Structural Analysis of Historic Construction – D’Ayala & Fodde (eds)© 2008 Taylor & Francis Group, London, ISBN 978-0-415-46872-5

Bond behaviour of CFRP and GFRP laminates on brick masonry

M. Panizza, E. Garbin, M.R. Valluzzi & C. ModenaDepartment of Structural and Transportation Engineering, University of Padova, Italy

ABSTRACT: In the last decade, innovative technologies have been developed using Fiber Reinforced Polymer(FRP) as strengthening and repair of masonry structures. Bond of FRP to substrate is crucial for the effectivenessof the technique especially to masonry substrate, which can have a wide variability. Few contributions are availableconcerning debonding problems on masonry. In this paper the main results of an experimental campaign on thelocal behaviour of externally bonded FRP, applied on clay bricks, are presented. Double-lap Push-pull ShearTestshave been performed by using carbon and glass fiber reinforcement. Experimental results, in terms of failure load,have been compared with predictive bond-strength models proposed in literature, mainly available for concrete.Based on the measured strength, interface fracture energy has been calibrated. A simplified analytical model,fitted on the experimental data, has been proposed as bond-slip law. Finally, a bilinear function, as commonlyadopted by some guidelines, has been calibrated.

1 INTRODUCTION

The application of externally bonded textiles is adeveloping technique for strengthening of masonrystructures. The bond between these products and sup-port is a crucial aspect to clarify, as it strongly influencethe effectiveness of the intervention.

In the last decade, bond of composite laminateson concrete substrate has been deeply investigated.The characterization of bond behaviour has been per-formed by means of different test set up. The mostcommonly used are: Single-lap Shear Test (Chajeset al. 1996, Täljsten 1997, Bizindavyi & Neale 1999),Double-lap Pull-pull Shear Test (Lee et al. 1999,Nakaba et al. 2001), Double-lap Push-pull Shear Test(Camli & Binici 2007) and Beam-typeTest (De Loren-zis et al. 2001). Wide reviews of available strength orbond-slip models were given by Chen & Teng (2001),Lu et al. (2005) and Karbhari et al. (2006).

On the other hand, few investigations concerningdebonding on masonry substrate are available, such asAiello et al. (2005) that investigated bond on naturalstones, and Briccoli Bati et al. (2007), that tested bondon solid clay bricks.

They adopted Double-lap Push-pull Shear Tests,also known in literature by different names, suchas Double-shear Push Test or Near-end SupportedDouble-shear Test (Yao et al. 2004). It consists inloading in tension two reinforcement strips, symmetri-cally connected to the support, in order to create shearstresses at the interface; the brittle support is subjectedto compressive stresses.

This set-up is based on the assumption that theapplied load is equally distributed on the two strips,but it is also particularly simple and suitable for theusual common available test machine.

This experimental procedure has been adopted forthe present work, that aims at giving a contribution tothe characterization of the bond behaviour of carbonand glass textiles externally bonded to clay substrate.

The results of five samples for high-strength car-bon reinforcement and five samples for alkali-resistantglass reinforcement are presented and discussed.The predictions of twenty-one bond-strength models,available in literature for concrete as parent material,have been compared with the measured strength. Thefracture energy of the composite-clay interface hasbeen evaluated through the experimental failure loads.A simplified bond-slip law has been proposed on thebasis of the data obtained from load and strain moni-tored during the tests; furthermore, a bilinear functionhas been also calibrated.

2 EXPERIMENTAL TESTS DESCRIPTION

2.1 Materials characterization

Solid clay bricks (nominal dimension 250 × 120 ×55 mm) were used as substrate, and the MBrace©

Wet lay-up System as reinforcement. High strengthcarbon fibers (CFRP) on five specimens and alkali-resistant glass fibers (GFRP) on other five specimenswere used.

763

Table 1. Bricks mechanical properties.

Mean cubic compressive strength 50.94 MPaMean direct tensile strength 2.37 MPaMean splitting tensile strength 3.99 MPaMean flexural tensile strength 5.46 MPaSecant elastic modulus 16,100 MPa

Table 2. Reinforcement components properties.

