CRACK GROWTH IN ADHESIVELY BONDED JOINTS SUBJECTED TO VARIABLE FREQUENCY FATIGUE LOADING

The main aim of this article is to investigate the effect of frequency on fatigue crack propagation in adhesively bonded joints. Adhesively bonded double-cantilever beam (DCB) samples were tested in fatigue at various frequencies between 0.1 and 10 Hz. The adhesive used was a toughened epoxy, and the substrates used were a carbon fibre-reinforced polymer (CFRP) and mild steel. Results showed that the crack growth per cycle increases and the fatigue threshold decreases as the test frequency decreases. The locus of failure with the CFRP adherends was predominantly in the adhesive layer, whereas the locus of failure with the steel adherends was in the interfacial region between the steel and the adhesive. The crack growth was faster, for a given strain energy release rate, and the fatigue thresholds lower for the samples with steel adherends. Tests with variable frequency loading were also carried out, and a generalised method of predicting crack growth in samples subjected to a variable frequency loading was introduced. The predicted crack growth using this method agreed well with experimental results.     

Fatigue Crack Growth in Epoxy/Aluminum and Epoxy/Steel Joints

The fatigue crack growth rate within epoxy/aluminum and epoxy/steel joints was evaluated as a function of a) type of surface pretreatment, b) water soak, c) fatigue cycle rate (Hz), d) adhesive thickness and e) type of epoxy adhesive.

For both adherends, aluminum and steel, a significant improvement in the fatigue behavior was obtained by use of a mercaptoester coupling agent. After an 8-day, 57°C water soak, the metal surfaces which were pretreated with coupling agent (CA) or by phosphoric acid anodization (PAA) still resulted in cohesive failure, while the controls had higher crack growth rate and showed greater scatter. The room-temperature cure matrix with CA-treated aluminum showed a less dramatic improvement, probably because of a known difference in the application procedure. For the steel joints and room-temperature adhesive the improvement in the fatigue behavior of CA-treated samples was maintained after the 8-day hot water soak. No significant change was found in the fatigue crack growth rate over a frequency range of 1 to 5 Hz, but a significant change was found as a function of the bondline thickness. The room temperature curing adhesive evaluated herein exhibited a much lower fatigue resistance than a heat-cured commercial structural adhesive FM-73.     

Effect of Adhesive Ductility on Cyclic Debond Mechanism in Composite-to-Composite Bonded Joints

An investigation of an adhesively bonded composite joint with a brittle adhesive was conducted to characterize both the static and fatigue debond growth mechanism under mode I and mixed mode I-II loadings. The bonded system consisted of graphite/epoxy adherends bonded with FM-400 adhesive. Two specimen types were tested: (1) a double-cantilever-beam specimen for mode I loading and (2) a cracked-lap-shear specimen for mixed mode I-II loading. In all specimens tested, failure occurred in the form of debond growth either in a cohesive or adhesive manner. The total strain-energy-release rate is not the criterion for cohesive debond growth under static and fatigue loading in the birttle adhesive as observed in previous studies with the ductile adhesives. Furthermore, the relative fatigue resistance and threshold value of cyclic debond growth in terms of its static fracture strength is higher in the brittle adhesive than its counterpart in the ductile adhesive.     

Fatigue Crack Growth in Epoxy/Aluminum and Epoxy/Steel Joints

The fatigue crack growth rate within epoxy/aluminum and epoxy/steel joints was evaluated as a function of a) type of surface pretreatment, b) water soak, c) fatigue cycle rate (Hz), d) adhesive thickness and e) type of epoxy adhesive.

For both adherends, aluminum and steel, a significant improvement in the fatigue behavior was obtained by use of a mercaptoester coupling agent. After an 8-day, 57°C water soak, the metal surfaces which were pretreated with coupling agent (CA) or by phosphoric acid anodization (PAA) still resulted in cohesive failure, while the controls had higher crack growth rate and showed greater scatter. The room-temperature cure matrix with CA-treated aluminum showed a less dramatic improvement, probably because of a known difference in the application procedure. For the steel joints and room-temperature adhesive the improvement in the fatigue behavior of CA-treated samples was maintained after the 8-day hot water soak. No significant change was found in the fatigue crack growth rate over a frequency range of 1 to 5 Hz, but a significant change was found as a function of the bondline thickness. The room temperature curing adhesive evaluated herein exhibited a much lower fatigue resistance than a heat-cured commercial structural adhesive FM-73.     

