You are on page 1of 8

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/285037232

Failure mechanisms of scarf-repaired composite laminates under tensile load

Article  in  Journal of the Brazilian Society of Mechanical Sciences and Engineering · November 2015
DOI: 10.1007/s40430-015-0460-z

CITATIONS READS

12 354

5 authors, including:

Cheng Xiaoquan Hasan Junaid Hasham


Beihang University (BUAA) Institute of Space Technology
110 PUBLICATIONS   937 CITATIONS    18 PUBLICATIONS   42 CITATIONS   

SEE PROFILE SEE PROFILE

Musharaf Abbas Wajid Khan


UNSW Sydney University of Hafr Al Batin
32 PUBLICATIONS   283 CITATIONS    7 PUBLICATIONS   24 CITATIONS   

SEE PROFILE SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Misalignments in Welding Processes View project

Failures in scarf-repaired composite laminates View project

All content following this page was uploaded by Musharaf Abbas on 02 April 2016.

The user has requested enhancement of the downloaded file.


J Braz. Soc. Mech. Sci. Eng.
DOI 10.1007/s40430-015-0460-z

TECHNICAL PAPER

Failure mechanisms of scarf‑repaired composite laminates


under tensile load
Yasir Baig1 · Xiaoquan Cheng1 · Hasan Junaid Hasham2 · Musharaf Abbas1,2 ·
Wajid Ali Khan3 

Received: 28 March 2015 / Accepted: 13 November 2015


© The Brazilian Society of Mechanical Sciences and Engineering 2015

Abstract  An investigation based on experimental and 1 Introduction


finite element (FE) studies was carried out to understand the
failure mechanisms of scarf-repaired T700SC/NY9200GA Fiber-reinforced polymer composite materials possess high
composite laminates under tensile load. Test specimens specific strength and stiffness, structure designable and
were categorized into 4 groups based on the type of scarf excellent corrosion resistance properties when compared to
patch (pre-preg and pre-formed) and scarf angles (6°and conventional materials such as metals [1, 2]. Most of these
10°). A 3-D progressive damage accumulation model was composites used for aerospace are quite expensive thereby
developed to understand the damage mechanisms and making frequent replacements of damaged structures with
predict the ultimate strength of repaired laminates. Dam- new ones un-economical. An alternative methodology is
age characteristics of specimens were observed by optical to use effective repair techniques to redress the damaged
microscope and scanning electron microscopy. Test results structure at reduced cost without compromising structural
have shown that the laminates repaired with pre-preg scarf integrity [3, 4].
patch and 6° scarf angle have higher failure strength. It was Presently the main composite structure repair tech-
also found that the fracture morphology of failed laminate nologies contain patch-bonded and bonded scarf repair
is different but failure mode was found to be the same. The [5]. Bonded scarf repair is used in composite structures
crack initiated from highly stressed adhesive regions and when high strength recovery is needed or when there is a
then circumferentially propagated throughout the adhesive requirement for a flush surface to satisfy aerodynamic or
interface thus causing adhesive failure. FE results were stealth requirements. Bonded scarf repair effects are mainly
found to be in good agreement with experimental results. affected by scarf angles, patches, adhesives and other fac-
tors. From 1980, many researchers have focused on com-
Keywords  Laminates · Adhesion · Finite element posites scarf repair, and associated experiments have been
analysis (FEA) · Progressive damage model conducted as well. Strength recovery of about 76 % [6],
80 % [7] and 90 % [8] was reported for scarf-repaired lami-
nates. Scarf angle effects have been investigated through
experiments and damage models variations are concluded
[9–11]. Some other discoveries are found in the recent
Technical Editor: Fernando Antonio Forcellini. experiments, such as if the outer ply is 90°, the repaired
laminate will have more strength than that with 0° outer ply
* Musharaf Abbas
matengr01@gmail.com
[5].
On the basis of the oblique lap repair, Soutis analyzed
1
Beijing University of Aeronautics and Astronautics, No. 37 scarf-repaired structure with FE model [12]. 2D numerical
Xueyuan Road, Beijing 100191, China model was developed for analysis of scarf-repaired struc-
2
International Islamic University, H‑10, Islamabad 44000, ture (scarf angles ranging from 1.1° to 9.2°) [13]. Hashin-
Pakistan Lee criterion was applied to determine the laminate failure
3
Wah Engineering College, Wah Cantt 47040, Pakistan and failure load was calculated for various scarf angles [8].

