- Email: [email protected]
Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach
Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Contents lists available at ScienceDirect
Theoretical and Applied Fracture Mechanics journal homepage: www.elsevier.com/locate/tafmec
Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach V. Anes a,b, R. Pedro c, E. Henriques a, M. Freitas a, L. Reis a,⇑ a
IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisbon, Portugal GI-MOSM, Instituto Superior de Engenharia de Lisboa, Instituto Politécnico de Lisboa, Rua Conselheiro Emídio Navarro, 1, 1959-007 Lisbon, Portugal c TAP Maintenance & Engineering, Hangar 5, 1704-801 Lisbon, Portugal b
a r t i c l e
i n f o
Article history: Received 15 May 2016 Revised 25 July 2016 Accepted 19 August 2016 Available online xxxx Keywords: Adhesive selection Adhesive micro cracks Dissimilar materials Bonded joint Low temperatures Stress analysis
a b s t r a c t Maintenance, repair and overhaul companies have been reporting corrosion failure events in the Airbus A320 CFM56-5b intakes. These intakes are attached to the power plant frame by a dissimilar material bonded joint, where liquid shim adhesive is used to avoid the dielectric formation between dissimilar materials. In previous works, the authors reported that the A320 intakes corrosion is a result of the adhesive inability to avoid the dielectric formation, which is a result of micro-cracks formation within the adhesive layer. The main reason that lead to these cracks is the adhesive aging and thermal cycling at very low temperatures, which quite often reach values lower than 50 °C. This paper studies the effect of negative thermal loading on dissimilar materials bonded joints. Two epoxy adhesives are studied and compared, namely the Hysol EA-934, which is the adhesive currently used in the A320 Airbus intakes, and the Hysol EA-9394, a second generation adhesive candidate to replace the actual adhesive. A numerical study was performed in order to simulate the adhesive joint using a finite element analysis commercial package, where several hypotheses were explored. These hypotheses where focused on the effects of several factors on the adhesive layer stress distribution. Factors such as temperature range, boundary conditions, variation of the coefficient of thermal expansion with temperature, and interfacial cracks between the adhesive layer and dissimilar adherend materials were analyzed. Results show that very low temperatures have a negative impact on the adhesives strength and micro-cracks formation may result from thermal loads below zero degrees Celsius, even for adhesives without any aging. Moreover, low temperatures in dissimilar materials bonded joints may create stress states that surpass the adhesive lap shear strength. Some conclusions are drawn regarding adhesive selection for dissimilar materials bonded joints in order to overcome these issues. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Several complex repair events related to adhesive failure of dissimilar bonded joints have been reported by maintenance, repair, and overhaul companies. Particularly, the Hysol EA-934, a first generation adhesive, has been related to galvanic corrosion between the Al 2024-T3 and the Ti6Al4v titanium alloy, which are two dissimilar adherend materials [1]. These two metallic materials have different electro-negativities, which is a suitable condition to spark the galvanic corrosion process if a dielectric between them is created. Besides the gap filling task, the main feature of the Hysol EA 934 adhesive is to avoid the direct contact between dissimilar metals and create a barrier to dielectric pro-
⇑ Corresponding author.
moters such as moisture, water, and corrosive agents. Usually, aerospace epoxy adhesives are able to endure very high temperatures, they may cure at a wide range of temperatures ranging from 70 to 177 °C. Moreover, they are able to maintain their performance beyond the adhesive glass temperature, they do not melt and in many cases, service temperatures higher than their glass temperature increases their ductility and peel strength [2]. However, high temperatures are used in repair actions, for example the adherends separation of an adhesively bound can be made by heating the joint over 200 °C in order to brake the chemical links and to allow the mechanical separation using a palette-knife, which is difficult task to make and normally a non-advised procedure in aerospace due to side effects in the structure. On the other side, low temperatures can be problematic for some adhesives, especially in the case of the so-called first generation adhesives. Low temperatures significantly promotes the reduction of these
E-mail address: [email protected] (L. Reis). http://dx.doi.org/10.1016/j.tafmec.2016.08.012 0167-8442/Ó 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
2
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
adhesives strength by reducing their ductility. One evidence of the low temperatures effect on the first generation adhesives strength is commonly experienced in the shop during repair actions. The Airbus A320 structural repair manual (SRM) demand the use of dry ice to brake the adhesive chemical bounds. The dry ice is placed over the bonded joint and with a palette-knife the two adherends are easily separated. This process is much more easy to perform and safe than the heating/palette-knife process. Moreover, no side effects such as warpage or change on the metallic parts quenching state are created. The problem is that during service, aircraft easily experiences temperatures lower than 50 °C. These very low temperatures make the adhesive brittle and weakens their chemical bonds during service. In this situation the adhesive integrity can be easily reduced due to the structural forces transmitted to the bonded joint, which can create micro-cracks in the adhesive layer. The hot temperatures required to break the adhesive bond are unlikely to be found in the field (up to 200 °C), normally aircraft structures are not subject to such high temperatures. However, very low temperatures able to break adhesive bonding can be often experienced by aircraft, especially in the case of first generation adhesives. In literature, it can be found several works related to dissimilar materials joints. The most common examples are magnesium to aluminum, and composite to aluminum adhesive joints [3–5]. In these works, the research focus has been on interface cracks, high and low temperatures, residual stresses, and failure characterization [6–10]. The effect of cryogenic temperatures on the adhesive layers, especially in micro-cracking formation, has been a subject of intense research [11–14], however most of the results cannot be extrapolated from one adhesive to another. Similarly to the low temperature topic, fracture mechanics is also a very active topic in adhesives characterization. The focus has been on interface cracks, crack initiation, crack growth, and crack propagation [4,15–19]. In many of these experiments, the single lap joint of similar and dissimilar adherends uses the ASTM D1002 standard to evaluate thermal effects and adhesive crack characterization [20–22]. Also, the single lap joint has been widely used in literature to numerical modeling adhesively bonded joints, including simulation of interfacial cracks [23–25]. Developments on these simulations has been an important outcome that has been supporting the development of failure criteria of adhesive joints under complex loadings [26]. This paper studies the Hysol EA934, and Hysol EA-9394 mechanical performance in a dissimilar joint made of two dissimilar metals typically found in aerospace structures, namely an aluminum alloy, and a titanium alloy. It was found that there is a lack of knowledge in literature regarding the behavior of these two adhesives in bonded joints of dissimilar materials, especially at very low temperatures. Very few works can be found in literature regarding the Hysol EA-9394 mechanical behavior [25], but none of them focus its behavior in dissimilar materials joints at low temperatures. On the other hand, results for the EA-934 adhesive are also very scarce in literature. The Hysol EA-934 adhesive is quite old, it was launched in the market for more than 20 years and starts to appear in the aeronautic industry some concerns about its failure to barrier the galvanic corrosion between dissimilar metals.
2. Materials and methods 2.1. Materials In this study two adhesives are analyzed in order to understand and predict their mechanical behavior under very low temperatures in the case of dissimilar materials joints. The two adhesives are the Hysol EA-934 and the Hysol EA-9394 from Henkel, a worldwide German company with more than 140 years and a strong
presence in the aerospace industry. These adhesives are the socalled liquid shims and are suitable for aerospace applications, especially in the rib-to-skin assemblies. They are epoxy-based materials and their main function is to eliminate gaps between composite parts, this gaps are usually less than 3 mm. For gaps wider than 3 mm solid shims are advised. The usual key features when selecting liquid shim adhesives are the pot life (working life or gel time), compressive strength, resistance to cyclic fatigue and optimal viscosity. Table 1 shows the Hysol EA-934 and Hysol EA9394 mechanical properties at room temperature, 25 °C. Their Young’s modulus and strength are very similar. Fig. 1 shows the adhesives stress-strain behavior at room temperature (25 °C). These results were obtained using the ASTM D618 standard, and they can be found in Refs. [27,28]. The EA-9394 adhesive has lower stresses at higher strains than the EA-934 adhesive. This result makes the Hysol EA-9394 adhesive a good candidate to be used in dissimilar bonded joints because it is able to accommodate higher strains with lower stresses, it is a more ductile adhesive. Dissimilar bonded joints experience relative strains in the bond region, therefore, a good adhesive for dissimilar material bonded joint must be able to accommodate a wide range of contact strains. Table 2 shows the lap shear strength and the coefficient of thermal expansion variation with temperature for the Hysol EA-934 and Hysol EA-9394 adhesives. Fig. 2 depicts the data shown in Table 1. Fig. 2(a) shows the variation of the EA-934 and EA-9394 lap shear strength with temperature. From this graph one can conclude that the Hysol EA-9394 adhesive has a lap shear strength 15% higher than the Hysol EA934. Fig. 2(b) shows the coefficient of thermal expansion of both adhesives, the Hysol EA-9394 has a coefficient of thermal expansion 40% lower in average than the Hysol EA-934. This feature and the higher capability do accommodate higher strains reinforce the hypothesis in which the Hysol EA-9394 has improved capabilities to be used in dissimilar materials bonded joints. Table 3 shows the mechanical properties of the two dissimilar metals considered in this study, namely the 2024-T3 aluminum alloy and the 6th grade titanium alloy Ti6Al4v. These two materials are widely used in aircraft structures and often are bonded with the Hysol EA-934 epoxy adhesive.
2.2. Numerical simulations This study is strongly based on numeric simulations of the ASTM D1002 specimen test. The numerical simulations were done using ANSYS, a FEM commercial package widely used by academia and industry. The numerical model was developed based on the guidelines given in [31]. The book provides technical know-how to simulate a variety of adhesively bonded joints using ANSYS. Moreover, issues regarding convergence, mesh, loads, etc. were already validated in this book. The study main objective was to analyze the mechanical behavior of the dissimilar materials bonded joint within the temperature range 50 °C to 22 °C. This joint comprises three different materials, namely an adhesive, an aluminum alloy, and a titanium alloy, these materials were already presented in the previous sub-section. All materials have different coefficient of thermal expansion (CTE), thus this joint will experience different stress patterns within each material, the strains at contact regions will be the same for both materials in contact but stresses (assuming non slip in the materials interface) at each material will be different. Fig. 3 shows the specimen geometry used in simulations, the adhesive layer has a 0.2 mm thickness, the upper subtract is made of 2024-T3 aluminum alloy, and it is 3 mm thick and 120 mm long. The lower subtract is made of titanium and also is 3 mm thick, and 120 mm long.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
3
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx Table 1 EA-934 and EA-9394 mechanical properties at room temperature (25 °C) [27,28].