Effect of Adhesive Ductility on Cyclic Debond Mechanism in Composite-to-Composite Bonded Joints

An investigation of an adhesively bonded composite joint with a brittle adhesive was conducted to characterize both the static and fatigue debond growth mechanism under mode I and mixed mode I-II loadings. The bonded system consisted of graphite/epoxy adherends bonded with FM-400 adhesive. Two specimen types were tested: (1) a double-cantilever-beam specimen for mode I loading and (2) a cracked-lap-shear specimen for mixed mode I-II loading. In all specimens tested, failure occurred in the form of debond growth either in a cohesive or adhesive manner. The total strain-energy-release rate is not the criterion for cohesive debond growth under static and fatigue loading in the birttle adhesive as observed in previous studies with the ductile adhesives. Furthermore, the relative fatigue resistance and threshold value of cyclic debond growth in terms of its static fracture strength is higher in the brittle adhesive than its counterpart in the ductile adhesive.     

Fatigue Crack Growth Rates in Adhesive Joints Tested at Different Frequencies

Mode I fatigue crack growth tests were conducted on joints bonded with a filled adhesive (A) at 20 Hz and 2 Hz and on joints bonded with a filled and toughened adhesive (B) at 20 Hz, 2 Hz, 0.2 Hz and 0.02 Hz. Strain energy release rate, G, and J-integral were evaluated based on elastic and elastoplastic finite element analyses (FEA) of the joints bonded with adhesive A and B, respectively. For the configurations considered, J was found to be path-independent and did not differ much from G. The fatigue crack growth rate (FCGR), da/dN, in the joints bonded with adhesive A was relatively independent of frequency while it increased with decreasing frequency at given δ for the joints bonded with adhesive B. The fatigue processes in both adhesives involved the cracking of the filler particles and subsequent linkage of the resultant microcracks. The process zone in adhesive B is larger than that in adhesive A and it increases with decreasing frequency. It is suggested that this variation in process zone size can account for the observed fatigue behaviour. The fatigue crack growth velocity, da/dt, was also calculated for the joints bonded with adhesive B and the variation of da/dt with test frequency at given δG is much smaller than the variation in da/dN, suggesting a creep effect in the fatigue crack growth.     

Analysis and design of adhesively bonded joints for fatigue and fracture loading: a fracture-mechanics approach

An experimental–computational fracture-mechanics approach for the analysis and design of structural adhesive joints under static loading is demonstrated by predicting the ultimate fracture load of cracked lap shear and single lap shear aluminum and steel joints bonded using a highly toughened epoxy adhesive. The predictions are then compared with measured values. The effects of spew fillet, adhesive thickness, and surface roughness on the quasi-static strength of the joints are also discussed. This fracture-mechanics approach is extended to characterize the fatigue threshold and crack growth behavior of a toughened epoxy adhesive system for design purposes. The effects of the mode ratio of loading, adhesive thickness, substrate modulus, spew fillet, and surface roughness on the fatigue threshold and crack growth rates are considered. A finite element model is developed to both explain the experimental results and to predict how a change in an adhesive system affects the fatigue performance of the bonded joint.     

Cohesive Zone Model Characterization of the Adhesive Hysol EA-9394

Abstract

The cohesive zone model approach is attractive for the analysis of failure of adhesively bonded structures. While the numerical implementation of cohesive elements has been well established, there remains a lack of cohesive material data. The present paper contributes to efforts to fill this void. An investigation of crack growth in the widely used structural adhesive Hysol EA-9394 is presented, and the adhesive is characterized by a cohesive zone law. Crack growth experiments were performed on specimens consisting of aluminum adherends bonded by use of the adhesive. Measurements of the surface topography leading reconstruction of fracture processes indicate that plastic deformation is absent during fracture. Thus, the cohesive zone law can directly be determined from the energy release rate and the material separation measured at the initial crack tip. The cohesive zone law is then applied in finite element model to predict crack growth. The predicted strain fields during crack growth are well matched to those obtained by digital image correlation measurements. An independent set of crack growth experiments was performed, and finite element models based on the cohesive law were used to predict the outcome of these experiments. Again good agreement between simulation and experiment was obtained. The results give confidence that the cohesive zone model parameters are transferable to the analysis of structures bonded with the adhesive Hysol EA-9394 in general. A comparison of the cohesive zone law for Hysol EA-9394 demonstrates that this adhesive possesses high strength and moderate toughness. Limits to the transferability regime are discussed.     

