Composite Structures: Sciencedirect

Composite Structures 200 (2018) 886–909
Contents lists available at ScienceDirect
Composite Structures
journal homepage: www.elsevier.com/locate/compstruct
3D printed architected polymeric sandwich panels: Energy absorption and T

structural performance
⁎
H. Yazdani Sarvestania, A.H. Akbarzadeha,b, , H. Niknama, K. Hermeneanc
a
AM3L Laboratory, Department of Bioresource Engineering, McGill University, Island of Montreal, QC H9X 3V9, Canada
b
Department of Mechanical Engineering, McGill University, Montreal, QC H3A 0C3, Canada
c
MACHINA Corp., Edmonton, AB T6H 2H3, Canada
A R T I C LE I N FO A B S T R A C T
Keywords: Lightweight architected cellular cores have been introduced as an advanced alternative to improve the overall
Architected 3D printed sandwich panels performance of sandwich structures. In this study, we implement semi-analytical and finite element approaches
Cellular cores and conduct experimental impact tests to evaluate the performance of 3D printed lightweight sandwich panels
Energy absorption with architected cellular cores of programmable six-sided cells. Changing the geometrical parameters of the cells
Low-velocity impact
leads to cellular cores of hexagonal, rectangular and auxetic topologies. A semi-analytical methodology is de-
Modified higher-order shear deformation
veloped for conducting structural and low-velocity impact analyses based on a modified higher-order shear
theory
deformation theory. The standard mechanics homogenization is implemented through finite element modelling
to accurately predict the effective mechanical properties of architected cellular cores. We apply explicit large
deformation finite element analysis using ANSYS to analyze the elasto-plastic behavior of architected sandwich
panels under a low-velocity impact. To experimentally corroborate the developed theoretical and computational
models and to evaluate the manufacturability of the architected sandwich panels, we use the fused deposition
modeling to 3D print samples of polylactic acid biopolymers. Uniaxial tensile test is first used to characterize the
polymer. We then conduct low-velocity impact tests to investigate the energy absorption capability of archi-
tected sandwich panels. X-ray micro-tomography is finally employed to study the microstructural features of
panels before and after the impact. The experimental and numerical results show that the auxetic sandwich panel
is potentially an appropriate candidate for energy absorption applications due to its high-energy absorption
capability and a minimum response force transferred from the 3D printed panel.
1. Introduction demand for improvement of structural, aerodynamic and thermal in-

sulation/heat exchanging performance of sandwich structures for a
Lightweight sandwich panels are nowadays extensively used in wide range of engineering applications. It has been recently shown that
aerospace, marine, automotive, windmills and construction industries the multifunctional performance of sandwich structures can be im-
due to their high flexural stiffness-to-weight ratio, excellent thermal proved by the engineering of microarchitecture of the cellular core in
insulation and high energy absorption capability [1]. Sandwich panels fabricating the sandwich structures [4]. The recent progress in ad-
consist of two thin but stiff face-sheets at the top and bottom of the vanced manufacturing technologies, e.g. additive manufacturing (3D
panel separated by a lightweight and relatively thick core. The light- printing) and laser cutting, has allowed manufacturing of architected
weight core connects the solid face-sheets to provide sandwich panels cellular cores which were impossible to be fabricated by conventional
with a high bending stiffness and buckling resistance [2] as well as a manufacturing processes, e.g. extrusion, milling and corrugation [5].
superb shear stiffness and energy absorption capability [3]. The core Using additive manufacturing for fabrication of sandwich structures
can be made of alternative materials but is usually in the form of cel- with architected cores leads to several advantages over the conven-
lular materials, e.g. foams or periodic lattices. In a sandwich panel, the tional sandwich structures with a hexagonal honeycomb or foam core.
face-sheets carry most of in-plane and bending loads while the cellular Among the advantages, we can refer to the programmability of multi-
core mainly carries transverse shear [1]. The complexity of material functional properties over a multiple length scale through controlling
and architecture of sandwich panels is growing due to the increasing the architecture and geometrical features of the cellular core [1].
⁎
Corresponding author at: AM3L Laboratory, Department of Bioresource Engineering, McGill University, Island of Montreal, QC H9X 3V9, Canada.
E-mail address: hamid.akbarzadeh@mcgill.ca (A.H. Akbarzadeh).
URL: https://www.mcgill.ca/bioeng/faculty-and-staff/abdolhamid-akbarzadeh-shafaroudi (A.H. Akbarzadeh).
https://doi.org/10.1016/j.compstruct.2018.04.002
Received 18 December 2017; Received in revised form 13 March 2018; Accepted 2 April 2018
Available online 05 April 2018
0263-8223/ © 2018 Elsevier Ltd. All rights reserved.
H. Yazdani Sarvestani et al. Composite Structures 200 (2018) 886–909
However, the durability of additively manufactured sandwich panels sandwich panel is affected by its: (a) geometry (length, width and
with a complex microarchitecture is yet to be explored for different thickness); (b) mass (relative density); (c) properties of the base ma-
range of materials [6]; a reason which motivated us to investigate the terial used to construct the cellular core and solid face-sheets; and (d)
performance of 3D printed architected sandwich structures in the cur- the core cell topology. The current study focuses specifically on the
rent research study. effect of cellular core topology and relative density on the energy ab-
The structural performance and energy absorption capability of an sorption capability and structural design of 3D printed architected
architected sandwich panel mainly depend on the material properties sandwich panels; hence, the total dimensions and material properties of
and geometrical features of solid face-sheet and the cellular core. the core and face-sheets are maintained constant throughout the study.
Among all cell topologies for the core of architected sandwich panels, This paper is organized as follows: Section 2 introduces different
hexagonal honeycomb has been commonly used and analyzed as a parameters used in the theoretical, numerical and experimental ana-
cellular core [7–9]. Sandwich panels with conventional honeycomb lyses. Section 3 presents the mathematical formulation of architected
cellular cores are stiff, lightweight and absorb high energy under im- sandwich panels subjected to a low-velocity impact load. Section 4
pact and shockwaves for applications in sports goods as well as auto- describes the homogenization technique to obtain the mechanical
motive and aerospace industries [10–12]. However, they have some properties of different cell topologies used in the theoretical formula-
issues due to their closed-cell architectures including gas retention, tions. Section 5 introduces the numerical model using ANSYS
leading to low thermal conductivity and moisture trapping [1]. Workbench 18.2 to analyze energy absorption capability. Section 6.1
Moisture trapped in the closed-cell cores increases the weight and shifts verifies the theoretical formulations. Section 6.2 reports the effects of
the center of gravity, a problem which can be resolved by using open- core topology on the structural response of panels. Section 6.3 presents
cell cores [1]. An alternative sandwich core is open-cell cores such as the effects of core topology on the energy absorption capability of
truss-like structures [13–15]. It has been shown that truss-like cellular sandwich panels subjected to a low-velocity impact load. Finally, the
cores, with low relative densities, possess significantly higher buckling effect of low-velocity impact loads on the microstructural damage of the
resistance and energy absorption capability compared to hexagonal 3D printed architected sandwich panels are studied in Section 6.4.
honeycomb cores [16–18].
More recently, auxetic cores have been examined due to their
unusual deformation mechanisms, which is negative Poisson’s ratio or 2. Sandwich panel and experimental test configuration
“auxetics” behavior. Traditionally, auxetic cellular cores have been
fabricated via a multi-step manufacturing process which involves rather Dimensions of the sandwich panel including length (a), width (b)
complex design of molds and heat compression processes [19]. 3D and total thickness (h) are shown in Fig. 1. The coordinate system (x, y,
printing technology is able to precisely fabricate auxetic cores with z) is located at the middle plane of the panel. The selected dimensions
complex geometries directly from computer-aided design (CAD) files are listed in Table 1, which are chosen based on ASTM standard D3763
[19]. When architected sandwich panels are used in aerospace and [50] and testing limitations. The programmable six-sided cells, which
automotive applications as an exterior frame, low energy impact pro- include rectangular, hexagonal and auxetic cells, are considered in this
tection of small objects hitting the frame is also an important factor to study for the cellular core of sandwich panels.
consider. Auxetic cores possess promising properties in the aforemen- Fig. 2a shows the unit cell configurations of programmable six-sided
tioned applications compared to other topologies of cellular solids. Due cells. To investigate the effect of topology and relative density of cel-
to the auxetics behavior, auxetic cellular cores reveal unique mechan- lular core on the energy absorption capability of the architected sand-
ical properties such as increased indentation resistance [20], shear re- wich panel, while maintaining total dimensions constant, the following
sistance [21], fracture toughness [22–24] and energy absorption ca- three independent design variables are defined for programmable cells:
pacity [25–28]. The auxeticity of the cellular core of sandwich angle between the cell walls θ, number of unit cells in the out-of-plane
structures could improve their specific mechanical properties, e.g. re- direction of the core Nh and the number of unit cells in the in-plane
duced deflection during flexural bending [25] and increased shear direction Nv. The values of three design variables considered in this
modulus [21], making these architected structures ideal for applica- study are shown in Table 2. The number of unit cells in the out-of-plane
tions as sensors and actuators [29]. The structural responses of sand- and in-plane directions (Nh and Nv) is selected in order to completely fill
wich panels with auxetic cores have already been studied for small the panel global dimensions (a and b). The cell wall angle θ is also
deformation conditions [30–32], large deformation situations [33,34], constrained to be an integer in order to reduce the design possibilities;
bending responses [35–37] and impact resistance [38,39]. Further- for a regular cell (positive Poisson’s ratio), θ > 90°; while for auxetic
more, it has been shown that auxetic cellular cores have better shear cells (negative Poisson’s ratio), θ < 90°. The constraint of θ > 65° has
properties compared to conventional lightweight cores of positive been imposed based on the wall thickness to prevent the overlap of the
Poisson’s ratio [40,41]. The auxetic core is one type of architected cell walls. Examples of 3D printed sandwich panels, manufactured by
cellular cores; recent studies have shown that architected 3D printed fused deposition modeling (FDM) made of polylactic acid (PLA) fila-
cores present a new class of highly optimized energy absorbers that ments, for different combinations of the independent design variables
offer more flexibility in controlling and tailoring their response under for the six-sided cells (for θ = 70°, 90° and 120°) are shown in Table 2.
impact loads than conventional sandwich structures [42–44]. Another
type of architected cellular cores is octet-cubic cells, which are maxi-
mally stiff in all directions and they possess low-density with maximum
achievable toughness and crush-resistance [45]. Although a few studies
have been performed on architected sandwich panels, e.g. flexural be-
havior [45–47] and out-of-plane compressive strength [48,49], less
attention has been paid to explore the energy absorption capabilities of
3D printed sandwich panels with architected cellular cores.
In the present study, we investigate the structural and energy ab-
sorption capabilities of 3D printed sandwich panels with architected
periodic cellular cores. The deformation mechanism of the lightweight
sandwich panels is studied to effectively design and optimize their
microarchitecture to simultaneously satisfy the energy absorption and Fig. 1. Geometry of an architected sandwich panel and the considered co-
structural requirements. The energy absorption capability of a ordinate system.
887
Table 1 performed on specimens fabricated in the orientation of 45° with re-

