AIAA 2013-4836

Guidance, Navigation, and Control and Co-located Conferences

August 19-22, 2013, Boston, MA
AIAA Modeling and Simulation Technologies (MST) Conference

An Investigation on the Tensile Behavior of Glass-reinforced

Aluminium Laminate

Chen Qi1, Guan Zhidong2, He Wei3 and Zhang Mi4

Beihang University, Beijing, 100191, China

An experimental and analytical study is carried out on GLARE laminates tensile

behavior, In the test results, the double-linear load-displacement curve and wave
deformation are found. By means of the user subroutine VUSDFLD of ABAQUS, one finite
element model is developed in the study to simulate the tensile behavior of GLARE (glass-
reinforced aluminum laminate). The models are based on progressive failure analysis
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

methodology, and take the random distribution of tensile strength in the reality into account.
The calculated results have good agreement with tests, which means the models could
predict the tensile behavior of GLARE exactly.

Nomenclature
GLARE = glass-reinforced aluminum laminate
ABAQUS = a software product of Dassault Systèmes Simulia Corp., Providence, RI, USA
VUSDFLD = user defined field
E = modulus
G12 = shear modulus
ρ = density
µ = the Poisson's ratio
Xt = the tension ultimate stress along the fiber direction
Xc = the compression ultimate stress along the fiber direction
Yt = the tension ultimate stress along the transverse direction
Yc = the compression ultimate stress along the transverse direction
S = in-plane shear failure stress
t = tension
c = compression
t = tension
c = compression

I. Introduction

G LARE (glass-reinforced aluminum laminate) is a new class of fiber metal laminates consisting of bonded thin
aluminum sheets and glass fiber/adhesive layers. It is manufactured by Stork Aerospace of the Netherlands, the
material is the culmination of 20 years of development work begun by Delft University of Technology and the
Netherlands National Aerospace Laboratory (NLR) 1.
This laminated structure provides the material with excellent fatigue, impact and damage tolerance characteristics.
GLARE also offers 15-30% weight savings over aluminum2. The first flying GLARE part is a bonded patch repair
installed on a C5-A Galaxy aircraft by the USAF in October 1995 at Kelly AFB. The first civil application is in a
bulk cargo floor of Boeing 777 and the bulkhead of the Bombardier Learjet 125 2,3. In 2003, GLARE is selected for
the upper fuselage skin structures of Airbus A380. Each A380 has about 380 square meter of GLARE. In addition to
being 10% less dense than aluminum--a weight saving of 794 kilogram--GLARE has proven superior in terms of
fatigue, damage, and fire resistance. GLARE is also used in the leading edge of tails of the A3801.

1
Ph.D Candidate, School of Aerospace Science and Engineering, No.37 Xueyuan Road. AIAA Student Member
2
Professor, School of Aerospace Science and Engineering, No.37 Xueyuan Road.
3
Ph.D Candidate, School of Aerospace Science and Engineering, No.37 Xueyuan Road.
4
Undergraduate student, School of Aerospace Science and Engineering, No.37 Xueyuan Road.
1
American Institute of Aeronautics and Astronautics

Copyright © 2013 by Qi Chen, Zhidong Guan, Wei He, Mi Zhang. Published by the American Institute of Aeronautics and Astronautics, Inc., with permission.
 In China, GLARE is a little different. we don't have the S-2 fiber for the time being, so we take the S-4 instead.
The differences between these two kind of fibers may result in some different characters of GLARE between
laminates made in China and abroad.

II. Experimental Scheme

A. Tested Materials
GLARE is a set, you can get different laminate through changing the fiber/resin system, the alloy type and
thickness, stacking sequence, fiber orientation, surface pretreatment technique, etc. Table 1 shows the laminates we
chose. There are 2 groups of specimen. Eight kinds of GLARE have been tested in all. The fiber used in group A is
unidirectional, and the fiber used in group B are biaxial. The glass fiber used as the reinforcement is S-4. Standard
test method for tensile properties of polymer matrix composite materials 5 is used. Laminates are 250×25mm in size
and 0.808-1.662mm in thickness. The Al 2024-T3 aluminum layer is 0.254mm in thickness, and it is 0.15mm in
thickness for R-4 glass unidirectional fiber prepreg. Specimens are summarized in table 1.
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