Adhesive MBrace© SaturantCharacteristic compressive strength >80 MPaCharacteristic direct tensile strength >50 MPaMaximum tensile strain 2.5%Tensile elastic modulus >3000 MPa

High-strength Carbon MBrace© C1-30Equivalent thickness 0.165 mmCharacteristic direct tensile strength 3430 MPaMaximum tensile strain 1.5%Tensile elastic modulus 230,000 MPa

Alkali-resistant Glass MBrace© G60-AREquivalent thickness 0.230 mmCharacteristic direct tensile strength 1700 MPaMaximum tensile strain 2.8%Tensile elastic modulus 65,000 MPa

The main bricks properties are summarized inTable 1 (Cartolaro 2004). Reinforcement systemproperties, obtained from producers datasheets, arereported in Table 2.

2.2 Experimental test set-up

The specimen was made by a single clay brick with twostrips of reinforcement externally bonded (wet lay-upsystem), symmetrically applied on the opposite widersurfaces. Each strip was 50 mm wide and bonded tothe brick for 200 mm (Fig. 1).

An unbonded length from the limit of the brick,equal to 30 mm, was imposed next to the loaded end,in order to minimize edge effects.

An universal mechanical press, Galdabini Sun60(maximum load 600 kN), was used as test machine(Fig. 2).

Each strip of reinforcement, made by a single layerof fibers and two layers of epoxy resin, was bonded,at the loaded end, to a steel support connected to thetest machine. Brick was connected to the machinethrough a steel frame, made by two plates linked bybolts (Fig. 2).

The load was applied axially and tests were con-trolled by a displacement rate of 0.2 mm/minute.

On the outer side of one of the two reinforcementstrip for each specimen, seven strain gauges wereapplied, distributed as follows: one on the unbondedzone, next to the loaded end of the reinforcement, and

Figure 1. Geometry of specimens.

Figure 2. Test machine (left) and a specimen ready fortesting (right).

Figure 3. Distribution of the strain-gauges.

six on the bonded one. To optimize the number ofinstruments and to monitor the whole bonded region,the strain-gauges were not equally distributed, but lessspaced near the loaded end (Fig. 3).

3 TEST RESULTS AND DISCUSSION

After test, all specimens revealed the complete detach-ment of the reinforcement from the support. The fail-ure involved the brick surface (Fig. 4), where curvedcracks and ripping of clay pieces were observed.

The failure loads, Pu, are given Tables 3–4 and plot-ted in Figure 5. It is possible to note as specimensstrengthened with CFRP showed better performancesthan GFRP one. The mean failure load in the first casewas around 35% higher than the latter.

764

Figure 4. Brick and composite surfaces after the test.

Table 3. Experimental results for carbon reinforcement.

Ef Pu Pu/2b′f σu

Specimen MPa N N/mm MPa

ShC1 164,419 31,884 318.8 1932ShC2 336,439 34,233 342.3 2075ShC3 284,991 35,325 353.3 2141ShC4 277,511 39,210 392.1 2376ShC5 338,456 40,301 403.0 2442

Mean value 280,363 36,191 361.9 2193Stand. dev. 70,696 3505COV 25.2% 9.7%

Table 4. Experimental results for glass reinforcement.

Ef Pu Pu/2b′f σu

Specimen MPa N N/mm MPa

ShG1 50,934 23,380 233.8 1017ShG2 87,014 27,940 279.4 1215ShG3 80,545 27,300 273.0 1187ShG4 102,598 26,400 264.0 1148ShG5 84,842 28,360 283.6 1233

Mean value 81,035 26,676 266.9 1160Stand. dev. 18,817 1985COV 23.2% 7.4%

It was assumed that strain is uniform on the com-posite cross section, and that is possible to refermechanical properties to the dry woven: this approachis accepted by many authors (Chen & Teng 2001)and by some guidelines, such as Italian CNR-DT200/2004.

Therefore, it was possible to calculate the nominaltensile stress on the textile, defined as the load andcross section area ratio. By coupling the strain mea-sures (obtained from the strain-gauge on the unbondedregion) with the load values, and by assuming the linear

Figure 5. Experimental failure loads per unit width.

elastic behaviour of the composite, it was also possibleto evaluate Young’s modulus of the reinforcement foreach sample through a best fitting (Eq. 1):

where σ = nominal tensile stress, P = axial loadon the composite strips, b′

f = single strip width,t = equivalent textile thickness, ε = axial strain mea-sured by the strain-gauge and Ef = composite elasticmodulus.

