Composites: Part B 30 (1999) 621–627

www.elsevier.com/locate/compositesb

Residual strength and fatigue life assessment of composite patch repaired

specimens
V.V. Rao a,*, R. Singh b, S.K. Malhotra c,1
a
Aeronautical Development Agency, PB No. 1718, Vimanapura Post, Bangalore 560 017, India
b
Aerospace Engineering, Indian Institute of Science, Bangalore 560 012, India
c
Indian Institute of Technology, Chennai 600 036, India
Received 20 January 1998; accepted 18 December 1998

Abstract
Design, analysis and technology for the integrity enhancement of damaged or underdesigned structures continues to be an engineering
challenge. Bonded composite patch repairs to metallic structures is receiving increased attention in the recent years. It offers various
advantages over rivetted doubler, particularly for airframe repairs. This paper presents an experimental investigation of residual strength
and fatigue crack-growth life of an edge-cracked aluminium specimen repaired using glass epoxy composite patch. The investigation begins
with the evaluation of three different surface treatments from bond strength viewpoint. A simple thumb rule formula is employed to estimate
the patch size. Cracked and repaired specimens are tested under static and fatigue loading. The patch appears to restore the original strength
of the undamaged specimen and enhance the fatigue crack growth life by an order of magnitude. q 1999 Elsevier Science Ltd. All rights
reserved.
Keywords: Fatigue life assessment

1. Introduction Life extension programs might demand a considerable

amount of preventive repairs to be performed. In short,
Aircraft structures get damaged during service, regardless repairs and modifications offer economic solutions to struc-
of how carefully they are designed, operated and maintained tural integrity enhancement for an airframe component in
due to fatigue, environment and accident. The care only service [2]. Most repairs are addition of material with or
reduces the rate of deterioration. Executing repairs to without surface treatment. The bonded patch repair offer
damaged airframes is unavoidable primarily due to creditable advantages over conventional methods of repair
economic considerations and to prevent further damage to the [3,4] which include (i) easy conformance to complex aero-
already damaged ones. It would be logical to leave a dynamic contour, (ii) improved fatigue behaviour, (iii)
damaged component unattended rather than performing a retarded corrosion and (iv) low parasitic stiffening along
bad repair, which could lead to situations of peril. Of the the cracked direction.
various damages, those due to fatigue account for majority Formal design and technological aspects of bonded
of failures [1]. Fatigue damage appears as fine cracks first composite patch repairs were initiated by Baker and his
and if permitted, they propagate to failure at alarmingly fast colleagues [5,6]. This technique is receiving ever-increasing
rates. From safety considerations, it is important to detect attention for the past decade and a half. In the present piece
these cracks and arrest/retard their growth to the extent of investigation at the Composites Technology Centre of the
possible. On the contrary, the design deficient components IIT, Madras, an attempt has been made to evaluate the resi-
would require reinforcement to restore strength and longevity. dual strength and life of repaired samples using glass fibre
composites as patching materials of commercially pure
aluminium sheet specimens with edge cracks. The study
* Corresponding author. Tel.: 1 91-80-5087337; fax: 1 91-80- begins with the evaluation of the various surface treatments
5232790.
and development of a method to arrive at the patch size.
E-mail addresses: vvrao@ada.ernet.in (V.V. Rao), malhotra@cc.iitm.
ernet.in (S.K. Malhotra) Notched and repaired specimens are then tested for strength
1
Tel.: 1 91-44-4458705; fax: 1 91-44-2301039. and fatigue life.
1359-8368/99/$ - see front matter q 1999 Elsevier Science Ltd. All rights reserved.
PII: S1359-836 8(99)00024-4
622 V.V. Rao et al. / Composites: Part B 30 (1999) 621–627

factor of safety.
Min‰W patch Lpatch su;patch ; Wpatch Lpatch tu;adhesive Š

$ Factor of safety × su;sheet atsheet :

In addition there may be constraints on the minimum
patch and adhesive thickness and the size of the patch (W,
L). A certain amount of patch material is desirable ahead of
the crack tip so as to add reasonable fatigue life to the
repaired configuration.

