Impact and Damping Behaviour of Composite Adhesive Joints

IMPACT AND DAMPING BEHAVIOUR OF COMPOSITE ADHESIVE JOINTS Submitted by Helder André Moreira Araújo MSc THESIS Supervised by: Lucas F. M. da Silva Co-supervised by: Eduardo A. S. Marques José J. M. Machado Mestrado Integrado em Engenharia Mecânica July 2016 Impact and Damping Behaviour of Composite Adhesive Joints Helder André Moreira Araújo Mestrado Integrado em Engenharia Mecânica July 2016 II III First of all, I would like to express my gratitude to Prof. Lucas da Silva, my supervisor, for the opportunity to collaborate with him in this work. I would like to thank Eduardo Marques and José Machado, my co-supervisors, for the time spent, and the complete availability during the development and execution of this dissertation. I would like to thank my colleagues Miguel and Pedro for the support given during the execution of the thesis. Also a special word for all the members of FEUP Adhesives Group that followed my work during these months. I would like to thank also to Prof. Raul Campilho for his technical help, working with CFRP and also working with Abaqus® software. Special thanks to Mr. Miguel Figueiredo from Laboratório de Ensaios Tecnológicos (LET) for his support and availability while performing the impact tests. My appreciation for Prof. José Dias Rodrigues for his availability and help, during the dynamic tests. Lastly and the most important, I would like to express my gratitude to my family and friends, especially my parents and girlfriend, for the support, encouragement and patience given during this long journey. IV The automotive industry has significantly increased the use of adhesive joints in structural automotive components, which has coincided with increase in the use of composite materials. The combined use of composites and adhesive bonding allows the production of structures with very high mechanical strength and reduced weight. Adhesive joints are especially suitable for this purpose, as the use of rivets and screws inherently damages the components by causing stress concentrations, which are especially damaging for composite materials. However, as one of the main automotive industry requirements is the safety of the occupants in the event of a collision, these adhesive joints must be able to sustain large impact loads, transmitting the load to the structure and without damaging the joint. This work aims to further explore the impact behaviour of composite adhesive joints. For this purpose, a characterization of the behaviour of single lap joints (SLJ) was performed, mainly for a ductile adhesive, and under quasi-static and impact conditions. The performance of experimental tests in these two conditions allowed to establish the comparison between behaviours. To allow a better interpretation of the SLJ results, a characterization of the adhesives’ and substrate properties was necessary. To accomplish this, tensile tests of the adhesives under different test speeds were performed. The shear properties were obtained from thick adherend shear test (TAST), performed under different test speeds. Finally, double cantilever beam (DCB) and End-Notched Flexure (ENF) were performed to determine the fracture toughness in mode I and II, respectively. DCB and ENF were performed for the ductile adhesive and carbon fibre reinforced polymer (CFRP), under different test speeds. Dynamic tests were also performed using vibration analysis to assess and compare the damping capabilities of the studied joints. To support and interpret all the experimental research, numerical models were developed with cohesive elements in Abaqus® software. These numerical simulations included both static and dynamic models. The experimental and numerical results, allowed to draw conclusions regarding the influence of the overlap length in the SLJs behaviour, under quasi-static and impact conditions. A considerable increase in the joint strength was noticed in impact tests. The numerical results presented are in accordance with the experimental results. V VI A indústria automóvel aumentou significativamente o uso de ligações adesivas, coincidindo com um uso mais extenso de materiais compósito. O uso simultâneo de ligações adesivas e materiais compósitos permite a construção de estruturas com elevada resistência mecânica e baixo peso. As juntas adesivas são especialmente adequadas para uso em materiais compósito, uma vez que o uso de técnicas de ligação mecânica, como rebites e parafusos, causa concentrações de tensão nos compósitos e reduz a resistência das estruturas. Uma vez que um dos principais focos da indústria automóvel é a segurança, estas juntas adesivas de materiais compósitos deverão ser capazes de suportar fortes cargas de impacto, transmitindo as forças para a estrutura sem quebrar a ligação prematuramente. Este trabalho procurou então explorar o comportamento ao impacto de juntas adesivas com substratos de material compósito. Para tal, foi executada uma caracterização do comportamento de juntas de sobreposição simples, principalmente usando um adesivo dúctil, em condições quasi-estáticas e de impacto. A realização dos testes nestas duas condições, permitiu a estabelecer uma comparação de comportamentos. Com o intuito de melhor interpretar os resultados das juntas de sobreposição simples, a caracterização dos adesivos e do substrato foi considerada necessária. Posto isto, foram realizados testes de tração do adesivo a diferentes velocidades de teste para obter as propriedades de tração. Foram ainda obtidas as propriedades ao corte, através de testes TAST (thick adherend shear test) a diferentes velocidades de teste. Finalmente, testes DCB (double cantilever beam) e ENF (end notched flexure) foram realizados para determinar a tenacidade à fratura em modo I e II, respetivamente. Os testes DCB e ENF testes foram realizados quer para o adesivo dúctil quer para os compósitos, a diferentes velocidades de teste. Adicionalmente, foram também realizados testes de vibração para caracterizar as características de amortecimento das juntas testadas. Modelos numéricos foram desenvolvidos com elementos coesivos no software Abaqus®. Estes modelos, que incluem simulações estáticas e dinâmicas, permitiram interpretar os resultados experimentais. Nos resultados obtidos foi possível aferir a influência dos diversos comprimentos de sobreposição no comportamento das juntas quer em condições quasi-estaticas, quer em condições de impacto. Um aumento considerável na resistência da junta foi registada nos ensaios ao impacto. Os resultados numéricos apresentados estão em concordância com os resultados experimentais. XIII FIGURE 1 – COMPARISON BETWEEN RIVETED AND ADHESIVE BONDED JOINTS [2]. .......................................................... 5 FIGURE 2 – REPRESENTATION OF FAILURE MODES: COHESION AND ADHESION [2]. ......................................................... 6 FIGURE 3 – STRESSES ON SLJS. (A) NORMAL STRESS, (B) SHEAR STRESS, (C) CLEAVAGE STRESS, (D) PEEL STRESS (ADAPTED FROM [3]). ............................................................................................................................................. 7 FIGURE 4 – ADHESIVE BONDED JOINTS CONFIGURATIONS: (A) SINGLE LAP JOINT, (B) DOUBLE LAP JOINT, (C) DOUBLE SCARF JOINT, (D) DOUBLE STEPPED-LAP JOINT (ADAPTED FROM [5]). ........................................................................... 7 FIGURE 5 – AUTOMOTIVE INDUSTRY APPLICATIONS OF COMPOSITES [8]. .................................................................... 11 FIGURE 6 – TYPES OF COMPOSITE MATERIALS (ADAPTED FROM [12]). ....................................................................... 12 FIGURE 7 – PROPERTIES COMPARISON [11].......................................................................................................... 13 FIGURE 8 – DIFFERENT PLIES ORIENTATION [6]. ..................................................................................................... 14 FIGURE 9 – OVERVIEW OF PLY-LEVEL FAILURE MODES [15]. ..................................................................................... 15 FIGURE 10 – FAILURE MODES WITH COMPOSITE ADHERENDS [5]. ............................................................................. 15 FIGURE 11 – OVERLAP LENGTH VARIATION ON SLJ [17]. ........................................................................................ 16 FIGURE 12 – EXPERIMENTAL FAILURE LOADS FOR SLJS WITH EPOXY AND POLYURETHANE ADHESIVES [16]. ....................... 17 FIGURE 13 – PEEL LOADING EFFECT IN COMPOSITE ADHERENDS [5]. .......................................................................... 17 FIGURE 14 – TRIANGULAR LAW FOR PURE MODE [18]. ........................................................................................... 19 FIGURE 15 – TRACTION-SEPARATION LAW WITH PURE AND MIXED LAWS [16]. ............................................................ 20 FIGURE 16 – SCHEMATIC BEHAVIOUR OF CFRP DUE TO DIFFERENT IMPACT LOADS VELOCITIES [24]. ............................... 22 FIGURE 17 – STRAIN RATE EFFECT ON LONGITUDINAL AND TRANSVERSE TENSILE STRENGTH AND MODULUS [27]. .............. 23 FIGURE 18 – FAILURE LOADS UNDER STATIC AND IMPACT TESTS OF THREE DIFFERENT ADHERENDS [30]. ........................... 24 FIGURE 19 – FORCE-DISPLACEMENT CURVES UNDER IMPACT AND QUASI STATIC TESTS [32]. ......................................... 25 FIGURE 20 – STRAIN-STRESS CURVE UNDER IMPACT AND QUASI-STATIC TESTS [33]. ..................................................... 25 FIGURE 21 – STRESS-STRAIN DIAGRAM OF A CRASH-SUITABLE ADHESIVE [2]. .............................................................. 26 FIGURE 22 – IMPACT PEEL TEST OF TWO STRUCTURAL ADHESIVES [2]. ....................................................................... 27 FIGURE 23 – SPECIFIC DAMPING CAPACITY [2]. ..................................................................................................... 29 FIGURE 24 – SPECIFIC DAMPING CAPACITY OF DIFFERENT MATERIALS [2]. .................................................................. 30 FIGURE 25 – VARIATION OF SPECIFIC DAMPING CAPACITY UNDER DIFFERENT OVERLAP RATIOS OF AV119 ADHESIVE JOINTS [2]. ..................................................................................................................................................... 31 FIGURE 26 – GEOMETRY OF BULK SPECIMENS ACCORDING TO EN ISO 527-2 [38]. ..................................................... 33 FIGURE 27 – MOULD FOR PRODUCING THE BULK SPECIMENS WITH STEEL PLATES AND SILICONE RUBBER FRAME [6]. ........... 33 FIGURE 28 – CURE CYCLE OF NAGASE ADHESIVES: (A) XNR6852E-3 AND (B) XNR3324FT. ........................................ 34 FIGURE 29 – TENSILE BULK SPECIMENS OF NAGASE XNR6852E-3 AFTER MACHINING. ................................................ 34 FIGURE 30 – TAST SPECIMENS’ GEOMETRY IN ACCORDANCE WITH ISO 11003-2. ...................................................... 35 FIGURE 31 – RELEASING AGENT APPLIED IN THE MOULD AND IN THE STEEL SPACERS. .................................................... 36 FIGURE 32 - SANDBLASTING MACHINE. ................................................................................................................ 36 XIV FIGURE 33 – INCORRECT TAST SPECIMEN PREPARATION [39] ................................................................................. 36 FIGURE 34 – TAST SETUP WITH EXTENSOMETER. .................................................................................................. 37 FIGURE 35 – FRACTURE MODES (I, II AND III) OF ADHESIVE JOINTS [41]. ................................................................... 38 FIGURE 36 – DCB SPECIMEN GEOMETRY (DIMENSIONS IN MM). ............................................................................... 39 FIGURE 37 – ADHESIVE APPLIED IN THE OPEN SUBSTRATES. ..................................................................................... 39 FIGURE 38 – EXCESS OF ADHESIVE AFTER CURE IN HOT PRESS. .................................................................................. 40 FIGURE 39 – CLEANED SUBSTRATE SIDE VIEW........................................................................................................ 40 FIGURE 40 – DCB SETUP. ................................................................................................................................. 40 FIGURE 41 - MEASUREMENT OF THE CRACK LENGTH (A).......................................................................................... 41 FIGURE 42 – ENF SETUP. ................................................................................................................................. 42 FIGURE 43 – REPRESENTATION OF LINEAR REGRESSION OF THE CORRECTION CRACK LENGTH FACTOR [47]. ....................... 44 FIGURE 44 – SCHEMATIC REPRESENTATION OF THE FPZ (ADAPTED FROM [32]). ......................................................... 45 FIGURE 45 – SQUARES OF PRE-PREG CFRP. ......................................................................................................... 46 FIGURE 46 – (A) HEATING THE PLIES (B) APPLYING PRESSURE. .................................................................................. 46 FIGURE 47 – RELEASING AGENT IS APPLIED IN THE MOULD. ...................................................................................... 47 FIGURE 48 – HOT PRESS. .................................................................................................................................. 47 FIGURE 49 – CURE CYCLE OF CFRP PLATES. ......................................................................................................... 47 FIGURE 50 – CUTTING THE PLATES. ..................................................................................................................... 48 FIGURE 51 – GEOMETRY OF CFRP DCB SPECIMEN (DIMENSIONS IN MM). ................................................................. 48 FIGURE 52 – DCB CFRP SPECIMEN WITH STEEL BLOCKS. ........................................................................................ 49 FIGURE 53 – DCB CFRP TESTING. ..................................................................................................................... 49 FIGURE 54 – ENF CFRP GEOMETRY SPECIMEN (DIMENSIONS IN MM). ...................................................................... 