.

Figure 3: (a) C-Scan inspection before and after 7 points bending test. (b) Reaction forces measured in the 7 support points.

REFERENCES

[1] J. Bertolini, Contribution à l’analyse expérimentale et théorique des ruptures de structures compos-ites en postflambement par décollement de raidisseurs. Thesis Université Paul Sabatier, Toulouse, April 2008 (in french) http://thesesups.ups-tlse.fr/ [2] ANSYS User Manual. Release 13.0. [3] G. Alfano, M.A. Crisfield, Finite Element Interface Models for the Delamination Anaylsis of Laminated Composites: Mechanical and Computational Issues, International Journal for Numerical Methods in Engineering, 50, 2001, pp. 1701-1736. [4] C. Sarrado, A. Turon, J.Renart, I. Urresti, Assessment of energy dissipation during mixed-mode delamination growth using cohesive zone models, Composites Part A-Applied Science and Manufacturing, 2012, pp. 2128-2136. [5] A. Turon, P.P. Camanho, J. Costa, J. Renart, Accurate simulation of delamination growth under mixed-mode loading using cohesive elements: Definition of interlaminar strengths and elastic stiffness, Composite Structures, 92, 2010, pp. 1857-1864.

0.00

0.50

1.00

1.50

2.00

2.50

3.00

0.00 0.50 1.00 1.50 2.00 2.50 3.00 3.50 4.00 4.50

Reat

ctio

n Fo

rces

(kN

)

Displacement between tooling plates (mm)

102102


Aalborg, 2013

QUANTIFICATION OF DAMAGE DUE TO ENVIRONMENTAL CONDITIONS ON CARBON FIBRE/EPOXY COMPOSITE SAMPLES

Enrique Guzman1, Joël Cugnoni1 and Thomas Gmür1

1Faculté des Sciences et Techniques de l’Ingénieur, Ecole Polytechnique Fédérale de Lausanne Station 9, 1015 Lausanne, Switzerland

Email: [email protected], web page: http://lmaf.epfl.ch

Keywords: Composite plates, Modal analysis, Identification, Properties, Finite elements

ABSTRACT

Polymer-based composite materials are massively being introduced as a primary material to the next generation of civil aircraft, due to their outstanding mechanical properties, especially their stiffness-to-mass ratio. Among these ultra-light materials, the carbon fibre/epoxy (CFRP) composite is one of the most frequently used in aeronautical engineering. Unfortunately, the behaviour of polymers when exposed to most of weathering conditions is still complex and ill-understood, mainly because of the number of simultaneous physical and chemical changes that take place. From a safety and reliability point-of-view, the use of CFRP could be compromised, due to the vulnerablity to weathering they exhibit, compared to more classic metallic alloys.

In order to study this question, the ageing can be approached using a simplified mathematical model to accept or refuse, by statistical means, if an environmental agent (or a combination of agents) does really degrade the inner structure of a composite sample. The performance of a mechanically critical airframe component is evaluated from its structural stiffness. From the manufacturing point-of-view, this depends on the resin elastic properties. Thus, the results considered here are the global elastic properties of a CFRP structure evaluated over several weeks of accelerated artificial ageing.

To establish the model, a non-destructive dynamic testing technique, the experimental modal analysis (EMA), combined to a numerical-experimental identification based on a finite element (FE) model, is used to characterize the material, from the natural frequencies of the part. The stiffness degradation of the samples subjected to the aforementioned ageing protocols is analytically described by a linear model with interactions. The model, obtained by a least-squares method, is analysed by statistical means, in order to evaluate the significance of the eventual variations versus the randomness of measurement errors. Then, only a reduced number of results are considered from the whole experimental campaign previously analysed. These results are chosen according to factorial design principles, based on three varying factors: the temperature (T), the relative humidity (RH) and the ultraviolet radiation intensity (UV) (figure 1(a)).

The ageing trend curves are shown in figure 1(b). After visual inspection of their evolution, the constitutive properties exhibit increasing (or decreasing) pattern over the time. This observation is supported by authors and experimenters that reported about accelerated ageing protocols. Indeed, this type of mathematical model is inspired from previous works on mechanical fatigue. This notion is then extended to the cyclic exposure to environmental conditions (see figure 1(a)), weathering being considered in this case like a sort of fatigue. After experimentation and modelling, it has been demonstrated that mass changes are essentially due to moisture absorption (up to 4% in 6 weeks), while T, RH and UV play a role in the loss of stiffness within the epoxy matrix (up to -13% in 6 weeks), the damage being more amplified for some specific combination of factors. Even if other factors related to the manufacturing and the ageing protocol itself may play a role to determine the ageing history of a mechanical part, the establishment of an ageing modelling method is a first step to quantify the resistance of a material to environmental agents.

103103

Enrique Guzman, Joël Cugnoni and Thomas Gmür

The degradation model is based on the following equation:

=

8

1i3210

0

)exp( ),,( + = =y tbxxxacEE

iiiε (1)

where E represents any of the measured elastic properties, while E0 is the corresponding initial value (before ageing). Therefore, y is a unitless ratio between the initial value of elasticity constants before and after ageing. The independent variables in the model are the normalized temperature T (x1), the normalized relative humidity RH (x2) and the normalized UV radiation intensity (x3). Coefficient c0 is the value of y for no ageing (a priori this is zero), while the ai are the parametric influences of each factor. Results are summarized in table 1, for ageing of several series of samples after 900 cycles. The combination function i(x1,x2,x3) = x1 x2 x3 quantifies the actions of the ageing factors. The exponents

, and are alternatively 0 or 1, thus there are 8 different combinations.

(a) (b)

Figure 1: (a) Time profile for each factor as simulated in a climatic chamber; (b) Evolution of the relative elastic properties as functions of time (in number of cycles).

Engineering constant

T RH UV T-RH T-UV RH-UV T-RH-UV

),,( 321 xxxiε a1 a2 a3 a12 a13 a23 a123

E1/E10 0.0129 0.0196 0.0183 -0.0200 -0.0139 -0.0268 0.0199 E2/E20 -0.0288 -0.0172 -0.0076 0.0497 0.0204 0.0121 -0.0151

G12/G120 -0.0336 -0.0104 -0.0063 0.0446 0.0134 -0.0058 0.0069

Table 1: Results of the parametric identification.

Using this multivariate model it is now possible to quantify the degradation of the composite for different ageing scenarios. The worst case conditions have been found to correspond to cycles with a combination of 45 to 135°C temperature, 0 to 300 W/m2 UV and 0 to 95 %RH.

The authors would like to acknowledge the partial financial support from the Swiss National Science Foundation, Grant No. 200021-143968/1.

REFERENCES

[1] S. Bondzic, J. Hodgkin, J. Krstina and J. Mardel, Chemistry of thermal ageing in aerospace epoxy composites, Journal of Applied Polymer Science 100(3), 2006, pp. 2210-2219.

[2] J. Cugnoni, T. Gmür and A. Schorderet, Inverse method based on modal analysis for characterizing the constitutive properties of thick composite plates, Computers and Structures85(17-18), 2007, pp. 1310-1320.

[3] J.R.White, Polymer ageing: physics, chemistry or engineering? Time to reflect, Comptes Rendus Chimie 9(11-12), 2006, pp. 1396-1408.

104104


Aalborg, 2013

Oral session 10 – Testing, material concepts and joining (4 presentations)

105105

106106

6th International Conference on Composites Testing and Model Identification O.T.Thomsen, Bent F.Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

SECUNDARY STRESS EFFECTS DURING LOAD INTRODUCTION INTO UNIDIRECTIONAL COMPOSITE TEST COUPONS

L. P. Mikkelsen and J. I. Bech

Department of Wind Energy, Technical University of Denmark P.O. Box 49, Frederiksborgvej 399, DK-4000 Roskilde, Denmark

Email: [email protected] and [email protected], web page: http://www.dtu.dk

Keywords: Unidirectional composites, fatigue tests, finite elements, thermography

ABSTRACT

The load introduction in the gripping region during tension and compression test of uni-directional high strength glass and carbon fiber reinforced composites is a big challenge due to the large difference in the tensile and compression strength in the order of 1GPa compared with the shear and transverse tensile strength on 40-60MPa. A difference which is even more pronounced during fatigue loading. Experimental testing typical show failure between the grips both for rectangular and waisted specimens, see figure 1a.Compared with a rectangular specimen, a waisted specimen will in addition exhibit splitting but will in many cases despite this give higher strength measurements. Nevertheless, many of the test samples will still fail between the grips. Therefore, even though the measured values can be considered as a conservative representation of the material strength, the measurements cannot be used comparing the axial strength of different material systems as it is not the pure axial strength of the material which is measured.

a) Failure modes b) Thermography

Figure 1: Rectangular and waisted compression/tension fatigue test samples.

Figure 1a) shows typical failure modes where the tabs are debonded from the test material inside and outside the grips resulting in final material failure outside the gauge section. For the rectangular test samples the debonds occurs in the full width of the specimens while for the waisted test sample, the debonded zone follows the projection of the gauge area into the clamped region, as splitting occur along this projection due to shear stresses distributing stresses from the narrow gauge section to the wide gripping section. For both cases, the final material fails by fiber breakage deep inside the gripping region.In figure 1b) a thermographic image of a test specimen during R=-1 fatigue cycling at 5 Hz is shown. A temperature increase of more than 10-20 degree Celsius and in some cases up to 50 degrees Celsius is found inside the gripping region during fatigue loading. A heat generation which could come from either hysteresis loss due extensive local straining of the material or friction from crack surfaces.

Material failure

Rectangular

Waisted

107

L.P.Mikkelsen and J. I. Bech

a) Two view of model b) Stress profile along 1x

Figure 2: The finite element model (1/8 due to symmetry) and stress predictions extracted along a path in the test material just below the interface between the test material and the tabs.

Several works, experimental [1] as well as FEM studies [2-3], address the issue of stress concentrations near the tip of the tabs attempting to optimize tab materials and geometries. Contrary, only few studies has focused on secondary stresses inside the grips even though these stresses may introduce tab debonding and premature failure. Figure 2a show a side and bottom (symmetry plane) view of the model where the green, red and gray region corresponds to the grips, tabs and test material, respectively. The grips is in the particular cases loaded by an equally large clamping and axial force on 50kN corresponding to an axial stress 11 800MPa in the gauge section. The grip is fixed against rotation and the interaction between the staggered teeth surface on the grips and the taps is taken into account using a Coulomb friction with a friction coefficient 1 . Figure 2b show the predicted stress profile extracted along a path in the gray test material just below the red tab material. The stiff grips results in a rather localized stress state in the much more flexible tab and test material with peaks around the end of the grips both regarding the shear stresses, 12 , and the transverse normal stresses, 22 . For the particular case, it is found that the shear

stresses vary between 12 [ 50;20]MPa in this point going from a compressive ( ) to a tensile ( )loading which is a critical range regarding fatigue loading specially including additional loading from the other stress components. The stress profile under the grips is found to be very sensitive to the properties of the interface between the taps and the grips but for all realistic cases a rather localized stress profile is found which are in strong contrast to the uniform shear stress condition used in e.g. [3]. An extensive finite element study has been performed compared with experimental measurements observations as the one shown in figure 1.

Acknowledgements: This research was supported by the Danish Centre for Composite Structure and Materials for Wind Turbines (DCCSM), grant no. 09-067212, from the Danish Strategic Research Council (DSF).

REFERENCES

[1] M. Hojo, Y. Sawada, H. Miyairi. Influence of clamping method on tensile properties of unidirectional CFRP in 0° and 90° directions — round robin activity for international standardization in Japan". Composites, 25, pp. 786-96, 1994. [2] T.A. Bogetti, J.W. Gillespie Jr and R. Byron Pipes. Evaluation of the IITRI compression test method for stiffness and strength determination. Composites Sci. Technol., 32, pp 57-76, 1988. [3] M. Xie and D.F. Adams. Effect of Specimen Tab Configuration on Compression Testing of Composite Materials. J.l of Composites Technology & Research, 17, pp 77-83, 1995.

0 10 20 30 40 50 60 70 80-100

-80

-60

-40

-20

0

20

40

Stre

sses

[MP

a]

Position [mm]

+12-12+22-22

Gripping area Tabs Gauge

Gripping area | Tabs | Gauge

Side view

Bottom view

1x

2x

1x3x

108


Aalborg, 2013

DISCONTINUOUS-PLY COMPOSITES FOR ENHANCED DUCTILITY

S. Pimenta1, P. Robinson1, G. Czél2, M.R. Wisnom2, H. Diao3 and A. Bismarck3

1The Composites Centre, Department of Aeronautics, Imperial College London South Kensington Campus, London SW7 2AZ, UK


2Advanced Composites Centre for Innovation and Science, University of Bristol Queens Building, University Walk, Bristol BS8 1TR, UK

3PACE Group, Department of Chemical Engineering, Imperial College London

South Kensington Campus, London SW7 2AZ, UK

Keywords: Ductile composites, Discontinuous-ply composites, Shear-lag theory.

ABSTRACT Continuous-fibre reinforced composites are remarkably stiff and strong, but tend to fail in a brittle manner; this leads to conservative design and makes them unsuitable for damage-tolerance demanding applications. Prepreg-based discontinuous composites are considerably more deformable during manufacturing than conventional ones, and have been shown notch insensitive [1]. However, the potential of cut-ply composites to exhibit enhanced ductility – due to the presence of discontinuities and the additional matrix shearing mechanism – has not been investigated. This work focuses therefore on developing a new analytical model for the mechanical response of discontinuous composites, and subsequently validating its predictions against experimental results. Consider the shear overlap (of length 2𝐿) represented in Figure 1a, composed by two stiff composite plies 𝒜 and ℬ (thickness 𝑡p, stiffness 𝐸p and strength 𝑋p), and an interlayer of soft matrix (thickness 𝑡m). The latter is assumed to have a generic piecewise linear constitutive response in shear (Figure 1b), so that each subdomain 𝑖 is characterised by its secant modulus 𝐺[ ] ⋛ 0. Considering a shear-lag stress transfer mechanism from ply 𝒜 to ℬ, longitudinal stresses 𝜎 𝒜(𝑥) and 𝜎ℬ(𝑥) (with Δ𝜎 = 𝜎 𝒜 − 𝜎ℬ) are related to matrix shear deformation 𝛾m(𝑥) and stresses 𝜏m(𝑥) by: − d𝜎 𝒜(𝑥)d𝑥 = d𝜎ℬ(𝑥)d𝑥 = 𝜏m(𝑥)𝑡p ∧ d𝛾m(𝑥)d𝑥 = 𝜎ℬ(𝑥) − 𝜎 𝒜(𝑥)𝑡m ⋅ 𝐸p ⇒

d Δ𝜎(𝑥)d𝑥 = 2 ⋅ 𝐺[ ] ⋅ Δ𝜎(𝑥)𝑡p ⋅ 𝐸p ⋅ 𝑡m . (1)

a) Shear overlap: geometry (half length 𝐿) and matrix subdomains. b) Piecewise linear matrix constitutive law.

Figure 1: Model development.

109109

S. Pimenta, P. Robinson, G. Czél, M.R. Wisnom, H. Diao and A. Bismarck

This differential equation defines the stress fields in each subdomain [𝑖] of the overlap (Figure 1a), as well as its process zone length 𝑙pz[ ] . Global fields are determined by imposing continuity of stresses and strains at the transitions (𝑥[ ]) between active subdomains ({𝑖, … , 𝑖 + 𝑛} in Figure 1a); their evolution with progressive loading is controlled by the combined process zone length 𝑙pz[ ,…, ]: 1. If 𝑙pz[ ,…, ] > 𝐿, the central domain [𝑖] is deactivated once 𝛾(𝐿) = 𝛾[ ]; 2. If 𝑙pz[ ,…, ] < 𝐿, a new edge subdomain [𝑖 + 𝑛 + 1] is activated once 𝛾 = 𝛾[ ]. The analytical formulation is validated against Finite Element (FE) models in Figure 2a. It is also shown that the response and failure mode of an overlap is controlled by its length: while short overlaps tend to reproduce the matrix response and fail in a strength-controlled way, longer overlaps fail either by unstable propagation of a crack between the two plies, or by ply fracture. Figure 2b compares model predictions and experimental results for a carbon-epoxy system (Hexcel IM7/8552) and three overlap geometries. Modelling assumed mechanical properties from the literature [2] and a bi-linear matrix response. Specimens were manufactured by laying-up pre-cut plies, producing 6 overlaps in the through-thickness direction; after curing, tests were performed under uniaxial tension, with optical strain monitoring on the lateral edge of each specimen. The model was able to reproduce the overall behaviour and strengths observed experimentally. The alignment between discontinuities (Figure 2b) was found to be crucial for achieving a good agreement between the model and experiments. While one configuration tested (2𝑡p = 0.25 mm and 2𝐿 =5.0 mm in Figure 2b) shows a non-linear behaviour with evidence of progressive delamination, further analysis revealed that, when using brittle resins, ductility will necessarily be very limited. Consequently, on-going work is exploring the use of thermoplastic matrices.

a) FE validation and effect of the overlap length. b) Experimental validation for IM7/8552.

Figure 2: Model results and experimental validation. The authors gratefully acknowledge the funding from EPSRC under the Programme Grant HiPerDuCT (High Performance Ductile Composite Technology, grant no. EP/I02946X/1).

REFERENCES [1] P. Feraboli, E.Peitso, T. Cleveland, P.B. Stickler and J.C. Halpin, Notched behavior of prepreg-

based discontinuous carbon fiber/epoxy systems, Composites: Part A, 40, 2009, pp. 289-299. [2] S.R Hallett, B.G. Green, W.G Jiang and M.R. Wisnom, An experimental and numerical

investigation into the damage mechanisms in notched composites, Composites: Part A, 40, 2009, pp. 613-624.

110110


Aalborg, 2013

THE EFFECT OF PRE-BOND MOISTURE AND TEMPERATURE ON THE FRACTURE TOUGHNESS OF BONDED JOINTS FOR COMPOSITE REPAIRS

S. Budhe1, A. Rodríguez-Bellido2, J. Renart1, J. Costa1

1AMADE. Escola Politècnica Superior. Universitat de Girona Campus Montilivi s/n. 17071 Girona, Spain


2Composite Technology, Materials and Processes, AIRBUS España S.L. Paseo John Lennon s/n. 28906 Getafe (Madrid), Spain


Keywords: Pre-bond moisture, Test Temperature, Fracture toughness, Moisture absorption

ABSTRACT

Composite structures, which are growingly being included in civil aircrafts, need to be repaired. Repair operations are based on bonding a composite patch on the damaged structure. The structural performance of the repaired component is mainly related to the quality of the bond. The part being repaired might have suffered different environmental histories leading, for example, to a variable moisture uptake. “In-service” repair operations preclude the use of the standard procedures for composite manufacturing (i.e. the use of an autoclave), so the quality of the bond between the patch and the damaged structure may be affected by the initial condition of the later. Adhesive strength generally shows temperature and moisture dependence. In the last years, there has been a growing demand, particularly in the aerospace industry, for bonding agents able to withstand high temperatures and moisture. The fracture toughness of the bond under mode I loading, GIC, is a property of the joint commonly taken as a quality indicator. A few studies have been reported on the pre-bond moisture effect and most of them concluded that the presence of moisture in the composite lead to the reduction in joint strength [2], while temperature showed a positive effect on GIC [3]. Unexpectedly, some results for certain bonded joints, claimed that strength increased with the moisture present in the adherents [1]. Matrix ductility [1, 3] is the most common explanation to the increase in the fracture toughness with temperature and moisture. Most of the previously published works were performed using dry and wet specimens in order to determine the effect of temperature on the fracture toughness behavior; only minor attention was paid to the pre-bond moisture of the adherents in the framework of the repair of composites. The focus of this communication is to characterize the effect of hygrothermal (temperature and pre-bond moisture) effects on the mode-I fracture toughness of composite joints prepared with two different laminating resins (LR1 and LR2) as bonding agents. This experimental study was performed by means of DCB specimens and their posterior fractographic analysis. Substrates of Carbon-Fibre Reinforced Polymers, CFRP, were produced in autoclave. To study the pre-bond moisture effect, some substrates were immersed in distilled water at 70°C for 336±12 hour and then dried for 1 hr at 80°C. This drying procedure is known to not remove completely the moisture in the material. Indeed, the final moisture content on the substrates was 1.25%, before bonding. The bonded joint specimens were manufactured by wet lay-up of fresh material over the pre-cured substrates. Half of the resulting specimens were tested after being manufactured (AR for as-received) while the other half were conditioned at 70ºC/85%HR until saturation of moisture content was reached (WET). The moisture uptake to reach saturation for specimens with pre-bond moisture conditioning was found to be approximately 0.56 % and 0.46 % for LR1 and LR2 respectively. For

111111

S. Budhe, A. Rodríguez-Bellido, J. Renart and J. Costa

the specimens not previously conditioned the moisture uptake at saturation was 0.67 % and 0.49 % for LR1 and LR2. The test temperature for WET specimens was 80°C and 120°C, whereas AR specimens were tested at Room Temperature (RT). The tests were performed according to ISO 15024 standard; each batch consisted on five specimens. A mean propagation value of GIC was determined by three methods, i.e Corrected Beam Theory, Modified Compliance Calibration and the Area methods. Due to industrial confidentiality, fracture toughness data for the different batches is presented as a normalized value to the fracture toughness of AR specimens tested at RT for each resin. A detailed discussion on the effects of moisture pre-conditioning and temperature on each resin will be presented in this communication. The discussion is supported by the fractographic inspection by visual means, optical microscopy and SEM.

