Bond Behavior of Fiber Reinforced Polymer Bars under Direct Pullout Conditions

allowed toteel barsr-reinforcedbars undererials. Thisent length,

so shown to

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.

Bond Behavior of Fiber Reinforced Polymer Barsunder Direct Pullout Conditions

Zenon Achillides1 and Kypros Pilakoutas2

Abstract: This paper examines the behavior of Eurocrete fiber-reinforced polymer~FRP! bars~glass, carbon, aramid, and hybrid! inconcrete under direct pullout conditions. More than 130 cube specimens were tested in direct pullout where no splitting wasdevelop. In normal concrete, the mode of bond failure of FRP bars was found to differ substantially from that of deformed sbecause of damage to the resin rich surface of the bar when pullout takes place. Bond strengths developed by carbon fibepolymer and glass fiber-reinforced polymer bars appear to be very similar and just below what is expected from deformed steelsimilar experimental conditions. The load slip curves highlight some of the fundamental differences between steel and FRP matpaper reports in detail on the influence of various parameters that affect bond strength and development such as the embedmtype, shape, surface characteristics, and diameter of the bar as well as concrete strength. The testing arrangement is alinfluence bond strength because of the ‘‘wedging effect’’ of the bars.

DOI: 10.1061/~ASCE!1090-0268~2004!8:2~173!

CE Database subject headings: Bonding strength; Fiber reinforced polymers; Bars; Concrete; Steel; Pull-out resistance.

usedhatt canele-re-

orce-ed to

gssfullfac-aretrated

con-cost-d

viron-ith-

restry,avior.bond’’ of

s re-orces

udiedimen-

ur-sionsel. Inonsid-

thet will

con-s andfortscon-signthe

e ofhe

ableket-

nderr of

llout;rein-underd ontail inalsoworksigniation

ingece.

ing,c.ukssionste bygingpos-002.

Introduction

In construction, steel reinforced concrete is the most widelystructural material in the world. However, it is well known tunder certain environments corrosion of steel reinforcemenlead to the deterioration or even the collapse of structuralments. Billions of dollars are spent every year worldwide inpairing and strengthening concrete structures whose reinfment has deteriorated due to corrosion. This has contributresearch that focuses on alternative solutions~Pilakoutas 2000!.

Fiber-reinforced polymer~FRP! materials offer a promisinsolution because for many years they have been used succein other industries~such as the automobile and sports manuturing industries! and, more recently, in construction. Theremany examples of structural applications that have demonsthe speed and convenience of strengthening and repairingcrete structures using advanced composites to produceeffective solutions~Head 1996!. In addition, structures reinforcewith FRP bars have been in service under aggressive enments in various parts of the world for more than 15 years wout any structural problems~Rostasy 1996!. Nevertheless, befoFRP materials are widely accepted in the construction induresearch must be done on all aspects of their structural behOne of the fundamental aspects of structural behavior isdevelopment, because bond is the key for the ‘‘cooperation

1Scientific Associate, National Technical Univ. of Athens, BuildMaterials Laboratory, Iroon Polytechnion, Zographon, Athens, GreE-mail: [email protected]

2Reader, Univ. of Sheffield, Dept. of Civil and Structural EngineerMappin St., Sheffield, S1 3JD, UK. E-mail: [email protected]

Note. Discussion open until September 1, 2004. Separate discumust be submitted for individual papers. To extend the closing daone month, a written request must be filed with the ASCE ManaEditor. The manuscript for this paper was submitted for review andsible publication on February 20, 2002; approved on October 2, 2This paper is part of theJournal of Composites for Construction, Vol. 8,
No. 2, April 1, 2004. ©ASCE, ISSN 1090-0268/2004/2-173–181/$18.00.
JOURNAL OF COMP

J. Compos. Constr. 20

y

reinforcing bars and concrete. An adequate level of bond iquired between reinforcement and concrete to transmit ffrom one to another~Pilakoutas et al. 1997!.

Bonding of steel reinforcement to concrete has been stextensively in the last 40 years, and a huge amount of expertal and analytical work has been published on this subject~CEB1982; Task 2000!. However, the design formulas of the most crent design codes of practice do not incorporate any provifor the use of alternative reinforcing materials other than stethe best cases, some provisions for epoxy-coated bars are cered. The introduction of FRP reinforcing bars has creatednecessity for the development of design specifications thaallow engineers to use these materials as reinforcement increte structures. In order to overcome this problem, engineerresearchers round the world are currently intensifying their efto understand how these new materials actually interact withcrete in order to be able to contribute to the formulation of decodes of practice. A part of this research effort wasEUROCRETE project~Clarke and Waldron 1996!. EUROCRETEwas a 4-year research project that investigated the usnonferrous ~FRP! reinforcement in concrete structures. TEUROCRETE project led to the development of a new durFRP bar, which is now commercially available in the marplace, and the production of design guidelines~Institution 1999!.

Two major experimental series of tests were conducted uthe EUROCRETE project to investigate the bond behavioFRP reinforcement in concrete structures~Achillides 1998!. In thefirst series, more than 100 specimens were tested in direct puwhereas, in the second, the bond splitting behavior of FRPforcing bars was examined in nine concrete beams testedfour-point bending. In this paper, the main emphasis is placethe pullout tests; whereas, beam tests will be presented in dea following publication. The work described in this paper ispart of the work of the European Union funded research netCONFIBRECRETE, which aims for the development of deguidelines for FRP reinforced concrete structures in assoc
with Task Group 9.3 of the International Federation of Concrete.
OSITES FOR CONSTRUCTION © ASCE / MARCH/APRIL 2004 / 173

04.8:173-181.

d per-tressllouts de

evel-ndi-they

f thempleh

ell-n begains

tion-ullou

foro findnves-erginn ofrnedof theengthefor-loadFor

pted.imensend ogate

d onrete.

fiber,pe ofect of

e onronglysteelbondateri-r the

t theube.e bariam-pped

actthen

was

e re-essiveube’sachwith

mentesin.sionnu-

amidehichrfacethe

ttle-ough0.75

d dif-forma-icitysionable 2the

Thespe-

ma-hich

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.
Pullout Testing
Pullout tests are used commonly in the assessment of bonformance of steel reinforcing bars in concrete. Although the sconditions developed in the concrete specimen during putests are rarely encountered in practice and the bond valueveloped under those tests differ substantially from those doped in reinforced concrete elements for most practical cotions, pullout tests have been widely adopted. For this reasonoffer an economical and simple solution for the evaluation obond performance of reinforcing bars and represent in a simanner the concept of anchoring a bar~Cairns and Abdulla1995!. However, it should be stressed that a good bond in wconfined pullout tests does not mean that a good bond caachieved in a concrete member where the cover resistance asplitting determines the ultimate failure load~Achillides et al.1997a!.

