Strength and Fracture Energy of Adhesives for the Automotive Industry

Strength and fracture energy of adhesives for the automotive industry AUTHOR: CARLOS M. S. CANTO - 080504114 SUPERVISOR: PROFESSOR LUCAS F. M. DA SILVA Departamento de Engenharia Mecânica FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO Oporto, Portugal, July of 2013 REPORT DEVELOPED AS A RESULT OF THE WORK DONE FOR THE MASTER THESIS IN MECHANICAL ENGINEERING. Strength and fracture energy of adhesives for the automotive industry AUTHOR: CARLOS M. S. CANTO - 080504114 SUPERVISOR: PROFESSOR LUCAS F. M. DA SILVA Departamento de Engenharia Mecânica FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO Oporto, Portugal, July of 2013 The present dissertation presents the work done for the acquisition of the master degree in mechanical engineering. It has been developed in conjunction with laboratory of adhesive and laboratory of technological trials from the department of mechanical engineering of FEUP. Carlos Maurício Sousa Canto FACULDADE DE ENGENHARIA DA UNIVERSIDADE DO PORTO Departamento de Engenharia Mecânica Rua Dr. Roberto Frias 4200 – 465 Porto Portugal [email protected] “It is not in the stars to hold our destiny but in ourselves.” William Shakespeare Strength and fracture energy of adhesives for the automotive industry i Abstract Adhesives used in automotive industry must be cheap and have good mechanical properties when bonding metal parts or composites and although there are disadvantages some adhesive bonds are stronger than the materials being bonded together. In some cases it may be the only solution to efficiently bond two surfaces together (ex: carbon fibber and all composites in general) and could be the solution for building lighter and more efficient cars. Nowadays, design is initiated using computer aided simulation and in order to predict the mechanical behaviour of adhesives bonds with a finite element analysis (FEA) a full characterization of the adhesive is necessary. In this study two different adhesives were characterized: a high elongation and high toughness epoxy adhesive, and a toughened epoxy adhesive. Failure strength tests were conducted using both structural adhesives. Tensile bulk tests were performed using long dogbone specimens to characterize the adhesives at room temperature. In addition, a study of size dependence of the bulk specimen was also carried out. Secondly, numerical models of double cantilever beam (DCB) specimens using cohesive zone models were developed using Abaqus®. This was used to develop an optimized smaller DCB specimen that allows a reliable characterization of adhesive fracture toughness. Afterwards fracture strength tests were used to characterize a toughened epoxy adhesive loaded in mode I and validate the numerical results. Lastly, a drop weight impact test was conducted to characterize the mechanical behaviour of a high elongation and high ductility adhesive with low yield strength steel adherends. Contents v 4.1.1 Numerical modelling ............................................................................................................... 37 4.1.2 Numerical results and discussion ............................................................................................ 40 4.2 Experimental DCB tests ................................................................................................................. 57 4.2.1 Experimental procedure .......................................................................................................... 57 4.2.2 Experimental results and discussion of DCB tests .................................................................. 59 5. IMPACT TESTS .................................................................................................. 65 5.1 Experimental procedure ................................................................................................................. 65 5.1.1 Adhesive .................................................................................................................................. 65 5.1.2 Substrates ................................................................................................................................ 65 5.1.3 Specimen manufacture ............................................................................................................ 65 5.1.4 Test procedure ......................................................................................................................... 66 5.2 Experimental results and discussion ............................................................................................. 66 6. CONCLUSIONS ................................................................................................... 69 7. FUTURE WORK .................................................................................................. 71 REFERENCES ............................................................................................................. 73 APPENDED PAPERS .................................................................................................. 75 Strength and fracture energy of adhesives for the automotive industry vi List of publications 1. D. F. S. Saldanha, C. Canto, L. F. M. da Silva, R. J. C. Carbas, F. J. P. Chaves, K. Nomura, T.Ueda, Mechanical characterization of a high elongation and high toughness epoxy adhesive, International Journal of Adhesion and Adhesives, accepted for publication. Strength and fracture energy of adhesives for the automotive industry vii List of tables Table 1 – Comparative values of stiffness and strength of common structural materials.* ............................................ 7 Table 2 – General properties of the most common structural adhesives. ** .................................................................. 9 Table 3 - Results of bulk tensile tests ........................................................................................................................... 30 Table 4 – Comparison of the stress intensity factor between the three cases studied (see Figure 32). ......................... 36 Table 5 - The adhesive properties used for the simulations ......................................................................................... 37 Table 6 – Compilation of the fracture toughness calculated using the CCM, CBT and CBBM. .................................. 53 Table 7 - Summary of the fracture toughness calculated using the CCM, CBT and CBBM. ....................................... 56 Table 8 - Mechanical properties of the steel used for the substrates of the DCB specimens. ....................................... 57 Table 9 – Values of the fracture toughness of adhesive SikaPower 4720 using the normal specimen. ........................ 59 Table 10 – Mechanical properties of the substrates in SLJ .......................................................................................... 65 Table 11 - Energy (J) and failure load (N) values obtained from the quasi-static and impact test. .............................. 67 Strength and fracture energy of adhesives for the automotive industry viii List of figures Figure 1 - Comparison of two stress distribution caused by tension on a sheet part, a) traditional riveted assembly and stress distribution, b) stress distribution on adhesive bonded sheets. [1] ....................................................................... 5 Figure 2 - Schematic models of viscoelastic behaviour, a) Maxwell model, b) Voigt model. ..................................... 10 Figure 3 – Deformations in loaded single-lap joints with rigid adherends. [8] ............................................................ 11 Figure 4 – Deformation in loaded single-lap joints with elastic adherends. [8] ........................................................... 12 Figure 5 – Typical stress distribution using Volkersen’s model for SLJ. ..................................................................... 12 Figure 6 - Typical shear stress and peel stress distribution using Goland and Reissner´s model for SLJ. ................... 12 Figure 7 - Examples of singularities in single lap joints and its contribution to the strain and stress results. [3] ......... 13 Figure 8 – The three modes of loading that can be applied to a crack. ......................................................................... 