Experimental Details
When joining large components as in the case of longitudinal fuselage
joints, it may not be feasible to control curing temperature evenly and
as such, an adhesive capable of both curing at room temperature and
elevated temperature is desirable. During FSW of FS weld-bonded joints,
the uncured adhesive will be subjected to an elevated temperature which
may locally accelerate the curing process. To assess the curing process
of the chosen adhesive differential scanning calorimetry (DSC) was used.
DSC analysis was performed on a Netzsch® DSC 200 F3
equipment on specimens with a mass of ≈50 mg, at a constant heating rate
of 20 K/min from 21ºC to 320ºC in an atmosphere of constant flow of 20
mL/min of N2. Figure 1 shows a representative curve of a
DSC analysis.
Figure 1 Representative curve of
DSC analysis of the uncured epoxy adhesive
Even though the adhesive may cure at room temperature, DSC analysis
showed that the majority of the curing process occurs at elevated
temperatures, with peek curing at ≈120ºC. An endothermic event is also
observed at about 200ºC in all samples tested, which is believed to be
evaporation of water, after exceeding the sealing limit of the sample
container (water vapor pressure at 200ºC is ≈15 Atm).
From the DSC analysis it may be inferred that full curing does not occur
at room temperature, leading the adhesive to have different mechanical
behavior with different curing conditions. To assess the tensile
mechanical properties of the adhesive, bulk tensile were performed at 1
mm/min crosshead speed in an Instron testing machine. The bulk tensile
specimens were made with 4 different curing conditions, room temperature
for 7 days, 120ºC for 1 hour as indicated in the adhesive data sheet,
165ºC and 200ºC for 30 minutes. After curing specimens were milled
according to ASTM D638 standard. In Figure 2, the resulting stress vs
strain curves for the 4 curing conditions are presented.
Figure 2 Representative Araldite
420 stress vs. strain curves with curing temperature
An increase in ultimate strength was observed with increasing curing
temperature, accompanied by a reduction in elongation at break. This
behavior may be due to increased cross-linking of the polymeric chains
with increasing cure temperature. The change is more significantly from
room temperature to 120ºC cure condition than from 120ºC to further
higher temperatures. This is consistent with the DSC analysis results
were most of the curing was shown to happen around 120ºC.
As during the welding process, the adhesive will be subjected to high
temperatures, having an adhesive with a high degradation temperature is
important. Thermogravimetric analysis (TGA) of the uncured epoxy
adhesive was made on a Netzsch® Tg209 F3 Tarsus at 20 K/min from 21ºC to
600ºC. Figure 3 presents resulting TG curves where the onset of
degradation was found to be at \(357_{-3}^{+2}\)ºC. Temperatures in
the adhesive during FS weld-bonding were reported to be between 200ºC
and 250ºC, which leads to conclude that no significant degradation will
occur during the welding procedure.6
Figure 3 TGA of uncured epoxy
adhesive
To determine shear strength and shear modulus, thick adherend shear test
(TAST) and Poisson ratio measurement were made according to ASTM D5656
– 10 and ASTM E132 – 17 respectively. Two curing conditions were
assessed, room temperature for 1 week and 120ºC for 1 hour. The
resulting (\(\tau_{u}\)) and shear modulus are presented in Table 1.
Given the complex loading case of single lap joints, with combination of
peel and shear load on the adhesive, fracture toughness of the adhesive
in mode I and II was assessed through double cantilever beam (DCB)
specimens and end notch flexure (ENF) specimens. In both tests,
specimens had the same dimensions, being composed of two steel beams
with 320 x 25 x 12.7 mm bonded by a layer of 0.2 mm thick adhesive,
differing only on the loading method. Specimens used for fracture
toughness assessment were cured at room temperature for 7 days and as an
approximation it was assumed that fracture toughness remained unchanged
with curing condition. However, it may be expected that some fracture
toughness is lost with increasing curing temperature and as such the
real values may be lower.7
DCB specimens were loaded at 1 mm/min crosshead speed and the resulting
load vs. displacement curves were used to plot the corresponding
R-curves using the compliance based beam method
(CBBM).8 A resulting representative R-curve is shown
in Figure 4. The critical fracture toughness in mode I measured was 3\(\pm\)0.37 N/mm. This value is relatively high compared with other
structural adhesives9,10 and continuous fiber
reinforced composites11, showing that the chosen epoxy
has high fracture resistance.
Figure 4 Representative adhesive
R-curve for mode I
A digital twin of the experimental procedure was created to confirm
measured experimental data using Abaqus. Cohesive zone modeling (CZM)
was used to model the adhesive failure. The load vs. displacement curves
were in good agreement as shown in Figure 5.
Figure 5 a) von Mises stress in 3D
DCB Abaqus model at 5 mm displacement and b) load displacement curve
comparison between numeric and experimental
For mode II, ENF testing was performed at 0.2 mm/min crosshead speed.
Similarly to the DCB tests in mode I, the ENF tests were also analyzed
through CBBM method, but in this case for mode II
loading.12 A representative R-curve obtained in the
ENF tests is presented in Figure 6. The critical fracture toughness
measured in mode II was 11.6\(\pm\)0.3 N/mm. However, the maximum
bending load was relatively high, which may have induced local
plasticization, and as such the measured mode II fracture toughness may
be artificially high. A parametric study was then used to find an
adequate adhesive fracture toughness in mode II, by keeping constant all
other material parameters and comparing numeric and experimental loads
vs displacement curves.