Geometric parameters of sandwich panels (Unit: mm). spect to the loading axis in order to provide an estimation of the ani-
Length (a) Width (b) Total thickness Core thickness Face-sheet sotropy of the sandwich panels (see Fig. 4a). Fig. 4b presents the stress-
(h) (tc) thickness (ts) strain curves of six 3D printed PLA coupons under tensile loading. The
averaged experimental properties of PLA are reported in Table 3.
100 100 50 40 5
2.3. Low-velocity impact tests

2.1. Sample preparation
The low-velocity impact tests are performed on sandwich panels
The MK2 3D printer of MACHINA Corp. with the layer height of with different 3D printed cores as shown in Table 2 using a drop weight
0.25 mm manufactures the architected sandwich panels. MK2 3D machine based on the guidelines given in the ASTM standard D3763
printer uses fused deposition modeling technology (FDM) to additively [50]. The impactor has a mass of 12 kg and a diameter of 25.4 mm with
manufacture architected samples of PLA polymers. FDM is the most a semi-spherical tip. During the impact test, the specimens are con-
commonly used 3D printing technology; polymeric filaments are heated strained between two parallel rigid supports with 75 mm diameter
and extruded through an extrusion head that deposits the molten holes in the center of two parallel supports (see Fig. 5). Sufficient
polymers in x- and y-coordinates, while the build table lowers the ob- clamping pressure is applied to prevent the specimen from slipping
ject layer by layer in the z-direction [51]. Considering the layer-by- during the experiments. The two measured parameters in the experi-
layer fabrication process of FDM 3D printing, the layer orientation and ment are impactor load and impact velocity over the time of the impact
toolpath of depositing the molten filaments are influential on the me- test. The load is measured by a load cell (penetration force sensor,
chanical properties of the 3D printed material. Two types of sandwich piezoelectric load sensor, with a load capacity 22.5 kN) and the velocity
panels (in-plane or out-of-plane cells) with three different topologies of of impactor is measured by a velocity detector using a photo detector
cellular cores (hexagonal, rectangular and auxetic), shown in Table 2, block and a flag [54]. The flag, a double pronged stainless steel piece, is
are 3D printed. The 3D printed panels are manufactured based on the mounted on the drop weight. The displacement is calculated according
geometrical dimensions given in Table 1. A few examples of 3D printed to the impact velocity, mass and load based on the following equations:
t 1 t
sandwich panels are shown in Fig. 3. It is worth mentioning that the z(t) = ∫0 v(t ) dt and v(t) = vi + g t − m ∫0 F (t ) dt , where F(t), m, g, vi, v
toolpath of 3D printing process can affect the damage mechanism of 3D (t) and z(t) are load at the instant of t, impactor mass, gravity accel-
printed architected sandwich panels subjected to an impact load. eration, impact velocity, the velocity of impactor at the instant of t and
displacement at time t, respectively. The change in the total energy of
the impactor is also calculated by the impactor’s velocity, mass and load
2.2. Material properties 1
according to the following equation; E(t) = 2 m(v i2−v(t )2) + m g z(t),
where E(t) is the energy of the impactor which is assumed to be equal to
In order to correctly characterize the panels, it is essential to mea-
the energy of the 3D printed sandwich panel at time t (t = 0 is associate
sure the base material properties. Dogbone tensile specimens in
with the instant that the impactor touches the top face of the panel).
agreement with the ASTM standard D638 [52] are 3D printed with the
same material used for all architected sandwich panels as seen in Fig. 4.
The tensile tests are performed using a MTS mechanical tester (C43 3. Mathematical formulation for theoretical analysis
frame) with a 10 kN load cell. Six specimens are used to perform uni-
axial tensile tests. It is often the case with additive manufacturing 3.1. Modified higher-order shear deformation theory
processes that material properties exhibit a certain degree of aniso-
tropy. The study performed by Yang et al. [53] showed that the 3D The kinematics of the deformation in the theoretical analysis of this
printing orientation of 45° could provide an accurate approximation for paper is limited to the linear elastic material. For elasto-plastic beha-
the evaluating the material properties, e.g. Young’s modulus and vior, we have employed a computational approach by implementing
Poisson’s ratio, of 3D printed materials. Here, tensile tests are therefore large deformation finite element simulation. It is well known that shear
Fig. 2. (a) Unit cell of architected six-sided cells as the core of architected sandwich panels and (b) Applied periodic boundary conditions for homogenization of the
unit cell or representative volume element.
888
Table 2
Architected sandwich panels with alternative cellular core architecture.
Cell type
Auxetic core ș = 70° Rectangular core ș = 90° Hexagonal core ș = 120°
Cell
(Nh × Nv) (Nh × Nv) (Nh × Nv)
direction
In-plane
(5 × 3) (5 × 5) (5 × 8)
Out-of-
plane
(3 × 2) (5 × 2) (8 × 2)
m m
deformation theories are suitable for predicting the global responses of
composite structures [55]. In addition, these theories are simpler and
εxx = u 0,x + ∑ z iui,x , εyy = v0,y + ∑ z ivi,y,
i=1 i=1
computationally less expensive than the 3D elasticity models or layer- n m
wise theories [56–58]. The modified higher-order shear deformation εzz = ∑ jz j − 1 wj, γxy = u 0,y + v0,x + ∑ z iui,y
theory with an arbitrary number of terms for displacement components j=1 i=1
m
is to provide sufficiently accurate structural results. The components of
the displacement fields at a material point in a sandwich panel can be
+ ∑ z ivi,x ,
i=1
expressed as [59]: n m n m
m
γxz = w0,x + ∑ z jwj,x + ∑ iz i − 1ui, γyz = w0,y + ∑ z jwj,y + ∑ iz i − 1vi
j=1 i=1 j=1 i=1
u (x ,y,z ,t ) = u 0 (x ,y,t ) + ∑ z iui (x ,y,t )
i=1 (1a) (2)
m
3.2. Kinetics of impactor
v (x ,y,z ,t ) = v0 (x ,y,t ) + ∑ z ivi (x ,y,t )
i=1 (1b)
In this study, it is assumed that the vibration of impactor can be
neglected. During the loading phase, the total contact force Pc can to be
n
related to the local contact indentation δ(t) as [60]:
w (x ,y,z ,t ) = w0 (x ,y,t ) + ∑ z jwj (x ,y,t )
j=1 (1c) Pc (t ) = K C [δ (t )]1.5 (3a)
where i and j are indices representing the number of summation terms and the local contact indentation δ(t) is defined as:
in displacement field, from i = 1 to a desirable number m for the dis- δ (t ) = S (t )−w (xs ,ys ,z s,t ) (3b)
placement components in x- and y-directions of u (x, y, z, t) and v (x, y,
z, t), respectively, and from j = 1 to a desirable number n for the dis- where S(t) represents the lateral displacement of the impactor; w (xs, ys,
placement component in the z-direction of w (x, y, z, t). The linear zs, t) also represents the sandwich panel deflection at the impact loca-
strain-displacement relations are given as follows [55]: tion (xs, ys, zs). The contact stiffness Kc can be defined by [60]:
−1
4 1−νi2 1⎞
KC = Ri ⎜⎛ + ⎟
3 ⎝ iE Ez⎠ (3c)
Out-of-plane rectangular core Out-of-plane hexegonal core Out-of-plane auxetic core

Fig. 3. Examples of 3D printed PLA sandwich panels fabricated by FDM technology.
889
(a) (b)
Fig. 4. (a) 3D printed dogbone test coupons before and after tensile tests, (b) Engineering stress-strain curves of 3D printed PLA coupons under tensile load.
Table 3
Experimental properties of 3D printed PLA samples.
Averaged mechanical properties Young's modulus (E ) Ultimate strength (S u) Ultimate strain (ε u) Plastic Poisson’s ratio (υp)
2.7 ± 0.1 GPa 46.2 ± 1.3 MPa 0.021 ± 0.001 0.38 ± 0.003
mi S¨ (t ) + Pc (t ) = 0, S (0) = 0, S ̇ (0) = V0 (5)

where mi and V0 are, respectively, the mass and initial velocity of the
impactor.
3.3. Equations of motion
The equilibrium equations of a sandwich panel subjected to a low-

velocity impact load are obtained by using the principle of virtual
displacement [55]:
t2
∫t1
(δU + δV −δK ) dt = 0
(6)
Fig. 5. Low-velocity impact test configuration for energy absorption analysis of
3D printed architected sandwich panels. where δU, δV and δK are the virtual strain energy, virtual work done by
external loads, and virtual kinetic energy, respectively. Virtual strain
where Ri is the radius of the impactor tip and Ei and νi are Young’s energy δU is defined as [55]:
modulus and Poisson’s ratio of the impactor, respectively; Ez is the h /2 b a
transverse Young’s modulus of the face-sheet of the architected sand-

δU = ∫−h/2 ∫0 ∫0 (σxx δεxx + σyy δεyy + σzz δεzz + σyz δγyz + σxz δγxz
wich panel. During the unloading phase, the contact force Pc is defined + σxy δγxy ) dxdydz (7a)
as [60]:
The total virtual work done by the external applied loads δV is:
2.5
δ (t )−δ0 ⎤
Pc (t ) = Pmax ⎡
⎢
δV = ∫A (Nxx0 w,x δw,x + N yy0 w,y δw,y + 2Nxy0 w,y δw,x ) dydx
⎣ max−δ0 ⎥
δ ⎦ (4)
− ∫ (K C [S (t )−w (xs ,ys ,z s,t )]1.5 δw (xs ,ys ,z s,t )) dydx (7b)
A
where Pmax and δmax are the maximum contact force and maximum
local indentation during the loading phase, respectively. The perma- Finally, the total virtual kinetic energy δK including the kinetic
nent indentation δ0 equals to zero when δmax remains below a critical energy of impactor is written as:
indentation during the loading phase. We should mention that δ0 is h /2 1
equal to zero since the current semi-analytical analysis is elastic. The δK = ∫−h/2 ∫A ρ (uδu
̇ ̇ + vδv ̇ ̇ ) dx dy dz + mi S 2̇
̇ ̇ + wδw
2 (7c)
equation of motion of the impactor can be written as:
where ρ (x, y, z) represents the sandwich panel density and the mass
890
moment of inertia Ikj are defined as: (R1ixz ,R1iyz ) = (B55