3. Surface treatment and bond strength evaluation

Overall quality of repair is governed by the bond strength

in addition to the patch size and the patch material strength.
This in turn is highly sensitive to the surface characteristics
achievable through the surface treatment. Surface treat-
ments of various types have been evaluated by Reinhart
Fig. 1. Specimen configuration (not to scale): (a) bond strength test and (b) [7]. He recommends phosphoric acid anodising as a method
patch strength and life test.
of surface treatment for aluminium alloys to achieve accep-
table levels of bond strength. He also mentions that use of
2. Patch size assessment chemicals in surface treatment can be prohibitive due to the
fact that they cannot be readily removed. In our investiga-
Designing a repair generally differs form structural tion, based on feasibility in our laboratory, two methods of
product design. Repairs can be corrective or preventive. surface treatments, viz. mechanical and chemical have been
Corrective repairs are generally damage specific, thus single evaluated.
in quantity and often require quick implementation. In
contrast the preventive repairs offer time to design and 3.1. Mechanical methods
test before scheduled implementation to a large number of Two processes have been of interest in this case: (i)
structures in service. Quick design and implementation of surface treatment by manual emery and (ii) electric disc
corrective repairs in the field, based on simple mathematics sander. In manual emery, the emery fineness of 60 gave
and a few variables is a highly desirable situation. Experi- an average surface roughness value of Ra ˆ 5 mm: In the
mental evaluation or computer simulation of each and every case of the electric sander, the fineness of 120 gave Ra ˆ
repair configuration is highly unrealistic and out of means. 2:5 mm roughness.
An approximate method to arrive at the size of the patch for
a given edge notch is proposed here. The method is based on 3.2. Chemical methods
virgin strength of aluminium, bond strength between alumi-
nium and composite patch, strength of the composite patch, A solution of potassium dichromate and sulphuric acid in
notch size and constraints on the geometry. It aims at reco- the ratio of 1:10 (28.5 and 285 g) is made and diluted with
vering the loss of strength due to the notch by the patch distilled water to a quantity of 1 l. The solution is heated to a
tensile strength and the adhesive shear strength with a factor temperature ranging between 60 and 708C and then the
of safety. specimens are soaked for 10–15 min. Later they are
Consider a notch of length a in a sheet of thickness removed, washed thoroughly in running water and dried
tsheet. The loss of strength will be su;sheet atsheet : If this is in hot air. This method of surface treatment gave a surface
repaired by a patch of size W patch Lpatch tpatch and a bond roughness value of 0.3 mm.
of thickness tadhesive, then the bond strength as well as For the three surface treatment methods considered, bond
the patch strength must exceed the strength loss by a strength tests were conducted using the specimen shown in

Table 1
Bond strength evaluation for various surface treatments

Surface treatment Surface roughness Ra (mm) Bond strength (MPa) of five specimens and their average

Manual Emery 5.0 4.5 4.0 6.6 4.8 5.1 5.0

Electric disc sander 2.5 8.5 9.1 10.1 8.9 8.9 9.1
Chromic acid etching 0.25 6.3 7.2 6.1 6.4 6.5 6.5
 V.V. Rao et al. / Composites: Part B 30 (1999) 621–627 623

