Composite Structures 123 (2015) 1–8

Contents lists available at ScienceDirect

Composite Structures

journal homepage: www.elsevier .com/locate /compstruct

Repairing impact damaged fiber reinforced composite pipes by externalwrapping with composite patches

http://dx.doi.org/10.1016/j.compstruct.2014.12.0170263-8223/� 2014 Elsevier Ltd. All rights reserved.

⇑ Corresponding author.E-mail addresses: mkara@konya.edu.tr (M. Kara), muyaner@selcuk.edu.tr

(M. Uyaner), aavci@selcuk.edu.tr (A. Avci).

Memduh Kara a,⇑, Mesut Uyaner b, Ahmet Avci c

a Necmettin Erbakan University, Seydis�ehir Ahmet Cengiz Engineering Faculty, Dept. of Metall. and Material Eng., 42370 Seydis�ehir, Konya, Turkeyb Selcuk University, Engineering Faculty, Dept. of Metall. and Material Eng., 42075 Konya, Turkeyc Selcuk University, Engineering Faculty, Dept. of Mechanical Eng., 42075 Konya, Turkey

a r t i c l e i n f o a b s t r a c t

Article history:Available online 16 December 2014

Keywords:Polymer-matrix composites (PMCs)Impact behaviorMechanical testingBurst strengthFailure behaviorComposite patch repair

Repairs made with composite patches on impact damaged fiber reinforced composite pipes offer distinctadvantages over traditional repairs in addition to reduced cost. In this study, effects of number of patchlayers on the burst pressure of low velocity impact damaged tubes that have been repaired with compos-ite patches were investigated. The tubes were pressurized up to 32 bar prior to impact. The pre-stressedglass fiber reinforced plastic tubes were damaged by applying low velocity impacts at different energylevels (5, 10 and 15 J). The damaged areas of the affected tubes were repaired with 2, 4 and 6 layers ofglass/epoxy fabrics. The repaired tubes were then failed catastrophically by being subjected to monotonicinternal burst tests based on ASTM D 1599–99 standards. Changes in the tubes’ burst pressures wererecorded and the resulting damages on the tubes were studied. It was found that, for all the energy levelsemployed in this study, a six-layered patch repairing is suitable for the retrofitting of impact damagedtubes.

� 2014 Elsevier Ltd. All rights reserved.

1. Introduction

Generally, metals subjected to low-velocity impact show plasticbehavior before failure and it may be possible to remove the effectsof deformation/damage by annealing and/or reworking the mate-rial. However; depending on the nature of the impact, non-visible,barely visible and visible irreversible failures may be occurred inthe fiber reinforced layered composites [1]. The repairs made ondamaged areas in order to achieve the original mechanical proper-ties tend to vary depending on the type of failure. Once damage isdetected and the effects on the residual properties of the structurehave been predicted, a decision must be made as to whether thiscomposite part should be repaired or replaced. There are caseswhere damage cannot be repaired. For instance, members thathighly stressed may not have sufficient strength after repair [2].

If the damage level is small enough to be mended, then therepair is executed. Several methods such as bolted collars andwelded collars are used for repairing damaged pipe lines. Recently,composite patches have seen an increasing usage in the repairingprocess too [3–6]. These patches are lighter, more resistant tocorrosion and easier applied than the conventional repairing

methods. In addition, an industry analysis showed that, on average,composite repair systems are 24% cheaper than welded sleeverepairs and 73% cheaper than replacing the damaged section ofpipe [7].