Substrate Conditioning

Specimen conditioning

Test temp RT Test temp 80°C Test temp 120°C Normalized GIC

LR1 No - preconditioned Immersion + Drying No - preconditioned Immersion + Drying LR2 No - preconditioned Immersion + Drying No - preconditioned Immersion + Drying

As received As received Wet (70°C/85%) Wet (70°C/85%) As received As received Wet (70°C/85%) Wet (70°C/85%)

100 176

- -

100 98

- -

- -

206 228

- -

240 145

- -

670 562

- -

362 327

Table.1. Normalized GIC values for the joints with and without pre-bond moisture

REFERENCES [1] B. R. K. Blackman, B. B. Johnsen, A. J. Kinloch, W. S. Teo, The effect of pre-bond moisture on

the fracture behaviour of adhesively bonded composite joints, Adhesion, 84, 2008, pp. 256-276 (doi:10.1080/00218460801954391).

[2] B. M. Parker, The effect of composite pre-bond moisture on adhesive-bonded CFRP-CFRP joint, Composites, 14, 1983, pp. 226-232 (doi:10.1016/0010-4361(83)90008-3).

[3] L.E.Asp, L, The effect of moisture and temperature on the interlaminar delamination toughness of carbon/epoxy composites, Composite science and technology, 58, 1998, pp. 967-977 (doi:10.1016/S0266-3538(97)00222-4).

[4] J. E. Robson, F. L. Matthews, A. J. Kinloch, The bonded repair of fibre composites: effect of composite moisture content, Composite science and technology, 52, 1994, pp. 235-246 (doi:10.1016/0266-3538(94)90208-9).

112112


Aalborg, 2013

EXTRACTING THE STRAIN SOFTENING RESPONSE OF COMPOSITES USING A DETAILED FINITE ELEMENT ANALYSIS AS A VIRTUAL DIGITAL IMAGE

CORRELATION TECHNIQUE

N. Zobeiry1, A. Forghani1, R. Vaziri1, A. Poursartip1, X. Xu2, S.R. Hallett2, M.R. Wisnom2

1 Composites Research Network, Dept. of Materials Eng. and Civil Eng., The University of British Columbia, Vancouver, BC, Canada

Email: [email protected], web page: http://www.crn.ubc.ca

2Advanced Composites Centre for Innovation and Science, University of Bristol Queen's Building, University Walk, Bristol, UK


Keywords: Damage, Strain Softening, Characterization, Full Field Measurement, Finite Element

ABSTRACT From both numerical and experimental standpoints, it is very desirable to develop a general methodology that can be used to determine the strain-softening response and characteristic laminate damage properties (e.g. fracture energy, damage height) of composite materials. These damage properties and the strain-softening response can be used to conduct finite element analysis to simulate the damage behaviour of laminated composites in large-scale structures. To characterize the damage behaviour of laminated composites, experiments such as over-height compact tension (OCT) [1] or compact compression (CC) [2] tests are usually conducted to produce a stable damage growth. Recently, a new methodology has been developed that identifies the strain-softening response of composite materials directly from these experiments [1]. Using the digital image correlation (DIC) technique, full-field displacement vectors of the specimen surface are measured during each test. Based on the acquired data combined with application of the basic principles of mechanics, a family of approximate stress-strain curves are obtained. This methodology provides insight into the details of damage propagation in composite materials and can be used to characterize the strain-softening response of composites. However, in cases where splitting and delamination become dominant modes of failure [e.g. 3], surface strains can no longer be correlated with strains through the thickness of the laminate and therefore techniques that are based on the surface measurements alone do not result in a full characterization of the damage behaviour. An alternative to the above approach is to conduct a numerical investigation into the progressive damage development in laminated composites. Recently, a detailed finite element analysis approach has been developed at the meso-scale to simulate the interaction of major failure mechanisms in laminated composites including fibre failure, matrix cracking, delamination and splitting [4]. Such a model can potentially be used to extract the strain-softening response of composites without the need to conduct experiments. In this study, OCT tests have been conducted on IM7/8552 quasi-isotropic laminates and using the DIC technique, surface displacements are obtained. Using these values and utilizing the recently developed approach [2], damage properties including a family of strain-softening curves that describe the overall damage behaviour of the laminates are obtained (Fig. 1). In parallel to the experiments, detailed finite element analyses of the OCT tests are performed. Surface displacements obtained from these analyses are then compared with the surface displacements measured using the DIC technique. In addition, the surface displacements from the finite element analysis are used to extract the damage properties and strain-softening response of the material (Fig. 2). The successful comparison between the strain-softening response obtained from the experiments and the FE analysis, serves to validate the

113113

N. Zobeiry, A. Forghani, R. Vaziri, A. Poursartip, X. Xu, S. Halleti, M. Wisnom

detailed FE analysis approach as a potential virtual testing technique to extract the damage response of laminated composites without the need to conduct experiments.

(a)

Stress Contour

Family of measured strain-softening responses

(b)

200

400

600

800

0.02 0.04 0.06 0.08

Average Strain

Stre

ss (M

Pa)

00

Figure 1: (a) Stress distribution around the damage zone in an OCT test, and (b) family of approximate

strain-softening responses obtained using a previously developed technique [2].

(a) (b)

0

200

400

600

800

0 0.02 0.04 0.06 0.08

Strain

Stre

ss (M

Pa)

Figure 2: (a) Detailed finite element analysis of the OCT specimen (b) Idealised strain-softening

response of the damage zone obtained from the detailed FE analysis.

REFERENCES

[1] I. Kongshavn and A. Poursartip, Experimental investigation of a strain-softening approach to predicting failure in notched fibre-reinforced composite laminates, Composites Science and Technology, 59, 1999, pp. 29-40.

[2] N. Zobeiry, Extracting the strain-softening response of composites using full-field displacement measurement, Ph.D. Thesis, The University of British Columbia, Vancouver, Canada, 2010.

[3] X. Li, S.R. Hallett, S.R., M.R. Wisnom, N. Zobeiry, R. Vaziri, A. Poursartip, Experimental study of damage propagation in overheight compact tension tests, Composites: Part A, 40, 2009, 1891-1899.

[4] X. Li, S.R. Hallett, M.R. Wisnom, Numerical investigation of progressive damage and the effect of layup in overheight compact tension tests, Composites Part A, 43, 2012, 2137-2150

114114


Aalborg, 2013

POSTER SESSIONS

115115

116116


Aalborg, 2013

Poster session 1 - Damage and Failure (9 presentations)

117117

118118


Aalborg, 2013

MODAL-BASED DAMAGE IDENTIFICATION IN COMPOSITE STRUCTURES USING SPATIAL WAVELET TRANSFORM

Andrzej Katunin1

1Institute of Fundamentals of Machinery Design, Silesian University of Technology Konarskiego 18A, 44-100 Gliwice, Poland


Keywords: Composite structures, Damage identification, Wavelet transform, Modal analysis

ABSTRACT Outstanding strength and stiffness properties with simultaneous low specific mass of polymer-based composites decided about their popularity in industrial applications. Many of these applications concerned with automotive, aircraft and spacecraft industries, where the polymer-based composites are used for responsible elements and constructions subjected to high stresses and vibrations. Therefore, such elements should be diagnosed in order to detect and identify occurred damages in possibly early stage of their development. There are a lot of methods and techniques used for damage identification in composite structures including infrared-based techniques, shearography, X-ray-based techniques etc. One of the biggest groups of techniques used for inspection of polymer-based composites is a group, which is based on vibration tests and modal analysis. Old modal-based techniques, e.g. tap test and operational modal analysis, give information about the damage only when the damage is large and do not give information about its location. However, the advanced signal processing methods could improve much the accuracy of the damage detection and identification. One of the promising signal processing method is the wavelet transform of measured vibration signals. Using the discrete wavelet transform of the modal shapes of a structure it is possible to separate a signal into two parts: approximation and detail coefficients. Then, the detail coefficients are analysed in order to detect and localize small singularities in a signal. The great practical importance of this approach has its generalization to the 2D problems, which give a set of approximation coefficients and three sets of detail coefficients (horizontal, vertical and diagonal). Previous studies in this field [1] had shown the great accuracy of damage identification both for numerical and experimental tests. The tests were preformed for square artificially damaged laminated composite plates clamped on the edges. Obtained results show that there are some factors which have a great influence on the accuracy of a method. The great importance during the analysis has a choice of appropriate wavelet, previous studies conducted [1,2] that the best wavelet family, with respect to other families of orthogonal compactly supported wavelets, is the B-spline wavelet family. During the application of the signal processing algorithm there are some unwanted derivative phenomena, e.g. boundary effect which causes singularities on the boundaries of the investigated domain. The methods of its reduction were analyzed in [3]. Another important factor is the choice of a type of wavelet transform. Preliminary studies shown that the discrete transform is the less time-consuming and the most accurate one in comparison with continuous wavelet transform, stationary wavelet transform and the wavelet transform with lifting scheme. In order to improve and test the proposed algorithm of signal processing for damage identification in composite structures some additional studies were carried out. It was noticed that the magnitudes of coefficients after wavelet transform are proportional to the amplitudes of modal shapes and thus, the damage identification is possible only in the regions with high displacement. For solving this problem it is essential to consider a few modal shapes in the analysis.

119119

Andrzej Katunin

The algorithm should be also applicable for damage identification in structures with high degree of non-linearity of their geometry. In order to test its applicability the numerical model of a composite rotor blade with longitudinal crack through the whole length was taken into consideration. In this case the projection on the plane was used for extracting only one (bending) displacements’ component. The results of analysis were presented in Figure 1.

Figure 1: Results of damage identification in a composite rotor blade.

As it could be noticed the results of analysis for the first modal shape (Figure 1.a) presented only a part of a true damage resulted by low amplitudes in the left-side region of a blade. Following the previous proposition the absolute values of all details coefficients for the first five bending modal shapes were added. Using such an approach the modelled crack was identified and clearly localized (Figure 1.b). For improving the presented algorithm it is planned to develop new spatial wavelets and use them for damage identification. The research project was financed by the Polish National Science Centre granted according the decision no. DEC-2011/03/N/ST8/06205.

REFERENCES [1] A. Katunin, Damage identification in composite plates using two-dimensional B-spline

wavelets, Mechanical Systems and Signal Processing, 25(8), 2011, pp. 3153-3167. [2] A. Katunin, The construction of high-order B-spline wavelets and their decomposition relations

for fault detection and localisation in composite beams, Scientific Problems of Machines Operation and Maintenance, 46, 2011, pp. 43-59.

[3] A. Katunin, Reduction of the boundary effect during structural damage identification using wavelet transform, Selected Engineering Problems, 3, 2012, in press.

120120


Aalborg, 2013

CHARACTERISATION OF BRIDGING MECHANISMS IN A SINGLE Z-PINNED COMPOSITE LAMINATE

Mehdi Yasaee1, James K. Lander2, Giuliano Allegri1 and Stephen R. Hallett1

1Advanced Composites Centre for Innovation and Science (ACCIS), University of Bristol, Aerospace Engineering, Queen’s Building, University Walk, Bristol, BS8 1TR, UK Email: [email protected], [email protected], web page:

http://www.bristol.ac.uk/composites/

2Rolls-Royce plc, PO Box 31, Moor Lane, Derby, DE24 8BJ Email: [email protected]

Keywords: Through thickness reinforcement, damage tolerance

ABSTRACT

The lack of reinforcement in the thickness direction of laminated composites leads to delamination damage as the dominant failure mode in various practical applications. A method to overcome this inherent weakness is Z-Pinning. This process involves the insertion of small diameter pins, made from fibrous composite or metals, through the thickness direction of the composite material. This is done prior to the final cure process and results in a composite structure capable of resisting delamination growth [1] and thus improving impact damage tolerance [2] and aiding the performance of joined structural composite parts, such as stringers [3]. Experimental studies on z-pinned composite laminates have typically characterised arrays of pins through standard fracture toughness tests [3-4] and bespoke pull-out and shear tests [4-5]. However due to pin to pin interaction and the large variation of the inserted pin quality and offset angles which arise from the manufacturing process, it is difficult to extract single pin behaviour from such tests. The purpose of this investigation was to characterise the behaviour of a single z-pin inserted through the thickness of a composite block directly. Mode I, mode II and mixed mode loading cases were considered. The data collected will be useful for the calibration of z- pin bridging laws that can then be implemented into a cohesive zone model for high fidelity finite element analysis.

Figure 1: Diagram of single z-pin test specimen

A schematic diagram of a z-pin specimen manufactured for this study is shown Fig. 1. Each specimen is made up of 64 plies of IM7/8552 pre-preg (Hexcel, UK) with a 16 m PTFE release film inserted at the mid-plane. A single 0.28mm diameter T300 carbon/BMI pin was inserted through the thickness. The two stacking sequences used are shown in .

121


Uni-Directional (UD)

Top Half [032/10]

Bottom Half [032]

Cross-Ply (XP)

Top Half [0/90]8s

Bottom Half [90/0]8s

Table 1: Layups used in experimental investigation

Representative load vs. displacement results of the UD and XP samples tested at different mode mixities are presented in Fig. 2. The results indicate a substantial difference between the bridging mechanisms of a single pin inside a UD and XP specimens.

Figure 2: Representative UD and XP samples load vs. displacement results

Analyses of the different mechanisms which contribute to the bridging response of each mode mixity and layup are discussed. The influence of the stacking sequence and the offset angle of the pins within the laminate are quantified and discussed. Acknowledgements The authors would like to acknowledge the support of Rolls-Royce plc for the support of this research through the Composites University Technology Centre (UTC) at the University of Bristol, UK.

REFERENCES [1] I.K. Partridge, D.D.R. Cartie, Delamination resistant laminates by Z-Fiber pinning: Part I

manufacture and fracture performance, Composites Part A: Applied Science and Manufacturing, 36, 2005, pp. 55-64 (doi: 10.1016/j.compositesa.2004.06.029).

[2] X. Zhang, L. Hounslow, M. Grassi, Improvement of low-velocity impact and compression after-impact performance by z-fibre pinning, Composites Science and Technology, 66, 2006, pp. 2785-2794 (doi: 10.1016/j.compscitech.2006.02.029).

[3] K.L. Rugg, B.N. Cox, R. Massabò, Mixed mode delamination of polymer composite laminates reinforced through the thickness by z-fibers, Composites Part A: Applied Science and Manufacturing, 33, 2002, pp. 177-190 (doi: 10.1016/S1359-835X(01)00109-9).

[4] D.D.R. Cartie, M. Troulis, I.K. Partridge, Delamination of Z-pinned carbon fibre reinforced laminates, Composites Science and Technology, 66, 2006, pp. 855-861 (doi: 10.1016/j.compscitech.2004.12.018).

[5] H. Liu, W. Yan, X. Yu, Y. Mai, Experimental study on effect of loading rate on mode I delamination of z-pin reinforced laminates, Composites Science and Technology, 67, 2007, pp. 1294-1301 (doi: 10.1016/j.compscitech.2006.10.001).

122


Aalborg, 2013

CHOICE OF AN ANALYTICAL SCHEME IN CORRELATING STRAIN ENERGY RELEASE RATE, CRACK LENGTH AND OPENING OF THE FACES OF AN

ADHESIVELY BONDED, THICK COMPOSITE DCB SPECIMEN

A. Bernasconi1, A. Jamil1

1Department of Mechanical Engineering, Politecnico di Milano via La Masa 1, Milan, Italy

Email: [email protected] web page: http://www.mecc.polimi.it

Keywords: Composites, Adhesives, Double Cantilever Beams DCB, Data reduction scheme

ABSTRACT In a previous paper [1], mode I static tests were conducted on double cantilever beams DCB made up of adhesively bonded composite material (see Figure 1.). The strain energy release rate SERR and compliance was correlated with the crack length using both numerical and analytical methods. In this paper we compare and discuss in detail the various analytical methods employed for the determination of crack length from the opening of the faces and evaluation of the strain energy release rate SERR, along with a detailed description of the method used to determine the flexural and the shear moduli of the laminate. The techniques developed by Krenk [2] was first utilised, which involved the approximation of the adhesive as a continuous distribution of independent springs, whereas, the adherents were modelled by classical beam theory. This resulted in a cubic relation between the opening δ and the crack length a. The SERR and opening δ were not in agreement with the results obtained from FEA. The effect of shear on the deformation and orthotropic behaviour of the material were considered as the main causes. To assess the influence of shear deformation, three point bending tests were performed at different span lengths, on specimen having same stacking sequence as that of the double cantilever beam DCB as seen in Figure 2. Flexural modulus E and out of plane shear modulus Gxy were evaluated by linear interpolation of load versus strain test data. These values were used to simulate numerically the bending of a nominal sized CRFP specimen, by using two-dimensional plane stress elements, and in three-dimensions using conventional and continuum shell elements. Deflections were evaluated analytically with the identification of contributions of bending and shear. The slopes so obtained from the load versus displacement data of numerical simulations were compared with the experimental ones and for shorter span lengths considerable effect of shear stiffness was observed. The Krenk’s formulation was therefore modified with shear in accordance with the works of Williams [3], which concludes that the shear stiffness is the main factor causing composite DCB specimens to deviate from the cantilever beam theory and was again compared with the results obtained by numerical simulations using the virtual crack closure technique VCCT as seen in Figure 3. Timoshenko beam on Winkler elastic foundation (TB on WEF), developed by Kondo [4], was further utilised for the evaluation of SERR and compliance, which effectively involves the shear stiffness considerations and was found to be in close agreement with the results obtained from FEA. Another data reduction scheme, known as Timoshenko Beam on a Two-Parameter Elastic Foundation (TB on 2PEF), developed by Shokrieh et al. [5] was utilized, which is an extension of the Timoshenko Beam TB by taking into account the effect of crack tip deformations under the interface peel and shear stresses as well as shear deformations at beams thus forming a two-parametric elastic foundation (with extensional and rotational springs). The results so obtained were also in agreement with the FE results obtained through VCCT.

123123

A. Bernasconi, A. Jamil

Figure 1. Configuration of double cantilever beam DCB

Figure 2. Schematic representation of the three-point bending test.

(a) (b)

Figure 3. (a) Face openings vs. the crack length for FEA & other analytical schemes (b) Crack length vs. SERR for FEA and other analytical schemes

ACKNOWLEDGEMENTS This work was funded by Regione Lombardia in the framework of the project ‘‘STIMA – Strutture Ibride per Meccanica ed Aerospazio’’- Bando Accordi Istituzionali - ID_Progetto : 14567 REFERENCES [1] A. Bernasconi, A. Jamil, F. Moroni, A. Pirondi, A study on fatigue crack propagation in thick

composite adhesively bonded joints, Int. J. Fatigue (2012), http://dx.doi.org/10.1016/j.ijfatigue.2012.05.018

[2] S. Krenk, Energy release rate of symmetric adhesive joints, Engineering Fracture Mech. 43, 1992, pp. 549-559.

[3] J.G. Williams, End corrections for orthotropic DCB specimens, Composite Science & Technology 35, 1989, pp. 367–376.

[4] K. Kondo, Analysis of double cantilever beam specimen, Advanced Composite Materials 4, 1995, pp. 355–366.

[5] M.M. Shokrieh, M. Heidari-Rarani , M.R. Ayatollahi, Calculation of GI for a multidirectional composite double cantilever beam on two-parametric elastic foundation, Aerospace Science and Technology 15, 2011, pp. 534–543.

20

30

40

50

60

70

0.1 0.6 1.1

Cra

ck L

engt

h a

[mm

]

Strain Energy Release Rate G [N/mm]

KrenkMod_KrenkTB on WEFTBM on 2PEFVCCT FEA

0

0.5

1

1.5

2

2.5

3

20 40 60 80 100

Ope

ning

[δ m

m]

Crack Length a [mm]

KrenkMod_KrenkTB on WEFTB on 2PEFVCCT FEA

124124


Aalborg, 2013

CFRP FATIGUE TESTING AND ISSUES FOR AERONAUTICAL APPLICATIONS

V. Dattoma, R. Nobile and F.W. Panella

1 Diaprtimento di Ingegneria dell’Innovazione – Università del Salento, Lecce - Italy Email: [email protected], [email protected], [email protected]

web page: http://www.dii.unisalento.it

Keywords: CFRP, Fatigue testing, bearing, interlaminar shear, open hole compression

ABSTRACT

CFRP parts are widely used in aeronautics since long time, but in the recent years main structural elements are also to be produced in composite materials, in order to gain improved resistance/weight ratios and reduce fuel costs normalized to number of passengers and with environmental benefices. The material certification for CFRP laminates to be applied on civil aeronautical structures is statically determined by the main constructors and the principal strength limit values are well known for any loading conditions in case high strength and high modulus pre-preg carbon fabrics of elevated quality are considered. Particular care is found to be critical in specific structural configurations, in case of bearing behaviour for bolted laminates for example under tensile and compression loading state, due to the hybrid material junction characteristics and by-pass problems. Other special cases are carefully studied for aircraft reliability purposes, such as the Open Hole compression properties and Interlaminar Shear strength, which are related to the most dangerous failure modes for composite made aeronautic parts, especially in case the working conditions are approaching the temperature material limits. In addition, the fatigue limit values are in general claimed to be extremely elevated for aeronautical composites, as compared to the tensile strength for example, giving rise to the concept of composites considered as fatigue failure free materials and with small safety factors to be needed. Practical experience and recent fatigue experiments showed a certain level of fatigue material decay is always present in different amounts and according to specific composite characteristics and stress states. In this work, the results of a large amount of fatigue tests are presented and described for the three most interesting ASTM testing and loading procedures: tension-tension Bearing by-pass (BEA tests), Open Hole Compression (OHC tests) and Interlaminar shear strength determination (ILSS tests). The current standard codes are well suited for the mentioned tests only in case of static load [1]; special care on specimen preparation is always recommended and special fixtures are clearly described to conduct tests in a standard and repetitive way. When fatigue loading is applied the test procedure is not directly applicable, cause several difficulties appear case by case. The experience acquired during the first fatigue tests and the conceived experimental solutions allowed to conduct the testing campaign in the most efficient way, producing interesting results and useful observations. The residual strength determination is also performed for not failed specimens and values are compared to the ultimate strength. In Figure 1, the test fixtures used for the fatigue test are reported. Test fixture used for Bearing test is very simple with respect to the one defined in standard ASTM D5961 and it consists of two plates connected by several bolts. Finally, while test fixture used for Short beam specimen is the same of the standard ASTM D2344, the test fixture used for Open Hole Compression is commonly used and well-known as CRAG. Particular attention is needed to ensure alignment of specimen in the test fixture and minimize load transfer through the anti-buckling plates, since for fatigue tests the strain gauges are not applied to check the stress distribution. All the fatigue tests were carried out on a servohydraulic test machine at a frequency of 5 Hz and a load ratio R = 0.1. A limit value of 106 cycles was considered for run-out tests. The first fatigue results, which are referred only to Bearing and Short beam specimens, are resumed in Figure 2 in terms of applied stress amplitude normalized to the corresponding ultimate strength of the specimen. The fatigue curves shows a very limited slope and consequently elevated scatter of the

125125

V. Dattoma, R. Nobile and F.W. Panella

experimental data. However, the Bearing specimens highlights better fatigue behaviour with respect to ILSS data, since the fatigue limit is approximately near to the 40% of the static strength (Fig. 2). The first results of OHC tests seems to produce similar results to BEA data. From these test results the more critical loading condition is realized in the Short Beam specimen.

a) b) c) Figure 1: Fixtures and test configuration: Bearing (a), Short Beam (b) and Open Hole Compression (c).

y = 43.8x-0.036

R² = 0.8943

y = 239.2x-0.128

R² = 0.8687

10

100

1.0E+03 1.0E+04 1.0E+05 1.0E+06 1.0E+07

Norm

aliz

ed S

tress

Am

plitu

de [%

]

Number of cycles

Fatigue strenght behaviour

ILSSBEA

Figure 2: Comparison of fatigue behaviour of Bearing specimen and Short Beam specimen.

