Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
net/publication/273178511
CITATIONS READS
5 167
5 authors, including:
Some of the authors of this publication are also working on these related projects:
Effect of damage on failure mode of multi-bolt composite joints using failure envelope method View project
All content following this page was uploaded by Shufeng Liu on 10 September 2015.
Original Article
Journal of Reinforced Plastics
and Composites
Abstract
Composite structures used in aircraft are vulnerable to low-velocity impact and replacing them with new ones requires
significant cost. Scarf-repair technique is an effective method to restore the stiffness and strength of damaged composite
structures. Previous studies on scarf-repaired composite laminates have focused on stiffness and strength recovery
efficiency, but limited attention has been directed toward the impact properties. However, scarf-repaired composite
laminates are also sensitive to low-velocity impact. The residual strength is worth investigating to ensure the safety of
scarf-repaired composite structures. In this study, experiments were conducted to test the residual compression
strength after impact. Two main factors, namely impact energy and location, were considered. Finite element models
were established to analyze the influences of these two factors and to predict the residual strength. Impact damages were
introduced by degrading the mechanical properties in the damaged area estimated through the C-scan photographs.
Results indicated that the damage area was larger when the impact was applied at the bondline of the top surface and the
residual strength was the smallest. Increasing the impact energy caused large damage area, in which the residual strength
declined. The predicted residual strength agreed with the experimental results well.
Keywords
Carbon fiber reinforced polymer laminates, scarf-repair, low-velocity impact, compression after impact, finite element
modeling
Many scholars have conducted a large amount of meet the damage tolerance restoration of the original
experimental and numerical researches on scarf- composite structure.18
repaired composite laminates over the past decades; All aforementioned studies have been conducted on
the majority of these studies focused on the recovery 2D scarf joint of composite laminate. However, in prac-
efficiency of static strength and stiffness after repair.8–12 tical engineering, a 3D scarf repair technique is typic-
To improve the strength and stiffness recovery effi- ally used to repair the damaged composite panels.18 In
ciency, numerical methods have been used to optimize this study, an experimental investigation was conducted
the repair parameters (scarf angle, adhesive thickness, to test the CAI strength of 3D scarf-repaired composite
overlap length, patch shape, etc.). However, a few con- laminates. Four impact locations were selected to
siderations have been given to the residual load- explore the influences of impact location on the
carrying capacity of scarf-repaired laminates subjected impact damage. Four different impact energies were
to low-velocity impact. Scarf-repaired laminates are loaded on the 3D scarf-repaired composite laminates
also sensitive to low-velocity impact. to identify the delamination threshold loads and
Harman and Rider13 used a 2D scarf joint to study impact damage characteristics under different impact
the damage tolerance and impact resistance of scarf- energies. Moreover, a 3D finite element (FE) model
repaired composite laminates. Their results indicated was constructed to predict the residual compression
that the impact damage pattern of scarf joint is strongly strength of the 3D scarf repaired composite laminates
influenced by the location of impact relative to the scarf with impact damages. A progressive damage model was
joint tip. The most severe damage may occur when the implanted to simulate the damage process during CAI
impact is some distance away from the tip, in which tests. Numerical results were then compared with
case the structure is minimally tolerant to low-velocity experimental results to verify effectiveness of the
impact. The compression after impact (CAI) strength numerical model.
of scarf joint laminates is relatively higher than the CAI
of original composite laminate, whereas the tensile after
impact (TAI) strength is lower than the TAI of original Experimental studies
composite laminate. CAI strength is more vulnerable to
hygrothermal environment condition than TAI
Specimens manufacture
strength. The effect of scarf angle was studied in The material selected for this study was carbon fiber
Hoshi et al.,14 in which three angles (3 , 4.5 , and 6 ) reinforced polymer laminate, which was made from
were chosen to conduct the experiments. The impact unidirectional carbon fiber epoxy prepregs T300/
damage of scarf joint increased with increasing scarf 5228A provided by Beijing Institute of Aeronautical
angle, which resulted in a decrease in the residual tensile Materials. To simulate practical engineering applica-
strength. To model the impact failure of the scarf joint tion, the stacking sequence was designed as [45/02/
of composite laminates, the authors assumed adhesive 45/90/45/02/45/0]s, and the total thickness was
failure as a more catastrophic failure mode than approximately 2.7 mm. The stack prepregs were first
composite failure and two adhesive damage models packed in a vacuum bag with a vacuum pressure of
(cohesive element model and elastic–plastic models) 980 mbar and then were cured at 180 C for approxi-
were used.15 The former one was to simulate the mately 2 h in an autoclave with a pressure of 0.6 MPa.