EA-934 EA-9394
Young’s modulus [MPa]
Tensile strength [MPa]
Compressive strength [MPa]
Tensile lap shear strength [MPa]
Elongation e%
3790 4237
40 46
65.5 68.9
25.5 28.9
1.2 1.66
Fig. 1. Hysol EA-934 and Hysol EA-9394 stress-strain mechanical behavior at room temperature (25 °C), modified from [27,28].
Table 2 Hysol EA-934 and Hysol EA-9394 mechanical properties [27,28]. Temp. [°C]
50 40 30 20 10 0 10 22
Hysol EA-934
Hysol EA-9394
Lap shear strength [MPa]
CTE [lm/m °C]
Lap shear strength [MPa]
CTE [lm/m °C]
21.7 22.2 22.7 23.2 23.7 24.2 24.7 25.5
44.1 49.0 53.9 58.7 63.6 68.4 73.3 80.6
23.1 23.9 24.6 25.4 26.2 27.0 27.7 28.9
31.0 34.4 37.8 41.3 44.7 48.1 51.5 56.6
In this study, the FEM simulations were performed considering a 2D approach, and the plain stress state. The tridimensional simulation of the ASTM D1002 specimen test will not bring any further insight regarding the bonded joint mechanical behavior studied here. The 2D approach will allow to optimize computational resources, which can be used to refine local mesh accordingly to the needs found during simulations. Moreover, several works in literature have been shown good results using the 2D approach in FEM simulation of adhesives [24]. 2.2.1. Numerical model A longitudinal section of the D1002 specimen is shown in Fig. 4, this section was simulated in Ansys. Two numerical models were implemented for the dissimilar materials bonded joint. The first one, simulates the adhesive joint
without any interface crack, the second model has a crack between the adhesive and the adherends, an interface crack. The first model aims to analyze the stress levels in the contact region between the adhesive and the adherends, these stress levels will be compared with the adhesive lap shear strength. In the second model, an interface crack is simulated between the adhesive and the adherends, the objective is to analyze the stress riser effect in the crack front in order to analyze the adhesive strength reduction due the interface crack presence. In this model the crack lays between the adhesive and the adherend and has a 0.1 mm wide. Fig. 5 shows the mesh performed for each model, this mesh was manually tuned in order to avoid any conflict between elements, their deformation, and aspect ratio. Fig. 5(a) shows the contact region between adhesive and adherends, Fig. 5(b) shows the mesh performed in the crack spot.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
4
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 2. Hysol EA-934 and Hysol EA-9394 mechanical properties, (a) lap shear strength, (b) coefficient of thermal expansion [27,28].
Table 3 Al 2024 T3 and Ti6Al4v mechanical properties [29,30].
Al 2024 T3 Ti6Al4V
Young’s modulus [MPa]
Yield stress [MPa]
CTE [lm/m °C]
73,100 113,800
345 880
23.2 8.6
2.2.2. Numeric models boundary conditions In both numerical models, with and without interface crack, it was performed a static simulation in order to estimate the stress distribution experienced by the adhesive layer. In simulations the room temperature was 22 °C, and the thermal load was performed by setting the final temperature within the 50 °C to 22 °C range. These simulations were performed with two different boundary conditions. In the first case, named here as boundary conditions number 1, the ASTM D1002 specimen was constrained at the most left side by constraining all degrees of freedom (constraining the titanium adherend at x = 0, please see Fig. 6(a)). In the specimen right end (at the aluminum adherend right end) the specimen is free to move in any direction. Fig. 6 shows the deformation at 50 °C as a results of the boundary condition number 1. Fig. 6(a) shows the entire specimen test and Fig. 6(b) shows the zoom of
the dissimilar material joint. The particular behavior of the dissimilar joint deformation results mainly from the different coefficient of thermal expansion of each material, which in turn lead to different thermal strains within each material. However, in the bonded region, all materials (aluminum, titanium, and epoxy adhesive) share the same thermal strains in the contact region, which is a consequence of a non-slipping contact (bonding) between materials. Moreover, the contact thermal strains in the bonding region are different from the thermal strains found on the free surface of each adherent, this feature creates a thermal strain gradient through the specimen longitudinal cross section. Therefore, due to this gradient the strongest material tend to curl the weakest material, as seen in Fig. 6. The objective of the boundary conditions number 1 is to eliminate any effect of the boundary constraints on the adhesive layer stress numerical results. This boundary condition will allow the simulation of the thermal effect on the stress distributions between the adherends and adhesive without additional influences from other sources that somehow may increase the stress level experienced by the adhesive layer. The second boundary condition, named here as boundary condition number 2, constraints all degrees of freedom at x = 0 (at the titanium adherend most left end), and constraints the UY degree of freedom at the most right
Fig. 3. ASTM D1002 specimen test, geometry and dimensions [27].
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
5
Fig. 4. Lap shear specimen simulation in 2D approach, longitudinal section of the lap shear strength specimen test (ASTM D1002).
Fig. 5. Numeric models for the dissimilar materials bonded joint, (a) numerical model without interface crack, (b) numerical model with interface crack.
Fig. 6. Lap shear specimen deformation under boundary conditions number 1 at
end of the aluminum adherend. This boundary condition lead to the deformation depicted in Fig. 7. Fig. 7(a) shows the deformation
50 °C, (a) specimen total deformation, (b) zoom on the specimen bonded area.
of the entire specimen test, and Fig. 7(b) shows the zoom of the dissimilar material bonded joint.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
6
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 7. Lap shear specimen deformation under boundary conditions number 2 at
50 °C, (a) specimen total deformation, (b) zoom on the specimen bonding area.
In this section the stress gradients for both adhesives at 50 °C are shown. The results are presented in the following way: first, the results for the Hysol EA-934 adhesive are shown and then for the Hysol EA-9394. The results demonstration starts with the stress distribution for boundary condition number 1 (fix-free), following the results for boundary condition number 2 (fix-UY constrained). Next the stress distributions of the adhesive layer with and without interface crack are shown. These results will help to understand the boundary condition influence on the adhesive stress distribution and stress level. The stress distribution are shown based on the von Mises equivalent stress. Under static loads this equivalent stress is suitable to represent biaxial stresses, which is the case of the dissimilar material joint studied here.
adhesive layer significant differences were not found, as it can be concluded by correlating Figs. 8(b) and 9(b). The interface crack stress distributions of the Hysol EA-934 adhesive are shown in Figs. 10 and 11 for the boundary conditions 1 and 2, respectively. The results correlation of both boundary conditions in the presence of a interface crack lead to conclude that the boundary condition number 2 cause higher stress levels in the D1002 specimen adherends than boundary condition number 1. However, comparing the stress levels experienced in the crack front by the adhesive layer, one can see that boundary condition number 2 cause lower stress levels on the crack front end, please see Figs. 10(b) and 11(b). Correlating the von Mises stress distribution in the Hysol EA934 adhesive layer for the cases with and without interface crack, one can conclude that only the case of boundary condition number 2 with interface crack show significant differences, showing less 10 MPa than the other loading cases analyzed in this section.
3.1.1. Hysol EA-934 stress distribution at 50 °C Fig. 8 shows the dissimilar material joint with a 0.2 mm EA-934 adhesive layer experiencing a thermal load of DT = 72 °C. The thermal load started at room temperature 22 °C and evolved until reaching 50 °C. Fig. 8(a) shows the joint deformation and its von Mises stress distribution for the boundary condition number 1 (fixfree). Fig. 8(b) shows the von Mises stress distribution for the adhesive layer only, without showing the joint adherends. Fig. 9(a) shows the dissimilar joint stress distribution for the boundary condition number 2. Comparing these results with the results depicted in Fig. 8(a), one can conclude that different boundary conditions lead to different stress distributions, however, in the
3.1.2. Hysol EA-9394 stress distributions at 50 °C Figs. 12 and 13 show the von Mises stress distribution for the Hysol EA-9394 adhesive layer subjected to boundary conditions 1 and 2, respectively. The von Mises stress distributions for the Hysol EA-9394 adhesive are very alike to the ones found in the Hysol EA-934 simulations. Also here the boundary condition do not affect the adhesive stress gradient. The similarly found between the Hysol EA-934 and Hysol EA-9394 results is explained by the similarly found in their Young’s modulus, 3.8 GPa from EA-934 against 4.2 GPa from EA-9394, thus the stress-strain relation in both adhesives is very similar in the elastic region. Differences can be found in their ten-
3. Results and discussion 3.1. Stress distributions at the bonded joint
Fig. 8. von Mises stress distribution in MPa for the Hysol EA-934 adhesive subjected to boundary condition number 1 (fix-free end) at distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
50 °C, (a) joint von Mises stress
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
7
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 9. von Mises stress distribution in MPa for the Hysol EA-934 adhesive subjected to boundary condition number 2 (fix-Uy constrained) at distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
Fig. 10. von Mises stress distribution in MPa for the EA-934 adhesive subjected to boundary condition number 1 (fix-free) at joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
50 °C with an interface crack a = 0.1 mm, (a)
Fig. 11. von Mises stress distribution in MPa for the EA-934 adhesive subjected to boundary condition number 2 (fix-Uy constrained) at a = 0.1 mm, (a) joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
sile strength, where the Hysol EA-9394 tensile strength is 15% higher than the Hysol EA-934 tensile strength. This conclusion can be extended to their lap shear strength, where the Hysol EA9394 has higher performance. Figs. 14 and 15 show the von Mises stress distribution for the Hysol EA-9394 adhesive in the cases of interface crack, and boundary conditions 1 and 2, respectively. Also here, the interface crack results are very similar in both adhesives and boundary conditions. From these results one can
50 °C, (a) joint von Mises stress
50 °C with an interface crack
conclude that the boundary conditions considered in this study do not significantly affect the adhesive stress distribution. The results shown in the previous figures are related to the 50 °C thermal loading, which is the bottom end of the temperature range considered here, which ranges from 50 °C to 22 °C. In the following sub-section, the von Mises and max shear stresses are shown for each loading conditions and for the entire temperature range. These results will allow to analyze if the pattern found at 50 °C
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
8
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 12. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 1 (fix-free) at (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
50 °C, (a) joint von Mises stress distribution,
Fig. 13. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 2 (fix-UY constrained) at distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
Fig. 14. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 1 (fix-free) at Joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
can be extrapolated to other temperatures within the temperature range considered here.