i i
,B 45 i
) ui + (B 45 i
,B 44 i
) vi + (B55 i
,B45 i
) w0,x + (B45 i
,B44 ) w0,y
h /2
+ (D55ji ,D45ji ) wj,x + (D45ji ,D44ji ) wj,y
(I1,I1i,I2i ,I1j,I2j ) = ∫−h/2 ρ (1,z i,z2i,z j,z2j) dz (8)
(13d)
where i = 1, 2, …, m, and j = 1, 2, …, n. Employing the fundamental (R xzj ,R yzj ) = (D55ji ,D45ji ) ui + (D45ji ,D44ji ) vi + (B55j ,B45j ) w0,x + (B45j ,B44j ) w0,y
lemma of calculus of variations, the equilibrium equations of the + (D55j ,D45j ) wj,x + (D45j ,D44j ) wj,y (13e)
sandwich panels are derived:
M1jzz = B13j u 0,x + D13
ij
ui,x + B26j u 0,y + D36
ij
ui,y + B36j v0,x + D36
ij
vi,x + B23j v0,y
δu 0 : Nxx ,x + Qxy,y = I1 u¨ 0 + I1i u¨ i (9a) ij
+ D23 vi,y + D33j wj (13f)
δv0: Nyy,y + Qxy,x = I1 v¨0 + I1i vï (9b)
where the stiffnesses introduced in Eqs. (13) are defined as:
δw0: Qxz,x + Q yz,y + N0̂ −K C δ01.5 = I1 w¨ 0 + I1j w¨ j (9c) h
(Amn ,Bmn,Dmn ) = ∫−2h Cmn (1,z ,z 2) dz m,n = 1,2,3,6
2
i i i i
δui : Mxx ¨ 0 + I2i u¨ i
,x + R xy,y −R1 xz = I1 u (9d) h h
i i j j
(Bmn ,Dmn ) = ∫ Cmn (z i,z 2i ) dz , (Bmn
2 ,Dmn ) = ∫ 2h Cmn (z j,z 2j ) dz m,n
− h2 − 2
i i i i i
δvi: Myy ,y + R xy,x −R1 yz = I1 v¨0 + I2 vï (9e)
= 1,2,6
̂ K C δ1.5
δwj: −M1jzz + R xzj ,x + R yzj ,y + Nj− j
j = I1 w¨ 0 + I2j w¨ j
h
(9f) (B13j ,D13j ) = ∫−2h C13 (jz j − 1 ,j 2 z 2j − 2) dz
2
where N0̂ and Nĵ which are nonlinear terms associated with buckling h
i i
loads are: (Bmn ,B mn ) = ∫−2h Cmn (iz i − 1,i 2z 2i − 2) dz m,n = 4,5
2
h
N0̂ = N xx
0 0
w0,xx + 2N xy 0
w0,xy + N yy w0,yy ij
(Dmn ji
,Dmn 2
) = ∫ Cmn (jz j − 1 z i,iz i − 1z j ) dz m,n = 4,5
− h2
Nj ̂ = 0 j
N xx z wj,xx + 0 j
2N xy z wj,xy + 0 j
N yy z wj,yy (10)
(14)
The generalized stress and moment resultants given in Eq. (9) are where z-direction is along the thickness direction as shown in Fig. 1.
defined as: Finally, the governing equations are derived by substituting Eq. (13)
h /2 into Eq. (9).
(Nxx ,Nyy,Q yz,Qxy,Qxz ) = ∫−h /2 (σxx ,σyy,σyz,σxy,σxz ) dz ,
h /2 h /2
i
(Mxx i
,Myy i
,R xy ) = ∫−h /2 (σxx ,σyy,σxy ) z idz , (R1iyz ,R1ixz ) = ∫−h /2 (σyz,σxz ) iz i − 1dz , 3.4. Spatial solution
h /2 h /2
(R yzj ,R xzj ) =∫ −h /2
(σyz,σxz ) z jdz , M1jzz =∫ σ jz j − 1 dz
−h /2 zz In this section, we introduce a numerical methodology for solving
(11) the coupled governing differential equations in the spatial coordinate
system for alternative boundary conditions of the architected sandwich
The three-dimensional constitutive law can be written as [55]:
panel. The boundary conditions can be simply-supported (S), clamped
(k ) (C), free (F), or a combination of any of these boundary conditions. The
σ (k ) ⎡ C11 C12 C13 0 0 C16 ⎤ ε (k )
⎛ σxx ⎞ ⎢C12 ⎛ xx ⎞ Galerkin method is adopted here for solving the governing differential
yy C22 C23 0 0 C26 ⎥ εyy
⎜ ⎟ ⎢ ⎥ ⎜ ⎟ equations to predict the performance of architected sandwich panels of
σzz C C23 C33 0 0 C36 ⎥ ε
⎜ ⎟ = ⎢ 13 ⎜ zz ⎟ arbitrary boundary conditions. Mathematical expressions for boundary
σ ⎢ 0 γyz
⎜ yz ⎟ 0 0 C44 C45 0 ⎥ ⎜ ⎟
⎢ ⎥ conditions are presented here [55,62]:
σ
⎜ xz ⎟ ⎢ 0 0 0 C45 C55 0 ⎥ ⎜ γxz ⎟
σxy
⎜γ ⎟
⎝ ⎠ ⎢C C26 C36 0 0 C66 ⎥ ⎝ xy ⎠
⎣ 16 ⎦ (12) (a) Simply-supported edges (S):
(k ) u = v = w = 0 at x = 0,a
where Cij represent
the stiffness components of face-sheets (k = 1 and
3) and cellular core (k = 2); the stiffness components of architected u = v = w = 0 at y = 0,b (15a)
cellular core are obtained by the standard mechanics homogenization
(see Section 4). It is worth mentioning that the effective orthotropic
(b) Clamped edges (C):
multifunctional properties of cores can be engineered by tuning the cell
wall thickness, cell wall angle, relative density of unit cell and material ∂w ∂w
u=v=w=0= ∂y
= ∂x
= 0 at x = 0,a
properties of solid materials constituting the core [61]. The generalized ∂w ∂w
stress and moment resultants in LWT are expressed in terms of the u=v=w=0= = = 0 at y = 0,b
∂y ∂x (15b)
displacement functions by substituting Eqs. (12) and (2) into Eq. (11):
i i i
(Nxx ,Nyy,Qxy ) = (A11 ,A12 ,A16 ) u 0,x + (B11,B12,B16 ) ui,x + (A16 ,A26 ,A66 ) u 0,y (c) Free edges (F):
i i i i i i
+ (B16,B26,B66 ) ui,y + (A16 ,A26 ,A66 ) v0,x + (B16,B26,B66 ) vi,x i
Nxx = Qxy = Mxx i
= R xy = R1ixz = R xzj = 0 at x = 0,a
+ (A12 ,A22 ,A26 ) v0,y + i
(B12 i
,B22 i
,B26 ) vi,y + (B13j ,B23j ,B36j ) wj i
Nyy = Qxy = Myy i
= R xy = R1iyz = R yzj = 0 at y = 0,b (15c)
(13a)
where i = 1, 2, …, m, and j = 1, 2, …, n. It should be mentioned that
i
(Qxz ,Q yz ) = (B55 i
,B45 i
) ui + (B45 i
,B44 ) vi + (A55 ,A 45 ) w0,x + (A 45 ,A 44 ) w0,y the number of governing equations for the proposed modified higher-
order shear deformation theory are 3 + 2 m + n equations and each of
+ (B55j ,B45j ) wj,x + (B45j ,B44j ) wj,y (13b) the boundary conditions in Eq. (15) provides minimum 3 + 2 m + n
i i i i i i i i i i i i
equations to be able to solve the governing different equations. To
(Mxx ,Myy ,R xy ) = (B11,B12,B16 ) u 0,x + (D11,D12 ,D16 ) ui,x + (B16,B26,B66 ) u 0,y
implement the Galerkin method, the displacement field is expressed in
i i i i i i i i i
+ (D16 ,D26 ,D66 ) ui,y + (B16,B26,B66 ) v0,x + (D16 ,D26 ,D66 ) vi,x the following form for arbitrary boundary conditions [55]:
i i i i i i ij ij ij
+ (B12,B22,B26 ) v0,y + (D12 ,D22 ,D26 ) vi,y + (D13 ,D23 ,D36 ) wj
(13c)
891
∞ ∞
∂Φk (x )
u 0 (x ,y,t ) = ∑ ∑ 0
Ukc (t ) ∂x
Φc (y ), ui (x ,y,t )
k=1 c=1
∞ ∞
∂Φk (x )
= ∑ ∑ i
Ukc (t ) ∂x
Φc (y )
k=1 c=1
∞ ∞
∂Φc (y )
v0 (x ,y,t ) = ∑ ∑ 0
V kc (t )Φk (x ) ∂y
, vi (x ,y,t )
k=1 c=1
∞ ∞
∂Φc (y )
= ∑ ∑ i
Vkc (t )Φk (x ) ∂y
k=1 c=1
∞ ∞
w0 (x ,y,t ) = ∑ ∑ 0
Wkc (t )Φk (x )Φc (y ), wj (x ,y,t )
k=1 c=1
∞ ∞
= ∑ ∑ Wkcj (t )Φk (x )Φc (y )
k=1 c=1 (16) Fig. 6. ANSYS Model for the low-velocity impact analysis.
where 0
Ukc , 0
V kc , 0
Wkc ,
and i
Ukc , i
Wkcj
(k and c are integer numbers) are
Vkc
unknown coefficients determined by satisfying the governing differ-
ential equations and associated boundary conditions given in Eq. (15).
In addition, admissible functions Φk and Φc could be defined as [55]:
Φk (x ) = α1cosh(λk x ) + α2cos(λk x )−ζk (α3sinh(λk x ) + α4sin(λk x ))
Φc (y ) = β1cosh(λ c y ) + β2cos(λ c y )−ζ c (β3sinh(λ c y ) + β4 sin(λ c y )) (17)
where λk and λc are defined in Table 4 for alternative boundary con-
ditions. By implementing the Galerkin method and using the approx-
imate displacement fields given in Eq. (16) and governing equations literature
given in Eq. (9), we can solve the governing differential equations by:
p ∂Φ (x )
0
⎧ R1 (Ukc 0
,V kc 0
,Wkc i
,Ukc i
,Vkc ,Wkcj ) ∂x Φl (y ) ⎫ literature
⎪ ⎪
∂Φ (y )
0
⎪ R2 (Ukc 0
,V kc 0
,Wkc i
,Ukc i
,Vkc ,Wkcj )Φp (x ) ∂ly ⎪
⎪ ⎪
b a ⎪ R3 (U 0 ,V 0 ,W 0 ,U i ,V i ,W j )Φp (x )Φl (y ) ⎪
kc kc kc kc kc kc
∫0 ∫0 ⎨ R (U 0 ,V 0 ,W 0 ,U i ,V i ,W j ) ∂Φp (x ) Φ (y ) ⎬
dxdy
⎪ 4 kc kc kc kc kc kc ∂x l ⎪
⎪ 0 0 0 i i j ∂Φl (y ) ⎪
⎪ R5 (Ukc,V kc,Wkc,Ukc,Vkc,Wkc )Φp (x ) ∂y ⎪
⎪ 0 0 0 i i j ⎪
⎩ R6 (Ukc,V kc,Wkc,Ukc,Vkc,Wkc )Φp (x )Φl (y ) ⎭
= 0 (p = 1,2,..., and l = 1,2,...) (18)
where Ri (i = 1, 2 and 3) are the residuals of the governing differential
equations, when the admissible displacement fields are substituted into Fig. 7. The verification of the developed semi-analytical model and finite ele-
the governing differential equations. Substituting Eq. (16) into gov- ment analysis for the low-velocity impact of a steel panel [64].
erning equations Eq. (9)) and then applying the Galerkin formulation
(Eq. (18)) lead to a system of 3 + 2 m + n differential equations:
Table 4
αj, λi and ζi for S-S, C-C, F-F, C-S, C-F and F-S boundary conditions [55].
Boundary conditions αj or βj Characteristic equations and values of μλi (a × λk or b × λc) ζi (ζk or ζc)
(j = 1, 2, 3, and 4)
S-S α1 = 0, α2 = 0 sin(μλi) = 0 1 or 1
α3 = 0, α4 = −1 kπ or cπ
C-C α1 = 1, α2 = −1 cos(μλi)cosh(μλi) = 1 cosh(μλi) − cos(μλi)
α3 = 1, α4 = −1 μλi = 4.730, 7.853, … sinh(μλi) − sin(μλi)
F-F α1 = −1, α2 = −1 cos(μλi)cosh(μλi) = 1 cosh(μλi) − cos(μλi)