impact damage, whereas the latter one was for residual According to the ASTM D7137 standard,19 the cured
strength after low-velocity impact. Kim et al. tested composite panels were cut into required dimensions
the impact properties of a 2D scarf-repaired laminate (150 mm long 100 mm wide) for test specimens,
with a tensile prestrain up to 5000 microstrain.16 where the 0 direction was along with the longitudinal
Results indicated that the prestrain has significant influ- direction. Damage was assumed to occur in the center
ences on the mixed damage of adhesive disbonding of each laminate. A scarf-repair technique was used to
and matrix cracking. Ichiya Takahashi et al. detected repair the damage. As shown in Figure 1, the damaged
the impact damage of scarf joint of composite lamin- area was removed by a scarf angle of 6 , and the
ates, especially for debondings of adhesive layer, bottom diameter was 10 mm. Subsequently, a corres-
with the use of lamb waves.17 The debonding was ponding patch with the same size of removed damaged
caused by the shear cracks and delaminations near region was machined from the same material as the
the impact point and propagated toward the bottom parent laminate. An adhesive film SY-14 M with a
surface. Wang et al. investigated the CAI strength of thickness of 0.2 mm was used to bond the patch and
2D multistepped joint of composite laminate through the parent laminate. After the parent laminate, the
numerical and experimental methods. The CAI patch, and the adhesive film were assembled properly,
strength of multistepped joints was about 8% under the repair was cured at 180 C under a vacuum pressure
the original composite laminates, which is close to of 980 mbar for 2 h.
Shufeng et al. 3
Impact
Impact Specimen energy Specimen
Experimental test results
Specimen location group (J/mm) number Impact test results and C-scan
Original laminates Center Group 1 4.45 2 The impact test results were presented and discussed in
6.67 2 detail in Cheng et al.,20 which showed the location of
impact point had specific influences on the impact
Scarf-repaired P1 Group 2 4.45 2
behavior of scarf-repaired laminates. With the impact
laminates 6.67 2 point moving from P1 to P3, the top of the load history
P2 Group 3 4.45 2 curves had more oscillations. The maximum contact
6.67 2 force was the smallest when impact location was at
P3 Group 4 4.45 2 P3. Correspondingly, the rebound height of the impac-
6.67 2 tor was the lowest, too. That means more energy is
P4 Group 5 6.67 2 absorbed when the scarf-repaired laminate is impacted
at P3. However, when the impact location was at P4,
the responses were more alike those of original
laminates.
The impact load history curves presented in Cheng
Impact test et al.20 indicated that there were damages occurred in
As shown in Table 1, the scarf-repaired laminate speci- the original laminates and scarf-repaired laminates
mens were divided into four test groups categorized by from the early load decreasing points. Subsequently,
four impact locations. Original laminates were also the contact force increased again and entered into the
tested as a control group. The original laminates were fluctuations. During this period, the damages, espe-
impacted at the geometrical center. The four groups of cially delamination damages, developed quickly.
scarf-repaired laminates were impacted at points P1, Considering 6.67 J/mm impact energy as an example,
P2, P3, and P4, as shown in Figure 1. Each group the C-scan images are presented in Figure 5 to show
was tested under two impact energy levels, namely the projected damage area caused by the low-velocity
4.45 and 6.67 J/mm. As shown in Figure 2, the impact impact in the five groups. Comparison of Group 1 and
test apparatus consists of a mass block, an impactor, an Group 2 revealed that the impact induced less damage
acquisition device, and a computer. The specimen was in the scarf-repaired laminates than in the original
clamped on the base support at four points around the laminates. Analysis suggested that the adhesive intro-
four corners. Between the mass block and impactor, a duced in the scarf-repaired laminates could absorb par-
load transducer was used to collect the impact contact tial impact energy and mitigate the development of the
load during the impact moment, the duration of which delamination damages. The damage of Group 2 was
Shufeng et al. 5
Figure 5. Ultra C-scan images and damage appearance of the five group specimens under 6.67 J/mm impact energy. (a) Group 1,
(b) Group 2, (c) Group 3, (d) Group 4, and (e) Group 5.