14 Figure 9 - A comparative representation of the fracture toughness of different materials as a function of density (CesEdupack® (Cambridge, UK)). ............................................................................................................................... 15 Figure 10 (continues)- CZM laws with triangular, exponential and trapezoidal shapes. [10] ...................................... 16 Figure 11 – Butt joint geometry with the load direction (dimensions in mm) (ASTM D 2095) .................................. 18 Figure 12 –Fracture bulk specimens, a) compact tension (CT), b) single-edge notched bending (SENB) [1] ............ 19 Figure 13 – Mode I double cantilever beam (DCB) adhesive-joint specimen. ............................................................. 20 Figure 36 - Schematic representation of the FPZ and crack equivalent concept. ......................................................... 21 Figure 14 – Mode I tapered double cantilever beam (TDCB) test specimen. ............................................................... 22 Figure 15 - Instrumented impact pendulum test. [12] .................................................................................................. 24 Figure 16 - ASTM block impact test (ASTM D950-78) [12] ....................................................................................... 24 Figure 17 - ISO 11343 wedge impact peel test specimen. [1] ...................................................................................... 25 Figure 18 - Exploded view of the mold to produce plate specimens under hydrostatic pressure. ................................ 28 Figure 19 - Dimensions of the bulk tensile specimen used in accordance with standard BS 2782 (dimensions in mm). ..................................................................................................................................................................................... 28 Figure 20 - Stress-strain curve with, Pliogrip 7400/7410 (PU), AV 119 (toughened epoxy), XNR6852 and SikaPower 4720 ............................................................................................................................................................................. 29 Figure 21 - True Stress-True strain curve with, Pliogrip 7400/7410 (PU), AV 119 (toughened epoxy), XNR6852 and SikaPower 4720. .......................................................................................................................................................... 29 Figure 22 – Bulk tensile specimens after test, a) XNR 6852 and b) SikaPower 4720 .................................................. 30 Figure 23 – Short tensile specimen according to EN ISO 527-2 (dimensions in mm) ................................................. 31 Figure 24 – Short specimen with transition radius of 25 (dimensions in mm) ............................................................. 31 Figure 25 – Short specimen with a 54 radius in the transition area (dimensions in mm) ............................................. 31 Figure 26 – Comparison of stress and strain curves between the EN ISO 527-2 short specimen and long dogbone specimen using XNR 6852. .......................................................................................................................................... 32 Figure 27 – Ductile fracture of the short specimen (EN ISO 527-2 standard) in the necking part of the specimen using XNR 6852 .................................................................................................................................................................... 33 Figure 28 - Comparison of stress and strain curve between the short specimen (25 radius) and long dogbone specimen using adhesive XNR 6852. ........................................................................................................................................... 33 Figure 29 - Ductile fracture of short specimen (25 radius) in the necking part of the specimen using XNR 6852. ..... 33 Figure 30 - Comparison of stress and strain curve between the short specimen (54 radius) and long dogbone specimen using SikaPower 4720. ................................................................................................................................................. 34 Figure 31 – Fragile fracture of the short specimen (54 radius) using SikaPower 4720. ............................................... 34 Figure 32 - Principal stress distribution in the horizontal direction (σ11) for the short tensile specimen, a) short tensile specimen according to EN ISO 527-2, b) short specimen with transition radius of 25mm, c) short specimen with a 35 radius in the transition area .......................................................................................................................................... 35 Figure 33 - Geometry of the DCB specimen (dimensions in mm). .............................................................................. 37 Figure 34 – Modelled DCB specimen with the finite element mesh, a) cohesive zone (red), b) view of the all specimen, c) boundary conditions. ............................................................................................................................... 38 Figure 35 - Traction separation law with linear softening available in Abaqus® ......................................................... 39 Figure 37 – Numerical P-δ of three different initial cracks. ......................................................................................... 40 Figure 38 – P-a curve for the three different initial crack lengths. ............................................................................... 40 List of figures ix Figure 39 – Numerical R-curves of the three different initial cracks lengths, a) CCM method, b) CBT method, c) CBBM method. ............................................................................................................................................................ 41 Figure 40 - Summary of the numerical fracture toughness as a function of initial crack length. ................................. 42 Figure 41 - Numerical P-δ of three different specimen length. .................................................................................... 43 Figure 42 – P-a curve for the three different specimen length. .................................................................................... 43 Figure 43 – Numerical R-curves of the three different specimen lengths, a) CCM method, b) CBT method, c) CBBM method. ........................................................................................................................................................................ 44 Figure 44 - Summary of the fracture toughness as a function of specimen length. ...................................................... 45 Figure 45 – Numerical P-δ curve for three different widths. ........................................................................................ 46 Figure 46 – Numerical P-a curve with three different widths. ..................................................................................... 46 Figure 47 – Comparison of the numerical fracture toughness in the plateau region of the R-curve for three different widths and three different methods: CCM, CBT and CBBM. ...................................................................................... 47 Figure 48 – Numerical P-δ curve of three different thicknesses. ................................................................................. 48 Figure 49 – Numerical P-a curve of three different substrates thicknesses. ................................................................. 48 Figure 50 – Numerical R-cures of the three different substrate thicknesses using: a) CCM method, b) CBT method, c) CBBM method. ............................................................................................................................................................ 49 Figure 51 – Comparison of the numerical fracture toughness in the plateau region of the R-curve for three different substrate thicknesses and three different methods: CCM, CBT and CBBM ................................................................ 50 Figure 52 – P-δ curve of two DCB specimens with different materials for adherends. ................................................ 51 Figure 53 - P-a curve for the two different materials. .................................................................................................. 