α3 = −1, α4 = −1 μλi = 4.730, 7.853, … sinh(μλi) − sin(μλi)
C-S α1 = 1, α2 = −1 tan(μλi) = tanh(μλi) cosh(μλi) − cos(μλi)

α3 = 1, α4 = −1 μλi = 3.927, 7.069, … sinh(μλi) − sin(μλi)
C-F α1 = 1, α2 = −1 cos(μλi)cosh(μλi) = −1 cosh(μλi) + cos(μλi)

α3 = 1, α4 = −1 μλi = 1.875, 4.694, … sinh(μλi) + sin(μλi)
F-S α1 = 1, α2 = 1 tan(μλi) = tanh(μλi) cosh(μλi) − cos(μλi)

α3 = 1, α4 = 1 μλi = 3.927, 7.069, … sinh(μλi) − sin(μλi)
892
Fig. 8. Distributions of non-dimensional transverse deflection (w ) along the x-direction of architected sandwich panels under the circular uniform distributed load:
(a) Out-of-plane cells and (b) In-plane cells.
0 0 0 iωkc t 0 0 iωkc t 0 0 iωkc t

0 0 i
{0}1 × m {0}1 × n ⎤ ⎧U¨ kc ⎫ Ukc (t ) = Ukc e , V kc (t ) = V kc e , Wkc (t ) = Wkc e
⎡ M11 M14
⎢0 ⎪ 0 ⎪ i
Ukc i iωkc t
(t ) = Ukc e i
, Vkc i iωkc t
(t ) = Vkc e , Wkcj (t ) = Wkcj eiωkc t
M22 0 {0}1 × m M25i
{0}1 × n ⎥ ⎪V¨kc ⎪ (20)
⎢ ⎥
⎪¨ ⎪ 0
⎢0 0 M33 {0}1 × m {0}1 × m M36j ⎥ ⎪W kc ⎪ Substitution of Eq. (20) into Eq. (19) yields an eigenvalue problem
⎢ i ⎥
⎢ M41
i
{0}1 × m {0}1 × m M44 {0}1 × m {0}1 × n ⎥ ⎨U¨ kci ⎬ in the absence of mechanical load and nonlinear in-plane force re-
⎪ i ⎪
⎢ {0} i
M52 {0}1 × m {0}1 × m M55i
{0}1 × n ⎥ sultant N .̂ The fundamental frequency of the architected sandwich
⎢ 1×m ⎥ ⎪V¨kc ⎪
⎢ {0}1 × n ⎪ j⎪ panels can be obtained by solving the following eigenvalue problem:
{0}1 × n M63j {0}1 × n {0}1 × n M66j ⎥ ⎦⎪ ¨ ⎪
⎣ ⎩Wkc ⎭
|[K ]−ω2 [M ]| = 0 (21)
⎡ K11 K12 K13 i
K14 i
K15 K16j 0
⎤ ⎧Ukc ⎫
⎢ ⎥ ⎪ 0 ⎪ ⎧0 ⎫
⎢ K21 K22 K23 i
K24 i
K25 K26j ⎥ ⎪V kc ⎪ ⎪ 0 ⎪ where ω is the fundamental frequency of the sandwich panel and |…|
⎢ K31 K32 K33 + N0̂ i
K34 i
K35 K36j ⎥ ⎪W 0 ⎪ ⎪ F13 ⎪
kc
represents determinant in Eq. (21). The smallest eigenvalue obtained
+⎢ i ⎥
i ⎬
= 0 from Eq. (21) is called the fundamental vibration frequency. Since prior
⎢ K 41 i
K 42 i
K 43 i
K 44 i
K 45 K 46j ⎥ ⎨Ukc ⎨ ⎬
⎢ i ⎪
⎥ i ⎪ ⎪0 ⎪ to impact, the architected sandwich panel is initially at rest, the initial
⎢ K51
i
K52 i
K53 i
K54 i
K55 K56j ⎥ ⎪Vkc ⎪ ⎪ F j ⎪ conditions are:
⎢ j ⎥ ⎪ j ⎪ ⎩ 16 ⎭
K K 62j K 63j K 64j K 65j K 66j + Nj ̂⎦ ⎩Wkc ⎭
⎣ 61 0
Ukc 0
(t = 0) = 0, V kc 0
(t = 0) = 0, Wkc (t = 0) = 0
(19a) i
Ukc i
(t = 0) = 0, Vkc (t = 0) = 0, Wkcj (t = 0) = 0 (22)
mi S¨ (t ) = −Pc (t ) (19b)
For example, CSCS refers to a sandwich panel with clamped edges at
F13 and F16j may be presented in terms of defined shape functions as: x = 0 and a; and simply-supported edges at y = 0 and b. It should be
∞ ∞ 1.5 mentioned that the admissible functions satisfy the essential boundary
⎡ 0 ⎤
F13 = K C [S (t )−w0 (xs ,ys ,t )]1.5 = K C ⎢S (t )− ∑ ∑ Wkc (t )Φk (xs )Φc (ys ) ⎥ conditions (i.e., generalized displacement) of the problem. Except for
⎣ k=1 c=1 ⎦ the simply-supported case, the admissible functions may not fully sa-
F16j = K C [S (t )−z jwj (xs ,ys ,t )]1.5 tisfy natural (i.e., generalized forces) boundary conditions.
∞ ∞ 1.5
⎡ ⎤
= K C ⎢S (t )−z j ∑ ∑ Wkcj (t )Φk (xs )Φc (ys ) ⎥
4. Homogenized properties of architected cellular core
⎣ k=1 c=1 ⎦
(19c) The elastic effective properties of the cellular core of architected
It is worth mentioning that Eq. (19) is a nonlinear equation which is sandwich panels are obtained here by implementing the standard me-
a function of w0 at the contact point. Iterative calculations are used by a chanics homogenization [61,63]. Opposed to analytical models which
MATLAB code to explore the results in the time domain. It should be only provide an estimated bound for elastic properties of advanced
mentioned that [M] and [K] are symmetric stiffness and mass matrices, materials, effective elastic stiffness of advanced materials can be ob-
respectively, and their components can be obtained by the integration tained with a high degree of accuracy by implementing the standard
of [M′] and [K′] matrices, using Eq. (18), as presented in Appendix A for mechanics homogenization. In this method, a representative unit cell is
arbitrary boundary conditions. In addition, N0̂ and Nj arê functions as- modeled in ANSYS mechanical APDL and is discretized with 8-node 3D
sociated with the axial mechanical buckling loads defined in Appendix Solid185 elements. Elastic periodic boundary conditions, presented in
A for alternative boundary conditions. For harmonic and free vibration Fig. 2b, are applied on the edges of the unit cell. Then, the unit cell is
analyses, solutions are sought in the following form [55]: subjected to six independent macroscopic unit strains as:
893
50
(Pa) 40
30
20
10
0
0 0.01 0.02 0.03 0.04 0.05 0.06 0.07 0.08 0.09 0.1
out-of-plane
= 70 = 90 = 120
Fig. 9. Comparison of von Mises stress distribution of 3D printed sandwich panels under the circular uniform distributed load obtained by semi-analytical solution
methodology and FEM implemented in ANSYS Workbench.
ε11 = [0 0 0 0 0 0]T , ε22 = [0 1 0 0 0 0]T , ε33 = [0 0 1 0 0 0]T , geometries considered in this paper is presented in the Appendix B. Fur-
ε12 = [0 0 0 1 0 0]T , ε13 = [0 0 0 0 1 0]T , ε23 = [0 0 0 0 0 1]T thermore, in order to corroborate the accuracy of the homogenized model
(23) compared to a detailed finite element modelling, we have compared the von
Mises stress predicted by two approaches in Fig. 9, which shows a max-
Solving the finite element model, the equivalent microscopic imum error of 5% for the von Mises stress predicted by standard mechanics
strains, εij , are obtained for each of the above macroscopic strains. homogenization compared with a detailed finite element modelling. It is
c
Knowing the values of εkl and εij local structural tensor, Mijkl can be worth mentioning that for elastic analysis of low-velocity impact responses,
obtained according to: we only need the effective stiffness and material density (mass). Linear
c homogenization can be used for obtaining the effective linear elastic
εij = Mijkl εkl (24)
properties. Based on homogenization, effective material density (ρm ) is
The effective stiffness matrix can be simply derived by taking the in- simply calculated by: ρm = ρ × ρs (ρm is density of cellular unit cell, ρ is
tegral of the microscopic stress over the unit cell and dividing by its volume: relative density of unit cell and ρs is the material density of constituent solid
1 materials). To model the low-velocity impact response in elasto-plastic re-
Cijkl =
VRVE
∫ Cijmn Mmnkl
c
dVRVE
(25) gime, the introduced linear homogenization is not appropriate and non-
linear homogenization, which incorporates large-deformation occurring in
where Cijmn (i, j, k, l, m and n = 1, 2 and 3) is stiffness matrix, VRVE re- elasto-plastic deformation, is required. To avoid the need for nonlinear
presents the volume of the unit cell or representative volume element homogenization, we limit our semi-analytical analysis to elastic domain,
(RVE). The range of different stiffness properties which can be obtained by and we use a detailed finite element modelling for elasto-plastic analysis.
894
Fig. 10. Design chart of architected sandwich panels for non-dimensional buckling loads (N ) vs. the cell wall angle (θ) for in-plane ((a) and (b)) and out-of-plane ((c)
and (d)) cellular cores.
895
5. Finite element modeling
The 3D explicit large-deformation FEM is conducted using the