The main aim of bond tests is to obtain the bond-slip relaship at the loaded and free ends of FRP bars subjected to a pload. A careful evaluation of the pullout arrangements usedsteel bars by previous researchers was conducted in order tthe most appropriate set-up for experimental purposes. The itigation resulted in two options for pullout tests: the Losb~1963! and the RILEM/CEB/FIP~1978! arrangements shownFig. 1, both of which are used commonly for the evaluatiobond for steel bars. However, the writers were initially conceabout the accuracy of the measurements of slip at the endbar obtained by the RILEM test, because the embedment lof the bar is at the very end of the cube. In this case, any dmation at the end of the concrete caused by the pulloutwould be recorded as bar slip although it is not actually slip.this reason, a modified version of Lowberg’s test was adoHowever, at a later stage of the testing series some specwere prepared having the bar-concrete contact area at thethe bar, similar to the RILEM arrangement, in order to investidifferences in bond development by the two arrangements.

The selection of primary variables for this study was baseexisting experience of bond behavior of steel bars in concThe most important factors examined were the type of barconcrete strength, diameter of the bar, shape of the bar, tybar surface characteristics, embedment length, and the effembedment length location in the concrete cube.

These factors were expected to have a different influencthe bond behavior of FRP bars because FRP materials are stanisotropic and have different mechanical properties thanbars. A proper evaluation of the influence of these factors ondevelopment is crucial to the understanding of how these mals cooperate with concrete in structural members and foestimation of adequate anchorage lengths~Achillides et al.

Fig. 1. Typical pullout tests

1997b!.

174 / JOURNAL OF COMPOSITES FOR CONSTRUCTION © ASCE / MARC


-

t

t

f

Specimen Preparation and Material Properties

Prior to casting, the FRP bars were properly marked so thaembedment length would lie in the middle of the concrete cThe embedment lengths were designed as multiples of thdiameter to facilitate comparisons between bars of different deters. The two ends of the bar in the concrete cube were wrawith several layers of cling film in order to form noncont~debonded! areas between bar and concrete. The bars werepositioned vertically in 150 mm cube molds, and concretecast around the bar.

Material Properties

ConcreteMore than 10 castings were made during the period of thsearch study. The concrete mixes had average comprstrengths in the range of 15.5–49.5 MPa. The concrete c~150 mm on each side! compressive strength values for ebatch of pullout tests are shown in the Appendix, togetherthe experimental results for each test.

Fiber-Reinforced Polymer MaterialsThe FRP bars tested were pultruded during the developstages of the EUROCRETE bar, using a vinyl ester hybrid rUp to 70% volume fraction was achieved during the pultruprocess. Three kinds of fibers were initially tried for the mafacture of four types of reinforcing bars: carbon, e-glass, arand hybrid~carbon plus glass together!. However, the bulk of thtesting was done using bars with the first two types of fiber, wseemed more promising for the purposes of the project. Sudeformations are created by the addition of a peel-ply onsurface of the bar during pultrusion, which is removed after sement and curing of the resin. This procedure created a rexternal surface on the bar, with an average peak height ofmm ~see Fig. 2!.

The bars that were tested in this experimental series haferent types of cross-sectional areas, sizes, and surface detion textures as shown in Table 1. Young’s modulus of elastfor the EUROCRETE FRP bars was evaluated by direct tentests, and the average values obtained are presented in T~Duranovic et al. 1995!. The characteristic tensile strength ofvarious bars is also given in the table.

Experimental Procedure

The pullout test schematic arrangement is shown in Fig. 3.concrete cube with the embedded FRP bar was placed in acially made steel frame that was positioned in the testingchine. The rig consisted of two steel plates 25 mm thick, w

Fig. 2. EUROCRETE glass fiber-reinforced polymer bar

were connected at the four edges with four rods 20 mm in diam-

H/APRIL 2004

04.8:173-181.

nterhreehole,cers

h thed one of

f theload

te, antactbear-

at then theThe

hine,the

seater

weree av-epre-dmen

ementslip

islr

d be-theneraled in

cube

the

-

ength

r castg an

andend

l

-P

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.

eter. The top plate had a hole 30 mm in diameter in its ceallowing the FRP bar to run through. On this plate were tadditional holes in a triangular arrangement around the mainwhich allowed three linear voltage displacement transdu~LVTD !, located at the loaded end of the specimen, to touctop surface of the concrete cube. A fourth LVDT was attachea small aluminum frame that was glued to the bottom surfacthe concrete cube to measure the slip at the free~unloaded! end ofthe bar. The bottom end of the rig was secured in the jaws otesting machine, which provided the reaction to the pulloutresisted by the specimen.

Between the concrete block and the bearing steel pla5-mm-thick wooden plate was introduced to secure the cobetween the top surface of the concrete block and the steeling plate. This was necessary because small irregularitiestop surface of the cube might introduce accidental bending obar during loading or movements caused by local crushing.test specimen was positioned in a universal testing macwhich applied direct tension in a deflection-control mode tospecimen~at a maximum rate of about 0.1 kN/s!. The test wastopped only when slip at the loaded end of the bar was grthan 8 mm.

Analysis of Measurements

The measurements obtained from the experimental resultsused to produce the bond-slip curves for each specimen. Therage displacement measured of the three LVDTs not only rsents bar slip but also the rebar extension above the embe

Table 1. Types of Bars Used in Pullout Tests

Type of bar Surface texture

Dimensionsof cross section

~mm!