Effect of Substrate Material on Fatigue Crack Propagation Rate of Adhesively Bonded DCB Joints

The effect of substrate material on the fatigue crack propagation rate was investigated using adhesively bonded DCB specimens with CFRP and aluminum substrates. The experimental results show that the increase in thickness of the adherend lowers the fatigue threshold, ΔG _th, and raises the crack growth parameter, n, irrespective of the substrate material, and that the crack growth parameter, n, for the aluminum joints is less than that for the CFRP joints. To elucidate the fatigue crack propagation behavior, fracture surface observation and finite element analysis have been conducted. Besides, Gurson's model is applied to the adhesive layer. SEM images show that numerous voids are formed in the fracture surface for the joints with aluminum substrate, but the growth of voids is suppressed for the joints with CFRP substrate. FEM results also show that the void area fraction for the joint with aluminum substrate is greater than that with CFRP substrate. Thus, the above experimental and numerical trends of voids correspond to the trends of the fatigue crack propagation behavior.     

Metallic multi-material adhesive joint testing and modeling for vehicle lightweighting

While adhesive bonding has been shown to be a beneficial technique to join multi-material automotive bodies-in-white, quantitatively assessing the effect of adherend response on the ultimate strength of adhesively bonded joints is necessary for accurate joint design.In the current study, thin adherend single lap shear testing was carried out using three sheet metals used to replace mild steel when lightweighting automotive structures: hot stamped Usibor® 1500 AS ultra-high strength steel (UHSS), aluminum (AA5182), and magnesium (ZEK 100). Six combinations of single and multi-material samples were bonded with a one-part toughed structural epoxy adhesive and experimentally tested to measure the force, displacement across the bond line, and joint rotation during loading. Finite element models of each test were analyzed using LS-DYNA to quantitatively assess the effects of the mode mixity on ultimate joint failure. The adherends were modeled with shell elements and a cohesive zone model was implemented using bulk material properties for the adhesive to allow full three-dimensional analysis of the test, while still being computationally efficient.The UHSS-UHSS joint strength (27.2 MPa; SD 0.6 MPa) was significantly higher than all other material combinations, with joint strengths between 17.9 MPa (SD 0.9 MPa) and 23.9 MPa (SD 1.4 MPa). The models predicted the test response (average R2 of 0.86) including the bending deformation of the adherends, which led to mixed mode loading of the adhesive. The critical cohesive element in the UHSS-UHSS simulation predicted 85% Mode II loading at failure while the other material combinations predicted between 41% and 53% Mode II loading at failure, explaining the higher failure strength in the UHSS-UHSS joint.This study presents a computational method to predict adhesive joint response and failure in multi-material structures, and highlights the importance of the adherend bending stiffness and on joint rotation and ultimate joint strength.     

Fatigue Crack Growth Rates in Adhesive Joints Tested at Different Frequencies

Mode I fatigue crack growth tests were conducted on joints bonded with a filled adhesive (A) at 20 Hz and 2 Hz and on joints bonded with a filled and toughened adhesive (B) at 20 Hz, 2 Hz, 0.2 Hz and 0.02 Hz. Strain energy release rate, G, and J-integral were evaluated based on elastic and elastoplastic finite element analyses (FEA) of the joints bonded with adhesive A and B, respectively. For the configurations considered, J was found to be path-independent and did not differ much from G. The fatigue crack growth rate (FCGR), da/dN, in the joints bonded with adhesive A was relatively independent of frequency while it increased with decreasing frequency at given δ for the joints bonded with adhesive B. The fatigue processes in both adhesives involved the cracking of the filler particles and subsequent linkage of the resultant microcracks. The process zone in adhesive B is larger than that in adhesive A and it increases with decreasing frequency. It is suggested that this variation in process zone size can account for the observed fatigue behaviour. The fatigue crack growth velocity, da/dt, was also calculated for the joints bonded with adhesive B and the variation of da/dt with test frequency at given δG is much smaller than the variation in da/dN, suggesting a creep effect in the fatigue crack growth.     

Fatigue Growth of Cohesive Defects in T-Peel Joints

This paper uses 2D and 3D finite element models to predict the stresses within bonded and weld-bonded T-peel joints. Epoxy adhesive is modelled as a homogeneous layer providing a perfect bond between aluminium adherends. Knowledge of the critical tensile stresses enables the likely region of fatigue crack initiation to be predicted. The long term reliability and durability of a joint depend directly on its fatigue strength. This research elucidates the region of cohesive crack initiation, the subsequent direction of crack propagation and the relative duration of the different stages of fatigue crack growth. The various stages of embedded, surface and through-width fatigue growth of cohesive defects within a T-peel joint are compared. This establishes fatigue life from crack initiation to final joint fracture for typical bonded and weld-bonded T-peel joints.     

