Effect of Stacking Sequence on Low-Velocity Impact Behavior of Metal Laminates

In this paper, the low-velocity impact behavior of metal laminates was studied experimentally and numerically. Metal laminates with different number of metal layers and different stacking sequences were investigated by examining the contact force, the contact duration, the dissipated energy and the transverse displacement as the main low-velocity impact responses. It was found that from stacking sequence perspective, the low-velocity impact responses of metal laminates were mainly affected by the volume fraction of metal layers, the material characteristics of the first and last metal layers and the number of metal layers. Increasing the number of metal layers in a constant thickness of metal laminates decreased the contact force and increased the contact duration and transverse displacement. The use of higher volume fraction of a metal material in a metal laminate caused the metal laminate to inherit more of the impact characteristics of that material. The results of this research can assist engineers to design metallic structures with desired low-velocity impact behaviors.


INTRODUCTION
Adhesively bonded metal laminates (ABML) as a specific type of adhesively bonded structures have been examined under various loading such as static [1], fatigue [2] and impact loading [3]. Bonding different materials using adhesive can be utilized to manufacture laminated structures with outstanding impact behavior. This outstanding behavior is achieved because of several mechanisms taking place in a laminate structure such as more wide spread plastic deformation. Structural impact is known as a complex problem from experimental, analytical and numerical perspectives. Many physical and material parameters participated in a structural impact may lead to a highly nonlinear problem [4]. Therefore, many studies are required to be undertaken to investigate the mechanical behavior of structures under impact loading. A wide range of studies has been carried out to study the adhesively bonded structures under impact loading. Higuchi et al. [5,6] studied single lap adhesive joints under impact loading and bending moment applied by dropping a weight numerically and experimentally. They studied the stress distributions in the single lap joints subjected to impact loading and found that decreasing the overlap length, the adhesive thickness and the substrate thickness increased the adhesive maximum stress. Sato and Ikegami [7] studied the dynamic deformation of some adhesive joints including the single lap, the tapered lap and the scarf joints under impact loading. They found that the stress concentration in the scarf joint was relatively smaller in comparison to the other joints. Also the stress occurred in the tapered lap joints was smaller than that in the single lap joint.
Prakash et al. [8] investigated the effect of adhesive layer thickness on the impact behavior of laminate structures under high velocity impact loading. They found that several high velocity impact outputs including the shear strain rate, the penetration depth and the back plate deformation were affected by the adhesive layer thickness. Arias et al. [9] investigated the penetration of multilayer structures under impact loading and developed a finite element model for predicting the penetration. Kihara [10] evaluated the shear strength of the adhesive layers in adhesive joints sub-jected to impact loading. They showed that the joint fracture depended on the level of the incident stress wave. Yu et al. [11] studied numerically the mechanical behavior of adhesively bonded joints under impact loading. They found that the adhesive layer thickness and the elastic modulus of the adhesive affected the impact behavior of the joints considerably.
Yildirim and Apalak [12] conducted an important study on impact behavior of two-layer adhesively bonded metal laminates. They found that the location of steel layers could considerably affect the contact force, contact duration and plastic dissipation level. Moreover, they studied the effect of adhesive layer thickness on mechanical behavior of the two-layer structures. It was found that the adhesive layer thickness had a minor effect on the contact force and the contact duration whereas increasing the adhesive layer thickness caused a reduction in the residual plastic strains in the plates and adhesive layer. Khoramishad and Bagheri Tofighi [13] studied the effect of geometrical and mechanical properties of the adhesive and metal layers on low-velocity impact behavior of metal laminates. They introduced yield stress and Youngs modulus as the main material parameters affecting the low-velocity impact behavior of adhesively bonded metal laminates.
ABML structures were investigated in several studies. However, the effect of stacking sequence of the metal layers needs more detailed investigations. In the present study the effect of stacking sequence of the metal materials on the main impact outputs including the contact force, the contact duration, the dissipated energy and the transverse displacement were investigated experimentally and numerically.
Current study was undertaken in several steps. .irst, the impact behavior of the two-and three-layer adhesively bonded metal laminates was studied experimentally. Then, a finite element model was developed and validated against the experimental results for studying the impact behavior of metal laminates. The effects of the number of metal layers and the stacking sequence of the metal layers were studied on the main low-velocity impact loading outputs. The results obtained from this research can provide useful guidelines for engineers to tailor the material characteristics of a structure subjected to low-velocity impact loading by bonding the layers together.