50 FIGURE 55 – ENF CFRP SETUP.......................................................................................................................... 50 FIGURE 56 – GEOMETRY OF SLJ SPECIMENS (DIMENSIONS IN MM). .......................................................................... 51 FIGURE 57 – MOULD WITH SLJ AND SPACERS. ...................................................................................................... 52 FIGURE 58 – CURE CYCLE OF CFRP SLJ OF XNR 6852 E-3 ADHESIVE. ...................................................................... 52 FIGURE 59 – SLJ STATIC SETUP. ......................................................................................................................... 53 FIGURE 60 – TOOLS DESIGNED FOR IMPACT TESTS ................................................................................................. 54 FIGURE 61 – SLJ IMPACT TEST SETUP. ................................................................................................................. 54 FIGURE 62 – SETUP FOR THE VIBRATION ANALYSIS. ................................................................................................ 55 FIGURE 63 - DYTRAN® IMPACT HAMMER. ............................................................................................................. 56 FIGURE 64 – 3 DB METHOD REPRESENTATION. ..................................................................................................... 56 FIGURE 65 – TRACTION-SEPARATION MIXED-MODE LAW AVAILABLE IN ABAQUS®. ........................................................ 57 FIGURE 66 - REPRESENTATIVE TRIANGULAR LAWS OF MODES I AND II OF ADHESIVE XNR 6852 E-3. ............................... 58 FIGURE 67 - REPRESENTATIVE TRIANGULAR LAWS OF MODES I AND II OF CFRP ADHEREND. ........................................... 59 FIGURE 68 – ELEMENTS INTRODUCED IN FE SOFTWARE FOR THE STATIC ANALYSIS. ....................................................... 59 FIGURE 69 – REPRESENTATIVE MESH IN THE OVERLAP AREA. .................................................................................... 60 FIGURE 70 - ELEMENTS INTRODUCED IN FE SOFTWARE FOR THE DYNAMIC ANALYSIS. .................................................... 60 XV FIGURE 71 - ELEMENTS INTRODUCED IN FE SOFTWARE FOR THE STATIC ANALYSIS OF CFRP DCB SIMULATIONS ................. 62 FIGURE 72 – REPRESENTATIVE MESH OF THE DCB SPECIMEN. .................................................................................. 62 FIGURE 73 - ELEMENTS INTRODUCED IN FE SOFTWARE FOR THE STATIC ANALYSIS OF CFRP ENF SIMULATIONS. ................ 63 FIGURE 74 – STRESS-STRAIN CURVES OF BULK TENSILE TESTS OF NAGASE XNR 6852 E-3 UNDER TWO STRAIN RATES. ........ 65 FIGURE 75 – FRACTURED BULK SPECIMENS OF NAGASE XNR6852E-3. ..................................................................... 65 FIGURE 76 – FRACTURED BULK SPECIMENS OF NAGASE XNR3324FT. ...................................................................... 66 FIGURE 77 – SHEAR STRESS-STRAIN CURVES OF NAGASE XNR 6852 E-3 UNDER TWO STRAIN RATES. .............................. 67 FIGURE 78 – FAILURE SURFACES OF TAST SPECIMENS OF NAGASE XNR 6852 E-3 UNDER TWO STRAIN RATES: ................ 67 FIGURE 79 – REPRESENTATIVE R-CURVES OF DCB SPECIMENS OF NAGASE XNR6852E-3 UNDER TWO STRAIN RATES: (A) 0.017 S-1 AND (B) 8.33 S-1. ..................................................................................................................... 69 FIGURE 80 – FAILURE SURFACES OF DCB SPECIMENS OF NAGASE XNR6852E-3 UNDER TWO STRAIN RATES: ................... 69 FIGURE 81 – ENF EXPERIMENT OF NAGASE XNR 6852 E-3 ................................................................................... 70 FIGURE 82 - PRESENCE OF THE EFFECT OF FIBRE BRIDGING ON DCB SPECIMEN. ........................................................... 71 FIGURE 83 – TWO REPRESENTATIVE R-CURVES OF DCB CFRP SPECIMENS UNDER TWO STRAIN RATES: ............................ 72 FIGURE 84 – REPRESENTATIVE R-CURVES OF CFRP USING DIFFERENT METHODS UNDER TWO STRAIN RATES: .................... 73 FIGURE 85 – FAILURE SURFACES OF THE DCB CFRP SPECIMENS ............................................................................... 74 FIGURE 86 – REPRESENTATIVE R-CURVES OF ENF CFRP SPECIMENS UNDER TWO STRAIN RATES:.................................... 75 FIGURE 87 – REPRESENTATIVE R-CURVES OF ENF CFRP SPECIMENS USING DIFFERENT METHODS UNDER 0.11 S-1. ............ 76 FIGURE 88 – ENF CFRP SPECIMEN TESTED. ......................................................................................................... 76 FIGURE 89 – REPRESENTATIVE P-Δ CURVES OF NAGASE XNR 6852 E-3 ADHESIVE FOR THE THREE OVERLAP LENGTHS STUDIED. .............................................................................................................................................. 77 FIGURE 90 – FAILURE SURFACES OF CFRP SPECIMENS OF NAGASE XNR 6852 E-3 OF THREE OVERLAP LENGTHS: (A) 12.5 MM, (B) 25 MM AND (C) 50 MM. ............................................................................................................. 78 FIGURE 91 - REPRESENTATIVE P-Δ CURVES OF NAGASE XNR 33254 FT ADHESIVE FOR THE THREE OVERLAP LENGTHS STUDIED. .......................................................................................................................................................... 79 FIGURE 92 - FAILURE SURFACES OF CFRP SPECIMENS OF NAGASE XNR 3324 FT OF THREE OVERLAP LENGTHS: (A) 25 MM AND (B) 50 MM. .................................................................................................................................... 79 FIGURE 93 – FAILURE LOADS OF THE NAGASE XNR 6852 E-3 AND NAGASE XNR 3324 FT OF THE STUDIED OVERLAP LENGTHS. ............................................................................................................................................. 80 FIGURE 94 – LOAD VS DISTANCE CURVES FOR NAGASE XNR6852E-3 FOR TWO OVERLAP LENGTHS: ............................... 81 FIGURE 95 – FAILURE LOADS FOR DIFFERENT FAILURE MODES OF 12.5 MM OVERLAP LENGTH SPECIMENS. ........................ 82 FIGURE 96 – FAILURE MODES OF IMPACT TESTS OF XNR6852E-3: (A) COHESIVE FAILURE OF 12.5 MM, (B) DELAMINATION FAILURE OF 12.5 MM, (C) DELAMINATION FAILURE OF 25 MM. ....................................................................... 82 FIGURE 97 – ENERGY ABSORBED VS FAILURE LOAD RESULTS OF NAGASE XNR6852E-3 FOR THE IMPACT TESTS................. 83 FIGURE 98 – COMPARISON BETWEEN AVERAGE FAILURE LOADS UNDER QUASI-STATIC AND IMPACT CONDITIONS. ............... 83 FIGURE 99 – COMPARISON OF THE QUASI-STATIC AND IMPACT RESULTS FOR NAGASE XNR6852E-3. ............................. 84 FIGURE 100 – INFLUENCE OF OVERLAP RATIO ON NATURAL FREQUENCIES................................................................... 85 FIGURE 101 – DAMPING RATIO OF STEEL AND CFRP SPECIMENS. ............................................................................. 85 XVI FIGURE 102 – VARIATION OF DAMPING RATIO WITH OVERLAP LENGTH AND TYPE OF ADHESIVE ....................................... 86 FIGURE 103 – NUMERICAL MODEL. .................................................................................................................... 87 FIGURE 104 – P- OF DCB EXPERIMENTS AND SIMULATIONS IN QS CONDITIONS. ....................................................... 87 FIGURE 105 – REPRESENTATIVE FIGURE OF THE ENF CFRP CRACK’S PROPAGATION ..................................................... 88 FIGURE 106 - EXPERIMENTAL VS NUMERICAL QUASI-STATIC RESULTS FOR NAGASE XNR 6852 E-3 ................................ 89 FIGURE 107 – EXPERIMENTAL RESULTS VS NUMERICAL PREDICTION FOR ADHESIVE NAGASE XNR 6852 E-3 ..................... 89 FIGURE 108 – REPRESENTATIVE FAILURE MODE OF NAGASE XNR 6852 E-3 SLJ OF 12.5 MM OF OVERLAP LENGTH. ......... 90 FIGURE 109 – AC YIELD PARAMETER OF A SLJ .................................................................................................... 90 FIGURE 110 – PEEL STRESS DISTRIBUTION ALONG THE OVERLAP LENGTH IN THE CFRP LAYER ......................................... 91 FIGURE 111 - EXPERIMENTAL VS NUMERICAL IMPACT RESULTS FOR NAGASE XNR 6852 E-3 WITH 12.5 MM OVERLAP LENGTH ................................................................................................................................................ 91 FIGURE 112 – FAILURE MODE OBTAINED IN IMPACT NUMERICAL ANALYSIS OF NAGASE XNR6852E-3 ADHESIVE WITH 12.5 MM OVERLAP LENGTH. ............................................................................................................................ 92 FIGURE 113 - EXPERIMENTAL VS NUMERICAL IMPACT RESULTS FOR NAGASE XNR 6852 E-3 WITH 25 MM OVERLAP LENGTH .......................................................................................................................................................... 92 XVII TABLE 1 – MECHANICAL PROPERTIES OF SOME STRUCTURAL ADHESIVES [2]. ................................................................. 9 TABLE 2 – TYPICAL CHARACTERISTICS OF EPOXIES ADHESIVES [5]. ............................................................................... 9 TABLE 3 – TYPICAL CHARACTERISTICS OF POLYURETHANE ADHESIVES [5]. ................................................................... 10 TABLE 4 – TYPICAL CHARACTERISTICS OF ACRYLIC ADHESIVE [5] ................................................................................ 10 TABLE 5 – MECHANICAL PROPERTIES OF SOME COMMON ADHERENDS. ...................................................................... 11 TABLE 6 – BASIC CHARACTERISTICS OF NAGASE XNR6852E-3 AND NAGASE XNR3324FT .......................................... 32 TABLE 7 – CFRP ORTHOTROPIC PROPERTIES FOR UNIDIRECTIONAL PLATES [40]. .......................................................... 45 TABLE 8 - COHESIVE PARAMETERS OF ADHESIVE AND ADHEREND............................................................................... 58 TABLE 9 - ELASTIC ORTHOTROPIC PROPERTIES OF THE CFRP [40]. ............................................................................ 58 TABLE 10 - COHESIVE PARAMETERS OF NAGASE XNR6852E-3 ADHESIVE UNDER DIFFERENT CROSS-HEAD DISPLACEMENT SPEEDS. ................................................................................................................................................ 61 TABLE 11 - COHESIVE PARAMETERS OF ADHEREND UNDER DIFFERENT SPEED TESTS. ...................................................... 62 TABLE 12 - TENSILE PROPERTIES FOR NAGASE XNR6852E-3 ADHESIVE UNDER DIFFERENT STRAIN RATES. ........................ 64 TABLE 13 - TENSILE PROPERTIES FOR NAGASE XNR3324FT ADHESIVE UNDER DIFFERENT STRAIN RATES. ......................... 65 TABLE 14 - SHEAR PROPERTIES FOR NAGASE XNR6852E-3 ADHESIVE UNDER DIFFERENT STRAIN RATES........................... 66 TABLE 15 - FRACTURE TOUGHNESS IN MODE I FOR NAGASE XNR6852E-3 ADHESIVE UNDER DIFFERENT STRAIN RATES. ...... 68 TABLE 16 - FRACTURE TOUGHNESS IN MODE II FOR NAGASE XNR6852E-3 ADHESIVE UNDER DIFFERENT STRAIN RATES. ..... 70 TABLE 17 - FRACTURE TOUGHNESS IN MODE I FOR CFRP UNDER DIFFERENT STRAIN RATES. ........................................... 71 TABLE 18 - FRACTURE TOUGHNESS IN MODE I FOR CFRP FOR THE CRACK INITIATION UNDER DIFFERENT STRAIN RATES. ....... 73 TABLE 19 - FRACTURE TOUGHNESS IN MODE I FOR CFRP FOR THE CRACK STABILIZATION UNDER DIFFERENT STRAIN RATES. .. 74 TABLE 20 - FRACTURE TOUGHNESS IN MODE II FOR CFRP FOR THE CRACK STABILIZATION UNDER DIFFERENT STRAIN RATES. . 75 TABLE 21 - FAILURE LOAD AND EXTENSION FOR THE NAGASE XNR6852E-3 ADHESIVE SLJS. ......................................... 77 TABLE 22 - FAILURE LOAD AND EXTENSION FOR THE NAGASE XNR3324FT ADHESIVE SLJS. .......................................... 78 TABLE 23 - FAILURE LOAD AND ABSORBED FOR THE NAGASE XNR6852E-3 ADHESIVE SLJS. ......................................... 80 Chapter 1 - INTRODUCTION 1 This dissertation presents a study on the behaviour of adhesively bonded joints under static and impact conditions. For this purpose, several experimental and numerical studies were performed, and the resultant data analysed and compared. A dynamic vibration analysis was also performed with the aim to assess the variation of the damping behaviour of adhesively bonded joints with the change of several different parameters. In this chapter, a brief introduction to adhesive bonding and composite materials is made, supported by the motivation and objectives of the thesis. The methodology followed in this work is also described. 1.1 Background and motivation The use of adhesively bonded joints in the automotive industry has increased substantially over the past decades. The necessity to reduce the weight of structures, driven by the need to reduce emissions and fuel consumption, has led to a significant increase in the application of composite materials, which require extensive use of adhesive bonding be used in complex structures. However, adhesively bonded joints used in the automotive industry must be able to resist to impact loads, providing high impact strength and high deformation before failure. The joints must therefore be able to absorb large amounts of energy during impact, but at the same time be capable to maintain the integrity of the structures. Composite materials are composed of two or more components, and they are characterized by combining the mechanical properties of the different components together. High stiffness, high strength and low density are common properties of this type of materials. The behaviour of composites when bonded is very different from metals. While a large number of studies have been carried out about impact loads in metals, just a few were done considering composites. In this field, the automotive and aerospace industry have been the leaders in development, focusing their efforts on improving the delamination resistance of CFRP laminates and fracture behaviour of adhesive joints. The demands of the customers of the automotive industry concerning acoustic comfort are also continuously increasing, which makes the reduction of noise and vibration a main concern in the automotive industry. Different oscillatory effects can be found: wheel shimmy, brake judder axle harshness, between others. This kind of oscillatory effects transmits to the structure, and it is necessary to include damping mechanisms on the structures. Due to this, it is necessary to study the damping behaviour. Chapter 1 - INTRODUCTION 2 1.2 Objectives The main goal of this thesis is to understand the behaviour of CFRP adhesively bonded joints under static and impact loading, characterizing their behaviour under these conditions. 