Most of the studies conducted on repair of damaged structureshave concentrated on planar plates failures. Repairs of damagedpipes and pressure vessels have been generally carried out by com-posite patches. Some of them are summarized as below:

Roberts [8] conducted experimental investigation on crackedsteel pressure vessel that was repaired with carbon fiber compositepatches. The researcher used standard tensile stress specimens tomeasure the effects of the repair. After the repair with the compos-ite patches, the cracked specimens were subjected to environmen-tal loads. Static and environmental loads were applied on thecracked steel specimens that were repaired with compositepatches and their behaviors studied. It was found that the crackpropagation was retarded and the life span of the specimensincreased. Hu et al. [9] repaired a cracked steel pressure vesselby using steel patches with epoxy glue. In a study by Wilson[10], a damaged steel pipe was repaired with carbon/epoxy wrap-ping. The energy release rate of the composite wrapping/steelinterface was obtained. A new laboratory specimen was createdto evaluate mixed mode debonding of composite over-wrappedpiping. Goertzen and Kessler [11] carried out dynamic andmechanical analysis in order to investigate mechanical and

Table 1Mechanical properties of the fiber and the resin.

E (GPa) rTS (MPa) q (g/cm3) et (%)

E-glass 73 2400 2.6 1.5–2Epoxy resin 3.4 50–60 1.2 4–5

Table 2Mechanical properties of the GRP tubes.

h: Fiber winding angle ±55�rt: Tangential failure stress (MPa) 428.96my: Poisson’s ratio 0.53Ey: Modulus of elasticity (GPa) 20.48Vf: Fiber volume fraction 0.50

Fig. 1. Low velocity impact test rig and hydraulic pump [18].

Table 3Mechanical properties and dimensions of the composite patches.

Ex = Ey: Modulus of elasticity (GPa) 22ry: Tensile strength (MPa) 292mxy = �ey /ex: Poisson’s ratio 0.16

Number of patch layers wp (mm) tp (mm)

2 Layered patches 0.30 1004 Layered patches 0.60 1006 Layered patches 0.90 100

Fig. 2. Repairing of the specimen with patch.

2 M. Kara et al. / Composite Structures 123 (2015) 1–8

thermal properties of the carbon/epoxy patch material they usedfor pipe repair. For this, they conducted a three-point bending test.In that study, effects of heating rate, frequency and measuringmethods on glass transition temperature were studied. Duellet al. [12] conducted a study in which a repair was made on steelpipes in order to reinforce them and stop their surface corrosion.They used carbon/epoxy composite as a repairing material. Theycarried out stress analysis for the damaged pipes having differentgeometries by using a three dimensional finite element method.They compared the experimental results with the ones obtainedfrom the finite element analysis method. In both results, it wasfound that the maximum stress occurs at the center of the dam-aged area. In a study by Gunaydin et al. [13], where experimentalinvestigation of the effects of composite patch repairing of sur-face-notched glass fiber reinforced plastic (GFRP) composite pipeson fatigue behavior of the pipes was made. The pipes have notchsize ratios of a/c = 0.2 and a/t = 0.75. The burst pressures of thepipes repaired with 100 mm wide patches with two, three, four,five, six and seven layers of the patch were higher than the burstpressure of the un-patched notched pipes. It was found that fatiguelife increases with the number of patch layers.

Composite patches are not only used for reinforcements andrepairing but can also be used in joining composite pipes end toend. Pang et al. [14], joined two 54� winding angle glass/epoxycomposite pipes end to end by winding them with a fiber rein-forced composite material. The joined pipes were subjected tointernal pressure and four-point bending tests. In a study by Pecket al. [15], two composite pipes were joined end to end by usingglass fabrics having different thicknesses and by using choppedglass fibers having different sizes with UV cured vinyl ester resin.The joined pipes were cured with UV lights. Mechanical propertiesof the pipes were determined with internal pressure and four-point bending tests. At the end of the tests, it was found that thepipes with three and five layers of glass clothes exhibit higherburst strengths than the eight layered glass cloth joints. This isso because; the latter had not undergone sufficient curing. In orderto determine the bending strength and bending rigidity of the joint,the researchers carried out bending tests. Li et al. [16], joined 54�winding angle glass/epoxy composite pipes end to end by usingfour different adhesives and a cross layer glass prepreg. To deter-mine the effectiveness of the joining method, the internal pressuretests and finite element analysis were conducted.

According to the authors’ knowledge there is no study that hasbeen encountered on the repairing of the damaged GFRP pipesunder low velocity impacts in literature.