51 Figure 54 – von Mises stresses during the DCB test simulation, a) to d) frames of the test from the beginning to the end and e) an amplification of the critical part of the specimen. .................................................................................. 52 Figure 55 (continues) - Numerical R-curves of the two different materials, a) CCM method, b) CBT method, c) CBBM method. ............................................................................................................................................................ 52 Figure 56 – Geometry of the small DCB specimen; final specimen (dimensions in mm) ............................................ 54 Figure 57 - Numerical P-δ of a normal and a small DCB specimen............................................................................. 54 Figure 58 – P-a curve for the normal and a small DCB specimen. .............................................................................. 54 Figure 59 – Numerical R-curves for the normal and small DCB specimen, a) CCM method, b) CBT method, c) CBBM method. ............................................................................................................................................................ 55 Figure 60 - von Mises stresses during the final DCB test simulation. .......................................................................... 56 Figure 61 - Geometry of the DCB specimens tested, a) small specimen, b) normal specimen. ................................... 58 Figure 62 - Schematic representation of the mold used to cure the DCB specimens with the respective legend. ........ 58 Figure 63 – Example of the P-δ obtained, specimen 4. ................................................................................................ 59 Figure 64 – Example of an R-curve obtained, specimen 4. .......................................................................................... 59 Figure 65 - Comparison of three R-curves using the CBBM method for different velocities. ..................................... 60 Figure 66 – Comparison of the fracture toughness of SikaPower 4720 with different displacement rates................... 60 Figure 67 – Example of the failure mode of DCB specimens with SikaPower 4720 using three different displacement rates, a) 0.2mm/min, b) 0.5mm/min and c) 2 mm/min. ................................................................................................ 61 Figure 68 – P-δ curve for the short and normal DCB specimens tested with SIKA® 4720. ......................................... 61 Figure 69 – R-curve of the small specimen and normal specimen using the CBBM method. ..................................... 62 Figure 70 - Two examples of the cohesive fracture surface of the specimens tested, a) small specimen and b) normal specimen. ..................................................................................................................................................................... 63 Figure 71 – Geometry of the SLJ used for the impact tests (dimensions in mm). ........................................................ 65 Figure 72 - Schematic mold for SLJ specimens. .......................................................................................................... 66 Figure 73 - Comparison of SLJ with mild steel adherends under two different strain rates. ........................................ 67 Figure 74 – Failure mode of the SLJ tested .................................................................................................................. 67 Strength and fracture energy of adhesives for the automotive industry x List of acronyms CT Compact tension DCB Double cantilever beam DMA Dynamic mechanical analysis ENF End notched flexure EVA Ethylene-vinyl acetate FEA Finite element analysis FPZ Fracture process zone PU Polyurethane SENB Single-edge notched bending SLJ Single lap joint CCM Compliance calibration method CBT Compliance beam theory CBBM Compliance-Based beam method Strength and fracture energy of adhesives for the automotive industry xi List of symbols δ Displacement ε Strain η Dumping constant σ Normal stress τ Shear stress a Crack length a0 Initial crack length aeq Equivalent Crack length b Substrate width C Compliance E Young´s modulus Ef Corrected flexure modulus G Shear modulus GI Fracture toughness in mode I GIC Critical fracture toughness in mode I GII Fracture toughness in mode II GIIC Critical fracture toughness in mode II h Substrate thickness k Rigidity KI Stress intensity factor KIC Critical stress intensity factor l Overlap length m Geometric factor P Applied load Tg Glass transition temperature Y Correction coefficient Δ Correction for the crack tip rotation ν Poisson´s ratio Literature review 6 On the other hand, some disadvantages are: a) the service temperature is limited, b) requires in most cases a surface preparation, c) due to a huge variety of adhesives available selecting a suitable one requires some experience, d) adhesive bonding is weak when loaded in tension. The two main cases to be avoided are cleavage and peel stresses, e) avoiding localized stresses on the adhesive is not always possible and can cause rupture of the adhesive, f) in consequence of its polymeric nature heat and humidity are very harmful, g) the joint cannot be built instantly and usually needs a holding mechanism, h) needs curing at high temperatures in most cases, i) although there has been improvements in the quality control of adhesive joints it is still a very hard task to accomplish.[1-3] The adhesives industry is very diverse with multiple applications in areas such as: aeronautical, aerospace, automotive, shoe, furniture and others. With many applications and a market share already well-established the future for adhesives looks promising. According to a study from Ceresana®, 2012, on adhesive markets, it is concluded that it is expanding with growth rates of about 2.9% for the next 8 years. This evolution is predicted as a result of the rapid increase in the demand of consumer goods in the Asia- Pacific region. 2.1 Adhesive properties Adhesives are polymeric by nature and are formed by large molecules, polymers, with small groups of atoms, monomers. The diversity of viable combinations for monomers is great and, as a consequence, the number of polymeric compounds that result are vast. On top of it, there are also mixed adhesives, resulting from a combination of several polymeric compounds. As a consequence the classification of adhesive is accomplished in different ways. Among the most common are: a) Polymer base; natural or synthetic. b) Chemical composition; thermoplastic, thermoset or rubber. c) Physical forms; one or multiple components, films, tape, powder. Strength and fracture energy of adhesives for the automotive industry 7 d) Chemical families; epoxy, polyamides and others. e) Function; structural or non-structural. [1, 4] 2.1.1 General properties In this thesis there is a particular interest in structural adhesives and their mechanical properties. A structural adhesive is an adhesive that transfer loads between adherends and usually have a shear strength higher than 5 MPa. Typically, structural adhesives are cross-linked/thermosetting polymers even though some thermoplastics are used. [7] The strength of properly made adhesive joints is directly related to the strength of the adhesive. It has also been proved that failure is unlikely to occur at the interface and only in cases of poor surface preparation it is likely to take place. [5] Furthermore, since in most cases the adherends (ex. metals and carbon fibber) have a higher rigidity than the adhesive, the displacement will be mainly due to strain in the adhesive. Some common strength properties are presented in Table 1. The choice of the adhesive for a particular application is not unique. Usually there are a variety of adhesives suitable and surface pre-treatments that can be applied to improve the joint performance. Also, the surface type can condition the adhesive selection for the task. For example, in the case of thermoplastic substrates, some adhesives may have a detrimental effect producing effects such as crazing, swelling, dissolutions or may be simply incompatible. On the other hand, some adhesives such as epoxies are versatile and will bond to different substrates.[6] Table 1 – Comparative values of stiffness and strength of common structural materials.