1.3 Research methodology A step by step research procedure was followed during this thesis in order to reach its main goal. The main steps undertaken are described below, presented by their chronological order: a. A literature review was made focusing on themes such as adhesive bonding, composite materials, impact tests as well as vibrations analysis; b. SLJ specimens with CFRP adherends were manufactured to perform quasi-static, impact and vibrational tests; c. DCB and ENF tests were performed to characterize the CFRP under two different strain rates; d. Bulk tensile, TAST, DCB and ENF tests were performed to characterize adhesives under two different strain rates, mainly to understand the influence of strain rate on the adhesive properties; e. Quasi-static and impact tests of SLJ specimens were performed at room temperature (RT); f. Numerical simulations were developed using Cohesive Zone Models (CZM) to evaluate the obtained experimental results under quasi-static and impact conditions; g. Vibrational analysis of SLJ specimens was also performed to determine the influence of substrates, overlap length and type of adhesive in the dynamic response. 1.4 Dissertation outline This dissertation is divided in seven chapters, including introduction and conclusions. In the introductory chapter, a brief description of the study is described, as well as background, motivations and objectives of the thesis. The literature review chapter is composed of a brief description of adhesive bonding, considerations regarding the mechanical properties of different types of adhesives. A detailed description of structural adhesives is made, since these are the adhesives used in this work. A study about the influence of different strain rates in the adhesive properties is also made. The properties of the substrate used, CFRP, under different strain rates are also analysed. Impact behaviour of adhesively bonded joints is considered. Crash resistant adhesives and their Chapter 1 - INTRODUCTION 3 characteristics are also an object of study. Finally, a study on the dynamical and vibration aspects of adhesively bonded joints properties is also made. The experimental details chapter details the preparation and procedures followed during practical tests. Specimens’ geometry and material is also referred for each test. The manufacture process for each specimen is described. Tests to characterize the adhesive and the substrates were made, in conjunction with the SLJ specimens intended to assess joint strength. The numerical details chapter reports the 2D finite element models built, both for the static and impact simulations. Also discussed are the DCB and ENF numerical models employed for the process of CFRP characterization. The experimental results chapter is composed of the results of the adhesive and CFRP characterization in addiction to static, impact and dynamic results of SLJ. Failure loads, fracture modes and its corresponding simulations are also presented. The last sections of this document are the conclusions and future work chapters. Here, conclusions are drawn regarding the main topics of research, complemented by a discussion of the most relevant suggestions for further work. Chapter 2 - LITERATURE REVIEW 4 2.1 Overview of adhesive joints An adhesive can be defined as a material that when applied on two different surfaces is capable of joining them permanently, through an adhesive bonding process. The two parts needed to be joined are commonly referred as substrates or adherends [1]. Considering the load carrying capacity, it is possible to define two different types of adhesives: structural and non-structural. Structural adhesives are capable of withstanding shear stresses ranging from 5 up to 50 MPa, depending on the adhesive type. Aging resistance is also important due of the requirement of adhesive to maintain its structural properties over time. The capacity to withstand stresses without losing its structural integrity is an important property as well. Non-structural adhesives are not capable of sustaining substantial loads. Structural adhesives can be found in different industries, such as the automotive, aerospace and construction industries, used to bond very distinct structures. Non-structural adhesives are also extensively used and can be commonly found in the shoes, sports, packing and furniture industries [2]. More traditional methods, such as mechanical fastening and fusion methods (for example: riveting and welding), have been used over the years as joining mechanisms. More recently, adhesives were added to this category, presenting a wide range of advantages when compared to the other existing techniques [3]:  Capability to join different materials, both metallic and non-metallic, with different thermal expansion coefficients;  More uniform stress distribution along the bonded region, providing a higher stiffness and better load transmission between the two adherends (Figure 1);  High dynamic-fatigue resistance;  Reduction of the weight of the structures and costs;  Good ability to join sheet material efficiently;  Possibility of automatization. Chapter 2 - LITERATURE REVIEW 5 Figure 1 – Comparison between riveted and adhesive bonded joints [2]. Some disadvantages are also inherent to this type of joining process such as [1], [4]:  Need to avoid peel and cleavage stresses, because they are concentrated on a small area, resulting in poor joint strength;  Bonding degradation under extreme environmental conditions such as high temperature and humidity;  Difficulty of evaluating bond quality;  Surface preparation is generally required;  Long curing cycles, especially when a high temperature of cure is necessary;  Necessity of using tools such as presses, ovens and autoclaves, which can increase process cost;  Low maximum service temperature. Two different failure modes can take place in adhesive joints, namely, adhesion failure and cohesion failure, which are illustrated on Figure 2. Adhesive failure occurs within the interfacial region and it is related to the intermolecular forces between two substances, with unsuitable surface preparations being a common cause for such type of failure. On the other hand, cohesive failure only considers the intermolecular forces inside one substance. Cohesive failure can be due to various factors such as: inadequate overlap length, thermal stresses and gross void defects. In addition, the intermolecular forces presented in cohesion and adhesion are both Van der Waals type forces [2]. Chapter 2 - LITERATURE REVIEW 12 2.2 Overview of composite materials Composite materials are made of dissimilar constituents, combined in order to obtain better properties than each component would show individually. These materials are widely used in aeronautical, aerospace and automotive industries, since they combine high strength with lower weight than metal alloys. Composites materials are also very capable under fatigue conditions, allowing to extend the number of life cycles of a structure. Other important properties of such materials are: high strength, high stiffness, corrosion resistance, thermal stability and fatigue resistance [10]. As stated above, composites are made of two main components: matrix and reinforcement (Figure 6). The matrix can be of ceramic, polymeric or metallic nature, and it is needed to keep the fibres in the appropriate direction and protected from abrasion and the environment. The reinforcement, which is usually the strongest element and provides the mechanical properties, can be made of fibres or particles [11]. Figure 6 – Types of composite materials (adapted from [12]). Some advanced fibre composites provide high strength, high stiffness and low weight, which can be achieved through the use of elements of the first row of the periodic table on the matrix. Some other characteristics that can be offered by this type of material are: high melting points, low thermal expansion coefficient and low density. Chapter 2 - LITERATURE REVIEW 13 2.2.1 CFRP properties As stated previously, composites are made of two components, namely, a matrix and a reinforcement. The carbon fibre provides the strength as a reinforcement, and the polymeric matrix allows to keep the fibres together [11]. This type of composite material has been increasingly studied and used in structural applications, due to its characteristics in comparison to metallic materials. A qualitative comparison between steel, aluminium and composite materials is made on Figure 7 [11]. Figure 7 – Properties comparison [11]. Some disadvantages of this type of material are:  High cost in comparison to the most commonly applied materials (steel and aluminium);  Possibility of failure delamination between plies, especially when under impact circumstances. The structure can exhibit reduced strength when subjected to impact loads. The direction of the fibres is crucial to define the mechanical properties of the composites, being stronger and stiffer in the parallel direction of the fibres and weaker in the perpendicular direction. This flexibility of fibres rearrangement can make this material very anisotropic, which Chapter 2 - LITERATURE REVIEW 14 can be useful when the loads are applied in different directions. In order to reach such characteristic, the plies of composite can be stacked in different directions, making it more isotropic when compared to the unidirectional stacked plies (Figure 8). Figure 8 – Different plies orientation [6]. The direction of the fibres also has a significant influence on the machining process, for instance, in an article published by Zhang, L. et al, the surface grinding forces of unidirectional composites were analysed. The grinding forces are higher in the normal direction than longitudinal forces and these ones are higher than transverse forces [13]. 2.2.2 Failure modes in composite materials The failure of composite materials is a complex subject due to the interactions between fibres and matrix. Completely different failure modes are observed when composites are compared to metallic materials. Instead of a single crack, in composites it is typical to observe a damaged area, where different mechanisms of failure can be presented (Figure 9): fibre breakage, fibre micro buckling, fibre pull-out, matrix cracking, delamination and debonding [14]. Considering the non-homogeneous characteristics of composites, three different failure modes should be taken in account:  Tensile failure in the fibre direction;  Tensile failure perpendicular to the fibre direction;  Interlaminar shear failure. Chapter 2 - LITERATURE REVIEW 15 Figure 9 – Overview of ply-level failure modes [15]. 2.2.3 Joints with composite adherends The anisotropic characteristics of composites requires an analysis of the different failure modes that can occur in adhesive joints that employ these materials as substrates. In the ASTM D5573 standard the different possible failure modes are described (Figure 10). Figure 10 – Failure modes with composite adherends [5]. Failure modes are fundamentally related to different parameters, such as: specimen geometry, quality of the bond and loading. A more detailed research was performed by Jangfen et al [16] analysing not only the influence of the overlap length, but also the adherends’ thickness, adherends’ width and scarf angle on single-lap joints, double-lap joints and scarf joints. As the overlap length increased, the ultimate failure load increased as well, as did the equivalent stiffness of the joint. In contrast, the average lap shear strength reduces with the overlap length (Figure 11). Analysing the surfaces Chapter 2 - LITERATURE REVIEW 16 of the tested joints, the fracture modes changed from cohesive in the adhesive to cohesive in the adherend with an increase in the overlap length. The adherend thickness, as the thickness increased, the failure load also increases, equivalent stiffness and average lap shear strength increased. In this case, the failure mode went from delamination to cohesive in the adhesive as the adherend thickness increased. Figure 11 – Overlap length variation on SLJ [16]. In Figure 11, it is clear that with brittle adhesives failure load does not vary linearly with the overlap length, and it is also noticeable that the increase in the failure load is smaller as the overlap length increased. In double lap joints, the variation on the adherend width was studied and the conclusion drawn is that doubling the adherend width, the failure load also doubled [16]. On the work developed by Neto et al [17], a comparison between a polyurethane adhesive (SikaForce® 7888) and a two-component epoxy adhesive (Araldite® AV138) using CFRP adherends was made. In Figure 12, it is possible to notice that in the case of epoxy adhesive, the failure load increases until it reached a plateau from the overlap length of 30 mm. From this point further, the failure load was dictated by the composite. For the ductile adhesive, for the failure load, a linear behaviour was noticed, since the failure mode was cohesive in the adhesive. Chapter 2 - LITERATURE REVIEW 17 Figure 12 – Experimental failure loads for SLJs with epoxy and polyurethane adhesives [17]. For CFRP adherends it is common to encounter high stress concentrations in the ends of each substrate, because of the peel loading applied in this region results in significant transverse tension (Figure 13). This fact explains the occurrence of delamination in some of the joints, where a strong adhesive leads to the failure of the composite. Figure 13 – Peel loading effect in composite adherends [5]. Chapter 2 - LITERATURE REVIEW 18 2.3 Strength prediction of single lap joints There are two different approaches to analyse and predict the stresses that occur on an adhesive joint subjected to a load. These are known as the analytical and the numerical approaches. Several useful analytical methods are available to predict the strength of adhesive joints. While these methods are simple and very accurate, they are limited in their scope, being restricted to well defined geometries and loading conditions. Due to the need to simulate dynamic behaviour, all of the simulation work in this thesis was performed using numerical methods, namely the finite element methods. 