Effect of vibration fatigue on modal properties of single lap adhesive joints

As most existing studies focus on developing models and theories describing the static strength of adhesive joints as a function of the fatigue loading, there is a lack of understanding on how the fatigue of the adhesive joint affects dynamic modal properties of the bonded structure. In applications such as automobile components, modal properties are critical in determining their dynamic performances. To investigate the relationship between modal properties of single lap joints (SLJs) and the cyclic-vibration-peel loading, this study first carries out vibration fatigue tests and subsequent modal response measurements using steel–aluminum SLJ specimens. It is experimentally demonstrated that modal frequencies of the SLJ structure tend to decrease with increasing vibration fatigue cycles. Furthermore, it is also shown that this trend is related to the fatigue characteristics of the adhesive layer. The fatigue degradation effects of Young's modulus and contact area between the adhesive and the adherends on modal frequencies are then investigated using a finite element model. Simulation results reveal that dramatic reductions in modulus and contact area values are required to result in the modal frequency shifting observed in experiments, which may not be always realistic. Although the findings in this study are informative, more research effort is needed to further identify the critical reason(s) for the experimental trend of decreasing modal frequencies with increasing vibration fatigue cycles.     

Metal alkoxide primers in the adhesive bonding of mild steel

The effects of metal alkoxide type and relative humidity on the durability of alkoxide-primed, adhesively bonded steel wedge crack specimens have been determined. Aluminum tri-secbutoxide, aluminum tri-tert-butoxide, tetrabutyl orthosilicate, and titanium(IV ) butoxide were used as alkoxide primers. Grit-blasted, acetone-rinsed mild steel adherends were the substrates bonded with epoxy and polyethersulfone. The two aluminum alkoxides significantly enhanced the durability of the adhesively bonded steel, while the titanium alkoxide showed no improvement in durability over a nonprimed control. The silicon alkoxide-primed samples gave an intermediate response. The failure plane in the adhesively bonded samples varied with the relative humidity during the priming process.     

An investigation on the initiation and propagation of damaged zones in adhesively bonded lap joints

In this study, the initiation and propagation of damaged zones in the adhesive layer and adherends of adhesively bonded single and double lap joints were investigated considering the geometrical non-linearity and the non-linear material behaviour of the adhesive and adherends. The modified von Mises criteria for adherends and Raghava and Cadell's failure criteria (J. Mater. Sci. 8, 225 (1973) [1]) including the effects of the hydrostatic stress states for the epoxy adhesive were used to determine the damaged adhesive and adherend zones which exceeded the specified ultimate strains. The stiffness of all finite elements corresponding to these zones was reduced so that they could not contribute to the overall stiffness of the adhesive joint. This approach simplifies to observe the initiation and propagation of the damaged zones in both the adhesive layer and adherends. A tensile load caused first the damaged adhesive zones to appear at the right free end of the adhesive-lower adherend interface and at the left free end of the adhesive-upper adherend interface, and then to propagate through the adhesive regions near the adhesive-adherend interfaces (interfacial failure). In the bending test, the damaged zone initiated at the left free end of the adhesive-upper adherend interface in tension, and similarly propagated through the adhesive regions close to the adhesive-adherend interface (interfacial failure). In the double-lap joint subjected to a tensile load, the damaged adhesive zones initiated first at the right free end of the adhesive-middle adherend interface and then propagated through the adhesive region near the adhesive-adherend interface. After the damaged zone reached a specific length it also grew through the adhesive thickness, and the adhesive joint failed. The SEM micrographs of fracture surfaces around the free edges of the overlap region indicated that the failure was interfacial. An additional damaged zone growth was observed in the side adhesive regions due to lateral straining, called the Poisson effect.     

