You are on page 1of 7

Composite Structures 81 (2007) 533–539

www.elsevier.com/locate/compstruct

Sandwich panels with Kagome lattice cores reinforced by carbon fibers


H.L. Fan, F.H. Meng, W. Yang *

Department of Engineering Mechanics, Tsinghua University, Beijing 100084, China

Available online 27 November 2006

Abstract

Stretching dominated Kagome lattices reinforced by carbon fibers were designed and manufactured. The sandwich panels were assem-
bled with bonded laminate skins. The mechanical behaviors of the sandwich panels were tested by out-of-plane compression, in-plane
compression and three-point bending. Different failure modes of the sandwich structures were revealed. The experimental results showed
that the carbon fiber reinforced lattice grids are much stiffer and stronger than foams and honeycombs. It was found that buckling and
debonding dominate the mechanical behavior of the sandwich structures, and that more complaint skin sheets might further improve the
overall mechanical performance of the sandwich panels.
Ó 2006 Elsevier Ltd. All rights reserved.

Keywords: Lattice structures; Sandwich panel; Carbon fiber; Experiments

1. Introduction ments have shown that the actual stiffness and strength
of the 3D lattice composites suffer by the inevitable deflec-
Sandwich panels assembled by skins and cores have tion of the carbon fiber reinforced rods near the intersect-
potential in naval and aeronautical applications since they ing nodes [16]. To overcome this deficiency, one may
offer high stiffness/strength-to-weight ratio [1]. The cores of explore 2D lattice grids reinforced by carbon fibers [16].
sandwich panels are conventionally made of foams or hon- Han and Tsai [17] introduced interlocked grid structures
eycombs. Recent researches have shown the higher weight with pultruded glass fiber ribs. Using unidirectional com-
efficiency of lattice structures than that of foams [2–10]. Of posite ribs, one could align all fibers along the rib direc-
critical importance in such designs is the principle of tions so that the stiffness and the strength of the material
stretching dominance, namely each element in the lattice can be utilized efficiently. Grids could have ribs running
is predominately subjected to tensile or compressive load- in several directions for multidirectional loadings. Accord-
ing. When that principle is observed, the mechanical behav- ing to the rib directions, various grids such as bi-grids, tri-
ior of the core scales linearly with the relative density q* of grids and quadri-grids are shown from Fig. 1a–c. Among
lattices [2]. On the other hand, the stiffness of open-cell them, tri-grids and quadri-grids have balanced longitudinal
foams scales as q*2 and the strength scales as q*1.5 and shear properties, whereas bi-grids offer little shear and
[1,2,11–14]. Several methods were developed to manufac- twisting stiffness. Bi-grids are simpler to manufacture
ture metallic lattice materials, including octet-truss materi- because they have a lower number of rib-crossings at the
als [3,4], pyramidal lattice materials and Kagome lattice intersections. On the other hand, tri-grids and quadri-grids
materials [5]. The interlacing method was attempted are difficult to be interlocked. To get a shear-resistant lat-
recently to manufacture an integrated three dimensional tice grid, rows of ribs of at least three different orientations
carbon structure of fiber reinforced lattice [15,16]. Experi- are required, while there could be only two rib-crossings at
the intersections. Consequently, the Kagome-grid shown in
*
Corresponding author.
Fig. 1(d) presents itself as the optimal choice.
E-mail addresses: yw-dem@tsinghua.edu.cn, yangw@zju.euc.cn (W. The interlocked cores reinforced by carbon fibers of the
Yang). Kagome-grid were manufactured and tested by Fan et al.

0263-8223/$ - see front matter Ó 2006 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2006.09.011
534 H.L. Fan et al. / Composite Structures 81 (2007) 533–539

Fig. 1. (a) Bi-grid. (b) Tri-grid. (c) Quadri-grid. (d) Kagome-Grid. Typical rib arrangements of grid structures.