EXPERIMENTAL WORK
Several tests were conducted on two-and threelayer metal laminates to investigate the low-velocity impact behavior of adhesively bonded metal laminates. The impact behavior of the adhesively bonded metal laminates was studied using the drop weight test method.

Manufacturing and Testing Procedure
Disk-like metal plates of 195 mm diameter and 2 mm thickness were bonded together using a structu-  tations stand for aluminum 1050-H14 and aluminum 2024-T3 metal layers, respectively. The aluminum surfaces were cleaned several times by Aston and fabrics. After stacking the metal and adhesive layers, the specimens were cured at room temperature of about 25°Ñ. The thickness of the adhesive layer was controlled as 0.30 ± 0.05 mm by placing small shims between the metal layers. .igure 1 shows typical two-layer and three-layer metal laminates.
A hemispherical rigid impactor of 16.2 mm diameter was dropped on the central point of the metal laminates from a height of 200 mm. Weight of the impactor was considered to be 6.

The Impact Tests Results
The low-velocity impact responses of metal laminates including the contact force, the contact duration, the permanent transverse displacement of the central point of the front and back faces, the radius of the permanent deformed area on the front face and the impactor kinetic energy were determined following to impact testing.
.igure 2 shows the typical contact force versus the contact duration for the metal laminate layups tested. The contact force for the three-layer structures was higher than for the two-layer structures because of the higher kinetic energy of the impactor for the three-layer structures in comparison with the two-layer structures. The input impact energies for the 1050-1050, 1050-2024, 1050-1050-2024 and 1050-2024-1050 laminates were 12.15, 20.00, 31.75 and 31.75 J, respectively. It should be mentioned that the drop weight apparatus used in this research could determine the contact force by measuring the instantaneous acceleration of the weight and multiplying it by the corresponding drop weight mass. The reason of why the input impact energies used for different laminates were different was because of the upper limit of the accelerometer beyond which the acceleration measurement was not possible. Therefore, the input energy used for the twolayer laminates had to be low enough. Moreover, in order to observe plastic deformation in the laminates the input energies used for the three-layer laminates had to be high enough.
As can be seen from .ig. 2, the contact force for 1050-1050-2024 was 6.54% higher than 1050-2024-1050, whereas the central point displacement of 1050-1050-2024 was 19.16% lower than 1050-2024-1050 according to Table 1. Since the impact loading caused bending in the laminate, the materials of the front and back layers which were farther from the neutral plane affected the laminate displacement more considerably in comparison with the middle layers. Therefore, placing aluminum 2024-T3 at the bottom of the laminate 1050-1050-2024 resulted higher difference in the lami- The low-velocity impact test outputs of the laminates including the permanent transverse displacement of the central point of the front and back faces, the radius of the permanently deformed area on the front face and the remaining impactor kinetic energy are listed in Table 1. The input kinetic energy imposed by the impactor was the same for the both three-layer laminates. Nevertheless, the 1050-2024-1050 laminate represented higher central displacement and absorbed higher energy in the course of impact loading in comparison with the 1050-1050-2024 laminate. The higher absorbed energy can be interpreted as higher amount of damage occurred in the laminate as a result of impact loading. However, the plastic zone radius was lower for the 1050-2024-1050 laminate in comparison with the 1050-1050-2024 laminate. This could be because the 2024 layer was closer to the impactor. In case of two-layer laminates, although the input kinetic energy for the 1050-2024 laminate was higher than the 1050-1050 laminate, higher displacement, plastic radius and absorbed energy were occurred for the 1050-1050 layup. The front face damages are shown in .ig. 3.