In this study, GFRP tubes that were wound with filament wind-ing method were used. The GFRP tubes were made of E-glass/epoxy material with ± 55� winding angle. The tubes were manufac-tured as [±55�]3 (six-layered) manner at Izoreel Company, Turkey.The manufactured test specimens were subjected to low velocityimpact tests by applying 32 bar internal pressure on them atenergy levels of 5, 10 and 15 Joules. 2, 4 and 6 layered patches wereapplied as a repair on the damaged area which emerged as a resultof various energy levels being exerted on the tubes. The repairedtubes were subjected to monotonic internal pressure tests basedon the ASTM D 1599–14 standard [17]. Then evaluation on theeffects of patching on burst pressure of the tubes and their failurebehavior was carried out.

2. Materials and methodology

2.1. Material properties

In this study, six-layered glass/epoxy composite pipe specimenswith ± 55� winding angle were used. The composite tubes were300 mm long, having inner diameters of 72 mm and 2.375 mm

Table 4The values obtained from the force histories (X ± Sx) [18].

Appliedpressure [bar]

Impactenergy [J]

Impactvelocity [m/s]

Maximum contactforce [N]

Contacttime [ms]

Maximumdisplacement [mm]

Maximum internalpressure [bar]

Impulseforce [Ns]

Absorbedenergy [J]

0 5 1.26 2033.98 ± 117.68 9.10 ± 0.26 3.67 ± 0.17 – 11.82 ± 0.61 3.81 ± 0.3610 1.78 2255.96 ± 24.03 11.48 ± 0.12 6.02 ± 0.06 – 16.14 ± 0.58 8.14 ± 0.4515 2.18 2658.46 ± 101.16 12.10 ± 0.29 7.83 ± 0.31 – 20.14 ± 0.71 11.96 ± 0.13

32 5 1.26 2250.68 ± 131.80 7.56 ± 0.33 3.20 ± 0.11 33.98 ± 0.13 11.24 ± 0.63 4.16 ± 0.3210 1.78 2962.49 ± 106.41 8.48 ± 0.70 4.94 ± 0.23 35.56 ± 0.32 15.74 ± 0.74 8.43 ± 0.5215 2.18 3572.38 ± 99.44 9.24 ± 0.19 6.38 ± 0.20 37.82 ± 0.37 19.92 ± 0.48 12.06 ± 0.43

M. Kara et al. / Composite Structures 123 (2015) 1–8 3

thick. Mechanical properties of the fibers and resin used in thecomposite tubes are given in Table 1 whereas the mechanical prop-erties of the tubes are presented in Table 2.

2.2. Low velocity impact tests

In order to inflict impact damage on the GFRP specimens sub-jected to internal pressure, a low velocity impact device and aninternal pressure unit were used. These two devices are shown inFig. 1. The impactor on the impact device had a semisphericaltap, a mass of 6.35 kg and a diameter of 24 mm. The test specimenswere placed on a V shaped bearing and subjected to 32 bar of pres-sure from the internal pressure apparatus, subsequently. Then lowvelocity impact tests at various energy levels were imposed on thespecimens. Each specimen received only one strike.

2.3. Repair of the impact damaged GFRP tubes

The damaged area of the tubes affected by low velocity impactsat varying energy levels was repaired by applying 2, 4 and 6 layersof patches. The patches used for the purpose were made of E-glassfabric having thickness of 0.15 mm, width of 100 mm and arealdensity of 200 g/m2. The prepared glass fabrics were wrapped overthe tubes after being saturated with epoxy resin by using a brush.The wrapping was made in such a way that the patching layerswould be in two, four and in six-layered manner. The repairingprocess was accomplished by curing the patched tubes in a furnace

Fig. 3. Transverse cross sections of damaged areas of the GRP spe

at 120 �C for 3 h. Mechanical properties and the dimensions of thecomposite patches are given in Table 3.