commercial software ANSYS Workbench 18.2. Fig. 6 presents the
ANSYS model for conducting the low velocity impact analysis. The face-
sheets and the core are meshed with quadrilateral and triangular ele-
ments and a convergence study is conducted to avoid mesh size de-
pendency of FEM results. The semi-spherical impactor is defined as a
rigid body and the dynamic load is applied by assuming an initial ve-
locity for the rigid impactor. The top surface of the panel is fixed in x-
and y-directions while the bottom surface is kept totally fixed with
circular holes at the middle of both surfaces as shown in Fig. 6. The
stress-strain curve of PLA is assumed elasto-perfectly plastic model for
FEM analysis. Geometric nonlinearity is also considered for the analysis
to reflect the large deformation that occurred during the velocity im-
pact test of 3D printed sandwich panels.
6. Results and discussion
The effects of core topology of 3D printed architected sandwich

panels on the structural responses and energy absorption capability are
investigated and the results of semi-analytical model, FEM and ex-
perimental testing are presented. The material properties of PLA used
Fig. 11. Design chart for architected sandwich panels for non-dimensional for the semi-analytical and finite element analyses are given in Table 3.
maximum deflection (wmax ) versus non-dimensional fundamental frequency The numerical results focus on the effect of relative density and core
(ω ).
topology on non-dimensional deflection, stress distribution, critical
buckling load, fundamental frequency (see Eq. (26)) and energy ab-
sorption.
Table 5
Effects of the cellular core topology on the first three natural frequencies and the associated mode shapes of architected sandwich
panels.
Cell angle
auxetic core ș = 70° rectangular core ș = 90° hexagonal core ș = 120°
Methods
1st FEM 179 (Hz) 19712 (Hz) 2475.6 (Hz)
natural
frequency Homogenized 1808.3 (Hz) 2059.6 (Hz) 2548.3 (Hz)
2nd FEM 2459.8 (Hz) 2971.9 (Hz) 3565.4 (Hz)
natural
Homogenized 2573.8 (Hz) 3120.5 (Hz) 3742.9 (Hz)
frequency
3rd FEM 2976.4 (Hz) 3562.5 (Hz) 4685 (Hz)
natural
frequency Homogenized 3130 (Hz) 3741.5 (Hz) 4914.5 (Hz)
1st
Mode
2nd
shapes
3rd
896
in-plane core
out-of-plane core
Fig. 12. Experimental force-displacement curves of 3D printed architected sandwich panels of alternative cellular core configurations, cell topologies and cell relative
densities.
Es h3 a2 a2 ρs σ computationally less expensive than the theories used in the literature.