Table 2 4, 8, 12, 16 and 20 mm long. These correspond to an

Material failure strength data a=W ratio of 0.1, 0.2, 0.3, 0.4 and 0.5, respectively. The
Material Failure strength (MPa) notches are produced by diamond tipped band saw,
resulting in a blunt notch of 1 mm width.
Five specimens Average 3. Based on the strength considerations of glass epoxy a
Aluminium 130 120 116 122 122 122
single layer of unidirectional composite patch should be
UD Gfrp 360 380 360 370 380 370 adequate. From fabrication considerations (hand layup
method), we selected a two layer patch giving us a patch
thickness of 1.4 mm.
Fig. 1(a). The tests were carried out as per ASTM D2733 4. We initiated our investigation with a square patch,
test standard. Specimens were fabricated by in situ layup of knowing very well that rectangular patches could be
gfrp layers on aluminium specimens after surface treatment more efficient. We hope to come up with a rule of
and acetone wash. They have been cured at 1208C for 4 h. thumb for the rectangular aspect ratio during our inves-
After the specimens have cooled to room temperature they tigation subsequent to this reported work. As per the
have been finished to final size as shown in the Fig. 1(a). thumb rule algorithm proposed in the sections above,
Two saw cuts were given one on the aluminium side and the size of the square patch W ˆ L† is calculated for
other on the composite side. Unlike the lap shear specimen various notch lengths with constraints as tpatch ˆ
this configuration produces a pure shear and no bending. 1:4 mm as minimum, W minimum ˆ 1:3a and a factor of
The results are obtained are given in Table 1. safety ˆ 1:5: For various notch lengths the safety factor
Electric disc sander method of surface treatment gives the obtained are presented in Table 3.
highest bond strength and is selected as the process for the 5. Square patches are cut to these sizes from a larger glass/
patching of notched specimens. epoxy sheet of 300 × 300 mm:
6. The notched specimens are surface treated with electric
3.3. Patch material selection disc sander while the emery is applied to the composite
patch.
The most popular materials for the composite patch are 7. Aluminium specimens and glass/epoxy patches are
boron/epoxy and carbon/epoxy. Till date most of the repairs cleaned with acetone to remove traces of grease or oil
carried out with reference to aluminium aircraft structures from the surface to be bonded.
were with either of these as patch materials. Boron/epoxy 8. The acetone cleaned specimens are dried, bonded with
offers an advantage in terms of lesser differential thermal epoxy adhesive and cured for four hours at 1208C. The
expansion coefficient with respect to aluminium. We configuration of the patch repaired specimens is shown
selected glass/epoxy for the reasons of cost and availability in Figs. 1(b) and 2.
of this material in our centre, knowing very well the fact that 9. The fibre volume fraction V f † obtained for the patch
this may not be a good choice of patch material as it is less material is approximately 0.32.
stiffer in comparison to aluminium and is not likely to give 10. Adhesive thickness is estimated from the measurements
large benefits in terms of life extension. Glass/epoxy and of bonded specimens and the constituent adherands
aluminium specimens were tested for tensile strength. The before the bond. It is found to vary between 0.1 and
values obtained are presented in Table 2. 0.2 mm.

4. Repair specimen preparation

4.1. Residual strength evaluation
For aluminium and fatigue life evaluation, the composite
patch repaired aluminium specimens are prepared as The notched and repaired specimens are tested on a 20 t
follows: UTM at a loading rate of 2 mm/min. Load level is increased
until the specimens fail either at the notch, at the bond or
1. Basic aluminium strip size is 200 × 40 × 1:5 mm3 : away from the repaired region. Fig. 3 and Table 4 show the
Aluminium strips are sheared to an approximate dimen- strength values obtained for various notched and repaired
sion and then finished to the right size. specimens. All the repaired specimens failed away form the
2. Single normal edge notch is added to the aluminium patched region. It is evident that the repairs fully recovered
strip. Five different notch lengths are considered viz. from the loss of strength due to presence of the notch. There
Table 3
has been no failure due to debonding or delamination. This
Patch sizes for various notch lengths suggests the efficiency of the bonding methodology. For
specimens with greater a=W ratio (0.4 and 0.5), the edges
a=W ratio 0.1 0.2 0.3 0.4 0.5 of the notch did open up due to out-of-plane deformation
Notch size (mm) 4 8 12 16 20
Patch size (mm) 11 16 19 22 26
towards the unpatched side. This problem can be solved by
patching symmetrically, i.e. on both sides with same volume
624 V.V. Rao et al. / Composites: Part B 30 (1999) 621–627