REFERENCES

[1] J.M. Hodgkinson (editor), Mechanical Testing of Advanced Fibre Composites, Woodhead Publishing, Abington Hall, 2000.

126126

6th International Conference on Composites Testing and Model Identification O.T. Thomsen, Bent F. Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

EFFECT OF PLY THICKNESS ON THE FATIGUE DELAMINATION GROWTH IN TAPERED LAMINATES: MEASUREMENTS AND ANALYSIS

Stefanos Giannis and Choothum Jeenjitkaew

Materials Engineering Research Laboratory Ltd., Wilbury Way, Hitchin SG4 0TW, UK

Email: [email protected], web page: http://www.element.com

Keywords: Composites, Ply Drop-off, Fatigue, Delamination, Testing, VCCT plates

ABSTRACT Tapered composite laminates formed by terminating or dropping-off some of the plies have received much attention from researchers, because of their potential for creating significant weight savings, by tailoring the thickness of the components. Nevertheless, ply terminations create geometry and material discontinuities that can act as potential sources for delamination initiation and propagation. There has been significant research work to understand the failure mechanisms induced by ply drop-offs in tapered laminates through the determination of the state of interlaminar stresses in the vicinity of ply drop-offs, the calculation of strain-energy release rate associated with delamination within the tapered region, and the direct modelling of delamination progress by using finite element analysis [1]-[3]. The delamination analysis of tapered composite laminates involves the determination of interlaminar stresses using finite element methods or analytical tools, the prediction of delamination onset location, and simulation of delamination propagation. In order to predict delamination onset and growth and, hence, the performance of the composite laminates, some form of failure predictive methodology needs to be applied. Two general approaches exist for this purpose. They are the strength-of-materials approach (stress, strength approach) and the strain-energy-release-rate approach (fracture mechanics approach) [4]. Although much of the research work up to now was driven by commercial and military aircraft and rotorcraft applications, the benefits that tapered composite laminates offer to a design engineer, can also been seen in the design of load bearing spars for wind and tidal turbine blades. The basic difference is that in the renewable energy sector usually much heavier prepregs are used in out-of-autoclave manufacture processes, which result in thicker plies and hence structures. In this work we aimed to experimentally quantify the effect of ply thickness on the performance of a tapered laminate by employing the same stacking sequence but prepreg materials of different fibre areal weights. In addition, by combining the material’s behaviour, obtained through established fracture mechanics tests, and finite element analysis, we have attempted to predict the observed behaviour, and hence validate a numerical approach for fatigue life prediction. A unidirectional HS-Carbon/epoxy material was used in three fibre areal weights: (A) 600 gsm, (B) 400 gsm and (C) 300 gsm. Tapered laminates were hand laid-up and oven cured at 80 °C for 5 hrs. Pressure was applied through a vacuum bag. The resulting cured ply thickness for the three laminates was 0.66 mm, 0.45 mm and 0.33 mm, respectively. The lay-up was [0°/0°/+45°/0°/-45°/0°3]S with the 0°, +45 , 0° and -45° plies dropping at approximately 30 mm intervals, and with a 0° ply used as a cover to provide a core (i.e. thin end) of eight plies in total. Quasi-static and fatigue tests were performed on all three configurations. Two PixeLINK™ digital cameras were employed to record photographs at specified time intervals during the tests to assist with identifying the damage initiation and ultimate failure more accurately. Both edges of the composite specimens were painted white and a grid was marked to enable tracking of the delamination onset and growth. All data (i.e. load, displacement, photographs) were recorded using a purpose built LabView® program. In Fig. 1(a), a sequence of photographs taken for a specimen tested at 45 kN maximum fatigue load and R=0.1 shows the exact location of damage initiation and the direction of the delamination propagation. Using the grid marked on each specimen the delamination growth rate was established. In

127127

Stefanos Giannis and Choothum Jeenjitkaew

all cases delamination initiated at the end of the 0° terminating ply closest to the core and propagated at either side of this ply towards the thick end of the laminate configuration. The experimental results in Fig. 1(b) showed a clear performance change as the ply thickness increased to 0.66 mm, for both the quasi-static and fatigue tests. Under static loading (N=1) the stress at delamination initiation was proportional to the inverse square root of ply thickness. Under fatigue loading the same proportionality factor was found. This was explained by means of the critical strain energy release rate at the delamination front.

(a) (b)

Figure 1: (a) Delamination initiation and growth for tapered laminate tested in fatigue

(b) Experimental results for cycles for onset of delamination

ACKNOWLEDGMENTS The authors would like to acknowledge the UK Technology Strategy Board for co-funding this research work under the BMAX (New Materials and Methods for Energy Efficient Tidal Turbines) project (Project No. 100514). Special thanks to AEL (http://www.aviationenterprises.co.uk) for manufacturing and preparing of all tapered composite specimens.

REFERENCES

[1] S.V. Hoa, B.L. Du and T. Vu-Khanh, Interlaminar Stresses in Tapered Laminates, Polymer Composites, 9, 1988, pp. 337-344.

[2] G.B. Murri, T.K. O'Brien and S.A. Salpekar, Tension Fatigue of Glass/Epoxy and Graphite/Epoxy Tapered Laminates, American Helicopter Society Journal, 38, 1993, pp. 29-37.

[3] M.R. Wisnom, M.I. Jones and W. Cui, Failure of Tapered Composites Under Static and Fatigue Tension Loading, AIAA Journal, 33, 1995, pp. 911-918.

[4] S. Giannis, Utilising Fracture Mechanics Principles for Predicting the Mixed-Mode Delamination Onset and Growth in Tapered Composite Laminates, Composite Structures, Under Review.

128128


Aalborg, 2013

IDENTIFICATION OF DAMAGE MODES IN CERAMIC MATRIX COMPOSITES BY

ACOUSTIC EMISSION SIGNAL PATTERN RECOGNITION

Nathalie Godin, Mohamed R’Mili, Pascal Reynaud, Gilbert Fantozzi, INSA-Lyon, Laboratoire MATEIS, Villeurbanne, France

[email protected], mohamed.rmili@insa- lyon.fr, pascal.reynaud@insa- lyon.fr, gilbert.fantozzi@insa- lyon.fr,

Keywords: CMC, Acoustic emission, Damage Mechanisms, Pattern Recognition

ABSTRACT

Ceramic matrix composites (CMC) are interesting structural materials for high temperature

applications, owing to their good mechanical properties at elevated temperatures. Cf/SiC composites have been developed for aeronautic and aerospace applications. Acoustic emission is a transient wave resulting from the sudden release of stored energy during damage. Detection and analysis of acoustic emission (AE) are powerful means for identification of damage phenomena and monitoring of their evolution. Damage of fiber reinforced ceramic matrix composites involves several phenomena at various length scales including matrix cracking, fiber debonding, fiber failures. The objective of the present paper is to propose a quantitative approach to damage identification based on signal analysis by pattern recognition. The signals recorded depend on the source, material microstructure and sensor features. However, in identical conditions, similarities exist among the AE signals originating from similar sources. Consequently, it can be considered that a signal represents a source and that the acoustic signature of a damage mode can be determined.

A 3D multi-layered composite made of self-healing [Si-B-C] matrix reinforced ex-PAN (High

Resistance) carbon fibres (40% volume fraction) was investigated. Test specimens with 4 mm thickness, 16 mm width, and 30 mm gauge length were tested. Static and cyclic fatigue tests were conducted in air at various temperatures (700°C, 1000°C and 1200°C) under uni-axial on axis tensile loading. The cyclic fatigue tests at 700°C and 1000°C were conducted under tensile/tensile sinusoidal loading with constant amplitude and 0.25 Hz frequency.

Two methods [1-3] for signals discrimination were implemented in order to analyse AE data and identify source mechanisms involved during fatigue at high temperatures on C/[Si-B-C] composites: an unsupervised classification (PCA + k-means) and a supervised classification (K-nearest neighbours). First, the unsupervised classification gave reproducible clustering solutions with 4 (denoted A, B, C and D) or 5 classes (denoted A, A’,B, C and D) in static fatigue (depending on testing conditions) and 4 (denoted A, B, C’ and D) classes in cyclic fatigue. Analysis of AE activity, mechanical behaviour and SEM observations led to the identification of 6 distinct types of AE signals, which were each associated with one main damage mechanism.

129129


Figure 1. Flow chart representation of the pattern recognition method proposed for the analysis of the AE data showing the main steps.

The unsupervised classification method was used to differentiate the signals generated during reference fatigue tests performed on C/SiC samples at high temperature. Each class of signals was then associated to relevant damage mode: Class A: Collective fibre breaks, Class A': individual fibre breaks, Class B: matrix cracking, Class C: fibre/matrix debonding, class C': yarn/yarn debonding, Class D: sliding at fibre/matrix interfaces and closure of matrix cracks after unloading. The training set for the supervised classification set was created by merging AE data collected during reference static fatigue tests (1200 ° C-150 MPa) and cyclic fatigue tests (700 ° C-0/130MPa). The library of signals was then used to identify the damage modes generated during fatigue tests performed at various temperatures. The use of a supervised classification allowed identifying AE signals with a better accuracy regardless of testing conditions (temperature, applied stress and loading mode). The 6 types of signals could be distinguished even when present in small quantity. It allows real-time identification of damage mechanisms regardless of testing conditions (temperature, applied load and loading mode).

REFERENCES [1] Moevus M, N. Godin, D. Rouby, R’Mili M, Reynaud P, Fantozzi G, Farizy G. Analyse of

damage mechanisms and associated acoustic emission in two SiC/[Si-B-C] composites exhibiting different tensile curves. Part II: Unsupervised acoustic emission data clustering. Comp Sci Technol; 68 (6),1258-1265 (2008).

[2] Momon S, Godin N, Reynaud P, R’Mili M, Fantozzi G. Unsupervised and supervised classification of AE data collected during fatigue test on CMC at high temperature. Composites Part A 43 254-260 (2012)

[3] Sibil A, Godin N, R’Mili M, Maillet E, Fantozzi G. Optimization of acoustic emission data clustering by a genetic algorithm method. Journal of Nondestructive Evaluation 31(2) 169-180 (2012).

Unsupervised

acquisition Fiber failure

Matrix cracking Sliding

debonding

Measure of similarity

K-means Algorithm

For k=2 to 10

Feature selection using hierarchical clustering feature extraction using Principal

data

Di

Dj

DB validation

Di

Dj Labelling of the

Di

Dj

Training data: Library of identified signals

d h i lib

130130

6th International Conference on Composites Testing and M odel Identification O.T.Thomsen, Bent F. Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

Lifetime prediction with acoustic emission during static fatigue tests on ceramic matrix composite at high temperature under air

Nathalie Godin, Mohamed R’Mili, Pascal Reynaud, Gilbert Fantozzi,

INSA-Lyon, Laboratoire MATEIS, Villeurbanne, France [email protected], mohamed.rmili@insa- lyon.fr, pascal.reynaud@insa- lyon.fr,

gilbert.fantozzi@insa- lyon.fr,

Keywords: CMC, Acoustic emission, lifetime prediction

ABSTRACT

Non-oxide ceramic matrix composites and more particularly SiCf/[Si-B-C] or Cf/SiC composites have been widely studied during the last decades and are very attractive candidates for many high temperature structural applications, thanks to their high temperature strength and low weight. Future engine applications in civil aircrafts are foreseen for such composites, but these applications require a very long lifetime in service conditions. Expected lifetimes in service conditions are tens of thousands of hours and therefore unattainable in laboratory tests. The objective of this paper is to propose an Acoustic Emission (AE) based approach to lifetime prediction for CMCs.

In this study, Cf/[Si-B-C] and SiCf/[Si-B-C] composites with self-healing matrix were

studied under static fatigue under air at temperatures of 700°C, 1000°C and 1200°C for the Cf/[Si-B-C] composite and at 450 °C and 500 °C for the SiCf/[Si-B-C] composite. AE was recorded during the tests. A main purpose of this study was to consider the possibility of predicting rupture time from damage evolution recorded by AE technique. It has been observed that an increase in the seismic activity, prior to large earthquakes, is described by a power law. The cumulative Benioff [1] “strain” (the sum of the square root of the energy released for sequential earthquakes) has been suggested as a precursory phenomenon of large earthquakes, increasing as an inverse power-law of time before the main shock. This precursory phenomenon is similar to the evolution of damage in a material during the tertiary creep stage. In this paper, results derived previously from the problem of seismic activation prior to earthquakes are applied on CMC. Under constant stress, micro cracks are created, which generate elastic waves in a manner similar to earthquakes.

Two approaches based on the analysis of liberated energy are applied [2]. First, a

coefficient RAE was defined in order to describe the evolution of AE activity during test. Then, the applicability of the Benioff law was examined in a qualitative way. It is generally accepted that the energy of an AE signal represents a part of the energy released at the source. The recorded AE signal energy is affected by distance of wave propagation, energy attenuation due to damage, coupling between sensor and material, and by sensor frequency response. The effects of attenuation due to propagation distance effects and material damage can be eliminated by combining AE signals energies recorded by sensors. The influence of coupling may lead to a different amount of energy received by each sensor even for a source located at equal distance. Hence, by comparing the amount of energy received at each sensor from any source located at mid-distance, recorded AE energies can be calibrated. The effects of attenuation due to damage are also corrected using calibration tests [3]. An equivalent energy of AE sources is calculated for each AE event from the signal energy received at two

131131


sensors. A method of real-time analysis of associated energy release has been developed. The AER coefficient decreases first (Fig.1), down to a minimum value for t = tm, and then it

increases up to the failure of the composite at time tc. It allows the identification of a characteristic time tm/tc at 55% of the measured rupture time. The model based on Benioff’s law provides a satisfactory approximation of the acceleration of the process of the AE release after the minimum of the AER coefficient. This result suggests that as the damage increases, especially after the minimum of AER coefficient, a new stage appears leading to ultimate fracture. For the SiCf/[Si-B-C] composite and the Cf/[Si-B-C] composite, this avalanche phenomenon is certainly linked to the delayed failure of fibres (slow crack growth, oxidation). This characteristic time reflects a local critical behavior described by the Benioff law and it indicates a second damage phase when subcritical crack growth in fibers is predominant. The present work shows the applicability of the Benioff law to describe energy release. Future work will focus on the use of the Benioff law as a predictive model. The present method was applied to CMCs and is applicable to any material as long as AE activity is sufficient to allow a statistical study of the energy received by each sensor.

Figure 1. Evolution of the RAE coefficient during the static load hold

REFERENCES [1] Bufe CG & Varnes DJ. Predictive modelling of the seismic cycle of the greater San Francisco

Bay region. J. geophys. Res. 1993; 98: 9871–9883. [2] O. O. Ochoa and J. N. Reddy, Finite element analysis of composite laminates, Solid Mechanics and its Applications, Vol. 7, Kluwer Academic Publishers, Dordrecht, 1992.

[2] Momon S, Moevus M, Godin N, R’Mili M, Reynaud P, Fantozzi G, Fayolle G. Acoustic emission and lifetime prediction during static fatigue tests on ceramic matrix composite at high temperature under air. Comp: Part A 2010;41:913-18.

[3] Maillet E., Godin N., R'Mili M., Reynaud P., Lamon J., Fantozzi G. Analysis of Acoustic Emission energy release during static fatigue tests at intermediate temperatures on Ceramic Matrix Composites: towards rupture time prediction. Composites Science and Technology, 72(9), 1001-1007 (2012).

132132


Aalborg, 2013

HIGH-SPEED DIC FOR BLAST TESTING OF COMPOSITE PANELS Søren. Giversen1, Christian. Berggreen1 and Benjamin. Riisgaard2

1Department of Civil Engineering, Technical University of Denmark Nils Koppels Allé, Building 403, 2800 Kgs. Lyngby

Email: [email protected], [email protected] , web page: http://www.byg.dtu.dk

2IDT Consultants Europe Amagerbrogade 32, 2300 Copenhagen

Email: [email protected], web page: http://www.idtce.com

Keywords: Blast-testing, Full-field deformation, 3D digital image correlation, transient materiel response, light-weight armor

ABSTRACT

Experiences in Afghanistan and Iraq have shown that land mines and road side bombs such as IEDs (Improvised Explosive Devises) and especially EFPs (Explosively Formed Penetrators) are among the major threats. If military vehicles (and thereby the personnel) are to be protected using add-on panels of armored steel (RHA), they would become too heavy. Either the total allowable vehicle weight will be exceeded or the maneuverability will be unacceptable. There is thus a need to identify and test alternative materials, which are lighter than steel and provide the same or a better protection. Such alternative material could be advanced composite materials [1]. Composites are known to possess greater stiffness to weight and strength to weight ratios compared to steel and it is therefore envisioned that such materials can offer improved performance regarding blast protection. The present study examines the response of composite panels subjected to blast loading. The dynamic structural response of such panels is difficult to monitor because of the high speed and transient nature of the deformations due to the severity of the shock front impact. An effective method to assess the dynamic structural response is by use of high-speed digital cameras in a stereo setup, applying Digital Image Correlation (DIC) on synchronized images of speckled patterned specimens to monitor the dynamic structural response during the loading history. Using image correlation to monitor the structural response, allows for a full field dynamic analysis of the structural deformation, effective strains, strain rates as well as the possibility of monitoring the initial failure mechanisms in case of fracture. To perform the image correlation the commercial software ARAMIS is used.

Figure 1

133133


Figure 2 The test facility consists of a 20ft container, a high speed 3D DIC system and a blast box in which test panels can be mounted, see figure 1. The blast box is constructed of several steel parts that are bolted together and has been designed to allow for testing under various load conditions. The test panels have in-plane dimensions of 700x700mm with a blast exposed area of 500x500mm. The panels are fixed along all edges and the blast box has been dimensioned for blast loads up to 500g TNT and allows for test with a standoff distance up to 300mm. The use of high speed DIC for blast experiments is a well-established test method reported by several authors [1-2]. Many of the performed experiments in the literature have been aimed at plates in various types of metallic material, but the measuring techniques can easily be used for composite plates as well. To obtain valid and accurate results special attention has been given to match speckle pattern with facet size and resolution of the cameras, see figure 2. The speckles applied to the plates should cover between 3-6 pixels [3] and the size of the speckles applied therefore becomes dependent on the field of view and the chosen resolution which again is a consequence of the required frame rate [1, 4]. With the designed setup it is thus possible to monitor the in and out of plane plate deformations and the inplane strain development. Figure 3 gives an example of the measured out of plane deformation.

Figure 3 [1]. P. L. Reu, T. J. Miller . Journal of Strain Analysis 43:673(2008) [2]. Spranghers K, Kakogiannis D, Ndambi JM, Lecompte D, Sol H . EPJ Web of

Conferences 6(2010) [3]. M. A. Sutton Image Corrolation for Shape, Motion and Deformation Measurements

Basic Concepts, Theory and Applications. Springer(2009) [4]. Schmidt T, Tyson J, Galanulis K, Revilock D, Matthew M 26th International Congress

on High-Speed Photography and Photonics 5580:174-185(2004)

134134


Aalborg, 2013

THE EFFECT OF VARIOUS DIFFUSE DAMAGE LEVELS ON THE TRANSVERSE CRACKING EVOLUTION FOR T700/M21 CROSS-PLY LAMINATED COMPOSITES.

Hedi Nouri1, Daniel Traudes1 and Gilles Lubineau1

1COHMAS Laboratory, Division of Physical Science and Engineering, King Abdullah University of Science and Technology (KAUST),

Thuwal 23955-6900, Kingdom of Saudi Arabia e-mail: [email protected] , web page: http://cohmas.kaust.edu.sa/Pages/Home.aspx

e-mail: [email protected] , web page: http://www.kaust.edu.sa/

Keywords: Laminated Composite, Experimental Characterization, Diffuse Damage, Transverse

Cracking.

ABSTRACT The damage evolution and the development of transverse cracking in cross-ply laminates subjected to static uniaxial loading have been extensively studied [1–5]. A reference bibliography model has been developed and implemented for simulating the two-dimensional strain distribution in transverse cracked cross-ply laminates.

A key point is to better understand their complex mechanisms of degradation, from the process to the final failure. This requires the development of multi-scale models than can capture the effect of fiber/matrix debonding, transverse cracking, delamination, and all potential interactions between these damages.