13
J Braz. Soc. Mech. Sci. Eng.

Fig. 1  Schematic diagram of
scarf-repaired Laminate

In another research work, tensile strength of scarf-repaired


structure was obtained through experiments and numerical
calculation and results were found to be in good agreement
[14]. Considering the non-linearity of adhesive and com-
posites, a 3D model was built to predict the failure load of
both repaired and un-repaired composite structure [15].
Scarf repair parameters have been thoroughly studied
previously by many researchers. Some concluded that the
optimum scarf angle is 5° under uni-directional compres-
sive load, whereas, the optimum angle is 4° under bi-direc-
tional compression [16]. Some others proposed that with
lower scarf angle, peel stresses will be higher [17]. It was
found that the use of outer ply will largely decrease the
peel stresses [18].
In conclusion, present experiments have paid more atten-
tion towards the properties (tensile strength, compression
strength, etc.) of repaired laminates instead of further inves- Fig. 2  Experimental test setup
tigation on damage propagation and failure mechanisms.
Moreover, 3D scarf repair structure is usually simplified by
2D scarf joints which under-estimates the load capacity of (±45)2/0]s. According the composite repair guide [4], scarf
parent remaining structure after adhesive failure. repair was conducted on the laminates. Thickness of lami-
In this paper, the laminates repaired with different types nates as shown in Fig. 1 was variable from 2.6 to 2.9 mm,
of patches (pre-preg and pre-formed) and scarf angles (6° and the scarf hole diameter (D) was 20 mm. Two differ-
and 10°) were tested under tensile load. Damage charac- ent scarf angles (6° and 10°) and two types of patches (pre-
teristics of specimens were observed by metalloscope and preg and pre-formed) were used in repair process. Adhe-
scanning electron microscopy (SEM). A comprehensive sive Sy-14 was used to bond the parent laminate and patch
FE code was developed to understand the damage mecha- together. The material and ply-stacking sequence for both
nism and to predict the ultimate strength of scarf-repaired the parent laminate and patch was the same.
laminates. Testing was carried out on electro-hydraulic servo material
testing machine (see Fig. 2). Specimens were continuously
loaded at a speed of 2 mm/min speed until complete failure.
2 Materials and experimental procedure

20 specimens categorized in 4 groups according to scarf 3 Test results


angle and patch type of size 300 × 100 mm were pre-
pared. The matrix material used was NY9200GA and fibre Test results (tensile strength and failure strain of repaired
was 700SC with ply-stacking sequence of [±45/90/03/ laminates) are shown in Table 1. Results show that

13
J Braz. Soc. Mech. Sci. Eng.

Table 1  Tensile test results of Group Scarf angle/deg. Patches Elastic modulus/GPa Tensile strength/MPa Failure strain/μɛ
repaired laminates
I 6 Pre-preg 57.46 456 7743
II 6 Pre-form 60.13 348 5502
III 10 Pre-preg 60.98 448 5852
IV 10 Pre-form 61.73 434 5668

Fig. 3  Top surface of laminates (pre-preg patch and scarf angle 10°) Fig. 4  Bottom surface of laminates (pre-preg patch and scarf angle
10°)

laminates repaired with pre-preg patch and 6° scarf angle


have the highest tensile strength of 456 MPa. It is also
obvious that laminates repaired with pre-preg patches have
higher tensile strength than those repaired by pre-formed
patches irrespective of the scarf angle used, however, the
difference in strength is very small.

4 Damage characteristics of laminates

4.1 Observation method

In addition to the visual observation, other means of inspec- Fig. 5  Failure pattern of bottom surface
tion such as metalloscope and scanning electron micros-
copy (SEM) were used to observe the damage in repaired
laminates after failure.

4.2 Observation of macro‑morphology

Most of the specimens were found to be fractured within


the repair region (see Figs. 3, 4, Fig. 5). Observing from
superior surface (large patch circle side), it can be seen
that all the laminates have been cracked along upper or
lower patch circle edge and patches are separated from the
parent laminate (see Fig. 3). However, no damages were Fig. 6  Tensile fracture pattern of repaired laminate with removed
found in parent laminate or patch along the cracked adhe- patch (pre-preg patch and scarf angle 6°)
sive region.
At about 1/2 patch circle (the most narrow parent sec- It can be seen from Figs. 4 and 5 that the outer plies have
tor), the parent laminate damage is characterized by com- been fractured with “X” pattern and damages are not found
posite fracture [19]. However, the failure pattern of top on smaller circular surface of the hole. For 6° scarf angle
surface and bottom surface of the laminate is not the same. (see Fig. 6), the crack was propagated though a short

13
J Braz. Soc. Mech. Sci. Eng.

Fig. 7  Tensile fracture pattern of repaired laminate with adhered


patch (pre-preg patch and scarf angle 6°)

Fig. 9  Fracture morphology of parent repair area

Fig. 8  Tensile fracture pattern of repaired laminate with adhered


patch (pre-formed patch and scarf angle 6°)

distance horizontally and then in an oblique direction to


cause complete failure of laminate.
Opening the damage area along the fracture crack, frac-
ture patterns are shown in Figs. 6, 7 and 8. It can be seen
from the figures that patches remain undamaged and cracks
propagate along the adhesive and laminates ultimately frac-
ture at the most narrow region of parent laminate. Traces of
adhesive on the patch surface can be observed as shown in
Figs. 7 and 8.