* *Information compiled from several text books and databases, illustrative only. Literature review 8 When selecting adhesives, some key factors to consider in the fabrication process are: joint performance (load, operating environment, durability), substrates type, adhesive form, costs, aesthetics, manufacturing process, application, health requirements and pretreatments. Finally testing and validation is recommended to ensure the process quality.[1] Since the diversity of adhesives available is extensive, an accurate choice is hard and will require experience. An overview of the most used structural adhesives is presented in Table 2 with a compilation of several general properties. 2.1.2 Temperature related properties a) Glass transition temperature The glass transition temperature, Tg, is the most important temperature in polymers and is a property of the amorphous part. It marks a transition from a glass-like structure to a rubber-like state. It is not a phase transition but a change in the derivative of the fundamental quantities with respect to temperature. Although in some polymer (linear and very regular) the transition is masked, this is not the case for amorphous polymers where above Tg the long coiled molecular chains can rearrange and extend. This behaviour is mostly unwanted in rigid structures because the viscoelastic nature of the polymer will result in fast stress relaxation, low modulus and strength. In conclusion, the structural adhesives are expected to work below their Tg.[1, 5] b) Decomposition temperature Using adhesives above this temperature will completely destroy the joint and only for a short time there will be relevant mechanical properties. This is important for military projectiles for it relies on the char strength for a short period of time.[5] c) Melting temperature Opposed to the glass transition the melting temperature is not very important for adhesives because crystalline melting does not take place in amorphous polymers. Although of little importance, some adhesives do exhibit cristallinity such as: ethylenevinyl acetate (EVA) and polyamides hot melts; polyvinylalcohol, polychloroprene and starch. [1] Strength and fracture energy of adhesives for the automotive industry 9 Table 2 – General properties of the most common structural adhesives. ** **Information compiled from several text books and databases, illustrative only. Literature review 10 d) Thermal expansion The thermal expansion coefficient of the adhesive is much higher than that of typical metallic substrates, which can lead to damaging interfacial stresses. The expansion can be reduced by the addition of mineral fillers.[1] 2.1.3 Viscoelasticity In the elastic domain of metals an imposed stress will generate an extension proportional to it. However, in polymers the tensile behaviour is strongly influenced by time and the instantaneous response will be a small fraction of the total deformation. Figure 2 shows two different models to describe viscoelasticity. The Maxwell model (Figure 2 - a) is described by: 𝑑𝜀 𝑑𝑡=1𝜂𝑆+1𝑘𝑑𝜎 𝑑𝑡 (1) In Equation 1, σ represents stress, ε the strain, η and k are constants relating to the dashpot viscosity and the rigidity of the spring respectively. According to Equation 1 if the deformation is constant (dε/dt = 0) the stress will decay to zero, commonly known as stress relaxation. The Voigt model (Figure 2 - b) is described by: 𝜎=𝜂𝑑𝜀 𝑑𝑡+𝑘𝜀 (2) In this model the deformation and its recovery is subjected to a time dependency, and this constant is commonly known as retardation time. Figure 2 - Schematic models of viscoelastic behaviour, a) Maxwell model, b) Voigt model. Strength and fracture energy of adhesives for the automotive industry 11 If the stress is applied during a period of time much smaller than the relaxation and retardation times the behaviour is determined by the spring. Typically, for polymers below the glass transition temperature, the relaxation time is infinitely long making the response elastic but time dependent. If the temperature is raised, the viscous component becomes increasingly important, especially above the Tg.[1, 5-7] 2.2 Analysis of adhesive joints For complex geometries, a finite element analysis (FEA) is preferable however, for a fast and easy answer a closed-form analysis is usually used.[1, 3, 8] 2.2.1 Analytical approach In the literature, most attention is given to single lap joint (SLJ) specimens for it is an efficient geometry to characterize an adhesive joint. For this geometry, generally, failure takes place in the adhesive and the stress distribution in that region was subjected to extensive study from many researchers. For the analysis, some simplifying assumptions are made: substrates deformation due to tension and bending only and adhesive stresses restricted to peel and shear are assumed to be constant across the adhesive layer. However, the stresses in the adhesive are not uniform because of differential straining and the eccentricity of the loading path.[2, 8, 9] a) Linear elastic analysis A common and simple analysis is to consider undeformable substrates with a constant shear stress state in the adhesive layer (Figure 3). The adhesive shear stress is given by Equation 3 where P is the remote load applied, b is specimen width and l is overlap length. [8, 9] Figure 3 – Deformations in loaded single-lap joints with rigid adherends. [8] Literature review 12 𝜏= 𝑃 𝑏∙𝑙 (3) b) Volkersen´s analysis The Volkersen’s analysis introduces a differential shear stress in the adhesive as a consequence of substrate deformation (Figure 4). It considers that the SLJ has no bending moment and therefore substrates are in pure tension. The adhesive is in pure shear.[8, 9] Substrates deformation is maximum near the adhesive overlap (point A) and minimum in the opposite end (point B). The reduction of strain along the overlap causes a nonuniform shear stress distribution in the adhesive (Figure 5). The Volkersen´s model does not take into account the effects of the adherend bending and shear deformation, both important aspects for a correct analysis of adhesive joint Figure 4 – Deformation in loaded single-lap joints with elastic adherends. [8] Figure 5 – Typical stress distribution using Volkersen’s model for SLJ. Figure 6 - Typical shear stress and peel stress distribution using Goland and Reissner´s model for SLJ. Strength and fracture energy of adhesives for the automotive industry 13 stress distribution. This is particularly important in adherends with low shear and transverse modulus.[3, 8, 9] c) Goland and Reissner analysis In this analysis a more sophisticated approach is done introducing the aspect of adherend bending and with it peel stresses in the adhesive layer. Figure 6 is an example of the adhesive shear and peel stress in a SLJ. [10] In summary, the classical analysis of Volkersen and Goland and Reissner were a big step forward in adhesive modelling and failure prediction. Nevertheless there are some limitations to these models. Firstly, variation of stresses along bondline thickness is not taken into account. Secondly, the peak shear stresses at the overlap ends are inaccurate as a correct representation should take into account the zero shear stress at the end of the overlap. Also, the complex stress field of the substrates is neglected to most extend. In order to improve these models, more work has been done and more complex models have been put forward increasing the accuracy of the stress distributions.[8, 9] 2.2.2 Numerical approach a) Continuum mechanics approach In continuum mechanics, one of the approaches is the strength of materials which accounts for the maximum stress and strain. It is among the most used. However it is sometimes inappropriate due to singularities inherent to the bonded joint and in such cases the refinement of the mesh will increase greatly the values obtained from the simulation for the strain and stress. Some common singularities are presented in Figure 7.[3] b) Fracture mechanics approach In the continuum mechanics approach, materials are considered to have no defects, in contrast with fracture mechanics analysis where a defect has to exist. The fracture mechanics approach studies the defects to predict if they will cause a catastrophic Figure 7 - Examples of singularities in single lap joints and its contribution to the strain and stress results. [3] Literature review 14 failure of the structure or if they can withstand the stresses throughout the service life of the component. In this analysis there are two types of criteria, stress intensity factor and energetic concepts. [11] It´s a relatively recent field of study and current research is being done to introduce time dependent effects such as viscoelasticity. These effects are important when traditional fracture mechanics are insufficient to accurately predict failure. In this case new computer aided technologies are emerging. [11] In the traditional approach to the design of structures there are two variables, applied stress and strength of the material but with the introduction of the failure criteria of fracture mechanics this has changed. Following the fracture mechanics approach one takes into account the stress, the flaw size and the fracture toughness of the material. The combinations of these three factors can be done with the energy criterion or the stress-intensity one.