It is well known that laminated composites can exhibit two main mechanisms of degradation, depending on the in-plane loading. Shear loading mainly induces so-called “diffuse” damage, mainly related to fiber/matrix debonding. Transverse loading is responsible for the development of transverse cracks that completely percolate throughout the whole thickness of the single ply. An important material parameter in composite design is the fracture toughness to transverse cracking that is commonly identified with multi cracking testing on cross-ply laminates. The primary objective of the present work was to identify the effect of pre-existing diffuse damage on this fracture toughness. The material studied was a [0/90]s carbon fiber pre-preg T700/M21.

The experimental part of this work was based on three stages. The first stage was the preparation of materials: Laminates plates of carbon fiber pre-preg were laid up in a [0/90]s configuration with dimensions 300 × 300 × 1 mm and cured in compression molding at the manufacturer’s specified single dwell temperature of 180 C. The second stage was to characterize the material in the shear direction, and to study the evolution of damage. Samples of [+45/-45]s were obtained by cutting with a numerical control machine. Static and quasi static tests were performed on the samples. Digital image correlation was used to determine the strain field evolution at the surface during the mechanical tests. The results of this stage were the shear damage evolution versus the applied strain or stress, and the evolution of transverse crack numbers after different levels of damage. In the third stage, we observed the evolution of transverse cracks on [0/90]s pre-damaged samples subjected to tensile stresses of 350, 450, and 550 MPa. This was accomplished using X-ray tomography coupled with an in situ mechanical testing machine. The samples in this stage were cut by water jet from the pre-damaged samples at different levels of shear damage (0, 0.05, 0.10, 0.15, and 0.20). With this microcracking data, the microcracking fracture toughness of the material was computed for each level of diffuse damage. Figure 1 presents X-ray images with die penetrant. The number of transverse cracks for the five chosen damage levels is indicated for the applied loads of 1400, 1800, and 2200N. The number of cracks increases with the level of damage.

135135

Hedi Nouri, Daniel Traudes and Gilles Lubineau

Figure 1: X-ray images with die penetrant. The number of transverse cracks for the five chosen damage levels is indicated for the applied loads of 1400, 1800, and 2200N. The number of cracks

increases with the level of damage.

REFERENCES [1] G. Lubineau, P. Ladeveze, D. Violeau, Durability of CFRP laminates under thermomechanical

loading:A micro–meso damage model, Composites Science and Technology 66 (2006) 983–992.

[2] J. Berthelot, Transverse cracking and delamination in cross-ply glassfiber and carbon-fiber reinforced plastic laminates: static and fatigue loading. APPL. MECH. REV. 2003;56(1):1–37.

[3] P. Ladeveze and E. LeDantec, Damage modeling of the elementary ply for laminated composites. Compos Sci Technol 1992;43(3):257–67.

[4] J. Nairn, Matrix microcracking in composites, in: R. Talreja, J.-A.E. Manson (Eds.), Polymer Matrix Composites, Comprehensive Composite Materials, vol. 2, Elsevier Science, 2000, pp. 403–432.

[5] L. Boniface, P. Smith, M. Bader, A. Rezaifard. Transverse ply cracking in cross-ply cfrp laminates. initiation or propagation controlled. Compos Mater 1997;31(11):1080–112.

136136


Aalborg, 2013

Poster session 2 – Manufacturing/processing & materials characterisation (11 presentations)

137137

138138


Aalborg, 2013

ON THE USE OF DIGITAL IMAGE CORRELATION TO DETERMINE THE PERMEABILITY AND COMPACTION LAW OF FABRICS IN VACUUM INFUSION

PROCESS

Joaquim Vilà1, Carlos González1,2, Javier Llorca1,2

1Instituto Madrileño de Estudios Avanzados de Materiales (IMDEA-Materiales), C/ Eric Kandel, 2 Parque Científico y Tecnológico, 28906 Getafe (Madrid) Spain

Email: [email protected], web page: http://www.imdea.org

2Departamento de Ciencia de Materiales, Universidad Politécnica de Madrid, E.T.S. de Ingenieros de Caminos, 28040 Madrid, Spain

Email: [email protected], web page: http://www.mater.upm.es

Keywords: VIC-3D, VARTM, Permeability, DM

ABSTRACT Vacuum Infusion process is an open mould injection process for manufacturing composite parts with reduced tooling costs in comparison with the standard resin transfer moulding techniques RTM. The procedure is based on the use of the vacuum driving force to infiltrate resin through a vacuum bagged fiber preform. The fact that a mould face is replaced by the vacuum bag, make this process hard to control from the viewpoint of thickness and void content. Moreover, the permeability factor, the basic fabric characteristic controlling the infiltration process, depends on the fabric characteristics as well as the fiber volume fraction or thickness. In this work, the infusion process and the compaction phenomena is analyzed by means of the digital image correlation technique (VIC-3D Correlated Solutions [4]). The deformation of the panel is obtained from the displacement of the vacuum bag measured with VIC-3D [1] and this information is used and post processed to obtain the permeability factors and their dependence with the fiber volume fraction [2]. To this end, the partial differential equations controlling the infusion process are solved with Mathematica [5] software and the permeability parameters -in-plane and through-the-thickness- are iteratively obtained to minimize the error with respect to the experimental compaction results. The partial differential equation for the fluid pressure as a function of spatial coordinates and time ),( trp are based on the continuity and Darcy’s equation as:

tp

v

pv

pK

f

f

tpp

v f

pK

Where fv is the fiber volume fraction, K the permeability tensor and the fluid viscosity. It should be noted that the permeability factor depend on the volume fraction of reinforcement (see Kozeny-Carman [3]) and this in turn to the applied pressure on the fiber bed. This equation is solved for arbitrary values of the permeability factors to minimize the difference with the measured thickness obtained with the VIC-3D technique. A typical experiment measurement with DIC-3D is plotted in Figure 1. E-Glass plain woven preforms (0.5 kg/m2) are infiltrated with corn syrup and the thickness evolution is measured with the DIC-3D technique, plotted at six different and equally spaced positions through the infused panel. In all the plots, the evolution of the thickness is similar and two stages, filling and post-filling, are easily observed. Initially, the thickness of the panel increases in the filling stage due to the load transfer from the fiber bed (supporting the vacuum pressure at the beginning of the experiment) to the corn syrup, until the steady state regime is attained when the fluid reaches the outlet gate. After that point, the inlet gate is closed and the post-filling stage initiated until the vacuum pressure is again re-established and the final thickness of the panel is obtained.

139139

Joaquim Vilà1, Carlos González1,2, Javier Llorca1,2

As it was mentioned previously, the evolution of the panel thickness with time is simulated and the resolution of the partial derivative pressure equations solved with the Mathematica software. The values of the permeability factors are iteratively changed until a good match between the experiments and the simulation is obtained. This procedure has been used to obtain the in-plane and through-the-thickness permeability factors and their dependence with the fiber volume fraction. To this end, two experiments are carried out: the first one is a standard infusion of the textile preform without the use of any distribution media (high permeability fabric used to speed-up the infusion of the resin). In this case, the fluid flow is constrained to be parallel to the plane of the laminate obtaining the in-plane permeability factor. After that, the experiment is repeated with the distribution media (with a priori known in-plane permeability) observing a mixed in-plane&out-of-plane flow which enable to measure the through-the-thickness permeability factor, Figure 2.

Figure 1: Experimental and numerical thickness evolution at six different positions equally spaced

through the infused panel.

Figure 2: Pressure pattern evolution in a representative section of the panel. The effect of the DM is

observed. The flow is initially favoured in the top of the panel until the final saturation is achieved. Max. pressure (Black), Min. pressure (White)

REFERENCES

[1] Q. Govignon, S. Bickerton, J. Morris, P.A. Kelly. Full field monitoring of the resin flow and

laminate properties during the resin infusion process. Composites: Part A 39 (2008) 1412-1426.

[2] Pavel Simacek, Ömer Eksik, Dirk Heider, John W. Gillespie Jr, Suresh Advani. Experimental validation of post-filling flow in vacuum assisted resin transfer molding processes. Composites: Part A 43 (2012) 370–380.

[3] N.C.Correia, F. Robitaille, A.C. Long, C.D. Rudd, P.Simacek, S.G. Advani. Analysis of the

vacuum infusion moulding process: I Analytical formulation. Composites: Part A, 36 (2005) 1645-1656.

[4] Correlated Solutions. www.correlatedsolutions.com [5] © Copyright 1988-2005 Wolfram Research, Inc. http://www.wolfram.com

140140


Aalborg, 2013

FABRIC PERMEABILITY TESTING AND THEIR USE IN INFUSION SIMULATION

Justas Sirtautas1, Anthony K. Pickett2

1,2Institute for Aircraft Design, University of Stuttgart Institut für Flugzeugbau, Pfaffenwaldring 31, 70569 Stuttgart, Germany

Email1: [email protected], web page: http://www.ifb.uni-stuttgart.de

Keywords: Permeability, Infusion, Simulation, Composites

ABSTRACT Manufacture of composites using Liquid Resin Infusion (LRI) methods is usually performed either via Resin Transfer Moulding (RTM) or Vacuum Assisted Resin Infusion (VARI). RTM uses sealed matched metal tooling in which the fabric preform is placed; resin is then pushed under pressure from one, or more, inlet ports toward the outlet ports. The flow is essentially 2D and large distances, possibly requiring large pressures, may be needed for large parts unless a sophisticated system of sequential inlet ports is used. The alternative VARI method overcomes the problem of infusion distance by using a high permeability flow media to distribute resin over one fabric surface; sealing membranes and vacuum then force the resin to flow primarily through the thickness. This method is advantageous for large parts since only low pressure one-sided tooling is required, but does have the disadvantage that fibre volume ratios may be slightly lower and only one smooth surface is possible from the moulding. RTM infusion simulation is relatively straightforward and can be treated as a 2D flow problem using Darcy’s law. However, VARI infusion simulation presents additional problems since flow is 3D and orthotropic permeability data is required. The fabric will also undergo thickness changes, depending on pressure, which will further modify fabric permeability and flow rates. This presentation outlines some ongoing test and simulation work to simulate 3D VARI infusion. The infusion study presented involves a demonstration module for an aircraft structure consisting of four C-Spars molded to an outer skin. For increased stiffness the central portion of the panel uses a sandwich construction; a section cut of the assembly is shown schematically in Fig. 1. Briefly, once flow ports are opened, the resin quickly flows into the central inlet pipe and flow radiates outwards through the surface flow media, Fig 2a (17sec). At 224 seconds it has advanced well into the C-Spars, Fig 2b, and at 311 seconds, Fig 2c, it has flowed through the thickness and started to flow under the sandwich panel.

Figure 1: Schematic cross section of the demonstrator.

141141

Justas Sirtautas and Anthony K. Pickett

(a) 17 sec (b) 224 sec (c) 311 sec

Figure 2: Small demonstrator infusion test views and corresponding filling time.

For infusion simulation new 3D capabilities in the infusion code PAM-RTM1 have been developed as part of a CEC European project INFUCOMP2. In this case 580.000 tetrahedral elements are used for the quarter symmetry model. Important input data determined experimentally include permeability for the flow media, fabrics and also for the fabric-to surfaces contacts; in this case the fabric-core interface was particularly important. Figure 3 shows the experimental setup used to determine permeability values and fit test to simulation results. For example the permeability’s for fabric and ‘fabric-core’ interface were found to be 1.5E-12 m2 and 3.7E-11 m2 respectively, indicating the sensitivity of this parameter.

(a) (b) (c)

Figure 3: a) test setup, b) foam-fabric interface test view and c) resin flow front test and fitted simulation results

Simulation results for the quarter symmetry model are shown in Fig. 4, where a good agreement to test measurement is found.

18 sec 237 sec 317 sec

Figure 4: Demonstrator quarter model: infusion simulation views and filling time.

1 PAM-RTMTM, ESI Software, 99 rue des Solets, SILIC 112, 94513 Rungis Cedex, France (2006). 2 CEC Framework VII project INFUCOMP (Contract 233926)

142142


Aalborg, 2013

SHRINKAGE AND THERMAL EXPANSION MODEL FOR A GLASS/EPOXY LAMINATE

Johnny Jakobsen1, Jens H. Andreasen1, Erik Appel Jensen1 and Ole T. Thomsen2



2Faculty of Engineering and Environment, University of Southampton

Keywords: Thermal Analysis, Property Identification, Shrinkage, Gelation

ABSTRACT Chemical shrinkage together with thermal expansion/contraction during curing of fibre composite materials is known to build up residual stresses. Residual stresses may lead to a knock down on the strength and fatigue properties of the material, as they could lead to premature failure of a part. Residual stresses are built up due to chemical shrinkage of the epoxy matrix and differences in thermal expansion behaviour between the glass fibre reinforcement and epoxy matrix. In addition the architecture of the fibre reinforcement has a great influence on these properties and the magnitude of residual stresses. Lastly the boundary condition’s which may restrict the material from deforming freely may contribute to the residual stress built up. Residual stresses in fibre composite materials and structures have been studied vastly for the past three decades and various approaches have been used to predict residual stresses [1-5]. An often taken approach to model the shrinkage behaviour of a composite is to measure the shrinkage of the neat resin and apply a rule of mixture type model to mimic the behaviour of the composite. The work presented in this abstract relates to the development of a model for predicting shrinkage and thermal expansion of a glass/epoxy composite. Since only the deformational change after gelation is relevant for predicting residual stresses in composite structures the point of gelation is measured. Shrinkage and thermal expansion is measured with modulated thermal mechanical analysis (MTMA) during several imposed temperature cycles (cf. Figure 1).

Figure 1: Dimensional change measurements of a glass/epoxy composite with

MTMA are shown in the figure. The MTMA measurements reveal that the dimensional change behaviour in the beginning of the imposed thermal cycle is highly complex and separation into a reversing and non-reversing

143143

Johnny Jakobsen, Jens H. Andreasen and Erik Appel Jensen

dimensional change illustrates that the complexity arise from competing effects (shrinkage, thermal expansion and crossing a glass transition). Dimensional change measurements of partially cured neat epoxy samples and glass/epoxy samples are used to establish models of chemical shrinkage and thermal expansion in the range from gelation and to fully cured. To construct the empirical models of the shrinkage and expansion it is important to identify the point of gelation. Measurements from two DMA techniques are used. The neat epoxy is analysed in small strain oscillatory shear using parallel plates and for the glass/epoxy samples gelation is identified with dual cantilever beam setup (cf. Figure 2).

Figure 2: Gelation of the epoxy resin is measured with two DMA techniques; a parallel plate torsion setup and a dual cantilever beam setup. Results from the DCB setup are shown in the figure.

REFERENCES [1] Bogetti, T. A. & Gillespie, J. W. (1992), 'Process-Induced Stress and Deformation in

Thick-Section Thermoset Composite Laminates', Journal of Composite Materials 26(5), 626-660.

[1] Hahn, H. & Pagano, N. (1975), 'Curing Stresses in Composite Laminates', Journal of Composite Materials 9(1), 91-106.

[2] Hahn, H. T. (1976), 'Residual Stresses in Polymer Matrix Composite Laminates', Journal of Composite Materials 10(4), 266-278.

[3] Ruiz, E. & Trochu, F. (2005), 'Numerical analysis of cure temperature and internal stresses in thin and thick RTM parts', Composites Part A: Applied Science and Manufacturing 36(6), 806 - 826.

[4] White, S. & Hahn, H. (1992), 'Process Modeling of Composite Materials: Residual Stress Development during Cure. Part I. Model Formulation', Journal of Composite Materials 26(16), 2402-2422.

[5] White, S. & Hahn, H. (1992), 'Process Modeling of Composite Materials: Residual Stress Development during Cure. Part II. Experimental Validation', Journal of Composite Materials 26(16), 2423-2453.

144144


Aalborg, 2013

EXAMINATION OF COMPRESSION AND SHEAR PROPERTIES OF GLASS/CARBON HYBRID LAMINATED COMPOSITES

Simin A. Oshkovr1, Mohsen Rezaei1, Christen M. Markussen2 Tom L. Andersen2, Francis Aviles1,3,

and Ole T. Thomsen1


Email: [email protected] (S.A. Oshkovr), web page: http://www.m-tech.aau.dk

2Department of Wind Energy, Technical University of Denmark P.O. Box 49, Frederiksborgvevj 399, DK-4000 Roskilde, Denmark

Email: [email protected], web page: http://www.dtu.dk

3 Materials Department, Centro de Investigación Científica de Yucatán, Calle 43 No. 130 Col. Chuburná de Hidalgo, Mérida, Yucatán, México

Email: [email protected], web page: http://www.cicy.mx

Keywords: Hybrid composites, Testing, Finite element analysis, DIC.

ABSTRACT Hybrid fibres consist of two or more fibre materials typically woven or stitched together in a fabric in an attempt to tailor the advantages of both fibres while mitigating their limitations. Continuous glass/carbon hybrid fibres are of particular importance for the structural composites community because they are expected to combine the strength and stiffness of the carbon fibres with the ultimate strain, processability, and economy of the glass fibres. This work investigates the compression and shear behaviour of unidirectional laminated composites made of non-crimp glass/carbon fibres and an epoxy resin, through testing and finite element analysis. Particular attention is paid to the influence of the hybrid architecture in the distribution of elastic stress and strain in the composite, which is further examined by digital image correlation. Unidirectional laminated composites made of glass/carbon fibre non-crimp hybrid fabrics and a thermosetting resin are investigated. The hybrid fabric architecture is sketched in Fig. 1, where the thicknesses of the glass/epoxy and carbon/epoxy tows are indicated as hg and hc, respectively. The specimen dimensions, loading and support configuration (L1 to L4, w,d1,d2) comply with those of the corresponding ASTM standard [1]. Laminated composites were fabricated using a standard resin infusion process.

Fig. 1 Schematic of a glass/carbon/epoxy V-notched shear (Iosipescu) hybrid specimen.

145145

Simin Oshkovr, Mohsen Rezaei, Christen Markussen, Tom Andersen, Francis Aviles, and Ole Thomsen

Compression failure of unidirectional composites may be dominant over tension in most structural applications of unidirectional composites, since it is known that the compression strength of these kind of composites is in the order of 50% their tensile strength [2]. Therefore, the compression behaviour of the hybrid composite is of main interest here and is examined by testing and the use of classical micromechanical models. Compression testing will utilize a recently developed compression test fixture which is able to control the ratio of end-to-shear loading, known as the “mechanical combined loading” fixture [3]. It is accepted that the compression failure of unidirectional composites is governed by the matrix shear properties (which acts as an elasto-plastic foundation against fibre micro-buckling) and the fibre misalignment, characterized by a misalignment angle ( [2,4]. There are several micromechanical models to predict the compression strength of unidirectional fibre reinforced composites based on its shear properties [2,4]. The Budiansky’s model, for example, is a micromechanical model which has been successfully used to predict the compressive strength of (non-hybrid) glass and carbon composites [2,4,5]. Assuming a linear elastic-perfectly plastic behaviour in shear, this model predicts the composite’s compression strength ( c) as a function of a statistical measurement of the composite’s shear modulus (G) and its shear yield strength ( y) as,

(1) ( )

Therefore, knowledge of the shear properties of the composite is instrumental to investigate its compression failure. One of the most accepted test methods to measure shear properties of laminated composites is the V-notched (Iosipescu) shear test [6], Fig. 1. However in the case of hybrid composites, the stress and strain distribution upon shear (and compression) loading may not be uniform given the hybrid architecture. For the Iosipescu shear specimen, this non-uniformity may extend beyond the (small) size of the specimen’s shear zone, which would compromise the validity of the test results to determine shear material properties of the composite. Therefore, using finite element and digital image correlation analyses, this work will investigate the stress/strain distribution and failure mechanisms of glass/carbon/epoxy hybrid composites during MCL compression and Iosipescu shear testing, in an attempt to produce recommendations on test specimen configurations that are meaningful to obtain reliable material properties. The authors wish to acknowledge the support of the Danish “Advanced Technology Foundation” under the “Blade King” project.

REFERENCES [1] ASTM standard D5739, Standard test method for shear properties of composite materials by the

V-notched beam method, West Conshohocken, PA, USA, 1998. [2] B. Budiansky, N.A. Fleck, Compressive failure of fibre composites, Journal of the Mechanics

and Physics of Solids, 41, 1993, pp. 183–211 (doi: 10.1016/0022-5096(93)90068-Q). [3] J.I Bech, S. Goutianos, T. L. Andersen, R. K. Torekov and P. Brøndsted, A new static and

fatigue compression test method for composites, Strain, 47, 2011, pp. 21-28 (doi: 10.1111/j.1475-1305.2008.00521.x).

[4] K.K. Kratmann, M.P.F. Sutcliffe, L.T. Lilleheden, R. Pyrz and O.T. Thomsen, A novel image analysis procedure for measuring fibre waviness in unidirectional fibre composites, Composites Science and Technology, 69, 2009, pp. 228-238 (doi: 10.1016/j.compscitech.2008.10.020).

[5] B. Budiansky, Micromechanics, Computer and Structures, 16, 1983, pp. 3–12. [6] D.E. Walrath, D.F. Adams, The Iosipescu shear test as applied to composite materials,

Experimental Mechanics, 23, 1983, pp. 105-110.

y

cG

y

c

1

146146


Aalborg, 2013

EVALUATION OF THE MECHANICAL PROPERTIES OF POLYMER CONCRETES UNDER VARIOUS CONDITIONS

Kyung-Chae Jung1, Ji-Hun Bae1 and Seung-Hwan Chang1*

1School of Mechanical Engineering, Chung-Ang University 221 HukSuk-Dong, Dongjak-Gu, Seoul 156-756, Republic of Korea

Corresponding author: [email protected], web page: prof.cau.ac.kr/~phigs

Keywords: Polymer concrete, Cure monitoring, Strain monitoring, FBG fiber optic sensor, Dielectrometry sensor

ABSTRACT

This paper aims to evaluate the material properties of polymer concretes controlling density, degree of cure, and weight fraction between aggregates and epoxy resin. The density of polymer concretes varied with mixing ratio because the number of internal pores depends on the weight fraction of epoxy resin. The real-time the degree of cure of epoxy resin was measured by a dielectrometry sensor to evaluate the curing time under room temperature. The dissipation factor representing the ratio of dissipation energy was measured by a LCR meter. The dissipation factor varied with phase change of polymer material. This factor decreased as the curing was progressed because the movement of dipoles and ions was diminished [1]. In this study, curing polymer concretes was carried out under room temperature circumstances. The three curing points were selected based on the degree of cure data. The compression test was performed using MTS-810 machine and compression platen. The test was carried out following the standard test procedure described in ASTM C 579-01. The specimens were fabricated under the same circumstances condition in the environment chamber (25 C). The compression speed was 1.25mm/min and the dimension of specimen was 20 20 60mm3. The test was carried out at the three curing points and results were compared one another. The test results are shown in Fig.1.