of Group 1 and Group 2 proved that the scarf-repaired along the thickness, which was equal to the total
laminates achieved high restoration efficiency in impact number of the ply sequence. The material direction of
resistance. When the impact location moved from P1 each layer was modified according to the stacking
to P3, the residual strength declined smoothly. The resi- sequence of the laminate. The adhesive film was also
dual strength was lowest when the impact location was modeled using a 3D Solid45 element. Given that the
at P3, i.e. the patch edge. However, when impacted thickness of the adhesive film was only 0.13 mm, it
at P4, the residual strength increased again and the was divided into two along its thickness to avoid a
results were similar to those of Group 2. These experi- large shape ratio of the element. Figure 8 shows the
mental results have a good agreement with boundary condition and applied loading. Encastre
Whittingham et al.,7 i.e. the impact at the termination was applied at the lower side, and axial compression
location of the repairing patch caused the largest load was applied at the top side. The longitudinal cen-
strength reduction. tral face was constricted with a symmetric condition
about y axis. The out-of-plane degree of the two rows
of nodes 10 mm away from the side face was restrained.
Numerical analysis
3D FE model Failure criteria and property degradation rules
A 3D FE model was created using the ANSYS FE The elastic properties and strength of the CCF300/
code. Considering the symmetry of the scarf-repaired 5228A lamina and SY-14M film are shown in
laminates, half of the structure (see Figure 7) was mod- Table 3. For the parent laminate and patch, modified
eled to save the calculation cost. For the parent lamin- Hashin-type failure criteria21,22 were adopted to detect
ate and patch, an eight-node element Solid45 was used. the four failure modes: matrix cracking in tension and
The parent laminate and patch were divided into 20 compression, fiber breakage in tension and
Figure 6. Compressive destruction appearances of scarf-repaired laminates after low-velocity impact. (a) Impact at P1, (b) impact at
P2, (c) impact at P3, and (d) impact at P4.
compression, fiber matrix shear-out, and delamination Fiber compressive failure (11 5 0)
in tension and compression. The failure criteria that
2
corresponded to each failure mode are listed in equa- 11
tions (1) to (7). 1 ð4Þ
XC
Matrix tensile cracking (22 4 0)
Fiber matrix shear-out failure (11 5 0)
Shufeng et al. 7
2 2 2
11 12 13
þ þ 1 ð5Þ
XC S12 S13
2 2 2
33 13 23
þ þ 1 ð6Þ
ZT S13 S23
uz=0
ux= uy= uz 0
Start
Building FE model
No
Stress analysis
No
Yes Degradation of material properties at the
Check for failure
damaged elements
In the above equations, 11, 22, and 33 represent material. Maximum shear stress criterion was used to
stresses in direction 1, 2, and 3, respectively; 12, 23, detect adhesive failure. Once the damage occurred, the
and 13 represent the shear stresses in each direction, stiffness decreased to a considerably small value.
respectively. In the coordinate system, the 1 - and 2-
axes are parallel and transverse to the fibers, respect-
Progressive damage analysis method
ively. XT and XC are the tensile and compressive
strengths in x direction, and YT and YC are the tensile To predict the ultimate strength and simulate the
and compressive strengths in y direction, respectively; fracture process under compressive load, an APDL
ZT and ZC are the tensile and compressive strengths program was used. A typical progressive damage
along thickness direction, respectively. S12, S23, and analysis flowchart is shown in Figure 9. After build-
S13 are the shear strengths in each plane, respectively. ing the FE model of scarf-repaired laminate, the
Property degradation rule is another important con- impact damages were introduced in by degrading
cern for progressive damage models (PDMs). The the material properties (using a factor of 0.228,29)
material should be degraded according to failure in the damaged area according to C-scan results.
