4.1. Numerical Analyses of Static Compression Using the Simplified Y Element
The smallest, repetitive subarea that can be separated from a structure is a cell with the base of an equilateral triangle, the sides of which are perpendicular to the walls of three adjacent honeycomb cells. A separated single cell (the Y element) is shown in
Figure 3.
As indicated in [
13,
14,
15], structures consisting of repeating elements can be analyzed by solely focusing on a model that has been simplified down to the level of a segment, using appropriate boundary conditions. In the case of the analyzed structures, these conditions were limited to the treatment of the sides of the subarea as planes of symmetry.
The geometric model shown in
Figure 4a, when prepared for discretization, consisted of three walls connected to each other in line with one of the longer edges. The walls were connected to each other at an opening angle of 120°.
Four-node, fully integral shell elements with five integration points across the element thickness were used to build the finite element mesh, as presented in
Figure 4b. The characteristic dimension of the element was 0.23 mm, and there were 100 of them at the wall height. The selection of the element size was preceded by an analysis of the impacts that changing this parameter would have. At levels below the dimensions described above, no improvements were observed. The Belytschko–Tsay elements were used. To carry out the static compression analysis, all degrees of freedom were fixed in the top nodes and in the nodes next to the plane that forces motion (except that which was translational in the
Z-axis). The boundary conditions, along with the symmetry conditions and the local coordinate systems, are shown in
Figure 4c.
In order to accurately reproduce the actual structure and identify imperfections, the tested samples were subjected to computed tomography (CT) studies.
A SkyScan 1174 tomograph was used for this task. The obtained point clouds were processed to reduce noise, and then they were used to generate a polygonised structure; discrete, triangular surfaces were created between the closest points. In this way, spatial models of the cores were obtained, which was useful for making measurements. The axonometric views of the CT models for the 1/8-0.0015 and 3/16-0.0015 samples are shown in
Figure 5.
Analyzing the actual structures of the cores via the creation of three-dimensional models using computed tomography allowed for the examination of the basic geometric parameters, e.g., the length of individual walls, the size of the cell, and the bending radii of the walls near the joints. Exemplary measurement results are shown in
Figure 6.
The results of the geometric measurements, averaged for individual samples, are presented in
Table 2.
On the basis of the data presented above, it can be seen that the double walls have a smaller width and the single walls have a higher width than those declared in the catalog cards. Moreover, the portion of the wall that is arched is a distinct part of it. The obtained measurement results allowed us to conclude that the bend radii of the walls depend mainly on their thickness, and that the dimensions are linked by a ratio of approximately 10:1.
In order to check the impact of the above-described geometric parameters, models of the structure of the sample designated as 1/8-0.0015 were built, one using a simple, idealized version, and the other constructed using the mapped shape of the walls according to the parameters described in
Table 2. Both versions are shown in
Figure 7, which demonstrates that, in addition to the visible rounding of the edges of the single wall (marked in red), the width of the double wall (marked in blue) has been reduced in favor of the single wall.
The introduced change in the geometry resulted in differences in the form of structural deformation that were apparent as early as the initial loss-of-stability phases. As can be seen in
Figure 8, there are seven ridges on the flat core walls, and there are six smaller ridges on the rounded sides.
Minor differences in the geometry of the compared models were also reflected in the stress–strain characteristics obtained via the compression test. The value of the critical stress decreased by 11.6% (from 8.65 MPa to 7.64 MPa), and the value of the mean breaking stress decreased by 15.5% (from 5.98 MPa to 5.05 MPa). The value of the longitudinal modulus of elasticity before loss of stability also changed by 28.1% (from 3.11 GPa to 3.24 GPa). These are significant differences, thus the change in the model’s geometry should be taken into account in simulations. A comparison of the discussed characteristics, with reference to those obtained experimentally [
21], is presented in
Figure 9.
The constitutive model MAT_003_PLASTIC_KINEMATIC [
19] was used to reflect the behavior of the material from which the tested Y element was made: aluminum alloy Al 5052-H39. It is a bilinear model, the stress–strain relationship of which is described by defining the tangent of the angle of the elastic part and the part of the kinematic strengthening. The parameters and mechanical properties of the material were taken from a report published by NASA [
22] because the parameters of the thin, multi-rolled foil significantly differ from the parameters of the material from which it was made. In [
22], the results of tests of the foil composed of the Al 5052-H39 alloy, as well as those of the foil fragments cut from the finished structure of the cellular core, are presented.