The impact tests for three different energies (5, 10 and 15 J) andthree different patch layers (2, 4 and 6-layered) were separatelyrepeated three times for each specimen dimension. That is,twenty-seven tests were totally performed. In Fig. 2, the sketchof the specimen damaged by low-velocity impact is seen afterrepairing it with patch.

2.4. Monotonic internal pressure tests

The impact damaged tubes which have been repaired were sub-jected to monotonic burst pressure tests based on the ASTM D1599–14 standard. During the tests, a hydraulic pump was usedto pressurize the pipe specimens. When the oil was allowed to flowinto the tubes, care was taken to make sure that the ultimate fail-ure of the tubes occurs at 60–70 s from the initial loading of thetubes and that the pressure loading is linear. In case the ultimatefailure occurred earlier or later than 60–70 s, then the test wasrepeated.

3. Results and discussion

3.1. Low velocity impact tests

The specimens are pressurized up to 32 bar of internal pressureand impacted by using a hemi-spherical indenter. To assure that

(a)

(b)

(c)

cimens under impact energy levels of (a) 5 J (b) 10 J (c) 15 J.

Impact energy = 5 J

160

180

200

220

240

260

280

300

Non-damaged Without patch 2-layered patch 4-layered patch 6-layered patch

Number of patch layers

Burs

t pre

ssur

e [b

ar]

Impact energy = 10 J

160

180

200

220

240

260

280

300

Non-damaged Without patch 2-layered patch 4-layered patch 6-layered patch

Number of patch layers

Burs

t pre

ssur

e [b

ar]

Impact energy =15 J

160

180

200

220

240

260

280

300

Non-damaged Without patch 2-layered patch 4-layered patch 6-layered patch

Number of patch layers

Burs

t pre

ssur

e [b

ar]

(a)

(c)

(b)

Fig. 4. Variations of burst pressure with number of patches for damaged specimensunder energy levels of (a) 5 J, (b) 10 J and (c) 15 J.

4 M. Kara et al. / Composite Structures 123 (2015) 1–8

only the impact energy is related to the failure mechanism withinthe specimen, after the first impact, the striking unit was held andhence prevented from inflicting further blows on the specimen.Data for force variations from the beginning to the end of the blowwere transmitted to the electronic device by using a force trans-ducer. With the help of the NI Signal Express TM software, time his-tories were obtained.

Entities like the maximum impact velocity, the maximum con-tact force, duration of contact, maximum displacement, maximumvalue of internal pressure; impulse force and the amount ofabsorbed energy, all of which obtained from the conducted lowvelocity impact tests at varying energy levels were determinedand presented in Table 4. It was found that all the values obtainedfrom different energy levels tend to increase as the impact energyincreases. These values were already discussed in [18].

3.2. Analysis of low velocity impact failure

In order to investigate the failures of impacted specimens, thetubes were sectioned from the damaged zones. After sanding andpolishing, cross sections were observed by using an optical micro-scope. Fig. 3 shows transverse cross sections of the GFRP speci-mens’ damage areas after being impacted with 5 J, 10 J and 15 J(Magnification: 8�). When the cross sections are studied, it isobserved that radial matrix cracks and delamination are vivid.The failures seem to increase as the impact energy increases. Forthin layered composites, the bending stresses on the nonimpactedface cause matrix cracking on the lower most layer and this initi-ates the matrix cracks and delamination giving a view of a reversepine tree appearance [2]. Although the thin walled compositestubes were used in this study, inverted pine tree damage patternsoccurred because prestressed tubes cannot be deformed easierthan the non-prestressed one. The damaged specimens wereexposed to ignition test ASTM D2584–11 standard [19] and thefibers were examined. No fiber breakage was observed on the spec-imens for each energy level.

