Professional Documents
Culture Documents
Composite Structures: Guang-Min Luo, Ya-Jung Lee
Composite Structures: Guang-Min Luo, Ya-Jung Lee
Composite Structures
journal homepage: www.elsevier.com/locate/compstruct
a r t i c l e i n f o a b s t r a c t
Article history: This investigation presents a quasi-static simulation method that adopts energy conservation and the
Available online 13 April 2008 results from the static crush to simulate the impact behavior of carbon-epoxy composite laminate plates.
These plates consist of adhered visco-elastic material, and formed constrained layered damped (CLD)
Keywords: components. The damping force is considered in the quasi-static simulation. Comparison with experi-
Visco-elastic mental results indicates that the proposed method yields accurate results and produces significant saving
Quasi-static in computing resources.
Constrained layered damping (CLD)
Ó 2008 Elsevier Ltd. All rights reserved.
0263-8223/$ - see front matter Ó 2008 Elsevier Ltd. All rights reserved.
doi:10.1016/j.compstruct.2008.04.009
G.-M. Luo, Y.-J. Lee / Composite Structures 88 (2009) 290–295 291
2
v-s curve without damping
1.5
0.5
velocity (m/s)
-0.5
-1
-1.5
-2
0 0.001 0.002 0.003 0.004 0.005 0.006 0.007
Fig. 3. Integration path and unloading path.
displacement (m)
Fig. 1. Velocity–displacement curve subjected to low-velocity impact without 2 v-s curve without damping
damping. v-s curve with damping : Iteration-1
1.6
v-s curve with damping : Iteration-2
1.2 v-s curve with damping : Iteration-3
relationship between impact velocity v and displacement s (v–s) as
defined by Huang and Lee [4]. If the v–s trend of CLD laminate plate 0.8
is similar to that without damping in Fig. 1, then the relationship velocity (m/s)
0.4
between damping force Fd = cv and displacement s is as given in
Fig. 2(b). 0
The energy conservation during the impact is considered.
-0.4
Assuming the dynamic energy lost from the impactor all trans-
ferred the work of mass-spring-damping system. The equation -0.8
can be expressed as: -1.2
Z s Z s
1
m½v20 v2 ðtÞ ¼ Fds þ F d ds ð1Þ -1.6
2 0 0 0 0.002 0.004 0.006 0.008 0.01 0.012
displacement (m)
Fig. 2 shows F–s and Fd–s curves, after obtaining the force–displace-
ment (F–s ) curve of static crush. The impact energy of the first dis- Fig. 4. The v–s trends after iteration.
placement increment s1 can be expressed as Eq. (2).
1 1 1 W 2 ¼ W 1 þ ½ðF 2 þ F 1 Þ ðs2 s1 Þ=2 þ ½cðv2 ðtÞ þ v1 Þ ðs2 s1 Þ=2
W 1 ¼ W k1 þ W d1 ¼ F 1 s1 þ cðv0 þ v1 ðtÞÞs1 ¼ m½v20 v21 ðtÞ ð2Þ
2 2 2 1
¼ m½v20 v22 ðtÞ ð4Þ
The velocity of the first step, v1(t), can be computed by solving Eq. 2
(2).
By solving Eq. (4), the velocity of the second increment, v2(t), can be
qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
expressed as Eq. (5).