[16]. The reinforced laminate ribs of [0/ ± 45/90]2 construc- in Fig. 3, where the curves earmarked by A and B refer to two
tion are made of T300/QY8911 carbon fibers. They have tests performed. According to the displacement curves, the
the thickness of 1 mm and the width of 18 mm. The length peak loadings are 469 kN and 405 kN, respectively. The
of struts in a representative unit cell is 20 mm. With such Young’s moduli are about 1.1 GPa to 1.5 GPa and
dimensions, the relative density of the grid core is just the strengths are about 22.7 MPa to 26.1 MPa in the tests.
0.086. According to the tests, the in-plane stiffness of the The failure modes of the grid under out-of-plane com-
grid cores is about 600 MPa, and the in-plane effective pression are depicted in Fig. 4. Delaminating, buckling
strength is 8.05 MPa. Comparisons given by Fan et al. and shearing were found in tests. In the top graph of
[16] showed that the carbon fiber reinforced grids are much Fig. 4, two ribs were intersected at the slots and were later-
stiffer and stronger than carbon foams and aluminum ally supported by each other. Delaminating and buckling
lattices. were confined and the ribs were sheared and snipped into
Due to relatively small density, the carbon fiber rein- two parts. In the bottom graph of Fig. 4, the failure modes
forced grids are ideal core materials of sandwich panels. for the thin free ribs were controlled by delaminating and
The sandwich panels were assembled with bonded laminate buckling, where ribs were seldom intersected.
skins, as shown in Fig. 2. The left portion of the top skin Theoretically, the stiffness of the grids are given by [16],
sheet is removed to reveal the interior structure of the 3q Es
panel. Ein ¼ ; Eout ¼ q Es ; ð1Þ
8
This work presents a preliminary experimental study on
the mechanical behavior of the carbon fiber reinforced where Ein and Eout are the in-plane and out-of-plane
sandwich grid panels. Young’s moduli of the lattice grid. Symbol Es denotes the
stiffness of the ribs, and q* is the relative density of the grid.
2. Out-of-plane compression of sandwich panels If the slots in the ribs were absent, the in-plane stiffness
would be about 37.5% of the out-of-plane stiffness accord-
2.1. Out-of-plane compression tests ing to Eq. (1). In actual tests for the grids with slotted ribs,
the ratio of Ein/Eout is about 0.4–0.5. The theoretical ratio
Out-of-plane compression tests were performed to study and the actual ratio are consistent in the sense that the slot-
the compressive stiffness and strength in the out-of-plane ted ribs are more likely to be twisted to buckle under out-
direction. The length and the width of the panel are of-plane compressions, leading to a slightly higher tested
130 mm and 140 mm, respectively. The thickness of the ratio. The strength ratio of rsin/rsout for the actual Kagome
skins of [±45/02/90/0]2 construction is 1.5 mm, and the total grids is about 0.30–0.35, with rsin and rsout being the in-
thickness of the panel is 21 mm. The tested curves are plotted plane and the out-of-plane compression strengths of the
lattice grid, respectively.

500
Compression load (kN)

400
A

300 B

200

100

0
0.0 0.2 0.4 0.6 0.8 1.0 1.2
Displacement (mm)

Fig. 2. Sandwich panel with interlocking Kagome lattice grid. Fig. 3. Out-of-plane compression curves of lattice grids.
H.L. Fan et al. / Composite Structures 81 (2007) 533–539 535

Fig. 4. Out-of-plane compression failure of lattice panels.