INITE ELEMENT MODEL
There are various analytical and computational models for simulating impact mechanics. Qiao et al. [14] noted various models for simulating impact mechanics and presented a comprehensive classification of the models. There are several analytical models such as the rigid-body dynamic model, the stress wave propagation in perfectly and not perfectly elastic materials and the nonlocal models. Among the important computatio-  nal models, finite element and finite difference models, mesh-free methods and nonlocal numerical methods such as peridynamics can be considered. In this paper the mechanical behavior of ABML structures were investigated by finite element modeling using LS-DYNA finite element code.
.inite element analyses were carried out for studying the effects of the stacking sequence of the metal layers on impact behavior of adhesively bonded metal laminates. Various metal laminates with different sequences of various materials including aluminum 6061-T6, steel, lead and aluminum 6061 were numerically modeled. The mentioned materials were chosen because of their diverse mechanical properties.
Disk-like laminated structures with a fixed diameter of 210 mm and a fixed whole thickness of 8.3 mm were considered. As the number of metal layers increased in a constant whole thickness, the thickness of each metal layer was decreased. Target laminates were collided by a spherical rigid impactor with 13 mm radius, 3 m/s velocity and 87 J kinetic energy. .or all cases the adhesive layer thickness was 0.1 mm and the peripheral boundaries were set as clamped. The metal layers were considered to be bounded together using the epoxy adhesive of EC 2214. The model of the plastic-kinematic-isotropic material was used for the both metal and adhesive layers. In the plastic-kinematic-isotropic model, the strain rate effect was considered by using CowperSymonds coefficients. A bilinear elastic-plastic behavior was assumed for all materials. The material parameters of the metals and the adhesive are presented in Table 2. In Table 2 ρ, E, ν, y σ and t indicate density, elastic modulus, Poissons ratio, yield stress and tangent modulus, respectively. The parameters C and p in Table 2 represent the CowperSymonds coefficients [15] relating the dynamic yield stress to static yield stress using Eq. (1): where d 0 0 , , , σ σ ε C and p represent the dynamic yield (d) (c) stress, the static yield stress, the strain rate and the CowperSymonds first and second coefficients, respectively.
A finite element axisymmetric model of adhesively bonded metal laminates is illustrated in .ig. 4. A twodimensional 4-noded element Thin shell 163 was employed to mesh the both adhesive and metal layers. This element can be used for axisymmetric models. A higher mesh density was utilized near the impact region to handle the highly nonlinear contact problem. Mesh convergence was studied considering the effect of mesh density on the contact force and the contact duration. The thickness of each adhesive layer was divided into 2 sections in all cases but the thickness of the metal layers was divided differently in the structures with different number of metal layers to obtain convergent results.
The finite element model used for studying the lowvelocity impact behavior of adhesively bonded metal laminates was validated based on the experimental results summarized in Sect. 2. .igure 5 compares the numerical and experimental variations of the contact force versus the contact duration for 1050-1050, 1050-2024, 1050-1050-2024 and 1050-2024-1050 laminates. As it can be seen from .ig. 5, the numerical results correlated reasonably well with the experimental results. The average difference between the numerical and experimental peak contact force was 7.5%.

LOW-VELOCITY IMPACT RESPONSES O. METAL LAMINATES
Metal laminates can be manufactured with different number and stacking sequences of metal layers. The number of metal layers in a constant whole thickness and also different stacking sequences can affect the impact behavior of adhesively bonded metal laminates considerably. The effects of the number of metal layers and different stacking sequence were studied using the validated finite element model.

Effect of Number of Metal Layers
.igure 6 shows the effect of number of metal layers on impact behavior of metal laminates. As can be seen from .ig. 6, by increasing the number of metal layers in a constant thickness of metal laminates, the contact force decreased whereas the contact duration and transverse displacement increased. However, a considerable change was not observed for the dissipated energy by increasing the number of metal layers.
According to previous studies [13], the yield stress and Youngs modulus of the metal layers were the key material parameters affecting the low-velocity impact responses. This can also be seen from .ig. 6, as the low-velocity impact responses varied in case of steel laminates more noticeably than aluminum laminates due to this fact that materials with relatively high yield stress and Youngs modulus are more sensitive to the number of metal layers in comparison to other materials [13]. In other words, the higher yield stress and Youngs modulus of metal layers, the higher sensitivity of the metal laminate to the number of metal layers in a constant whole thickness.