2.3.1 Numerical methods The analytical methods are a simple way to evaluate the stresses on SLJs, nevertheless they are based on some assumptions in order to simplify its evaluation. When more complex geometries are considered or more elaborate materials, such as composites, are used, it is usually more convenient to employ numerical methods. The finite element method (FEM) is one of the most important numerical techniques. This method has been recently combined with fracture mechanics to create cohesive damage models, which are able to accurately model the failure of an adhesive joint. Cohesive damage model The necessity to better define the fracture mechanics in the fracture process zone led to the creation of a computational tool called the Cohesive Zone Model (CZM). The combination of the stress criteria with the fracture mechanics data made possible to determine the crack initiation and growth. In order to implement this method, varied property data of the adherends and the adhesives is required, such as the stiffness in tension (E) and shear (G), cohesive strength in tension and shear (𝑡𝑛 0 and 𝑡𝑠0, respectively), the tensile (𝐺𝐼𝐶) and shear toughness (𝐺𝐼𝐼𝐶), that can be obtained experimentally. This data provides an accurate estimation of fracture laws, which leads to a good prediction of strength of bonded joints. Chapter 2 - LITERATURE REVIEW 19 Figure 14 – Triangular law for pure mode [18]. As it can be seen in Figure 14, this method relates stresses with relative displacements. These laws can present different shapes, such as triangular, exponential and trapezoidal. The selection of a law type should be made having in consideration the specimen geometry and material behaviour. Generally, the trapezoidal law is mainly used on ductile materials, and the triangular law is used on brittle and composite materials. In the numerical analysis develop by Avendaño et al [19]., a trapezoidal shape was used to study the behaviour of SLJs of a ductile adhesive. This triangular scheme (Figure 14) can be seen as the sum of two different areas, one related to the elastic behaviour and another related to fracture energy. The first one is limited by the cohesive strengths (𝜎𝑡,𝑖) that correspond to the peak on the Figure 14. Once the critical relative displacement (𝛿𝑜,𝑖) is exceeded the failure starts, defining the second area, that can be associated to the critical failure energy. In this region, it can be verified that as the crack propagates, energy is dissipated. So, having defined all the properties required, it is possible to find the maximum relative displacement (𝛿𝑚𝑎𝑥,𝑖) [18], [20]. This analysis can be made for three different modes (I, II and III), corresponding to tension, in plane shear and out of plane shear stresses, respectively. Figure 15 shows a puremode (tension and shear) law, and a mixed mode law, which is an arrangement of mode I and mode II. Chapter 2 - LITERATURE REVIEW 20 Figure 15 – Traction-separation law with pure and mixed laws [17]. Degraded joints can also be studied with this model, as shown by Loh et al [21] , who studied the effect of moisture and its diffusion in adhesive joints by combining this method with diffusion-stress finite element analysis (FEA) [2]. The traction separation laws have an initial elastic behaviour; therefore, it becomes a linear damage propagation performance. In case of elasticity, this is modelled by the elastic constitutive matrix, 𝐾, Equation 1 [17]. 𝑡={𝑡𝑛 𝑡𝑠}=[𝐾𝑛𝑛 𝐾𝑛𝑠 𝐾𝑛𝑠 𝐾𝑠𝑠]{𝜀𝑛 𝜀𝑠}=𝐾𝜀 (Eq. 1) In the case of thin adhesive layers, several considerations can be made: 𝐾𝑛𝑛=𝐸, 𝐾𝑠𝑠= 𝐺 and 𝐾𝑛𝑠=0. The complete separation and mixed-mode failure displacement are predicted by a linear power law form Equation 2 of the required energies for failure in the pure modes [17]. 𝐺𝑛 𝐺𝑛𝐶+𝐺𝑠 𝐺𝑠𝐶=1 (Eq. 2) Chapter 2 - LITERATURE REVIEW 21 2.4 Impact behaviour of adhesive joints Loads that are applied during a very short period of time are mainly designated as impact loads. A significant number of different studies on this subject have been carried out over the last years due to the importance of such loads on a wide range of different fields. For instance, in the automotive industry impact loads can occur on crash accidents, and it is necessary to minimize the consequences of an extreme situation like this. Distinct mechanical systems need to be analysed differently according to the kind of load they are subjected to. The impact load can be applied on a cyclic way or it can also be applied on a unique impulse, like a car collision. Regarding this, different mechanical parameters should be studied, in the first case the force or stress, and in the second case the absorbed energy, these are the parameters that better define the adhesive and joint performance. One of the main issues that arises about the behaviour of adhesives under impact conditions is the strain rate dependence of the adhesive and substrates, as a non-brittle adhesive under quasi-static conditions can exhibit brittle behaviour under impact conditions. To fully understand the adhesive and substrate behaviour, experimental tests are highly desirable. The experimental tests most commonly used to study the impact are: pendulum, falling weight and Hopkinson bar [2]. 2.4.1 CFRP behaviour under impact loads Several studies have been carried out to study the behaviour of CFRP under impact damage. The work developed by Liang Tao et al [22] analysed how different impact loads affected a CFRP plate using a non-destructive detection and evaluation technique called: eddy current pulsed thermography (ECPT). It was shown that when the impact energy is low, it is difficult to detect visually any damaged area on the surface. Nevertheless, this technique provided a deeper analysis to interior defects caused by the applied loads. It was concluded that CFRP is sensitive to impact damage, and that this has big influence on its strength. In order to improve the mechanical properties of Fibre-Reinforced Plastics (FRPs), Stelldinger et al [23] evaluate how the inclusion of a rubber layer in a CFRP laminate could improve the impact damage resistance. Instead of plates, the specimens used were tubes of CFRP, since tubular shapes are more vulnerable to damage. It was concluded that the use of a soft rubber inside the CFRP near to the damaged side lead to higher damage threshold loads. Also in the case of low-energy impacts, it was shown that the propagation of damage was smaller in the case of CFRP integrated with rubber. Chapter 2 - LITERATURE REVIEW 28 2.6 Dynamical analysis of adhesive joints Vibration can be defined as “the variation with time of the magnitude of a quantity which is descriptive of the motion or position of a mechanical system, when the magnitude is alternately greater and smaller than some average value or reference” as stated in ISO 2041-1975. This kind of behaviour is generally related to dynamic loads applied to machinery and structures. Most of the times, excessive vibration of the structure is undesirable since it can irretrievably damage the structure once resonance frequency is reached. If this happens, large deformations or even failure can occur. Excessive exposure to vibration and acoustic radiation can also be harmful for humans [35]. The most effective method to reduce vibration magnitude is to improve the damping capabilities when designing structures and machines. Due to the now widespread use of composites materials and adhesive joints in industries like automotive and aerospace, it is necessary to understand the behaviour of composite adhesive joints when undergoing vibrational conditions [35]. Concerning adhesive bonding and its damping capacity, it is necessary to consider three main aspects: vibration amplitude, static strength and stiffness. Adhesives with higher damping coefficient are also related with undesired mechanical properties such as low stiffness and strength. On the other hand, structural adhesives, such as modified epoxies and polyurethanes, are able to combine high damping capacity and high strength, mainly because of its viscoelastic behaviour [2]. Adhesives can present different behaviours with the variation of temperature, and this also has a significant effect on its damping capacity. Three main regions need to be pointed: a “glassy” region where low values of loss factor are achieved; transition region where the loss factor reaches its maximum value, this region is located nearby glass transition temperature; and a third region where low values of loss factor are presented [2]. In order to optimize the damping capacity of this joining method it is necessary to study different variables: adhesive properties, joint geometry and temperature. Chapter 2 - LITERATURE REVIEW 29 2.6.1 Damping parameters Damping can be expressed by using different parameters: specific damping capacity, damping ratio, logarithmic decrement and loss factor. 2.6.1.1 Specific damping capacity The specific damping capacity (SDC) is defined as the ratio between energy dissipated per cycle and the maximum elastic energy stored per cycle per unit volume, and is given by Equation 3. 𝜓=𝑑𝑊 𝑊 (Eq. 3) This entity is expressed in percentage and it is represented in Figure 23. Figure 23 – Specific damping capacity [2]. 2.6.1.2 Damping ratio Damping ratio (Equation 4) is defined by the ratio between the effective damping and the critical damping. 𝜉=𝑐 𝑐𝑐=𝑐 2𝑚𝜔𝑛 (Eq. 4) When 𝜉=0 the system is non-damped, when 𝜉<1 the system is underdamped, when 𝜉=1 the system is critically damped and finally when 𝜉>1 the system is overdamped [35]. Chapter 2 - LITERATURE REVIEW 30 2.6.2 Variation of damping analysis In literature, different aspects were studied in order to define the damping behaviour of lap joints under those variables: overlap ratio, temperature, bondline thickness and type of adhesive. On a research performed by Adams et al. [2] four different adhesives were used, employing a high strength low-damping steel for adherends. Since the flexural bending vibration is the most common type of vibration in lap joints, this type of vibration has to be studied and therefore a free-free beam method was considered. Primarily, it is necessary to differentiate the damping capacity of each lap joint component individually, and as can be seen in Figure 24 the specific damping capacity in this experiments comes mainly from the adhesives, since a low-damping steel was used for adherends. According to this, adhesives were the main source of damping in this case. Figure 24 – Specific damping capacity of different materials [2]. 2.6.2.1 Overlap ratio The overlap ratio is one of the parameters that can influence the damping of a lap joint. This parameter consists on the ratio between the overlap length and the total length of the specimen. Considering that the adhesive is the main damping factor in this system, it would be expectable that a joint with a ratio of 1 would exhibit a higher damping capacity. In experimental conditions (Figure 25), it was observed that this is not true, and an overlap ratio of 0.25 is a common value for higher damping capacity in first mode of vibration [2]. Chapter 2 - LITERATURE REVIEW 31 Figure 25 – Variation of specific damping capacity under different overlap ratios of AV119 adhesive joints [2]. It was also noticed that an increase in the overlap ratio leads to an increase in the natural frequency, which is also an important parameter to have in consideration when designing a structure using this joining method [2]. 2.6.2.2 Temperature Variations in temperature also have significant influence on the damping capacity of adhesive joints. As stated previousily, the polymeric materials have a property named glass transition temperature, 𝑇𝑔. The loss factor is maximum, as well as damping capacity, at this temperature. For such reason, it is necessary to know the service temperature of the joint in study, because by varying the temperature is possible reach higher values of damping according to needs [2]. 2.6.2.3 Bondline thickness Bondline thickness can be another way to control the damping capacity of an adhesive joint. But this is only true with low shear modulus adhesives, where when increasing the bondline thickness an increase in damping is also noticed. On the other hand, in the case of stiffer adhesives, the increase in thickness does not influence the damping behaviour significantly. It is referred by Kaya et al. [36] that with the increase of the adhesive thickness a decrease in the natural frequencies occurs. 2.6.2.4 Type of adhesive In the work developed by Loureiro et al. [7] two different adhesives were used: AV138 (brittle epoxy adhesive) and Sikaflex 256 (ductile polyurethane adhesive). The aim of this work was to characterize the influence of the type of adhesive on the damping characteristics of SLJs. It was concluded that ductile adhesive had higher damping ratio that the brittle one. Chapter 3 - EXPERIMENTAL DETAILS 32 3.1 Materials characterization and experimental procedure The characterization of materials used during the course of this work (adhesives and substrates) is fundamental, in order to better understand their properties and characteristics. This is also crucial when a finite element analysis (FEA) is to be performed, as accurate material properties must be introduced in the model to obtain reliable results [37]. 