Accelerated ageing behaviour of the adhesive bond between concrete specimens and CFRP overlays

Construction and Building Materials 25 (2011) 523–538

Contents lists available at ScienceDirect

Construction and Building Materials journal homepage: www.elsevier.com/locate/conbuildmat

Accelerated ageing behaviour of the adhesive bond between concrete specimens and CFRP overlays Karim Benzarti a,⇑, Sylvain Chataigner b, Marc Quiertant a, Céline Marty a, Christophe Aubagnac b a b

Université Paris Est, Laboratoire Central des Ponts et Chaussées (LCPC), 58 Bld Lefebvre, 75732 Paris Cedex 15, France LRPC d’Autun, Boulevard de l’Industrie, BP 141, 71 405 Autun, France

a r t i c l e

i n f o

Article history: Available online 17 September 2010 Keywords: Concrete infrastructure Composite Adhesive bond Epoxy Hydrothermal ageing Moisture Pull-off Shear test Plasticization Glass transition

a b s t r a c t In this paper, the durability of the adhesive bond between concrete and carbon fibre reinforced polymers (CFRP) strengthening systems has been investigated under accelerated ageing conditions, i.e., at 40 °C and 95% relative humidity. Mechanical characterizations were carried-out on control and exposed CFRP strengthened concrete specimens, in order to assess the evolutions of the adhesive bond properties during hydrothermal ageing. Results from different experimental campaigns are presented and reveal significant evolutions (decrease in the adhesive bond strength and/or change in the failure mode) depending on various parameters, such as the surface preparation of concrete, the presence of a carbonated concrete layer, the nature of the CFRP overlay (carbon fibre sheets or pultruded CFRP plates), the ageing behaviour of the bulk epoxy adhesive itself, or the test configuration used to evaluate the adhesive bond strength (pull-off or shear loading test). Moisture diffusion from the superficial layer of concrete (i.e., diffusion of interstitial pore solution) towards the adhesive joint is suspected to be a key factor driving the degradation process during hydrothermal ageing. Ó 2010 Elsevier Ltd. All rights reserved.

1. Introduction Externally bonded composites systems are commonly used worldwide for the strengthening of civil structures [1–3]. In France, for instance, carbon fibre reinforced polymers (CFRP) are increasingly being considered for the rehabilitation and upgrading of deteriorating or under-strength infrastructures [4]. If the technique effectiveness has been clearly demonstrated and is now accepted by all stakeholders in the construction sector, the long term durability of the adhesive bond between the FRP and concrete is still a crucial issue [5]. Environmental factors, such as moisture and temperature cycles, exposure to marine or other aggressive environments, may indeed affect the integrity of both the interfacial bonds (substrate/polymer interactions) and the polymer material itself (adhesive joint in the case of bonded composite plates, or polymer matrix in the case of laminates made from impregnated fabrics according to the wet lay-up process), leading to a decrease in the overall performances of the composite overlay. At the moment, such long term effects are taken into account in the design guidelines either by introducing semi-empirical safety coefficients on tensile properties of FRP reinforcements and shear characteristics of bonded interfaces [6,7], or through reliabilitybased approaches [8]. ⇑ Corresponding author. Tel.: +33 1 40 43 52 51; fax: +33 1 40 43 65 14. E-mail addresses: [email protected], [email protected] (K. Benzarti). 0950-0618/$ - see front matter Ó 2010 Elsevier Ltd. All rights reserved. doi:10.1016/j.conbuildmat.2010.08.003

In the last decade, many researches were undertaken on the durability of ambient temperature cured resins used in construction [9,10], FRP reinforcements [11,12], and concrete/concrete or FRP/concrete bonded interfaces [13–23]. In these studies, property evolutions of specimens exposed to various accelerated ageing conditions were assessed, especially under freeze–thaw or dryheat cycles, immersion in salt or alkali solutions, salt fog spray, or under fixed hydrothermal environments; although these environments were not representative of the actual in-service conditions, they allowed to investigate the effects of selected or coupled parameters. As regards the bonded interfaces, substantial decreases in bond strength over ageing time (up to 40–50% loss in some cases) were observed under wet environments [13– 15,18–23], which was usually attributed to extensive moisture plasticization of the low Tg polymer adhesive and to additional breakage of interfacial bonds. However, a broad range of FRP and adhesive systems was used by the different authors as well as various bond test methods (pulloff, beam test, single or double-lap-joint shear tests, slant shear, etc.) which makes it difficult to compare the relative intensities of the property evolutions. It was shown indeed, by Momayez et al. [24] and Aiello and Leone [25], that the measured bond strength between the concrete substrate and repair materials is highly dependent on the test method, due to large variations in the stress field at the bonded interface; measured values were found to be up to 8 times larger from one method to the other

524

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

[24]. The selection of the test methods is therefore a key issue with regard to the pertinence of durability studies. If the pull-off test is easy to carry-out and is widely applied for quality control on working site, it is also known to be affected by small misalignment between the specimen and the loading axis [24,26]; stress concentrations due to partial coring of the test zone can also result in erroneous results and large scatters [26]. According to Aiello and Leone [25], the single lap shear test represents a good alternative to more complicated methods and could also be used for characterizations in the field. The present experimental approach aims at providing new insights or at least confirmation of previous trends regarding the hydrothermal ageing behaviour of FRP strengthened concrete specimens, and the influence of selected parameters. The influence of the test method is also investigated as explained in the next section.