Cross-sectionaarea

~mm2!

GFRP round Rough 13.5 143.13GFRP round Rough 8.5 56.27GFRP round Medium rough

hde f,0.25 mm10.5 86.59

GFRP square Rough 8.538.5 72.25GFRP round Smooth 16 201.06Carbon round Rough 13.5 143.13Carbon round Rough 8 50.26Carbon ring Roughout, smoothin dout521, din510 267.82

Carbon square Rough 8.538.5 72.25Aramid round Rough 13.5 143.13Aramid round Rough 8.5 56.27Aramid square Rough 8.538.5 72.25Hybrid round Rough 13.5 143.13

Note: GFRP5glass fiber-reinforced concrete.

Table 2. Mechanical Properties of Fiber-Reinforced Polymer~FRP!Bars

Property GFRP CFRP AFRP HFRP

Young’s modulus~MPa!

45,000 115,000 67,000 51,000

Tensile strength~MPa!

.1,000 .1,500 .1,500 .1,000

Note: GFRP5glass fiber-reinforced polymer; CFRP5carbon fiberreinforced polymer; AFRP5aramid fiber-reinforced polymer; and HFR
5hybrid fiber-reinforced polymer.
JOURNAL OF COMP


t

length. Hence, the elongation of the length of the bar,l a , from thetransducer support point to the level of the bonded bar~see Fig. 3!is subtracted from the measurements. The bond-slip displacat the unloaded end of the bar is obtained directly from themeasurement of the bottom LVDT.

The average bond stress,tav , at any stage during loadingthe recorded pullout load on the bar,F, divided by the nominasurface area of the embedment length,L, of the bar. For a circulabar diameterd, this is given by the relationship

tav5F/~pdL! (1)

The maximum average bond value,t* , is obtained by Eq.~1!when the pullout load reaches its maximum value (Fmax) duringthe test.

Experimental Results

The results of 131 cube tests are used to evaluate the bonhavior of FRP reinforcing bars to concrete. A summary ofexperimental results is presented in the Appendix. The gecoding notation applied to the rough surface bars embeddconcrete cubes is as follows:1. The first number of the code indicates the concrete

compressive strength in MPa;2. The first letter denotes the kind of reinforcing bar used in

test@G for glass fiber-reinforced polymer~GFRP! bar, C forcarbon fiber-reinforced polymer~CFRP!, A for aramid fiberreinforced polymer~AFRP!, and H for hybrid#;

3. The next letter denotes the type of the bar cross section~r forround and s for square!;

4. The second number indicates the ratio of embedment lto the bar diameter~applies only in round bars!; and

5. The last letter denotes the size of the bar diameter~D for13.5 mm, d for 8 or 8.5 mm andd for 10.5 mm!.

For example, 45Gr8D designates a 13.5 mm GFRP round bain concrete with a compressive strength of 45 MPa and havinembedment length equal to 8 times the bar diameter.

Bond Stress versus Slip Curve

Fig. 4 shows typical plots of bond stress versus slip of GFRPCFRP reinforcing bars. As seen from the curves, the loaded

Fig. 3. Pullout test arrangement

slips almost at the beginning of loading, as soon as the chemical


04.8:173-181.

sis-roundr thisesti-

ments, it isbarss canmanye

t wasn be-ationthe

ver-signifibondit ap-creteadheesion

un-tressondbondend

onar re-

RPo the

partther

topeci-rs be-r testsovecus-

ignedidedh theird onlength

ns ofilureperi-hen

uringsur-

ode.bar

Nev-to be

r thanand

f theter theon ofthatlass

ced

o barbars

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.
adhesion breaks between bar and concrete. According toCEBBulletin 151~CEB 1982!, in the case of steel bars the bond retance offered by adhesion is assumed to be rather small, a0.5–1.5 MPa. However, the bulletin does not state whethevalue is derived from experimental data or it is a theoreticalmation.
Because the required level and accuracy of slip measurefor this purpose are not determined anywhere in the literaturedifficult to determine accurately the adhesion between FRPand concrete. In addition, high accuracy in slip measurementhardly be expected at the beginning of the tests becausefactors ~such as local bending! influence the accuracy of thloaded end slip measurements at that stage. Therefore, idecided to estimate the level at which the chemical adhesiotween an FRP bar and concrete breaks, using visual examinof the bond-slip curves. This was accomplished by findingpoint, at the initial loading stage, at which the slope of the aage bond stress versus the loaded end slip curve decreasescantly. The experimental results show that the adhesivestrength seems to depend only on bar diameter; whereas,pears to be independent of the types of bar fibers and conused in the specimens. Fig. 5 shows the average value ofsion, where smaller diameter bars developed greater adhwith the surrounding concrete than larger bars.

Another important observation from the tests is that theloaded end slip remains practically zero until the bond sreaches quite high levels, compared with the ultimate bstrength. The results showed that the average ratio of thestress to the maximum bond strength when the unloadedstarted to slip is around 80% for both G and CFRP bars~Achil-lides 1998!. Although the ratio was initially assumed to dependthe elastic modulus of the bar, the results suggested no clelationship between the two quantities.

The initial slope of the loaded end bond-slip curve for CFbars is stiffer than that of GFRP bars and is clearly related t

Fig. 4. Typical bond-slip envelopes for glass fiber-reinforpolymer and carbon fiber-reinforced polymer embedments

Fig. 5. Average values of adhesive bond strength with respect ttype and diameter



-

-

difference in the elastic moduli of the two bars. However, afrom the elastic modulus, the initial slope is influenced by oparameters that will be examined later.

The residual~postmaximum! bond strength value appearsbe more than 60% of the peak bond value in most of the smens. This seemed to be an important attribute of FRP bacause the residual stress of deformed steel bars under similawas only 20 to 40% of the maximum value. However, the abvalue may be unrealistic and will be explained after the dission on the mode of bond failure.