Table 1 Test matrix

Group laminate Lay-up number 0°fiber volume fraction, %

LS-01 Al/0/0/Al 6 37
LS-02 Al/90/90/Al 6 0
A
LS-03 Al/0/0/Al/0/0/Al 6 44
LS-04 Al/90/90/Al/90/90/Al 6 0
LS-05 Al/0/90/Al 6 19
LS-06 Al/0/90/Al/90/0/Al 6 22
B
LS-07 Al/0/90/0/Al 6 31
LS-08 Al/0/90/0/Al/0/90/0/Al 6 36

B. Testing Set-up
Before the tension testing, determine the specimen area, at three places in the gage section, and report the area as
the average of these three determinations. record the average area in units of mm2. Each specimen is evaluated with
back-to-back axial transducers, place the specimen in the grips of the testing machine Instron 8802. Set the speed of
testing to the head displacement rate of 2 mm/min, and apply the force to the specimen until failure, while recording
data and the mode and location of failure of the specimen. After the testing, calculate the ultimate tensile strength
and report the results to three significant figures.

III. Results and Discussion

A. Calculation results
Tensile strength, tensile modulus and Poisson rate are summarized in Table2-4.

Table 2 Tensile ultimate strength

Tensile
Standard Coefficient of
Group strength Average
deviation variation, %
, MPa
LS-01 957 12.94 1.35
LS-02 280 8.59 3.07
A
LS-03 1070 15.22 1.42
LS-04 267 5.16 1.94
LS-05 607 8.33 1.37
LS-06 659 6.14 0.93
B
LS-07 750 26.94 3.59
LS-08 883 5.67 0.64

2
American Institute of Aeronautics and Astronautics
 Table 3 Tensile elastic modulus

Tensile
Standard Coefficient of
Group modulus Average
deviation variation, %
, GPa
LS-01 59.1 0.24 0.41
LS-02 39.9 1.43 3.58
A
LS-03 50.0 0.69 1.39
LS-04 37.1 1.99 5.37
LS-05 46.7 4.00 8.55
LS-06 48.2 0.55 1.13
B
LS-07 52.8 2.07 3.92
LS-08 45.9 1.33 2.90
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

Table 4 Poisson rate

Poisson Standard Coefficient of

Group Average
rate deviation variation, %
LS-01 0.293 0.00550 1.87
LS-02 0.176 0.00159 0.90
A
LS-03 0.294 0.00294 1.00
LS-04 0.163 0.00115 0.70
LS-05 0.199 0.00985 4.94
LS-06 0.197 0.00265 1.35
B
LS-07 0.208 0.00949 4.55
LS-08 0.194 0.00280 1.45

B. Calculation results Discussion

Test results show that great differences lie in the tensile behavior. The 0 degree fiber volume fraction has an
effect on the failure mode of the laminate. For LS-02 and LS-04, they only have 90 degree fiber, which means they
have no 0 degree fiber, the failure modes they have are same, which fiber has delamination and neat fracture. For
LS-05, which has 19% of 0 degree fiber, delamination and the fracture of aluminium ply on one side are found. For
LS-06, which has 22% of 0 degree, fiber the fracture of fiber and aluminium ply in the middle is found. For LS-07,
the fracture of fiber and outside aluminium plies is found, which has 31% of 0 degree fiber. For LS-08, the fracture
of fiber is found, which has 36% of 0 degree fiber. For LS-03 and LS-01, which have 37% and 44% of 0 degree
fiber,the fracture of fiber and delamination are found, and the seamark is seen on aluminium surfaces. Picture 1
shows the seamark we found. The more the 0°fiber volume fraction is, the more the strength is.

Figure 1 Failure mode

C. Load-displacement Curves
After the tests, two different kinds of groups of load-displacement curves are received as shown in figure 2 and
figure 3.

3
American Institute of Aeronautics and Astronautics
 10000

8000

6000
Load, N

4000

LS-04
LS-02
2000
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

0
0 2 4 6 8 10 12 14 16 18
Displacement, mm

Figure 2 Load-displacement curves of LS-02 and LS-04

40000

35000 LS-01
LS-03
30000 LS-05
LS-06
25000 LS-07
LS-08
Load, N

20000

15000

10000

5000

0
0 1 2 3 4 5 6 7 8 9
Displacement, mm

Figure 3 Load-displacement curves of LS-01, LS-03, LS-05, LS-06, LS-07 and LS-08

D. Load-displacement Curves Discussion

There are two kinds of load-displacement curves, metal type and double-linear type. Figure 2 gives the metal
type load-displacement curves. As the figure shows, both laminates' curves are controlled by the aluminum ply,
because the curves are similar with aluminum's curve. although we find LS-04 ultimate load is higher than LS-02,
according to table 2, we can find both the ultimate strength and the elastic modulus of LS-02 are higher than LS-04.
From figure 3 we can see six kinds of GLAREs double-linear curves. These six laminates each has 90 degree
glass fiber plies, besides 90 degree, LS-05~LA-08 also have 0 degree. During the first linear stage, the laminates are
controlled by metal laminate, because the stress of the knee point of each laminates is arouund 230MPa. Afier the
knee point, the laminates are controlled by fiber layers, so during the second linear stage, the slopes of curves are
totally different from the first stage. They are all lower than the slopes of first stage, because the modulus of fiber is
lower than the one of aluminum.