[5, 11] There are three modes of loading that produce a singularity at the crack tip (Figure 8). Energy criterion The energy approach states that fracture will occur when the energy available for the crack growth is sufficient to overcome the resistance of the material. The material resistance can take into account the surface energy, plastic work, or other energy dissipation associated with the propagation of the crack.[11] The present version of the approach was developed by Irwin which is defined as the rate of change in potential energy with the crack area for a linear elastic material. At the moment of fracture GI = GIC (critical energy release rate which is a measure of fracture Figure 8 – The three modes of loading that can be applied to a crack. Strength and fracture energy of adhesives for the automotive industry 15 toughness). As an example of the method, for a crack of length 2a in an infinite plate subject to a remote tensile stress, the energy release rate is given by: GI=πσ2a E (4) Where E is Young’s modulus, σ is the remotely applied stress, and a is the half-crack length.[11] Since a well-designed adhesive joint will fail cohesively, it is reasonable to assume that the fracture toughness is, to some extent, dependent on the adhesive bulk toughness. The fracture toughness is an important aspect of design and has a great variation as a consequence of temperature, geometry and material. Most adhesives are polymers with intermediate fracture toughness and, in most cases, are one order of magnitude lower than the metal and alloys (Figure 9). Stress intensity approach According to this criterion one assumes the material will fail locally at a critical combination of stress and strain commonly known as the critical intensity factor, KIC. For a situation similar to the energy approach presented above, infinite long plate subject to a remote tensile stress: Figure 9 - A comparative representation of the fracture toughness of different materials as a function of density (CesEdupack® (Cambridge, UK)). Literature review 22 𝐸𝑓 can be obtained using: 𝐸𝑓=(𝐶0−12(𝑎0+|Δ|) 5𝑏ℎ𝐺𝑥𝑦 )−18(𝑎0+|Δ|)3 𝑏ℎ3 (14) The crack equivalent concept is: 𝑎𝑒𝑞=1 6𝛼𝐴−2𝛽 𝐴 (15) where the coefficients are: 𝛼= 8 𝑏ℎ3𝐸𝑓 ;β= 12 5𝑏ℎ𝐺𝑥𝑦 ; 𝛾=−𝐶 (16) 𝐴=((1−108𝛾+12√3(4𝛽3+27𝛾2𝛼) 𝛼 )𝛼2)1/3 (17) Effect of adhesive layer If the bondline thickness is too low for full development of a plastic zone the fracture toughness will change. As a recommendation, the thickness should be between 0.1 mm and 1 mm. This is also applicable in the case of TDCB. [1, 12] c) Tapered double cantilever beam (TDCB) test This method was developed to enable long term measurements of adhesive damage propagation without the need to measure the crack length. The height of the beam changes along the adhesive layer to ensure a constant change of compliance as a function of crack length (Figure 15).[1] The manufacturing of these specimens is more expensive and complex, requiring a CNC machine to account for the non-linear height profile. Figure 15 – Mode I tapered double cantilever beam (TDCB) test specimen. Strength and fracture energy of adhesives for the automotive industry 23 Values of 𝐺𝐼𝐶 can be determined using a simple beam theory, Equation 18. 𝐺𝐼𝐶=4𝑃2𝑚 𝐸𝑠𝑏2 (18) Where P is the load, 𝐸𝑠 the substrate elastic modulus, 𝑏 the substrate width, and 𝑚 the geometry factor defined previously. Or using a more complex but accurate method, corrected beam theory. This method formulation is presented below. 𝐺𝐼𝐶=4𝑃2𝑚 𝐸𝑠𝑏2∙(1+0.43(3 𝑚𝑎)13) (19) 2.3.3 Impact tests The impact is an important feature when designing an adhesive joint for the automotive industry due to the passenger safety regulations and manufacturer quality standards. As a result the behaviour of adhesive joints in high strain rates is a major consideration in order to know how the strength of the joint reduces varies. a) Instrumented pendulum impact test An instrumented pendulum was developed by Harris and Adams to impact a single lap joint or a solid adhesive specimen. The fixture is presented in Figure 16 showing the specimen clapped to the machine’s piezoelectric force transducer that in turn is connected to the frame. The other end is free although there is a journal bearing block to guide the specimen during the test. [12] The strength of the joint is calculated with the load cell and the energy is measured from the pendulum swing after impact. The movement of the end clamp can be instrumented to record the acceleration and thus monitor the position recording the specimen’s behaviour. The energy absorbed by the adhesive rupture is small compared to the energy required to deform the metallic substrates. On the other hand, a low ductility adhesive can have high lap shear strength when using high yield strength substrates and fail with low load loads with ductile substrates. Literature review 24 b) Block impact test This test applies a condition of impact loading, mainly shear, on the test rig similar to that of the Izod resilience measurement (Figure 17). The specimen is fabricated using two blocks, a larger block that will be attached to the base and a smaller block on top of the adhesive layer. This smaller block will be struck during the test by a pendulum in a direction parallel to the bonded surface. In the case of misalignment, the distribution of the shear and peel stresses is strongly influenced. Also, the elastic energy of the steel block may not be negligible in some cases. For these reasons this method is not suitable for the measurement of the energy absorption of the adhesive and can only be used for comparative studies. Figure 17 - ASTM block impact test (ASTM D950-78) [12] Figure 16 - Instrumented impact pendulum test. [12] Strength and fracture energy of adhesives for the automotive industry 25 c) Impact wedge-peel test Impact wedge peel test is a realistic test on an adhesive joint with two ductile adherends bonded together to form a Y shape. The two strips used are 90 mm long, 20 mm wide and the thickness can range from 0.6 to 1.7 mm. The bonding length is 30 mm without pre-cracking or crack initiator. The impact is applied to the shackle, Figure 18, with a pendulum and test is conducted with a speed of 2 or 3 m/s. There are usually two different behaviours of adhesive joints when tested with this apparatus, stable or unstable crack propagation. In the first the crack grows more rapidly than the speed of the wedge and is typically encountered when testing at low temperatures or brittle adhesives. In the second case, the crack tip grows ahead of the wedge with a constant offset. In this last case the force-time history exhibits an initial peak, sudden impact and crack initiation, followed by a plateau where it is possible to calculate the cleavage force. Figure 18 - ISO 11343 wedge impact peel test specimen. [1] Strength and fracture energy of adhesives for the automotive industry 27 3. Failure strength tests 3.1 Bulk tensile test 3.1.1 Adhesives The epoxy adhesive XNR6852 was used, supplied by NAGASE CHEMTEX® (Osaka, Japan). This adhesive is a one-part system that cures at 150 ºC for 3 h. This adhesive has a linear structure, which allows greater freedom of movement to the chains, unlike the network structure of a conventional epoxy adhesive. Along with the development of this adhesive, NAGASE CHEMTEX® has produced others with the same technique. The epoxy resin of XNR6852, when pure, is a conventional thermosetting resin due to generating cross-linking during polymerization. A technological advance in the epoxy adhesive has been done and a no cross-linking polymer has been produced through the introduction of phenols. Thus, the reaction process is changed and in this new process the epoxy resin and phenol are polymerized linearly by a consecutive reaction getting a no cross-linking polymer. As a consequence, this polymer has some features of thermoplastic polymers due to the resulting linear structure [18]. Also, the epoxy adhesive SikaPower 4720 was used, supplied by SIKA® (Portugal, Vila Nova de Gaia). This adhesive is a two-part system that cures at room temperature for 24 hours. 