Table 1
Constants of cellular materials
Properties Carbon foam [18] Aluminium lattice [3] CFRC grid in-plane Out-of-plane
A B A B C
3
Density (g cm ) 0.71 0.45 0.367 0.138
Relative density 0.507 0.321 0.136 0.086
Modulus(GPa) 2.73 1.11 1.11 0.60 1.10 1.50
Specific stiffness (106 m2 s2) 3.84 2.48 3.02 4.348 7.97 10.87
Strength (MPa) 39.70 10.30 7.10 8.05 26.10 22.70
Specific strength (103 m2 s2) 55.92 22.89 19.34 58.33 191.91 166.91

2.2. Comparisons 3. In-plane compression of sandwich panels

As inferred in Eq. (1), the specific stiffness and the spe- The in-plane mechanical properties of the Kagome lat-
cific strength of lattice materials are independent of the rel- tice grid cores were given by Fan et al. [16]. The present sec-
ative density. On the other hand, those of foams increase tion is focused on the in-plane tests for the sandwich panels
monotonically with the relative density. Various specific with Kagome lattice cores. The height and the width of the
stiffnesses E/q and the specific strengths r/q are compared panel are 130 mm and 140 mm, respectively. Under in-
in Table 1 and Fig. 5. The out-of-plane compression stiff- plane compression, the strength and stiffness of the lattice
ness and strength of the Kagome grids reinforced by core can be neglected when compared with the solid skins.
carbon fibers are superior when comparing with the values In Fig. 6, the peak loadings of curve A and curve B are
of other materials. In Fig. 5, all data are normalized by the 114 kN and 111.7 kN, respectively. The elastic properties
in-plane specific stiffness or specific strength of grids. It of two curves are similar and the in-plane effective stiffness
showed that the grids are lighter, but stiffer and stronger, of the panel is 2.9 GPa in tests. The effective in-plane
than the aluminum lattice material and carbon foams. strengths of the panel for curves A and B are 38.8 MPa
and 38.0 MPa, respectively.

(σy/ρ) /(σy/ρ) gridA


grid B 120
(E/ρ)/(E/ρ)gridA
(E/ρ)/(E/ρ)gridA ,(σy/ρ) /(σy/ρ) gridA

A: Face debonding
3
grid C
100
grid C
Compression load (kN)

80
2
grid B

60

grid A
1
40
carbon foam B
aluminium lattice carbon foam A
20
0 B: Face sheet local delamination
0.1 0.2 0.3 0.4 0.5
ρ* 0
0 1 2 3 4 5 6 7 8
Displacement (mm)
Fig. 5. Specific stiffness and specific strength of different cellular materials.
Grid tests A, B and C are defined in Table 1. Fig. 6. In-plane compression curves of sandwich panels.
536 H.L. Fan et al. / Composite Structures 81 (2007) 533–539

The minimum critical buckling stress rxc of the skins


that are perfectly adhered to the sandwich core is derived
by Hetenyi [19] as
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
2Ef Ec t
rxc ¼ ; ð2Þ
3ð1  m2 Þ c
where Ef and Ec are the moduli of the skins and the cores.
The testing value of Ef is 20.3 GPa and that of Ec is
1.1 GPa. Symbols t and c denote the thickness of the skins
and the height of the cores, respectively. The Poisson’s
ratio of the skin is denoted by m.
According to Eq. (2), the critical buckling stress of the
panel would be 159 MPa. The compression strength of
the panel from the actual tests is only about 38 MPa. It
means that if the bonding is perfect, elastic buckling would
Fig. 7. Combined buckling and debonding mode of failure for the panel never happen and the compression strength of the panel
under in-plane compression.
should be controlled by the local delamination of the panel
skins. The occurrence of the buckling mode at much lower
compression strength suggests the weakness of the
Two different failure modes were detected by the tests as bonding.
shown in Figs. 7 and 8. Shown in Fig. 7 is the combined
buckling and debonding mode occurred in the in-plane 4. Bending behavior of sandwich panels
tests. As the in-plane compression proceeds, the skins
remain intact but tend to buckle and to bend outward. 4.1. Bending experiments
The bonding layers are not strong enough to transmit the
constraints of the cores. Debonding occurs abruptly and Three-point bending experiments were carried out for
the skins burst apart with a sharp sound. The load–dis- the sandwich panels. The span of the simple supported
placement curve drops straightly after debonding. Another sandwich panel is 360 mm and is denoted by symbol L.
failure mode was observed in the tests as shown in Fig. 8. The width of the panel is 130 mm and is denoted by symbol
Instead of elastic buckling, local delamination of the lami- b. The thickness of the face sheets is 1.5 mm and is denoted
nate skins appeared near an end of the panel, transforming by symbol t. The height of the panel is 21 mm. The ends of
the panel end to a broom-like damage state. The loading the panel were enwrapped by glass fiber clothes to prevent
decreased abruptly, but the residual strength made the face sheet debonding. The enwrapped length is 40 mm.
panel to maintain a loading capacity of about 20% of the Dimensions of the sandwich panel in three-point bending
peak value. tests are shown in Fig. 9.
The deflection curves of the tests shown in Fig. 10 dis-
play four representative regimes. They are elastic bending,
as shown in Fig. 11 (a); debonding, as shown in Fig. 11 (b);
core shearing, as shown in Fig. 11 (c); and subdued loading
capacity.
Three testing curves in Fig. 10 are labeled as A, B and C,
respectively. They share similar elastic properties, with the
bending rigidities varying from 1.84 KN m2 to 2.0 kN m2.
The face sheet is a cross-ply laminate of [±45/02/90/0]2
construction, with the theoretical stiffness of 79.3 GPa.