Tables 3–4 also report the composite elastic mod-ulus values, Ef , and the nominal tensile stresses atfailure, σu. The experimental mean elastic moduliresult higher than the producer’s values (22% for car-bon reinforcement and 25% for glass one). The meanmaximum stress reached by reinforcement is around64% of tensile strength for carbon, and 68% for glass.

As the reinforcement axial stiffness per unit width,Ef tf , was known for each specimen, failure loads perunit width, P/2b′

f , were tabulated versus the axial stiff-ness. Trend lines were fitted, referring to all data or toeach set (carbon and glass). The expression adoptedfor the trend lines is given in Equation 2:

where c1 e c2 are regression constants, which valuesare reported in Table 5. It can be observed that thefitting of all data shows a better correlation than thefitting of each single set.

By assuming the relationship between the failureload and the square root of the axial stiffness per unitwidth (Eq. 7), it was possible to reduce the numberof free parameters in Equation 2, by imposing theexponent value c2 equal to 0.5. Results (carbon dataset, glass data set and all data) are given in Table 5,whereas Figures 6–7 compare the trend lines with theexperimental data. Regression coefficient for GFRPare slightly higher than CFRP (around 16%) and this

765

Table 5. Regression constants of load vs axial stiffness(both per unit width) trend lines.

Data set c1 c2 R2

All data 12.876 0.310 0.876Glass data only 27.197 0.234 0.622Carbon data only 35.063 0.218 0.439All data (Square Root) 1.759 0.5 n.a.Glass data (Square Root) 1.953 0.5 n.a.Carbon data (Square Root) 1.681 0.5 n.a.

Figure 6. Failure loads per unit width versus reinforcementaxial stiffness per unit width: experimental data and trendlines.

Figure 7. Failure loads per unit width versus reinforcementaxial stiffness per unit width: experimental data and trendlines based on the axial stiffness square root.

could be significant for the fracture energy evaluationas given in the following.

4 PREDICTION OF STRENGTH

Many predictive models have been developed to esti-mate the failure load of the composite-to-concretebonded joint. Twenty-one of them were applied inthis study, in order to make a comparison with theexperimental results of the tests on clay substrate.

Among them, the models of Tanaka, Hiroyuki andWu, Maeda, Khalifa (reported in Chen & Teng 2001),

Table 6. Predictions of failure load.

CFRP GFRP

Model Pu/2b′f Error Pu/2b′

f ErrorN/mm N/mm

Tanaka 166 −54.0% 166 −37.6%Hiroyuki and Wu 158 −56.2% 158 −40.6%Maeda 254 −29.7% 174 −34.9%Khalifa 248 −31.4% 170 −36.2%Yang 192 −47.0% 143 −46.3%Sato 415 +14.6% 116 −56.5%Iso 271 −25.1% 161 −39.5%

Izumo 557 +53.9% 224 −15.9%Neubauer and R. 283 −21.8% 179 −32.7%Chen and Teng 245 −32.2% 156 −41.6%

Monti et al. 321 −11.3% 204 −23.6%Lu et al. Bilinear 220 −39.3% 139 −47.7%Brosens and V. G. 359 −0.9% 228 −14.7%CNR−DT 200 263 −27.4% 167 −37.5%

Nakaba et al. 350 −3.3% 222 −16.7%Savoia et al. 328 −9.4% 208 −22.0%

Neubauer and R. 266 −26.4% 169 −36.6%Dai and Ueda (1) 326 −9.9% 207 −22.5%Dai and Ueda (2) 322 −11.0% 202 −24.2%Lu et al. Precise 220 −39.3% 139 −47.7%Lu et al. Simplif. 220 −39.3% 139 −47.7%Mean experim. 362 – 267 –

Yang, Sato, Iso (reported in Lu et al. 2005), expressfailure load as the product of an area and a nominalaverage tangential stress, τu (Eq. 3).