modes. Chang et al.23,24 and Tan25,26 presented degrad- Displacement load was applied on the right side of
ation rules for laminated composite with open holes. the repaired laminates and the load was increased by
However, they used two different sets of degradation 0.02 mm every load step. During each load step, 15
rules. These two sets of degradation rules are listed in substeps were set to detect the damages and to
Table 4. Chang’s complete degradation model and ensure that no element failure would occur under
Tan’s degradation model extended by Camanho and the corresponding displacement load condition.
Matthews27 were compared in this study for laminate When element failure happened, their properties
Table 3. Mechanical properties of CCF300/5228A unidirectional lamina and SY-14M adhesive film.
CCF300/5228A lamina
Matrix cracking in tension E22 ¼ G12 ¼ G23 ¼ v12 ¼ 0 E22 ¼ 0.2E22, G12 ¼ 0.2G12, G23 ¼ 0.2G23
Matrix cracking in compression E22 ¼ G12 ¼ G23 ¼ v12 ¼ 0 E22 ¼ 0.4Eyy, G12 ¼ 0.4G12, G23 ¼ 0.4G23
Fiber breakage in tension E11 ¼ E22 ¼ E33 ¼ G12 ¼ G23 E11 ¼ 0.07E11, E22 ¼ 0.07E22, E33 ¼ 0.07E33,
¼ G13 ¼ v12 ¼ v23 ¼ v13 ¼ 0 G12 ¼ 0.07G12, G13 ¼ 0.07G13, v12 ¼ 0.07v12,
v23 ¼ 0.07v23, v13 ¼ 0.07v13
Fiber breakage in compression E11 ¼ E22 ¼ E33 ¼ G12 ¼ G23 ¼ G13 E11 ¼ 0.14E11, E22 ¼ 0.14E22, E33 ¼ 0.14E33,
¼ v12 ¼ v23 ¼ v13 ¼ 0 G12 ¼ 0.14G12, G13 ¼ 0.14G13, v12 ¼ 0.14v12,
v23 ¼ 0.14v23, v13 ¼ 0.14v13
Fiber–matrix shear-out failure G12 ¼ v12 ¼ 0 G12 ¼ v12 ¼ 0
Delamination in tension E33 ¼ G23 ¼ G13 ¼ v23 ¼ v13 ¼ 0 E33 ¼ G23 ¼ G13 ¼ v23 ¼ v13 ¼ 0
Delamination in compression E33 ¼ G23 ¼ G13 ¼ v23 ¼ v13 ¼ 0 E33 ¼ G23 ¼ G13 ¼ v23 ¼ v13 ¼ 0
Shufeng et al. 9
4000 4000
3000 3000
2000 2000
1000 1000
0 0
0 10000 20000 30000 40000 50000 60000 70000 80000 90000 0 10000 20000 30000 40000 50000 60000 70000 80000 90000
Compression load (N) Compression load (N)
Figure 11. Strain–load curves obtained from experiments and numerical analysis using degradation rules SET 1 (a) and SET2 (b).
Ply No. 223.70 MPa 232.88 MPa 238.52 MPa 244.52 MPa
3 (0°)
11(90°)
20(45°)
Figure 14. Damage propagation in the scarf-repaired laminate subjected to impact at P1.
Figures 13 and 14 show the damage propagations in damage propagation differed in the location of the ini-
the adhesive film and the scarf-repaired laminate under tial failure and the direction. As the impact point
compressive loadings after low-velocity impact at P1, moved away from the center, the initial failure occurred
respectively. As shown in Figure 13, the damage in the around the initial impact damage and the propagation
adhesive film originated around the 0 plies in the top direction rotated toward impact location.