3.2. Adhesive layer stress results for the temperature range 22 °C
50 °C to
In this section the results for the loading conditions discussed in the previous section are shown for the temperature range 50 °C to 22 °C. Moreover, one additional verification is explored, in this
50 °C, (a) joint von Mises stress
50 °C with an interface crack a = 0.1 mm, (a)
section the variation of the adhesive coefficient of thermal expansion (CTE) with temperature is considered and analyzed its effect on the adhesive strength predictions. The CTE variation with temperature is difficult to obtain from literature. In the current study, the constant CTE approach effect on the adhesive layer stress distributions is analyzed. The constant CTE approach is usually performed in literature, and it is of interest understand if this approach is conservative or not in the adhesive strength estimates at very low temperatures.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
9
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 15. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 2 (fix-UY constrained) at a = 0.1 mm, (a) joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
50 °C with an interface crack
Table 4 Hysol EA-934 stress levels for constant CTE at the most left bonded region of the adhesive layer. Temp. [°C]
50 40 30 20 10 0 10 22
EA-934 [28]
Hysol EA-934 Crack - BC1 CTE constant
Hysol EA-934 No Crack – BC1 CTE constant Max von Mises [MPa]
Max Shear Sxy [MPa]
Max von Mises [MPa]
Max Shear Sxy [MPa]
Lap shear strength [MPa]
60 60 53 45 36 27 23 0
35 35 31 26 21 15 13 0
60 60 55 53 45 32 24 0
35 35 32 31 26 19 14 0
22 22 23 23 24 24 25 25
3.2.1. Hysol EA-934 stress results Table 4 shows the stress results for the EA-934, boundary condition number 1, constant CTE at room temperature (22 °C), and with and without interface crack conditions. The last column of Table 3 shows the EA-934 lap shear strength variation with temperature. Tables 5 and 6 show the maximum stress levels in the adhesive layer for the conditions of variable CTE. These results were obtained for boundary conditions 1 and 2, with and without interface crack. The results shown in Tables 5 and 6 confirm the hypothesis in which the boundary conditions selected in this study do not influences significantly the stress results on the adhesive layer. The suspicious was that the high thermal load analyzed in the previous section (about 50 °C) could somehow biased the results due to high stress in the adhesive layer, which are near the adhesive tensile strength.
3.2.2. Hysol EA-9394 stress results for the thermal load range 50 °C to 22 °C Table 7 shows the EA-9394 results for a constant CTE, boundary condition number 1 (i.e. fix-free end), and in the condition of with/ without interface crack. The same stress level pattern seen in the Hysol EA-934 case can be identified here, i.e. interface cracks do not significantly influence the stress results. Tables 8 and 9 show the Hysol EA-9394 stress results for the condition of CTE variable with temperature, and adhesive layer with/without interface crack. Also here, the results have the same pattern shown in Tables 4 and 5. Based on the results shown in Tables 4–9, one can conclude that the adherend boundary conditions has little effect on the adhesive layer stress gradient. Also, the interface crack between adherends and adhesive layer do not have a strong influence on the stress values. However, the CTE variation with temperature does have a strong effect on the adhesive layer stress estimates.
Table 5 Hysol EA-934 stress levels for the condition of a CTE variable with temperature, and specimen test without interface crack. Temp. [°C]
CTE lm/m °C [28]
BC1 Max von Mises [MPa]
Hysol EA-934 | no adhesive interface crack 50 44.14 57 40 49.00 52 30 53.86 47 20 58.72 41 10 63.58 33 0 68.44 26 10 73.30 23 22 79.14 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [28]
29 27 25 22 18 14 13 0
58 53 48 41 34 26 23 0
30 27 25 22 18 14 13 0
22 22 23 23 24 24 25 25
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
10
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Table 6 Hysol EA-934 stress levels for the condition of a CTE variable with temperature and adhesive layer with a = 0.1 mm interface crack, at most left region of the specimen bonded region. Temp. [°C]
CTE lm/m °C [28]
BC1 Max von Mises [MPa]
Hysol EA-934 | 0.1 mm adhesive layer interface crack 50 44.14 55 40 49.00 55 30 53.86 53 20 58.72 48 10 63.58 41 0 68.44 30 10 73.30 23 22 79.14 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [28]
30 27 24 21 18 13 10 0
55 55 53 49 41 31 23 0
31 27 24 21 18 13 10 0
22 22 23 23 24 24 25 25
Table 7 Hysol EA-9394 stress levels for constant CTE, at most left bonded region of the adhesive layer. Temp. [°C]
50 40 30 20 10 0 10 22
EA-9394 [27]
EA-9394 Crack - BC1 CTE constant
EA-9394 No Crack – BC1 CTE constant Max von Mises [MPa]
Max Shear Sxy [MPa]
Max von Mises [MPa]
Max Shear Sxy [MPa]
Lap shear strength [MPa]
43 44 44 38 31 25 20 0
25 26 26 22 18 15 11 0
46 46 41 39 33 27 21 0
27 26 24 23 19 16 12 0
23 24 25 25 26 27 28 29
Table 8 Hysol EA-9394 stress levels for the condition of CTE variable with temperature, and adhesive layer without interface crack. Temp. [°C]
CTE lm/m °C [27]
BC1 Max von Mises [MPa]
Hysol EA-9394 | no adhesive interface crack 50 31.02 53 40 34.43 48 30 37.85 42 20 41.26 36 10 44.68 30 0 48.10 25 10 51.51 19 22 55.61 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [27]
26 24 21 19 16 13 11 0
54 49 43 37 31 25 19 0
27 24 22 19 16 13 11 0
23 24 25 25 26 27 28 29
Table 9 Hysol EA-9394 stress levels for constant CTE, and 0.1 mm interface crack at the most left bonded region of the adhesive layer. Temp. [°C]
CTE lm/m °C [27]
BC1 Max von Mises [MPa]
Hysol EA-9394 | 0.1 mm adhesive interface crack 50 31.02 53 40 34.43 51 30 37.85 47 20 41.26 42 10 44.68 34 0 48.10 28 10 51.51 15 22 55.61 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [27]
26 23 21 18 15 11 4 0
54 51 48 42 34 28 23 0
26 24 21 18 15 11 9 0
23 24 25 25 26 27 28 29
3.2.3. CTE effect on the adhesive layer maximum stress The coefficient of thermal expansion (CTE) influence on the adhesives stress level is depicted in Figs. 16 and 17 for the cases with/without interface crack. Fig. 16 shows the results for the Hysol EA-934, and Fig. 17 shows for the Hysol EA-9394, respec-
tively. The lines depicted on these figures (Figs. 16 and 17) are based in the stress levels and loading conditions shown in Tables 4–9. In the following analysis, the full black lines of Figs. 16 and 17, indicate the adhesive lap shear strength variation with temperature, these lines were obtained based on Refs. [27,28]. The
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
11
Fig. 16. Hysol EA-934 shear stress vs shear strength results based on the data of Tables 4–6, (a) CTE constant, (b) CTE variable with temperature.
Fig. 17. Hysol EA-9394 shear stress vs shear strength results based on the data of Tables 7–9, (a) CTE constant, (b) CTE variable with temperature.
dash and dash-point lines show the numerical estimates for the maximum stress in the adhesive layer for a constant and variable CTE, respectively. The idea is to inspect if in any situation (thermal loading) the adhesive local stresses surpass the adhesive shear strength. If it is the case the adhesive will not stand for that thermal loading. This analysis is easily made by comparing the dashed lines with the black full line for each case. In this comparison, the temperature at the horizontal axis where the dashed and dashed-point lines intercept the full black line is called here as minimum allowable temperature. Therefore, for temperatures lower than the minimum allowable temperature the adhesive will experience shear stresses higher than the adhesive shear strength, thus in this situation the adhesive rupture is expected. In the constant CTE case, the dashed lines depicted in Fig. 16(a) and (b) show that the interface crack increases the minimum allowable temperature before failure from 14 °C to 8 °C. Moreover, the dash-point lines of Fig. 16(a) and (b) show that a variable CTE does not strongly influences the minimum allowable temperature before failure in both bonding cases, i.e. with and without interface crack. In the CTE variable case, the minimum allowable temperature in both bonding cases is very similar and is settle around 25 °C. For the Hysol EA-934 adhesive, entering with the CTE variation in the numeric calculations reduces the minimum allowable temperature. In the case of adhesive layer without interface crack the minimum allowable temperature is reduced from 14 °C to 25 °C, in the adhesive layer with interface crack
the minimum allowable temperature is reduced from 8 °C to 25 °C. The Hysol EA-9394 results are shown in Fig. 17 for a constant/variable CTE, and adhesive layer with/without interface crack. As seen in the Hysol EA-934 adhesive, a constant CTE lead to higher minimum allowable temperatures in both adhesive layer bonding conditions (adhesive with and without interface crack). The Hysol EA-9394 CTE variation in simulations reduce the minimum allowable temperature before failure from 30 °C to 40 °C in both adhesive bonding cases. This can be seen by correlating the interception of both dashed and dashed-point lines with the full black line (adhesive lap shear strength). As conclusion, the CTE variation with temperature reduces the estimate value for the minimum allowable temperature before failure, thus the CTE at room temperature is a more conservative approach. The CTE variation with temperature is an information difficult to obtain from literature, thus it is pointed out here that using a CTE at room temperature is a conservative approach in adhesive selection for dissimilar bonded joints. From Figs. 17 and 18, one can conclude that the Hysol EA-9394 adhesive has lower allowable temperatures before failure than the Hysol EA-934 adhesive. Also the Hysol EA-9394 adhesive is less sensitive to the CTE variation. 3.2.4. Interface crack analysis Figs. 18 and 19 show the evolution of the maximum shear stress experienced by the adhesive layer with and without interface crack
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
12
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
between the adhesive layer and the adherends. These results show that the Hysol EA-934 is sensitive to interfaces cracks if a constant CTE is considered, however, if variable CTE is considered, the Hysol EA-934 adhesive becomes insensitive to interface crack, please see Fig. 18(b). On the other side, the Hysol EA-9394 adhesive is insensitive to interface cracks for constant or variable CTE, because the minimum allowable temperature before failure is the same in both CTE cases, please see Fig. 19(a) and (b). From here, it can be conclude that the Hysol EA-9394 has better properties against micro-cracks formation at very low temperatures. Fig. 20 shows a performance comparison between the two adhesives considered in this study. This performance is evaluated in terms of marginal stresses in the temperature range 50 °C to 22 °C. Marginal stress is the adhesive allowable stress reserved for structural forces such as drag forces. This stress is calculated by subtracting the thermal stress values at each temperature from the adhesive lap shear strength at same temperature. The result is the allowable stress that the adhesive can experience before its failure. Therefore, the total load experienced by the dissimilar materials bonded joint will be the summation of thermal loads plus structural loads. Thus, it is desired to have higher marginal stresses, in order to accommodate higher service loads without experience an adhesive failure. Fig. 20(a) and (b) shows the marginal stresses for constant CTE. In both bounding conditions (with/without crack) the Hysol EA9394 marginal stresses are higher than the ones found for the Hysol EA-934 adhesive. Also, it can be clearly seen the temperature in which the adhesive failure due thermal loading is predicted, this temperature is easily identified on the lines interception with the temperature horizontal axis. Negative marginal stresses do not have physical meaning because at this point the adhesive layer already failed, they also show that the thermal stresses are higher than the adhesive lap shear strength. The full and dashed lines interception with the temperature axis, depicted in Fig. 20 (a) and (b) show no variation of temperature for the Hysol EA9394 adhesive but for the Hysol EA-934 adhesive it can be seen a variation, which confirms the Hysol EA-934 sensitivity to the interface crack for a constant CTE. Fig. 20(c) and (d) shows the marginal stresses of both adhesives for a variable CTE. The Hysol EA-9394 insensitivity to interface cracks is also confirmed here where the minimum temperature is maintained in 40 °C in both bonding cases, i.e. with and without crack. Moreover, the Hysol EA-934 marginal stresses are also lower than the EA-9394 marginal stresses.