3.1.1 Adhesive characterization Two adhesives were selected for this work: a one-component epoxy based adhesive, Nagase-Chemtex XNR6852E-3, supplied by Nagase Chemtex® (Osaka, Japan) and a brittle adhesive, a two-part epoxy based Nagase-Chemtex XNR3324FT, supplied Nagase Chemtex® (Osaka, Japan). Only a partial characterization procedure was performed for the brittle adhesive, because it was concluded in an early phase that this adhesive was actually not suitable for the adherends used and led to significant adhesion problems. Basic characteristics for both adhesives are described in Table 6. The adhesives used in this work were characterized under two different displacement rates: 1 and 100 mm/min. For that purpose, tensile and shear tests were performed as well as fracture energy tests in modes I and II. Table 6 – Basic characteristics of Nagase XNR6852E-3 and Nagase XNR3324FT XNR6852E-3 XNR3324FT Type of adhesive Epoxy Epoxy Physical forms One-part system Paste Two-part system Paste Cure cycle 3 hours at 150°C 24 hours at RT Comments Good damping characteristics Crash resistant Carbon adhesion 3.1.1.1 Tensile tests In order to characterize the tensile properties of the adhesive, bulk tests were performed. The specimen geometry used is in accordance of EN ISO 527-2, and it can be seen in Figure 26. Chapter 3 - EXPERIMENTAL DETAILS 33 Figure 26 – Geometry of bulk specimens according to EN ISO 527-2 [38]. The longer dogbone-shape was used in these tests, as it simplifies the process of measuring the strain. Manufacture To manufacture the bulk sheet plates, a steel mould was used (Figure 27), with a silicon rubber frame to prevent the adhesive from flowing out, a design based on French standard NF T 76-142. To prepare for use, the mould was manually abraded with sandpaper to remove remnants of previous use and then degreased with acetone. The last step in the preparation of the mould, was the application of three layers of mould release agent (Loctite® Frekote® 770-NC). This step is done in order to promote easier removal of the cured adhesive plate from the mould. Figure 27 – Mould for producing the bulk specimens with steel plates and silicone rubber frame [6]. Chapter 3 - EXPERIMENTAL DETAILS 34 Once the mould was prepared, a high speed centrifugal mixing machine (SpeedMixer® DAC 150.1 FVZ-K) was used to mix the adhesive components (in the case of two-part adhesives) or to increase the temperature of adhesive (for the single part adhesive, XNR6852E-3). The single part adhesive requires this heating process as it is extremely viscous and an amount of preheating promotes easier application. The adhesives were applied on the mould cavity using a spatula. A generous amount of adhesive was used in order to prevent the appearance of voids. The mould was then closed and a hot press was used to simultaneously apply curing temperature and pressure on the mould, according to the specific cure cycle of each adhesive. The cure cycle for each adhesive is presented in Figure 28. (a) (b) Figure 28 – Cure cycle of Nagase adhesives: (a) XNR6852E-3 and (b) XNR3324FT. Once the cure cycle was completed, the bulk sheet plates were machined with the required dogbone shape (Figure 29). To allow the measurement and logging of the displacement of the specimens using a video extensometer system, two dark lines were drawn on the specimens perpendicularly to the longitudinal direction of the specimens. Figure 29 – Tensile bulk specimens of Nagase XNR6852E-3 after machining. Chapter 3 - EXPERIMENTAL DETAILS 35 Experimental procedure The tensile tests were performed using an INSTRON® model 3367 universal testing machine (Norwood, Massachusetts, USA) with a maximum load capacity of 30 kN. The specimens were mounted on the machine using two wedge grips. This test consists on applying a load on the longitudinal direction of the specimen until failure. Two displacement rates were tested at room temperature. In order to obtain average representative values of the tensile properties, three specimens for each condition were tested. An optical method was used to record the displacement of the specimen, with a digital camera taking pictures of the gauge length between the two lines during the test, allowing to record all the stages until failure. An image processing and analysing software was used to obtain the strain for each specimen. The width and thickness of each specimen were measured in order to calculate the stress of each strain stage. The properties obtained from this test were: Young’s modulus and tensile strength. Young’s modulus was determined considering the slope of the stress-strain curve between 0.05% and 0.25% strain, which is usually in the elastic behaviour of structural adhesives. Tensile strength was obtained directly from the stress-strain curve [2]. 3.1.1.2 Shear properties Thick adherend shear tests (TAST) were employed to characterize the shear properties of the ductile adhesive Nagase XNR 6852 E-3. The specimen is composed of two steel substrates (DIN C45 E), joined by the adhesive layer. The stiffness of the steel substrate combined with the short overlap introduces an almost pure shear stress in the adhesive layer. The geometry used is in accordance to ISO 11003-2 (Figure 30). Figure 30 – TAST specimens’ geometry in accordance with ISO 11003-2. To ensure the alignment of the specimens during the manufacturing stage, a mould with guiding pins is used. Sandpaper was used to remove small pieces of adhesives from previous uses. Acetone was used to degrease and clean the surface and to achieve better adhesion of the mould release agent. After these procedures, the mould is then ready for use (Figure 31). Chapter 3 - EXPERIMENTAL DETAILS 36 Figure 31 – Releasing agent applied in the mould and in the steel spacers. Figure 32 - Sandblasting machine. The next step is the TAST specimens’ surface preparation. Since the specimens are reusable, cleaning was required to remove any remaining adhesive particles from the substrates. For that purpose, the area where the adhesive is applied was sandblasted (Figure 32). Aluminium oxide particles were projected onto the substrates surface using compressed air, at 4 bar of pressure. Besides cleaning the surface, this treatment also enhances the adhesion of the surface. Finally, to clean and degrease the surface, acetone was used. The next step was the application of adhesive on the substrates. The top and bottom substrates were placed side by side and a portion of adhesive was spread in both parts. The substrates were then joined and the spacers were placed between the substrates, to avoid the presence of adhesive in undesired areas (Figure 33). Figure 33 – Incorrect TAST specimen preparation [39] The pins in the mould provided the alignment between the two substrates of each specimen. The mould was closed and placed in the hot plate press. The cure cycle was the same as the one applied when bulk adhesives were manufactured. Once the cure cycle was completed, the specimens were extracted from the mould and the excess of adhesive on the sides of the specimens was removed. Lastly, the spacers were carefully removed from the specimens. Chapter 3 - EXPERIMENTAL DETAILS 37 The machine used to perform the tests was INSTRON® model 3367 universal testing machine. Special fixtures using pins were used to connect the TAST specimens to the universal testing machine. Figure 34 – TAST setup with extensometer. An extensometer was used to obtain the load-displacement curve (Figure 34). Since the extensometer measures the adhesive and adherends displacement, it is also necessary to include a correction factor on the measured displacements. The tests were performed at room temperature under two different displacement rates: 1 and 100 mm/min. 3.1.1.3 Fracture toughness determination In fracture mechanics two different criteria can be studied: stress concentration, factor based criterion, and the energetic criterion. In its work, Campilho [40] determined that the second criteria is more suitable for the prediction of adhesive joints failure. The energetic criterion states that the crack’s propagation occurs only when energy release rate is higher than critical energy release rate, which is a material property [40]. 𝐺>𝐺𝑐 (Eq. 5) Considering the energetic criterion, different approaches should be followed in order to determine different fracture energies. There are three fracture modes: Mode I, Mode II and Mode III. The main difference between them is how the load is applied (Figure 35). In Mode I the load is applied perpendicularly to the plane of the crack, conducting to the tensile mode. In mode II, a shear load is applied parallel to the plane of the crack and perpendicular to its front, in-plane shear Chapter 3 - EXPERIMENTAL DETAILS 44 Figure 43 – Representation of linear regression of the correction crack length factor [47]. Compliance Based Beam Method This method is based on beams’ theory, which allows to define the compliance of adherends. Since this theory does not take in consideration the stress concentration and rotation of the adherends near the crack, equivalent bending modulus is used [48]. The failure energy is given by Equation 12. 𝐺𝐼𝐶 =6𝑃2 𝐵2ℎ(2𝑎𝑒 2 ℎ2𝐸𝑓+1 5𝐺13) (Eq. 12) Where, 𝑃− 𝐴𝑝𝑝𝑙𝑖𝑒𝑑 𝑙𝑜𝑎𝑑 𝐸𝑏− 𝐸𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡 𝑏𝑒𝑛𝑑𝑖𝑛𝑔 𝑚𝑜𝑑𝑢𝑙𝑢𝑠 𝐵− 𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛′𝑠 𝑤𝑖𝑑𝑡ℎ 𝑎𝑒− 𝐸𝑞𝑢𝑖𝑣𝑎𝑙𝑒𝑛𝑡 𝑐𝑟𝑎𝑐𝑘′𝑠 𝑙𝑒𝑛𝑔𝑡ℎ ℎ−𝑆𝑝𝑒𝑐𝑖𝑚𝑒𝑛′𝑠 𝑡ℎ𝑖𝑐𝑘𝑛𝑒𝑠𝑠 𝐺13−𝑆ℎ𝑒𝑎𝑟 𝑚𝑜𝑑𝑢𝑙𝑢𝑠 Contrarily to the previous methods, as it can be seen in Equation 12, instead of the use of real crack’s length, this method has its own definition of equivalent crack length. This measurement depends only of the specimen’s compliance during the test. Equivalent crack’s length is measured until half of the Fracture Process Zone (FPZ), which is defined by the existence of multiple micro-cracks and plasticization’s ahead the major crack that absorb some of the energy (Figure 44). For ductile adhesives, the energy dissipated in the FPZ is higher. This method is based on the beam theory of Timoshenko [48], [49], [50]. Chapter 3 - EXPERIMENTAL DETAILS 45 Figure 44 – Schematic representation of the FPZ (adapted from [32]). 3.1.2 Carbon fibre reinforced plastic characterization As it was previously stated, the tensile properties of CFRP are known to not vary significantly with the strain rate. Due to this fact, characterization of Young’s modulus and tensile strength was not performed in these experimental procedures. There is, however, limited information regarding strain rate dependence of fracture energy, which required the use of DCB and ENF testing to obtain this data. The adherends used on this work consisted on unidirectional laminates of CFRP (Carbon Fibre Reinforced Plastic) using a carbon/epoxy pre-preg (SEAL® Texipreg HS 16 RM). A unidirectional disposal is used, since it provides better properties on the solicitation direction. The mechanical properties of this material were already determined by Campilho in 2009 and can be seen on Table 7 [40]. Table 7 – CFRP orthotropic properties for unidirectional plates [40]. Elastic Modulus (MPa) Poisson´s Ratios Shear Modulus (MPa) E1 = 1.09E5 ν12 = 0.342 G12 = 4315 E2 = 8819 ν13 = 0.342 G13 = 4315 E3 = 8819 ν23 = 0.380 G23 = 3200 3.1.2.1 Manufacture Unidirectional laminates of CFRP were used to manufacture the specimens. They were produced by hand lay-up and then cured in a hot press. Each plate with 300x300 mm dimensions was fabricated a stack of 0° lay-ups, allowing it to exhibit excellent properties in the fibre direction. Each ply of carbon/epoxy pre-preg (SEAL® Texipreg HS 160 RM) had 0.15 mm of thickness. For each experiment different plate thicknesses were manufactured. Chapter 3 - EXPERIMENTAL DETAILS 46 CFRP plates were manufactured according to the following steps: 1. The carbon/epoxy pre-preg roll was removed from the freezer and left until it reached ambient temperature. 2. The pre-preg roll was cut into 300 mm squares using a blade (Figure 45); Figure 45 – Squares of pre-preg CFRP. 3. Two different plies were disposed side by side; 4. Both plies were uniformly heated, promoting an easier adhesion; Figure 46 – (a) Heating the plies (b) applying pressure. 5. The laminates were stacked by hand lay-up, maintaining the 0° orientation between plies. A small hand tool was used to apply some pressure on the stacked plies and remove any air bubbles (Figure 46); 6. The metal mould was sanded and cleaned with acetone. 7. Three layers of mould release agent were applied, with the aim of preventing the CFRP plate from sticking to the mould (Figure 47); Chapter 3 - EXPERIMENTAL DETAILS 47 Figure 47 – Releasing agent is applied in the mould. 8. Calibrated metal tape was used to regulate the mould height and therefore the final plate thickness; 9. The plates were then cured in a hot press according to the cure cycle (Figure 48 and Figure 49). Figure 48 – Hot press. Figure 49 – Cure cycle of CFRP plates. 10. After the cure cycle was completed, the plates were cut according to specimens’ geometry. The machine used was a diamond disc cutting was the model DV 25 Batisti Meccanica (Figure 50). Chapter 3 - EXPERIMENTAL DETAILS 48 Figure 50 – Cutting the plates. 3.1.2.2 Fracture toughness 3.1.2.2.1 Double Cantilever Beam – Mode I The geometry of the CFRP DCB specimens was based on the ISO 15024 [51] standard, from 2001, which is used to determine, in mode I, the interlaminar fracture toughness, GIC, for unidirectional reinforced fibre-reinforced plastic composites. Each DCB specimen was manufactured using twenty layers of CFRP, each 0.15 mm thick. Following the indications from the standard ISO 15024 [51] the dimensions (Figure 51) of each DCB specimen are the following: length of 125 mm, width of 25 mm and thickness of 3 mm. A pre-crack (a0) of 45 mm was introduced, using a layer of Teflon with a thickness of 0.03 mm. This layer of Teflon is introduced with the aim of inducing stable crack-propagation. Figure 51 – Geometry of CFRP DCB specimen (dimensions in mm). The ISO 15024 standard [51] also states that the crack-opening mode due to a load applied perpendicular to the plane of delamination using the DCB specimen can be achieved with the use of hinges or load blocks. For this work it was decided to use load blocks, which were bonded to the carbon plates using Araldite® 420 A/B adhesive (Figure 52). Chapter 3 - EXPERIMENTAL DETAILS 49 Figure 52 – DCB CFRP specimen with steel blocks. Five specimens per configuration were tested for strain rate analysis, accordingly the specification of ISO 15024 [51]. To help with the visualization of the crack propagation, the sides of the specimen were spray painted with white paint. The DCB tests have been performed in an INSTRON® model 3367 universal testing machine (Norwood, Massachusetts, USA) with a capacity of 30 kN. All the specimens have been pre-cracked using a pre-testing procedure. In this procedure, the specimen was loaded until the crack starts to propagate away from the Teflon insert. As soon this natural crack growth was detected, the specimen was unloaded. The length of initial pre-crack is then registered and a paper scale glued to the specimen starting from the tip of the pre-crack. This scale allows the measurement of crack growth while the test being performed (Figure 53). During the test the onset of stable delamination growth is monitored and the delamination initiation and propagation readings are recorded. The crack length of tests performed at room temperature and for the case of cross-head speed displacement of 2 mm/min was recorded by taking pictures every 5 seconds while for those tests with a strain rate of 100 mm/min a video was logged instead. Figure 53 – DCB CFRP testing. Chapter 3 - EXPERIMENTAL DETAILS 50 3.1.2.2.2 End-Notched Flexure – Mode II The same manufacture procedure described for the DCB specimens was followed for ENF specimens. Since there is no standard for this experiments, the geometry of the specimens was defined by studying the behaviour of numerical models in Abaqus®, based on the work developed by Moura et al. [52] The purpose of this step was to check if a stable crack propagation could be achieved with a given geometry. The final geometry used is represented on Figure 54. Figure 54 – ENF CFRP geometry specimen (dimensions in mm). Each plate of ENF specimens was made using twenty layers of CFRP. The pre-crack was induced introducing a layer of Teflon halfway during the manufacture process, after half of the CFRP plies were stacked. With the aim to correctly measure the crack propagation, the side of the specimen was painted using a spray white paint. The ENF tests have been performed in an INSTRON® model 3367 universal testing machine (Norwood, Massachusetts, USA) with a capacity of 30 kN (Figure 55) Figure 55 – ENF CFRP setup. A described for the DCB procedure, a pre-crack was created in each specimen by means of a pre-testing procedure where the specimen is loaded up until the crack propagates. In the Chapter 3 - EXPERIMENTAL DETAILS 51 case of composite materials, and to avoid unstable crack propagation, Equation 13 should be attended [53]. 