Strength prediction of epoxy adhesively bonded scarf joints of dissimilar adherends

In this study, strength of epoxy adhesively bonded scarf joints of dissimilar adherends, namely SUS304 stainless steel and YH75 aluminum alloy is examined on several scarf angles and various bond thicknesses under uniaxial tensile loading. Scarf angle, θ=45°, 60° and 75° are employed. The bond thickness, t between the dissimilar adherends is controlled to be ranged between 0.1 and 1.2 mm. Finite element (FE) analysis is also executed to investigate the stress distributions in the adhesive layer of scarf joints by ANSYS 11 code. As a result, the apparent Young's modulus of adhesive layer in scarf joints is found to be 1.5-5 times higher than those of bulk epoxy adhesive, which has been obtained from tensile tests. For scarf joint strength prediction, the existing failure criteria (i.e. maximum principal stress and Mises equivalent stress) cannot satisfactorily estimate the present experimental results. Though the measured stress multiaxiality of scarf joints proportionally increases as the scarf angle increases, the experimental results do not agree with the theoretical values. From analytical solutions, stress singularity exists most pronouncedly at the steel/adhesive interface corner of joint having 45-75° scarf angle. The failure surface observations confirm that the failure has always initiated at this apex. This is also in agreement with stress-y distribution obtained within FE analysis. Finally, the strength of scarf joints bonded with brittle adhesive can be best predicted by interface corner toughness, Hc parameter.     

Effect of Temperature on the Quasi-static Strength and Fatigue Resistance of Bonded Composite Double Lap Joints

Fibre reinforced polymer composites (FRP's) are often used to reduce the weight of a structure. Traditionally the composite parts are bolted together; however, increased weight savings can often be achieved by adhesive bonding or co-curing the parts. The reason that these methods are often not used for structural applications is due to the lack of trusted design methods and concerns about long-term performance. The authors have attempted to address these issues by studying the effects of fatigue loading, test environment and pre-conditioning on bonded composite joints. Previous work centered on the lap-strap joint which was representative of the long-overlap joints common in aerospace structures. However, it was recognised that in some applications short-overlap joints will be used and these joints might behave quite differently. In this work, double-lap joints were tested both quasi-statically and in fatigue across the temperature range experienced by a jet aircraft. Two variants on the double-lap joint sample were used for the testing, one with multidirectional (MD) CFRP adherends and the other with unidirectional (UD) CFRP adherends. Finite element analysis was used to analyse stresses in the joints. It was seen that as temperature increased both the quasi-static strength and fatigue resistance decreased. The MD joints were stronger at low temperatures and the UD joints stronger at high temperatures. It was proposed that this was because at low temperature the strength was determined by the peak stresses in the joints, whereas, at high temperatures, strength is controlled by creep of the joints which is determined by the minimum stresses in the joint. This argument was supported by the stress analysis.     

Adhesive failure of sandwich structures under combined fatigue and thermal loading

Adhesive failure of a sandwich beam made of steel and PTFE is studied under thermal and vibration environment. Superposed stresses model is presented in order to predict the effect of the combination environment. Then, the stress distributions are evaluated by an experimental method. Results show that the proposed model is effective for the study of thermal stress. The stress on the steel beam in the initial crack region is slightly low under thermal environment. Vibration excitation generates cyclic loading on the beam and stress distribution exhibits sensitivity to location and loading frequency. It is observed that the high-stress peak will be provided in combined environment. The fatigue crack growth is monitored, and it is shown that the adhesive failure is strongly temperature-dependent with invariable dynamic excitation. High temperature leads to both the high value of stress intensity and an increase in fatigue. Moreover, stick-slip behavior is observed at the interface of adhesive and base plate.     

Effect of Temperature on the Quasi-static Strength and Fatigue Resistance of Bonded Composite Double Lap Joints

Fibre reinforced polymer composites (FRP's) are often used to reduce the weight of a structure. Traditionally the composite parts are bolted together; however, increased weight savings can often be achieved by adhesive bonding or co-curing the parts. The reason that these methods are often not used for structural applications is due to the lack of trusted design methods and concerns about long-term performance. The authors have attempted to address these issues by studying the effects of fatigue loading, test environment and pre-conditioning on bonded composite joints. Previous work centered on the lap-strap joint which was representative of the long-overlap joints common in aerospace structures. However, it was recognised that in some applications short-overlap joints will be used and these joints might behave quite differently. In this work, double-lap joints were tested both quasi-statically and in fatigue across the temperature range experienced by a jet aircraft. Two variants on the double-lap joint sample were used for the testing, one with multidirectional (MD) CFRP adherends and the other with unidirectional (UD) CFRP adherends. Finite element analysis was used to analyse stresses in the joints. It was seen that as temperature increased both the quasi-static strength and fatigue resistance decreased. The MD joints were stronger at low temperatures and the UD joints stronger at high temperatures. It was proposed that this was because at low temperature the strength was determined by the peak stresses in the joints, whereas, at high temperatures, strength is controlled by creep of the joints which is determined by the minimum stresses in the joint. This argument was supported by the stress analysis.