Figure 6 Representative adhesive
R-curve for mode II
The Abaqus model showed that indeed 11.6 N/mm was an overestimation of
the critical fracture toughness in mode II and by an iterative process
that 9 N/mm resulted in better agreement with the experimental load vs.
displacement curves as shown in Figure 7. This value is still relatively
high fracture strength when compared with adhesives reported in the
literature.13,14 There was a small difference in terms
of stiffness between numerical model and experimental data which was
consistent in all numeric runs and is probably due to the experimental
loading configuration. As the ENF test is performed in 3-point bending,
the machine is operated in compression and the displacement values
measured include all the slack within the system, while in the numerical
model no such limitations exist.
Figure 7 a) Shear stress in 3D ENF
Abaqus model at the onset of damage, b) displacement curve comparison
between numeric and experimental.
The adhesive mechanical properties are summarized in Table 1, for two
curing conditions.
Table 1 - Araldite 420 mechanical
properties
The alloy used in this study was the AA6082-T6. The chemical composition
is show in Table 2 and the mechanical properties in Table 3.
Table 2 Chemical composition of
AA6082-T6(% mass)15
Table 3 AA6082-T6 mechanical
properties15
All welding procedures were performed on a dedicated FSW
ESAB® LEGIO 3UL numerical control machine. In FS
weld-bonding, the welding procedure was done with the adhesive in a
non-cured stated and right after adhesive lay-up and joint closing.
Calibrated metal strips with 0.2 mm thickness were strategically
positioned in-between the shim plates to assure a more uniform adhesive
thickness.
Prior to bonding surfaces to be bonded were degreased and sanded. In the
case of adhesive bonded joints phosphoric acid anodization (PAA)
according to ASTM D3933 - 98(2017) standard was used, while FS
weld-bonded joints were subjected to chemical treatment with
3M® AC-130, which is a sol-gel anodization replacement
normally intended for aeronautical repair 16.
The FSW tool used had 5 mm diameter cylindrical threaded pin with 3 mm
length and 16 mm diameter grooved shoulder. The FSW process parameters
used are listed in Table 4. These were chosen based on literature review
and past experience. Various levels of downward force were tested to
assess its effect on joint performance and maximize joint strength.
Table 4 FSW process parameters
All joint configurations were tensile tested with three specimens each.
Tensile testing was done in an Instron® 5566 machine
at 1 mm/min crosshead speed. Joint efficiency was calculated dividing
maximum axial load by the substrate cross-section outside of the overlap
as in previous works.3,17 Figure 8, compares the joint
efficiency of the joint configurations tested.
Figure 8 Joint efficiency of FSW
and FS weld-bonded joints with differing downward force
When comparing FS weld-bonded joints to FSW it was possible to observe
an improvement of 20-30 % in most cases. It was possible to observe
that for FSW joints the in downward force results in an increase of
joint strength. This may be related with higher thermal input which
leads to further softening of the workpiece. The further softening of
the workpiece may result in better mixing and as such diminishing the
hook defect size, as presented in.18 For FS
weld-bonded joints the trend is not as clear as in FSW joints, as it
increases from 400 to 450 kgf but diminishes from then on. The reasoning
for this decrease may be due to high downward force leading to excessive
adhesive thinning or possibly the higher temperature may lead to
degradation of the surface to bond and/or the adhesive. FS weld-bonded
joints also showed higher dispersion in terms of joint strength than FSW
which may be due to variations in surface treatment, as the joint
strength is very sensitive to the bonding strength. Figure 9, compares
the highest strength FSW and FS weld-bonded joints with adhesive bonded
joints.
Figure 9 Stress vs. displacement
of highest strength FSW and FS weld-bonded joints along the adhesive
bonded.
In Figure 10, the microhardness profile and joint cross-section of the
best performing FS weld-bonded joint is presented. It is possible to
observe that the hook defect formed by the upward flow of material
generated in the advancing side is present. The size and shape of this
defect is critical to overlap FSW joint strength.18Along with the hook defect it is also observable a cold lap defect on
the retreating side of the weld, which is a result of the initial upward
flow under shearing effect of the pin followed by a downward flow in
order to fill the space at the bottom of the pin. However, these defects
become less critical in FS weld-bonded joints when compared to FSW as
the adhesive increases the effective joined overlap and reduces loading
the weld edges.
Figure 10 a) Microhardness profile
and b) joint cross-section of FS weld-bonded joint with 450 kgf
A loss of hardness is presented in the microhardness profile, which is
due to the loss of T6 condition during welding. Temperatures increase
towards the top of the joint due to contact between workpiece and tool
and as such, a wider cross-section of the joint has lower hardness at
the top.
An assessment of the fatigue strength of FSW, FS weld-bonded and
adhesive bonded joints was then made at a load ratio of R=0.1 in an
Instron 8874 machine. For this study both FSW and FS weld-bonded joints
were made using 450 kgf. Figure 11 shows the resulting S-N curves
including 50% and 5% probability of failure calculated using
ProFatigue software.19 The S-N curve regarding the
AA6082-T6 base material reported in 20 was also
included.
Figure 11 a) p-S-N curves of the
three joint type and b) failure modes
FSW joints showed lower fatigue strength than adhesive and FS
weld-bonded joints as it would be expected given the lower quasi-static
strength. The FS weld-bonded showed similar fatigue strength to adhesive
bonded joints. Adhesive bonded joints still had the highest fatigue
strength of all joints tested, which is to be expected given the more
continuous stress distribution in these joints. At 106cycles adhesive bonded joints have a 50% probability of failure at 56.4
MPa, while FS weld-bonding and FSW at the same number of cycles the 50%
probability of failure is achieved at 45.1 MPa (79.9% of adhesive
bonded joints) and 23.5 MPa (41.6% of adhesive bonded joints). The
failure modes were consistent with the quasi-static ones, with the
adhesive failing in adhesive/cohesive way, the FSW failing through the
hook defect and the FS weld-bonded ones failing through the adhesive
layer followed by cracking through the hook.