4
American Institute of Aeronautics and Astronautics
 IV. Finite Element Model

A. Problem Description and material characteristics

Figure 4 The mesh of finite element mode

Within ABAQUS, a finite element model has been established in the study. In the model, the geometry, material
and boundary conditions are the same with those described in the test. Figure 4 shows the geometry of the laminate.
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

The laminate is subjected to uniaxial tension in the longitudinal direction. The laminate is made of aluminum and 0
degree/ 90 degree glass fiber-reinforced epoxy.
The material behavior of aluminum is assumed to be isotropic elastic-plastic with isotropic hardening. The
isotropic hardening data are listed in Table 5, and the mechanical properties are summarized in table 6.

Table 5 Isotropic hardening data for aluminum

Plastic strain/% 0.000 0.016 0.047 0.119 0.449 1.036

Yield stress/MPa 300 320 340 355 375 390
Plastic strain/% 2.130 3.439 5.133 8.000 14.710
Yield stress/MPa 410 430 450 470 484

Table 6 Mechanical properties of Aluminum 2024-T3

E/GPa G12/GPa ρ/g·mm-3 µ /MPa /MPa /MPa /MPa S/MPa

68 27 2.7 0.33 441 269 434 310 269
The material behavior of the glass fiber-reinforced epoxy layers is assumed to be orthotropic, with stiffer
response along the fiber direction and soften behavior in the matrix. The elastic properties-longitudinal modulus, E1t;
transverse modulus, E2t; shear modulus G12 and G23; and Poisson's ratios µ12 and µ23- are listed in table 7. The
subscript "1" refers to the longitudinal direction(or fiber direction), and the subscripts "2" and "3" refer to the two
transverse directions orthogonal to the fiber direction. The damage initiation and evolution behavior is also assumed
to be orthotropic. Table 7 also lists the ultimate values of the longitudinal failure stresses, Xt and Xc; transverse
failure stresses, Yt and Yc ; and in-plane shear failure stress, S. The subscripts"t" and "c" refer to tension and
compression, respectively.

Table 7 Mechanical properties of glass fiber-reinforced epoxy layers

E1t/MPa E2t/MPa E3t/MPa µ12 µ13 µ23 G12/MPa G13/MPa G23/MPa

51900 7180 7180 0.29 0.29 0.4 2975 2975 3000
E1c/MPa E2c/MPa E3c/MPa Xt /MPa Xc /MPa Yt /MPa Yc /MPa S/MPa
49667 11232 11232 1480 696 47.4 137 120
The plate is loaded with displacement boundary conditions applied at the right edge. To simplify the
postprocessing, the displacement loading is applied at a reference point and an equation constraint is used to
constrain the displacement along the loading direction between the right edge and the reference point. The loading
direction (along the global X-direction) aligns with the fiber direction of the 0 degree fiber-reinforced epoxy layer.

B. Material model implementation

5
American Institute of Aeronautics and Astronautics
 In the VUSDFLD model, the damage initiation criteria are expressed in terms of stress. The VUSDFLD model
uses two damage variables to describe damage in the fiber and matrix. Damage in the fiber is initiated when the
following criterion is reached:
  2   2   2
 11    12    13   1 ( 11  0) (1)
 X T   S12   S13 
FT 1  
2
2
  11 
  1 ( 11  0)
 X C 
Damage in the matrix is initiated when the following criterion is reached:
 2 2
  22   33  ( 232   22 33 )   12    13 
2

        1 ( 22   33  0) (2)

2
 YT  ( S23 )  S12   S13 
FT22  
  22   33   YC   (   )2  2   
2 2 2
   
 Y     1 
22 33
 23 22 33
  12    13   1 ( 22   33  0)
  2S23  
2 2
 C 4( S23 ) ( S 23 )  S12   S13 

where  11 is the longitudinal direction stress,  22 and  33 are the transverse directions stresses.  12 ,  13 and  23
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

are the shear stresses.