Figure 1: Degree of cure and Compression test results.

Compressive strengths were measured according to the mass density of the specimens and as a result, those strengths varied with the density and mixing ratio of specimens. To estimate the mechanical strain and the coefficient of thermal expansion of the polymer concretes an FBG fiber optic sensor was embedded in the specimen [2]. The FBG fiber optic sensor was set under the successively increasing temperature environment from -20ºC to 50ºC to measure the wavelength shift data which were used to calculate the coefficient of thermal expansion of the polymer concretes. By manipulating the wavelength changes and other parameters the coefficient of thermal expansion was calculated according to the temperature range. The strain was calculated by Eq.1 with wavelength shift data acquired from the sensor.

147147

Kyung-Chae Jung1, Ji-Hun Bae1 and Seung-Hwan Chang1*

GFWLWL 0/W

(1)

Where , WL, WL0, and FG stand for strain, wavelength shift, reference wavelength, and gage factor respectively. Also, the strain induced by thermal stress on interfacial between the specimen and sensor was measured to calculate the coefficient of thermal expansion of polymer concrete using Eq.2

FB

B

GB

BTePs

B

BB

B

BBBBss )1(

1 (2)

Where Pe, T, and represent the strain-optic coefficient, temperature change, and wavelength shift data respectively. The subscription G means the case that FBG fiber optic sensor was attached on the face of the specimen and F stands for non-attaching case [3]. To compensate the sensor’s self-variation caused by temperature changes, tests were carried out in glued and free cases. Test results are shown in Fig.2.

Figure 2: Wavelength shift of FBG fiber optic sensor varied with temperature variation.

Those data were used to estimate the interfacial behavior of the polymer concrete and cement concrete under a certain temperature environment by using a finite element analysis.

This research was supported by the project of “Development of Construction and Maintenance Technology for Low-Carbon Green Airport Pavements” funded by the Ministry of Land, Transport and Maritime Affairs (MLTM) and the Korea Institute of Construction & Transportation Technology Evaluation and Planning (KICTEP). The research project was conducted under the Center for Green Airport Pavement Technology (CGAPT) of Chung-Ang University. We are very grateful for their strong supports.

REFERENCES

[1] H. S. Kim, S. H. Yoo and S. H. Chang, In situ monitoring of the strain evolution and curing

reaction of composite laminates to reduce the thermal residual stress using FBG sensor and dielectrometry, Composites Part B: Engineering, 44, 2013, pp. 446-452.

[2] K. Kesavan, K. Ravisankar, S. Parivallal, P. Sreeshylam and S. Sridhar, Experimental studies on fiber optic sensors embedded in concrete, Measurement, 43, 2010, pp. 157-163.

[3] Y. L. Lo and H. S. Chuang, Measurement of thermal expansion coefficients using and in-fibre Bragg-grating sensor, Measurement Science and Technology, 9, 1998, pp. 1543-1547.

148148

6th International Conference on Composites Testing and Model Identification Yasmine Abdin, S. V. Lomov, Atul Jain, Ignaas Verpoest, Harry van Lenthe

Aalborg, 2013

GEOMETRICAL CHARACTERIZATION & MICRO-STRUCTURAL MODELLING OF SHORT STEEL FIBER REINFORCED COMPOSITES

Yasmine Abdin1, Stepan V. Lomov1, Atul Jain2, Harry van Lente3, Ignaas Verpoest1,

1Department of Metallurgy and Materials Engineering, KULeuven Kasteelpark Arenberg 44, Bus 2450, BE-3001 Leuven, Belgium & SIM program Nanoforce

Email: [email protected], web page: http://www.mtm.kuleuven.be/Onderzoek/Composites/Composieten

2LMS International

Researchpark ZI, Interleuvenlaan 68, BE-3001 Leuven, Belgium

3Biomechanics Section, KULeuven Celestijnenlaan 300c, Bus 2419, BE-3001 Leuven, Belgium

Keywords: Composites, Random Steel Fibers, Wavy fibers, Computed Tomography, Characterization

ABSTRACT

Short steel fiber reinforced polymers (SSFRP), composed of high strength stainless steel fibers mixed in a polymer matrix, have been recently introduced as an alternative to glass and carbon fiber reinforced composites. Even though density of the steel fiber is much higher, it offers advantages over conventional fibers by its inherent strength and ductility. In addition to outstanding mechanical behavior steel fibers are used for improving shielding properties. Injection molding is a manufacturing technique allowing shaping of complex geometries at low manufacturing cost. Injection molding of steel fibers leads to high waviness of the fibers when embedded in the matrix. Knowledge of the geometry of fibers in terms of fiber length, waviness and orientation is necessary to relate geometrical parameters to the overall mechanical response [1]. Due to waviness and orientation of short steel fibers, two-dimensional imaging techniques for obtaining geometrical parameters cannot be adopted. Micro-Computed Tomography (Micro-CT) is a radiographic, non-destructive and contact-free technique to locate and size volumetric details in three dimensions. The present work attempts to demonstrate the application of micro-CT imaging to the analysis of wavy fiber composites. Steel fiber reinforced polycarbonate plates with 0.05% volume fraction are used for this study. X-ray tomograms are scanned in a high-resolution nano-CT system (Phoenix, Nanotom). The overall scanning resolution (voxel size) is 3 μm. A 3D software is used (Mimics, Materialise, Belgium) for reconstruction and analysis of microCT images. The software is typically used for biomedical applications. A methodology is developed for the characterization of fiber diameter, fiber length distribution (FLD) and fiber orientation distribution (FOD) is presented. The obtained parameters are further used in a micro structural model for generation of representative volume elements (RVEs) of wavy short fiber composites. Figure 1(a) shows the reconstructed 3D model on Mimics software. As shown in Figure 1(b) separate masks can be built allowing the segmentation of single fibers. Using MedCad module, a centerline can be fitted for each individual fiber tracing its profile. Based on the distance between control points along the centerline, the fiber length can be obtained. With post-processing of the centerline data points, the fiber orientation distribution can be obtained (Figure 1(c)).

149149

Y. Abdin, S.V.Lomov, A. Jain, I. Verpoest, H. van Lenthe

(a) (b) (c) Figure 1: Procedure for characterization of fiber length and orientation distribution of SSFRP. (a) 3D reconstructed model in Mimics software (b) segmentation of single fibers, fitting of centerline and

measurement of fiber length (c) post-processing for measurement of fiber orientation. Figure 2 illustrates the obtained fiber length distribution (FLD). It can be shown that the FLD is best presented with a Weibull function. Moreover, it was found that the fiber orientation distribution (FOD) can be presented by aquasi- random 2D orientation tensor.

The obtained parameters are used in a geometrical model for generation of representative volume elements of short wavy fibers as shown in Figure 3.

Figure 2: Fiber length distribution in the steel

fiber reinforced polycarbonate sample. Figure 3: Geometrical model of internal structure

of SSFR: RVE generation.

REFERENCES [1] E. Rezakhaniha, A. Agianniotis, J. T. C. Schrauwen, A. Griffa, D. Sage, C. V. C. Bouten, F. N.

Van de Vosse, M. Under, N. Stergiopulos. Experimental investigation of collagen waviness and orientation in the arterial adventitia using confocal laser scanning microscopy, Biomechanics and Modeling in Mechanobiology, 11, 2012, pp. 461-473.

[3] F. Pfeifer, J. Kastner, R. Freytag. Method for three-dimensional evaluation and visualization of the distribution of fibres in glass-fibre reinforced injection molded parts by μ-X-ray computed tomography, Proceedings of the 17rd Conferene on Nondestructive Testing, Shanghai, China, October 25-29, 2008.

150150

6th International Conference on Composites Testing and Model IdentificationO.T.Thomsen, Bent F.Sørensen and Christian Berggreen (Editors)

Aalborg, 2013

STRAIN GAUGE APPLICATION IN SOFT MATERIAL TESTING

S. Zike and L. P. Mikkelsen

Department of Wind Energy, Technical University of DenmarkP.O. Box 49, Frederiksborgvevj 399, DK-4000 Roskilde, Denmark

Email: [email protected] and [email protected], web page: http://www.dtu.dk

Keywords: Strain gauge, Strain measurements, Composites, Polymer modeling

ABSTRACT

Strain gauges are commonly used for strain measurements; however, their performance depends on several physical and mechanical effects [1-2]. Some of these limitations are associated to the stiffness of the material tested, since larger elastic modulus of the strain gauge causes strain reduction in the softer test sample [3]. Attachment of the strain gauge also includes strain distortions around edges, where the strains are transmitted from the test sample to the gauge [3]. These phenomena are attributed to the effect known as "reinforcement effect" [1]. As a result of the reinforcement effect, strain gauges measure lower strains than the strains experienced by the test sample in the absence of the strain gauge. In figure 1, contour plots of the 3D model show both strain reduction below the gauge and strain distortions around the edges.

Figure 1: Strain fields obtained by the 3D model at = 0.35 % (specimen dimensions 85 x 12 x 10 mm3)

In the present study, correction methods of the gauge factor for a strain gauge are proposed. Gauge factor shows the relation between the relative electrical resistance change of the strain gauge and the strain of the underlying material. It is common that gauge factor is found from a calibration on a relatively stiff material. Nevertheless, the gauge factor will depend on the stiffness of the underlying material, thus ideally the calibration should be done on a similar material as tested. Experimental and numerical methods are used to determine the correction coefficient, C, for test sample with material stiffness ranging from 1 to 200 GPa. In addition, a parameter study of specimen and strain gauge geometrical dimensions is included. Based on a full 3D finite element simulation the design of significantly less stiffness dependent strain gauge for use on thick test samples is proposed. Digital image correlation method was used to observe experimentally strain field distortions in the test samples.

Main conclusions of the study suggest that the correction coefficient of the gauge factor is greatly influenced by the test sample stiffness and thickness, as well as the length of the strain gauge. In figure 2, the 2D model predicted dependency of the correction coefficient on the test sample thickness and stiffness is shown, where C represents the ratio of the elastic modulus determined by strain gauge, Esg, over the

151

S.Zike and L.P.Mikkelsen

elastic modulus of the material without attached strain gauge, Espec. Results indicated that for test sample with tspec = 1 mm and Espec = 1 GPa, the correction coefficient can reach almost 2.2, which corresponds to 120 % error on the strain gauge measurement. It can also be seen that even moderately stiff materials are prone to errors, e.g. if Espec = 20 GPa and tspec = 1 mm then the expected the correction coefficient is 1.055, i.e. strain gauge measurement error is 5.5 %. Furthermore, it is found that reduction of the correction coefficient by thicker specimen is limited, thus strain gauge measurement errors for soft materials cannot be eliminated using thick test samples.

Figure 2: The 2D results of correction coefficient affected by specimen thickness and stiffness, when the specimen attached to strain gauge HBM LY11-10/350

Evaluation of the strain gauge length revealed that shorter strain gauges are prone to larger strainmeasurement errors than the longer ones. This was explained with strain distortions around the gauge ends, which occupy larger area relatively to the gauge length for shorter strain gauges. Thus to reduce the effect of the edge induced strain distortions, the improved design of the strain gauge pattern was presentedby elongating the end-loops representing gauge edges. Improved design of the strain gauge reduced the correction coefficient by 50 % for specimen with Espec = 1 GPa, tspec = 10 mm and attached to strain gauge HBM LY11-10/350. Furthermore, it was observed that the correction coefficient values decrease with increasing plastic deformation of the strain gauge. Digital image correlation method measurements indicated similar observations to those obtained by the 3D simulation model.

Acknowledgements: This research was supported by the Danish Centre for Composite Structure and Materials for Wind Turbines (DCCSM), grant no. 09-067212, from the Danish Strategic Research Council (DSF).

REFERENCES

[1] R.B. Watson, Bonded Electrical Resistance Strain Gages, Handbook of Experimental Solid Mechanics, Springer, 2008. [2] A. Ajovalasit, L. D’Acquisto, S. Fragapane, B. Zuccarello, Stiffness and Reinforcement Effect of Electrical Resistance Strain Gauges, Strain, 43, 2007, pp. 299-305 (doi: 10.1111/j.1475-1305.2007.00354.x). [3] P. Stehlin, Strain distribution in and around strain gauges, The Journal of Strain Analysis for Engineering, 7, 1972, pp. 228-235 (doi: 10.1243/03093247V073228).

152


Aalborg, 2013

METHODOLOGY OF MATERIAL PARAMETERS IDENTYFICATION IN SANDWICH PANELS VERSUS COMPUTER SIMULATION

M. Chuda-Kowalska1, A. Garstecki1

1Institute of Structural Engeenering, Poznan University of Technology pl M. Sklodowskiej-Curie 5, 60-965 Poznan, Poland

Email: [email protected] web page: http://www. ikb.poznan.pl/monika.chuda.kowalska


Keywords: Polyurethane foam core, Material parameter identification, Sandwich panels, Shear test

ABSTRACT Two main objectives of the experimental study of sandwich panels can be distinguished: the first one to provide data for direct use of producers and design engineers and the second one to provide material parameters for advanced computer analyses. In the first case tests are carried out mainly on full scale panels. The interpretation of experimental results and identification of material parameters is often based on Timoshenko beam theory. The advantage of this approach is that it is simple and well complies with the theory used by design engineers. However, this approach does not make possible to analyse some crucial failure mechanisms i.e. debonding, wrinkling of facings, appearance of blisters etc. To analyse these phenomena more advanced computer models must be used, i.e. FEM employing shell elements for the facings. This approach requires material parameter identification basing not only on tests on full scale panels but also on tests using specimens cut out from the panel. In the paper both approaches to experimental testing are discussed. The tests are also simulated in numerical way. Identification of mechanical parameters for typical materials of the core (polyurethane foam, mineral wool, polystyrene, metallic foam) is a complex problem. They usually exhibit greater deviations than the parameters of the cover plate which is made of steel. On the other hand these parameters strongly influence the displacements of the plate and play decisive role in local stability of the compressed facing. Therefore, many papers have appeared recently, which take up various problems of behaviour and identification parameters for many types of materials [1-3]. Because of low density, thermal insulation or cushion properties many applications of sandwich structures uses rigid cellular foams as a core material. Their mechanical properties depend on a lot of parameters which are characterised in detail by Mills [4]. In the paper the main attention is focussed on material parameters of the core. Sandwich panels consisting of thin but relatively rigid external faces and a soft, polyurethane core are analysed. Behaviour of these structures depends strongly on the shear rigidity of the material in the core. A group of methods used in identification of the shear modulus of the core proposed in the literature [5, 6] is based on the bending tests of the panels with measurement of the transverse displacement w. In this method very simple is preparation of samples, testing set-up, interpretation of results and estimation of parameters. The authors observed that it can lead to disadvantageous size effects. Therefore, a new method was proposed and described in [7]. In this method a similar bending test is carried out, but instead of transverse displacement, we directly measure two angles of rotation which appear in Timoshenko beam theory. These angles are measured in the vicinity of a support. The first one is the angle of cross-section rotation, the latter one is the slope of the panel. The shear modulus is calculated directly from the difference between these angles. To complete the study of the shear modulus of the core other tests were also performed, namely tension/compression tests (with and without confinement of transverse displacements), a double-lap shear test and a torsion test of cylindrical samples. The results obtained from numerous tests compared with numerical simulations will be presented on the conference. The results of tests demonstrated small scatter within the same type of test, but evident differences between types of tests. One of the goals of the study is to explain these differences in identified parameters.

153153

M. Chuda-Kowalska, A. Garstecki

The stiffness and strength properties of foam using micromechanical methods was evaluated by Subramanian in [8]. To better understand and interpret the deformation process and failure mechanisms of the PU material microscopic studies are also undertaken. Fig. 1 presents the microscopic photos of the cells before and after the compression test. One can observe permanent, large in micro scale deformation of cells and the failure concentrated at the boundaries of cells. This picture is in good agreement with observations on macro-scale. In the advanced state of loading in compression test specimens deformed non-symmetrically showing large transverse displacements.

Figure 1: Micro observation of cell shape and damage.

REFERENCES

[1] M. Yang, P. Qiao, Quasi-static crushing behavior of aluminium honeycomb materials, Journal of Sandwich Structures & Materials, 10 (2), 2008, pp.133-160.

[2] H. Harders, K. Hupfer, J. Rösler, Influence of cell wall shape and density on the mechanical behaviour of 2D foam structures, Acta Materialia, 53, 2004, pp. 1335-1345.

[3] S. Zang, J.M. Dulieu-Barton, R.K. Fruehmann, A methodology for obtaining material properties of polymeric foam at elevated temperatures, Experimental Mechanics, 52(1), 2012, pp 3-15.

[4] N.J. Mills, Polymer foams handbook. Engineering and Biomechanics Applications and Design Guide, Butterworth – Heinemann, 2007.

[5] R. Juntikka, S. Hallstorm, Shear characterization of sandwich core materials using four-point bending, Journal of Sandwich Structures & Materials, 9 (1), 2007, pp. 67-94.

[6] EN 14509 Self-supporting double skin metal faced insulating panels – Factory made products – Specifications. Warsaw 2007.

[7] M. Chuda-Kowalska, Z. Pozorski, A. Garstecki, Experimental determination of shear rigidity of sandwich panels with soft core, Proc. of 10th International Conference Modern Buildings Materials, Structures and Techniques, Vol I, (Eds. P. Vainiûnas and E.K. Zavadskas), Vilnius, Lithuania, VGTU 2010, pp. 56-63.

[8] N. Subramanian, B.V. Sankar, Evaluation of micromechanical methods to determine stiffness and strength properties of foams, Journal of Sandwich Structures & Materials, 14 (4), 2012, pp. 431-447.

154154


Aalborg, 2013

VALIDATION OF SHORT DYNAMIC SPECIMEN GEOMETRY FOR IDENTIFICATION OF RATE DEPENDENT MODEL ON A LARGE RANGE OF STRAIN RATE

Julien Berthe1,2,3, Mathias Brieu2,3 and Eric Deletombe1

1ONERA - The French Aerospace Lab, F-59045, Lille, France

Email: {julien.berthe;eric.deletombe}@onera.fr, web page: http://www.onera.fr

2Univ Lille Nord de France, F-59000 Lille, France

web page: http://www.univ-lille-nord-de-france.fr/

3ECLille, LML, F-59650 Villeneuve D’Ascq, France

Email: [email protected], web page: http://www.ec-lille.fr

Keywords: Strain rate dependency, Identification, Spectral Viscoelastic model, Specimen geometry

ABSTRACT The identification and/or the validation of CFRP models, particularly for strain rate dependent models, require experimental results at various strain rates. These tests are classically performed on various testing devices, for example creep tests are performed on conventional testing devices with normative specimens and dynamic tests are performed on hydraulic jacks with non-normative specimens, due to a lack of normative procedure for high strain rate tests. Usually, shorter specimens are used to perform dynamic tests in order to adjust the specimen mechanical behaviour to the device capabilities and to reach higher strain rates. These shorter specimens may lead to inconsistencies between high and low strain rate tests, inter alia in the identification of the shear modulus, only because of the modification of the geometry [1]. In this work, an experimental campaign is performed on the T700GC/M21 composite laminate material for various kinds of loading: creep, static and dynamic tests. The objective is to identify a viscoelastic model of the UD ply in a large range of strain rate [2]. In order to avoid the inconsistency between low and high strain rate tests, a validation study of a short geometry for dynamic tests has been performed with controlled strain rate tests and full field measurements with Stereo-Digital Image Correlation system. Strain fields and stress-strain curves are compared for the proposed and normative geometry to validate or reject experimental results from the shorter one. A geometrical criterion for [ 45°] laminate is build up with the results of the validation study to avoid inconsistencies. The validated geometries are used to perform dynamic tests on an hydraulic jack (see Figure 1), as well as static and creep tests on conventional testing devices.

155155

Julien Berthe, Mathias Brieu and Eric Deletombe

Figure 1: Stress-strain curves for dynamic tests at various testing speeds for [ 45°] laminate.

Results of this campaign are used to identify a multi-spectral viscoelastic model of the ply [2]. After identification of the parameters, the model is able to describe the creep, static and dynamic viscoelastic behaviour of the T700GC/M21 UD ply (see Figure 2).

Figure 2: Comparison between model and experiments for dynamic tests on the left hand side, and for a multiple steps creep test on the right hand side for [ 45°] laminate.

The authors gratefully acknowledge the funding from the DGA (French Ministry of Defence) for this PhD work.