part of the laminates and then propagated to the
bottom part gradually. After the occurrence of adhesive
failure, the failure of laminates occurred in the 0 plies
Predicted CAI strength
in the top surface of the patch laminates. In the lower The predicted failure loads of the five groups under
0 plies and the middle 90 plies of the laminates, dam- 6.67 J/mm impact energy were summarized and com-
ages happened in the parent laminates around the pared with experimental results, as shown in Table 5
bonding line and spread to the side of the specimens. and Figure 15. The 3D FE model using the progressive
For 0 and 90 plies, the damages were mainly fiber failure model could predict the CAI strength of scarf-
failure and matrix failure in compression, respectively. repaired laminates well. The predicted results using
The major failure mode in the inner 45 plies was the degradation rule SET1 had a low accuracy whereas
fiber–matrix shearing out. For the other group, the those using SET2 had a high accuracy. The reasons
Shufeng et al. 11
Table 5. Compressive strength of original laminate and scarf-repaired laminate subjected to 6.67 J/mm impact.
50
results also showed that the CAI strength of the
repaired laminates decreased the most when they were
40
impacted at the patch edge. It indicated that the bond-
30 line area of scarf-repaired laminates was sensitive to
20 low-velocity impact, and the impact-induced delamin-
10 ation in the adhesive and parent laminate around the
patch edge caused large decrease in CAI strength.
0
original repaired repaired repaired repaired Third, a 3D FE model implanted with PDM was
plate P1 P2 P3 P4 used to predict the residual strength of the scarf-
repaired laminates subjected to 6.67 J/mm low-velocity
Figure 15. Comparison between the predicted failure load and impact energy. The model compared two material deg-
the experimental failure load. radation rules and found that the modified Tan’s deg-
radation rules had great accuracy in predicting CAI
for the results maybe that the material in the impact strength. However, the degradation rules insignificantly
damaged area was degraded throughout the thickness affected the predicted structural stiffness. The numer-
in the model, while delamination, matrix cracking, and ical model could agree with the experimental results
fiber breakage failure introduced by impacts distributed well in the initial stiffness and final failure strength.
along the thickness in a certain way. Degradation rule Fourth, the impact-induced damages accelerate the
SET2 defined that the damaged material maintained a damage propagation in the scarf-repaired laminates,
part of the original stiffness to obtain a result approach- and the failure behavior under compression load
ing the real strength. The numerical results showed the varies with the impact location. The damages initially
same trending with the experimental results, i.e. the occur around the impact-induced damage locus and the
compressive strength decreased most severely when propagation turns to be unsymmetrical when the
the impact location was at the bonding line between impact point is not at the center.
the patch and parent laminate. The CAI strength of
the scarf-repaired groups was close to that of the ori- Conflict of interest
ginal laminate plate. Therefore, the scarf-repair tech-
None declared.
nique could recover the damage tolerance to the level
of the original plate.
Funding
Financial support from National Natural Science Foundation
Conclusions of China (NSFC) on this study under Project No.11472024 is
In this study, the damage tolerance of scarf-repaired gratefully acknowledged.
laminates subjected to low-velocity impact was investi-
gated by experimental and numerical methods. First, it References
was shown in the experimental data that the scarf- 1. Soutis C. Fiber reinforced composites in aircraft construc-
repaired laminates had good impact resistance when tion. Prog Aerosp Sci 2005; 41: 143–151.
2. Falzon BG. Impact damage and repair of composite 16. Kim MK, Elder DJ, Wang CH, et al. Interaction of
structures. Aeronaut J 2009; 113: 431–445. laminate damage and adhesive disbonding in composite
3. Armstrong KB, Bevan LG and William C. Care and scarf joints subjected to combined in-plane loading and
Repair of Advanced Composites. Warrendale, PA: Social impact. Compos Struct 2012; 94: 945–953.
of Automotive Engineers, 2005, pp.495–500. 17. Takahashi I, Ito Y, Takeda S, et al. Impact damage
4. Duong CN and Wang CH. Composite Repair: Theory and detection on scarf-repaired composites using Lamb
Design. London: Elsevier Publications, 2007, pp.8–10. wave sensing. In: Proc 16th Int Conf on composite mat
5. Caminero MA, Pavlopoulou S, Lopez-Pedrosa M, et al. (ICCM-16), Kyoto, Japan, 2007.
Analysis of adhesively bonded repairs in composites: 18. Wang CH, Venugopal V and Peng L. Stepped flush
damage detection and prognosis. Compos Struct 2013; repairs for primary composite structures. J Adhesion
95: 500–517. 2014; 91: 95–112.