4. Conclusion In this work the mechanical performance of two different adhesives was evaluated for the case of dissimilar materials bonded joints subjected to very low temperatures. The performance comparison was made by simulating in a commercial finite element package the ASTM D1002 specimen subjected to thermal loadings in the temperature range of 850 °C to 22 °C. The two dissimilar materials were the aluminum alloy 2024 T3 and the 6th grade titanium alloy Ti6Al4v. Bonded joints using these two materials can be found quite often in aerospace industry, one example is the joint that attaches the Airbus A320 intakes to the power plant frame. In the correlation performed herein, the influence of three factors on the adhesive strength was analyzed. First, the influence of the boundary conditions on the stress distribution experienced by the adhesive layer was analyzed. Second, the influence of an interface crack between the adhesive layer and the adherends was studied, and finally it was analyzed the influence of the variation of the coefficient of thermal expansion on the adhesive stress predictions. Regarding the boundary conditions variation, the results lead to conclude that this variation do not significantly affect the stress distribution within the adhesive layer. Despite the adherends change their deformed shape and stress distributions with the boundary condition variation, both adhesives show no sensitivity to the boundaries conditions considered herein. Regarding the bonding condition with interface crack, results show that the Hysol EA-934 adhesive is more sensitive to the interface crack presence than the Hysol EA-9394 adhesive, the Hysol EA-9394 shows no response to the interface crack presence. The CTE variation with temperature shows in both adhesives a strong influence on the adhesive stress distribution. It can be concluded that the CTE variation with temperature reduces the minimum allowable temperature before failure about 10 °C in average, comparatively to the results computed with a constant CTE at room temperature (22 °C). The Hysol EA-9394 adhesive shows an improved performance compared to the Hysol EA-934 in dissimilar materials bonded joints at low temperatures. This performance results mainly due to the Hysol EA-9394 low coefficient of thermal expansion that is more close to the adherends CTE than the Hysol EA-934 CTE is, which at room temperature it is twice the value of the Hysol EA-9394 CTE. Also, the Hysol EA-9394 higher lap shear strength and high capability to accommodate higher strains are two factors that justifies the better performance found here for the Hysol EA-9394 adhesive. In mechanical design stages of dis-
Fig. 18. Hysol EA-934 interface crack effect. (a) CTE constant, (b) CTE variable with temperature.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
13
Fig. 19. Hysol EA-9394 interface crack effect. (a) CTE constant, (b) CTE variable with temperature.
Fig. 20. EA-934 and EA-9394 marginal stress vs temperature for constant and variable CTE.
similar materials bonded joints it is advised the selection of adhesives with lower CTE in order to guarantee higher marginal stresses. Moreover, in simulations, the use of a constant CTE obtained at room temperature shows to be a conservative approach, thus the lack of information regarding the CTE variation with temperature of commercial adhesives has no deep impact on the bonded joint safety.
Acknowledgements This work was supported by FCT, through IDMEC, under LAETA project UID/EMS/50022/2013. The first author gratefully acknowledge financial support from FCT - Fundação para Ciência e Tecnologia (Portuguese Foundation for Science and Technology), for the Ph. D. Grant PD/BD/52344/2013.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
14
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
References [1] V. Anes, R.S. Pedro, E. Henriques, M. Freitas, L. Reis, Galvanic corrosion of aircraft bonded joints as a result of adhesive microcracks, Procedia Struct. Integr. 1 (2016). [2] B. Ellis et al., Chemistry and Technology of Epoxy Resins, Springer, 1993. [3] D.R. Lefebvre, D.A. Dillard, A stress singularity approach for the prediction of fatigue crack initiation in adhesive bonds. Part 1: Theory, J. Adhes. 70 (1999) 119–138. [4] D.R. Lefebvre, D.A. Dillard, J.G. Dillard, A stress singularity approach for the prediction of fatigue crack initiation in adhesive bonds. Part 2: Experimental, J. Adhes. 70 (1999) 139–154. [5] S.H. Chowdhury, D.L. Chen, S.D. Bhole, X. Cao, P. Wanjara, Lap shear strength and fatigue behavior of friction stir spot welded dissimilar magnesium-toaluminum joints with adhesive, Mater. Sci. Eng. A 562 (2013) 53–60. [6] M.D. Banea, L.F.M. da Silva, Adhesively bonded joints in composite materials: an overview, Proc. Inst. Mech. Eng. Part L J. Mater. Des. Appl. 223 (2009) 1–18. [7] Y. Ryoji, C. Sang-Bong, Efficient boundary element analysis of stress intensity factors for interface cracks in dissimilar materials, Eng. Fract. Mech. 34 (1989) 179–188. [8] L.F.M. Da Silva, R.D. Adams, Adhesive joints at high and low temperatures using similar and dissimilar adherends and dual adhesives, Int. J. Adhes. Adhes. 27 (2007) 216–226. [9] B.H. Rabin, R.L. Williamson, S. Suresh, Fundamentals of residual stresses in joints between dissimilar materials, MRS Bull. 20 (1995) 37–39. [10] S. de Freitas, J. Sinke, Failure analysis of adhesively-bonded skin-to-stiffener joints: metal-metal vs. composite-metal, Eng. Fail. Anal. (2015). [11] M. Konstantakopoulou, A. Deligianni, G. Kotsikos, Failure of dissimilar material bonded joints, Nanomater. Join. (2016) 103. [12] J.F. Timmerman, M.S. Tillman, B.S. Hayes, J.C. Seferis, Matrix and fiber influences on the cryogenic microcracking of carbon fiber/epoxy composites, Compos. Part A: Appl. Sci. Manuf. 33 (2002) 323–329. [13] B.C. Ray, Adhesion of glass/epoxy composites influenced by thermal and cryogenic environments, J. Appl. Polym. Sci. 102 (2006) 1943–1949. [14] S.G. Kalarikkal, B.V. Sankar, P.G. Ifju, Effect of cryogenic temperature on the fracture toughness of graphite/epoxy composites, J. Eng. Mater. Technol. 128 (2006) 151–157. [15] M.D. Banea, L.F.M. da Silva, R. Campilho, Effect of temperature on tensile strength and mode I fracture toughness of a high temperature epoxy adhesive, J. Adhes. Sci. Technol. 26 (2012) 939–953.
[16] Y. Ryoji, L. Jin-Qiao, X. Jin-Quan, O. Toshiaki, O. Tomoyoshi, Mixed mode fracture criteria for an interface crack, Eng. Fract. Mech. 47 (1994) 367–377. [17] A.J. Kinloch, C.F. Korenberg, K.T. Tan, J.F. Watts, Crack growth in structural adhesive joints in aqueous environments, J. Mater. Sci. 42 (2007) 6353–6370. [18] B.R.K. Blackman, A.J. Kinloch, M. Paraschi, On the mode II loading of adhesive joints, Eur. Struct. Integr. Soc. 32 (2003) 293–304. [19] S.J. Bennett, K.L. Devries, M.L. Williams, Adhesive fracture mechanics, Int. J. Fract. 10 (1974) 33–43. [20] D.R. Mulville, R.N. Vaishnav, Interfacial crack propagation, J. Adhes. 7 (1975) 215–233. [21] A.M.G. Pinto, A.G. Magalhães, R.D.S.G. Campilho, M. de Moura, A.P.M. Baptista, Single-lap joints of similar and dissimilar adherends bonded with an acrylic adhesive, J. Adhes. 85 (2009) 351–376. [22] D. Ren, L. Liu, Y. Li, Investigation on overlap joining of AZ61 magnesium alloy: laser welding, adhesive bonding, and laser weld bonding, Int. J. Adv. Manuf. Technol. 61 (2012) 195–204. [23] S.-G. Kang, M.-G. Kim, C.-G. Kim, Evaluation of cryogenic performance of adhesives using composite–aluminum double-lap joints, Compos. Struct. 78 (2007) 440–446. [24] M. Davis, D. Bond, Principles and practices of adhesive bonded structural joints and repairs, Int. J. Adhes. Adhes. 19 (1999) 91–105. [25] L.F.M. Da Silva, R.D.S.G. Campilho, Advances in Numerical Modelling of Adhesive Joints, Springer, 2012. [26] C.T. Sun, W. Qian, The use of finite extension strain energy release rates in fracture of interfacial cracks, Int. J. Solids Struct. 34 (1997) 2595–2609. [27] T.R. Guess, E.D. Reedy, M.E. Stavig, Mechanical Properties of Hysol EA-9394 Structural Adhesive, SAND95-0229, Sandia Natl. Lab, Albuquerque, New Mex, 1995. [28] G. Anderson, K. DeVries, Predicting strength of adhesive joints from test results, Int. J. Fract. (1989). [29] T. Dursun, C. Soutis, Recent developments in advanced aircraft aluminium alloys, Mater. Des. (2014). [30] G. Welsch, R. Boyer, E. Collings, Materials Properties Handbook: Titanium Alloys, 1993. [31] M. Wahab, The Mechanics of Adhesives in Composite and Metal Joints: Finite Element Analysis with ANSYS, 2014.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
Contents lists available at ScienceDirect
Theoretical and Applied Fracture Mechanics journal homepage: www.elsevier.com/locate/tafmec
Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach V. Anes a,b, R. Pedro c, E. Henriques a, M. Freitas a, L. Reis a,⇑ a
IDMEC, Instituto Superior Técnico, Universidade de Lisboa, Av. Rovisco Pais, 1, 1049-001 Lisbon, Portugal GI-MOSM, Instituto Superior de Engenharia de Lisboa, Instituto Politécnico de Lisboa, Rua Conselheiro Emídio Navarro, 1, 1959-007 Lisbon, Portugal c TAP Maintenance & Engineering, Hangar 5, 1704-801 Lisbon, Portugal b
a r t i c l e
i n f o
Article history: Received 15 May 2016 Revised 25 July 2016 Accepted 19 August 2016 Available online xxxx Keywords: Adhesive selection Adhesive micro cracks Dissimilar materials Bonded joint Low temperatures Stress analysis
a b s t r a c t Maintenance, repair and overhaul companies have been reporting corrosion failure events in the Airbus A320 CFM56-5b intakes. These intakes are attached to the power plant frame by a dissimilar material bonded joint, where liquid shim adhesive is used to avoid the dielectric formation between dissimilar materials. In previous works, the authors reported that the A320 intakes corrosion is a result of the adhesive inability to avoid the dielectric formation, which is a result of micro-cracks formation within the adhesive layer. The main reason that lead to these cracks is the adhesive aging and thermal cycling at very low temperatures, which quite often reach values lower than 50 °C. This paper studies the effect of negative thermal loading on dissimilar materials bonded joints. Two epoxy adhesives are studied and compared, namely the Hysol EA-934, which is the adhesive currently used in the A320 Airbus intakes, and the Hysol EA-9394, a second generation adhesive candidate to replace the actual adhesive. A numerical study was performed in order to simulate the adhesive joint using a finite element analysis commercial package, where several hypotheses were explored. These hypotheses where focused on the effects of several factors on the adhesive layer stress distribution. Factors such as temperature range, boundary conditions, variation of the coefficient of thermal expansion with temperature, and interfacial cracks between the adhesive layer and dissimilar adherend materials were analyzed. Results show that very low temperatures have a negative impact on the adhesives strength and micro-cracks formation may result from thermal loads below zero degrees Celsius, even for adhesives without any aging. Moreover, low temperatures in dissimilar materials bonded joints may create stress states that surpass the adhesive lap shear strength. Some conclusions are drawn regarding adhesive selection for dissimilar materials bonded joints in order to overcome these issues. Ó 2016 Elsevier Ltd. All rights reserved.