Bond Failure Mode of Fiber-Reinforced Polymer Bars

In the described experiments, all FRP bars failed in the despull-through mode of bond failure. The concrete cubes provadequate confinement of the bars that enabled them to reacmaximum bond strength. No signs of splitting cracks appearethe cube specimens because the diameter and embedmentof the bars were relatively small compared with the dimensiothe cube specimens. However, by comparing the mode of faof FRP bars to that of steel deformed bars under similar exmental conditions, an important difference was observed. Wsufficient confinement is provided to a deformed steel bar dpullout, shear cracks develop between the bar ribs and therounding concrete before the bar fails in a pull-through mWhen this kind of failure happens, the bond strength of thedepends mainly on the strength of the surrounding concrete.ertheless, the bond strength of FRP bars does not appearcontrolled by concrete strength.

In the case of FRP bars and for concrete strengths greate30 MPa, bond failure occurs partly on the surface of the barpartly in the concrete by peeling part of the surface layer obar. Fig. 6 shows a characteristic GFRP specimen sample aftest. The cube was split after the test for a closer investigatithe actual mode of bond failure. It is obvious from this figurea white powder~consisting of crushed resin and chopped g

Fig. 6. Glass fiber-reinforced polymer specimen after test

Fig. 7. Shear failure at surface layer of fiber-reinforced polymer

H/APRIL 2004

04.8:173-181.

em-fibers

een7. InemsilurelativencreteailureFRPdam

des in

pears. Theantly

heredrepre-ilure

of thepartFig.

ction, apad by

causelt, the

zonee bareight

ltingntsdmen

nta-ental

in theults

FRPpthsofem tobondt thatulloutdgingnce.rencetwo

r in

arethe

ll berage

ber-

fiber-er

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.
fiber! is attached to the concrete cube at the location of thebedment length. In addition, the bar was scratched, and tinycould be seen on the surface of the bar by the naked eye.
The failure appears to develop at a critical interface betwsuccessive layers of fibers, as shown schematically in Fig.both cases, the shear strength between fibers and resin secontrol the bond capacity of FRP bars. The height of the fainterface from the bar axis is assumed to depend on the revalue of shear strength between fibers and resin and the coshear strength. In the current experiments, the height of the finterface of GFRP bars is assumed to be lower than that of Cbars, because the GFRP bar surface was more extensivelyaged.

For lower strength concrete~around 15 MPa!, FRP bars failein a mode similar to deformed steel bars. The concrete crushfront of the deformations of the bar, and the bond strength apto be controlled mainly by the shear strength of the concretebond strength values developed in this case were significlower than the ones developed in higher strength concrete.

Residual Bond Strength of Fiber-Reinforced PolymerBars

As a result of the type of bond failure of FRP bars in higstrength concrete (f cu.30 MPa), it is believed that the recordresidual bond stress, shown in Fig. 4, does not necessarilysent the real value of frictional stress developed at the fainterface. This is due to the fact that when the damaged partbar is slipping out of the cube during pullout the undamagedthat follows enters the embedment length zone, as shown in8, and adds additional resistance to the pullout load. This aenhances the recorded bond strength value that representsfrom the frictional stress, the additional resistance producethe wedging action of the undamaged bar.

The above phenomenon is unimportant for steel bars bebond failure happens in the surrounding concrete. As a resuunbonded part of the bar that enters the embedment lengthdoes not contribute significantly to the bond resistance of thbecause the bond failure interface is approximately at the hof the tips of bar deformations.

To quantify the value of the residual bond stress resusolely from frictional action, an additional series of experimewere conducted where some of the specimens had the embelength at the very end of the cube, as shown in Fig. 1~similar tothe RILEM pullout test!. The test arrangement and instrumetion used were similar to the ones used during the experimseries presented previously.

The results of the additional series of tests are presentedbottom half of Table 5 in the Appendix. An analysis of the res

Fig. 8. Wedging action by undamaged part of bar

showed that the residual bond stress was much lower in this case

JOURNAL OF COMP


to

-

rt

t

with the embedment length at the end of the specimen~Achillides1998!. The difference was more important in GFRP than in Cbars~Fig. 9!, and this can be attributed to differences in the deof the bond failure interface~Fig. 7!. Nevertheless, the positionthe embedment length in the concrete cube does not seinfluence the maximum bond stress developed or the initialstiffness of round FRP bars. This can be explained by the facthe unloaded end slip values recorded up to the maximum pload were less than 1 mm in all the specimens, and the weeffect could not be significantly activated in such a short distaOn the contrary, in some cases of square bars, a small diffein the maximum recorded bond values was observed for thepositions of embedment length. This will be discussed latemore detail.

Factors That Influence Bond Behavior ofFiber-Reinforced Polymer Bars

Important factors that influence bond behavior of FRP barsexamined in the following section, and proper evaluation oflevel of influence of these factors on bond development wihelpful to the formulation of equations for adequate ancholengths for FRP reinforcement.

Fig. 9. Ratio of residual to maximum bond stress for round fireinforced polymer bars

Fig. 10. Bond stress versus embedment length for glassreinforced ~GFRP! polymer and carbon fiber-reinforced polym~CFRP! round bars
,

04.8:173-181.

nt. 10.oundt lineis not

thsm-

FRPrage

(ison,dmen9andhich

sur-t thef val-rengtlowerelill re-

andbondquiteamid

it isdif-

ment

lvareorth

are95%3

es theeult ofbed-

stiff-ter forumedn the

cretelout.

entthanof a

er-

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.

Type of Bar Fiber

The maximum average bond stresst* versus the embedmelengthL developed for GFRP and CFRP bars is shown in FigThe examined specimens were 8 and 8.5 mm in diameter rbars of rough surface. By using linear regression, the best-fipassing through all data points was obtained, even though itintended to show that a linear relationship exists betweent* andembedment length. Over the small range of embedment leng~2to 10 times the diameter!, linear regression is used only for coparison purposes

From Fig. 10 it can be seen that both carbon and glassbars exhibited similar bond behavior. Their maximum avebond stress at the embedment length of 8 diametersL564 mm) was 11.9 and 12.0 MPa, respectively. By compardeformed steel bars having the same diameter and embelength developedt* equal to 16.5 MPa~for concrete strength 3MPa!. From the above figure, it can be deduced that GFRPCFRP bars developed about 72% of steel’s bond strength, wis quite remarkable considering the different nature of theirface and type of bond failure. However, it has to be noted thaabove percentage is not representative for the entire range oues of concrete strength because the influence of concrete ston bond strength of FRP and steel bars is not the same. Forconcrete strengths~around 30 MPa!, the bond strength of stebars will decrease; whereas, the FRP bar bond strength wmain practically the same.