3.3. Monotonic internal burst pressure tests of the repaired GFRP tubes

3.3.1. Burst pressureFig. 4 shows variations of burst pressure with the number of

patch layers for the GFRP pipe specimens; the specimens havebeen repaired with the patches after being damaged by theimpacts of energy levels varying at 5 J, 10 J and 15 J, respectively.In addition, for each energy level, comparison was made betweenburst pressure values of undamaged specimens and those damagedbut without patching. It is clearly shown from the Fig. 4 that for allthe energy levels applied in this study, damaged specimens thathave been repaired with two layered patches exhibited no positiveinfluence on the burst pressure. As for the specimens with four-layered patches, it seems that their burst pressures have increased,however; the values reached required values only at the energylevel of 5 Joules but not otherwise. For 10 J and 15 J energy levels,the specimens with four-layered patches did not give expectedresults. The required results could be obtained for all the energylevels by six-layered patches.

3.3.2. Burst damageDuring the monotonic internal burst pressure tests of the GFRP

tubes, five important failure steps were observed. These failuresteps are whitening initiation, dense whitening, leakage initiation,oil jet formation and ultimate failure [18]. Due to the fact that thespecimens were tested under open ended conditions, the axialstresses on the specimens may be neglected. As the internal pres-sure applied on the specimens starts to increase, the lengths of the

specimens tend to shorten while their diameters enlarge. As thepressure increased, whitening initiation was observed and this ten-dency kept increasing. The whitening had caused separation offibers off the matrix interface and led to delamination [20].Together with this, matrix cracks formed in between the matrixlayers on the specimens progressed and the first leakage occurred.As the internal pressure kept increasing, the leakage turned into oiljet and the ultimate failure occurred when the tubes failedcatastrophically.

M. Kara et al. / Composite Structures 123 (2015) 1–8 5

In Figs. 5–7, photographs showing ultimate failure modes of thespecimens after the burst tests are shown, prior to the tests, thespecimens impacted with 5 J, 10 J and 15 J of energy levels; thephotos also show the damaged areas having no patch, two layeredpatching, four-layered patching and six-layered patching,respectively.

Photographs showing the ultimate failure modes were obtainedby passing light through the specimens in order to obtain highresolution on the photos. Therefore; the ultimate failure modeson the specimens could be clearly seen from those photos. Thewhitening on the GFRP tube specimens observed during the burstpressure tests appeared as lack of the transparency on the speci-mens. As such, when the light was passed through the specimensand their photos taken, areas with whitening looked like dull. Itis important that such a phenomenon is taken into account duringevaluation of the failure progresses of the specimens subjected tothe tests [18].

Ultimate failure images of the impacted but not repaired spec-imens at are presented in Figs. 5a, 6a and 7a. Impact damage hashad an effect on the failure propagation of the specimens sub-jected to burst tests at various impact energy levels. In this study,it was found that for all the energy levels applied, whiteninginitiation occurs on the specimen when the value of internal

(a)

(b)

(c)

(d)

Fig. 5. Ultimate failure photographs for the specimen under the influence of 5 Joules opatches, (c) 4-layered patch and (d) 6-layered patch.

pressure reaches 170 bar. The whitening progressed as the inter-nal pressure kept increasing. Radial cracking and delaminationwhich are the results of impact damage propagated and fiber sep-arations occurred on the matrix interfaces. When the internalpressure reached the value of 260 bar, the specimen impactedwith 5 J of energy failed catastrophically without experiencingleakage or formation of strong oil jet. The main reason for theburst of the specimen is the low damage level inflicted by theimposed impact. On the other hand, the specimens impacted withhigher levels of 10 J and 15 J did not failed catastrophically. Inthese specimens, as the internal pressure increase the radialcracks and the delamination progressed rapidly through the crosssection of the tube and the pressurized oil filled in the delamina-tion region and finally reached to the surface. The matrix crackson the surface progressed and oil leakage started at the impactarea. For the specimen impacted with 10 J, a strong oil jet formedat a pressure of 221 bar and reached an ultimate failure while thespecimen with 15 J impact damage experienced the same situa-tion at a pressure of 195 bar.