The basic parameters of aluminum Al 5052-H39, as used to make cores with a honeycomb topology in the untreated form [
23], in the form of foil [
22], and in the form of a fragment of the core [
24], are presented in
Table 3. After plastic processing, the material is characterized by significantly lower strength parameters. In the form of a film, it has a 19% lower Young’s modulus, a 30% lower yield point, and a 25% lower tensile strength. The sample cut from the core fragment has a Young’s modulus that is up to 46% lower, a 32% lower yield point, and a 35% lower tensile strength.
Due to the fact that the developed Y-element models were used to determine the mechanical properties of the statically compressed aluminum honeycombs without the need for experimental testing, the properties of the Al 5052-H39 foil were considered in the subsequent numerical calculations.
4.2. Numerical Analysis of the Honeycomb Structures Using the Homogenous Model
The aim of the tests described below was to assess the effectiveness of using the results of the simulation of the Y element in models that describe the global response of this type of material without a detailed analysis of the behavior of their internal structure.
Geometric models, as shown in
Figure 5, were analyzed in order to identify fully closed cells and the possible damage resulting in the formation of open cells. Then, the volumes of the solids bounded by the closed walls of the cores were measured. Measurement results for the same types of samples were averaged. A visualization of an example of the discussed process is presented in
Table 4. The samples with smaller core sizes had a noticeably higher proportion of closed-cell volume, considering that they had the same global dimensions.
The previous experimental research of honeycomb structures under various levels of strain rate loading carried out by the authors [
21] was used to develop the assessments for this paper. By presenting the experimentally obtained, absolute increases in plateau stress [
21] in the domain of the air volume locked in by the cell walls, an almost-linear characteristic was obtained, which can be observed in
Figure 10. This observation may indicate that the main factor influencing the final shape of the stress–volumetric strain characteristic is the increase in the pressure of the air trapped inside the cells, especially for dynamic testing (strain rate of 3.8 × 10
2 and 3.3 × 10
3 1/s).
Another argument for the correctness of this assertion is the conclusion made on the basis of the observation of images recorded with a high-speed camera in the dynamic test using a split Hopkinson pressure bar (3.3 × 10
3 1/s). As shown in
Figure 11, the gas began to leak from the inside of the sample during the final compression phase (t = 333.30 µs). It became visible due to a dust cloud—a mixture of air escaping from the inside of the sample and fine particles of the damaged core and resin. This phenomenon took place in each case after reaching approximately 50% of the deformation; it continued after the maximum displacement of the initiating bar face, and this continued even after its plane lost contact with the sample plane. The described phenomenon may prove that, during the test, strong air compression takes place inside the core, which is released only after complete compression.
The above conclusion is important because none of the constitutive models available in commercially used computing environments take into account this kind of phenomenon’s influence on the change in the global mechanical properties of the structure.
Therefore, in order to describe the behavior of aluminum honeycomb structures in terms of strain rates higher than quasi-static, it is necessary to take an approach that will account for both the reaction of the deformed, thin-walled core and that of the air pressure inside the cells.
Assuming, as a simplification, that the air closed inside the cells is compressed without heat exchange, the increase in pressure ∆P can be described by a simple equation [
25],
where
P0 is the initial pressure (atmospheric).
Then, the indirectly measured stress value of the real structure would be the sum of the stress value in the structure (treating it as homogeneous according to the assumption made) and the pressure at a given time step. Due to the fact that the deformation of the core takes place only in the direction of displacement, the change in volume can be treated as a change in the height of the sample; as a result, the expected value of the measured stress is described by the formula [
25]
where
is the stress in the core of the sample treated as a continuous medium, and
is the initial core height.
The assumptions formulated above lead to the conclusion that the air trapped inside the cells causes an increase in the noted stress value, regardless of the strain rate. There are studies [
24] which prove that the key aspect linking the structure’s response with the duration of its destruction are air leakages resulting from imperfections in the structure and the material discontinuities formed in the process of deformation. The assumption can be modified in accordance with research carried out by Xu et al. [
24] and Hu et al. [
25], in which it was shown that the cross-sectional area of a honeycomb block also changes during axial compression. The air leakage δ is determined by the relationship [
21]
Therefore, the pressure value can be determined by the formula
where
is the volumetric strain.