As can be seen from Tables 3–5 in the Appendix, aramidhybrid FRP bars developed about 85% and 90% of thestrength of glass and carbon FRP, respectively, which is alsosatisfactory. It must be noted that the development of the ar

Fig. 11. Influence of embedment length ont*

Fig. 12. Influence of concrete strength ont* for fiber-reinforcedpolymer bars



t

h

bar during the EUROCRETE project was not completed, andnot expected that well-manufactured AFRP bars will behaveferently from glass or carbon FRP.

The previous experimental results appear to be in agreewith results published by other researchers~Chaallal andBenmokrane 1993; Larralde and Silva-Rodriqez 1993; Ma1995; Nanni et al. 1995; Tepfers and Karlsson 1997!, despite thfact that different types of FRP rods were used. It is also wnoting that the bond strength of epoxy-coated bars, whichmainly used as anticorrosive reinforcement, varies from 67–of that of deformed steel bars~Chaallal and Benmokrane 199!,which is comparable to the bond strength of FRP bars.

Embedment Length

An increase in embedment length, shown in Fig. 11, decreasmaximum average developed bond stress value~t* !. The sameffect is also reported for steel bars and is thought to be a resthe nonlinear distribution of bond stress on the bar. The emment length also has significant influence on the initial bondness of FRP bars. The rate of increase of bond stress is greasmaller than for larger embedment lengths. Again, this is assto be caused by the nonlinear distribution of bond stresses obar.

Concrete Strength

Results from this study show that the strength of the conaffects the actual mode of bond failure of the bar during pulFig. 12 showst* versus concrete strengths for an embedmlength of 6d. For concrete with compressive strength greater30 MPa, the bond failure interface happens at the surface

Fig. 13. Indicative distribution of normal stresses on a fibreinforced polymer bar cross section subjected to axial load

Fig. 14. Bars with different types of surface deformations

H/APRIL 2004

04.8:173-181.

th ofength

takesar is

aver-s bee

mayngth,

o de-r em-

eterwith

king

ffnessrface.thererfaceal

ic dis-lag’’

erageer inaver

-the

evelo, be-tcloses theeriora

ns onaviorof thebondme-ma-ls.

ma-

ndixFRP

ed

e bars

notFRPave aonddiedangesults

-meter.

ailedpro-reach

FRPsteel, fail-creterface

ars isto beresinan 15, andth of

ear toroundd for

ondybrid

th is

than25%

aused

wasior to

cialof theEU

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.

FRP bar. Consequently, in such concrete, the bond strengFRP bars does not depend much on the value of concrete strHowever, for lower concrete strengths~around 15 MPa!, the bondfailure mode changes. In this case, the failure interfaceplace in the concrete matrix, and the bond behavior of the bdirectly related to the concrete strength

Bar Diameter

Larger diameter bars, shown in the Appendix, develop lessage bond strength than smaller diameter bars, and, as hashown, they loose their adhesive bond earlier. Three factorsbe responsible for their lower bond strength: embedment lePoisson effect, and shear lag.

Embedment LengthLarger diameter bars require longer embedment lengths tvelop the same normal bond stress. As shown earlier, largebedment lengths reduce the average bond strength.

Poisson EffectThe Poisson effect can lead to a slight reduction in bar diamas a result of longitudinal stress. This bar reduction increasesbar size, which can lead to reduced frictional/mechanical locstresses.

Shear LagShear stiffness of FRP bars depends mainly on the shear stiof the resin and the shear strength at the resin-fiber inteWhen an FRP bar is pulled in tension through the surface,can be some differential movement between the core and sufibers, which results in a nonuniform distribution of normstresses through the cross section of the bar. A diagrammattribution of these stresses is shown in Fig. 13. This ‘‘sheareffect leads to higher surface normal stresses,sMax , which ‘‘gov-ern’’ the bond strength of the bar; whereas, the calculated avstress,sAve , is lower. The difference in these stresses is greatlarge diameter bars and is expected to reduce the estimatedage bond strength.

Cross-Sectional Shape of Bar

Square~838 mm! and round~8 or 8.5 mm diameter! cross sections were examined. By comparing the results for bars withsame embedment lengths, the square bars appeared to dsuperior bond strength values than round bars, up to 25%cause the wedging effect~reported previously! is more importanin the case of square bars because of their sharp edges. Aexamination of square bar specimens after pullout supportabove assumption because the bar edges appeared to detmore than the rest of the bar surface.

Surface Deformations

Preliminary research showed that the presence of deformatiothe surface of FRP bars plays a significant role on bond behbecause smooth bars appeared to develop only 10–20%bond stress of deformed bars. In relation to steel bars, thestrength of FRP bars is assumed to depend mainly on thechanical interlock of surface deformations and the concretetrix, rather than on the chemical adhesion of the two materia

In order to investigate the influence of the height of defor
tions on the bond strength of FRP bars, GFRP bars having differ-
JOURNAL OF COMP


.

n

-

p

r

te

ent bar deformation heights were tested. Table 4 of the Appeshows the experimental results for two types of 10.5 mm Ground bars~G24,G30! having smaller surface deformations~0.20and 0.25 mm, respectively! than the standard 8.5 mm deformbars used in this series~deformation average height 0.75 mm!. Acomparison of the average bond values developed from thesis shown in Fig. 14.