The repair must restore the strength of the part to withstandthe design ultimate loads. The repair patch must carry the loadacross the hole and restore stiffness and strength to the damagedarea [2]. Two-layered patch did not meet these expectations,

Burst pressure: 260 bar

Burst pressure: 259 bar

Burst pressure: 283 bar

Burst pressure: 291 bar

f impact energy; the damaged areas are (a) no repaired; repaired with (b) double

6 M. Kara et al. / Composite Structures 123 (2015) 1–8

because it is relatively thin. Thus, whitening initiated at the patchregion and tube at the same time (Fig. 8).

Shortly after the initiation of whitening, two-layered patchespeeled from the tubes due to the shear stresses resulted from thecircumferential expansion. Separation of the patches from the tubeled to carrying of the whole load by the impacted specimen itself.That’s why, in this study, the specimen with two-layered patch, atall energy levels, experienced similar failure behavior towardsinternal pressures as a specimen without patch. Ultimate failureimages of specimens with two-layered patch for 5 J, 10 J and 15 Jimpact energies are shown in Figs. 5(b), 6(b) and 7(b), respectively.

Circumferential expansion for the specimen under the influenceof various impacts was restricted by the four-layered patch duringburst pressure tests. This condition has affected the failure propa-gation on the specimens. In this case, the whitening initiated athigher pressures (190 bar). Therefore, other failure mechanismstook place at higher pressures as well. As the internal pressure keptincreasing, the patch separated from the tube as a result of theincreasing shear stresses and non-reversible circumferentialexpansion. During the test conducted with 5 J impacted specimenwith four-layered patch, the patch split and then the tube failedcatastrophically from the damage area when the pressure reached

(a)

(b)

(c)

(d)

Fig. 6. Ultimate failure photographs for the specimen under the influence of 10 Joulespatches, (c) 4-layered patch and (d) 6-layered patch.

to 283 bar. Because the highest stress on the pipe under internalpressure occurs on the circumferential direction, splitting of thepatches under the influence of this stress took place on axial direc-tion. The region of the patch ruptured is the place where theimpact damage occurred. During the monotonic internal burstpressure test, the specimen experienced the ultimate failure with-out undergoing leakage initiation or formation of strong oil jet. Themain reason for this is the low level of formation of the radialcracks and delamination on the specimen due to the 5 J impactdamage. Debonding and delamination are shown in Fig. 5(c). Inaddition, it is seen that the patch and the specimen have split alongthe axial direction as a result of the internal pressure due to thefiber breakage. The radial matrix cracks and delamination formedintensively on the specimens with 10 J and 15 J impact energy lev-els. For the specimens with four layered patches, leakage initiationappeared on the impacted area as soon as patches split. This is sobecause; the increase in diameter has led to increasing radialmatrix cracks and delamination; where the liquid oil began reach-ing onto the surface. As the leakage intensified, together with thesplitting of the patches, the ultimate failure took place in termsof strong oil jet. While the 10 J impacted and then repaired with4 layered patch specimens experienced ultimate failure at a pres-

Burst pressure: 221 bar

Burst pressure: 220 bar

Burst pressure: 263 bar

Burst pressure: 278 bar

of impact energy; the damaged areas are (a) no repaired; repaired with (b) double

(a)

(b)

(c)

(d)

Burst pressure: 195 bar

Burst pressure: 196 bar

Burst pressure: 230 bar

Burst pressure: 276 bar

Fig. 7. Ultimate failure photographs for the specimen under the influence of 15 Joules of impact energy; the damaged areas are (a) no repaired; repaired with (b) doublepatches, (c) 4-layered patch and (d) 6-layered patch.

whitening

M. Kara et al. / Composite Structures 123 (2015) 1–8 7

sure of 263 bar, similar specimens but impacted with 15 J exhib-ited theirs ultimate failure at 230 bar. This difference is the out-come of damages from the impact. Ultimate failure images forthe specimens that impacted 10 J and 15 J and then both repairedwith four-layered patches are shown in Fig. 6(c) and Fig. 7(c),respectively.