Effect of Stacking Sequence
Various stacking sequences of the metal laminates were considered and their effects were studied on the impact responses in order to determine how stacking sequence affected the low-velocity impact behavior of metal laminates. .or each impact response, two-to four-layer metal laminates were investigated to study the effects of various stacking sequences of the metal layers.
The metal laminates were coded in a way so they can be easily identified. The digits 1, 2, 3 and 4 represent aluminum 6061-T6, steel, lead and aluminum 6061 materials, respectively. Moreover, the digits of the designation code represent closest to farthest metal layer to the impactor from left to right. .or instance, code 23 represent a two-layer metal laminate made of steel and lead layers and the steel layer was struck by the impactor.
.igure 7 demonstrates the impact responses of the metal laminates including the contact force versus the contact duration, the contact force versus the transverse displacement and the impactor kinetic energy variations in the course of impact loading for all possible layups of two-layer metal laminates made of aluminum 6061-T6 and aluminum 6061 having different mechanical properties. According to .ig. 7, it was observed that a laminate with dissimilar metal layers (i.e. 14 and 41 laminates) represented impact responses in between the impact responses obtained for the laminates with similar metal layers (i.e. 11 and 44 laminates). Moreover, it can be seen from .ig. 7 that the sequence of the metal layers can influence the structure responses. However, this influence depends on the difference of the mechanical properties of the layers. It was previously found by the researchers [13] that in low-velocity impact behavior of metal laminate, the yield stress and the Youngs modulus of the layers were the key material parameters.
.igure 8 compares the low-velocity impact responses of different two-to four-layer metal laminate layups. To examine the effects of the position of changing the two key material parameters of the yield stress and the Youngs modulus in the laminate on low-velocity impact behavior of adhesively bonded metal laminates, aluminum 6061-T6 and lead were selected in .ig. 8 because of their considerable differences in yield stress and Youngs modulus.
Various layups were considered so the influencing factors related to the effects of stacking sequence on impact responses of the metal laminate can be recognized. The volume fraction, the materials of the first and last metal layers and the number of metal layers were found to be the influencing factors in low-velocity impact behavior of metal laminates from stacking .igure 8a shows that the volume fraction of the material in the metal laminates influences the contact force. This can be concluded by comparing the contact forces obtained for the laminates 13 and 133 in which the volume fraction of aluminum 6061-T6 was reduced by 33% from the laminate 13 to the laminate 133. It was found that the contact force decreased by 10.6% from the laminate 13 to the laminate 133. It should be noted that in all laminates modeled, the overall thickness was kept constant, therefore, in the laminate 133 each layer thickness was reduced by a factor of two third in comparison with the laminate 13. In addition to the volume fraction of the material in the metal laminate, the sequence of the materials plays an important role in contact force. Comparing the laminates 331, 313 and 133 indicates that by shifting the material with higher stiffness and yield stress closer to the top layer, the contact force increased. The maximum contact force among the metal laminates studied in .ig. 8a obtained for the laminate 113 in which two factors exist, namely the volume fraction of the material with higher stiffness and yield stress and positioning a material with higher stiffness and yield stress as the top layer. .or the laminate with minimum contact force (i.e. 331), exactly the opposite conditions were the case.

Contact Duration
The contact duration of various laminates made of aluminum 6061-T6 and lead layers are compared in .ig. 8b. The contact duration was found to be affected by the volume fraction of the materials. The contact duration obtained for the laminate 133 was 13.5% higher than the laminate 13 due to 33% higher volume fraction of lead in the laminate 133 in comparison with the laminate 13. Moreover, it was found that the material of the first and last layers was also influencing on the contact duration. As can be seen from .ig. 8b, the contact durations obtained for the both of laminates 133 and 331 were lower than the contact duration of the laminate 313. Therefore, positioning a layer with high stiffness and yield stress on the top or bottom of a laminate caused shorter contact duration. This was because one of the loading mechanisms took place when a clamped laminate being impacted was bending and clearly in bending the effect of the outer layers are of more dominance.