It is clear from Fig. 14 that G24 and G30 FRP bars didperform as well as the standard deformed EUROCRETE Gbar. It can be concluded that FRP bars must necessarily hminimum height of deformations to develop satisfactory bbehavior to concrete. The deformation height was not stuextensively in this research project because only a limited rof FRP bars was available for testing. However, the above reappear to agree with observations by Malvar~1995!, who suggests that surface deformations of about 5.4% of the bar diaare sufficient to provide adequate bond behavior to concrete

Conclusions

All the specimens in the experimental series of pullout tests fin a pull-through mode of failure because the concrete cubevided adequate confinement to the bars that enabled them totheir maximum bond strength. The mode of bond failure ofbars in most cases differs from the mode of bond failure ofdeformed bars. For concrete strengths greater than 30 MPaure occurs partly on the surface of the bar, not just in the conas in the case of steel bars, by peeling away part of the sulayer of the bar. Consequently, the bond strength of FRP bnot controlled as much by concrete strength but appearsinfluenced by the interlaminar shear strength just below therich surface layer of the bar. For concrete strengths less thMPa, the concrete is crushed in front of the bar deformationsthe bond strength is controlled mainly by the shear strengconcrete.

Chemical adhesion in the FRP bars and free end slip appbe correlated and they happen when the bond stress is a80% of the bond strength. This is much higher that expecteconventional steel reinforcement.

No significant difference was found between the bstrengths developed by GFRP and CFRP bars. Aramid and hdevelopment bars showed slightly lower bond strengths.

In the pull-out test, an increase in the embedment lengaccompanied by a decrease in bond strength.

Smaller diameter bars develop higher bond strengthslarger diameter bars; whereas, square bars develop up tohigher bond strengths than round bars. The wedging effect cby sharp edges seems to be responsible for this difference.

A minimum average height of deformations of 0.75 mmfound to be necessary to develop satisfactory bond behavconcrete.

Acknowledgments

The writers would like to acknowledge the technical and finanassistance offered by the Center for Cement and ConcreteUniversity of Sheffield, the EUROCRETE project, and the
TMR CONFIBRECRETE research network.

04.8:173-181.

55555

5

55

5

ion;

e

55

555

55

55

1

1

1

1

-ma-of

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.

Appendix. Experimental Results

Table 3. Experimental Results I

CodeD

~mm!L

~mm!Fmax

~kN!d1

~mm!d2

~mm!t*

~MPa!f cu

~MPa!

37Gr2DI 13.5 27 14.5 — — 12.7 3737Gr2DII 13.5 27 16.7 — — 14.6 3737Gr4DI 13.5 54 31.7 — — 13.8 3737Gr4DII 13.5 54 27.3 — — 11.9 3749Gr6DI 13.5 81 43.0 — — 12.5 4949Gr6DII 13.5 81 44.7 — — 13.0 4946Gr6DI 13.5 81 43.1 — — 12.6 4646Gr6DII 13.5 81 45.3 — — 13.2 4646Gr6DIII 13.5 81 26.9 — — 7.8 4646Gr6DIV 13.5 81 48.2 — — 14.0 4649Gr8DI 13.5 108 51.2 — — 11.2 4949Gr8DII 13.5 108 45.3 — — 9.9 4946Gr10DI 13.5 135 48.4 — — 8.5 4646Gr10DII 13.5 135 53.5 — — 9.3 4630Cr2DI 13.5 30 13.5 — — 10.6 3030Cr3DI 13.5 45 21.2 — — 11.1 3030Cr4.5DI 13.5 60 27.5 — — 10.8 3030Cr5.5DI 13.5 75 23.4 — — 7.4 3046Cr6DI 13.5 81 40.4 — — 11.8 4646Cr6DII 13.5 81 50.1 — — 14.6 4645Gr2D 13.5 27 13.4 0.58 0.57 11.7 445Gr4D 13.5 54 23.0 0.66 0.47 10.0 445Gr6D 13.5 81 41.0 0.91 0.48 11.9 445Gr8D 13.5 108 40.5 0.96 0.21 8.9 445Gr10D 13.5 135 51.9 1.03 0.32 9.1 445Cr2D 13.5 27 15.1 0.43 — 13.2 4545Cr4D 13.5 54 32.1 0.44 0.34 14.0 445Cr6D 13.5 81 30.1 0.44 0.25 8.8 4545Cr8D 13.5 108 44.8 0.55 0.32 9.8 445Cr10D 13.5 135 44.2 0.53 0.19 7.7 445Ar6D 13.5 81 34.8 0.42 0.33 10.1 4545Hr6D 13.5 81 37.3 0.72 0.23 10.9 4545Gr10d 8.5 81 16.5 0.66 0.40 8.1 445Cr10d 8 81 15.3 0.49 0.25 7.5 4545Cc 21 81 46.1 — — 8.6 4545Gsm 16 81 4.9 0.65 — 1.2 4545Hsm 8 81 2.6 0.19 0.02 1.3 4545Gs 838 81 24.5 0.99 0.41 8.9 4545Cs 838 81 25.0 0.57 0.32 9.0 4545As 838 81 14.8 — — 5.4 45

Note: D5diameter of bar/dimensions of bar’s cross sectL5embedment length;Fmax5maximum pull-out load;d15loaded endslip at Fmax; d25unloaded end slip atFmax; andt*5maximum averagbond stress.



Table 4. Experimental Results II

CodeD

~mm!L

~mm!Fmax

~kN!d1

~mm!d2

~mm!t*

~MPa!f cu

~MPa!