6. Katnam KB, Da Silva LFM and Young TM. Bonded 19. ASTM D7137/D7137M-07. Standard test method for
repair of composite aircraft structures: a review of scien- compressive residual strength properties of damaged
tific challenges and opportunities. Prog Aerosp Sci 2013; polymer matrix composite plates.
61: 26–42. 20. Cheng XQ, Zhao WY, Liu SF, et al. Damage of scarf-
7. Whittingham B, Baker AA, Harman A, et al. repaired composite laminates subjected to low-velocity
Micrographic studies on adhesively bonded scarf repairs impacts. Steel Compos Struct 2014; 17: 199–213.
to thick composite aircraft structure. Compos Part A Appl 21. Hashin Z. Failure criteria for unidirectional fiber com-
Sci Manuf 2009; 40: 1419–1432. posites. J Appl Mech 1980; 47: 329–334.
8. Hu FZ and Soutis C. Strength prediction of patch- 22. Tserpes KI, Papanikos P and Kermanidis Th. A three-
repaired CFRP laminates loaded in compression. dimensional progressive damage model for bolted joints
Compos Sci Technol 2000; 60: 1103–1114. in composite laminates subjected to tensile loading.
9. Kumar SB, Sivashanker S, Bag A, et al. Failure of aero- Fatigue Fract Eng Mater Struct 2001; 24: 663–675.
space composite scarf-joints subjected to uniaxial com- 23. Chang FK and Chang KY. A progressive damage model
pression. Mater Sci Eng A Struct 2005; 412: 117–122. for laminated composites containing stress concentra-
10. Pinto AMG, Campilho RDSG, de Moura MF, et al. tions. J Compos Mater 1987; 21: 834–855.
Numerical evaluation of three-dimensional scarf repairs 24. Chang KY, Liu S and Chang FK. Damage tolerance of
in carbon-epoxy structures. Int J Adhes Adhes 2010; 30:
laminated composites containing an open hole and sub-
329–337.
jected to tensile loadings. J Compos Mater 1991; 25:
11. Wang CH and Gunnion AJ. On the design methodology
274–301.
of scarf repairs to composite laminates. Compos Sci
25. Tan SC. A progressive failure model for composite lamin-
Technol 2008; 68: 35–46.
ates containing openings. J Compos Mater 1991; 25:
12. Cheng XQ, Yasir B, Hu RW, et al. Study of tensile failure
556–577.
mechanisms in scarf repaired CFRP laminates. Int J
26. Tserpes KI, Labeas G, Papanikos P, et al. Strength pre-
Adhes Adhes 2012; 41: 177–185.
diction of bolted joints in graphite/epoxy composite
13. Harman AB and Rider AN. Impact damage tolerance of
laminates. Compos Part B Eng 2002; 33: 521–529.
composite repairs to highly-loaded, high temperature
27. Camanho PP and Matthews FL. Progressive damage
composite structures. Compos Part A Appl Sci Manuf
2011; 42: 1321–1334. model for mechanically fastened joints in composite
14. Hoshi H, Nakano K and Iwahori Y. Study on repair of laminates. J Compos Mater 1999; 33: 2248–2280.
CFRP laminates for aircraft structures. In: Proc 16th Int 28. Cheng XQ, Zhang ZL and Wu XR. Post-impact com-
Conf on composite Mat (ICCM-16), Kyoto, July 2007. pressive strength of small composite laminate specimens.
15. Herszberg I, Feih S, Gunnion AJ, et al. Impact damage Acta Mater Compos Sin 2002; 19: 8–12.
tolerance of tension loaded bonded scarf repairs to CFRP 29. Cheng XQ, Ali AM and Li ZN. Residual strength of
laminates 2007. In: Proc 16th Int Conf on composite Mat stitched laminates after low velocity impact. J Reinf
(ICCM-16), Kyoto, July. Plast Compos 2009; 28: 1679–1688.