𝑎0 𝐿>0.7 (Eq. 13) Where L is the mid length of the specimen. Crack length was also monitored during the test procedure. For the tests performed at room temperature and for the case of 2 mm/min cross-head speed displacement, pictures were taken every 5 seconds, and for the strain rate of 100 mm/min a video was recorded. 3.1.3 Single lap joint testing To better study the behaviour of the adhesive and substrates in a joint, SLJ specimens were manufactured and tested. The tests were performed both at quasi static and under impact. The substrates were 2.1 mm thick CFRP plates and manufactured according to the procedure described above. The geometry chosen is representative of automotive industry applications, with thin substrates and single overlap bonding. Specimens were produced with three overlap length, to study the variation of static and impact behaviour in different overlap lengths. The geometry of the specimen is described in Figure 56. The free length was kept constant, in order to allow direct comparison between the results of the various overlap lengths. The thickness of the adhesive layer used was 0.2 mm. Figure 56 – Geometry of SLJ specimens (dimensions in mm). To manufacture the specimens, the same mould used in DCB adhesive specimens manufacture was used. After cleaning with acetone, three layers of mould release agent were applied on the mould surface.To ensure the correct positioning of the specimens in the mould and the correct adhesive thickness, steel spacers were used. Three layers of mould release agent were also applied to the spacers. Chapter 3 - EXPERIMENTAL DETAILS 52 Figure 57 – Mould with SLJ and spacers. Sandblasting was used to improve the adhesion of CFRP specimens to the adhesive, combined with the application of acetone to degrease the surface. Before application the adhesive was introduced on a mixture machine to reduce its viscosity and simplify the application process. Once the adhesive was applied in the substrates, the mould was closed and it was introduced on the hot plate press. The cure cycle performed was according to the diagram shown in Figure 58. Figure 58 – Cure cycle of CFRP SLJ of XNR 6852 E-3 adhesive. As soon as the cure cycle was completed, the mould was removed from the hot press and the excess adhesive on the sides of the SLJ was removed. Holes were drilled on each end of the specimens, to allow the assembly of the specimen on the machine holding device. For the static tests, CFRP square tabs were added to the specimens. For the impact tests, two tabs were added on each end of the specimens: on side used 1 mm thick steel tabs and the other side used 3 mm tabs aluminium tabs. These tabs are used to enhance the grip during testing, avoiding slippage of the specimen in the clamps during testing. Additionally, they also reinforce the clamped area and prevent hole failure during impact testing. In both cases, Araldite® Chapter 3 - EXPERIMENTAL DETAILS 53 420 A/B was used to join the tabs to the specimens. A cure cycle of 2h at 50°C was used for bonding the tabs. The static tests of specimens with 12.5 mm and 25 mm of overlap length have been performed in an INSTRON® model 3367 universal testing machine (Norwood, Massachusetts, USA) with a capacity of 30 kN. The static tests of 50 mm overlap length have been performed in a Material Test System® 810 universal testing machine with a capacity of 100 kN. The use of different testing machines was necessary as the failure loads of the specimens with the larger overlap length were higher than the maximum capacity of the first machine. All tests were performed at room temperature and with a constant cross head rate of 1 mm/min. At least three specimens were tested for each condition, and for each joint a load-displacement curve was obtained. In both machines, the assembling system was the same. Two grips with 4 screws and a centre pin were used as a holding device. (Figure 59) Figure 59 – SLJ static setup. The impact tests have been performed in a Rosand® Instrumented Falling weight impact tester, type 5 H.V. (Stourbridge, West Midlands, U.K.). This device consists of a mass that is dropped from a certain height, until it impacts the specimen. The mass and the height establish the energy and the speed, respectively, that is applied on the specimen during the impact test. The energy and speed can be calculated using Equation 14 and Equation 15. 𝐸𝑝=𝑚𝑔ℎ (Eq. 14) 𝐸𝐾=1 2𝑚𝑣2 (Eq. 15) The load is fully transmitted to the SLJ by the vertical guide (Figure 61). A load cell attached to the impactor is used to measure the load and time. Chapter 4 - NUMERICAL DETAILS 60 Figure 69 – Representative mesh in the overlap area. The numerical results are presented and compared with the experimental results in the next chapter. 4.3 Impact analysis of SLJs A two-dimensional explicit model was created in order to predict and simulate the behaviour of CFRP SLJs under impact conditions (Figure 70). The explicit model allows the simulation of high speed tests, including drop weight tests and crash analysis. To simulate the drop weight, an additional mass was added to the non-clamped edge. The value of the mass was changed accordingly to the impact energy required. Figure 70 - Elements introduced in FE software for the dynamic analysis. The boundary conditions were similar to those used in the static tests, replacing the displacement with the predefined field of velocity type, which was 2 m/s, in accordance with the practical drop weight experiments. Experimental tests for characterization of the adhesive and CFRP were performed under two different speed tests: quasi-static and 100 mm/min. This experimental procedure allows extrapolation from these values and the calculation of each property under impact speed conditions. The logarithmic law used in this process is given by Equation 17. 𝑃𝑟𝑜𝑝𝑒𝑟𝑡𝑦 𝑡𝑜 𝑑𝑒𝑡𝑒𝑟𝑚𝑖𝑛𝑒=𝐴 ln(𝜀󰇗)+𝐵 (Eq. 17) Where A and B are constants obtained from experimental data, and 𝜀󰇗 is the strain rate in s-1. Chapter 4 - NUMERICAL DETAILS 61 4.3.1 Conversion of machine cross-head displacement speed to strain rate During static testing procedures, the parameter that controls the test speed is the crosshead displacement speed, (usually with mm/min as units). However, to correctly estimate the effect of the testing speed, it is more accurate to use the actual strain rate of the tested material. Therefore, it is required to perform the conversion from crosshead displacement to the strain rate (which has s-1 as units). The fundamental concept of strain rate [54] is characterized as being the change in strain (deformation) of a specific material with respect to time. Equation 18 which can be used to determine the strain rate is the following: 𝜀󰇗(𝑡)=(𝐿(𝑡)−𝐿0 𝐿0)=𝑣(𝑡) 𝐿0 (Eq. 18) Where, 𝐿0 is the original length, 𝐿(𝑡) corresponds to the length at each time 𝑡, and 𝑣(𝑡) corresponds to the speed at which the ends are moving away from each other. The value of 𝐿0 being considered for the DCB and ENF specimens is 0.15 mm as it is the value of each ply thickness of CFRP and interlaminar fracture is being considered. In the case of bonded specimens, where an adhesive layer is present, the thickness of this layer is the value considered in this conversion. Table 10 - Cohesive parameters of Nagase XNR6852E-3 adhesive under different cross-head displacement speeds. Properties QS 100 mm/min Impact E [MPa] 1728 1728 1728 G [MPa] 665 645 603  [MPa] 51.5 60.5 77.7  [MPa] 44.9 44.0 42.9 𝐺𝑛0 [N/mm] 6.4 7.1 8.0 𝐺𝑠0 [N/mm] 51 58.2 64.1 From Table 10, it is possible to see that all the properties are strain rate dependent. In the literature review chapter, it is stated that the tensile properties of CFRP do not vary with the strain rate. Due to this fact, tensile properties were considered constant with an increase in the testing speed, Table 11. Chapter 4 - NUMERICAL DETAILS 62 Table 11 - Cohesive parameters of adherend under different speed tests. Properties QS 100 mm/min Impact E [MPa] 108000 108000 108000 G [MPa] 4315 4315 4315  [MPa] 40 40 40  [MPa] 35 35 35 𝐺𝑛0 [N/mm] 0.59 0.53 0.39 𝐺𝑠0 [N/mm] 1.17 1.04 0.82 4.4 Static analysis of fracture toughness in mode I of CFRP To support the characterization process of the fracture toughness of the CFRP, a two dimensional explicit model was created using Abaqus®. The main goal of this procedure was to validate the data obtained from the experimental work. The tests at room temperature and crosshead displacement of 2 mm/min were simulated using triangular cohesive elements based on the mechanical and cohesive properties of CFRP. Figure 71 - Elements introduced in FE software for the static analysis of CFRP DCB simulations A layer of cohesive elements with 0.15 mm of thickness was placed in the middle of the specimen, since the crack propagation occurred in the middle of the specimen’s thickness (Figure 71). The properties used are the same described before for the quasi-static conditions of CFRP. Figure 72 – Representative mesh of the DCB specimen. The mesh of the model was refined near the areas where the crack propagation occurred, which is in the middle of the adhesive layer (Figure 72). Chapter 4 - NUMERICAL DETAILS 63 4.5 Static analysis of fracture toughness in mode II of CFRP As performed for the mode I analysis, a two dimensional explicit model was created using Abaqus® to support the characterization process of the fracture toughness of the CFRP. The main goal of this practice was to validate the data obtained from the experimental work. The tests at room temperature and cross-head displacement of 2 mm/min were simulated using triangular cohesive elements based on the mechanical and cohesive properties of CFRP. The model used in ENF simulations was similar to the one used in DCB simulations. Only the length of the specimen and the pre-crack length were different. Figure 73 - Elements introduced in FE software for the static analysis of CFRP ENF simulations. A 0.15 mm thick layer of cohesive elements was placed in the middle of the specimen, since the crack propagation occurred in the middle of the specimen’s thickness (Figure 73). The properties used were the same described before for quasi-static conditions of CFRP. The mesh was again refined in the middle of the adhesive layer, as the crack propagation occurred only in this area. Chapter 5 - RESULTS AND DISCUSSION 64 5.1 Experimental results – Adhesives characterization 5.1.1 Bulk tensile tests For Bulk tensile tests, 6 specimens of each adhesive for each stage of strain rate were tested. The strain rate for quasi-static conditions was 0.004 s-1, which is equivalent to a displacement rate of 1 mm/min, and for the higher speed tests, strain rate was 0.42 s-1, which is equivalent to a displacement rate of 100 mm/min. All the tests were performed at room temperature. From these tests, only two adhesive properties were obtained: Young’s modulus (E) and tensile strength (  This occurred since those were the only properties required in the FE simulation, and the extensometer used did not allowed to accurately measure the strain to failure of each experiment. XNR6852E3 Table 12 - Tensile properties for Nagase XNR6852E-3 adhesive under different strain rates. Strain rate (s-1) Tensile strength (MPa) Young’s modulus (MPa) 0.004 51.5 ± 4.0 1728.3 ± 246.9 0.42 60.5 ± 4.6 1675.9 ± 40.5 In Table 12 the measured tensile properties of XNR 6852 E-3 are presented, as well as their standard deviations. It is possible to notice a decrease of the stiffness, and an increase of the tensile strength with the increasing strain rate. Considering the high scatter in the Young’s modulus value in quasi-static conditions and the fact that the value with higher strain rate is inside the standard deviation of quasi-static experiments, it can be reasonably assumed that this property is constant with the variation of strain rate. This is illustrated in Figure 74, where two representative curves for each stage are presented. Chapter 5 - RESULTS AND DISCUSSION 65 Figure 74 – Stress-strain curves of bulk tensile tests of Nagase XNR 6852 E-3 under two strain rates. According to the data in Figure 74, it is possible to conclude that there is no significant variation in Young’s modulus with strain rate. There is, however, a noticeable increase in the tensile strength. Figure 75 show the aspect of bulk specimen of XNR6852E-3 after failure. Significant plastic deformation can be seen, which indicates the ductile nature of this adhesive. Figure 75 – Fractured bulk specimens of Nagase XNR6852E-3. XNR3324FT Table 13 - Tensile properties for Nagase XNR3324FT adhesive under different strain rates. Strain rate (s-1) Tensile strength (MPa) Young’s modulus (MPa) 0.004 47.8 ± 4.1 5320.9 ± 1495.9 0.42 64.0 ± 1.85 6933.3 ± 406.5 From Table 13, an increase of the tensile strength is noticeable. An increase in the Young’s modulus also occurs, but in this case, the large standard deviation does not allow to statistically confirm this trend. Figure 76, show the aspect of bulk specimen of XNR3324FT after failure. Almost no plastic deformation can be seen, which indicates the fragile nature of this adhesive. Chapter 5 - RESULTS AND DISCUSSION 66 Figure 76 – Fractured bulk specimens of Nagase XNR3324FT. In both adhesives, the failure occurred between the drawn marks, which ensures that the collected data is accurate. 