Fig. 8. Local delamination mode of failure for the panel under in-plane
compression. Fig. 9. Dimensions of the sandwich panel in three-point bending test.
H.L. Fan et al. / Composite Structures 81 (2007) 533–539 537

12 Loading capacity The weakness of the sandwich panel lies in the adhesion
C strength between a face sheet and a core. The upper face
10
B
sheet was compressed during bending. With the develop-
A Debonding ment of the deflection, the face sheet tended to buckle
Loading (kN)

8 and debonding occurred due to the insufficient adhesion


Elasticity to the core, as shown in Fig. 11 (b). The scatter in the adhe-
6
sion strengths leads to large differences for the peaks of the
loading curves. The peak loadings are 8.88 kN, 9.9 kN and
4
Core shear 10.59 kN for curves A, B and C, respectively.
2
The debonding of the upper face skin led to a sudden
Subdued loading capacity drop of the load, while the lower face sheet and the core
0 were still adhered together. The loading has to be sup-
0 2 4 6 8 10 12 ported by the weakened panel including the lower face
Deflection (mm)
sheet and the core. Then the reduction of the loading
Fig. 10. Bending curves of sandwich panels. was slacked after about 50% load drop from the peak,
as shown in Fig. 10. As the deflection grew larger, the
The stiffness of the grid in the core is only 0.6 GPa and grid core bore larger shearing force. Finally, it was
hardly contributes to the bending rigidity of the sandwich snipped into two parts, as shown in Fig. 11(c). As the
panel. The theoretical rigidity (EI)eq of the sandwich panel consequence, the adhesion between the lower face sheet
is [20] and the core was also gone, leading to another abrupt
drop in the loading capacity of the panel. The face sheets
Ef btc2 remained intact and were braced by the snipped cores.
ðEIÞeq ¼ ; ð3Þ
2 Severely damaged panel still offered subdued loading
where b denotes the width of the face sheet. According to capacity, as shown in Fig. 10. For curve C, the residual
Eq. (3), the predicted rigidity of the panel is 2.5 kN m2, loading capacity of the panel is about 1.3 kN, or 12.3
about 25% larger than the tested values. percent of the peak loading.