3.1.2 Tensile strength test a) Experimental procedure Specimen manufacture The bulk tensile specimens were produced by curing the adhesive between steel plates of a mold (Figure 19) with a silicone rubber frame according to the French standard NF T 76-142. A silicone rubber frame was used to avoid the adhesive from flowing out. The dimensions of the adhesive plate after cure were defined from the internal dimensions of the silicone rubber frame. Then, dogbone specimens were machined from the bulk sheet plates (Figure 20). Failure strength tests 28 Test procedure The bulk tensile test was performed in an INSTRON® model 3367 universal test machine (Norwood, Massachusetts, USA) with a capacity of 30 kN, at room temperature and constant displacement rate of 1 mm/min. An extensometer to record the displacement was also used. Loads and displacements were recorded up to failure. Four specimens of each were tested. b) Experimental results and discussion Figure 21 and Figure 22 present a comparison of a tensile curve between toughened epoxy; AV 119 from Hunstman® [19], a polyurethane (PU); Pliogrip 7400/7410 from Ashland Specialty Chemicals® [20], and the studied adhesives; XNR 6852 and SikaPower 4720. The values of tensile strength determined in this test for XNR 6852 correspond to the values expected for a conventional epoxy adhesive (Table 3). In Figure 19 - Exploded view of the mold to produce plate specimens under hydrostatic pressure. Figure 20 - Dimensions of the bulk tensile specimen used in accordance with standard BS 2782 (dimensions in mm). Strength and fracture energy of adhesives for the automotive industry 29 contrast the two part epoxy, SikaPower 4720, has a low tensile strength more typical of polyurethane or a natural rubber. On top of it, the maximum strain is small and is far from the 100% strain of XNR 6852. In conclusion, XNR 6852, has a maximum strain much higher than a conventional toughened epoxy adhesive and a higher strength than a polyurethane adhesive. The stress-strain curves of polymeric materials are not linear in tension and have usually low rigidity in the elastic domain. Despite the evident non-linear behaviour, the Figure 21 - Stress-strain curve with, Pliogrip 7400/7410 (PU), AV 119 (toughened epoxy), XNR6852 and SikaPower 4720 Figure 22 - True Stress-True strain curve with, Pliogrip 7400/7410 (PU), AV 119 (toughened epoxy), XNR6852 and SikaPower 4720. Failure strength tests 30 Young´s modulus is used to describe most adhesive as it is simple to determine. It is worth mentioning that the shear modulus is relatively linear.[5] The Young’s modulus obtained for XNR 6852 is approximately half of a typical toughened epoxy (Table 3) and it is a consequence of the addition of the phenols. This property can have some advantages to the vibration damping [21] because of its smaller rigidity. On the other hand, SikaPower 4720 has a normal Young’s modulus for a toughened epoxy. Table 3 - Results of bulk tensile tests Before fracturing, adhesive XNR 6852 deforms in a ductile manner (Figure 23, a) suffering a reduction of area and acquiring an opaque colour, behaviour typical of thermoplastic polymers. This behaviour is an improvement in the properties of epoxy adhesives demonstrating an increased ductility of the material. As for the SikaPower 4720, it has a very fragile behaviour with little deformation (Figure 23, b). 3.1.3 Optimization of the bulk tensile test specimen a) Experimental procedure Specimen manufacture In order to increase productivity of specimens for the durability project that follows a reduction of specimen size is required. The reason is the many hours that take to cure the adhesive and the low number of long dogbone specimens produced with a single a) b) Figure 23 – Bulk tensile specimens after test, a) XNR 6852 and b) SikaPower 4720 Strength and fracture energy of adhesives for the automotive industry 31 bulk plate. To produce the short specimens the same manufacturing process as for the long dogbone was employed. A first attempt was made with standard EN ISO 572-2, short specimen, represented in Figure 24. This geometry is suited for ductile adhesive such as polyurethanes. The geometry in Figure 25 is not in accordance with any standard and was developed with the purpose of eliminating, as far as possible, the concentration of stress. In order to do so, a less abrupt transition was used with higher (double) radius. A further improvement of the previous geometry with an increased radius and increased cross section area was developed. Figure 26 shows the geometry of this specimen. In all cases, the specimen thickness was 2 mm. Figure 24 – Short tensile specimen according to EN ISO 527-2 (dimensions in mm) Figure 25 – Short specimen with transition radius of 25 (dimensions in mm) Figure 26 – Short specimen with a 54 radius in the transition area (dimensions in mm) Fracture tests 38 The shear modulus was calculated with Equation 14 using a Poisson’s ratio (ν) of 0.4. It is only valid for simple stress fields with no shear. 𝐺𝑥𝑦=𝐸 2(1+𝜈) (21) A general/static analysis was chosen in the step module library. Also, an output field for the degradation of the adhesive was programed in order to evaluate the adhesive condition. No interactions were input in the program. The boundary conditions are presented in Figure 35, c). The specimen arms were modelled with plane-strain 4-node quadrilateral solid (CPE4R: A 4-node bilinear plane strain quadrilateral, reduced integration, hourglass control). In order to have a square cohesive finite element with a 0.2 mm edge, the seed edges function was used. The mesh was constructed taking advantage of the automatic capabilities of Abaqus®. a) Cohesive zone model The experimental values of toughness and strength present in Table 5 were used to configure the triangular CZM model presented below. Figure 35 – Modelled DCB specimen with the finite element mesh, a) cohesive zone (red), b) view of the all specimen, c) boundary conditions. Strength and fracture energy of adhesives for the automotive industry 39 The area under the traction separation law, in both mode I and mode II, is equal to the respective fracture energy. Under pure mode I or mode II, damage propagation occurs at a specific integration when the stress is released in the traction-separation law. Initially it assumes a linear elastic behaviour followed by a linear evolution of damage. Although it is pure mode I crack propagation, an energy criterion was chosen for mixed mode crack propagation. The linear energetic criterion for complete separation chosen is presented in Equation 22. 𝐺𝐼 𝐺𝐼𝐶+𝐺𝐼𝐼 𝐺𝐼𝐼𝐶=1 (22) b) Data analysis To calculate the critical fracture energy in mode I, GIC, three different methods were used: Compliance Calibration Method (CCM), Compliance Beam theory and Compliance-based beam method (CBBM). Figure 36 - Traction separation law with linear softening available in Abaqus® Fracture tests 40 4.1.2 Numerical results and discussion a) Initial crack length Three initial crack lengths were studied to investigate the effect on the toughness measurement. The three initial cracks used in Abaqus® were: 20 mm, 56 mm and 120 mm. The P-δ curve presented below shows a high rigidity for the specimen with a short crack, a softer linear loading for the intermediate and a further decrease in compliance for the bigger crack. Also the maximum load for each crack decreased accordingly to the crack length. All the graphs coincide once the crack length is the same. To complement the P-δ curve Figure 38 shows the variation of the applied load with the crack length. Again, for the same crack length the applied load coincides. Figure 37 – Numerical P-δ of three different initial cracks. Figure 38 – P-a curve for the three different initial crack lengths. Strength and fracture energy of adhesives for the automotive industry 41 a) b) c) Figure 39 – Numerical R-curves of the three different initial cracks lengths, a) CCM method, b) CBT method, c) CBBM method. The R-curves calculated from the numerical data acquired are presented below. The CCM and the CBT have a peak in the initial toughness (Figure 39). This peak is higher in the CCM method but rapidly tends to the exact value. In contrast the CBT has a smaller peak but a