15Gr2D 13.5 27 3.2 0.79 0.65 2.8 1515Gr4D 13.5 54 7.1 0.44 0.42 3.1 1515Gr6D 13.5 81 6.7 0.80 0.70 1.9 1515Gr8D 13.5 108 11.5 0.67 0.47 2.5 115Gr10D 13.5 135 14.8 0.81 0.42 2.6 115Cr2D 13.5 27 4.2 0.44 0.43 3.7 1515Cr4D 13.5 54 8.0 0.48 — 3.5 1515Cr6D 13.5 81 10.8 0.30 0.16 3.1 115Cr8D 13.5 108 13.0 0.40 0.29 2.8 115Cr10D 13.5 135 12.7 0.49 0.41 2.2 115Ar6D 13.5 81 4.2 1.17 1.11 1.2 1515Hr6D 13.5 81 7.6 0.65 0.53 2.2 1515Gr10d 8.5 81 4.8 0.70 0.51 2.3 115Cr10d 8.0 81 4.4 0.43 0.28 2.2 115Ar10d 8.0 81 4.9 0.37 0.23 2.4 1515Cc 21.0 81 15.6 0.74 0.62 2.9 115Gsm 16.0 81 0.3 0.37 0.36 0.1 115Hsm 8.0 81 1.4 0.09 — 0.7 1515Gs 838 81 8.7 0.42 0.14 3.1 1515Cs 838 81 5.6 0.48 — 2.0 1515As 838 81 7.4 0.24 0.00 2.7 1541Gr6dI 8.5 48 15.6 1.27 1.16 12.2 441Gr6dII 8.5 48 11.9 1.01 0.80 9.3 4141Gr8dI 8.5 64 21.9 1.07 0.75 12.8 441Gr8dII 8.5 64 24.8 1.07 0.71 14.5 4141Gr10dI 8.5 80 25.6 1.31 0.65 12.0 441Gr10dII 8.5 80 28.8 — — 13.5 4141Cr6dI 8 48 16.9 0.43 0.37 14.0 4141Cr6dII 8 48 15.4 0.46 0.31 12.7 4141Cr8dI 8 64 22.1 0.59 0.37 13.7 4141Cr8dII 8 64 22.2 0.51 0.02 13.8 4141Cr10dI 8 80 25.7 0.73 0.41 12.8 441Cr10dII 8 80 28.7 0.53 0.34 14.3 41

41G24/6dI 10.5 60 10.4 0.24 0.96 5.3 41

41G24/6dII 10.5 60 8.0 0.23 0.02 4.0 41

41G24/8dI 10.5 80 11.0 0.39 0.04 4.2 41

41G24/8dII 10.5 80 13.3 0.51 0.16 5.0 41

41G24/10dI 10.5 100 12.4 0.37 0.08 3.8 41

41G24/10dII 10.5 100 18.7 0.55 0.13 5.7 41

41G30/6dI 10.5 60 9.2 0.27 0.04 4.7 41

41G30/6dII 10.5 60 9.1 0.26 0.06 4.6 41

41G30/8dI 10.5 80 18.3 0.60 0.18 6.9 41

41G30/8dII 10.5 80 10.6 0.37 0.10 4.0 41

41G30/10dI 10.5 100 20.2 0.51 0.05 6.1 41

41G30/10dII 10.5 100 22.0 — — 6.7 41

Note: Cc5CFRP bar with ring cross section (r out521 mm, r in

510 mm); Gsm5GFRP smooth surface bar; Hsm5hybrid smooth surface bar; G245GFRP round bar with different type of surface defortions ~24 rovings!; and G365GFRP round bar with different typesurface deformations~36 rovings!.

H/APRIL 2004

04.8:173-181.

Dtural

licte

rlicty,

ts

in-u

-

c-

rring,

ti-’ures

-orce-

ragef9.ing

, T.re-res,5.

-.

i-il-

ut

--

-lle-nne,

ntlicIn-

66

66

66

ndndear.

Dow

nloa

ded

from

asc

elib

rary

.org

by

Ista

nbul

Uni

vers

itesi

on

04/2

8/14

. Cop

yrig

ht A

SCE

. For

per

sona

l use

onl

y; a

ll ri

ghts

res

erve

d.

References

Achillides, Z. ~1998!. ‘‘Bond behaviour of FRP bars in concrete.’’ Phthesis, Centre for Cement and Concrete, Dept. of Civil and StrucEngineering, Univ. of Sheffield, Sheffield, U.K.

Table 5. Experimental Results III

CodeD

~mm!L

~mm!Fmax

~kN!d1

~mm!d2

~mm!t*

~MPa!f cu

~MPa!

39GSAI 838 64 32.1 0.88 0.67 15.7 3939GSAII 838 64 32.5 0.70 0.19 15.9 3939GSAIII 838 64 31.9 0.87 0.16 15.6 3939Gr8dI 8.5 64 20.1 0.92 0.63 11.7 3939Gr8dII 8.5 64 20.3 0.95 0.65 11.9 3939Gr8dIII 8.5 64 16.6 0.63 0.34 9.7 3939Cr8dI 8 64 19.4 0.57 0.45 12.0 3939Cr8dII 8 64 15.7 0.64 0.07 9.8 3939Cr8dIII 8 64 16.9 0.57 0.50 10.5 3939GSDI 838 64 25.5 0.96 0.37 12.5 3939GSDII 838 64 31.5 1.00 0.42 15.4 3939GSDIII 838 64 22.5 0.95 0.44 11.0 3939CSI 838 64 31.4 0.45 0.20 15.3 3939CSII 838 64 30.3 0.50 0.00 14.8 3939CSIII 838 64 27.8 0.40 0.00 13.6 3939GsI 838 64 26.6 0.84 0.64 13.0 3939GsII 838 64 26.0 0.94 0.70 12.7 3939CsI 838 64 24.6 0.57 — 12.0 3939CsII 838 64 21.9 0.60 0.39 10.7 3939GSTI 10310 81 7.5 6.80 6.20 2.4 3939GSTII 10310 81 11.0 6.65 5.10 3.4 3939Ar8dI 8 64 18.0 0.55 0.24 11.2 3939Ar8dII 8 64 15.4 0.49 0.22 9.6 39