It is assumed that
δ is the core failure time of the
t’ function and differentiating Equation (4) over time, assuming that one obtains
and the leakage rate is
Based on the relations presented above, it can be noticed that the pressure value and the rate of the leakage are related to the strain rate
. Researchers developing analytical models for this type of issue [
12], using the assumption that the strain rate is constant during the test, derived a relationship that links the pressure change inside the core with the strain and the leakage rate
Equation (16) does not contain any unknown parameters other than the leakage rate. The authors of [
25] state that this rate should be determined empirically by comparing the stress–volumetric strain results obtained during the compression of the samples with the cores tightly closed between covers, and those with openings releasing air during the test. The leakage intensity is very similar when the results of testing honeycomb structures with the same t/d ratio are compared, and its value depends on the strain rate. The obtained characteristics of the leakage intensity as a function of the strain rate for one of the cases analyzed in [
25] (1/8-5052-0.001) is presented in
Figure 12. In this case, the leakage rate increased almost in direct proportion to the value of the strain rate. This means that a tenfold increase in strain rate resulted in at least a tenfold increase in the leakage rate.
To perform simulation tests of the uniaxial compression of the homogenous honeycomb materials, a model with a cylinder geometry with a diameter of 25 mm and a height of 10 mm was used, reflecting the global geometry of the sample core. It consisted of 525 elements with a hexagonal topology.
The material model MAT_26_HONEYCOMB, available in the LS-DYNA system, was used to map the core, which was treated as a homogeneous material [
19].
In MAT_26, in the uncompressed state (initial state, loss of stability, and progressive folding), the material retains its orthotropic properties, and the stress tensor components remain unconnected from each other so that the strain component in one local direction does not cause reaction forces in the others. Modules of the longitudinal and shear stiffness in particular directions depend on the given modules of the initial stiffness and the stiffness of a fully compressed (compacted) structure. These dependencies are as follow [
19]:
where
and
E is the modulus of elasticity of the core material;
G is the shear modulus of the core material;
,
, and
are the modules of elasticity of the uncompressed cores;
,
, and
are the shear modules of the uncompressed cores;
is the relative volume (the ratio of the current volume to the initial volume), and
is the relative volume at which the core is considered fully compressed and it transforms into a linear elastic characteristic (the relative volume of total compaction).
In addition, the material model requires the definition of a set of curves that present the material characteristics obtained in the experimental tests: compression in each of the basic directions and shear in each of the base planes. There are two ways to define these characteristics. The first one is to determine the magnitude of stresses as a function of the relative volume (
V). It is also possible to determine the magnitude of stresses as a function of volumetric strains, defined as
Finally, the components of the stress tensor are calculated according to the following relationship:
After the process of updating the stress values is completed, they are converted into a global form.
For the modeling of individual structure types, the parameters presented in
Table 4 were adopted. In each case, the
σij – εij characteristics were developed by selecting characteristic points from the curves obtained by the FEM simulation using the Y-element model.
The development of a model capable of taking into account the change in the nature of the response of the change of the initial conditions causing the strain-rate increase was based on one of the available methods describing the behavior of air-filled elements.
The model should allow us to describe the change in gas pressure, along with the change in the volume inside where it was located. It should also take into account the leakage in the calculation of the pressure changes and the application to the surface and the spatial boundary of the element. All of the possibilities mentioned above are offered by one of the simplest models: AIRBAG_SIMPLE_AIRBAG_MODEL [
19]. The current value of the pressure acting on the boundaries of the vessel domain is calculated using the equation of state [
19]:
where
P is the pressure,
is the density, and
e is the internal energy of gas.
The γ coefficient is the adiabatic exponent: the ratio of specific heat at a constant pressure
to specific heat at a constant volume
[
25]
The rate of changes in the air mass
m inside the volume, with time, is described by the relationship [
26]
The value of the air mass, which flows in via subsequent time steps mi, is defined by the appropriate characteristic. There are two options for determining the mass of the outflowing air mo: by defining the area of the holes through which the air leaks and their shape coefficient, and by defining the mass characteristics over time. There is also a possibility of making the size of the surface of the holes and the shape factor dependent on the value of the pressure inside.
In the discussed approach, the components of the energy balance [
26]
are as follows:
is energy change caused by the mass of inflowing gas,
is the change in energy caused by the mass of the outflowing gas,
is work achieved by the pressure per volume change.
The described boundary condition was applied to the surfaces of the elements on the outer walls of the cylinder, as shown in
Figure 13. This allowed for a direct transfer of the forces resulting from the increase in pressure to the sample boundaries. The parameters used to describe the air enclosed inside the sample are presented in
Table 5. All parameters related to the influence of the air mass were omitted.
The initial boundary conditions are presented in
Figure 14. An explicit scheme of the integration of the equations of motion was used for the calculations.