REFERENCES

[1] D. Delsart, Composite helicopter structural crashworthiness. Tech. Rep; ONERA/DLR

Cooperation II – First year progress report. ONERA RT 99/52 DMSE/Y; 1999. [2] J. Berthe, M. Brieu, E. Deletombe, Improved viscoelastic model for laminate composite under

static and dynamic loadings. Journal of Composite Materials 2012, In press, DOI:10.1177/0021998312451294

156156


Aalborg, 2013

INTEGRATION OF MICROSTRUCTURED OPTICAL FIBRES INTO CARBON FIBREREINFORCED PLASTIC MATERIALS – DETERMINATION OF THE INITIAL STRAIN

STATE

C. Sonnenfeld1,2, G. Luyckx3, F. Collombet4, Y-H. Grunevald5, B. Douchin4, L. Crouzeix4, M.Torres4, S. Sulejmani1, T. Geernaert1, K. Chah6, P. Mergo7, H. Thienpont1, F. Berghmans1

1Brussels Photonics Team (B-PHOT), Vrije Universiteit BrusselPleinlaan 2, B-1050 Brussel, Belgium

Email: [email protected], web page: http://www.b-phot.org/

2CRISMAT – CNRT-Matériaux / ENSICAEN, UMR 6508, UMS 3318, University of Caen Basse-Normandie, 6, Boulevard Maréchal Juin, F-14050 Caen, France

3Materials Science and Engineering Department, Ghent UniversitySint-Pieternieuwstraat 41, B-9000 Ghent, Belgium

4INSA, UPS, Institut Clément Ader, University of Toulouse133 C, Avenue de Rangueil, F-31077 Toulouse, France

5Composites, Expertise & Solutions4, Rue Georges Vallerey, F-31320 Castanet Tolosan, France

6Electromagnetism and Telecom Department, University of Mons,31, Boulevard Dolez, B-7000 Mons, Belgium

7Department of Optical Fibre Technology, Maria Curie-Sklodowska UniversityPl. Marii Curie-Sklodowskiej, 520-031 Lublin, Poland

Keywords: Residual strain, Cure cycle monitoring, Fibre Bragg grating, Microstructured optical fibre

ABSTRACT

The appearance of excessive residual strains during the manufacturing of composite structures is amajor issue, as this can result in strength reduction, the creation of cracks or even delamination. Themain origin of residual strain is thermal strain arising from the difference in thermal expansioncoefficient between the main composite constituents, namely the matrix material and thereinforcement fibres. Fiber Bragg grating (FBGs) based optical fibre sensors are a powerful tool toperform internal measurements of strain and temperature. Fibre optic sensors appear to haveinteresting features to measure the build-up of residual strains during the cure cycle when embeddedinside the composite part without disturbing the material structure [1]. Research in this field has so farbeen limited to the measurement of a wavelength shift attributed to longitudinal (in-plane) strain [2]and to temperature. An extra grating [3] or another type of sensor is used to compensate the influenceof the temperature variations. All of those techniques require knowing the temperature at the exactlocation of the grating which is rather difficult in large carbon fibre reinforced plastic (CFRP) part andwith high temperature variation. However, during the cure cycle the residual strain consists of thermalstrain induced in all three directions. Fibre optic sensor based methods for determining the transverse(out-of-plane) strain have not been extensively investigated so far.

In this work, we describe how the use of a combination of two types of optical fibre sensors allowsidentifying the residual strains built up during the cure cycle. First we rely on FBGs in a highlybirefringent microstructured optical fibre (MOF) specifically designed to be insensitive to temperatureeffects and to identify transverse strain components with a sensitivity ten times larger than that ofconventional optical fibres [4]. The temperature insensitivity of the sensor stems from the very low

157157

C. Sonnenfeld, G. Luyckx, F. Collombet, Y-H. Grunevald, B. Douchin, L. Crouzeix, M. Torres, S. Sulejmani, T. Geernaert,K. Chah, P. Mergo, H. Thienpont, F. Berghmans

sensitivity of the phase modal birefringence of a MOF to temperature changes [5]. The transversestrain is encoded in the spectral distance between the two reflected resonance wavelengths of theFBG in such a fibre. Second, we use single mode fibres protected from the transverse effects toidentify the longitudinal strain in combination with a system that corrects for the effects of temperaturechanges.

These two types of FBG-based sensors were embedded in a CFRP material. The composite laminatewas produced by the vacuum bag autoclave technique. The lay-up was made of M10/T300 prepregmaterial (Hexcel) with a thickness of 6 mm. The optical fibre sensors were integrated at criticallocations in the composite part in a specific fibre network to be able to monitor the residual straincreation at several locations. The composite structure was also instrumented with severalthermocouples to monitor the temperature in the material at the FBG locations. The entire cure cyclewas monitored by following the Bragg wavelength changes and the wavelength separation of theFBGs. Figure 1 presents the spectral distance versus the cure cycle for the MOF. It evidences a firstdrop of which is linked to the polymerization onset. Moreover it features a large decrease of thepeak separation linked to the build-up of the residual strains during the consolidation phase.Eventually, using the sensor signal of the MOF we will be able to assess the transversal strain in thefinal composite piece while using the sensor signal of the second type of FBGs allows determining thelongitudinal strains as well.

Figure 1: Variation of the temperature and of the peak separation ( ) during the entire curing cycle.

C. Sonnenfeld and T. Geernaert are supported by the Research Foundation-Flanders (FWO-Vlaanderen). The Vrije University Brussel would also like to acknowledge the Methusalem andHercules Foundations Flanders and the Interuniversity Attraction Poles (IAP)-Belgian Science Policy.

REFERENCES

[1] C.M. Lawrence, et al., Determination of process induced residual stress in composite materialsusing embedded fiber optic sensors. Proceedings of SPIE, Smart structures and materials:smart sensing processing and instrumentation (Eds.K.A. Murphy, D.R. Huston), San Diego,March 03, 1997, 3042, pp. 154-65.

[2] M. Mulle, et al., Assessment of cure residual strains through the thickness of carbon–epoxylaminates using FBGs, Part I: Elementary specimen, Composite Part A 40, 2009, pp. 94-104.

[3] Kim, et al., Simultaneous measurement of temperature and strain based on double cladding fiberinterferometer assisted by fiber grating pair. IEEE Photon. Tech. Lett. 20, 2008, 1290-1292.

[4] C. Sonnenfeld, et al., Microstructured optical fiber sensors embedded in a laminate compositefor smart material applications, Sensors 11, 2011, pp. 2566-2579.

[5] S. Sulejmani, et al., Control over the pressure sensitivity of Bragg grating-based sensor in highlybirefringent microstructured optical fibers, IEEE Photonic. Tech. L. 24, 2012, pp. 527-529.

158158

A MICRO-COMPUTED TOMOGRAPHY TECHNIQUE TO STUDY THE INTERACTION BETWEEN THE COMPOSITE MATERIAL AND AN EMBEDDED OPTICAL FIBER SENSOR

G. Chiesura1, G. Luyckx1, N. Lammens1, W. Van Paepegem1, J. Degrieck1

M. Dierick2, L. Van Hoorebeke2

1Ghent University, Department of Material Science and Engineering,

Technologiepark 903, 9052 Gent-Zwijnaarde, Belgium Email: [email protected] web page: http://www.ugent.be/en

2Ghent University, UGCT - Department of Physics and Astronomy,

Proeftuinstraat 86, 9000 Ghent, Belgium

Keywords: μCT, Optical fiber sensor, Ormocer, Automated fiber placement

ABSTRACT

Over the last decade, there is growing interest in condition monitoring of large composite structures. Several industrial applications (e.g. Aerospace, Wind industry, Naval industry, Civil infrastructure) are looking for reliable methods, capable of investigating damage evolution during the entire lifetime of the structures employed. In the Wind Energy (WE) sector, for example, there is a need to decrease the cost of the energy production, and therefore they are searching for ways to optimize the Operation and Maintenance (O&M) phase of their wind turbines. Since the WE market is moving towards Offshore application, the difficulty and thus the cost of O&M is increasing. Among all different sensing techniques, conditions monitoring using Optical Fiber Sensors (OFS) appear to be the most suited, because of their high accuracy (±1 μ ), their immunity to electromagnetic interference and their small intrusive character when embedded in composite materials [1]. Furthermore, OFS technology has also been proven useful as a monitoring tool during composite manufacturing [2]. Although their small intrusive character, still questions are raised on the quality of embedding, the position of the sensor after production, and the to be maintained accuracy of the embedded sensor during the whole life cycle of the composite structure. The present work, therefore, aims to show the potential of micro-computer tomography (μCT) to answer these questions. High-resolution 3D X-ray micro-tomography is a relatively new technique, which allows investigating the internal structure of samples without actually opening or cutting them. The physical parameter, providing the information about the structure, is the X-ray attenuation coefficient μ, which depends on the local composition of the material of the sample and on the energy of the X-rays. Digital radiographs of the sample are made from different orientations by rotating the sample along the scan axis from 0 to 360 degrees [3]. After collecting all the projection data, the reconstruction process is producing 2D horizontal cross-sections of the scanned sample. μCT has some advantages over other non-destructive technology (NDT): e.g. the high scanning resolution (~2 μm, strongly focused) allow you to clearly identify the damaged zones, the possibility to reconstruct a 3D volume of the investigated region makes it easy to interpret, and an important advantage is the possibility to monitor the specimen each time in between two fatigue test cycles without the need for a “post-mortem analysis”. The μCT gives you information on the embedding process itself: the correct placement of your OFS in the embedding process assures you an accurate measurement (correct interpretation of the strain measured), reducing the possibility of having asymmetric stresses on your sensor (premature damage). For example, the layup of the composite plays an important role in the embedding as can be seen in Fig. 1. In this work, we have fatigue cycled several cross ply carbon fibre reinforced plastic (CFRP) laminates and followed up the damage evolution around the embedded OFS; all OFS are Ormocer coated.

159159

G. Chiesura, G. Luyckx, N. Lammens, W. Van Paepegem, J. Degrieck

Figure 1: left, 3D-tomography of three samples [0°,90°]2s carbon pps laminate with in the middle section an embedded optical fibre. Right, the microbending to which an OF is subjected in a fabric carbon/PPS laminate. In Fig. 2 a high resolution μCT – 2D section and a 3D rendering, respectively – are presented. A transverse crack in the proximity of the OFS, as well as the coating surrounding the OFS is clearly visible.

Figure 2: left, 2D cross section taken from a micro-CT of an embedded OFS on a cross-ply CFRP. Right, 3D micro-CT reconstruction of an embedded OFS on a cross-ply CFRP. By using the μCT technique, it was shown that the quality of the embedding of an OFS in a CFRP can be controlled during the whole life cycle of the composite structure beginning at the stage of the production. This allows us to conclude that using μCT, the quality of different embedding techniques and procedures can be evaluated. The overall goal is to define a reliable embedding method able to ensure adequate accuracy and repeatability that may be implemented in industry. This has partially already been achieved with one of our industrial partner, Airborne (NL), through an automated optical fiber placement process (AFP). The authors wish to acknowledge the support of SONACA S.A. (Société Nationale de Construction Aérospatiale SA) and the European Commission for funding the FP7 SmartFiber project.

REFERENCES [1] G. Luyckx, E. Voet, N. Lammens, J. Degrieck, Strain Measurements of Composite Laminates with

Embedded Fibre Bragg Gratings: Criticism and Opportunities for Research, Sensors, 11, 2011, pp. 384-408.

[2] P. Parlevliet, E. Voet, H. Bersee, A. Beukers, Process Monitoring with FBG sensors during vacuum infusion of thick composite laminates, Proceedings of ICCM 16 Conference, Kyoto Japan, 2007.

[3] B.C. Masschaele, V. Cnudde, M. Dierick, P. Jacobs, L. Van Hoorebeke, J. Vlassenbroeck, UGCT: New x-ray radiography and tomography facility, Nuclear Instruments & Methods in Physics Research Section a-Accelerators Spectrometers Detectors and Associated Equipment, 580(1), 2007, pp. 266-269.

160160


Aalborg, 2013

Poster session 3 – Material concepts, modelling and applications (10 presentations)

161161

162162


Aalborg, 2013

A STUDY ON THE LOW-VELOCTY IMPACT CHARACTERISTICS OF IMPACT LIMITER MATERIALS FOR NUCLEAR SPENT FUEL TRANSPORT CASK

J.H. Kim1, K.B. Shin*2 and W.S. Choi3

1Graduate School of Mechanical Design Engineering, Hanbat National University 125, Dongseo-daero, Yuseong-Gu, Daejeon, 305-719, Korea

Email: [email protected], web page: http://www.hanbat.ac.kr

2Department of Mechanical Engineering, Hanbat National University 125, Dongseo-daero, Yuseong-Gu, Daejeon, 305-719, Korea

Email: [email protected], web page: http://www.hanbat.ac.kr

3Division of Fuel Cycle System, Korea Atomic Energy Research Institute 1045, Daedeok-daero, Yuseong-Gu, Daejeon, 305-353, Korea

Email: [email protected], web page: http://www.kaeri.re.kr

Keywords: Balsa wood, Urethane foam, Low-velocity impact, Nuclear spent fuel transport cask

ABSTRACT

A spent nuclear fuel (SNF) transport cask is required by international atomic energy agency (IAEA) safety standard for the transportation of radioactive materials. The SNF transport cask should be without any damaged during transport. For this reason, the nuclear spent fuel cask has been protected using impact limiter made of sandwich panels[1]. The balsa wood, red wood and urethane foam were mainly used as the core materials for impact limiter. The balsa and red wood materials have been applied to KN-18 SNF transport cask as impact absorbing material in Korea. The carbon steel has been used as impact limiter of AGN1 SNF transport cask in Italy[2,3]. However, the impact limiter of new developing SNF transport cask should be designed in accordance with the required impact absorbing capacity and size. Thus, the mechanical and impact characteristics according to the core materials applied to impact limiter is evaluated and certified. In this study, the mechanical properties and low-velocity impact test was conducted for balsa wood and urethane foam considered as the core materials for impact limiter of new developing SNF transport cask. Also, it was done for sandwich panels with steel facesheets. For urethane foam, three categories of tensile, compressive and shear mechanical test were conducted. For balsa wood, nine mechanical properties were measured by the reason of orthotropic property having the different materials properties in different orthogonal directions. All mechanical tests were conducted according to the ASTM standards. In order to evaluate the impact characteristics of balsa wood, urethane foam core and their sandwich panels, the low-velocity impact test was conducted for impact energy levels of 1J, 3J and 5J. The low-velocity impact test was performed using Instron DYNATUP 8250 drop weight impact tester. The experimental results showed that both of urethane foam and balsa wood except growth direction (z-direction) had similar impact responses. Figure 1 shows that the impact test instrument and specimens. Also, in order to verify the proposed finite element model for impact limiter of SNF transport cask, low-velocity impact analysis of core materials and their sandwich panels were carried out using explicit finite element analysis code LS-DYNA 3D. The low-velocity impact analysis results were

163163

J.H. Kim, K.B. Shin and W.S. Choi

compared with experimental results for damage areas and contact force-time history. The numerical results showed that low-velocity impact responses of wood and urethane foam materials and their sandwich panels were in a good agreement between numerical and experimental results. Figure 2 shows the comparisons of experimental and FE analysis results for impact test of sandwich panels.

(a) Test instrument (b) specimens

Figure 1: The impact test instrument and specimens.

Figure 2: The comparisons of experimental and FE analysis results for impact test.

REFERENCES [1] J. H. Ku, K. S. Seo, H. Y. Kang and Y. J. Kim, A FEM Analysis of the Dynamic Behaviour of

Spent Nuclear Fuel Transport Cask under Oblique Drop Impact, Transactions of the KSME, 69, 1995, pp. 3252-3259.

[2] K. S. Kim, J. S. Kim, K. S. Choi, T. M. Shin and D. Y. Yun, Dynamic impact characteristics of KN-18 SNF transport cask – Part 2: Sensitivity analysis of modelling and design parameters, Annals of Nuclear Energy, 37, 2010, pp. 560-571 (doi: 10.1016/j.anucene.2009.12.024).

[3] D. Aquaro, N. Zaccari, M. Di Prinzio and G. Forasassi, Numerical and experimental analysis of the impact of a nuclear spent fuel cask, Nuclear Engineering and Design, 240, 2010, pp. 706-712 (doi: 10.1016/j.nucengdes.2009.12.018).

164164


Aalborg, 2013

POST-FIRE MECHANICAL PROPERTIES OF MARINE SANDWICH COMPOSITES

Luan TRANVAN, Vincent LEGRAND*, Pascal CASARI, Frédéric JACQUEMIN

LUNAM Université - Université de Nantes - Ecole Centrale Nantes Institut de Recherche en Génie Civil et Mécanique (GeM) UMR CNRS 6183

Equipe Etat Mécanique et Microstructure 37 boulevard de l'Université - BP 406 44602 Saint-Nazaire cedex – France

e-mail : [email protected], web page: http://www.gem.ec-nantes.fr

Keywords: Sandwich composite, Balsa core, Fire damage, Mechanical properties.

ABSTRACT Composite materials are used in marine industry for the building of a wild variety of boats. These materials are in general constituted with thin polymeric skins (polyester, vinylester or epoxy) reinforced with fibres (glass, carbon or Kevlar) surrounding a core made by an ultra-light material (balsa or polyurethane foam). However, the use of this kind of material in marine industry requires precaution, in particular concerning fire damage. Indeed, sandwich composites are highly inflammable, badly resist to heating and emit toxic stuff during combustion. Then, these materials are subject to a severe control and it is mandatory to well know their thermo-mechanical properties before every application [1-3].

Figure 1: The ATLAS cone calorimeter and detail about

a sandwich composite sample during fire testing. In this work, we focus on the analysis of the thermo-mechanical properties of sandwich composite materials (polyester skins with E-glass fibres and balsa core). As the present application is for marine industry, we studied dry samples but also hygroscopicaly aged (to saturation) samples. Thanks to a cone calorimeter (fig.1), fire damages were checked on both materials (dry and wet) to determine the combustion kinetic (fig. 2). Fire-tested composite experiments were performed at 750°C and for various ignition times. Thus, sandwich composites were mapped during combustion in terms of structural variations (delamination, cracking…) and of mechanical properties changes in particular in flexion deformation (fig.3). Cone calorimeter and 3-points flexion measurements allowed to highlight the exponential behaviour of the slide of the dry and wet sandwich composites mechanical properties. Mechanical damages are observed during the first 150 s of ignition time when cracking growth and delamination appears progressively from 200 s. These results were completed with thermogravimetric analysis to better

165165


understand the structure-properties relationships and to finely detail the thermal behaviour of balsa core inside the global composite material. Activation energies of the dry and wet balsa samples were notably determined.

Figure 2: Dry sandwich composite, cone calorimeter: fire damage kinetic at 750°C and determination

of the combustion speeds for the separated component elements.

Figure 3: Wet sandwich composite, 3-points flexion: Young modulus variation as a function of the ignition time. The fire-tested sample profiles at 750°C are reported for ignition times up to 2500 s.

REFERENCES [1] A. P. Mouritz and Z. Mathys, Post-fire mechanical properties of marine composites, Composite

Structure, Vol. 47, 1999, pp. 643-653. [2] A. P. Mouritz and C. P. Gardiner, Compression properties of fire-damaged polymer sandwich

composites, Composites: Part A, Vol. 33, 2002, pp. 609-620. [3] S. Feih, Z. Mathys, G. Mathys, A. G. Gibson, M. Robinson and A. P. Mouritz, Influence of

water content on failure of phenolic composites in fire, Polymer Degradation and Stability, Vol. 93, 2008, pp. 376-382.

166166

DEVELOPMENT OF AUTOMATED FINITE ELEMENT MODELS FOR LARGE WIND TURBINE BLADES

Mathijs Peeters1 and Wim Van Paepegem1

1 Department of Mechanics of Materials and Structures, Ghent University Technologiepark-Zwijnaarde 903, 9052 Zwijnaarde, Belgium

Email: [email protected] Email: [email protected], web page: http://www.composites.ugent.be/

Keywords: Wind turbine blades, Finite elements, Automation

ABSTRACT

The design of wind turbine blades to date is done using design codes employing advanced analytic models. While these models are very good for performing many iteration steps in the design process, little is known about the exact stress distribution within the blade under different loads and conditions. Such conditions may include transportation, power production, emergency stops, etc. For this purpose a finite element model can be used. However, such a model is only as good as it is detailed. A real life wind turbine blade is typically built from sandwich materials, consisting of a foam or balsa core and glass or carbon fibre epoxy laminate on the top and bottom. Such a blade has a composite layup consisting of a large number of plies. As a consequence, adding all the detail of each ply to a finite element model can be a rigorous process. In this work, a finite element model of a real life 50 metre long glass fibre epoxy composite wind turbine blade has been developed. The model employs shell elements with top offset positioned on the outer mould layer. For this purpose, a python script was built to automate the partitioning and layup creation process in the finite element code AbaqusTM. Based on an excel sheet, each ply is automatically added to the model, starting and ending at the exact specified millimetre. The use of shell elements results in a stepwise thickness transition of the laminate. This aligns with reality, since the local laminate thickness is the result of a discrete number of layers. As a consequence, each ply drop-off is accurately represented in the model.

Figure 1: Rendered thickness of the shell elements at the root of the blade. Notice the smooth,

stepwise thickness transition. A modal analysis was conducted on the obtained model. The resulting Eigen frequencies were in agreement with those predicted in the design. The result of a computational fluid dynamics analysis was also applied to the model, under the form of a distributed pressure across the blade’s surface.

167167

Mathijs Peeters and Wim Van Paepegem

Design (Hz)

Simulation at 14.1

rpm (Hz) 1st

Eigen mode

0.74-0.91 0.76

2nd Eigen mode

1.0-1.35 1.21

Figure 2: from top to bottom:1) the first Eigen mode corresponding to a flap wise Eigen shape, 2) the second Eigen mode corresponding to an edgewise Eigen shape, 3) the first torsional Eigen mode.

Because the generation of the model is automated, the model is parametric and can hence be used for optimization purposes. This possibility was demonstrated using a genetic algorithm, using the tip displacement under a uniform surface traction load as fitness function.

Figure 3: (left): FEM in the Abaqus/CAE pre-processor, each yellow plane performs an intersection.

(Right): cross section of the blade mesh.

An important influence on the stiffness and the general response of the blade are the adhesive bonds. These are present between the shear webs and spar caps and at the leading and trailing edges. As these undergo a complex loading and shear effects are very large, solid elements should be employed to model the adhesive. The bonds are modelled using a multi-layer orphan mesh, which extrudes the shell elements of the laminate on the top and bottom of the shear webs which is connected to the adhesive. On each of the bonds, a tie constraint is applied to the outer surfaces to constrain them to the blade’s shell or shear web. An option is used to set the distance between the outer surface of the adhesive layer and the blade’s shell to zero before the simulation starts. This makes it possible to obtain a smooth, well connected adhesive.