1. Introduction Several complex repair events related to adhesive failure of dissimilar bonded joints have been reported by maintenance, repair, and overhaul companies. Particularly, the Hysol EA-934, a first generation adhesive, has been related to galvanic corrosion between the Al 2024-T3 and the Ti6Al4v titanium alloy, which are two dissimilar adherend materials [1]. These two metallic materials have different electro-negativities, which is a suitable condition to spark the galvanic corrosion process if a dielectric between them is created. Besides the gap filling task, the main feature of the Hysol EA 934 adhesive is to avoid the direct contact between dissimilar metals and create a barrier to dielectric pro-
⇑ Corresponding author.
moters such as moisture, water, and corrosive agents. Usually, aerospace epoxy adhesives are able to endure very high temperatures, they may cure at a wide range of temperatures ranging from 70 to 177 °C. Moreover, they are able to maintain their performance beyond the adhesive glass temperature, they do not melt and in many cases, service temperatures higher than their glass temperature increases their ductility and peel strength [2]. However, high temperatures are used in repair actions, for example the adherends separation of an adhesively bound can be made by heating the joint over 200 °C in order to brake the chemical links and to allow the mechanical separation using a palette-knife, which is difficult task to make and normally a non-advised procedure in aerospace due to side effects in the structure. On the other side, low temperatures can be problematic for some adhesives, especially in the case of the so-called first generation adhesives. Low temperatures significantly promotes the reduction of these
E-mail address: [email protected] (L. Reis). http://dx.doi.org/10.1016/j.tafmec.2016.08.012 0167-8442/Ó 2016 Elsevier Ltd. All rights reserved.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
2
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
adhesives strength by reducing their ductility. One evidence of the low temperatures effect on the first generation adhesives strength is commonly experienced in the shop during repair actions. The Airbus A320 structural repair manual (SRM) demand the use of dry ice to brake the adhesive chemical bounds. The dry ice is placed over the bonded joint and with a palette-knife the two adherends are easily separated. This process is much more easy to perform and safe than the heating/palette-knife process. Moreover, no side effects such as warpage or change on the metallic parts quenching state are created. The problem is that during service, aircraft easily experiences temperatures lower than 50 °C. These very low temperatures make the adhesive brittle and weakens their chemical bonds during service. In this situation the adhesive integrity can be easily reduced due to the structural forces transmitted to the bonded joint, which can create micro-cracks in the adhesive layer. The hot temperatures required to break the adhesive bond are unlikely to be found in the field (up to 200 °C), normally aircraft structures are not subject to such high temperatures. However, very low temperatures able to break adhesive bonding can be often experienced by aircraft, especially in the case of first generation adhesives. In literature, it can be found several works related to dissimilar materials joints. The most common examples are magnesium to aluminum, and composite to aluminum adhesive joints [3–5]. In these works, the research focus has been on interface cracks, high and low temperatures, residual stresses, and failure characterization [6–10]. The effect of cryogenic temperatures on the adhesive layers, especially in micro-cracking formation, has been a subject of intense research [11–14], however most of the results cannot be extrapolated from one adhesive to another. Similarly to the low temperature topic, fracture mechanics is also a very active topic in adhesives characterization. The focus has been on interface cracks, crack initiation, crack growth, and crack propagation [4,15–19]. In many of these experiments, the single lap joint of similar and dissimilar adherends uses the ASTM D1002 standard to evaluate thermal effects and adhesive crack characterization [20–22]. Also, the single lap joint has been widely used in literature to numerical modeling adhesively bonded joints, including simulation of interfacial cracks [23–25]. Developments on these simulations has been an important outcome that has been supporting the development of failure criteria of adhesive joints under complex loadings [26]. This paper studies the Hysol EA934, and Hysol EA-9394 mechanical performance in a dissimilar joint made of two dissimilar metals typically found in aerospace structures, namely an aluminum alloy, and a titanium alloy. It was found that there is a lack of knowledge in literature regarding the behavior of these two adhesives in bonded joints of dissimilar materials, especially at very low temperatures. Very few works can be found in literature regarding the Hysol EA-9394 mechanical behavior [25], but none of them focus its behavior in dissimilar materials joints at low temperatures. On the other hand, results for the EA-934 adhesive are also very scarce in literature. The Hysol EA-934 adhesive is quite old, it was launched in the market for more than 20 years and starts to appear in the aeronautic industry some concerns about its failure to barrier the galvanic corrosion between dissimilar metals.
2. Materials and methods 2.1. Materials In this study two adhesives are analyzed in order to understand and predict their mechanical behavior under very low temperatures in the case of dissimilar materials joints. The two adhesives are the Hysol EA-934 and the Hysol EA-9394 from Henkel, a worldwide German company with more than 140 years and a strong
presence in the aerospace industry. These adhesives are the socalled liquid shims and are suitable for aerospace applications, especially in the rib-to-skin assemblies. They are epoxy-based materials and their main function is to eliminate gaps between composite parts, this gaps are usually less than 3 mm. For gaps wider than 3 mm solid shims are advised. The usual key features when selecting liquid shim adhesives are the pot life (working life or gel time), compressive strength, resistance to cyclic fatigue and optimal viscosity. Table 1 shows the Hysol EA-934 and Hysol EA9394 mechanical properties at room temperature, 25 °C. Their Young’s modulus and strength are very similar. Fig. 1 shows the adhesives stress-strain behavior at room temperature (25 °C). These results were obtained using the ASTM D618 standard, and they can be found in Refs. [27,28]. The EA-9394 adhesive has lower stresses at higher strains than the EA-934 adhesive. This result makes the Hysol EA-9394 adhesive a good candidate to be used in dissimilar bonded joints because it is able to accommodate higher strains with lower stresses, it is a more ductile adhesive. Dissimilar bonded joints experience relative strains in the bond region, therefore, a good adhesive for dissimilar material bonded joint must be able to accommodate a wide range of contact strains. Table 2 shows the lap shear strength and the coefficient of thermal expansion variation with temperature for the Hysol EA-934 and Hysol EA-9394 adhesives. Fig. 2 depicts the data shown in Table 1. Fig. 2(a) shows the variation of the EA-934 and EA-9394 lap shear strength with temperature. From this graph one can conclude that the Hysol EA-9394 adhesive has a lap shear strength 15% higher than the Hysol EA934. Fig. 2(b) shows the coefficient of thermal expansion of both adhesives, the Hysol EA-9394 has a coefficient of thermal expansion 40% lower in average than the Hysol EA-934. This feature and the higher capability do accommodate higher strains reinforce the hypothesis in which the Hysol EA-9394 has improved capabilities to be used in dissimilar materials bonded joints. Table 3 shows the mechanical properties of the two dissimilar metals considered in this study, namely the 2024-T3 aluminum alloy and the 6th grade titanium alloy Ti6Al4v. These two materials are widely used in aircraft structures and often are bonded with the Hysol EA-934 epoxy adhesive.
2.2. Numerical simulations This study is strongly based on numeric simulations of the ASTM D1002 specimen test. The numerical simulations were done using ANSYS, a FEM commercial package widely used by academia and industry. The numerical model was developed based on the guidelines given in [31]. The book provides technical know-how to simulate a variety of adhesively bonded joints using ANSYS. Moreover, issues regarding convergence, mesh, loads, etc. were already validated in this book. The study main objective was to analyze the mechanical behavior of the dissimilar materials bonded joint within the temperature range 50 °C to 22 °C. This joint comprises three different materials, namely an adhesive, an aluminum alloy, and a titanium alloy, these materials were already presented in the previous sub-section. All materials have different coefficient of thermal expansion (CTE), thus this joint will experience different stress patterns within each material, the strains at contact regions will be the same for both materials in contact but stresses (assuming non slip in the materials interface) at each material will be different. Fig. 3 shows the specimen geometry used in simulations, the adhesive layer has a 0.2 mm thickness, the upper subtract is made of 2024-T3 aluminum alloy, and it is 3 mm thick and 120 mm long. The lower subtract is made of titanium and also is 3 mm thick, and 120 mm long.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
3
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx Table 1 EA-934 and EA-9394 mechanical properties at room temperature (25 °C) [27,28].
EA-934 EA-9394
Young’s modulus [MPa]
Tensile strength [MPa]
Compressive strength [MPa]
Tensile lap shear strength [MPa]
Elongation e%
3790 4237
40 46
65.5 68.9
25.5 28.9
1.2 1.66
Fig. 1. Hysol EA-934 and Hysol EA-9394 stress-strain mechanical behavior at room temperature (25 °C), modified from [27,28].