w = w0 , N = Ncr , ω=ω , σ=
a4q0 Es h3 h Es q0 (26) For semi-analytical solution, we perform a convergence study to obtain
the required number of terms m (n = m) used for displacement com-
where q0 is the applied transverse load and Es and ρs are Young’s ponents in Eq. (1). We find that increasing the number of terms of m
modulus and material density of constitutive solid materials (i.e.. PLA leads to a decrease in difference between the obtained results. The semi-
used herein for 3D printing of architected sandwich panels), respec- analytical results converge for m = n = 12. Therefore, we consider
tively. them for the number of terms in the semi-analytical results presented
afterwards.
6.1. Verification
To validate the accuracy of the developed methodology, numerical 6.2. Effects of core cell topology on structural response
results are compared with the results given in the Ref. [64] for the low-
velocity impact of a simply-supported square steel panel. For the vali- In this section, we study the effects of core topology and density of
dation purpose, the mechanical and geometrical properties of the iso- 3D printed sandwich panels on structural responses including de-
tropic panel are taken from the Ref. [64], while for the rest of numerical formation and stress distributions, natural frequency and buckling
studies, material properties of PLA given in Table 3 are used. Fig. 7 loads. Fig. 8 presents the distributions of non-dimensional in-plane
presents the deflection-time history of the midpoint of the panel and the deformations (w ) along x-direction (z = h/2 and y = b/2) of 3D printed
impactor displacement obtained from the proposed semi-analytical sandwich panels under the circular uniform distributed load of
method and FEM compared to those obtained in the Ref. [64]. As seen q = −10 Pa over the circular area which has the same diameter as the
in Fig. 7, the results of the present study are in a good agreement with impactor (i.e., 25.4 mm). The sandwich panels with the relative density
those predicted in the Ref. [64]. It should be noted that a finite element of ρ = 0.3 have eight different core topologies including cell wall angle
analysis using brick elements and employing a Hertzian-type contact θ = 70°, 90°, 120° and 165° in out-of-plane and in-plane directions.
law was conducted in the Ref. [64]. We should mention that the im- Simply-supported boundary conditions are considered in four edges of
pactor returns back to its initial position after the impact, which means the sandwich panels (SSSS). It should be noted that the theoretical
the displacement of impactor should become zero. However, the approach and homogenization technique (see Sections 3 and 4) are used
sandwich panel experiences vibrations after the impact and accordingly to obtain the results. It is seen from Fig. 8 that for the out-of-plane
displacement doesn’t become zero at the time that impactor comes to its architected sandwich panels, auxetic cores (θ = 70°) make the panels to
original location (zero displacement) as seen in Fig. 7. It is worth deflect more in comparison with the rectangular (θ = 90°) and hex-
mentioning that the developed theory here is simpler and agonal cellular cores (θ = 120°) for the out-of-plane configurations of
897
in-plane core in-plane core in-plane core
out-of-plane core out-of-plane core out-of-plane core
Returned energy
Absorbed energy
time to absorb
energy
Fig. 13. Energy-time history obtained from the experimental impact test on 3D printed architected sandwich panels of alternative cellular core configuration, cell
topology, and cell relative density.
cells. However, hexagonal cellular cores (θ = 120°) show the highest increasing the cell wall angle for sandwich panels with in-plane cell
deflection for the in-plane configuration of cells. configurations, as shown in Fig. 10b. As shown in Fig. 10c, the buckling
The comparison of von Mises stress distribution of 3D printed load in y-direction (Nyy ) decreases for the cell wall angles in the range of
sandwich panels under the circular uniform distributed load calculated 65° ⩽ θ ⩽ 125° and then increases for the higher cell wall angles for
by semi-analytical solution and FEM is presented in Fig. 9. The nu- architected sandwich panels with out-of-plane cell configurations. In
merical results are shown along the x-direction (z = h/3 and y = b/2) addition, as presented in Fig. 10d, the buckling load in x-direction (Nxx )
of the SSSS sandwich panel with the relative density of ρ = 0.1 and core increases and then decreases for the higher cell wall angles for archi-
topologies of out-of-plane auxetic, rectangular and hexagonal shapes. tected sandwich panels with out-of-plane cell configurations. There are
There is a good agreement between the semi-analytical results using the some trends by comparing Fig. 10 and Fig. B1. For example, a similar
homogenization technique and the predictions of FEM using ANSYS behavior is seen in Fig. 10b and the effective Young’s modulus in x-
Workbench. Fig. 9 shows that the magnitudes of von Mises stress of the direction (i.e., E11, see Fig. 2) in Fig. B1 or between Fig. 10a and c with
architected sandwich panels for ρ = 0.1 with the auxetic and hexagonal the effective Young’s modulus in z-direction (i.e., E33, see Fig. 2) in Fig.
cores are greater than those of rectangular cores. B1 and another similar trend is seen in Fig. 10d with the effective
Fig. 10 demonstrates the non-dimensional critical buckling loads Young’s modulus in y-direction (i.e., E22, see Fig. 2) in Fig. B1. It is
(N ) versus the cell wall angle (varying from θ = 65° to θ = 165°) of the worth mentioning that 3D printed sandwich panels, opposed to con-
core of architected sandwich panels for the cellular core’s relative ventional sandwich composites made by bonding a face-sheet to a core
density variation from 0.1 to 0.7. The buckling loads in x- and y-di- through adhesion, are less sensitive to the failure of interaction between
rections (Nxx and Nyy ) for the sandwich panels with in-plane cores are the face-sheets and cellular core during the buckling since there is no
shown in Fig. 10a and b; the buckling loads in the in x- and y-directions adhesion between the cellular core and face-sheets [1].
(Nxx and Nyy ) of the sandwich panel with out-of-plane core are shown in Fig. 11 shows the non-dimensional maximum deflection (wmax )
Fig. 10c and d. It is worth mentioning that some sandwich panels with versus the non-dimensional fundamental frequency (ω ) of architected
specific cell wall angles are not possible to be 3D printed for higher sandwich panels with a cellular core of in-plane configuration as the
relative densities due to the minimum thickness, which can be 3D print relative density increases from ρ = 0.1 to 0.7 and the cell wall angle
with MACHINA Corp. 3D printer, named as “inadmissible geometry”. increases from θ = 65° to 165°. The arrows show the trend of the curves
As shown in Fig. 10a, the buckling loads in y-direction (Nyy ) of in-plane (deflection versus natural frequency) by increasing the cell wall angle θ,
architected panels monotonically decrease for all cell wall angles in low while relative density of core remains constant; or by increasing the
relative densities (ρ = 0.1 and 0.2). These buckling loads decrease and relative density of core, while the cell wall angle θ is kept constant. For
then increase by increasing the cell wall angle for higher relative den- example, the blue curve varies from θ = 65° to 165° for the core’s re-
sities (ρ ≥ 0.3). The buckling load in the x-direction (Nxx ) increases by lative density of ρ = 0.1. As shown in Fig. 11, increasing the cell wall
898
In-plane cores
= FEM = 20 % FEM = 30 % FEM
Experiment Experiment Experiment
Energy Absorption %
Energy Absorption %
Energy Absorption %
Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ
Out-of-plane cores
= FEM = 20 % FEM = 30 % FEM
Experiment Experiment Experiment
Energy Absorption %
Energy Absorption %
Energy Absorption %
Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ
Fig. 14. Experimental and FEM results for the energy absorption capability of 3D printed architected sandwich panels of alternative cellular core configuration, cell
topology, and cell relative density.
angle causes a higher fundamental frequency and maximum deflection. contact force versus displacement diagram. As our experimental results
Table 5 presents the first three natural frequencies and their cor- show, for the out-of-plane configuration of cores, the architected
responding mode shapes for SSSS architected sandwich panels obtained sandwich panels with the auxetic cores have the lowest contact force,
by the semi-analytical solution using the homogenization technique and while the hexagonal cores reveal the highest contact forces (see Table
FEM. The core relative density is ρ = 0.1 and the cell topologies include C1). For core topologies of in-plane configuration, contact forces are
out-of-plane auxetic, rectangular and hexagonal cores. A maximum almost the same for alternative core topologies, a phenomenon which
discrepancy of 6% is found between the homogenized results and the can also be observed in FEM results. In general, it is seen that the
detailed finite element analysis. auxetic core in the out-of-plane direction has much lower force and
stress level at the same deformation level, compared with the other core
6.3. Effects of cellular core topology on energy absorption capability topologies. Experimental time-history of displacement and contact
force of 3D printed architected sandwich panels of alternative core
In this section, we examine the effects of the topology and relative configuration, cell topology and cell relative density are presented in
density of cellular core of architected sandwich panels on historical Figs. C1 and C2 in Appendix C.
contact forces, displacements of sandwich panels and energy absorption Fig. 13 presents the experimental, numerical (finite element) and
when the sandwich panel is subjected to a low-velocity impact. We semi-analytical results for the energy-time history for alternative core
implement the numerical (FEM) and experimental testing to conduct configuration, cell’s topology and cell’s relative density. There is a good
the study. An effective energy absorbing sandwich panel absorbs all of agreement between the results of experimental tests and numerical
the kinetic energy of impactor while keeping the force on the body analyses. In the energy absorption-time history, the amounts of ab-
below a certain critical level [54]. For experimental study, the sand- sorbed and returned (released) energies during the impact test can be
wich panel is subjected to a 3 J low-velocity impact test unless other- observed. The absorbed energy is the energy mostly dissipated by
wise mentioned. various failure mechanisms such as delamination and cracking [54].
The experimental contact force-displacement curves (at the tip of The returned (released) energy is the elastic energy. Herein, we define
absorbed energy
impactor) of 3D printed architected sandwich panels are presented in the energy performance based on the ratio of [54] (see
maximum impact energy
Fig. 12 for alternative core’s configuration (out-of-plane and in-plane), Fig. 13). The semi-analytical results developed in Section 3 are also
cell topologies and cell relative densities. It should be noted that the given in Fig. 13. The theoretical results are not fully matched with FEM
integrated area under the entire force-displacement curve represents and experimental data since the semi-analytical results have been de-
the energy absorption for a sandwich panel under an impact load. veloped based on elastic stress-strain relations and do not incorporate
Fig. 12 shows that the core topology has a significant effect on the
899
out-of-plane core in-plane core
T T
Fig. 15. Energy absorption capability of 3D printed architected sandwich panels versus non-dimensional maximum deflection, maximum normal stress and required
time to completely absorb energy for sandwich panels of alternative cellular core configuration, cell topology and cell relative density.
900
Fig. 16. FEM results for the energy absorption capability of 3D printed architected sandwich panels of in-plane cellular core configuration, alternative cell topology,
and cell relative density (from ρ = 0.1 to 0.7).
the plastic deformation. To verify the semi-analytical solutions, FEM worth mentioning that even though the auxetic cellular cores in out-of-
has been implemented again by applying a fully elastic constitutive law. plane configuration absorb slightly less energy, they have the largest
Fig. 14 shows the experimental and numerical energy absorption toughness (i.e., absorbing energy through plastic deformation without
capabilities of 18 sandwich panels with different core topologies including fracture) [45] (see Fig. 12). For other relative densities in the range of
the out-of-plane and in-plane auxetic, rectangular and hexagonal cores ρ = 0.2 to 0.7, the auxetic cores in out-of-plane configuration show larger
(cell wall angle θ = 70°, 90° and 120°, respectively); and the relative deflection, more time to absorb energy and smaller energy absorption
densities of ρ = 0.1, 0.2 and 0.3. There is a good agreement between the capability. For the in-plane configuration and the relative densities in the
results of experimental tests and numerical analyses. For all (Out-of-plane range of ρ = 0.2 to 0.7, the auxetic cores show the same time to absorb
or in-plane configurations) considered relative densities of cellular cores, energy as others and slightly smaller energy absorption capability. An-
the rectangular and hexagonal cores provide a slightly higher energy ab- other advantage of the auxetic core is showing a lower contact force
sorption capability than the auxetic core except for the auxetic core with compared to other cores, which can be crucial for minimizing the impact
the relative density of ρ = 0.1. It should be noted that the auxetic cellular forced applied to the protected structures that could be of interest in shock
core with the relative density of ρ = 0.1 in the in-plane configuration is the absorber applications (see Fig. C2). Therefore, the auxetic sandwich panels
preferred core, in terms of energy absorption, for the architected sandwich are better potential candidates for energy absorption applications. In the
panel subjected to this specific impact energy. 25% and 29% improve- meantime, rectangular and hexagonal cells can be ideal for the core of
ments are observed during the experimental tests in energy absorption sandwich panels where high stiffness/strength-to-weight-ratios are re-
performance by using auxetic cores for architected sandwich panels in quired such as automotive applications (semi-structural parts), building
comparison with hexagonal and rectangular ones, respectively. The ar- and construction.
chitecture of the preferred cellular core may differ for dissimilar impact Fig. 16 presents the energy absorption capability of architected
energies. The impact energy applied to the architected sandwich panels is sandwich panels predicted by FEM for in-plane core configurations for
primarily absorbed through plastic deformation and damage of core and alternative core topologies with the relative densities in the range of
face-sheets as shown in Fig. 17. ρ = 0.1 to 0.7. The impact energy varies for different densities as seen
Fig. 15 shows the energy absorption capability of the architected in Fig. 16. As mentioned earlier, the auxetic core can show its super-
sandwich panels versus non-dimensional maximum deflection, non-di- iority in the energy absorption over the other core’s cell topologies
mensional maximum normal stress (σzz ) and the time required to absorb when the impact energy is large. As seen in Fig. 14, except for ρ = 0.1,
energy for alternative cell topologies and core configuration. As shown in the auxetic sandwich panels absorb slightly less energy in comparison
Fig. 15, the energy absorption capability and maximum normal stress in- with the hexagonal sandwich panels. However, as seen in Fig. 16, all
crease significantly and the maximum deformation and required time to auxetic sandwich panels for different densities absorb much more en-
absorb energy decrease by increasing the relative density. This observation ergy in comparison with rectangular and hexagonal sandwich panels up
corresponds to the increased stiffness, which is because of increase in the to 35% and 33%, respectively. The reason is that the auxetic sandwich
relative density. The architected sandwich panel with the in-plane con- panels with higher densities need more impact energy to show their
figuration of auxetic cores and the relative density of ρ = 0.1, however, large deflections and consequently, absorb much more energy in com-
reveals a different behavior. This architected sandwich panel absorbs parison with other architected sandwich panels. In addition, we study
significantly more energy compared to those of the other topologies. It is the effects of different impact energy on energy absorption of an
901
H. Yazdani Sarvestani et al.
Table 6
Comparison of damage zone of impacted architected sandwich panels of a hexagonal cellular core with the relative density of ρ = 0.3 subjected to impact energies of 20 J and 30 J for both out-of-plane and in-plane
configurations of cellular cores obtained by experimental tests and numerical simulation.
Impact energy Experimental out-of-plane hexagonal cores Numerical out-of-plane hexagonal cores Experimental in-plane hexagonal cores Numerical in-plane hexagonal cores
20 J
mm m m mm
9.2 4.8 8 mm
9 .16 4.7
902
30 J
m m m
5m 45 mm mm
12. 12. 7.1 7
mm 7.0
Composite Structures 200 (2018) 886–909
(a) the impacted part
3D printing defects
(b) before impact test
Circumferential
Delamination
cracks
Radial cracks
(c) after impact test
Fig. 17. (a) The impacted part of the sandwich panel with vertical hexagonal core under 30J impact energy. (b) and (c) Micro-CT tomography images of the center
part of the sandwich panel with vertical hexagonal cores (before and after impact) and the relative density of ρ = 0.3 subjected to 30J impact energy (maximum
resolution is 12 μm).
903
architected sandwich panel presented in Appendix C, Fig. C3. process for design and manufacturing of structural elements. As a re-
Table 6 compares the damage zone of architected sandwich panels sult, the effect of process (3D printing) induced defects on the durability
with out-of-plane and in-plane configurations of the hexagonal cores with of 3D printed products needs a series of detailed experimental analyses
the relative density of ρ = 0.3 when subjected to the impact energies of hybrid by numerical predicative damage evolution models which
20 J and 30 J. It is found that the circular damage zone of the architected should be a subject of future research studies.
sandwich panel is wider for the out-of-plane configuration of cellular core
compared to the in-plane configuration; a phenomenon due to the lower
7. Concluding remarks
stiffness of sandwich panels with out-of-plane configuration compared to
the in-plane configuration. Circumferential cracks are seen in the top face-
We conduct a comprehensive theoretical, numerical and experi-
sheet of the sandwich panel with the out-of-plane core. We can conclude
mental studies to determine the optimized geometry of architected 3D
that the damage created by an impact load on the sandwich panel with the
printed sandwich panels with a periodic cellular core. We 3D print
in-plane core is more localized in comparison with that of the sandwich
architected sandwich panels of three different cell topologies made of
panel with the out-of-plane core. In addition, there is a good agreement for
biopolymers and conduct experimental low-velocity impact tests on
the observation of the damage zone of the sandwich panels due to the
samples to estimate their energy absorption capability. The energy
impact load obtained by FEM and experimental tests.
absorption capacity of the 3D printed architected sandwich panels are
determined by integrating the load-displacement curves found from the
6.4. Effects of low-velocity impact on the microstructural damage of
computational and experimental response of the low-velocity impact
architected sandwich panels
analysis. The results show that if relative density of the auxetic cellular
core is selected appropriately for a specific value of impact energy, the
A 3D printed architected sandwich panel of a hexagonal core of in-
sandwich panels with auxetic cores can have a higher level of energy
plane configuration and ρ = 0.3 has been experimentally tested under a
absorption capability up to 33% compared to the rectangular and
30 J impact energy. As shown in Fig. 17a, the center part of the 3D
hexagonal sandwich panels (Fig. 16). For example for a low-energy
printed panel is cut to study its microstructure after the impact test
impact condition (3J), it is desired to use a low-density auxetic sand-
compared to its microstructure before the impact. SkyScan 1172 micro-
wich panel (ρ = 0.1) so that the auxetic core can show its potential in
CT system is used to conduct X-ray micro-tomography imaging and to
absorbing energy compared to other cell topologies. For a high-energy
explore the microstructural damage caused by the impact test as well as
impact condition (66 J), it is desired to use a high-density auxetic
existing 3D printing defects. The maximum resolution used for the
sandwich panel (ρ = 0.7). In the application of a low-energy impact
imaging is 12 μm. Figs. 17b and c illustrate the micro tomography
protection, it is highly desirable that the impact protectors absorb as
images of the center part of the sandwich panel before and after impact
much energy as possible at a minimum response force transferred to
test. As seen in Fig. 17b, there are some 3D printing defects in a part of
structure. Architected sandwich panels with auxetic cellular cores ex-
core walls of architected panels. Morphological mismatches are seen in
hibit a desirable combination of high-energy absorption capability with
the X-ray imaging between as-designed (defect-free) and as-manu-
a low peak value of response force (Fig. C2). This study implement a
factured (imperfect) cores of the panels. These geometric imperfections
numerical methodology corroborated by the experimental testing for
are generated during FDM 3D printing and could have effects on the
developing a new generation of 3D printed sandwich structures, which
mechanical properties and energy absorption capability of the manu-
simultaneously satisfy structural and energy absorption requirements.
factured panels. Fig. 17c reveals delamination occurred within the
core’s cell walls between the 3D printed layers in addition to cracks
developed in circumferential and radial directions of panels. The Acknowledgments
maximum size of the surface cracks in the 3D printed sample is 52 μm.
These cracks are propagated in the planar and transverse directions. A.H. Akbarzadeh acknowledges the financial supports provided by
While the radial cracks shown in Fig. 17c may be caused by the cutting McGill University and the Natural Science and Engineering Research
process, the circumferential cracks and delamination are likely caused Council of Canada (NSERC) through NSERC Discovery Grant RGPIN-
by impact and 3D printing manufacturing process and not by the cut- 2016-04716 and NSERC Engage Grant EGP 507009-16 in collaboration
ting process. We should mention that the durability of 3D printed with MACHINA Corp., Edmonton, AB, Canada. Authors also acknowl-
products, especially 3D printed sandwich structures is an important edge discussion with Prof. R. Boukhili and Ms. R. Ouadday at École
issue which could affect the applicability of additive manufacturing Polytechnique de Montréal for the low-velocity impact test.
Appendix A
The components of [K] and [M] matrices in Eq. (19) are calculated by the numerical integration of components of [K′] and [M′] matrices using
Eq. (18):
K ′11 = A11 ϕ‴m ϕn + 2A16 ϕ″m ϕ′n + A16 ϕ′m ϕ″n
K ′12 = A16 ϕ″m ϕ′n + (A12 + A16 ) ϕ′m ϕ″n + A26 ϕm ϕ‴n K ′22 = A66 ϕ″m ϕ′n + 2A26 ϕ′m ϕ″n + A22 ϕm ϕ‴n
K ′13 = 0 K ′23 = 0
i i i i
K ′14 = B11 ϕ‴m ϕn + 2B16 ϕ″m ϕ′n + B66 ϕ′m ϕ″n K ′i24 = B16
i i
ϕ‴m ϕn + (B12 i
+ B66 i
) ϕ″m ϕ′n + B26 ϕ′m ϕ″n
i i i i i
K ′15 = B16 ϕ″m ϕ′n + (B12 + B66 ) ϕ′m ϕ″n + B26 ϕm ϕ‴n K ′i25 = B66
i i
ϕ″m ϕ′n + 2B26 i
ϕ′m ϕ″n + B22 ϕm ϕ‴n
j
K ′16 = B13j ϕ′m ϕn + B36j ϕm ϕ′n j
K ′26 = B36j ϕ′m ϕn + B23j ϕm ϕ′n
904
K ′33 = A55 ϕ″m ϕn + 2A 45 ϕ′m ϕ′n + A 44 ϕm ϕ″n + N0̂