smaller slope until the exact value. Lastly the CBBM is highly Fracture tests 42 influenced by the initial crack length and, in the plateau region, the value of fracture toughness does not converge. The fracture toughness was calculated using the plateau region of the R-curve and the results are compiled in Figure 40. From all the methods used the CBBM is the most affected by the initial crack length with an increased toughness. Although CBT has a similar behaviour, it is more consistent and is not affected to the same extend as CBBM method. The accurate method was CCM giving precise results in all three tests with different initial crack lengths. Figure 40 - Summary of the numerical fracture toughness as a function of initial crack length. Strength and fracture energy of adhesives for the automotive industry 43 b) Specimen length A crucial parameter is the length of the specimen because the chamber has a limitation of 120 mm from the loading line. As a result, three specimen lengths were used, 290 mm, 240 mm and 200 mm. The P-δ curve of the simulation is presented below. To complement the previous graph, the force versus crack length is presented in Figure 42. Figure 41 - Numerical P-δ of three different specimen length. Figure 42 – P-a curve for the three different specimen length. Fracture tests 44 a) b) c) Figure 43 – Numerical R-curves of the three different specimen lengths, a) CCM method, b) CBT method, c) CBBM method. The results from the Abaqus® simulations are similar and only small variations between the three lengths can be perceived in the Figure 41 and Figure 42. Strength and fracture energy of adhesives for the automotive industry 45 The results from the CCM method are the only that have a shape with a big deviation from the normal specimen. In the two other methods the difference in the shapes of the R-curves is only due to the breaking of the specimen. The fracture toughness is in this case almost constant and it can be assumed that there is little influence of this parameter. In conclusion, if it is guaranteed enough length for a stable propagation then the fracture toughness in the plateau region is not influenced by the length of the specimen. Figure 44 - Summary of the fracture toughness as a function of specimen length. Fracture tests 46 c) Specimen width Three different widths were used, 50 mm, 25 mm and 10 mm. The failure load increased proportionally with the width and is consistent with the FEA formulation (Figure 45). In other words, there is no influence in the simulation and the parameters that influence it. Figure 45 – Numerical P-δ curve for three different widths. Figure 46 – Numerical P-a curve with three different widths. Strength and fracture energy of adhesives for the automotive industry 47 In all methods the shape of the R-curves was identical as was the fracture toughness in the plateau region (Figure 47). The decrease of the width has little influence on the numerical results. However, diffusion in the bondline of the DCB specimen can take up to two years and a reduction of the section can decrease greatly the period of the durability study. Figure 47 – Comparison of the numerical fracture toughness in the plateau region of the R-curve for three different widths and three different methods: CCM, CBT and CBBM. Fracture tests 54 f) Final specimen After the evaluation of different parameters a final specimen was designed. The small specimen geometry is presented in Figure 56 and a similar finite element analysis as the previously presented was conducted. Both P-δ curves shown below have a similar shape and although there is a smaller initial crack in the short specimen the force up to failure is reduced due to the smaller width. Figure 57 - Numerical P-δ of a normal and a small DCB specimen. Figure 58 – P-a curve for the normal and a small DCB specimen. Figure 56 – Geometry of the small DCB specimen; final specimen (dimensions in mm) Strength and fracture energy of adhesives for the automotive industry 55 a) b) c) Figure 59 – Numerical R-curves for the normal and small DCB specimen, a) CCM method, b) CBT method, c) CBBM method. From the three models used to calculate fracture toughness only the CBT and CBBM worked and gave reasonable R-curves (Figure 43). Fracture tests 56 The shape of the CCM method presented in Figure 59, a), does not have either a plateau or the shape that would be expected for stable damage propagation in the adhesive. In contrast, the two other methods accurately calculated the fracture toughness of the adhesive (Table 7). The stresses in the specimen were studied. The tensions in the final specimen are equal to the normal DCB specimen and as a result the same steel can be used effectively. The distribution of tension in the final specimen is presented below. The final specimen has the required dimensions of less than 120 mm from the loading line to the wall and will fit in the chamber to be used in the subsequent durability project. Figure 60 - von Mises stresses during the final DCB test simulation. Table 7 - Summary of the fracture toughness calculated using the CCM, CBT and CBBM. Strength and fracture energy of adhesives for the automotive industry 57 4.2 Experimental DCB tests 4.2.1 Experimental procedure a) Adhesive The epoxy adhesive SikaPower 4720 was used, supplied by SIKA® (Portugal, Vila Nova de Gaia). This adhesive is a two-part system that cures at room temperature for 24 hours. It has a tensile strength of 25 MPa and an elongation at break of 4%. b) Substrates In the DCB tests, a high tensile strength steel (DIN 40 CrMnMo 7) was used to avoid plastic deformation of the substrates. The general properties of the steels used are presented in Table 8. c) Specimen manufacture The geometry of the normal and final DCB test specimens of the numerical study of this thesis were used and are given again in Figure 61 for convenience. In order to prepare the specimens, firstly the surface of the substrates was grit blasted and degreased with acetone prior to the application of adhesive. To guarantee the adhesive bondline thickness, spacers were inserted between the adherends on both ends. On one end, two steel plates and a razor blade of 0.1 mm was inserted to introduce a pre-crack and guarantee cohesive failure propagation from the beginning of the test. On the other end, one steel plate was inserted to guarantee a bondline thickness of 0.2mm. Adhesive was applied in both adherends before assembly and were set in a mold for correct alignment while curing (Figure 62). Lastly, the joints were left under 2 MPa pressure for 24h at room temperature in a hydraulic hot plates press. After curing the spacers were removed along with any excess adhesive. The bondline thickness was controlled using an optic microscope. Table 8 - Mechanical properties of the steel used for the substrates of the DCB specimens. Fracture tests 58 d) Test procedure The specimen were tested according to standard ASTM D3433 in a INSTRON® model 3367 universal test machine (Norwood, Massachusetts, USA) with a capacity of 30 kN, at room temperature and constant displacement rate of 0.2, 0.5 and 2 mm/min. The specimen was loaded to measure the behaviour of the adhesive to fracture in mode I. Pictures were recorded during the testing with 5 s intervals using a 10 MPixel digital camera. These images allowed the measurement of the crack length during its growth. Loads and displacements were recorded up to failure. Four specimens were tested for each geometry. a) b) Figure 61 - Geometry of the DCB specimens tested, a) small specimen, b) normal specimen. Figure 62 - Schematic representation of the mold used to cure the DCB specimens with the respective legend. Strength and fracture energy of adhesives for the automotive industry 59 4.2.2 Experimental results and discussion of DCB tests a) Characterization of fracture toughness Four normal DCB tests were conducted with a displacement rate of 0.5 mm/min to characterize the adhesive toughness in mode I. One typical P-δ curve obtained with this method is presented in Figure 63. For each normal specimen an R-curve was calculated similar to the curve shown in Figure 64. The final value of 1.53 N/mm (value determined using the average of the techniques showed in Table 9) is much higher than conventional toughened epoxy adhesives (0.3- 0.6 N/mm) and comparable to that of a polyurethane adhesive (1.2-2.9 N/mm). [28] Table 9 – Values of the fracture toughness of adhesive SikaPower 4720 using the normal specimen. Figure 64 – Example of an