Table 2 Hysol EA-934 and Hysol EA-9394 mechanical properties [27,28]. Temp. [°C]
50 40 30 20 10 0 10 22
Hysol EA-934
Hysol EA-9394
Lap shear strength [MPa]
CTE [lm/m °C]
Lap shear strength [MPa]
CTE [lm/m °C]
21.7 22.2 22.7 23.2 23.7 24.2 24.7 25.5
44.1 49.0 53.9 58.7 63.6 68.4 73.3 80.6
23.1 23.9 24.6 25.4 26.2 27.0 27.7 28.9
31.0 34.4 37.8 41.3 44.7 48.1 51.5 56.6
In this study, the FEM simulations were performed considering a 2D approach, and the plain stress state. The tridimensional simulation of the ASTM D1002 specimen test will not bring any further insight regarding the bonded joint mechanical behavior studied here. The 2D approach will allow to optimize computational resources, which can be used to refine local mesh accordingly to the needs found during simulations. Moreover, several works in literature have been shown good results using the 2D approach in FEM simulation of adhesives [24]. 2.2.1. Numerical model A longitudinal section of the D1002 specimen is shown in Fig. 4, this section was simulated in Ansys. Two numerical models were implemented for the dissimilar materials bonded joint. The first one, simulates the adhesive joint
without any interface crack, the second model has a crack between the adhesive and the adherends, an interface crack. The first model aims to analyze the stress levels in the contact region between the adhesive and the adherends, these stress levels will be compared with the adhesive lap shear strength. In the second model, an interface crack is simulated between the adhesive and the adherends, the objective is to analyze the stress riser effect in the crack front in order to analyze the adhesive strength reduction due the interface crack presence. In this model the crack lays between the adhesive and the adherend and has a 0.1 mm wide. Fig. 5 shows the mesh performed for each model, this mesh was manually tuned in order to avoid any conflict between elements, their deformation, and aspect ratio. Fig. 5(a) shows the contact region between adhesive and adherends, Fig. 5(b) shows the mesh performed in the crack spot.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
4
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 2. Hysol EA-934 and Hysol EA-9394 mechanical properties, (a) lap shear strength, (b) coefficient of thermal expansion [27,28].
Table 3 Al 2024 T3 and Ti6Al4v mechanical properties [29,30].
Al 2024 T3 Ti6Al4V
Young’s modulus [MPa]
Yield stress [MPa]
CTE [lm/m °C]
73,100 113,800
345 880
23.2 8.6
2.2.2. Numeric models boundary conditions In both numerical models, with and without interface crack, it was performed a static simulation in order to estimate the stress distribution experienced by the adhesive layer. In simulations the room temperature was 22 °C, and the thermal load was performed by setting the final temperature within the 50 °C to 22 °C range. These simulations were performed with two different boundary conditions. In the first case, named here as boundary conditions number 1, the ASTM D1002 specimen was constrained at the most left side by constraining all degrees of freedom (constraining the titanium adherend at x = 0, please see Fig. 6(a)). In the specimen right end (at the aluminum adherend right end) the specimen is free to move in any direction. Fig. 6 shows the deformation at 50 °C as a results of the boundary condition number 1. Fig. 6(a) shows the entire specimen test and Fig. 6(b) shows the zoom of
the dissimilar material joint. The particular behavior of the dissimilar joint deformation results mainly from the different coefficient of thermal expansion of each material, which in turn lead to different thermal strains within each material. However, in the bonded region, all materials (aluminum, titanium, and epoxy adhesive) share the same thermal strains in the contact region, which is a consequence of a non-slipping contact (bonding) between materials. Moreover, the contact thermal strains in the bonding region are different from the thermal strains found on the free surface of each adherent, this feature creates a thermal strain gradient through the specimen longitudinal cross section. Therefore, due to this gradient the strongest material tend to curl the weakest material, as seen in Fig. 6. The objective of the boundary conditions number 1 is to eliminate any effect of the boundary constraints on the adhesive layer stress numerical results. This boundary condition will allow the simulation of the thermal effect on the stress distributions between the adherends and adhesive without additional influences from other sources that somehow may increase the stress level experienced by the adhesive layer. The second boundary condition, named here as boundary condition number 2, constraints all degrees of freedom at x = 0 (at the titanium adherend most left end), and constraints the UY degree of freedom at the most right
Fig. 3. ASTM D1002 specimen test, geometry and dimensions [27].
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
5
Fig. 4. Lap shear specimen simulation in 2D approach, longitudinal section of the lap shear strength specimen test (ASTM D1002).
Fig. 5. Numeric models for the dissimilar materials bonded joint, (a) numerical model without interface crack, (b) numerical model with interface crack.
Fig. 6. Lap shear specimen deformation under boundary conditions number 1 at
end of the aluminum adherend. This boundary condition lead to the deformation depicted in Fig. 7. Fig. 7(a) shows the deformation
50 °C, (a) specimen total deformation, (b) zoom on the specimen bonded area.
of the entire specimen test, and Fig. 7(b) shows the zoom of the dissimilar material bonded joint.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
6
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 7. Lap shear specimen deformation under boundary conditions number 2 at
50 °C, (a) specimen total deformation, (b) zoom on the specimen bonding area.
In this section the stress gradients for both adhesives at 50 °C are shown. The results are presented in the following way: first, the results for the Hysol EA-934 adhesive are shown and then for the Hysol EA-9394. The results demonstration starts with the stress distribution for boundary condition number 1 (fix-free), following the results for boundary condition number 2 (fix-UY constrained). Next the stress distributions of the adhesive layer with and without interface crack are shown. These results will help to understand the boundary condition influence on the adhesive stress distribution and stress level. The stress distribution are shown based on the von Mises equivalent stress. Under static loads this equivalent stress is suitable to represent biaxial stresses, which is the case of the dissimilar material joint studied here.
adhesive layer significant differences were not found, as it can be concluded by correlating Figs. 8(b) and 9(b). The interface crack stress distributions of the Hysol EA-934 adhesive are shown in Figs. 10 and 11 for the boundary conditions 1 and 2, respectively. The results correlation of both boundary conditions in the presence of a interface crack lead to conclude that the boundary condition number 2 cause higher stress levels in the D1002 specimen adherends than boundary condition number 1. However, comparing the stress levels experienced in the crack front by the adhesive layer, one can see that boundary condition number 2 cause lower stress levels on the crack front end, please see Figs. 10(b) and 11(b). Correlating the von Mises stress distribution in the Hysol EA934 adhesive layer for the cases with and without interface crack, one can conclude that only the case of boundary condition number 2 with interface crack show significant differences, showing less 10 MPa than the other loading cases analyzed in this section.
3.1.1. Hysol EA-934 stress distribution at 50 °C Fig. 8 shows the dissimilar material joint with a 0.2 mm EA-934 adhesive layer experiencing a thermal load of DT = 72 °C. The thermal load started at room temperature 22 °C and evolved until reaching 50 °C. Fig. 8(a) shows the joint deformation and its von Mises stress distribution for the boundary condition number 1 (fixfree). Fig. 8(b) shows the von Mises stress distribution for the adhesive layer only, without showing the joint adherends. Fig. 9(a) shows the dissimilar joint stress distribution for the boundary condition number 2. Comparing these results with the results depicted in Fig. 8(a), one can conclude that different boundary conditions lead to different stress distributions, however, in the
3.1.2. Hysol EA-9394 stress distributions at 50 °C Figs. 12 and 13 show the von Mises stress distribution for the Hysol EA-9394 adhesive layer subjected to boundary conditions 1 and 2, respectively. The von Mises stress distributions for the Hysol EA-9394 adhesive are very alike to the ones found in the Hysol EA-934 simulations. Also here the boundary condition do not affect the adhesive stress gradient. The similarly found between the Hysol EA-934 and Hysol EA-9394 results is explained by the similarly found in their Young’s modulus, 3.8 GPa from EA-934 against 4.2 GPa from EA-9394, thus the stress-strain relation in both adhesives is very similar in the elastic region. Differences can be found in their ten-
3. Results and discussion 3.1. Stress distributions at the bonded joint
Fig. 8. von Mises stress distribution in MPa for the Hysol EA-934 adhesive subjected to boundary condition number 1 (fix-free end) at distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
50 °C, (a) joint von Mises stress
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
7
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 9. von Mises stress distribution in MPa for the Hysol EA-934 adhesive subjected to boundary condition number 2 (fix-Uy constrained) at distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
Fig. 10. von Mises stress distribution in MPa for the EA-934 adhesive subjected to boundary condition number 1 (fix-free) at joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
50 °C with an interface crack a = 0.1 mm, (a)
Fig. 11. von Mises stress distribution in MPa for the EA-934 adhesive subjected to boundary condition number 2 (fix-Uy constrained) at a = 0.1 mm, (a) joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherents.
sile strength, where the Hysol EA-9394 tensile strength is 15% higher than the Hysol EA-934 tensile strength. This conclusion can be extended to their lap shear strength, where the Hysol EA9394 has higher performance. Figs. 14 and 15 show the von Mises stress distribution for the Hysol EA-9394 adhesive in the cases of interface crack, and boundary conditions 1 and 2, respectively. Also here, the interface crack results are very similar in both adhesives and boundary conditions. From these results one can
50 °C, (a) joint von Mises stress
50 °C with an interface crack
conclude that the boundary conditions considered in this study do not significantly affect the adhesive stress distribution. The results shown in the previous figures are related to the 50 °C thermal loading, which is the bottom end of the temperature range considered here, which ranges from 50 °C to 22 °C. In the following sub-section, the von Mises and max shear stresses are shown for each loading conditions and for the entire temperature range. These results will allow to analyze if the pattern found at 50 °C
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
8
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 12. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 1 (fix-free) at (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
50 °C, (a) joint von Mises stress distribution,
Fig. 13. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 2 (fix-UY constrained) at distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
Fig. 14. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 1 (fix-free) at Joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
can be extrapolated to other temperatures within the temperature range considered here.
3.2. Adhesive layer stress results for the temperature range 22 °C
50 °C to
In this section the results for the loading conditions discussed in the previous section are shown for the temperature range 50 °C to 22 °C. Moreover, one additional verification is explored, in this
50 °C, (a) joint von Mises stress
50 °C with an interface crack a = 0.1 mm, (a)
section the variation of the adhesive coefficient of thermal expansion (CTE) with temperature is considered and analyzed its effect on the adhesive strength predictions. The CTE variation with temperature is difficult to obtain from literature. In the current study, the constant CTE approach effect on the adhesive layer stress distributions is analyzed. The constant CTE approach is usually performed in literature, and it is of interest understand if this approach is conservative or not in the adhesive strength estimates at very low temperatures.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
9
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Fig. 15. von Mises stress distribution in MPa for the EA-9394 adhesive subjected to boundary condition number 2 (fix-UY constrained) at a = 0.1 mm, (a) joint von Mises stress distribution, (b) zoom on the adhesive layer stress distribution without showing the joint adherends.