blunt notch width of 1 mm can be directly thought of as a

crack tip hole of 1 mm diameter.
The fatigue testing was carried out on a closed-loop elec-
tro-hydraulic Material Testing System (MTS 810) machine.
The maximum load is taken as 3.5 kN which equals 50% of
the residual strength value of the patch repaired specimen
for a=W ˆ 0:5: The load ratio Pmin =Pmax † is selected as 0.1.
The frequency of the test was fixed at 5 Hz.
The notched specimens failed at a crack length of 24 mm.
In the repaired specimens, the crack growth was monitored
up to 10 mm (4 mm beyond the patch boundary). None of
the specimens failed until then. At this stage the specimens
were subjected to NDE by ultrasonic C-scan. This did not
reveal any debond or delamination. The fatigue tests were
continued until failure. Specimens failed at different crack
Fig. 2. Photograph of the notched and repaired specimen. lengths varying between 32–38 mm, by ending up into two
pieces. It is very hard to interpret how failure precipitated.
of the patch material. However our primary interest is in Most probably the debonding occurred first followed by the
single sided patches. ligament tearing. Else there is a finite probability that in one
odd specimen the crack could have grown to the other edge
and the patch still holding the two pieces together.
4.2. Fatigue life evaluation Fig. 4 and Table 5 present the fatigue life of the notched
and repaired specimens with a=W ˆ 0:5 (20 mm notch
From the static testing, it has been seen that the residual length and a repair patch of 26 × 26 mm†: If there were a
strength is fully recovered by patching. We now look at the crack in the specimen and left unattended the life of the
other important requirement of fatigue life enhancement specimen would have been just about 500 cycles, the fatigue
achievable through patch repairs. life that corresponds to crack growth only. If this crack was
The fatigue life of these notched and repaired specimens attended to by just drilling a hole at the crack tip then it
would have two components (i) crack initiation life at the would add a crack initiation life (4750 cycles) to the speci-
notch and (ii) the crack growth life. Patching should add life men leading to a life extension factor of about 10 (500–
to these components by reducing the stress concentration at 5250). Putting a patch after drilling a hole enhances the
the notch and later reducing the stress intensity at the crack crack initiation life by about 3 times (4750–14 430) and
tip. The fatigue precracking of the specimens is deliberately the crack growth life by about 67 times (500–33 570).
ignored for the reason that in real life repair, one would drill Had the cracked sheet been patch repaired without drilling
a small hole at the crack tip and then put a patch [8,9]. The a stop hole, the life extension could have been only 67 times

Fig. 3. Residual strength of notched and repaired specimen.

 V.V. Rao et al. / Composites: Part B 30 (1999) 621–627 625

Table 4
Residual strength of notched and repaired specimens

Specimen a=W Failure load (N) % Recovered

Five specimens Average

Notched 0.1 6800 6400 6900 6600 6300 6600 –

Repaired 7300 7200 7300 7100 7300 7240 99.2
Notched 0.2 6000 5700 5800 5600 5900 5800 –
Repaired 7300 7300 7200 7100 7000 7180 98.4
Notched 0.3 5000 5400 5200 5100 4800 5100 –
Repaired 7300 7200 7300 7300 7300 7280 99.7
Notched 0.4 4500 4900 4200 4000 4650 4450 –
Repaired 7300 7300 7300 7200 7300 7280 99.7
Notched 0.5 3650 3750 3000 4000 4100 3700 –
Repaired 7300 7300 7300 7300 7300 7300 100

(500–33 570). Thus the total maximum life extension band of 1.5 mm width along the crack indicating small
achievable by the combination of stop hole and a patch debonding regions. The region obviously carries very
is about 100 (500–48 000). All this is clear from Table 5. high shear stress as the bond transfers loads primarily at
One of the reasons for the scatter in the fatigue life of the ends.
repaired specimens is probably due to the difference in the
adhesive thickness. The increase in the fatigue life is
however small since glass/epoxy with a relatively lower 5. Strength tests on fatigue specimens
Young’s modulus offering lesser restraint on crack opening.
A stiffer patch material would reduce crack tip stress inten- Residual strength tests were done on specimens sub-
sity factor much more than a softer material leading to a jected to fatigue loading to evaluate loss of repair
larger increase in the crack growth life, up to 1000 times strength with usage. Five specimens were loaded to
[10,11]. failure after 14 000, 33 000 and 43 100 cycles each
As suggested by Baker [7], the tested specimens were (Table 6 and Fig. 5). The results obtained are: after
subjected to thermal stripping of the patch. Most of 14 000 cycles, the specimens failed away from the patch
the area under the patch was found to be shining without indicating adequate residual strength of the patch. At
any traces of oxidation discoloration, except for a narrow this stage and subsequent to this, the strength of the

Fig. 4. Fatigue crack growth life of notched and repaired specimen.