cs1 ðcs1 Þ2 4mF 1 s1 þ 4m2 v20 4mcv0 s1
v1 ðtÞ ¼ ð3Þ c
2m v2 ðtÞ ¼ ðs2 s1 Þ
2m
The damping force Fd1 = cv1(t) and energy W1 of the first increment sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
h c i2 2 s2 s1 1
can all be obtained once v1(t) is known. ðs2 s1 Þ W1 þ ðF 1 þ F 2 þ cv1 Þ mv20
The energy corresponding to the next displacement increment 2m m 2 2
s2 can thus be written as Eq. (4). ð5Þ
From Eq. (5), if v2(t) is known, then the damping force Fd2 = cv2(t) When the t–v–s–F arrays of CLD laminate plate are obtained, the
and impact energy W2 corresponding to s2 are obtained. The veloc- obtained v–s arrays gain a modified Fd–s curve to substitute for
ity array vi(t) and the corresponding energy Wi can be calculated by Fig. 2(b). Iterating quasi-static method with modified Fd–s curve,
repeating the above steps. The general formula of velocity is given and new t–v–s–F arrays can be obtained. Fig. 4 shows the results
by Eq. (6), where i P 2. of the iterations and proves that the v–s trend of CLD laminate
sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi
ffi
c h c i2 2 si si1 1 2
vi ðtÞ ¼ ðsi si1 Þ ðsi si1 Þ W i1 þ ðF i1 þ F i þ cvi1 Þ mv0 ð6Þ
2m 2m m 2 2
The dissipation of dynamic energy cause the impactor velocity to plate is similar to that without damping. The first iteration result
slow down gradually until it reaches zero, i.e. vi(t) = 0. This case named the iteration-1, the second iteration result named the iter-
indicates that the impact is finished, and the CLD laminate plate ation-2, and the third iteration result named the iteration-3. The
has the maximum displacement. The dynamic energy of impactor given v–s curve of iteration-1 can be utilized in the second itera-
is absorbed by the CLD laminate plate. The energy equation of the tion, and the new v–s curve obtained form iteration-2 utilized in
impact-ending is presented as Eq. (7) the third iteration. The iteration results in Fig. 4 show that twice
Z s Z s iterations have obtained the convergence result. Hence, the t–v–
1
mv20 ¼ Fds þ F d ds ð7Þ s–F arrays calculated in this research always used the iteration-2
2 0 0
v–s curve.
The rebound route of the impactor was based on Miki [10]. If dam-
age to the CLD laminate plate occurs during impact, then the resid- 3. Dynamic simulation results by quasi-static method
ual stiffness declines during the rebound. Miki recommended
returning the unloading path to its origin. Fig. 3 shows the unload- A dynamic experiment was performed using the sizes of speci-
ing path. mens and impactor, in this paper, boundary condition, material
With the velocity, displacement and impact force arrays for the and laminate form of specimens as in [11]. The carbon fiber was
impact process, including rebound process, are given. The time Toray T700 24K, and the matrix was Fiber Cote E765expoxy. The
increment array Dt can then be obtained. length and width of the specimens were 0.12 and 0.05 m, respec-
Dt ¼ Ds=v ¼ ðsi si1 Þ=v ð8Þ tively. The thickness was 0.00214 m, and the constrained layered
damping (CLD) was 3M damping foil SJ-2552. Two stacking se-
The t–v–s–F arrays of CLD laminate plate subjected to low-velocity quences [45/45]4s and [90/0]4s, respectively, named angle-ply
impact are then obtained, where t denotes the time, v denotes the (PA) and cross ply (PC) were adopted. Table 1 shows the specimens
impactor velocity, s denotes the displacement of impactor, and F de- and impactor data.
notes the impact force. The static crush F–s data were then adopted to perform the
quasi-static simulation. Fig. 5 shows the force–displacement (F–s)
Table 1 curves of static crush from the reference [11].
Specimen and impactor data The dynamic drop tests were performed by GRC-8250 Dynatup
Name Specification Dimension from the General Research Corporation. The impactor masses were
2.95 and 5.73 kg, and the two initial impact velocities were1.41
Laminate specimen PC [90/0]4s Span = 0.12 m
PA [45/45]4s Width = 0.05 m and 3.05 m/s.
Thickness = 0.00214 m The impact force data were measured by a hemi-spherical steel
Constrained layered 3M damping foil SJ- Backing thickness = 0.25 mm impactor, namely the 208A-33 force sensor, from PCB Piezotronics.
damping 2552 Total thickness = 0.38 mm
The impact force data were then recorded by the dynamic ampli-
Impactor PCB 208A-33 Hemisphere/
fier PCB 494A. Fig. 6 illustrates the drop and measurement
diameter = 0.0127 m
machine.
a b
1200 2000
Experiment curve of angle-ply plate Experiment curve with CLD
ABAQUS simulation with UMAT
1800
1000 1600 ABAQUS simulation with UMAT
1400
800
force (N)
force (N)
1200
600 1000
800
400 600
400
200
200
0
0 0
0 5 10 15 20 25 30 35 0 5 10 15 20 25 30 35
Fig. 6. Drop test and measurement apparatuses of impact experiment: (a) GRC-8250 dynatup and (b) PCB 494A amplifier and 208A-33 force sensor.