5.1.2 Thick Adherend Shear Tests For the shear tests, 6 specimens of XNR6852E-3 were tested for each stage of strain rate. The strain rate for quasi-static conditions was 0.033 s-1, which is equivalent to a displacement rate of 1 mm/min, and for the higher speed tests strain rate was 3.33 s-1, which is equivalent to a displacement rate of 100 mm/min. All the tests were performed at room temperature. From these tests, two adhesive properties were obtained: Shear modulus (G) and shear strength (   Table 14 - Shear properties for Nagase XNR6852E-3 adhesive under different strain rates. Strain rate (s-1) Shear strength (MPa) Shear modulus (MPa) 0.033 44.9 ± 1.3 665 3.33 44.0 ± 2.29 645 In Table 14 the measured shear properties of XNR 6852 E-3 are presented, as well as their standard deviations. It is possible to notice a residual decrease in the shear strength, with an increase of strain rate. Due to problems in acquiring reliable shear strain data, the shear stress-strain curves did not present a realistic value of shear modulus. Shear modulus was obtained instead using Equation 19, using the average Young’s modulus presented on Table 12, as well as a Poisson’s ratio of 0.3, typical of an epoxy adhesive. 𝐺= 𝐸 2(1+𝜈) (Eq. 19) A slight decrease in shear modulus was observed. These values follow the same trend as the Young’s modulus. On Figure 77 it is possible to observe the behaviour that was stated previously: a decrease in both shear modulus and shear strength. Chapter 5 - RESULTS AND DISCUSSION 67 Figure 77 – Shear stress-strain curves of Nagase XNR 6852 E-3 under two strain rates. On Figure 78 the failure surfaces of quasi-static and strain rate tests are presented. (a) (b) Figure 78 – Failure surfaces of TAST specimens of Nagase XNR 6852 E-3 under two strain rates: (a) 0.033 s-1 and (b) 3.33 s-1. As it can be seen in Figure 78, adhesive failure occurred when XNR6852E3 was applied in steel adherends. Considering this fact, if the adhesive adhered well to this material, higher Chapter 5 - RESULTS AND DISCUSSION 68 values of shear properties would be obtained. The same failure mode was obtained for the higher strain rate. Nevertheless, this value was considered in further analysis. 5.1.3 Double Cantilever Beam For DCB tests, 5 specimens of XNR6852E-3 were tested for each stage of strain rate. The strain rate for quasi-static conditions was 0.017 s-1, which is equivalent to a displacement rate of 0.2 mm/min, and for the higher speed tests strain rate was 8.33 s-1, which is equivalent to a displacement rate of 100 mm/min. All the tests were performed at room temperature. From these tests, the fracture toughness in mode I (GIC) was obtained for the adhesive under different strain rates. Table 15 - Fracture toughness in mode I for Nagase XNR6852E-3 adhesive under different strain rates. Strain rate (s-1) GIC (N/mm) 0.017 6.37 ± 0.22 8.33 7.12 ± 0.32 The average values are presented in Table 15. It is possible to notice a slight increase in the fracture toughness with the strain rate. Use of CBBM allowed to obtain the resistance curve (R-curve) of each experiment. The values on the table were obtained from the plateau in the Rcurve of each experiment, which occurred when the crack propagation stabilized. Representative R-curves of each strain rate stage are presented in Figure 79. (a) GI [N/mm] Chapter 5 - RESULTS AND DISCUSSION 69 (b) Figure 79 – Representative R-curves of DCB specimens of Nagase XNR6852E-3 under two strain rates: (a) 0.017 s-1 and (b) 8.33 s-1. As it can be seen in Figure 80, the adhesive did not adhere perfectly to the steel adherends used. Analysing the fracture surfaces, it is concluded that mixed failure in the adhesive layer occurred, with a combination of adhesive and cohesive failure. It is expected that if the adhesive adhered better to the substrates used, the value for the fracture toughness would be higher. (a) (b) Figure 80 – Failure surfaces of DCB specimens of Nagase XNR6852E-3 under two strain rates: (a) 0.017 s-1 and (b) 8.33 s-1. GI [N/mm] Chapter 5 - RESULTS AND DISCUSSION 76 The propagation of the crack is difficult to measure in high strain rate ENF experiments, because of its abrupt propagation. Due to this fact, the analysis using CCM, CBT and DBT was only made for quasi-static conditions. High speed conditions used CBBM. A representative Rcurve of quasi-static tests with the different methods is presented in Figure 87. Figure 87 – Representative R-curves of ENF CFRP specimens using different methods under 0.11 s-1. From Figure 87 it is possible to see that all the methods are in accordance. Since the tests were performed with CFRP specimens, delamination occurred between the plies. In Figure 88 a picture of a tested specimen is presented. Figure 88 – ENF CFRP specimen tested. In this work, ENF tests were performed to characterize the fracture energy in mode II for the composite. Nevertheless, it is important to refer that standardization for this mode is already available for the calibrated end-loaded split test (C ELS), ISO 15114:2014. GII [N/mm] Chapter 5 - RESULTS AND DISCUSSION 77 5.3 Experimental results – SLJ 5.3.1 Quasi-static tests For the static tests of SLJs, 3 specimens were tested for each overlap length of adhesive. Two adhesives were tested and the three overlap lengths analysed were: 12.5, 25 and 50 mm. The tests were performed with a strain rate of 0.08 s-1 which is equivalent to a displacement rate of 1 mm/min. The load and displacement curves were extracted from the recorded files provided by the testing machine. XNR6852E-3 In Table 21, the results for the failure load are presented for each case, as well as their standard deviations. Table 21 - Failure load and extension for the Nagase XNR6852E-3 adhesive SLJs. Overlap length (mm) XNR6852E-3 Failure Load (kN) Extension (mm) 12.5 10.2 ± 0.4 0.78 ± 0.03 25 21.3 ± 1.4 1.13 ± 2.23 50 38.9 ± 2.0 1.58 ± 0.68 From Table 21 it is possible to notice an increase in the failure load with the increasing overlap length. In Figure 89 the representative force-displacement curves are shown. Figure 89 – Representative P-δ curves of Nagase XNR 6852 E-3 adhesive for the three overlap lengths studied. Load [kN] Displacement [mm] Chapter 5 - RESULTS AND DISCUSSION 78 In all the experiments performed under static conditions, the same failure surface occurred: cohesive failure in the substrates. Representative pictures for each case are presented in Figure 90. (a) (b) (c) Figure 90 – Failure surfaces of CFRP specimens of Nagase XNR 6852 E-3 of three overlap lengths: (a) 12.5 mm, (b) 25 mm and (c) 50 mm. XNR 3324 FT For the brittle adhesive, only two overlap lengths were tested under static conditions. The results of the failure load and its extension are presented on Table 22. Table 22 - Failure load and extension for the Nagase XNR3324FT adhesive SLJs. Overlap length (mm) XNR3327FT Failure Load (kN) Extension (mm) 25 6.5 ± 0.5 0.28 ± 0.01 50 10.8 ± 1.3 0.5 ± 0.08 The failure load, as well as the extension both increase with the increasing overlap length. Representative force-displacement curves are shown on Figure 91 for each overlap length condition. Chapter 5 - RESULTS AND DISCUSSION 79 Figure 91 - Representative P-δ curves of Nagase XNR 33254 FT adhesive for the three overlap lengths studied. As it happened with the ductile adhesive, delamination of the substrates occurred. (a) (b) Figure 92 - Failure surfaces of CFRP specimens of Nagase XNR 3324 FT of three overlap lengths: (a) 25 mm and (b) 50 mm. To summarize the quasi-static tests, the average load and its standard deviation of Nagase XNR 6852 E-3 and Nagase XNR 3324 FT as a function of the overlap length are presented in Figure 93. Displacement [mm] Chapter 5 - RESULTS AND DISCUSSION 80 Figure 93 – Failure loads of the Nagase XNR 6852 E-3 and Nagase XNR 3324 FT of the studied overlap lengths. 5.3.2 Impact tests For impact tests of SLJs, 7 specimens were tested for two overlap lengths of Nagase XNR6852E-3: 12.5 mm and 25 mm. The tests of the two overlap lengths were performed on a strain rate of 10000 s-1 which is equivalent to a displacement speed of 2 m/s, and 15000 s-1 which is equivalent to a displacement speed of 3 m/s, respectively. The different test speeds used are explained by the fact that, for specimens with longer overlap length, more energy was required to break the bond. As the maximum mass capacity of the machine was already being used (26 kg), the only method available to increase the energy was increasing the test speed. The load and displacement curves were extracted from the recorded files provided by the testing machine. Impact tests had been only performed for the two smaller overlaps, since the longer overlap lengths could exceed the capacity of the load cell used. In the Table 23, the results for the average failure loads as well as the average absorbed energy are presented, for each case. Standard deviations are also presented. Table 23 - Failure load and absorbed for the Nagase XNR6852E-3 adhesive SLJs. Overlap length (mm) XNR6852E-3 Failure Load (kN) Absorbed energy (J) 12.5 14.3 ± 2.8 17.13 ± 7.2 25 32.7 ± 2.9 82.3 ± 29.4 From Table 23 is possible to notice a clear increase in the average failure load and absorbed energy, with the increasing overlap length. The wide dispersion of the results is common of the impact tests. In Figure 94, representative force-displacement curves of impact tests are presented for both overlap lengths. Load [kN] Chapter 5 - RESULTS AND DISCUSSION 81 (a) (b) Figure 94 – Load vs Distance curves for Nagase XNR6852E-3 for two overlap lengths: (a) 12.5 mm and (b) 25 mm. In all the experiments the same trend occurs, an increase in the applied load until the maximum force, which was considered as the failure load. Analysing the failure mode, it was noticed that in the case of 12.5 mm overlap length both cohesive failure in the adhesive and delamination occurred. From Figure 95, it is possible to see that the higher value of the failure load occurred in case of delamination in the composite, this means that when the adhesive withstands the load that was applied, the failure happened on the substrate. In the case of 25 mm overlap length, severe delamination occurred in all the specimens (Figure 96). Load [kN] Load [kN] Displacement [mm] Displacement [mm] Chapter 5 - RESULTS AND DISCUSSION 82 Figure 95 – Failure loads for different failure modes of 12.5 mm overlap length specimens. Representative pictures of the fracture surface for each case are presented in Figure 96. (a) (b) (c) Figure 96 – Failure modes of impact tests of XNR6852E-3: (a) cohesive failure of 12.5 mm, (b) delamination failure of 12.5 mm, (c) delamination failure of 25 mm. In impact testing, the energy absorbed is also a very important parameter, as it describes the ability of the material to sustain the impact damage. Due to this, the average energy absorbed as a function of the maximum average failure load is represented in Figure 97. Chapter 5 - RESULTS AND DISCUSSION 83 Figure 97 – Energy absorbed vs Failure load results of Nagase XNR6852E-3 for the impact tests. To conclude this section, a comparison between the quasi-static and impact results is made. Here, it can be concluded that a clear increase in the failure loads occurred for both overlap lengths (Figure 98). Figure 98 – Comparison between average failure loads under quasi-static and impact conditions. An overall perspective is presented in Figure 99. Chapter 5 - RESULTS AND DISCUSSION 84 Figure 99 – Comparison of the quasi-static and impact results for Nagase XNR6852E-3. 5.3.3 Dynamic tests The influence of different parameters on the damping capacity was studied in this chapter. The parameters under study were substrate type, overlap length and type of adhesive. In total, 14 specimens were produced: a single substrate of CFRP, a single substrate of steel, 6 SLJ of steel substrates and 6 SLJ of CFRP substrates. In the case of adhesively bonded joints three overlap lengths for two adhesives were produced. The total length was kept for all the adhesively bonded specimens. Due to the use of specimens with constant total length, the concept of overlap ratio is used in this section. This consists on ratio between the overlap length and the specimen total length. Analysing the variation of CFRP natural frequencies of the specimens, it is noticeable that with an increase in the overlap ratio, natural frequencies also increased, seen in Figure 100. Load [kN] Chapter 5 - RESULTS AND DISCUSSION 85 Figure 100 – Variation of the natural frequencies with the overlap ratio. This result means that an increase in stiffness is proportional to the increase in the mass. The same behaviour is not that significant in the case of steel adherends. These slight variations on natural frequencies need to be taken in account when dimensioning a structure. The damping ratio of CFRP and steel was obtained using two specimens with the same dimensions and testing conditions. In Figure 101, it is possible to notice the better damping properties of CFRP when compared to steel, as expected. The damping ratio of CFRP specimens was more than 3 times higher. Due to this, in terms of damping properties, the use of CFRP is more suitable for car structures. Figure 101 – Damping ratio of Steel and CFRP specimens. Chapter 5 - RESULTS AND DISCUSSION 92 As it can be seen in Figure 111, the failure load obtained from the numerical simulation is similar to the experimental failure load. Figure 112 – Failure mode obtained in impact numerical analysis of Nagase XNR6852E-3 adhesive with 12.5 mm overlap length. In Figure 112, the failure mode can be observed. As it was expected from the experimental tests, delamination in the composite substrates occurred as well. Figure 113 - Experimental vs Numerical impact results for Nagase XNR 6852 E-3 with 25 mm overlap length In Figure 113 a comparison between experimental and numerical impact tests for 25 mm of overlap length is established. As it can be seen, the failure loads are in accordance in both cases. In