The six-layered patch restricted the circumferential expansionduring burst pressure tests. For this reason, whitening initiated at

whitening

patch tube

whitening

Fig. 8. Whitening on the two-layered patches.

patch tube

patch tube

whitening

(a)

(b)

Fig. 9. Whitening on the six-layered patches.

8 M. Kara et al. / Composite Structures 123 (2015) 1–8

the outside of the patch region (Fig. 9(a). The damage begun at thepatch region after internal pressure reached a certain value (Fig. 9(b).

The restriction of circumferential expansion for the six-layeredpatch is large because it is stiffer than the four-layered one. For allthe energy levels applied in this study, the ultimate failureoccurred by bursting of the specimens. The specimens with six-lay-ered patch reached their ultimate failure at a pressure of 291 bar,278 bar and 276 bar for the specimens impacted with 5 J, 10 Jand 15 J, respectively.

The specimen exerted with 15 J of impact damage experiencedmore damage as compared to the specimens inflicted with 5 J or10 J. However; the damages inflicted include no perforations.Although the specimens had no perforations, their burst pressuresdecreased to about 32%. It is therefore, very important to repairthe impact damage or limit the propagation of the damage. Theimpact damaged specimen with two layered patch could not bringany limitation to circumferential expansion. That is why a doublepatch repairing is not suitable. As for a four-layered patch repair,the circumferential expansion on the specimen was restricted. Nev-ertheless, for a specimen inflicted with 15 J of impact damage, thisrestriction was not at a desired level. This is so because at 15 Jimpact energy the specimen has undergone serious damage andthe pressurized oil had propagated through this damaged area ontothe specimen surface. The best result for specimens with 15 J ofimpact was obtained with six-layered patches repairing. Due tothe patch thickness on six layered patches, the circumferentialexpansion in the specimens becomes substantially restricted withrespect to the other layer types and hence the value of the burstpressure that causes ultimate failure becomes very close to theburst pressure of an undamaged specimen.

4. Conclusions

1. The cross section of the specimen damaged with 5 J of impactenergy is characterized by very little radial matrix cracks anddelamination in between the layers. As the impact energyincreases, the radial cracks and delamination tend to increaseas well. For all the energy levels dealt with in this study, no fiberdamages or perforation of the specimens were observed as aresult of imposed impacts.

2. While the ultimate failures of the specimens damaged with 5 Jof impact energy occurred by bursting of the specimens, thespecimens damaged with 10 J and 15 J of impact energy levelsexhibited their ultimate failures by experiencing formation ofstrong jets of oil gushing out of the specimens.

3. As the number of patch layers on patch repaired GFRP compos-ite tubes increases, the burst pressure of those specimens tendto improve as compared with unrepaired ones. However; for allthe energy levels, a double patch repairing did not restoredesign burst pressures of the specimens. Two-layered patchesrepairing proved unfertile for all the specimens in terms ofburst pressures of the repaired pipe specimens, while applica-tion of four-layered patches repairing is fruitful only for speci-mens damaged with 5 J of energy levels but not for the rest ofthe specimens and six-layered patches proved successful forall the specimens considered in this study.

4. During the monotonic internal burst pressure tests, ultimatefailures for specimens damaged with 5 J of energy occurred byway of specimen explosion in all of the double, four and six lay-ered patches. However; for specimens affected with 10 J ofimpact energy, only the specimens with four- and six-layeredpatches exhibited the explosion as their sign of ultimate failure

with those having two layered patches gushing oil jets as theirsign of ultimate failure. In case of specimens with 15 J of impactenergy, the ultimate failure for two- and four-layered speci-mens occurred by gushing oil jets while the specimens repairedwith six-layered patches experienced bursting as their ultimatefailure sign.

5. The results show that repairing by external wrapping with com-posite patch supplied sufficient retrofitting of impact damagedcomposite tubes.

Acknowledgments

This study was carried out as a PhD thesis by Memduh KARA inthe Graduate School of Natural and Applied Science at the Univer-sity of Selcuk, Konya, Turkey. This work was also supported by Sel-cuk University Scientific Research Projects under Grant Numbers09101030.