C. Modeling considerations
An Abaqus/Explicit analysis with mass scaling is adopted and three dimensional 8-node reduced-integration
elements (C3D8R) are used for modeling the laminates. For GLARE, the damage in the glass fiber layers is
simulated using the model implemented in user subroutine VUSDFLD. In the VUSDFLD model, the damage
initiation criteria are expressed by Hashin6. In the test, ultimate strength is not well-distributed, but in the ABAQUS
model, it is well-distributed. The ultimate strength is set to obey normal distribution, whose average value is fiber-
reinforced epoxy’s tensile ultimate strength, and the variance is 1, to model the reality.

D. Results and discussion

Tensile moduli of eight kinds of GLARE are calculated by the analytical model. Table 8 shows the comparison
of calculated tensile moduli and test results, it can be seen that in table 8 the calculated moduli and test results agree
well. Compared with test results, the relative errors range from 2.7%~9.3%.

Table 8 Tensile modulus compare to test results

Test result, Calculation Error ,

specimen
MPa results, MPa %
LS-01 59.1 55.7 5.8
LS-02 39.9 43.6 9.3
LS-03 50.0 54.0 8.0
LS-04 37.1 39.6 6.7
LS-05 46.7 49.8 6.6
LS-06 48.2 46.9 2.7
LS-07 52.8 48.3 8.5
LS-08 45.9 45.7 0.4
Figure 5 presents the comparison of Load-displacement curves between calculated and test results. From the
figure, we can find that the typical double-linear curve is obtained through ABAQUS simulation. Simple metal type
is also received. Both analytical results and test results indicate that there are two phases of GLARE load-
displacement. During the first phase, all GLARE laminates' curves are linear. After the knee point, the situations are
different. For laminates who only have 90 degree fiber, they behave like metal, because the load-displacement
curves are controlled by aluminum. For others, they still have linear curves, because the controller is glass fiber.

6
American Institute of Aeronautics and Astronautics
 18000

16000

14000

12000 LS-02 TEST

LS-02 ABAQUS
Load, N 10000 LS-01 TEST
LS-01 ABAQUS
8000

6000

4000
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

2000

0
0 1 2 3 4 5 6 7 8 9 10
Displacement, mm

Figure 5 The comparison of load-displacement curves

the comparison of fiber failure between the test result and calculated result can be seen from figure 6, the major
failure mode of GLARE is fiber broken. And figure 7 shows the matrix delamination failure of the test result and
calculated result. both results show the same failure position of lamintes and the same area of matrix failure. These
agreements lend support to the validities of the analytical model.

Figure 6 Failure mode of fiber

Figure 7 Failure mode of matrix

V. Conclusion
In the present study, both experimental tests and analytical model are carried out to investigate the tesile
behavior of 8 kinds of GLARE laminates. The following conclusion can be drawn from the study: Under tensile
loads, the major failure mode of GLARE is fiber broken and matrix delamination. The analytical model is applicable
on calculating the fiber/metal stress/strain distributions of GLARE, the applicability is verified by finite element
model. For glass fiber epoxy, Hashin failure criterion is suitable to predict the ultimate load of the fiber, and the
calculated results agree well with test results. Both anaalytical model and test results indicate that the load-
displacement curves have double-linear character and the wave deformation, but for those laminates who only have
90 degree glass fiber, they have metal type load-diplacement curves.

References
1 "Reducing A380 weight, Glare is key," Flight International Supplement, 20-26 May, 2003
2 Vlot A., Vogelesang L.B. and de Vries T. f., “Towards application of fibre metal laminates in large aircraft,” Aircraft
Engineering and Aerospace Technology, Vol. 71, No. 6, 1999, pp. 558, 570.

7
American Institute of Aeronautics and Astronautics
 3 Vlot A., "Impact loading on fiber metal laminates," International Journal of Impact Engineering, Vol. 18, No. 3, 1996, pp.

291, 307.
4 Wu G., Yang JM., "The mechanical behavior of GLARE laminates for aircraft structures," JOM, Vol. 57, No. 1, 2005, pp.

72, 79.
5 ASTM Committee, Designation: D 3039/D3039M-08, “Standard test method for tensile properties of polymer matrix

composite materials,” United States.

6 Hashin Z, “Failure Criteria for Unidirectional Fiber Composites,” Journal of Applied Mechanics, Vol. 47, No. 2, 1980, pp.

329, 334.
Downloaded by MONASH UNIVERSITY on October 9, 2015 | http://arc.aiaa.org | DOI: 10.2514/6.2013-4836

8
American Institute of Aeronautics and Astronautics