36GSAI 838 64end 29.2 0.92 0.07 14.3 3636GSAII 838 64end 27.3 0.73 0.35 13.4 3636GSAIII 838 32end 13.5 0.18 0.28 13.3 3636GSAIV 838 32end 17.4 0.54 0.37 17.2 3636Gr4dI 8.5 32end 10.1 0.80 0.75 11.8 336Gr4dII 8.5 32end 11.8 0.51 0.40 13.8 336Gr4dIII 8.5 32mid 10.2 0.50 0.31 11.9 3636Gr4dIV 8.5 32mid 10.1 0.96 0.94 11.8 3636Cr4dI 8 32end 8.7 0.57 0.62 10.8 336Cr4dII 8 32end 10.1 0.72 0.54 12.6 336Cr4dIII 8 32mid 10.2 0.44 0.37 12.7 3636Cr4dIV 8 32mid 8.6 0.33 0.33 10.7 3636CSI 838 64end 27.2 0.50 0.36 13.3 3636CSII 838 64end 27.2 0.27 — 13.3 3636GsI 838 64end 18.8 1.13 0.49 9.3 3636GsII 838 64end 22.1 0.83 0.76 10.9 3636CsI 838 32end 10.5 0.43 0.39 10.4 3636CsII 838 32end 14.0 0.60 0.38 13.9 3636StI 8 64end 26.7 0.88 0.87 16.6 336StII 8 64end 26.5 0.29 0.78 16.5 3

Note: GSA5GFRP square~different kind of bar surface deformations aresin!; CS5CFRP square~different kind of bar surface deformations aresin!; GSD5GFRP square~different kind of resin!; GST5GFRP squarsmooth bar with twisted shape; and St5high strength steel deformed b

Achillides, Z., Pilakoutas, K., and Waldron, P.~1997a!. ‘‘Modelling of

JOURNAL OF COMP


FRP rebar bond behaviour.’’Proc., 3rd Int. Symp. on Non-Metal(FRP) Reinforcement for Concrete Structures, 14–16, Japan ConcreSociety, Sapporo, Japan, 423–430.

Achillides, Z., Pilakoutas, K., and Waldron, P.~1997b!. ‘‘Bond behaviouof FRP bars to concrete.’’Proc., 3rd Int. Symp. on Non-Metal(FRP) Reinforcement for Concrete Structures, Japan Concrete SocieSapporo, Japan, 341–348.

Cairns, J., and Abdullah, R.~1995!. ‘‘An evaluation of bond pullout tesand their relevance to structural performance.’’Struct. Eng.,73~11!,179–185.

CEB Bulletin 151. ~1982!. ‘‘Bond action and bond behaviour of reforcement.’’ State-of-the-Art Report, Comitee Euro-international dBeton Bulletin 151, Paris.

Chaallal, O., and Benmokrane, B.~1993!. ‘‘Pullout and bond of glassfibre rods embedded in concrete and cement grout.’’Mater. Struct.,Vol. 26~April !, 167–175.

Clarke, J., and Waldron, P.~1996!. ‘‘The reinforcement of concrete strutures with advanced composites.’’Struct. Eng.,74~3!, 283–288.

Duranovic, N., Pilakoutas, K., and Waldron, P.~1995!. ‘‘General testingarrangement on R.C. beams.’’Rep. No. CCC/94/0017A, Centre foCement and Concrete, Dept. of Civil and Structural EngineeUniv. of Sheffield, Sheffield, U.K.

Head, P. R.~1996!. ‘‘Advanced composites in civil engineering—A crical overview at this high interest, low use stage of development.’2ndInt. Conf. Advanced Composite Materials in Bridges and Struct,Montreal, 3–16.

Institution of Structural Engineers.~1999!. ‘‘Interim guidance on the design of reinforced concrete structures using fibre composite reinfment.’’ IstructE, The Institution of Structural Engineers, London.

Larralde, J., and Silva-Rodriguez, R.~1993!. ‘‘Bond and slip of FRPrebars in concrete.’’J. Mater. Civ. Eng.,5~1!, 30–40.

Losberg, A.~1963!. ‘‘Force transfer and stress distribution at anchoand curtailment of reinforcement.’’Rep. No. 608, Chalmers Univ. oTechnology, Dept. of Building Technology, Goteborg, Sweden, 4

Malvar, J. L.~1995!. ‘‘Tensile and bond properties of GFRP reinforcbars.’’ J. ACI Materials,92~3!, 276–285.

Nanni, A., Al-Zaharani, M., Al-Dulaijen, S., Bakis, S., and Boothby~1995!. ‘‘Bond of FRP reinforcement to concrete—Experimentalsults.’’ Non-Metallic (FRP) Reinforcement for Concrete StructuProc., 2nd Int. RILEM Symp.~FRPRC5-2!, L. Taerwe, ed., 137–14

Pilakoutas, K.~2000!. ‘‘Composites in concrete construction.’’FailureAnalysis of Industrial Composite Materials, A. Gdoutos, K. Pilakoutas, and C. Rodopoulos, eds., McGraw-Hill, New York, 449–497

Pilakoutas, K., Achillides, Z., and Waldron, P.~1997!. ‘‘Non-ferrous re-inforcement in concrete structures.’’Innovation in Composite Materals and Structures, M. B. Leeming and B. H. V. Topping, eds., CivComp Ltd, Edinburgh, Scotland, 47–58.

RILEM/CEB/FIP. ~1978!. ‘‘Bond test for reinforcing steel 2: Pullotest.’’ Recommendation RC6, RILEM/CEB/FIP, Bagneux, France.

Rostasy, F. S.~1996!. ‘‘FRP: The European perspective.’’Fiber Composites in Infrastructure, 1st Int. Conf. in Infrastructure, H. Saadatmanesh and M. Ehsani, eds., 12–20.

Task Group Bond Models.~2000!. ‘‘Bond of non-metallic reinforcement.’’ Chapter 7,Bond of Reinforcement in Concrete, Rep. fib butin 10, International Federation for Structural Concrete, LausaSwitzerland, 315–394.

Tepfers, R., and Karlsson, M.~1997!. ‘‘Pull-out and tensile reinforcemesplice tests using FRP C-bars.’’Proc., 3rd Int. Symp. on Non-Metal(FRP) Reinforcement for Concrete Structures, Japanese Concrete
stitute, Sapporo, Japan, Vol. 2, 357–364.

04.8:173-181.

Bond Behavior of Fiber Reinforced Polymer Bars under Direct Pullout Conditions

Documents

Transcript of Bond Behavior of Fiber Reinforced Polymer Bars under Direct Pullout Conditions