50 °C with an interface crack
Table 4 Hysol EA-934 stress levels for constant CTE at the most left bonded region of the adhesive layer. Temp. [°C]
50 40 30 20 10 0 10 22
EA-934 [28]
Hysol EA-934 Crack - BC1 CTE constant
Hysol EA-934 No Crack – BC1 CTE constant Max von Mises [MPa]
Max Shear Sxy [MPa]
Max von Mises [MPa]
Max Shear Sxy [MPa]
Lap shear strength [MPa]
60 60 53 45 36 27 23 0
35 35 31 26 21 15 13 0
60 60 55 53 45 32 24 0
35 35 32 31 26 19 14 0
22 22 23 23 24 24 25 25
3.2.1. Hysol EA-934 stress results Table 4 shows the stress results for the EA-934, boundary condition number 1, constant CTE at room temperature (22 °C), and with and without interface crack conditions. The last column of Table 3 shows the EA-934 lap shear strength variation with temperature. Tables 5 and 6 show the maximum stress levels in the adhesive layer for the conditions of variable CTE. These results were obtained for boundary conditions 1 and 2, with and without interface crack. The results shown in Tables 5 and 6 confirm the hypothesis in which the boundary conditions selected in this study do not influences significantly the stress results on the adhesive layer. The suspicious was that the high thermal load analyzed in the previous section (about 50 °C) could somehow biased the results due to high stress in the adhesive layer, which are near the adhesive tensile strength.
3.2.2. Hysol EA-9394 stress results for the thermal load range 50 °C to 22 °C Table 7 shows the EA-9394 results for a constant CTE, boundary condition number 1 (i.e. fix-free end), and in the condition of with/ without interface crack. The same stress level pattern seen in the Hysol EA-934 case can be identified here, i.e. interface cracks do not significantly influence the stress results. Tables 8 and 9 show the Hysol EA-9394 stress results for the condition of CTE variable with temperature, and adhesive layer with/without interface crack. Also here, the results have the same pattern shown in Tables 4 and 5. Based on the results shown in Tables 4–9, one can conclude that the adherend boundary conditions has little effect on the adhesive layer stress gradient. Also, the interface crack between adherends and adhesive layer do not have a strong influence on the stress values. However, the CTE variation with temperature does have a strong effect on the adhesive layer stress estimates.
Table 5 Hysol EA-934 stress levels for the condition of a CTE variable with temperature, and specimen test without interface crack. Temp. [°C]
CTE lm/m °C [28]
BC1 Max von Mises [MPa]
Hysol EA-934 | no adhesive interface crack 50 44.14 57 40 49.00 52 30 53.86 47 20 58.72 41 10 63.58 33 0 68.44 26 10 73.30 23 22 79.14 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [28]
29 27 25 22 18 14 13 0
58 53 48 41 34 26 23 0
30 27 25 22 18 14 13 0
22 22 23 23 24 24 25 25
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
10
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
Table 6 Hysol EA-934 stress levels for the condition of a CTE variable with temperature and adhesive layer with a = 0.1 mm interface crack, at most left region of the specimen bonded region. Temp. [°C]
CTE lm/m °C [28]
BC1 Max von Mises [MPa]
Hysol EA-934 | 0.1 mm adhesive layer interface crack 50 44.14 55 40 49.00 55 30 53.86 53 20 58.72 48 10 63.58 41 0 68.44 30 10 73.30 23 22 79.14 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [28]
30 27 24 21 18 13 10 0
55 55 53 49 41 31 23 0
31 27 24 21 18 13 10 0
22 22 23 23 24 24 25 25
Table 7 Hysol EA-9394 stress levels for constant CTE, at most left bonded region of the adhesive layer. Temp. [°C]
50 40 30 20 10 0 10 22
EA-9394 [27]
EA-9394 Crack - BC1 CTE constant
EA-9394 No Crack – BC1 CTE constant Max von Mises [MPa]
Max Shear Sxy [MPa]
Max von Mises [MPa]
Max Shear Sxy [MPa]
Lap shear strength [MPa]
43 44 44 38 31 25 20 0
25 26 26 22 18 15 11 0
46 46 41 39 33 27 21 0
27 26 24 23 19 16 12 0
23 24 25 25 26 27 28 29
Table 8 Hysol EA-9394 stress levels for the condition of CTE variable with temperature, and adhesive layer without interface crack. Temp. [°C]
CTE lm/m °C [27]
BC1 Max von Mises [MPa]
Hysol EA-9394 | no adhesive interface crack 50 31.02 53 40 34.43 48 30 37.85 42 20 41.26 36 10 44.68 30 0 48.10 25 10 51.51 19 22 55.61 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [27]
26 24 21 19 16 13 11 0
54 49 43 37 31 25 19 0
27 24 22 19 16 13 11 0
23 24 25 25 26 27 28 29
Table 9 Hysol EA-9394 stress levels for constant CTE, and 0.1 mm interface crack at the most left bonded region of the adhesive layer. Temp. [°C]
CTE lm/m °C [27]
BC1 Max von Mises [MPa]
Hysol EA-9394 | 0.1 mm adhesive interface crack 50 31.02 53 40 34.43 51 30 37.85 47 20 41.26 42 10 44.68 34 0 48.10 28 10 51.51 15 22 55.61 0
BC1 Max Shear Sxy [MPa]
BC2 Max von Mises [MPa]
BC2 Max Shear Sxy [MPa]
Lap shear strength [MPa] [27]
26 23 21 18 15 11 4 0
54 51 48 42 34 28 23 0
26 24 21 18 15 11 9 0
23 24 25 25 26 27 28 29
3.2.3. CTE effect on the adhesive layer maximum stress The coefficient of thermal expansion (CTE) influence on the adhesives stress level is depicted in Figs. 16 and 17 for the cases with/without interface crack. Fig. 16 shows the results for the Hysol EA-934, and Fig. 17 shows for the Hysol EA-9394, respec-
tively. The lines depicted on these figures (Figs. 16 and 17) are based in the stress levels and loading conditions shown in Tables 4–9. In the following analysis, the full black lines of Figs. 16 and 17, indicate the adhesive lap shear strength variation with temperature, these lines were obtained based on Refs. [27,28]. The
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
11
Fig. 16. Hysol EA-934 shear stress vs shear strength results based on the data of Tables 4–6, (a) CTE constant, (b) CTE variable with temperature.
Fig. 17. Hysol EA-9394 shear stress vs shear strength results based on the data of Tables 7–9, (a) CTE constant, (b) CTE variable with temperature.
dash and dash-point lines show the numerical estimates for the maximum stress in the adhesive layer for a constant and variable CTE, respectively. The idea is to inspect if in any situation (thermal loading) the adhesive local stresses surpass the adhesive shear strength. If it is the case the adhesive will not stand for that thermal loading. This analysis is easily made by comparing the dashed lines with the black full line for each case. In this comparison, the temperature at the horizontal axis where the dashed and dashed-point lines intercept the full black line is called here as minimum allowable temperature. Therefore, for temperatures lower than the minimum allowable temperature the adhesive will experience shear stresses higher than the adhesive shear strength, thus in this situation the adhesive rupture is expected. In the constant CTE case, the dashed lines depicted in Fig. 16(a) and (b) show that the interface crack increases the minimum allowable temperature before failure from 14 °C to 8 °C. Moreover, the dash-point lines of Fig. 16(a) and (b) show that a variable CTE does not strongly influences the minimum allowable temperature before failure in both bonding cases, i.e. with and without interface crack. In the CTE variable case, the minimum allowable temperature in both bonding cases is very similar and is settle around 25 °C. For the Hysol EA-934 adhesive, entering with the CTE variation in the numeric calculations reduces the minimum allowable temperature. In the case of adhesive layer without interface crack the minimum allowable temperature is reduced from 14 °C to 25 °C, in the adhesive layer with interface crack
the minimum allowable temperature is reduced from 8 °C to 25 °C. The Hysol EA-9394 results are shown in Fig. 17 for a constant/variable CTE, and adhesive layer with/without interface crack. As seen in the Hysol EA-934 adhesive, a constant CTE lead to higher minimum allowable temperatures in both adhesive layer bonding conditions (adhesive with and without interface crack). The Hysol EA-9394 CTE variation in simulations reduce the minimum allowable temperature before failure from 30 °C to 40 °C in both adhesive bonding cases. This can be seen by correlating the interception of both dashed and dashed-point lines with the full black line (adhesive lap shear strength). As conclusion, the CTE variation with temperature reduces the estimate value for the minimum allowable temperature before failure, thus the CTE at room temperature is a more conservative approach. The CTE variation with temperature is an information difficult to obtain from literature, thus it is pointed out here that using a CTE at room temperature is a conservative approach in adhesive selection for dissimilar bonded joints. From Figs. 17 and 18, one can conclude that the Hysol EA-9394 adhesive has lower allowable temperatures before failure than the Hysol EA-934 adhesive. Also the Hysol EA-9394 adhesive is less sensitive to the CTE variation. 3.2.4. Interface crack analysis Figs. 18 and 19 show the evolution of the maximum shear stress experienced by the adhesive layer with and without interface crack
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
12
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
between the adhesive layer and the adherends. These results show that the Hysol EA-934 is sensitive to interfaces cracks if a constant CTE is considered, however, if variable CTE is considered, the Hysol EA-934 adhesive becomes insensitive to interface crack, please see Fig. 18(b). On the other side, the Hysol EA-9394 adhesive is insensitive to interface cracks for constant or variable CTE, because the minimum allowable temperature before failure is the same in both CTE cases, please see Fig. 19(a) and (b). From here, it can be conclude that the Hysol EA-9394 has better properties against micro-cracks formation at very low temperatures. Fig. 20 shows a performance comparison between the two adhesives considered in this study. This performance is evaluated in terms of marginal stresses in the temperature range 50 °C to 22 °C. Marginal stress is the adhesive allowable stress reserved for structural forces such as drag forces. This stress is calculated by subtracting the thermal stress values at each temperature from the adhesive lap shear strength at same temperature. The result is the allowable stress that the adhesive can experience before its failure. Therefore, the total load experienced by the dissimilar materials bonded joint will be the summation of thermal loads plus structural loads. Thus, it is desired to have higher marginal stresses, in order to accommodate higher service loads without experience an adhesive failure. Fig. 20(a) and (b) shows the marginal stresses for constant CTE. In both bounding conditions (with/without crack) the Hysol EA9394 marginal stresses are higher than the ones found for the Hysol EA-934 adhesive. Also, it can be clearly seen the temperature in which the adhesive failure due thermal loading is predicted, this temperature is easily identified on the lines interception with the temperature horizontal axis. Negative marginal stresses do not have physical meaning because at this point the adhesive layer already failed, they also show that the thermal stresses are higher than the adhesive lap shear strength. The full and dashed lines interception with the temperature axis, depicted in Fig. 20 (a) and (b) show no variation of temperature for the Hysol EA9394 adhesive but for the Hysol EA-934 adhesive it can be seen a variation, which confirms the Hysol EA-934 sensitivity to the interface crack for a constant CTE. Fig. 20(c) and (d) shows the marginal stresses of both adhesives for a variable CTE. The Hysol EA-9394 insensitivity to interface cracks is also confirmed here where the minimum temperature is maintained in 40 °C in both bonding cases, i.e. with and without crack. Moreover, the Hysol EA-934 marginal stresses are also lower than the EA-9394 marginal stresses.