4.3 Observation of fracture micro‑morphology

On the bases of macro-morphology observation, fracture


morphology is observed by SEM. For pre-preg patch repair, Fig. 10  SEM micrograph showing adhesive fracture of pre-formed
there was no obvious damage found on the oblique plane of patch repaired laminate SEM micrograph
parent scarf area. Few resin-rich areas and resin fragments
were found scattered as shown in Fig. 9. Parent plies ori-
entation is also visible and failure is characterized by fibre 4.4 Observation of non‑damaged adhesive area
fracture.
Moreover, adhesive film fracture pattern is similar to It was observed that the laminates crack along adhesive on one
that of thermo-set resin [19]. For pre-formed patch repaired half of circular region of patch and the opposite half of circu-
laminates, the damage type is shown in Fig. 10. It was lar region has no obvious damages under tensile fracture. Two
observed that adhesive film is fractured at the dimple edges specimens repaired with pre-preg patch and pre-formed patch,
found on the patch oblique plane. respectively were chosen for metalloscope observation in the

13
J Braz. Soc. Mech. Sci. Eng.

Fig. 13  Stress-displacement curves of scarf repaired laminates

Fig. 11  Voids in adhesive of pre-preg patch repair Material test parameters for both the parent laminate
and patch were the same. Material properties in Y-direction
and Z-direction were assumed to be the same. Ply direction
was defined by changing the material coordinate system for
each ply angle. For each ply of parent laminate and patch,
single layer of Solid46 element was used whereas adhesive
layer was divided into four layers of elements in thickness
direction for more accurate analysis of damage in adhesive
layer.
Load was applied in the form of displacement and
reaction forces and mean stresses were calculated after
every load-step to get stress-displacement curve. Analysis
was performed in 35 load steps. Hashin criterion [22] was
used to judge the matrix and fibre failure whereas, delam-
ination criterion reported by F.K.Chang [23] and modu-
lus attenuation criterion in the literature [24] was applied.
Adhesive failure was judged by maximum shear stress
criterion.
Fig. 12  Voids in adhesive of pre-formed patch repair
5.2 Comparison of experimental and FE simulation
results
non-damaged area of adhesive. As can be seen from Figs. 11
and 12, there are a lot of voids in adhesive, the voids of pre- Results obtained by FE simulation were found to be in
formed patch adhesive are larger than that of pre-preg patch. good agreement with experimental results so the developed
progressive damage model can be used to perform the dam-
age behavior analysis.
5 Finite element analysis of damage behavior Result for the laminates repaired by pre-preg and and
pre-form patches have some difference, however, results
5.1 Modeling and meshing were almost the same for both 6° and 10° scarf angles.
Therefore, damage propagation of laminates repaired by
Using ANSYS APDL, the damage propagation of scarf pre-formed patch and scarf angles 6° is calculated. The
repair laminate was modeled to predict the ultimate stress-displacement curve of repaired laminates using
strength and to understand the damage mechanism of scarf- pre-formed patch and 6° scarf angle under tensile load-
repaired composite laminates [20, 21]. One-fourth of scarf- ing is shown in Fig. 13. The curve is mainly divided into
repaired structure was modeled for its symmetry in calcula- two stages from initial loading to ultimate failure. There is
tion. Parent and patch were modeled with 3-D laminated a linear relationship between the displacement and stress
element SOLID46, and adhesive was modeled with iso- until adhesive failure occurs in the first stage. First stage
tropic element SOLID45. is completed at the moment when adhesive is completely

13
J Braz. Soc. Mech. Sci. Eng.

Fig. 14  Damage propagation in repaired laminate

failed. In the second stage, stress is redistributed and parent The difference in tensile strength for 6° scarf angle and
laminate bears the load and then fails shortly. 10° scarf angle was not very obvious.
The FE simulation results (see Fig. 14) show that cracks The crack was initiated from the highly stress concen-
initiate from the interface of adhesive and laminate (or trated region of adhesive and then propagated circumfer-
patch), and propagate along with the adhesive until its com- entially through the adhesive region. Reducing the stress
plete failure, and then propagate from adhesive to laminate concentration within adhesive film may help to improve the
free edge to cause final failure of parent laminate. ultimate tensile strength of repaired laminates.
Parent laminate was failed shortly after the adhesive
failure. It means that ultimate strength of repaired lami-
6 Conclusions nate is mainly dependent upon the strength of bonded
region.
Four groups of T700SC/NY9200GA laminates were Voids and defects in adhesive film were more severe
repaired with two different types of patches (pre-preg and in case of pre-preg patch as compared to pre-formed
pre-formed) and scarf angles (6° and 10°) and tested under patch.
tensile loading. Following conclusions were drawn: FE simulation results were found to be in good agree-
Laminates repaired with pre-preg patches achieved ment with experimental results, therefore, FE model can be
the highest tensile strength irrespective of the scarf angle used to understand the damage mechanism and predict the
employed. ultimate tensile strength of scarf-repaired laminates.