Fig. 7 [11] shows the damaged PA and PC specimens of static cases with velocity of 1.41 m/s and impactor mass of 5.73 kg. Figs.
crush test, and Fig. 8 shows the damaged PA and PC specimens 9, 10 also show the time and velocity arrays calculated by the
of drop test results. The results of static crush and drop tests indi- quasi-static method. The maximum impact force at v = 1.41 m/s
cate that the failure forms of the drop test were similar to those of occurred at an impact velocity of zero. At this time, the failure
the static crush test under low-velocity impact. According to Fig. 8 of CLD plate was not very serious, and the impactor would
the failures in the PA specimen comprise several local failures (in rebound.
±45° direction) in the width direction, and the failures of CLD ad- The simulation results of the quasi-static and ABAQUS-UMAT
hered to the PA specimens were spread widely and irregularly. algorithms were very similar to the experiment results. Compari-
However, the failures of PC specimen were based on fiber-snap son with ABAQUS-UMAT further confirms the accuracy of the
in the width direction, and the failure of CLD adhered to PC speci- quasi-static method.
mens were similar to those of the PC specimens themselves. The greater impact force let to serious failure, and the evalua-
Figs. 9–12 show the results of quasi-static simulation and tion of ABAQUS-UMAT did not converge, at an impact velocity of
experiments were shown in. Besides quasi-static simulation, the 3.05 m/s. Therefore, only the quasi-static simulation results and
FEM package ABAQUS-UMAT was also adopted to simulate the experiment results were compared for v = 3.05 m/s case. Figs. 11,
behavior of low-velocity impact. The stiffness modifications for 12 show the comparison results.
the ABAQUS-UMAT simulation were based on the suggestions of The quasi-static simulation results were also closed to the
Luo and Lee [11]. experiment results for v = 3.05 m/s. Because of the greater impact
The CLD laminated plates at an impact velocity of 1.41 m/s force led the serious fracture after impact, the rebound stage of
presented no obvious failures, and the dynamic evaluation easily the PA and PC specimens were uncertainly. For PA specimens
converged in ABAQUS-UMAT. However, at an impact velocity of ([45/45]4s), the strength in the length direction was weaker than
3.05 m/s, the failure of CLD laminate plates was very obvious, and PC specimens, and the CLD laminate plate were damaged during
the evaluation of ABAQUS-UMAT did not converge. The ABAQUS the impact, so the impactor did not rebound in this case. For PC
evaluation in this study only considered the case of v = 1.41 m/s. specimens ([90/0]4s), the stronger strength in the length direction
Fig. 9 presents the results of the cases with velocity of 1.41 m/s led a little rebound stage of the impactor after impact. Figs. 11,
and impactor mass of 2.95 kg. Fig. 10 presents the results of the 12 show the results with v = 3.05 m/s.
Fig. 7. Front view and rear view of damaged PA and PC specimens [11].
294 G.-M. Luo, Y.-J. Lee / Composite Structures 88 (2009) 290–295
a b
1000 Pah1w1-dyna experiment 2 1500 Pch1w1-dynamic experiment
2
impact velocity
impact force (N)
700
velocity (m/s)
velocity( m/s)
500 0 0
400 -0.5 600 -0.5
300
-1 -1
200 300
100 -1.5 -1.5
0 -2 0 -2
0 0.004 0.008 0.012 0.016 0.02 0 0.003 0.006 0.009 0.012 0.015
time (s) time (s)
Fig. 9. Quasi-static simulation, v = 1.41 m/s, m = 2.95 kg: (a) PA specimen and (b) PC specimen.
a Pah1w2-dyna experiment
b 1800 2
1000 2 Pch1w2-dyna experiment
ABAQUS simulation
ABAQUS simulation
900 Quasi-static simulation
1.5 1600
impact velocity
Quasi-static simulation 1.5
800 impact velocity
1400
1
impact force (N)
impactforce(N)
1
700
velocity (m/s)
velocity (m/s)
1200
600 0.5
1000 0.5
500 0
800 0
400 -0.5 600
300 -0.5
-1 400
200
-1
-1.5 200
100
0 -2 0 -1.5
0 0.005 0.01 0.015 0.02 0.025 0.03 0.035 0 0.005 0.01 0.015 0.02
time (s) time (s)
Fig. 10. Quasi-static simulation, v = 1.42 m/s, m = 5.73 kg: (a) PA specimen and (b) PC specimen.