i i i
K ′34 = B55 ϕ″m ϕn + B45 ϕ′m ϕ′n K ′i44 = D11
i i
ϕ‴m ϕn + 2D16 i
ϕ″m ϕ′n + D66 i
ϕ′m ϕ″n −B55 ϕ′m ϕn
i i i
K ′35 = B45 ϕ′m ϕ″n + B44 ϕm ϕ″n K ′i45 = D16
i i
ϕ″m ϕ′n + (D12 i
+ D66 i
) ϕ′m ϕ″n + D26 i
ϕm ϕ‴n −B 45 ϕm ϕ′n
j
K ′36 = B55 ϕ″m ϕn + 2B45j ϕ′m ϕ′n
j
+ B44j ϕm ϕ″n j
K ′46 ij
= (D13 −D55ji ) ϕ′m ϕn + (D36
ij
−D45ji ) ϕm ϕ′n
i i i i i
K ′55 = D66 ϕ″m ϕ′n + 2D26 ϕ′m ϕ″n + D22 ϕm ϕ‴n −B 44 ϕm ϕ′n
j
K ′56 = ij ji ij ji
(D36−D45) ϕ′m ϕn + (D23−D44 ) ϕm ϕ′n j
K ′66 = D55j ϕ″m ϕn + 2D45j ϕ′m ϕ′n + D44j ϕm ϕ″n −D33j ϕm ϕn + Nj ̂
i
M ′11 = I1 ϕ′m ϕn, M ′22 = I1 ϕm ϕ′n , M ′33 = I1 ϕm ϕn, M ′14 = I1i ϕ′m ϕn, M ′i41 = I1i ϕ′m ϕn, M ′i44 = I2i ϕ′m ϕn
j
M ′i25 = I1i ϕm ϕ′n , i
M ′55 = I2i ϕm ϕ′n , M ′36 = I1j ϕm ϕn, j
M ′63 = I1j ϕm ϕn, M ′66
j
= I2j ϕm ϕn, M ′52
i
= I1i ϕm ϕ′n
N0̂ = 0
N xx ϕ″m ϕn + 0
2N xy ϕ′m ϕ′n + 0
N yy ϕm ϕ″n Nj ̂ = N xx
0 j 0 j
z ϕ″m ϕn + 2N xy 0 j
z ϕ′m ϕ′n + N yy z ϕm ϕ″n (A1)
Appendix B
Fig. B1 presents the effective mechanical properties of different architected cellular materials used here as the core of the architected sandwich
panel. The cell wall angle of architected cellular core may vary between θ = 70° to θ = 160°, from auxetic to hexagonal cells. The variation of
normalized effective mechanical properties, i.e. Young’s modulus (E/Es ), Poisson’s ratio (ν / νs ) and shear modulus (G/Es ), is obtained by the standard
mechanics homogenization for a wide range of the cell wall angle (70° ⩽ θ ⩽ 160°) and the relative density.
Fig. B1. Normalized effective properties of the RVE representing the core of architected sandwich panels predicted by the standard mechanics homogenization for
cell wall angle ranging from 70° to 160°: Young’s modulus, Shear modulus and Poisson’s ratio.
905
Appendix C
Fig. C1 presents the experimental displacement-time history curves of 18 3D printed architected sandwich panels. We consider two core con-
figurations of in-plane and out-of-plane; for each core configuration three cell topologies of auxetic (θ = 70°), rectangular (θ = 90°) and hexagonal
(θ = 120°) with three relative densities of ρ = 0.1, 0.2 and 0.3 are considered. The total deformation (x-, y- and z- directions) of panels subjected to
an impact load predicted by FEM is also shown in Fig. C1. It is found that the peak value of displacement in z-direction remains almost the same for
the sandwich panels with in-plane cores regardless of the core’s cell topology for ρ = 0.2 and 0.3, while it changes for the different cores in out-of-
plane configuration of architected sandwich panels. For example, the peak displacement of the out-of-plane auxetic sandwich panels is respectively
45% and 72% higher than the sandwich panel with rectangular and hexagonal cell topologies for ρ = 0.1. The configurations at the time right before
the impact and the maximum peak load experienced by the panel with out-of-plane auxetic, rectangular and hexagonal cores are shown in Table C1.
Fig. C2 shows the experimental contact force-time history curves of for the aforementioned 18 3D printed architected sandwich panels. The
maximum shear stresses of the selected sandwich panels subjected to the impact load obtained by FEM are also presented. It is seen from Fig. C2 that
the peak impact forces are almost the same for the sandwich panels with in-plane cores no matter what the core topology is, while the peak forces are
different with the sandwich panels with the out-of-plane cores for different core topologies. For example, the peak force of the out-of-plane auxetic
sandwich panel is less than those of the out-of-plane rectangular and hexagonal sandwich panels. Compared to the other two core topologies, the
contact force of the architected sandwich panel with an out-of-plane core takes more time to be relaxed and reach zero in comparison with the other
two panels. As shown in Fig. C2, this time is almost the same for all panels with in-plane cores.
Fig. C3a and b show the experimental time history of energy absorption for an architected sandwich panel with the hexagonal cellular core and
the relative density of ρ = 0.3 with the in-plane configuration subjected to the following impact energies: 8 J, 12 J and 15 J; for the out-of-plane
configuration of hexagonal core, the sandwich panel is subjected to 20 J and 30 J impact energies. The overall trends of the curves are similar for all
impact energies; increasing the impact energy is accompanied by an increase in the energy absorption capability.
Fig. C1. Experimental displacement–time history of architected sandwich panels with alternative topologies and relative density of cellular core.
906
Table C1
Configuration of out-of-plane 3D printed architected sandwich panels with auxetic, rectangular and hexagonal cellular cores for ρ = 0.1 right before and after impact.
Core Before impact After impact
Auxetic
Rectangular
Hexagonal
907
Fig. C2. Experimental contact force–time history curves of 3D printed architected sandwich panels of alternative cellular core configurations, cell topologies and cell
relative densities.
in-plane
in-plane
in-plane
out-of-plane
out-of-plane
Fig. C3. Experimental energy absorption-time history of a 3D printed architected sandwich panel with the relative density of ρ = 0.3 subjected to different impact
loads: (a) in-plane hexagonal core with the impact energies of 8 J, 12 J and 15 J and (b) in-plane and out-of-plane hexagonal cores with the impact energies of 20 J
and 30 J.
908
Appendix D. Supplementary data
Supplementary data associated with this article can be found, in the online version, at http://dx.doi.org/10.1016/j.compstruct.2018.04.002.
References [34] Fu M, et al. Nonlinear shear modulus of re-entrant hexagonal honeycombs under
large deformation. Int J Solids Struct 2016;80:284–96.
[35] Hou Y, et al. The bending and failure of sandwich structures with auxetic gradient
[1] Schaedler TA, Carter WB. Architected cellular materials. Annu Rev Mater Res cellular cores. Compos A Appl Sci Manuf 2013;49:119–31.
2016;46:187–210. [36] Yang L, et al. A comparison of bending properties for cellular core sandwich panels.
[2] Gibson LJ, Ashby MF. Cellular solids: structure and properties. Cambridge Mater Sci Appl 2013;4(08):471.
University Press; 1999. [37] Hou Y, et al. Graded conventional-auxetic Kirigami sandwich structures: flatwise
[3] Allen HG. Analysis and design of structural sandwich panels: the commonwealth compression and edgewise loading. Compos B Eng 2014;59:33–42.
and international library: structures and solid body mechanics division. Elsevier; [38] Imbalzano G, et al. A numerical study of auxetic composite panels under blast
2013. loadings. Compos Struct 2016;135:339–52.
[4] Akbarzadeh A, et al. Dynamic eigenstrain behavior of magnetoelastic functionally [39] Imbalzano G, et al. Three-dimensional modelling of auxetic sandwich panels for
graded cellular cylinders. Compos Struct 2014;116:404–13. localised impact resistance. J Sandwich Struct Mater 2017;19(3):291–316.
[5] Bitzer T. Honeycomb technology: materials, design, manufacturing, applications [40] Scarpa F, Tomlinson G. Theoretical characteristics of the vibration of sandwich
and testing. Springer Science & Business Media; 2012. plates with in-plane negative Poisson's ratio values. J Sound Vib
[6] Dinwiddie RB, et al. Thermographic in-situ process monitoring of the electron-beam 2000;230(1):45–67.
melting technology used in additive manufacturing. Proc SPIE K [41] Lira C, Scarpa F. Transverse shear stiffness of thickness gradient honeycombs.
2013;87050:87050K–9K. Compos Sci Technol 2010;70(6):930–6.
[7] Petras A, Sutcliffe M. Failure mode maps for honeycomb sandwich panels. Compos [42] Stephani G, et al. Iron based cellular structures–status and prospects. Adv Eng Mater
Struct 1999;44(4):237–52. 2006;8(9):847–52.
[8] Rathbun H, et al. Performance of metallic honeycomb-core sandwich beams under [43] Li MZ, Stephani G, Kang KJ. New cellular metals with enhanced energy absorption:
shock loading. Int J Solids Struct 2006;43(6):1746–63. Wire-Woven Bulk Kagome (WBK)-Metal Hollow Sphere (MHS) Hybrids. Adv Eng
[9] Buitrago BL, et al. Modelling of composite sandwich structures with honeycomb Mater 2011;13(1–2):33–7.
core subjected to high-velocity impact. Compos Struct 2010;92(9):2090–6. [44] Goehler H, et al. Functionalized metallic hollow sphere structures. Adv Eng Mater
[10] Masters I, Evans K. Models for the elastic deformation of honeycombs. Compos 2014;16(3):335–9.
Struct 1996;35(4):403–22. [45] Berger J, Wadley H, McMeeking R. Mechanical metamaterials at the theoretical
[11] Meraghni F, Desrumaux F, Benzeggagh M. Mechanical behaviour of cellular core for limit of isotropic elastic stiffness. Nature 2017;543(7646):533–7.
structural sandwich panels. Compos A Appl Sci Manuf 1999;30(6):767–79. [46] Altan G, Kovan V. Flexural behavior of 3D printed honeycomb sandwich structures
[12] Becker W. Closed-form analysis of the thickness effect of regular honeycomb core with waste filler material. Mater Testing 2016;58(10):833–8.
material. Compos Struct 2000;48(1):67–70. [47] Li T, Wang L. Bending behavior of sandwich composite structures with tunable 3D-
[13] Sypeck DJ, Wadley HN. Cellular metal truss core sandwich structures. Adv Eng printed core materials. Compos Struct 2017;175:46–57.
Mater 2002;4(10):759–64. [48] Dikshit V et al. Out of plane compressive strength of 3D printed vertical pillared
[14] Sypeck DJ. Cellular truss core sandwich structures. Appl Compos Mater corrugated core structure. 2016.
2005;12(3):229–46. [49] Turner AJ. Low-velocity impact behavior of sandwich panels with 3D printed
[15] Hu Y, et al. Fabrication and mechanical behaviors of corrugated lattice truss polymer core structures. Wright State University; 2017.
composite sandwich panels. Compos Sci Technol 2016;125:114–22. [50] Association US. ASTM D3763–2006 Standard Test Method for High Speed Puncture
[16] Wadley HN, Fleck NA, Evans AG. Fabrication and structural performance of peri- Properties of Plastics using Load and Displacement Sensor. USA: USA Standards
odic cellular metal sandwich structures. Compos Sci Technol 2003;63(16):2331–43. Association International; 2006.
[17] Cote F, et al. Structural response of pyramidal core sandwich columns. Int J Solids [51] Gibson I, Rosen D, Stucker B. Additive manufacturing technologies: 3D printing,
Struct 2007;44(10):3533–56. rapid prototyping, and direct digital manufacturing. Springer; 2014.
[18] Xiong J, et al. Shear and bending performance of carbon fiber composite sandwich [52] Standard A., D638,“ Standard test method for tensile properties of plastics,” ASTM
panels with pyramidal truss cores. Acta Mater 2012;60(4):1455–66. International, West Conshohocken, PA, 2010, DOI: 10.1520/D0638-10, 2010.
[19] Vogiatzis P, et al. Topology optimization of multi-material negative Poisson’s ratio [53] Yang L, et al. Design of auxetic sandwich panels for structural applications. Solid
metamaterials using a reconciled level set method. Comput Aided Des freeform fabrication symposium 2013.
2017;83:15–32. [54] Ouadday R, et al. Experimental and numerical investigation on the impact behavior
[20] Lakes R, Elms K. Indentability of conventional and negative Poisson′s ratio foams. J of dual-core composite sandwich panels designed for hydraulic turbine applica-
Compos Mater 1993;27(12):1193–202. tions. Compos Struct 2017.
[21] Lakes R. Foam structures with a negative Poisson's ratio. Science [55] Reddy JN. Mechanics of laminated composite plates and shells: theory and analysis.
1987;235:1038–41. CRC Press; 2004.
[22] Yang W, et al. Review on auxetic materials. J Mater Sci 2004;39(10):3269–79. [56] Sarvestani HY, Akbarzadeh A, Hojjati M. Hygro-thermo-mechanical analysis of
[23] Grima JN, Evans KE. Auxetic behavior from rotating squares. J Mater Sci Lett fiber-steered composite conical panels. Compos Struct 2017;179:146–60.
2000;19(17):1563–5. [57] Sarvestani HY, Hojjati M. Three-dimensional stress analysis of orthotropic curved
[24] Choi J, Lakes R. Fracture toughness of re-entrant foam materials with a negative tubes-part 2: laminate solution. Eur J Mech-A/Solids 2016;60:339–58.
Poisson's ratio: experiment and analysis. Int J Fract 1996;80(1):73–83. [58] Sarvestani HY, Hojjati M. Failure analysis of thick composite curved tubes. Compos
[25] Evans KE. Auxetic polymers: a new range of materials. Endeavour Struct 2017;160:1027–41.
1991;15(4):170–4. [59] Yazdani Sarvestani H, Naghashpour A. Analysis of free edge stresses in composite
[26] Alderson A, Alderson K. Auxetic materials. Proc Inst Mech Eng G: J Aerospace Eng laminates using higher order theories. Indian J Mater Sci 2014;2014.
2007;221(4):565–75. [60] Wang Z-X, Xu J, Qiao P. Nonlinear low-velocity impact analysis of temperature-
[27] Mohsenizadeh S, et al. Crashworthiness assessment of auxetic foam-filled tube dependent nanotube-reinforced composite plates. Compos Struct 2014;108:423–34.
under quasi-static axial loading. Mater Des 2015;88:258–68. [61] Akbarzadeh A, et al. Electrically conducting sandwich cylinder with a planar lattice
[28] Hou S, et al. How does negative Poisson’s ratio of foam filler affect crashworthiness? core under prescribed eigenstrain and magnetic field. Compos Struct
Mater Des 2015;82:247–59. 2016;153:632–44.
[29] Evans KE, Alderson A. Auxetic materials: functional materials and structures from [62] Sarvestani HY, Akbarzadeh A, Mirabolghasemi A. Structural analysis of size-de-
lateral thinking!. Adv Mater 2000;12(9):617–28. pendent functionally graded doubly-curved panels with engineered micro-
[30] Lee J, Choi J, Choi K. Application of homogenization FEM analysis to regular and architectures. Acta Mech 2018:1–27.
re-entrant honeycomb structures. J Mater Sci 1996;31(15):4105–10. [63] Hassani B, Hinton E. A review of homogenization and topology opimization
[31] Assidi M, Ganghoffer J-F. Composites with auxetic inclusions showing both an II—analytical and numerical solution of homogenization equations. Comput Struct
auxetic behavior and enhancement of their mechanical properties. Compos Struct 1998;69(6):719–38.
2012;94(8):2373–82. [64] Shariyat M, Farzan F. Nonlinear eccentric low-velocity impact analysis of a highly
[32] Yang S, et al. A comparative study of ballistic resistance of sandwich panels with prestressed FGM rectangular plate, using a refined contact law. Arch Appl Mech
aluminum foam and auxetic honeycomb cores. Adv Mech Eng 2013;5:589216. 2013:1–19.
[33] Wan H, et al. A study of negative Poisson's ratios in auxetic honeycombs based on a
large deflection model. Eur J Mech-A/Solids 2004;23(1):95–106.
909