626 V.V. Rao et al. / Composites: Part B 30 (1999) 621–627

Table 5
Residual life of notched and repaired specimens for a=W ˆ 0:5

Specimen Fatigue life Crack length (mm) Fatigue life cycles

Five specimens Average

Notched Initiation – 3560 4940 5170 4670 5120 4750

Repaired Initiation – 12 830 15 170 15 290 14 970 13 890 14 430
Notched Growth 4 410 470 590 510 520 500
Repaired Growth 10 31 460 26 760 30 720 32 540 21 920 28 680
Notched Total 4 4170 5410 5760 5270 5640 5250
Repaired Total 10 44 290 41 930 46 010 47 510 35 810 43 110
Repaired To failure 12–18 50 270 49 980 51 020 51 620 47 110 48 000

Table 6
Residual strength loss due to fatigue cycling

Fatigue cycles Crack length (mm) Failure load (N)

Five specimens Average

14 000 0–1 7220 7290 7360 7360 7360 7320

33 000 5–7 5870 5660 6070 6670 6070 6070
43 100 8–10 4790 4990 5390 5330 5120 5120

Fig. 5. Residual strength loss in repaired specimen with fatigue cycles.

patch degraded as expected. The failure probably pre- by the adequate recovery of static strength and reasonable
cipitates with debonding of the patch first followed by increase in the fatigue life, considering that it is a rather less
the ligament tear. The scatter in the fatigue crack growth stiff glass epoxy patch.
life can be once again attributed to variations in adhesive The selected surface treatment method (electric disc
thickness. sander) as against the chemical method (PAA), which is
the most sought after one, appears to be a good choice.
This method of surface treatment has tremendous impetus
6. Concluding remarks for site repair.
The reduction in the residual strength of the repaired
The technique of assessing the size of the patch in a specimens with fatigue cycles has been reasonable and as
simple manner has been robust enough, which is revealed expected.
 V.V. Rao et al. / Composites: Part B 30 (1999) 621–627 627

Acknowledgements composite patch repair. Proc Ninth Int Conf Fracture 1997;ICF-
9:2159–2166.
[5] Baker AA, Jones R, editors. Bonded patch repair of aircraft structures
Author acknowledges with pleasure the thought provok- Dordrecht: Martinus Nijhoff, 1988.
ing discussions with Prof. C.R.L. Murthy and Mr M.R. Bhat [6] Baker AA. Fiber composite repair of cracked metallic aircraft compo-
of the Aerospace Engineering IISc, Bangalore and their help nents—practical and basic concepts. Composites 1987;18:293–308.
in fatigue testing. [7] Reinhart TJ. Surface treatments for bonded repairs of metallic compo-
nents. In: Baker AA, Jones R, editors. Bonded patch repair of aircraft
structures, Dordrecht: Martinus Nijhoff, 1988.
References [8] Lai MO. Fatigue performance of repaired aluminium panels. Int J
Damage Mech 1993;2:199–204.
[1] Shin CS, Wang CM, Wong PS. Fatigue damage repair: comparison of [9] Baker AA. Growth Characterisation of fatigue cracks repaired with
possible methods. Intl J Fatigue 1996;18:535–546. adhesively bonded boron/epoxy composite. Proc Ninth Int Conf Frac-
[2] Singh R. Repairs to damage tolerant structures: issues and a point. ture 1997;ICF-9:117–128.
Aero Soc India J 1996;48:169–186. [10] Umamaheshwar TVRS. Fatigue crack growth studies under repair
[3] Baker AA. Repair of cracked or defective metallic aircraft compo- patches. ME dissertation, Indian Institute of Science, Bangalore,
nents with advanced fibre composites—an overview of Australian December 1997.
work. Composite Structures 1984;2:153–181. [11] Mahadesh Kumar A. Finite element modelling of repair patches. ME
[4] Mahadesh Kumar A, Singh R. 3D finite element modelling of a dissertation, Indian Institute of Science, Bangalore, December 1997.