In an effort to expand the flexibility of this type of model, to explore the possibility of other element types, such as continuum shell and solid elements and to make it possible to include other complex details of real life blades, such as the adhesive bonds at the leading and trailing edges, a script is under development which directly generates the mesh for a finite element model of a wind turbine blade starting from aerofoil data files.

168168


Aalborg, 2013

BEND-TWIST COUPLING IDENTIFICATION IN COMPOSITE BEAMS

Vladimir Fedorov1 and Christian Berggreen2

1Department of Wind Energy, Technical University of DenmarkP.O. Box 49, Frederiksborgvevj 399, DK-4000 Roskilde, Denmark

Email: [email protected], web page: http://vindenergi.dtu.dk

2Department of Civil Engineering, Technical University of DenmarkBrovej, Building 118, DK-2800 Kgs. Lyngby, Denmark

Email: [email protected], web page: http://byg.dtu.dk

Keywords: Bend-twist coupling, Composite beam, FEM, DIC

ABSTRACT

The present study is aimed at identification and quantitative evaluation of bend-twist coupling effects in composite beams. The study motivation was to incorporate coupling effects in modern long wind turbine blades [1] which at a certain point lead to a simpler problem of bend-twist coupling identification in small-scale uniform composite beams.

Currently Euler-Bernoulli beam formulation was used for describing bend-twist coupling effects in a generic beam. Consider a beam that can take on only a torque and a bending moment in one principle direction. Then, the relation between the generalized torque and moment acting at a beam cross-section and the cross-section bending ( ) and twist ( ) deformations can be written as: = (1)

According to the above formulation, a bend-twist coupling coefficient was introduced [2]: = (2)

The novelty of the method is in evaluation of bending stiffness and torsional stiffness together with the coupling coefficient by obtaining both bending and twist responses of a beam in two load cases: Beam is loaded by a tip bending moment and by a tip torque. This can be done numerically, bydeveloping 3-D beam finite element (FE) model of high detail (using shell or solid elements), or experimentally, using full-field digital image correlation (DIC) measurements. Thus, it is also possible to validate the beam finite element models of high detail against experimental results in terms of Euler-Bernoulli formulation.

While application of tip bending moment to a beam in a FE model can be done with little efforts, accurate implementation of this load case experimentally is a challenge and it was addressed in the present study. A load application system with three hydraulic axes was developed for accurate application of torque and bending moment by mean of two actuators running simultaneously. The entire test setup was built based on a four-column testing machine with the tested beam specimensmounted vertically, Fig. 1.

Measurements on the beam deformations were done based on the full 3-D deformation field obtained numerically or measured by DIC system, Fig. 1. Several approaches to setup and configure a DIC system for measurements on beam specimens are presently addressed. Calculations of cross-section bending displacements, rotations and twist angles along the beam specimen were done based on theassumption of all the beam cross-sections remain rigid. Thus, homogeneous rigid-body transformation,

169169

Vladimir Fedorov and Christian Berggreen

that is three translations and three rotations, was considered and an algorithm [3] based on the least square fit method was used. Beam cross-section deformations and according to eq. (1) were calculated using the obtained cross-section displacement and rotation distributions along the tested beam length.

Two load cases were of a particular interest of this study: Pure bending (a bending moment is appliedat the beam tip) and pure torsion (only a tip torque is applied). Beam bending and twist responses in these load cases were obtained as and for the bending load case and and for the torsion load case. By solving the system of four equations (1), evaluations on the beam bending stiffness ,torsional stiffness and bend-twist coupling coefficient were finally obtained for the numerical models and experiments.

Small-scale bend-twist coupled composite beams made of glass-fiber reinforced plastics were tested experimentally and modeled using FE. Straight walls open (I-) and closed (box-) cross-sections were selected for the beam specimens, with the beam flanges containing fibers in the only direction. An example of bend-twist coupling coefficient for box-beams with different fiber directions (0°, 15° and 25°) in the uni-directional flanges is shown in Fig. 2.

Figure 1: Test setup and DIC measurements. Figure 2: Coupling coefficients for box-beams.

REFERENCES

[1] V. Fedorov, N. Dimitrov, C. Berggreen, S. Krenk, K. Branner and P. Berring “Investigation of structural behavior due to bend-twist couplings in wind turbine blades”. Proceedings of the 17th

International Conference of Composite Materials (ICCM), Edinburgh, UK, 2009.[2] D. Lobitz and P. Veers “Aeroelastic behavior of twist-coupled HAWT blades”. Proceedings of

the AIAA/ASME Wind Energy Symposium, Reno, Nevada, 1998.[3] K.S. Arun, T.S. Huang and S.D. Blostein, Least-squares fitting of two 3-D point sets, IEEE

Transactions on Pattern Analysis and Machine Intelligence, 9, 1987, pp. 698-700.

0 200 400 600-0.1

0

0.1

0.2

0.3

0.4

0.5

Beam length coordinate, mm

Ben

d-tw

ist c

oupl

ing

coef

ficie

nt

0 UD

15 UD

25 UD

170170


Aalborg, 2013

A NEW DAMAGE TOLERANT DESIGN APPROACH FOR SANDWICH PANELS LOADED IN FATIGUE

Georgios Martakos1, J.H.Andreasen2 and O.T.Thomsen 3



2Email: [email protected], web page: http://www.m-tech.aau.dk

3Email: [email protected], web page: http://www.m-tech.aau.dk

Keywords: Composite sandwich plates, fracture mechanics, fatigue, damage tolerance

ABSTRACT A new experimental procedure for crack propagation in composite sandwich structures under fatigue loading conditions is proposed and used to evaluate the performance of a damage tolerant design approach [1]. The investigation concerns crack propagation in all 3 dimensions and thus, composite sandwich panels are utilized as specimens. Few investigations have been conducted on crack propagation inside a sandwich panel [2] as the use of beam specimens is often more convenient and results are easier to assess. Sandwich structures, though, are commonly used in large panels and less often as beam elements making failure inside them a multidimensional problem. Testing sandwich panels instead of beams to evaluate the performance of the damage tolerant approach in fatigue appears to be a more resourceful experimental investigation. The panel specimens consist of glass fiber laminates while the core structure includes the main core foam material, the peel stopper and the insert, (Fig. 1). The panels are mounted in a specially designed rig that restricts displacement in the lateral direction and imposes simple support boundary conditions on all four edges of the plate, (Fig. 1).

Figure 1: Experimental set-up for testing sandwich panels and cut view of the core structure. The structure is loaded in fatigue and an initial debond is propagating outwards, towards the support of the panels. A Circular peel stopper is built into the panels core structure. The peel stopper confines a circular area in which the initial debond should propagate freely but not exceed its limits, “spreading” into the rest of the panel. A thin circular Teflon layer is included at the center of the panel in the lower interface in order to induce an initial debond in the structure. The fatigue loading conditions are induced close to the maximum load limit of the panel, 80% of the maximum load. The limit refers to the maximum load needed to propagate the initial debond and is determined quasistaticaly. The experimental investigation aims at evaluating the overall effect of the peel stopper on the fatigue life of the specimens. To achieve this goal, panels that do not contain peel

171171

Georgios Martakos, Jens H.Andreasen and Ole T.Thomsen

stoppers are also tested for comparison reasons. Results from first experiments showed that the peel stopper is able to extend the fatigue life of the specimens. The experimental procedure is designed such that axisymmetric behavior can be approximated around the area of the peel stopper. A cylindrical steel insert in the middle of the plate is utilized to apply the load, a circular initial debond is introduced on the structure and a circular peel stopper is used to confine the propagating crack. By doing so, the initial debond propagates in a circular shaped crack front and “strikes” the peel stopper at the same time. Axisymmetric behavior is evaluated through a series of numerical analyses by comparing displacement and stresses responses of an axisymmetric model with a full rectangular plate model (Fig. 2). The analysis indicates that a certain radius exists around the center of the panel on which the deviation of the two responses does not exceed a small percent. The peel stopper is well inside this radius.

Figure 2: Transverse shear stress resultant comparison in the core material. The comparison is made for the axisymmetric model (right and up) and the 3D plate model (right and down) in the diagonal

and one of the main axes (X-axis). For the purpose of assessing the results of the experiment, the compliance of the specimen was obtained. The compliance is derived and expressed as a function of the crack propagation radius, R while the specimen retains its axisymmetric behavior. Using the compliance of the test specimen, the energy release rate can be obtained for the propagating crack [3-4]. Monitoring energy release rate can be essential for assessing the results of crack propagation in fatigue and can provide insight for the performance of damage tolerant design. Finally, the compliance of the structure also relates the vertical displacement of the steel insert, measured by the test machine, with the crack propagation length (radius, R) inside the panel. An approximation can thus be made for the position of the crack while fatigue testing is running.

REFERENCES

[1] J. Jakobsen, E. Bozhevolnaya, T. Thomsen New peel stopper concept for sandwich structures, Composites Science and Technology, (2007), Vol. 67 (15-16), pp. 3378-3385.

[2] R. Moslemean, C.Berggreen, A.A. Karlsson Face/Core debond propagation in sandwich panels under cyclic loading Part-II experimental valiadation, Book of abstracts 10th International Conference On Sandwich Structures ICSS 2012 (Eds. P. Casari), Nantes, France, 27-29 August 2012, Universire de Nantes, pp. 43-44.

[3] Avilés F, Carlsson LA. Analysis of the sandwich DCB specimens for debond characterization. Engineering Fracture Mechanics,75, 2008, pp.153–68.

[4] Quispitupa A, Berggreen C, Carlsson LA. On the analysis of a mixed mode bending (MMB) sandwich specimen for debond fracture characterization. Engineering Fracture Mechanics,76, 2009, pp.594–613

172172


Aalborg, 2013

CHANGES IN MECHANICAL BEHAVIOUR OF A GLASS FIBRE REINFORCED EPOXY BY ADDING POLYAMIDE 6 NAN-OFIBRES

I. De Baere1, B. De Schoenmaker2, S. Van der Heijden2, W. Van Paepegem1 and K. De Clerck2

1 Department of Materials Science, Ghent University Technologiepark-Zwijnaarde 903, 9052 Zijwnaarde, Belgium

Email: [email protected], web page: http://www.composites.ugent.be

2 Department of Textiles, Ghent University Technologiepark-Zwijnaarde 907, 9052 Zwijnaarde, Belgium

Email: [email protected], web page: http://www.ugent.be/ea/textiles/en

Keywords: Glass-epoxy, Nano-fibres, Mechanical properties, Tensile test, Damage behaviour

ABSTRACT

Owing to their light weight and high stiffness and strength, fibre-reinforced epoxy resin composites are widely used in industry. However, an epoxy matrix is a brittle material, so an improvement of the interlaminar space would be interesting. Therefore, secondary (sub-)micron reinforcements are often incorporated in the matrix. Since it is difficult to obtain a homogeneous dispersion of these nano-particles with common techniques, the mechanical improvement of the composites is only moderated. Thermoplastic nano-fibrous structures can tackle this dispersion issue, as they can be inserted by other means. Therefore, this study investigated the effect of polyamide 6 nano-fibrous structures on the mechanical properties of a glass fibre/epoxy composite. The nano-fibres are produced using a multi-nozzle electrospinning set-up, using an in house developed technology [1]. These nano-fibres are then incorporated in the glass fibre/epoxy composite either as stand-alone interlayered structures (noted NF-I) or directly spun on the glass fibre reinforcement (noted NF-C). By doing so, the dispersion issue is taken care of. The composite plates for this study were manufactured by vacuum assisted resin transfer moulding (VARTM) using a closed steel mould. Both ways of nano-fibre incorporation have no negative effect on the impregnation of the epoxy. Fig. 1 (a) illustrates the cross section of the interlayered structures, whereas Fig. 1 (b) illustrates the nano-fibre rich epoxy. Compared to a standard VARTM glass/epoxy plate, it can be noted that the interlayer has a non-negligible thickness and the same can be said concerning the deposited version, although the effect is lesser in this case. Given the fact that all plates were 3 mm in thickness because of the closed mould, the thickness of the glass fibre reinforcement will be smaller when thicker interlayers are present. Also, different transfer of shear loads between neighbouring glass reinforcement is expected, compared to the basic plate without reinforcement.

(a) Cross section at larger scale (b) the nano-fibre containing interlaminar space

Figure 1 SEM observations of the nano-fibres in the glass/epoxy composite.

173173

I. De Baere, B. De Schoenmaker, S. Van der Heijden, W. Van Paepegem and K. De Clerck

To determine the influence of the effects of the added nano-structures, tensile tests were conducted on [0/90°]2s and on [±45°]2s stacking sequences and this for both the interlayered, the deposited nano-fibres and for the base plate. For the [0/90°]2s, the effects on the Young’s modulus, the Poisson’s ratio throughout the test and the ultimate tensile strength were assessed. For the [±45°]2s, both the in-plane shear behaviour as the Poisson’s ratio was observed. With respect to Young’s Modulus Exx, the differences between the three types of composite were so small that they are most likely caused by scatter in the production process. For the evolution of the Poisson’s ratio xy, however, a significant and reproducible difference was seen, especially for larger stress levels, meaning when damage was already present in the material. The evolution of xy as function of the strain was highest for the deposited nano-fibres and the lowest for the interlayered plate. With respect to the failure strength, for the basic glass/epoxy, an average value of 550 MPa was found, compared to 581 MPa for the interlayered and 611MPa for the deposited nano-fibres. Therefore, an increase in strength is present when adding the polyamide nano-fibres. To examine how the nano-fibres contributed to this increase in strength, the polished surfaces of the failed specimens where examined. Fig. 2 illustrates the crack growth in the 90° layers. For the standard glass-epoxy composites, the transverse cracks shift into a delamination upon hitting the 0°layer (Fig. 2 (a)). However, for the deposited nano-fibres, it seems that the crack is stopped in the deposited region and does not shift into a delamination (Fig. 2(b)). Similar effects were seen over the entire specimen and also for the interlayered nano-fibres.

(a) benchmark glass-epoxy (b) Deposited nano-fibres Figure 2: Microscopical investigation of the failed specimens

For the [±45°]2s stacking sequence, although there is not much difference between the interlayered and the basic glass/epoxy, the influence of the deposited nano-fibres cannot be neglected. For a given shear strain, the corresponding shear stress is significantly higher for the deposited version. With respect to the shear strain, both nano-reinforced versions show a value of 4.7 GPa, compared to 4.0 GPa for the basic plate. For the Poisson’s ratio, now both the nano-reinforced plates showed a lower evolution of

xy as function of the strain xx, meaning less narrowing of the specimen, with the lowest values for the interlayered version.

In conclusion, the polyamide 6 nano-fibres have an influence on the evolution of some mechanical properties, especially with growing damage. This influence is most likely due to their capacity of preventing transverse cracks from growing into delaminations and by changes in microstructure they impose by their non-negligible thickness.

REFERENCES

[1] P. Westbroek, T. Van Camp, S. De Vrieze, K. De Clerck, PCT/EP2008/056050, 2008

174174


Aalborg, 2013

PSEUDO-GRAIN DISCRETIZATION AND FULL MORI TANAKA FORMULATION FOR RANDOM HETEROGENEOUS MEDIA: PREDICTIVE ABILITIES FOR STRESSES IN INDIVIDUAL INCLUSIONS AND THE MATRIX

Atul Jain1,2, Stepan V. Lomov2, Yasmine Abdin2, Ignace Verpoest2, Wim Van Paepegem3

1LMS International Researchpark Z1

Interleuvenlaan 68 B-3001 Leuven, Belgium

Email: [email protected] web page: http://www.lmsintl.com/

2Department of Metallurgy and Materials Engineering, KULeuven Kasteelpark Arenberg 44 - bus 2450

B-3001 Heverlee Belgium

Email: [email protected], web page: http://www.mtm.kuleuven.be/English/

3 Department of Material Science and Engineering Ghent University

Gent Belgium

Keywords: Short fiber composites, mean field homogenization, Mori-Tanaka formulation, Finite

elements

ABSTRACT For modelling damage in short fibre composites, both the predictions of the effective properties and the stresses in the individual inclusions and in the matrix are necessary. Mean field theorems are usually used to calculate the effective properties of composite materials, most common among them is the Mori-Tanaka formulation [1]. Owing to occasional mathematical and physical admissibility problems with the Mori-Tanaka formulation, a pseudo-grain discretized Mori-Tanaka formulation (PGMT) was proposed [2]. This paper compares predictive capabilities for stresses in individual inclusions and matrix as well as the average stresses in the inclusion phase for full Mori-Tanaka formulation and PGMT. The predictions of average stresses inside inclusions and the matrix by both Mori-Tanaka formulation and PGMT are compared to solutions of full-scale FE models for a wide range of configurations. A short fiber composite consisting of glass fiber inclusions and polyamide matrix was considered for all calculations.

Figure 1 Finite element model of RVE containing 30 inclusions having uniform random distribution of inclusions and a volume fraction of 0.25. Notice that the structure is periodic and

inclusions intersecting a face of cube also appear on the opposite face.

175175


A range for representative volume elements (RVE) was analysed both by Mori-Tanaka formulation and PGMT. RVE considered had various volume fractions of inclusions from 0.01% to 0.25%, from fully aligned inclusions to various orientation distributions and length distribution.

Figure 2 Inclusion average stresses in the global loading direction, random orientation of inclusions , applied load is 1% uniaxial strain , vf = 0.1, with orientation tensor (a11=0.51, a22=0.49); Fig 2a. Average of stresses in axial direction, S11. Fig 2b. Average of stress in

transverse to loading direction S22.

Mori-Tanaka formulation gave good predictions of average stresses in individual inclusions, even when the basic assumptions of Mori-Tanaka were reported to be too simplistic [3], while the predictions of PGMT were off significantly in all the cases. However, the predictions of the matrix stresses by the two methods were found to be very similar to each other in all the cases considered. A Beneveniste [4] type interpretation of the PGMT will be provided to explain the discrepancies in prediction of stresses in individual inclusions by PGMT. The average value of stress averaged over the entire inclusion phase was also very close to each other. It is thus expected that the effective properties predicted by both methods will be similar. Conclusion Mori-Tanaka method is the first choice homogenization scheme especially when the stresses in individual inclusions are important for further analysis and modelling.

REFERENCES

[1] Mori, T. and K. Tanaka, Average Stress in Matrix and Average Elastic Energy of

Materials with Misfitting Inclusions. Acta Metallurgica, 1973. 21(5): p. 571-574. [2] Doghri, I. and A. Ouaar, Homogenization of two-phase elasto-plastic composite

materials and structures - Study of tangent operators, cyclic plasticity and numerical algorithms. International Journal of Solids and Structures, 2003. 40(7): p. 1681-1712.

[3] Benveniste, Y., G.J. Dvorak, and T. Chen, On Diagonal and Elastic Symmetry of the Approximate Effective Stiffness Tensor of Heterogeneous Media. Journal of the Mechanics and Physics of Solids, 1991. 39(7): p. 927-946.

[4] Benveniste, Y., A new approach to the application of mori-tanaka theory in composite-materials. Mechanics of Materials, 1987. 6(2): p. 147-157.

176176


Aalborg, 2013

SMA/GFRP COMPOSITE PLATES: PASSIVE DAMPING AND INTERFACE STRENGTH

M. Bocciolone1, M. Carnevale1, A. Collina1, N. Lecis1, A. Lo Conte1, B. Previtali1 C.A. Biffi2, P.Bassani2, A. Tuissi2

1Politecnico di Milano, Department of Mechanical Engineering,

Via La Masa, 1, 20156 Milano, Italy. Email: antonietta.loconte, web page: https://www.polimi.it

2National Research Council CNR, Institute for Energetics and Interphases, Corso Promessi Sposi 29, 23900 Lecco, Italy.

Email: [email protected], web page: http://www.ieni.cnr.it

Keywords: Shape Memory Alloy, Hybrid laminated composite, Structural damping, Interface adhesion, Fatigue.