Fig. 11. Bending behaviors of the sandwich panel: (a) elastic deformation; (b) buckling and debonding; (c) core shear.
538 H.L. Fan et al. / Composite Structures 81 (2007) 533–539

4.2. Analysis load Pcs leading to debonding (given in the last row of Ta-
ble 2) is smaller than the peak loads in tests. Therefore, the
The stresses in skins and cores of the sandwich panel shear stress sc sustained by the core is dictated by
were given by Allen [20] as Q
PL sc ¼ < sa ; ð11Þ
rf ¼ ; ð4Þ bc
4btc where Q is the shear force of the panel and sa is the shear
and strength of the adhesion layer. The latter is far smaller than
PL Ec the shear strength of the grid. According to Eq. (11), the
rc ¼ ; ð5Þ maximum shear stresses of the curves vary from
4btc Ef
2.26 MPa to 1.90 MPa. The theoretical shear strength of
where rf denotes the stress of the skin, and rc denotes the the adherent is 37.1 MPa. With a density of 0.086, the aver-
normal stress of the core. Symbol P denotes the concen- age adhesion area is 8.6% of the area of the face sheet. If
trate load. one considers the adhesion area where two ribs intersect,
The shear stress varies through the face and the core in a that area is down to 5% of the overall area of the face sheet.
parabolic fashion. If the faces are much stiffer and thinner Accordingly, the effective shear strength is reduced to
than the core, however, the shear stress can be approxi- 1.86 MPa. Thus, the shear stress of the adhesive layer
mated by a linear variation through the face and a constant exceeds the effective shear strength and debonding occurs
value through the core: at the cross section where two ribs intersect.
P
sc ¼ ; ð6Þ
2bc 5. Conclusions
where sc denotes the shear stress of the core.
Conventionally, the failure modes of sandwich panels Carbon fiber reinforced Kagome lattice grids were made
are sorted into skin failure, core failure and debonding by the interlocked method. The grids were sandwiched by
[20–23]. The skin failure modes include face yielding and two carbon fiber reinforced laminates to assemble sandwich
face wrinkling. The face yielding is controlled by [20,21] panels. In-plane compression, out-of-pane compression and
three-point bending tests were carried out to study the
P fy ¼ 4ryf btðc þ tÞ=L; ð7Þ
mechanical behaviors of the grids and sandwich panels.
where Pfy represents the critical loading, and ryf denotes According to the experiments, the mechanical behavior
the yielding strength of the skin. The critical load Pfw in of the sandwich panels with carbon fiber reinforced grids
face wrinkling is [20,21] can be concluded as follows.
1=3
P fw ¼ 2:28Ef E2=3
c btc=L: ð8Þ (1) High specific strength and specific stiffness can be
The core shear and local indentation are the main modes of achieved in comparison with other cellular materials,
core failure. The peak load Pcs in core shear is [20,21] rendering the carbon fiber reinforced grids weight-
efficient.
P cs ¼ 2scs bc; ð9Þ
(2) The failure modes of the structures are revealed by
where scs is the shear strength of the core. The critical load experiments. The detected modes show that the fail-
Pi in local indentation is [20,21] ure process in the carbon fiber reinforced grids and
pffiffiffi pffiffiffiffiffiffiffiffiffiffiffiffi the assembled sandwich panels could be ductile to
P i ¼ 2bt ryf ryc þ abryc ; ð10Þ
some extent. If that is achieved, the loads can be effec-
where ryc denotes the yielding strength of the core and a tively transferred to other parts of the structure if one
denotes the contacting width of the load with the skin. part is disabled.
These criteria are listed in Table 2, along with the corre- (3) Debonding is the main weakness of the carbon fiber
sponding predictions for failure loads. The loads leading reinforced sandwich panels. The strength of the carbon
to skin or core failures are all larger than the recorded fiber reinforced face sheets and grid cores are higher
peaks in tests, which are listed in the right column of Table than the strength in the adhesion area. Therefore, the
2. When using the strength of adhesion layer, however, the adhesion always fails first when the face sheets are
subjected to either compression or bending. The adop-
Table 2 tion of more complaint skin sheets might improve the
Comparison of failure modes of sandwich panels overall mechanical performance of the sandwich panels.
Failure mode Criterion Pcr(kN) Ptest-max(kN)
Face yielding Pfy = 4rfybt(c + t)/L 12.1
Face wrinkling
1=3
P fw ¼ 2:28Ef Ec2=3 btc=L 70.0 8.88 (A); Acknowledgement
Core shear Pcs =p2sffiffifficsbc 21.8 9.90 (B);
pffiffiffiffiffiffiffiffiffiffiffiffiffi
Local indentation P i ¼ 2bt ryf ryc þ abryc 15.2 10.59 (C). We are grateful to the support of the Key Project (No.