numerical analysis the same failure by delamination occurred. Displacement [mm] Load [kN] Chapter 6 - CONCLUSIONS 93 An in-depth analysis of the impact behaviour of composite adhesive joints was performed during the course of this work. A list of the conclusions drawn from this work follows:  A crash resistant adhesive (XNR6852E-3) was fully characterized at room temperature and under two strain rates, with the determination of tensile and shear strengths, tensile and shear moduli and mode I and mode II fracture energies. This data was used to extrapolate the mechanical behaviour of the adhesive under impact strain rates;  Another adhesive (XNR3324FT) was also partly characterized, but it was not fully studied as it was proven to be unsuitable for the intended application and inferior in performance to XNR6852E-3;  The fracture energy of the composite material used as substrate was characterized in mode I and mode II, at room temperature and under two different strain rates. This data was used to extrapolate the mechanical behaviour of the composite under impact strain rates. Finite element models using cohesive elements were used to validate this data;  Single lap joints were manufactured using carbon fibre substrates. These joints were tested under static condition and under impact conditions using varied overlap lengths. For static conditions, joints using XNR3324FT and XNR6852E-3 were tested. Due to the low performance of the joints using XNR3324FT, impact testing was performed only with joints containing XNR6852E-3 adhesive;  During single lap joint testing, it was found that, for the XNR6852E-3 adhesive, an increase in the overlap length led to an increase in the failure load, under both testing conditions, quasistatic and impact;  Independently of the overlap length, the same failure mode (delamination of the CFRP substrates) occurred in the quasi-static tests. On the other hand, under impact conditions, and for the overlap length of 12.5 mm, both failure modes occurred: cohesive in the adhesive and delamination;  The linear increase of the failure load combined with consistent failure by delamination of the static tests was explained by the fact that the adhesive is fully plasticized immediately before failure of the composite. This was demonstrated using numerical models.  Failure loads under impact are significantly higher for both overlap lengths tested. This is expected and mainly due to the strain rate sensitivity exhibited by adhesive and the resin of the composite; Chapter 6 - CONCLUSIONS 94  Finite element models were used to simulate and validate the static and impact behaviour of the SLJ specimens. The simulations used cohesive zone modelling, using properties derived from the characterization procedures performed during the course of this work. The models were found to be in agreement with the experimental results;  Joints with CFRP substrates were found to have a higher damping ratio than the joints with steel substrates, being more than three times higher. The optimum overlap ratio to improve damping characteristics was also investigated and was found to be 0.22. This can be explained due to a balance between the shear and peel stresses in the end of the overlaps combined with the damping effect of the adhesive. The XNR6852E-3 adhesive was found to have a higher damping ratio than the XNR3324FT adhesive. Chapter 7 - FUTURE WORKS 95 Due to the complexity of the impact phenomena, there are many experimental procedures that can be performed to further explore the behaviour of the adhesive joints when subjected to high strain rates. A few ideas are listed in this section  Due to the wide range of temperatures that a vehicle can be subjected to, it would be important to characterize the adhesives and composite substrates in tests that combine extreme temperatures with high strain rates. Single lap joints should also be tested under this conditions to better understand the joint behaviour;  Instead of a single adhesive per single joints, it would be also important to further explore the behaviour of mixed-adhesive joints or graded joints under impact conditions. With careful adhesive selection, significant improvements can be found;  As the failure of the joints was mostly by delamination, the use of techniques that mitigate this type of failure should also be explored. Usage of fibre metal laminates or geometrical modifications that reduce the peel stresses in the composite are of special interest. REFERENCES 96 [1] Ebnesajjad S.: Adhesives Technology Handbook - Second Edition, Norwich, NY, William Andrew Publishing (2009) [2] da Silva L.F.M., Öchsner A., Adams R.D.: Handbook of adhesion technology, Heidelberg: Springer-Verlag (2011) [3] Kinloch A.J.: Adhesion and adhesives science technology, London: Chapman And Hall (1987) [4] Lees W.A.: Adhesives in engineering design, Design Council (1984) [5] Banea M.D., da Silva L.F.M.: Adhesively bonded joints in composite materials: an overview. Proceedings of the Institution of Mechanical Engineers, Part L: Journal of Materials Design and Applications, 223(1), p. 1-18 (2009) [6] Varajão J.F.D.: Impact of CFRP Adhesive joints for the Automotive Industry, Master Thesis, Faculdade de Engenharia da Universidade do Porto (2015) [7] Loureiro A.L., da Silva L.F.M., Sato C., Figueiredo M.A.V.: Comparison of the Mechanical Behaviour Between Stiff and Flexible Adhesive Joints for the Automotive Industry. The Journal of Adhesion, 86(7), p. 765-787 (2010) [8] Systems D.A.: Structural Bonding of Lightweight Cars, in Crash durable, safe and economical, T.D.C. Company, Editor, (2015) [9] Sawa T., Suga H.: Behaviors of single-lap adhesive joints subjected to static and impact loads. in Intl. Conf. on Experimental Mechanics: Advances and Applications. 1997. International Society for Optics and Photonics. [10] Kelly A., Mortensen A.: Composite Materials: Overview A2 - Buschow, K.H. Jürgen, in Encyclopedia of Materials: Science and Technology (Second Edition), R.W. Cahn, et al., Editors, Elsevier: Oxford. p. 1361-1371 (2001) [11] Strong A.B.: Fundamentals of composites manufacturing: materials, methods and applications, Society of Manufacturing Engineers (2008) [12] Matthews F.L., Davies G., Hitchings D., Soutis C.: Finite element modelling of composite materials and structures, Elsevier (2000) [13] Zhang L., Ren C., Ji C., Wang Z., Chen G.: Effect of fiber orientations on surface grinding process of unidirectional C/SiC composites. Applied Surface Science, 366, p. 424-431 (2016) [14] Knauss W.G., Gonzalez L.: Global failure modes in composite structures, in NASA Grant NAG 1-1975, (2001) [15] Laffan M., Pinho S., Robinson P., McMillan A.: Translaminar fracture toughness testing of composites: A review. Polymer testing, 31(3), p. 481-489 (2012) REFERENCES 97 [16] Li J., Yan Y., Zhang T., Liang Z.: Experimental study of adhesively bonded CFRP joints subjected to tensile loads. International Journal of Adhesion and Adhesives, 57, p. 95104 (2015) [17] Neto J.A.B.P., Campilho R.D.S.G., da Silva L.F.M.: Parametric study of adhesive joints with composites. International Journal of Adhesion and Adhesives, 37, p. 96-101 (2012) [18] Gonçalves J.P.M., Moura M.F.S.F.d., Castro P.M.S.T.d., Marques A.T.: Interface element including point‐to‐surface constraints for three‐dimensional problems with damage propagation. Engineering Computations, 17(1), p. 28-47 (2000) [19] Avendaño R., Carbas R.J.C., Marques E.A.S., da Silva L.F.M., Fernandes A.A.: Effect of temperature and strain rate on single lap joints with dissimilar lightweight adherends bonded with an acrylic adhesive. Composite Structures, 152, p. 34-44 (2016) [20] de Moura M.F.S.F., Gonçalves J.P.M., Marques A.T., de Castro P.M.S.T.: Prediction of compressive strength of carbon–epoxy laminates containing delamination by using a mixed-mode damage model. Composite Structures, 50(2), p. 151-157 (2000) [21] Loh W.K., Crocombe A.D., Wahab M.M.A., Ashcroft I.A.: Modelling interfacial degradation using interfacial rupture elements. The Journal of Adhesion, 79(12), p. 1135-1160 (2010) [22] Liang T., Ren W., Tian G.Y., Elradi M., Gao Y.: Low energy impact damage detection in CFRP using eddy current pulsed thermography. Composite Structures, 143, p. 352-361 (2016) [23] Stelldinger E., Kühhorn A., Kober M.: Experimental evaluation of the low-velocity impact damage resistance of CFRP tubes with integrated rubber layer. Composite Structures, 139, p. 30-35 (2016) [24] Cantwell W., Morton J.: Comparison of the low and high velocity impact response of CFRP. Composites, 20(6), p. 545-551 (1989) [25] Harding J., Welsh L.M.: A tensile testing technique for fibre-reinforced composites at impact rates of strain. Journal of Materials Science, 18(6), p. 1810-1826 (1983) [26] Taniguchi N., Nishiwaki T., Kawada H.: Tensile strength of unidirectional CFRP laminate under high strain rate. Advanced Composite Materials, 16(2), p. 167-180 (2007) [27] Korber H.: Mechanical response of advanced composites under high strain rates, Master Thesis, Faculdade de Engenharia da Universidade do Porto (2010) [28] Ochola R., Marcus K., Nurick G., Franz T.: Mechanical behaviour of glass and carbon fibre reinforced composites at varying strain rates. Composite Structures, 63(3), p. 455-467 (2004) [29] Hou J., Ruiz C.: Measurement of the properties of woven CFRP T300/914 at different strain rates. Composites Science and Technology, 60(15), p. 2829-2834 (2000) [30] Harris J., Adams R.: An assessment of the impact performance of bonded joints for use in high energy absorbing structures. Proceedings of the Institution of Mechanical Engineers, Part C: Journal of Mechanical Engineering Science, 199(2), p. 121-131 (1985) REFERENCES 98 [31] Goglio L., Rossetto M.: Impact rupture of structural adhesive joints under different stress combinations. International Journal of Impact Engineering, 35(7), p. 635-643 (2008) [32] Canto C.M.S.: Strength and fracture energy of adhesives for the automotive industry, Master Thesis, Faculdade de Engenharia da Universidade do Porto (2013) [33] Kadioglu F., Adams R.D.: Flexible adhesives for automotive application under impact loading. International Journal of Adhesion and Adhesives, 56, p. 73-78 (2015) [34] Burchardt, Olaf Schulenburg J., Linnenbrink M.: New building blocks for lightweight structures. ADHESION ADHESIVES&SEALANTS, 6(4), p. 22-27 (2009) [35] Rodrigues J.D.: Apontamentos de Vibrações de Sistemas Mecânicos, FEUP (2015) [36] Kaya A., Tekelioğlu M.S., Findik F.: Effects of various parameters on dynamic characteristics in adhesively bonded joints. Materials Letters, 58(27–28), p. 3451-3456 (2004) [37] Banea M.D.: High Temperature Adhesives for Aerospace Applications, Doctoral Thesis, Faculdade de Engenharia da Universidade do Porto (2011) [38] da Silva L.F.M., de Magalhaes A.G., de Moura M.F.S.: Juntas adesivas estruturais, Publindustria (2007) [39] Monteiro J.P.R.: Caracterização das propriedades mecânicas e de fratura de um adesivo estrutural de alta ductilidade, Master Thesis, Instituto Superior de Engenharia do Porto (2015) [40] Campilho R.D.S.G.: Repair of composite and wood structures, Instituto Superior de Engenharia do Porto (2009) [41] Chaves F.J., Da Silva L., De Moura M., Dillard D., Esteves V.: Fracture mechanics tests in adhesively bonded joints: a literature review. The Journal of Adhesion, 90(12), p. 955992 (2014) [42] Lopes R.M.: Comparação dos ensaios de DCB e TDCB para determinação da tenacidade em tração de adesivos estruturais, Master Thesis, Instituto Superior de Engenharia do Porto (2014) [43] da Silva L.F.M., Öchsner A.: Modeling of Adhesively Bonded Joints, Springer Berlin Heidelberg (2008) [44] Saldanha D., Canto C., Da Silva L., Carbas R., Chaves F., Nomura K., Ueda T.: Mechanical characterization of a high elongation and high toughness epoxy adhesive. International Journal of Adhesion and Adhesives, 47, p. 91-98 (2013) [45] Wang Y., Williams J.: Corrections for mode II fracture toughness specimens of composites materials. Composites Science and Technology, 43(3), p. 251-256 (1992) [46] Yoshihara H., Kawamura T.: Mode I fracture toughness estimation of wood by DCB test. Composites Part A: applied science and manufacturing, 37(11), p. 2105-2113 (2006) REFERENCES 99 [47] de Moura M., Gonçalves J., Chousal J., Campilho R.: Cohesive and continuum mixedmode damage models applied to the simulation of the mechanical behaviour of bonded joints. International Journal of Adhesion and Adhesives, 28(8), p. 419-426 (2008) [48] de Moura M.F.S.F., Campilho R.D.S.G., Gonçalves J.P.M.: Crack equivalent concept applied to the fracture characterization of bonded joints under pure mode I loading. Composites Science and Technology, 68(10–11), p. 2224-2230 (2008) [49] de Moura M.F.S.F., Gonçalves J.P.M., Magalhães A.G.: A straightforward method to obtain the cohesive laws of bonded joints under mode I loading. International Journal of Adhesion and Adhesives, 39, p. 54-59 (2012) [50] Dias G., de Moura M., Chousal J., Xavier J.: Cohesive laws of composite bonded joints under mode I loading. Composite Structures, 106, p. 646-652 (2013) [51] ISO B.: 15024: 2001. Fibre-reinforced plastic composites. Determination of mode I interlaminar fracture toughness, GIC, for unidirectional reinforced materials. BSI (2002) [52] De Moura M., Campilho R., Gonçalves J.: Pure mode II fracture characterization of composite bonded joints. International Journal of Solids and Structures, 46(6), p. 15891595 (2009) [53] Carlsson L., Gillespie J., Pipes R.: On the analysis and design of the end notched flexure (ENF) specimen for mode II testing. Journal of composite materials, 20(6), p. 594-604 (1986) [54] Rittel D.: On the conversion of plastic work to heat during high strain rate deformation of glassy polymers. Mechanics of Materials, 31(2), p. 131-139 (1999) [55] Bui V., Marechal E., Nguyen-Dang H.: Imperfect interlaminar interfaces in laminated composites: interlaminar stresses and strain-energy release rates. Composites Science and Technology, 60(1), p. 131-143 (2000) [56] De Moura M., Campilho R., Amaro A., Reis P.: Interlaminar and intralaminar fracture characterization of composites under mode I loading. Composite Structures, 92(1), p. 144-149 (2010) [57] Iwamoto M., Ni Q.-Q., Fujiwara T., Kurashiki K.: Intralaminar fracture mechanism in unidirectional CFRP composites: Part I: Intralaminar toughness and AE characteristics. Engineering Fracture Mechanics, 64(6), p. 721-745 (1999)