G.-M. Luo, Y.-J. Lee / Composite Structures 88 (2009) 290–295 295
a b
1200 Pah2w1-dyna experiment 3.5 2000 Pch2w1-dyna experiment 4
Quasi-static simulation
Quasi-static simulation
1000 impact velocity 3 3
impact velocity
impact force (N) 1600
velocity(m/s)
velocity (m/s)
800
2 1200 1
600
1.5
800 0
400
1
-1
200 0.5 400
-2
0 0
0 0.003 0.006 0.009 0.012 0.015 0 -3
0 0.003 0.006 0.009 0.012 0.015 0.018
time (s)
time (s)
Fig. 11. Quasi-static simulation, v = 3.05 m/s, m = 2.95 kg: (a) PA specimen and (b) PC specimen.
a b
Pah2w2-dyna experiment
1200 Quasi-static simulation 3.5 2000 Pch2w2-dyna experiment 3.5
impact velocity 1800 Quasi-static simulation 3
1000 impact velocity
1600 2.5
3
1400
800 2
velocity (m/s)
force (KN)
force (KN)
velocity (m/s)
1200
1.5
600 2.5 1000
1
800
400 0.5
600
2 0
400
200
200 -0.5
0 -1
0 1.5 0 0.004 0.008 0.012 0.016 0.02
0 0.002 0.004 0.006 0.008 0.01 0.012
time (ms) time(ms)
Fig. 12. Quasi-static simulation, v = 3.05 m/s, m = 5.73 kg: (a) PA specimen and (b) PC specimen.
4. Conclusions [3] Belingardi G, Vadori R. Low velocity impact tests of laminate glass-fiber-epoxy
matrix composite material plates. Int J Impact Eng 2002;27:213–29.
[4] Huang, Lee. A quasi-static method for the simulation of composite laminated
The failures of CLD laminated plate subjected to low-velocity are plates subjected to low-velocity impact. J Soc Naval Architects Marine Eng
very complex. The use of a finite element package and experiments 2002;21:83–91.
[5] Huang, Lee. Quasi-static simulation of composite-laminated shells subjected to
to realize the failures of CLD laminate plate is always time-consum-
low-velocity impact. J Reinf Plates Compos 2005;24:763–74.
ing. This study adopts the quasi-static method in order to simplify [6] Michelle S, Hoo Fatt, Park Kyong S. Dynamic models for low-velocity impact
the description of failure behavior through the static crush results. damage of composite sandwich panels. Compos Struct 2001;52:335–51.
[7] Gustin J, Mahinfalah M, Jazar GN, Aagaah MR. Low-velocity impact of sandwich
The quasi-static simulation rule established in this paper con-
composite plates. Exp Mech 2004;44:1741–2765.
siders the damping effect. The impact force and the velocity varia- [8] Jiang, Shu. Local displacement of core in two-layer sandwich
tion of impactor can be obtained quickly using the quasi-static composite structures subjected to low velocity impact. Compos Struct 2004;
method through the relations of force and displacement from the 71:53–60.
[9] Malekzadeh K, Khalili MR, Mittal RK. Analytical prediction of low-velocity
static crush simulation. Finally, experimental results are adopted impact response of composite sandwich panels using new TDOF spring-mass -
to verify the accuracy of quasi-static simulation. The simulation re- damper model. J Compos Mater 2006;40:1671–89.
sults from the quasi-static method and experiment are very close, [10] Miki M, Fukuda T, Motogi S, Houjiyou M. Composite Materials. Kyoritsu
Publishing Co., Inc.; 1997. p. 202–3.
and confirm the feasibility and accuracy of the proposed quasi-sta- [11] Luo, Lee. Static crush experiments and simulations of laminated composite
tic method. plates and shells with constrained layered damping. Compos Struct
2008;85:64–9.
References
[1] Sun, Liou. Investigation of laminated plates under impact dynamic loading
using a three-dimensional hybrid stress finite element method. Comp Struct
1989;33:879–84.
[2] Reddy YS, Reddy JN. Three-dimensional finite element progressive failure
analysis of composite laminates under axial extension. J Compos Technol Res
1993;15:73–87.