where bf is the reinforcement width, Le is an effec-tive length and τu a tangential stress, different frommodel to model. The models of Izumo (Lu et al. 2005),Neubauer and Rostàsy, Chen and Teng (both reportedin Chen & Teng 2001) give other expressions for thefailure load. Finally, eleven ones provide an estima-tion of the fracture energy value Gf , that has beencorrelated with failure load through Equation 7. In par-ticular, the models of Monti (Lu et al. 2005), Lu et al.(2004), herein Lu Bilinear, Brosens and Van Gemert(Karbhari et al. 2006) and Italian Research Council(CNR-DT 200/2004) utilize a bilinear bond-slip law;the models of Nakaba et al. (2001) and Savoia et al.(2003a) adopt a Popovics curve as bond-slip law; themodels of Neubauer and Rostàsy (Lu et al. 2005),Dai & Ueda (2003), Dai et al. (2005) and Lu et al.(2004), Precise and Simplified models, take on othertypes of bond-slip function.

The results are given in Table 7 and shown inFigure 8. It can be seen that all the predictions, exceptfor the models of Sato and Izumo in case of carbon

766

Figure 8. Ratio of predicted failure loads vs mean experi-mental value for carbon and glass reinforcement.

Table 7. Evaluation of fracture energy.

Gf from Eq. 7 Gf from S.R. fitReinforcement type N/mm N/mm

Carbon fibers 1.42 1.41Glass fibers 1.91 1.91

reinforcement, underestimate the mean experimentalfailure load. Moreover, all formulations provide anestimation closer to test results in case of CFRP thanGFRP, except for the Tanaka and Hiroyuki models.However, results show large differences from modelto model: they vary between 44% and 154% of exper-imental mean failure load, for carbon reinforcement,and between 43% and 85%, for glass one.

5 FRACTURE ENERGY CALIBRATION

The interface fracture energy mode II, Gf , is definedby Equation 4 as the definite integral of the tangentialstress, τ, expressed as function of the mutual slip ofcomposite and substrate, s:

One of the first analytical models of the concrete-composite bond strength was derived by Täljsten(1996), starting both from a linear approach, basedon the beam theory, and from a non-linear approach,related to fracture mechanics. In case of most ofthe epoxy adhesives commonly used, a simplifiedformulation, as reported in Equation 5, was obtained.

where bf = reinforcement width, Gf = interface frac-ture energy, αT = constant value, Ectc = axial stiffnessper unit width of the concrete substrate.

Yuan (Chen &Teng 2001) proposed a modified con-stant value (Eq. 6) that takes into account the width (bfand bc) ratio of the bonded materials:

In most cases, the constant value αT, or αW, has a slightinfluence on the calculation. Many authors, such asSavoia et al. (2003a) and Dai et al. (2005), report thefollowing formula (Eq. 7) without any constant:

By applying these formulas to the experimental data ofthis work, it emerges that taking or not into account theparameters αT or αW leads to a difference lower than2%. It has to be noticed that Equation 7, demonstratedin some cases (Wu et al. 2002; Dai et al 2005), isassumed in every case of regular interface law (Savoiaet al. 2003b).

It is considered significant also for the clay substrateadopted in the present work.

Accordingly with Equation 7, it was possible to cal-ibrate fracture energy Gf through the mean values offailure load and elastic modulus; results are reported inTable 6. The estimated value, for glass reinforcement,is around 35% higher than carbon one.

Moreover, the fitting parameter c2 given in Table 5,where c1 was imposed equal to 0.5 (Square Rootbased fitting), allowed to evaluate Gf , as shown inEquation 8. The results (Table 6) show no signifi-cant difference from values obtained by means ofEquation 7.

6 CALIBRATION OF A BOND-SLIP LAW

Figure 9, from 9a to 9e, shows possible shapesof the bond-slip function: (a) cut-off, adopted byNeubauer and Rostàsy (Chen & Teng 2001); (b) bilin-ear, assumed by some guidelines like fib Bulletin14 (2001) and CNR-DT 200 (2004), and by Monti(Lu et al. 2005), Brosens and Van Gemert (Karb-hari et al. 2006) and Lu (Lu et al. 2004); (c) rigidwith linear softening, by Chen & Teng (2001); (d)a single function, as the Popovics curve chosen bySavoia et al. (2003a) and Nakaba et al. (2001), oran exponential curve obtained by Dai et al. (2005);(e) two different non-linear functions for ascendingand descending branch, for instance the expressionsadopted by Lu et al. (2004) or Dai & Ueda (2003).