ABSTRACT

Over the past few years, due to the increasing technological demand for the passive suppression of mechanical vibrations, there has been a growing interest in developing high damping materials. In particular, this paper focuses on light, slender structures in which the Glass Fiber Reinforced Plastic is used as structural material, requiring, however, further improvement in terms of damping performance. To enhance the damping of a Glass Fiber Reinforced Polymer (GFRP) plate through passive vibration suppression, a shape memory alloy (SMA) can either be bonded or integrated into the part. Shapes such as fibers, ribbons and films have been the object of numerous research and development projects and have confirmed the feasibility of improving damping performance by using SMA [1]. Unlike GFRP, SMA alloys have a higher storage modulus and a higher specific damping [2]. Due to its higher storage modulus and, assuming that the interface guarantees the proper load transfer, the SMA material is capable of storing more specific elastic energy than the GFRP and, as consequence, is able to take maximum advantage of its higher specific damping in order to enhance the structural damping of the hybrid composite. Therefore the partial substitution of the GFRP material with SMA material in various shapes and forms is expected to significantly increase the damping properties of the hybrid composite. On the other hand, when during the fabrication process the hybrid composite is cooled to room temperature, high residual stresses are expected in SMA elements, due to the mismatch between the thermal expansion coefficients of the SMA material and of the GFRP laminated composite. As consequence the pullout of SMA fibers or the delamination of ribbons and films from the GFRP laminated composite can occur easily. Though seemingly promising, another architecture using SMA textiles or yarns as smart fibers, is not free from interfacial failures. Recent papers [3] report studies on a GFRP composite filled with SMA short fibers or particles. In these discontinuous SMA composites, the residual stresses are dispersed due to the random distribution of SMA fibers in the GFRP laminated composite. However, short fibers or particles have a low aspect ratio, to the point where this composite shows a low degree of load transfer between the SMA reinforcement and the GFRP laminated composite. In recent papers [2, 4] the authors have shown the design optimization and fabrication of a hybrid composite material, in the form of beam, made from GFRP laminates and reinforced with two thin sheets of shape memory alloy (Figure 1). The two SMA sheets are embedded below the upper and lower surface of the beam. The thin SMA sheets are laser-patterned in order to improve adhesion between the SMA sheets and the GFRP laminated composite, to avoid the delamination of the hybrid composite and to maximize the load transfer between the GFRP laminated composite and the SMA reinforcements. This new concept of a hybrid composite has original traits with respect to the solutions proposed in literature. The hybrid composite proposed has to be characterised, at temperature below martensite

177177

M. Bocciolone, M. Carnevale, A. Collina, N. Lecis, A. Lo Conte, B. Previtali, C.A. Biffi, P.Bassani, A. Tuissi

finish temperature, by higher damping properties than those of traditional GRFP laminates. This increase is not, however, obtained to the detriment of the stiffness and weight of the component, which do not change significantly. Consequently, natural frequencies of the component are preserved. While not affecting none of the natural frequencies of the beam, the solution chosen can enhances its structural damping by as much as 40%. In this paper the experience gained in our previous papers will be used to design a SMA/GFRP hybrid composite in the shape of a plate with optimized structural damping with respect to its first vibrational mode. The new hybrid composite plate will be initially investigated by means of numerical simulation. In particular, the thickness of the SMA layers and their circular pattern geometry will be optimized by numerical calculation of the natural frequencies and the related loss factor of the hybrid composite. In order to evaluate the damping factor numerically, thermo-mechanical characterization of the SMA alloy are available. A solution combining the thickness and pattern geometry of the SMA inserts will be proposed. A prototype of the proposed hybrid composite plate will be manufactured in order to perform dynamic experimental test for the measurement of the structural damping in the intended range of small amplitudes. Particular attention will also be paid to the adhesion between SMA inserts and host composite. The strength of the interface will be tested statically, by means of pullout tests on small samples of SMA sheets partially embedded in the GFRP, and for fatigue load by means of preliminary fatigue tests at different load amplitude using other three samples of the hybrid composite plate.

Layers of fiber glass/epoxy resin [-45/+45]2

Layers of fiber glass/epoxy resin [-45/+45]2 gla

Thin sheet of SMA

Bulk of fiber glass/epoxy resin [-45/+45]11

1

3

+45° +-45° 2

Figure 1: Architecture of the hybrid composite beam.

Figure 2: Geometry of the SMA/GFRP plate with similar architecture of beam of Figure 1.

REFERENCES

[1] J.Van Humbeek, S.Kustov, Active and Passive Damping of Noise and Vibrations through Shape

Memory Alloys : Applications and Mechanisms, Smart Mater. Struct., 2005, 14, 171-185. [2] C.A. Biffi, P. Bassani, A. Tuissi, M. Carnevale, N. Lecis, A. Lo Conte, B. Previtali, Flexural

Vibration Suppression of Glass Fiber/CuZnAl SMA Composite, Functional Materials Letters, Vol. 5, No. 1 (2012) 1250014 (4 pages).

[3] Q. Q. Ni, R. Zang, T. Natsuki, M. Iwamoto, Stiffness and Vibration Characteristics of SMA/ER3 Composites with Shape Memory Alloy Short Fibers, Composites Structures, 2007, 79, 501-507.

[4] P. Bassani, C.A. Biffi, M. Carnevale, N. Lecis, B. Previtali, A. Lo Conte, Passive damping of slender and light structures, Materials and Design 45 (2013) 88–95.

[+45/-45]n GFRP host composite Embedded sheets of patterned SMA

178178


Aalborg, 2013

MECHANICAL AND MORPHOLOGICAL PROPERTIES OF TALC FILLED HIGH DENSITY POLYETHYLENE

Adib Kalantar Mehrjerdi and Mikael Skrifvars

School of Engineering, University of Borås, SE-501 90 Borås, Sweden Correspondence to: [email protected], web page: http://www.hb.se

Keywords: Composites, High-density polyethylene, Talc, Thermo-physical properties, Mechanical properties

ABSTRACT By adding particulate fillers to polymers, the properties can be improved. Many particles have been used, such as metals, carbon, glass fiber, and ceramic, each of which can give different qualities and properties for the intended end use. Carbon black (CB) has been used as an economical additive in many thermoplastics and thermosets compounds; it is also a very effective additive for improvement of the outdoor stability of plastics. It has been shown that for high density polyethylene-carbon black (HDPE/CB) composites, even at a level of 0.05 wt-% of CB, the composite has very good UV-screening strength and this can be enhanced further by adding up to 5 wt-% of CB.1 The drawback is, however, that carbon black can reduce the mechanical properties of HDPE at higher loadings.2 Talc is a mineral composed of hydrated magnesium silicate arranged in three disc-shape layers. In the middle, there is a layer of magnesium-oxygen/hydroxyl octahedra, while the two outer layers are composed of silicon-oxygen tetrahedra. These layers are kept together only by van der Waals’ forces, and the layers have the ability to slip over each other easily, which makes talc the softest known mineral, measured as 1 on the Mohs hardness scale. Due to its unique characteristics such as softness, chemical inertness, and its rather low price, talc has been used for many years as attractive filler in a wide range of industries. In recent years, there has been interest in investigating the effect of talc in polypropylene (PP) blends—not only as filler because of financial considerations, but also due to some of its functional properties. On the other hand, the most commonly used plastics, such as polyethylene (PE) and PP, are considered to be thermal insulators with low thermal conductivity. There are many new applications such as electronic packaging, pipe networks, heat exchangers, and domestic appliances, in which an increase in the heat transfer properties would be an advantage. In this study, HDPE precompounded with 2.5 wt-% CB was blended in a compounder with up to 35 wt-% talc loadings. Specimens were then made by injection moulding for testing of the thermo-physical and mechanical properties. The mechanical properties of the composites were studied by tensile testing and by impact testing. The thermal conductivity, thermal diffusivity, and specific heat were evaluated by the transient plane source (TPS) method, and the thermal stability of blends was examined by thermal gravimetrical analysis (TGA). The specific density and morphology were also measured and analyzed. The aim was to evaluate the effect of talc on the studied properties, in order to find the most optimal composition. The results indicated that CB causes a significant decrease in the toughness, while talc not only enhances the thermal conductivity and thermo-physical properties of the composites but can also play a role in compensating for the negative effects of CB on impact resistance. The experimental data showed that the presence of CB reduces the impact resistance of HDPE by up to 34%, while addition of up to 8 wt-% talc can return this value to close to that of pure HDPE. Carbon black proved to be an effective additive for enhancement of thermal stability, while it had a negative effect on the mechanical properties, particularly impact resistance. We have emphasized that due to its plate-like shape and evident aspect ratio, talc is a promising particulate to enhance the mechanical and thermo-physical characteristics of PE. The improvement in toughness perpendicular to the direction of flow

179179

was more pronounced, while the tensile at break and yield tensile remained unchanged when increasing talc addition. The thermal conductivity, the thermal diffusivity, and the specific density of the composites were enhanced, the specific heat capacity of the composites decreased, which can increase production speed. We are grateful to the Muovitech International Group in Borås, Sweden, and Västra Götalandsregionen, FoU-kort programme number RUN-625-0295-12 in Sweden for the financial support for this research study.

REFERENCES 1. S. Bigger and O. Delatycki, Journal of Materials Science, 24, 1946-1952 (1989). 2. J. Z. Liang and Q. Q. Yang, Journal of Reinforced Plastics and Composites, 28, 295-304 (2009).

180180


Aalborg, 2013

MIXED MODE DELAMINATION IN HYBRID LAMINATE UNDER DMMB TEST

Ji-Woong Kang1, Oh-Heon Kwon2 and Jung-Hoon Kwak2

1Faculty of Health Science, Daegu Haany University 1, Haanydae-ro, Gyeongsan, Gyeongsanbukdo, 712-715 Korea

Email: [email protected], web page: http://www.dhu.ac.kr

2Department of Safety Engineering, Pukyong National University 365, Sinseon-ro, Nam-gu, Busan, 608-739 Korea

Email: [email protected], web page: http://www.pknu.ac.kr

Keywords: Hybrid laminate composite, Delamination, Mixed mode bending, Fracture toughness

ABSTRACT

Recently, a wind power energy system has been developed actively among the renewable energy which is a solution for the global energy problem. A rotor blade is the most part in the wind power system because it revolves and has high weight. The box spar and tail parts are composed of the CFRP and GFRP hybrid laminate composites for the lightweight of the blade. However, CFRP/GFRP hybrid laminates have often damage as like the delamination condition and cracks at the interface of laminates. Due to the delamination or the interfacial crack tip behaviour at the hybrid materials, fracture occurs under mixed mode conditions, especially mode I and mode II. Therefore, there is a need for the evaluation of the mixed mode during the delamination of CFRP/GFRP hybrid laminate interface. This paper shows the results of an experimental examination of the delamination fracture toughness in a CFRP/GFRP hybrid laminate composites.

Fracture toughness experiments and estimation are performed by using DMMB(Dissimilar mixed mode bending) specimen. The materials used in the test are a commercial woven type CFRP (Carbon fiber reinforced plastic) prepreg(CF3327) and UD type GFRP(Glass fiber reinforced plastic) prepeg(HD224A). A CFRP/GFRP hybrid laminate composite are composed by CFRP(10plies)/GFRP(10plies) laminate and CFRP(10plies)/GFRP(7plies) laminate. The thickness of CFRP(10plies) and GFRP(10plies) layer is 2.5mm and 3.0mm. The thickness of CFRP(10plies) and GFRP(7plies) layer is 2.5mm and 2.5mm, respectively. Also the fulcrum location which is the loading point distance from a crack end opening point is a loading parameter. It is changed from 30 to 50mm on the specimen of length 120mm because it defines the ratio of mode I to mode II. In this study, the effects of the fulcrum location as a loading parameter are evaluated in the viewpoint of energy release rate in mode I and mode II contribution. The fracture experiments were carried out in a small scale universal servo-hydraulic machine(H5KS). And the crack advanced length was recorded with a travel telescope jointed by a stereo microscope.

The evaluation of the mode I and II energy release rate were conducted by the equations from the references[1,2]. The total strain energy release rate is given by Eq. (1) and mode I and II contributions are thus given by Eqs. (2,3).

IIG

IGG GG (1)

32h21E

131h11E

1(2W

2χh)(a2I6P

IG

χχ(2)

)2h21E1h11(E2h

132h21E

β31h11E

α(2W

2a 2II6P

IIG

E(3)

181181

Ji-Woong Kang, Oh-Heon Kwon and Jung-Hoon Kwak

Table 1 shows mechanical properties for plain woven CFRP and Unidirectionl GFRP prepreg The crack extension as the delamination length were measured. These results show that the delamination crack initiate at the lower displacement and load according to the increasing of the fulcrum location ratio. The variation of the strain energy release rate for mode I and mode II contributions are shown in Figure 1. The total energy release rate fracture toughness Gc are obtained as 20.94, 8.98, 4.57 and 1.74 J/mm2 for the fulcrum ratio 0.3, 0.35, 0.4 and 05, respectively. And energy release rates are almost constant according to the crack extension. Also mode mixities GII/GI are obtained from 0.35 to 0.22 Thus delamination crack behaviour on the hybrid laminate composite is mainly effected by mode I fracture even though the dissimilar materials with slightly different thickness of CFRP and GFRP laminate layers under mixed mode bending.

Engineering constant

Unit CFRP(Woven) GFRP(UD)

E1 [GPa] 54.3 43.312 – 0.10 0.31

G12 [GPa] 28.39 -u [GPa) 1.527 1.102

Thickness [mm] 2.5 3

Table 1: The mechanical properties of a plain woven CFRP and UD GFRP prepeg.

0

2

4

6

8

10

12

14

16

18

20

0 2 4 6 8 10 12 14 16 18

Ene

rgy

rele

ase

rate

, G

I&

GII

(J/m

m2 )

Crack extension length, Δa(mm)

C30-I

C30-II

C35-I

C35-II

C40-I

C40-II

C50-I

Figure 1: The energy release rate to the crack extension according to a fulcrum point.

This work was supported by the Pukyong National University Research Fund in 2011(PK-2011-2-28).

REFERENCES

[1] W.O.Soboyejo, G.Y.Lu, S. Chengalva, J.Zhang and V.Kenner, A modified mixed-mode bending specimen for the interfacial fracture testing of dissimilar materials, Fatigue Fracture Engineering Materials structure, 22, 1999, pp. 799-

[2] G.V.Marannano and A.Pasta, An analysis of interface delamination mechanism in orthotropic and hybrid fiber metal composite laminates, Engineering Fracture mechanics, 74, 2007, pp. 612-626.

182182


Aalborg, 2013

AUTHOR INDEX

A Abdin, Y. ............................................. 149, 175

Allegri, G. .................................................... 121

Alton, C. ........................................................ 45

Amacher, R. ................................................... 33

Andersen, T.L. ............................................. 145

Andreasen, J.H. ................................... 143, 171

Anglada, M. ................................................... 97

Arteiro, A. ....................................................... 3

Auguste, E. .................................................... 15

Avilés, F ................................................ 87, 145

Ayadi, Z. ........................................................ 13

B Bae, J-H. ...................................................... 147

Baere, I. De ............................................ 43, 173

Barrio, A. ..................................................... 101

Bassani, P. ................................................... 177

Battley, M. ..................................................... 83

Baumann, J. ................................................... 61

Bech, J.I. ...................................................... 107

Bennati, S. ..................................................... 41

Berggreen, C. ........................... 71, 81, 133, 169

Berghmans, F. ............................................. 157

Bernasconi, A. ............................................. 123

Berthe, J. ...................................................... 155

Biffi, C.A. .................................................... 177

Bismarck, A. ................................................ 109

Bocciolone, M. ............................................ 177

Botsis, J. ............................................ 33, 39, 91

Boyd, S.W. .................................................... 37

Branner, K. .................................................... 81

Brieu, M. ...................................................... 155

Brunner, A.J. .................................................. 73

Brøndsted, P. .................................................. 65

Budhe, S. ...................................................... 111

C Camanho, P.P. .................................................. 3

Canal, L.P. ............................................... 91, 95

Carraro, P. ...................................................... 55

Casari, P. ...................................................... 165

Catalanotti, G. .................................................. 3

Chah, K. ....................................................... 157

Chang, S-H. ................................................. 147

Carnevale, M. ............................................... 177

Chiesura, G. ........................................... 89, 159

Choi, W.S. .................................................... 163

Chripunow, A. ............................................... 79

Christiansen, J.deC. ....................................... 31

Chuda-Kowalska, M. ................................... 153

Clerck, K.D. ................................................. 173

Collina, A. .................................................... 177

Collombet, F. ............................................... 157

Conte, A.L. .................................................. 177

Costa, J. ............................................ 53, 63, 111

Crammond, G. ............................................... 37

Crouzeix, L. ................................................. 157

Cugnoni, J. ....................................... 33, 39, 103

Czel, G. ........................................................ 109

D Daggumati, S. ................................................ 43

Dattoma, V. .................................................. 125

Degrieck, J. ...................................... 43, 89, 159

183183

Deletombe, E. .............................................. 155

Devivier, C. ................................................... 69

Diao, H. ....................................................... 109

Dierick, M. .................................................. 159

Doroudgarian, N. ........................................... 97

Douchin, B. ................................................. 157

Drozdov, A.D. ............................................... 31

Dulieu-Barton, J.M. ................................. 37, 71

E Enfedaque, A. ................................................ 45

Eriksen, R. ..................................................... 71

F Fantozzi, G. ......................................... 129, 131

Farmand-Ashtiani, E. .................................... 39

Fedorov, V. .................................................. 169

Feindler, N. .................................................... 61

Fisicaro, P. ..................................................... 41

Forghani, A. ................................................. 113

Fuhr, J.-P. ...................................................... 61

G Garstecki, A. ................................................ 153

Geernaert, T. ................................................ 157

Giannis, S. ................................................... 127

Gibson, G. ....................................................... 5

Giversen, S. ................................................. 133

Godin, N. ............................................. 129, 131

Gmür, T. ...................................................... 103

Gorbatikh, L. ................................................. 29

González, C. .............................. 21, 45, 95, 139

Goutianos, S. ................................................. 65

Grunevald, Y-H. .......................................... 157

Gude, M. ........................................................ 99

Guzman, E. .................................................. 103

H Hallett, S.R. ................................... 59, 113, 121

Härtel, F. ........................................................ 61

Heijden, S.V.D. ............................................ 173

Hochard, C. .................................................... 43

Hoorebeke, L.V. .......................................... 159

Høgh, J. .......................................................... 81

I

J Jacobsen, T.K. .................................................. 7

Jakobsen, J. .................................................. 143

Jacquemin, F. ............................................... 165

Jain, A. ................................................. 149, 175

Jamil, A. ....................................................... 123

Jeenjitkaew, C. ............................................. 127

Jensen, E.A. ................................................. 143

Joffe, R. .................................................... 27, 97

Jones, M.I. ...................................................... 59

Jung, K-C. .................................................... 147

K Kang, J.-W. .................................................. 181

Katunin, A. ................................................... 119

Kawai, M. ...................................................... 49

Kim, J.H. ...................................................... 163

Klitkou, R. ..................................................... 31

Knoll, J.B. ...................................................... 99

Koch, I. .......................................................... 99

Koschichow, R. .............................................. 99

Ku-Herrera, JdeJ. ........................................... 87

Kuiken, J.J: .................................................... 51

Kwak, J.-H. .................................................. 181

Kwon, O.-H. ................................................ 181

L Laffan, M. J. ................................................... 19

Lahuerta, F. .................................................... 51

Lammens, N. .......................................... 89, 159

Lander, J.K. .................................................. 121

Laurin, F. ....................................................... 15

Lecis, N. ....................................................... 177

184184


Aalborg, 2013

Legrand, V. .................................................. 165

Lente, H.V. .................................................. 149

Limberger, H.G. ............................................ 91

Linde, P. ........................................................ 63

Llorca, J. .................................... 21, 45, 95, 139

Lomov, S.V. ...................... 25, 29, 43, 149, 175

Loukil, M.S. .................................................. 13

Lubineau, G. ................................................ 135

Luyckx, G. ..................................... 89, 157, 159

M Mahdi, S. ....................................................... 63

Manshadi, B.D. .............................................. 91

Markussen, C.M. ......................................... 145

Martakos, G. ................................................ 171

Mavel, A. ....................................................... 15

Melro, A.R. ...................................................... 3

Mergo, P. ..................................................... 157

May-Pat, A .................................................... 87

Mayugo, J. A. ................................................ 53

Mehrjerdi, A.K. ........................................... 179

Mestra, A. ...................................................... 97

Michaud, V. ................................................... 91

Middendorf, P. ............................................... 61

Mikkelsen, L.P. ................................... 107, 151

R’Mili, M. ........................................... 129, 131

Molina-Aldareguia, J.M. ............................... 45

Muñoz, R. ...................................................... 21

N Nasri, N. ........................................................ 39

Nijssen, R.P.L. ............................................... 51

Nobile, R. .................................................... 125

Nouri, H. ...................................................... 135

O Oshkovr, S.A. .............................................. 145

P Paepegem, W. Van ........ 43, 159, 167, 173, 175

Panella, F.W. ................................................ 125

Peeters, M. ................................................... 167

Pickett, A.K. ................................................ 141

Pierron, F. ...................................................... 69

Pimenta, S. ............................................. 17, 109

Pinho, S.T. ............................................... 17, 19

Poursartip, A. ............................................... 113

Previtali, B. .................................................. 177

Pupure, L. ....................................................... 27

Pupurs, A. ...................................................... 65

Q Quaresimin, P. ............................................... 55

R Raijmaekers, S. .............................................. 51

Rask. M. ......................................................... 75

Rasmussen, S. ................................................ 75

Renart, J. ........................................ 53, 101, 111

Rezaei, M. .................................................... 145

Reynaud, P. .......................................... 129, 131

Riisgaard, B. ................................................ 133

Robinson, P. ........................................... 19, 109

Rodríguez-Bellido, A. .................................. 111

Romanov, V.S. ............................................... 29

Ruder, M. ....................................................... 79

S Sans, D. .......................................................... 53

Schoenmaker, B.D. ...................................... 173

Schulte, K. ..................................................... 99

Segurado, J. .................................................... 95

Shin, K.B. .................................................... 163

Sirtautas, J. ................................................... 141

Sket, F. ........................................................... 45

Skrifvars, M. ................................................ 179

Sonnenfeld, C. ............................................. 157

Stang, H. ........................................................ 81

Stubbing, R.M. ............................................... 83

185185

Sulejmani, S. ............................................... 157

Sørensen, B.F. ............................................... 75

T Tabatabaei, S.A. ............................................ 25

Thienpont, H. ............................................... 157

Thomsen, O:T: ............................ 143, 145, 171

Toftegaard, H. ............................................... 75

Torres, M. .................................................... 157

Tranvan, L. .................................................. 165

Traudes, D. .................................................. 135

Turon, A. ....................................................... 63

Tuissi, A. ..................................................... 177

U Urresti, I. ..................................................... 101

V Valvo, P.S. ..................................................... 41

Varna, J. ............................................. 13, 27, 65

Vaziri, R. ..................................................... 113

Verpoest, I. ........................ 25, 29, 43, 149, 175

Vilà, J. ......................................................... 139

Violakis, G. ................................................... 91

Voet, E. .......................................................... 89

W Waldbjørn, J. ................................................. 81

Westphal, T. .................................................. 51

Wisnom, M.R. ......................... 59, 69, 109, 113

Wittrup-Schmidt, J. ....................................... 81

X Xu, J. ............................................................. 43

Xu, X ........................................................... 113

Y Yano, K-i. ...................................................... 49

Yasaee, M. ................................................... 121

Z Zike, S. ........................................................ 151

Zobeiry, N. .................................................. 113

Zubillaga, L. .......................................... 63, 101

186186

comptest 2013 - book of abstracts · emission, optical fiber sensors) open a very broad field of...

Documents