R-curve obtained, specimen 4. Figure 63 – Example of the P-δ obtained, specimen 4. Fracture tests 60 1.50 1.75 2.00 0.2 mm/min 0.5 mm/min 2 mm/min GIc (N/mm) Test displacement rate (mm/min) b) Displacement rate Three displacement rates were compared, 0.2, 0.5 and 2 mm/min to study the effect of strain rate on the fracture toughness (Figure 65). For the two lowest displacement rates, 0.2 and 0.5 mm/min, the values of fracture toughness were almost the same (Figure 64). On the other hand the displacement rate of 2 mm/min had an increase in toughness which can be a result of the adhesive’s viscoelastic behaviour. The rupture was cohesive in all tests and a picture of the facture surface is presented in Figure 67. As a note, the fracture surface of the DCB specimens with 2 mm/min displacement had less rugosity than the specimens tested with slower velocities and is further proof that viscoelastic behaviour has to be taken into account. Figure 65 - Comparison of three R-curves using the CBBM method for different velocities. Figure 66 – Comparison of the fracture toughness of SikaPower 4720 with different displacement rates. Strength and fracture energy of adhesives for the automotive industry 61 c) Comparison of two different DCB specimens The initial rigidity of the specimens is not the same for all the cases (Figure 68). A part of the reason is the variation of the initial crack length for the normal and short DCB specimens. The normal specimens have a mean value of 46.35 mm (45 idealized) and a standard deviation of 0.68 for the initial crack length. There was a better result for the short specimen of 19.93 mm (20mm idealized) of crack length with a standard deviation of 0.18 mm. Figure 68 – P-δ curve for the short and normal DCB specimens tested with SIKA® 4720. Figure 67 – Example of the failure mode of DCB specimens with SikaPower 4720 using three different displacement rates, a) 0.2mm/min, b) 0.5mm/min and c) 2 mm/min. Fracture tests 62 Throughout the work done with the normal and small specimens it is clearly much easier to control the manufacturing process for the small specimens, ex. bondline thickness and the initial crack length. Only the CBBM method was used in the analysis of the small DCB specimens. The first reason was due to the better results in the numerical study and on top of it a technical difficulty. The initial crack of the small DCB specimen was obstructed by the machines holding mechanism making it impossible to monitor the beginning of the failure propagation. The initial fracture toughness is very high for the small specimen (Figure 69). Since it was a toughened adhesive the initial crack introduced by the blade may not have been sufficient to introduce a high enough stress concentration factor. Also the concept of linear elastic fracture mechanics implies that plasticity should be limited to a small region ahead of the crack tip and a long enough crack should exist. Most likely neither verify in the testing of the small specimen. There was a stable propagation in all small specimens. [11] Figure 69 – R-curve of the small specimen and normal specimen using the CBBM method. Strength and fracture energy of adhesives for the automotive industry 63 The rupture was cohesive in all tests and a picture of the facture surface is presented in Figure 70. Figure 70 - Two examples of the cohesive fracture surface of the specimens tested, a) small specimen and b) normal specimen. Strength and fracture energy of adhesives for the automotive industry 71 7. Future work Study the small bulk specimens with other adhesives and compare with the results from the long dogbone. Improve the results for the short DCB specimen studying other parameters. Validate the numerical findings of the DCB simulations. Using bulk specimens developed to study the water aging process of the both, NAGASE CHEMPTEX® XNR 6852 and SikaPower® 4720. Strength and fracture energy of adhesives for the automotive industry 73 References [1] da Silva L.F.M., Öchsner A., Adams R.D., Handbook of Adhesion Technology, 1 ed., Springer, Berlin, 2011. [2] Kinloch A.J., Adhesion and Adhesives, Science and Technology., 1 ed., Chapman and Hall, Great Britain, 1987. [3] 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 (2009) 1-18. [4] Hartshorn S.R., Structural Adhesives, Chemistry and Technology., Hartshorn, 1986. [5] Adams R.D., Wake W.C., Structural Adhesive Joints in Enginering, ELSEVIER APPLIED SCIENCE PUBLISHERS LTD, 1984. [6] Adams R.D., Adhesive bonding, Science, technology and applications, Woodhead Publishing Limited, 2000. [7] Lees W.A., Adhesives in Engineering Design, Springer-Verlag, The Design Council, 1984. [8] da Silva L.F.M., das Neves P.J.C., Adams R.D., Spelt J.K., Analytical models of adhesively bonded joints—Part I: Literature survey, International Journal of Adhesion and Adhesives, 29 (2009) 319-30. [9] da Silva L.F.M., das Neves P.J.C., Adams R.D., Wang A., Spelt J.K., Analytical models of adhesively bonded joints—Part II: Comparative study, International Journal of Adhesion and Adhesives, 29 (2009) 331-41. [10] da Silva L.F.M., Öchsner A., Modeling of Adhesive Bonded Joints, Springer, 2008. [11] Kanninen M.F., Popelar C.H., Advanced Fracture Mechanics, Oxford University Press, 1985. [12] da Silva L.F.M., Dillard D.A., Blackman B.R.K., Adams R.D., Testing Adhesive Joints, Best practices., WILEI-VCH, 2012. [13] Kanninen M.F., Popelar C.H., Advanced Fracture Mechanics, Oxford University Press. [14] Robinson P., Das S., Mode I DCB testing of composite laminates reinforced with z-direction pins: a simple model for the investigation of data reduction strategies, Engineering Fracture Mechanics, 71 (2004) 345-64. [15] de Moura M.F.S.F., Gonçalves J.P.M., Chousal J.A.G., Campilho R.D.S.G., "Cohesive and continuum mixed-mode damage models applied to the simulation of the mechanical behaviour of bonded joints, International Journal of Adhesion and Adhesives, 2008. [16] de Moura M.F.S.F., Gonçalves J.P.M., Chousal J.A.G., Campilho R.D.S.G., Cohesive and continuum mixed-mode damage models applied to the simulation of the mechanical behaviour of bonded joints, International Journal of Adhesion and Adhesives, 28 (2008) 419-26. [17] Banea M.D., Silva L.F.M.d., Campilho R.D.S.G., Effect of Temperature on Tensile Strength and Mode I Fracture Toughness of a High Temperature Epoxy Adhesive, Journal of Adhesion Science and Technology, 26 (2012) 939-53. [18] Tsujimura T.Y., High toughness composites based on insitupolymerizable thermoplastic epoxy-resin, 29th International Conference and Forum -SAMPE Europe SEICO 08, Paris, 2008. [19] Karachalios E.F., Adams R.D., da Silva L.F.M., Single lap joints loaded in tension with high strength steel adherends, International Journal of Adhesion and Adhesives, 43 (2013) 81-95. References 74 [20] Gordon K., Quasi-static strength and fatigue life of hybrid (bonded/bolted) composite single-lap joints, Composite Structures, 72 (2006) 119-29. [21] Banea M.D., da Silva L.F.M., Static and fatigue behaviour of room temperature vulcanising silicone adhesives for high temperature aerospace applications. Statisches Verhalten und Dauerfestigkeitsanalyse von vulkanisierten Silikonklebstoffen für Luftfahrtanwendungen bei hohen Temperaturen, Materialwissenschaft und Werkstofftechnik, 41 (2010) 325-35. [22] Wang Y., Williams J.G., Corrections for mode II fracture toughness specimens of composites materials, Composites Science and Technology, 43 (1992) 251-6. [23] Yoshihara H., Kawamura T., Mode I fracture toughness estimation of wood by DCB test, Composites Part A: Applied Science and Manufacturing, 37 (2006) 2105-13. [24] 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 (2008) 2224-30. [25] 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, 2008. [26] Banea M.D., L. F. M. Silva, and R. D. S. G. Campilho, Effect of temperature on tensile strength and mode I fracture toughness of a high temperature epoxy adhesive, Journal of adhesivion science and technology, 2011. [27] Yan C., Mai Y.-W., Yuan Q., Ye L., Sun J., Effects of substrate materials on fracture toughness measurement in adhesive joints, International Journal of Mechanical Sciences, 43 (2001) 2091-102. [28] da Silva L.F.M., Carbas R.J.C., Critchlow G.W., Figueiredo M.A.V., Brown K., Effect of material, geometry, surface treatment and environment on the shear strength of single lap joints, International Journal of Adhesion and Adhesives, 29 (2009) 621-32. Strength and fracture energy of adhesives for the automotive industry 75 Appended papers