767

Figure 9. Some bond-slip law shapes available in literature.

Therefore, it is commonly assumed that bond ofcomposite laminates exhibits a softening behaviour,with an ascending branch followed by a descendingone, and presenting no residual stress for wider slip.

To calibrate the bond-slip law on the experimentalresults, this combined approach was adopted: tan-gential stress and interface slip points (τ–s) wereobtained from strain-gauges monitoring, while thefracture energy value, Gf , was calculated from failureloads trough Equation 7.

Fracture energy represents a restraint for the bond-slip function (Eq. 4) and allows to reduce the numberof free parameters involved in the calibration process.

Equations 9–11 briefly report the main relations(obtained from simple equilibrium and compatibil-ity considerations) between reinforcement strain ε,interface tangential stress τ and slip s, supposing todisregard the slip component of the substrate, suffi-ciently stiffer than composite. The notation x indicatesthe coordinate along the central axis of the bondedregion.

To calculate, from strain measured in discrete positionsalong the reinforcement, the corresponding tangentialstress and slip values, Equations 9–10 were modified.

In the present work, the discrete formulas given inEquations 12–13 (Valluzzi et al. 2003) were used; theyallow to manipulate data from devices not uniformlyspaced.

Figure 10. Calibrated bond-slip laws (CFRP).

Figure 11. Calibrated bond-slip laws (GFRP).

where the notation i-nth indicates the strain-gaugeposition. The orientation of the x axis makes i increas-ing from loaded end (x = 0) to free end (x = 200 mm).

Hence, it was possible to couple slip values with thecorresponding tangential stresses.

As above explained, it is assumed that the bond-sliplaw should show an ascending segment and a softeningbehaviour. Instead of using two different mathemati-cal expressions for the ascending and the descendingbranch, a single function was chosen. Although therecould be a slight loss of adherence to experimentaldata, it reduces the required parameters making easierthe fitting process.

The proposed law, easy to integrate and derive, isgiven in Equation 14.

where A and B are regression constants, τ is theinterface tangential stress and s the composite slip.

Applying the calibrated fracture energy value, it waspossible to have a fitting function that depends on justone parameter, as shown in Equation 15.

768

It is useful to rewrite the law, herein labeled UniPdcurve, in a normalized form (Eq. 16).

where s0 = 1 / B and τmax = τ (s0) are the coordinatesof the point of maximum tangential stress.

After the optimization of the Unipd curves, in caseof carbon reinforcement and glass one, it was possibleto calibrate a bilinear law, whose analytical form isreported in Equation 17.

where sf is the ultimate strain, related to null stress.This form is commonly proposed by some guidelines(fib Bulletin 14 2001; CNR-DT 200 2004).

Since the bilinear function depends on more param-eters, to reduce them the peak tangential stress value,obtained from the fitted Unipd curve, and the cali-brated fracture energy were imposed.

Figures 10–11 show the optimized curves and theexperimental stress-slip data. It can be noticed that car-bon reinforcement interface seems to be slight stifferthan glass one.

Tables 8–9 report the significant values (fractureenergy, peak tangential stress with related slip, andultimate slip) calculated by the fitting process; theyhave been compared with the values estimated throughthe eleven models based on the fracture energy predic-tion, mentioned in chapter 4. Estimated values varyinto a quite wide range.

It has to be noticed that not all models providesignificant differences from carbon to glass reinforce-ment; in particular, the difference between the fractureenergy values experimentally calibrated, in case ofCFRP and GFRP, is not in agreement with mostpredictions.

7 CONCLUSIONS

The bond behaviour of the composite-clay brick inter-face has been investigated by means of Double-lapPush-pull Shear Tests, for both high-strength carbon(CFRP) and alkali-resistant glass (GFRP) reinforce-ment.

Far from being exhaustive, the results show a betterperformance of carbon reinforcement than glass one,around 36% higher in the first case.

The experimental strength has been compared withtwenty-one predictive models developed for concrete

Table 8. Significant values for local bond of CFRP.