4. Conclusion In this work the mechanical performance of two different adhesives was evaluated for the case of dissimilar materials bonded joints subjected to very low temperatures. The performance comparison was made by simulating in a commercial finite element package the ASTM D1002 specimen subjected to thermal loadings in the temperature range of 850 °C to 22 °C. The two dissimilar materials were the aluminum alloy 2024 T3 and the 6th grade titanium alloy Ti6Al4v. Bonded joints using these two materials can be found quite often in aerospace industry, one example is the joint that attaches the Airbus A320 intakes to the power plant frame. In the correlation performed herein, the influence of three factors on the adhesive strength was analyzed. First, the influence of the boundary conditions on the stress distribution experienced by the adhesive layer was analyzed. Second, the influence of an interface crack between the adhesive layer and the adherends was studied, and finally it was analyzed the influence of the variation of the coefficient of thermal expansion on the adhesive stress predictions. Regarding the boundary conditions variation, the results lead to conclude that this variation do not significantly affect the stress distribution within the adhesive layer. Despite the adherends change their deformed shape and stress distributions with the boundary condition variation, both adhesives show no sensitivity to the boundaries conditions considered herein. Regarding the bonding condition with interface crack, results show that the Hysol EA-934 adhesive is more sensitive to the interface crack presence than the Hysol EA-9394 adhesive, the Hysol EA-9394 shows no response to the interface crack presence. The CTE variation with temperature shows in both adhesives a strong influence on the adhesive stress distribution. It can be concluded that the CTE variation with temperature reduces the minimum allowable temperature before failure about 10 °C in average, comparatively to the results computed with a constant CTE at room temperature (22 °C). The Hysol EA-9394 adhesive shows an improved performance compared to the Hysol EA-934 in dissimilar materials bonded joints at low temperatures. This performance results mainly due to the Hysol EA-9394 low coefficient of thermal expansion that is more close to the adherends CTE than the Hysol EA-934 CTE is, which at room temperature it is twice the value of the Hysol EA-9394 CTE. Also, the Hysol EA-9394 higher lap shear strength and high capability to accommodate higher strains are two factors that justifies the better performance found here for the Hysol EA-9394 adhesive. In mechanical design stages of dis-
Fig. 18. Hysol EA-934 interface crack effect. (a) CTE constant, (b) CTE variable with temperature.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
13
Fig. 19. Hysol EA-9394 interface crack effect. (a) CTE constant, (b) CTE variable with temperature.
Fig. 20. EA-934 and EA-9394 marginal stress vs temperature for constant and variable CTE.
similar materials bonded joints it is advised the selection of adhesives with lower CTE in order to guarantee higher marginal stresses. Moreover, in simulations, the use of a constant CTE obtained at room temperature shows to be a conservative approach, thus the lack of information regarding the CTE variation with temperature of commercial adhesives has no deep impact on the bonded joint safety.
Acknowledgements This work was supported by FCT, through IDMEC, under LAETA project UID/EMS/50022/2013. The first author gratefully acknowledge financial support from FCT - Fundação para Ciência e Tecnologia (Portuguese Foundation for Science and Technology), for the Ph. D. Grant PD/BD/52344/2013.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012
14
V. Anes et al. / Theoretical and Applied Fracture Mechanics xxx (2016) xxx–xxx
References [1] V. Anes, R.S. Pedro, E. Henriques, M. Freitas, L. Reis, Galvanic corrosion of aircraft bonded joints as a result of adhesive microcracks, Procedia Struct. Integr. 1 (2016). [2] B. Ellis et al., Chemistry and Technology of Epoxy Resins, Springer, 1993. [3] D.R. Lefebvre, D.A. Dillard, A stress singularity approach for the prediction of fatigue crack initiation in adhesive bonds. Part 1: Theory, J. Adhes. 70 (1999) 119–138. [4] D.R. Lefebvre, D.A. Dillard, J.G. Dillard, A stress singularity approach for the prediction of fatigue crack initiation in adhesive bonds. Part 2: Experimental, J. Adhes. 70 (1999) 139–154. [5] S.H. Chowdhury, D.L. Chen, S.D. Bhole, X. Cao, P. Wanjara, Lap shear strength and fatigue behavior of friction stir spot welded dissimilar magnesium-toaluminum joints with adhesive, Mater. Sci. Eng. A 562 (2013) 53–60. [6] M.D. Banea, L.F.M. da Silva, Adhesively bonded joints in composite materials: an overview, Proc. Inst. Mech. Eng. Part L J. Mater. Des. Appl. 223 (2009) 1–18. [7] Y. Ryoji, C. Sang-Bong, Efficient boundary element analysis of stress intensity factors for interface cracks in dissimilar materials, Eng. Fract. Mech. 34 (1989) 179–188. [8] L.F.M. Da Silva, R.D. Adams, Adhesive joints at high and low temperatures using similar and dissimilar adherends and dual adhesives, Int. J. Adhes. Adhes. 27 (2007) 216–226. [9] B.H. Rabin, R.L. Williamson, S. Suresh, Fundamentals of residual stresses in joints between dissimilar materials, MRS Bull. 20 (1995) 37–39. [10] S. de Freitas, J. Sinke, Failure analysis of adhesively-bonded skin-to-stiffener joints: metal-metal vs. composite-metal, Eng. Fail. Anal. (2015). [11] M. Konstantakopoulou, A. Deligianni, G. Kotsikos, Failure of dissimilar material bonded joints, Nanomater. Join. (2016) 103. [12] J.F. Timmerman, M.S. Tillman, B.S. Hayes, J.C. Seferis, Matrix and fiber influences on the cryogenic microcracking of carbon fiber/epoxy composites, Compos. Part A: Appl. Sci. Manuf. 33 (2002) 323–329. [13] B.C. Ray, Adhesion of glass/epoxy composites influenced by thermal and cryogenic environments, J. Appl. Polym. Sci. 102 (2006) 1943–1949. [14] S.G. Kalarikkal, B.V. Sankar, P.G. Ifju, Effect of cryogenic temperature on the fracture toughness of graphite/epoxy composites, J. Eng. Mater. Technol. 128 (2006) 151–157. [15] M.D. Banea, L.F.M. da Silva, R. Campilho, Effect of temperature on tensile strength and mode I fracture toughness of a high temperature epoxy adhesive, J. Adhes. Sci. Technol. 26 (2012) 939–953.
[16] Y. Ryoji, L. Jin-Qiao, X. Jin-Quan, O. Toshiaki, O. Tomoyoshi, Mixed mode fracture criteria for an interface crack, Eng. Fract. Mech. 47 (1994) 367–377. [17] A.J. Kinloch, C.F. Korenberg, K.T. Tan, J.F. Watts, Crack growth in structural adhesive joints in aqueous environments, J. Mater. Sci. 42 (2007) 6353–6370. [18] B.R.K. Blackman, A.J. Kinloch, M. Paraschi, On the mode II loading of adhesive joints, Eur. Struct. Integr. Soc. 32 (2003) 293–304. [19] S.J. Bennett, K.L. Devries, M.L. Williams, Adhesive fracture mechanics, Int. J. Fract. 10 (1974) 33–43. [20] D.R. Mulville, R.N. Vaishnav, Interfacial crack propagation, J. Adhes. 7 (1975) 215–233. [21] A.M.G. Pinto, A.G. Magalhães, R.D.S.G. Campilho, M. de Moura, A.P.M. Baptista, Single-lap joints of similar and dissimilar adherends bonded with an acrylic adhesive, J. Adhes. 85 (2009) 351–376. [22] D. Ren, L. Liu, Y. Li, Investigation on overlap joining of AZ61 magnesium alloy: laser welding, adhesive bonding, and laser weld bonding, Int. J. Adv. Manuf. Technol. 61 (2012) 195–204. [23] S.-G. Kang, M.-G. Kim, C.-G. Kim, Evaluation of cryogenic performance of adhesives using composite–aluminum double-lap joints, Compos. Struct. 78 (2007) 440–446. [24] M. Davis, D. Bond, Principles and practices of adhesive bonded structural joints and repairs, Int. J. Adhes. Adhes. 19 (1999) 91–105. [25] L.F.M. Da Silva, R.D.S.G. Campilho, Advances in Numerical Modelling of Adhesive Joints, Springer, 2012. [26] C.T. Sun, W. Qian, The use of finite extension strain energy release rates in fracture of interfacial cracks, Int. J. Solids Struct. 34 (1997) 2595–2609. [27] T.R. Guess, E.D. Reedy, M.E. Stavig, Mechanical Properties of Hysol EA-9394 Structural Adhesive, SAND95-0229, Sandia Natl. Lab, Albuquerque, New Mex, 1995. [28] G. Anderson, K. DeVries, Predicting strength of adhesive joints from test results, Int. J. Fract. (1989). [29] T. Dursun, C. Soutis, Recent developments in advanced aircraft aluminium alloys, Mater. Des. (2014). [30] G. Welsch, R. Boyer, E. Collings, Materials Properties Handbook: Titanium Alloys, 1993. [31] M. Wahab, The Mechanics of Adhesives in Composite and Metal Joints: Finite Element Analysis with ANSYS, 2014.
Please cite this article in press as: V. Anes et al., Bonded joints of dissimilar adherends at very low temperatures - An adhesive selection approach, Theor. Appl. Fract. Mech. (2016), http://dx.doi.org/10.1016/j.tafmec.2016.08.012