References

[1] Reid SR, Zhou G. Impact Behaviour of Fiber-Reinforced Composite Materialsand Structures. United States of America: CRC Press, Woodhead Pub.; 2000.303 s.

[2] Abrate S. Impact on Composite Structures. Cambridge: Cambridge UniversityPress; 1998. 135–160.

[3] Lukacs J, Nagy G, Török I. Experimental and numerical investigations ofexternal reinforced damaged pipelines. Procedia Eng 2010;2:1191–200.

[4] Shouman A. An experimental and numerical assessment of composite repairedpipes under a combined loading state [MS thesis]. Dalhousie University,Halifax, Nova Scotia, Canada; 2010.

[5] Toutanji H, Dempsey S. Stress modeling of pipelines strengthened withadvanced composites materials. Thin-Walled Struct 2001;39:153–65.

[6] Khawaja IA. Repair techniques for locally buckled energy pipelines using fibrereinforced polymer composites [MS thesis]. University of Alberta, Edmonton,Alberta, Canada; 2003.

[7] Koch GH, Brongers MP, Tompson NG. Corrosion cost and preventativestrategies in the United States. Federal Highway Administration, Office ofInfrastructure Research and Development. 2001. p. 260–11

[8] Roberts PD. Crack growth retardation by carbon fiber composite patching: Anapplication to steel pressure vessel repair [M.S. thesis], University of Alberta,Edmonton, Alberta; 1995.

[9] Hu YQ, Li PN, Ju DY, Hong-Liang Pan HL. Experimental investigation on acracked body with adhesive bonded reinforcement. Int J Pressure Vessel Piping1990;41:193–206.

[10] Wilson JM. Characterization of a carbon fiber reinforced polymer repair systemfor structurally deficient steel piping [Ph.D. thesis]. University of Tulsa; 2006

[11] Goertzen WK, Kessler MR. Dynamic mechanical analysis of carbon/epoxycomposites for structural pipeline repair. Composites: Part B 2007;38:1–9.

[12] Duell JM, Wilson JM, Kessler MR. Analysis of a carbon composite overwrappipeline repair system. Int J Pressure Vessels Piping. 2008;85:782–8.

[13] Gunaydin B, Daghan B, Avci A. Fatigue behavior of surface-notched compositepipes repaired by composite patches. Int J Damage Mech 2013;22(4):490–8.http://dx.doi.org/10.1177/1056789512450596.

[14] Pang SS, Li G, Jerro HD, Peck JA, Stubblefield MA. Fast joining of compositepipes using UV curing FRP composites. Polymer Compos ProQuest Sci J2004;25(3):298.

[15] Peck JA, Jones RA, Pang SS, Li G, Smith BH. UV-Cured FRP joint thickness effecton coupled composite pipes. Compos Struct 2007;80:290–7.

[16] Li G, Davis D, Stewart C, Peck J, Pang SS. Joining composite pipes using hybridprepreg welding and adhesive bonding. Polymer Compos ProQuest Sci J2003;24(6):697.

[17] ASTM Standard D1599-14, Standard Test Method for Resistance to Short-TimeHydraulic Pressure of Plastic Pipe, Tubing, and Fittings. ASTM International,West Conshohocken, PA, 2014. doi: http://dx.doi.org/10.1520/D1599-14,www.astm.org.

[18] Kara M, Uyaner M, Avci A, Akdemir A. Effect of non-penetrating impactdamages of pre-stressed GRP tubes at low velocities on the burst strength.Composites Part B 2014;60:507–14.

[19] ASTM Standard D2584. Standard Test Method for Ignition Loss of CuredReinforced Resins. ASTM International, West Conshohocken, PA, 2011. doi:http://dx.doi.org/10.1520/D2584-11, www.astm.org.

[20] Venkata MK Akula. Constitutive modeling of damaged unidirectionalcomposite laminae, University of Wyoming, ProQuest, UMI DissertationsPublishing, 2007. 3291044. p. 43.