Gf τmax s0 sfCurve N/mm MPa mm mm

UniPd fitting 1.42 7.22 0.072 –Bilinear fitting 1.42 7.22 0.034 0.392

Monti et al. 1.11 5.37 0.046 0.415Lu et al. Bilinear 0.52 3.73 0.048 0.280Brosens and V. G. 1.39 2.71 0.012 1.025CNR 0.75 7.46 0.056 0.200Nakaba et al. 1.32 7.08 0.065 –Savoia et al. 1.16 7.08 0.051 –Neubauer and R. 0.77 5.69 0.270 –Dai and Ueda (1) 1.15 8.58 0.103 –Dai and Ueda (2) 1.12 6.41 0.061 –Lu et al. Precise 0.52 3.73 0.054 –Lu et al. Simplif. 0.52 3.73 0.048 –

Table 9. Significant values for local bond of GFRP.

Gf τmax s0 sfCurve N/mm MPa mm mm

UniPd fitting 1.91 6.33 0.111 –Bilinear fitting 1.91 6.33 0.048 0.603

Monti et al. 1.11 5.37 0.046 0.415Lu et al. Bilinear 0.52 3.73 0.048 0.280Brosens and V. G. 1.39 2.71 0.012 1.025CNR 0.75 7.46 0.056 0.200Nakaba et al. 1.32 7.08 0.065 –Savoia et al. 1.16 7.08 0.051 –Neubauer and R. 0.77 5.69 0.270 –Dai and Ueda (1) 1.15 7.10 0.107 –Dai and Ueda (2) 1.10 5.69 0.067 –Lu et al. Precise 0.52 3.73 0.054 –Lu et al. Simplif. 0.52 3.73 0.048 –

substrate. All predictions, except two in case of CFRP,underestimate the results of the tests; the models,except two in case of GFRP, seem to work better forthe carbon reinforcement. However, the strength pre-dictions vary into an wide range (between 44% and154% of experimental mean failure load for CFRP,43% and 85% for GFRP).

From the measured failure loads, different fractureenergy values have been derived, around 35% higherin case of glass reinforcement than carbon one.

To analyze stress and slip from strain-gauges mea-surement, when instruments are not uniformly spaced,discrete equations have been used, consistent withcentral finite difference methods.

Moreover, a mathematical function easy to integrateand derive is proposed as bond-slip law. This functionhas been fitted in case of both carbon and glass rein-forcement; beside these fittings, two bilinear functions

769

have been also calibrated. The optimized functionsseem to show an interface local behaviour of CFRPslightly stiffer than GFRP.

However, further investigations are needed, to betterinquiry the bond behaviour of composite-clay bricksinterface; this is a first step in order to take intoaccount, in the future, the role of the mortar joints,characteristic of masonry structures.

Moreover, the reliability of the experimental set-upneeds to be verified; despite of its simplicity, the actualdistribution of the load should be more clarified, as itcould cause undesired bending moments inside the testspecimen.

ACKNOWLEDGEMENTS

The authors acknowledge Andrea Cartolaro and thetechnical staff of the Laboratory of Material Testingof the Department of Construction and Transportationof the University of Padova, where tests have beenperformed.

This activity has been partially supported by theNational Italian Project ReLUIS.

All materials, including fibers and adhesion system,have been supplied by Modern Advanced Concrete(MAC S.p.A.) of Treviso, Italy.

REFERENCES

Aiello, M.A. & Sciolti, M.S. 2005. Bond analysis of masonrystructures strengthened with CFRP sheets. Constructionand Building Materials, 20: 90–100.

Bizindavyi, L. & Neale, K.W. 1999. Transfer Lengths andBond Strengths for Composites Bonded to Concrete.ASCE Journal of Composites for Construction, 3 (4):153–160.

Briccoli Bati, S., Rovero, L., Tonietti, U. 2007. Adesione frablocchi in laterizio e rinforzo in CFRP (in Italian). Materi-ali ed Approcci Innovativi per il Progetto in Zona Sismicae la Mitigazione della Vulnerabilità delle Strutture; proc.nat. workshop ReLUIS, Salerno, Italy.

Camli, U.S. & Binici, B. 2007. Strength of carbon fiberreinforced polymers bonded to concrete and masonry.Construction and Building Materials, 21: 1431–1446.

Cartolaro, A. 2004. Volte in muratura di mattoni rinforzatecon FRP. Studio sperimentale dei meccanismi resistentialle interfacce (in Italian). MS Dissertation, University ofPadova, Italy.