Maximum Transverse Displacement
It was found that for the maximum transverse displacement, the same factors were significant as for the contact duration. Similarly, the material volume fractions and the material characteristics of the first and last layers were determined as the dominant factors in maximum transverse displacement. Therefore, the maximum and minimum transverse displacements were obtained for the laminates 313 and 113, respectively (see .ig. 8c).

Dissipated Energy
Typical variations of the impactor kinetic energy with respect to time are shown in .ig. 7c. The impactor kinetic energy level remained unchanged after leaving the target laminate. The difference between the initial and the final impactor kinetic energy when the impactor left the laminate can be considered as the dissipated energy. The majority of this dissipated energy was spent on the metal layers plastic deformation as other mechanisms of energy dissipation such as penetration, perforation, delamination and crack growth were considered to be negligible in low-velocity impact loading.
Khoramishad and Bagheri Tofighi [13] showed that among the mechanical properties of the layers in a metal laminate, the Youngs modulus and the yield stress were the most influencing material parameters on the main impact responses in the metal laminates, however, the yield stress imposed more considerable effect on the dissipated energy. Moreover, it was shown that increasing the yield stress and Youngs modulus resulted different effects on dissipated energy. This was in contrary to the other impact responses on which increasing the material stiffness and yield stress had similar effects. A higher yield stress value caused a lower dissipated energy, while a higher Youngs modulus value resulted in a higher dissipated energy. However, for the cases compared in .ig. 8d in which the laminates were made of aluminum 6061-T6 and lead, the difference between the yield stress values of the two materials was 27-fold while this difference for the Youngs modulus was 4-fold, therefore the effect of yield stress was predominant. Therefore, higher volume fraction of lead with lower yield stress in com-parison with aluminum 6061-T6 caused an increase in the dissipated energy. That was the reason why the dissipated energy for 133 was more than 13.
Comparing the dissipated energies of the laminates 331, 313 and 133 indicates that by shifting the material with higher yield stress closer to the top layer, the dissipated energy decreased. The dissipated energy value for 331 was about 10% higher than that for 133. .igure 8d shows that among the laminates compared the maximum dissipated energy gained for the 331 laminate while the minimum dissipated energy obtained for the 113 laminate. That was because in the 331 laminate the maximum fraction volume of lead with low yield stress was present and the lead layers were positioned closer to the impactor. Whereas, in the 113 laminate, the minimum fraction volume of lead was present and the aluminum 6061-T6 layers with high yield stress were positioned closer to the impactor.

CONCLUSIONS
Mechanical behavior of adhesively bonded metal laminate structures was investigated in this paper using both experimental investigations and finite element modeling. According to the experimental results, the volume fraction of the materials and stacking sequence of the metal layers could affect the impact behavior of the structures.
To further investigate the effects of stacking sequence on the low-velocity impact responses of the metal laminates a finite element model was developed and validated against the experimental results. The contact force, the contact duration, the dissipated energy and the transverse displacement were considered as the main low-velocity impact responses. The results showed that by increasing the number of metal layers in a constant thickness, the contact force decreased whereas the contact duration and transverse displacement increased. However, the effect of the number of metal layers was found to be rather negligible in dissipated energy.
Two-layer metal laminates were studied under lowvelocity impact loading and it was observed that the metal laminates with dissimilar layers represented impact responses in between the impact responses of the corresponding laminates with similar metal layers. Moreover, the results obtained for two-to four-layer metal laminates showed that there were several factors influencing the low-velocity impact responses of adhesi-vely bonded metal laminates. The material of the first and the last metal layer, volume fraction of the metal materials employed in adhesively bonded metal laminates and the number of metal layers were the important factors. Considering a material with high Youngs modulus value for the first and last metal layer caused a reduction in the transverse maximum displacement and the contact duration of the structure due to the effective role of the outer metal layers on the bending deformation of metal laminate. Placing a metal layer with high yield stress as the first metal layers caused an increase in the contact force and caused a reduction in the dissipated energy. It was also found that the volume fraction of the metal layers significantly affected the low-velocity impact responses. Considering the influencing factors and their corresponding effects on the low-velocity impact responses of the metal laminate can assist engineers in tailoring the metal laminate structures subjected to low-velocity impact loading.