13
J Braz. Soc. Mech. Sci. Eng.

References 12. Soutis C, Hu FZ (1998) Failure analysis of scarf-patch-repaired


carbon fiber/epoxy laminates under compression. AIAA J
38(4):737–740
1. Hull D (1987) An introduction to composite materials. Cam-
13. Odi RA, Friend CM (2004) An improved 2-D model for bonded
bridge University Press, Cambridge
composite joints. Int J Adhes Adhes 24(5):389–405
2. Hou WG, Zhang WF, Ding ML, Liu X, Wang ZR (2010) Ther-
14. Campilho RDSG, de Moura MFSF, Domingues JJMS (2007)
mal deplying technology of damaged area in carbon fiber/bis-
Stress and failure analyses of scarf repaired CFRP lami-
maleimide resin laminates. In: The proceedings of 2010 Asia-
nates using a cohesive damage model. J Adhes Sci Technol
Pacific international symposium on aerospace technology, vol. 2.
21(9):855–870
pp 865–868
15. Tzetzis D, Hogg PJ (2008) Experimental and finite element anal-
3. Hou WG, Zhang WF, Tang QY (2012) Progress in patch repair
ysis on the performance of vacuum-assisted resin infused single
of aerospace composite structures. Proc. SPIE 8409, Third inter-
scarf repairs. Mater Des 29(2):436–449
national conference on smart materials and nanotechnology in
16. Yu M, Xu XW (2008) Study of the compression strength of scarf
engineering. doi:10.1117/12.923795
patch repaired composite structures. J China Univ Min Technol
4. Chen SJ (2001) Guide of composite structure repair. Aviation
37(5):711–714 [in Chinese]
Industry Press, Beijing [in Chinese]
17. Adkins DW, Pipes RB (1985) End effects in scarf joints. Compos
5. Wang CH, Gunnion AJ (2008) On the design methodology
Sci Technol 22(3):209–221
of scarf repairs to composite laminates. Compos Sci Technol
18. Gunnion AJ, Herszberg I (2006) Parametric study of scarf joints
68(1):35–46
in composite structures. Compos Struct 75(1–4):364–376
6. Found MS, Friend MJ (1955) Evaluation of CFRP panels with
19. Dong Z, Zhong PD, Tao CH (2003) Failure analysis. National
scarf repair patches. Compos Struct 32(1–4):115–122
defense industry, Beijing [in Chinese]
7. Jones JS, Graves SR (1984) Repair techniques for Celion/
20. Wang YL (2011) Study on scarf repair to composite laminates.
LARC-160 graphite/polyimide composite structures. Langley
Master Degree thesis, Beihang University
Research Center, United States NASA-CR-3794
21. Wang YL, Cheng XQ (2010) Influence of scarf angle on compos-
8. Meng FH, Chen SJ, Tong XY (2001) Selection of the design
ite scarf repair under a tensile load. Transformation of economic
parameters in laminate repair. Acta Materiae Compos Sin
development mode and self-dependent innovation, the twelfth
18(4):123–127 [in Chinese]
annual meeting of China association for science and technology,
9. Kumar SB, Sridhar I, Sivashanker SI et al (2006) Tensile failure
vol. II
of adhesively bonded CFRP composite scarf joints. Mater Sci
22. Hashin Z (1980) Failure criteria for unidirectional fiber compos-
Eng B Solid State Mater Adv Technol 132(1–2):113–120
ites. J Appl Mech 47:329–334
10. Kumar SB, Sivashanker SI, Osiyemi SO et al (2005) Failure of
23. Change FK, Springer GS (1986) The strengths of fiber rein-
aerospace composite scarf-joints subjected to uniaxial compres-
forced composite bends. Compos Mater 20:30–45
sion. Mater Sci Eng A 412(1–2):117–122
24. Cheng X Q Study on compressive properties of composite lami-
11. Pipes RB, Adkins DW, Deaton J (1982) Strength and repair of
nates and honeycomb core sandwich panels after low speed
bonded scarf joints for repair of composite materials. NASA
impact. Ph.D thesis, Beihang
Langley Research Center, United States. MSG 1304

13

View publication stats

You might also like