Composite Structures: Sciencedirect

Uploaded by

Copyright:

Available Formats

Composite Structures: Sciencedirect

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Composite Structures: Sciencedirect

Uploaded by

Copyright:

Available Formats

Composite Structures 200 (2018) 886–909

Contents lists available at ScienceDirect

3D printed architected polymeric sandwich panels: Energy absorption and T

1. Introduction demand for improvement of structural, aerodynamic and thermal in-

Table 1 performed on specimens fabricated in the orientation of 45° with re-

2.3. Low-velocity impact tests

Out-of-plane rectangular core Out-of-plane hexegonal core Out-of-plane auxetic core

mi S¨ (t ) + Pc (t ) = 0, S (0) = 0, S ̇ (0) = V0 (5)

3.3. Equations of motion

The equilibrium equations of a sandwich panel subjected to a low-

transverse Young’s modulus of the face-sheet of the architected sand-

moment of inertia Ikj are deﬁned as: (R1ixz ,R1iyz ) = (B55

F-F α1 = −1, α2 = −1 cos(μλi)cosh(μλi) = 1 cosh(μλi) − cos(μλi)

C-S α1 = 1, α2 = −1 tan(μλi) = tanh(μλi) cosh(μλi) − cos(μλi)

C-F α1 = 1, α2 = −1 cos(μλi)cosh(μλi) = −1 cosh(μλi) + cos(μλi)

F-S α1 = 1, α2 = 1 tan(μλi) = tanh(μλi) cosh(μλi) − cos(μλi)

0 0 0 iωkc t 0 0 iωkc t 0 0 iωkc t

5. Finite element modeling

The 3D explicit large-deformation FEM is conducted using the

6. Results and discussion

The eﬀects of core topology of 3D printed architected sandwich

Es h3 a2 a2 ρs σ computationally less expensive than the theories used in the literature.

in-plane core in-plane core in-plane core

out-of-plane core out-of-plane core out-of-plane core

Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ Ό = 70Ŵ Ό = 90Ŵ Ό = 120Ŵ

out-of-plane core in-plane core

out-of-plane core in-plane core

out-of-plane core in-plane core

(a) the impacted part

(c) after impact test

K ′33 = A55 ϕ″m ϕn + 2A 45 ϕ′m ϕ′n + A 44 ϕm ϕ″n + N0̂

in-plane core in-plane core in-plane core

out-of-plane core out-of-plane core out-of-plane core

in-plane core in-plane core in-plane core

out-of-plane core out-of-plane core out-of-plane core

Appendix D. Supplementary data

You might also like