Chajes, M.J., Finch, W.W.Jr., Januszka, T.F. & Thomson, T.A.1996. Bond and Force Transfer of Composite MaterialPlates Bonded to Concrete. ACI Structural Journal, 93(2): 295–303.

Chen, J.F. & Teng, J.G. 2001. Anchorage Strength Models forFRP and Steel Plates Bonded to Concrete. ASCE Journalof Structural Engineering, 127 (7): 784–791.

CNR-DT 200/2004. Guide for the Design and Construc-tion of Externally Bonded FRP Systems for StrengtheningExisting Structures. National Research Council, Italy.

Dai, J.G. & Ueda,T. 2003. Local Bond stress Slip relations forFRP Sheets-Concrete Interfaces. Fibre-Reinforced Poly-mer Reinforcement for Concrete Structures FRPRCS-6;proc. int. symp., Singapore.

Dai, J.G., Ueda, T. & Sato, Y. 2005. Development of theNonlinear Bond Stress-Slip Model of Fiber ReinforcedPlastics Sheet-Concrete Interfaces with a Simple Method.ASCE J. of Composites for Construction, 9 (1): 52–62.

De Lorenzis, L., Miller, B. & Nanni, A. 2001. Bond ofFRP laminates to concrete. ACI Materials Journal, 98 (3):256–264.

fib Bulletin 14 2001. Externally bonded FRP reinforcementfor RC structures – Technical Report on the design anduse of externally bonded fibre reinforced polymer rein-forcement for reinforced concrete structures. FédérationInternationale du Béton.

Karbhari, V.M., Niu, H. & Sikorsky, C. 2006. Review andComparison of Fracture Mechanics-based Bond StrengthModels for FRP-strengthened Structures. J. of ReinforcedPlastics and Composites, 25 (17): 1757–1794.

Lee, Y.J., Boothby. T.E., Bakis. C.E. & Nanni. A. 1999. SlipModulus of FRP Sheets Bonded to Concrete. ASCE J. ofComposites for Construction, 3(4): 161–167.

Lu, X.Z., Teng, J.G., Ye, L.P. & Jiang, J.J. 2004. Bond-slip models for FRP sheet/plate-to-concrete interfaces.Advanced Polymer Composites for StructuralApplicationsin Construction – ACIC 2004; proc. 2nd int. conf., Surrey,UK.

Lu, X.Z., Teng, J.G., Ye, L.P. & Jiang, J.J. 2005. Bond-slip models for FRP sheets-plates bonded to concrete.Engineering Structures, 27: 920–937.

Nakaba, K., Kanakubo, T., Furuta, T., Yoshizawa, H. 2001.Bond Behavior between Fiber-Reinforced Polymer Lam-inates and Concrete. ACI Structural Journal, 98 (3):359–367.

Savoia, M., Ferracuti, B. & Mazzotti, C. 2003a. Nonlinear bond-slip law for FRP-concrete interface. Fibre-Reinforced Polymer Reinforcement for ConcreteStructures FRPRCS-6; proc. int. symp., Singapore.

Savoia, M., Ferracuti, B. & Mazzotti, C. 2003b. Una legge diinterfaccia non lineare per placcaggi con lamine in FRP(in Italian). 16th AIMETA Congress of Theoretical andApplied Mechanics; proc., Ferrara, Italy.

Täljsten, B. 1996. Strengthening of concrete prisms using theplate bonding technique. Int. J. of Fracture, 82: 253–266.

Täljsten, B. 1997. Defining anchor lengths of steel andCFRP plates bonded to concrete. Int. J. of Adhesion andAdhesives, 19: 319–327.

Valluzzi, M.R., Tinazzi, D., Garbin, E. & Modena, C. 2003.FEM modelling of CFRP strips bond behaviour for bedjoints reinforcement techniques. Computer Methods inStructural Masonry; proc. 6th int. conf., Rome, Italy.

Wu, Z.,Yuan, H. & Niu, H. 2002. StressTransfer and FracturePropagation in Different Kinds of Adhesive Joints. ASCEJournal of Engineering Mechanics, 128 (5): 562–573.

Yao, J., Teng, J.G. & Chen, J.F. 2004. Experimental study onFRP-to-concrete bonded joints. Composites: Part B, 36:99–113.

770