Debonding P cs ¼ 2sa bc 8.7
106015) from the Chinese Ministry of Education.
H.L. Fan et al. / Composite Structures 81 (2007) 533–539 539

References [12] Christensen RM. Mechanics of low density materials. J Mech Phys
Solids 1986;34:563–78.
[1] Gibson LJ, Ashby MF. Cellular solids: structure and proper- [13] Warren WE, Kraynik AM. The linear elastic properties of open-cell
ties. Cambridge: Cambridge University Press; 1997. foams. J Appl Mech 1988;55:341–6.
[2] Deshpande VS, Ashby MF, Fleck NA. Foam topology bending [14] Warren WE, Kraynik AM. Linear elastic behavior of a low density
versus stretching dominated architectures. Acta Mater Kelvin foam with open cells. J Appl Mech 1997;64:787–94.
2001;49:1035–40. [15] Fan HL, Yang W, Wang B, et al. Design and manufacturing of a
[3] Deshpande VS, Fleck NA, Ashby MF. Effective properties of the composite lattice structure reinforced by continuous carbon fibers.
octet-truss lattice material. J Mech Phys Solids 2001;49:1747–69. Tsinghua Sci Tech 2006;11(5):515–22.
[4] Wallach JC, Gibson LJ. Mechanical behavior of a three-dimensional [16] Fan HL, Meng FH, Yang W. Mechanical behaviors and bending
truss material. Int J Solids Struct 2001;38:7181–96. effects of carbon fiber reinforced lattice materials. Arch Appl Mech
[5] Evans AG, Hutchinson JW, Fleck NA, et al. The topology design of 2006;75(10–12):635–47.
multifunctional cellular metals. Prog Mater Sci 2001;46:309–27. [17] Han DY, Tsai SW. Interlocked composite grids design and manu-
[6] Vasiliev VV, Barynin VA, Rasin AF. Anisogrid lattice structures – facturing. J Compos Mater 2003;37:287–316.
survey of development and application. Compos Struct [18] Sihn S, Rice BP. Sandwich construction with carbon foam core
2001;54:361–70. materials. J Compos Mater 2003;37:1319–36.
[7] Huybrechts S, Tsai SW. Analysis and behaviour of grid structures. [19] Hetenyi M. Beams on elastic foundation; theory with applications in
Compos Sci Technol 1996;56:1001–15. the fields of civil and mechanical engineering. Ann Arbor: The
[8] Hohe J, Beschorner C, Becker W. Effective elastic properties of University of Michigan press; 1946.
hexagonal and quadrilateral grid structures. Compos Struct [20] Allen HG. Analysis and design of structural sandwich pan-
1999;46:73–89. els. Oxford: Pergamon; 1969.
[9] Hohe J, Becker W. Effective elastic properties of triangular grid [21] Petras A, Sutcliffe MPF. Failure mode maps for honeycomb panels.
structures. Compos Struct 1999;45:131–45. Compos Struct 1999;44(4):237–52.
[10] Kim TD. Fabrication and testing of thin composite isogrid stiffened [22] Muc A, Zuchara P. Buckling and failure analysis of FRP faced
panel. Compos Struct 2000;49:21–5. sandwich plates. Compos Struct 2000;48(1–3):145–50.
[11] Ashby MF, Evans AG, Fleck NA, et al. Metal foams: a design [23] McCormack TM, Miller R, Kesler O, et al. Failure of sandwich
guide. Boston: Butterworth–Heinemann; 2000. beams with metallic foam cores. Int J Solids Struct 2001;38:4901–20.

You might also like