2. Objectives and outlines of the experimental program In the present paper, the main results of two distinct experimental campaigns devoted to accelerated ageing of FRP strengthened concrete specimens are presented. Constant hydrothermal ageing conditions were intentionally chosen, i.e., a temperature of 40 °C and a relative humidity of at least 95%, in order to avoid complex coupling effects induced by either hydro/thermal cycling or specific aggressive environments (salt fog, salt or alkaline solutions). Those exposure conditions were not supposed to be representative of the in-service ageing environment, but were mainly intended to study moisture effects on the properties of the concrete/FRP bonded interfaces. The chosen temperature of 40 °C allowed to accelerate both the post-curing process of the epoxy adhesive system and the water sorption kinetics (in comparison to the same processes at room temperature), while remaining just below the usual glass transition temperature range of ambient curing epoxies used in construction. A first set of experiments was concerned with hydrothermal ageing of FRP strengthened concrete slabs and the assessment of adhesive bond property evolutions by means of pull-off tests. Ageing behaviours (over a period of 20 months) of several model concrete/composite interfaces were compared and analysed, in order to get information: – regarding the influence of various parameters on the bond durability, such as the concrete surface preparation, a superficial carbonation of the concrete substrate or the nature of the FRP system, – on mechanisms responsible for the property evolutions of concrete/FRP bonded interfaces under wet environments. In the second experimental campaign, accelerated ageing behaviours of both FRP strengthened concrete blocks and the constitutive materials taken separately (concrete, bulk epoxy adhesive and composite element) were investigated over a period of 13 months. Adhesive bond properties were assessed by means of two different techniques, i.e., the pull-off method and the singlelap-joint shear test, in order to get comparative data. The ‘‘sensitivity” of the two test methods was discussed in the light of apparent property evolutions of both the bonded interfaces and the separate constituents. As regards the shear tests, changes in the load transfer mechanism induced by humid ageing at the bonded interface were also investigated. For both experimental campaigns, the interval period between two test sessions was adjusted according to the observed evolutions of mechanical properties (see details in Table 4). The final periods of exposure (20 and 13 months, respectively) were depen-

dant on the number of ageing specimens which could be stored in the climatic chambers, hence on the size of the test specimens.

3. Durability of concrete/composite interfaces: first experimental study based on pull-off characterizations 3.1. Preparation of the test specimens In the framework of a first experimental campaign, specific test specimens based on FRP reinforced concrete slabs, and providing various model bonded interfaces were prepared. The preparation steps were as follows: – concrete slabs of dimensions 30  30  5 cm3 were cast from a Portland cement (CEM I 52.5 PMES from Lafarge, Le Havre, France) and silico-calcareous aggregates (0/4 and 5/20 mm), with a conventional water-to-cement ratio of 0.5. These specimens were then immersed in water for a maturation period of 28 days. The resulting material exhibited an average compressive strength of 35 ± 2 MPa, as determined by compression tests (EN 12390-3 European standard) on cylindrical 16  32 cm samples casted with the same concrete (three repeated tests). – these mature concrete slabs were then divided into two series which were subjected to specific surface preparations commonly used on working sites. Half of the specimens was sandblasted using a manual equipment, whereas the other half was ground with a rotary diamond bur tool. For both preparation methods, a 1–2 mm thick layer was removed from the concrete surfaces; however grinding led visually to a much smoother surface finish than sand-blasting. Machined surfaces were then carefully vacuum cleaned in order to remove dust particles. – for each of the previous series, half of the specimens was also subjected to a carbonation process: slabs were stored for 1 month under controlled air/CO2 atmosphere at 20 °C and 65% relative humidity, in order to form a carbonated surface layer of thickness 10 mm, – all abovementioned slabs were finally strengthened by two types of commercially available FRP systems: for each type of prepared concrete surface, half of the available specimens was reinforced using the wet lay-up process based on carbon fibre sheet (CFS), and the remaining half by pultruded unidirectional CFRP plates of thickness 1.2 mm and width 50 mm (Fig. 1a and b, respectively). A single composite layer was used in the two cases. Bonding was achieved at room temperature using two bi-component epoxy formulations, denoted A and B, which are prescribed for the CFS and CFRP plate systems, respectively. Implementation was carried-out by experienced staff from FRP supplier companies. Characteristics of the various materials used in the preparation of the FRP strengthened slabs of series A and B (concrete, commercial adhesives, composites reinforcements) are summarized in Tables 1–3. Finally, eight types of concrete/composite bonded interfaces were obtained, as listed below: – – – – – – – –

sand-blasted concrete/CFS, sand-blasted concrete/CFRP plate, ground concrete/CFS, ground concrete/CFRP plate, sand-blasted and carbonated concrete/CFS, sand-blasted and carbonated concrete/CFRP plate, ground and carbonated concrete/CFS, ground and carbonated concrete/CFRP plate.

525

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

a ‘‘complete” hardening of the epoxy adhesives (in terms of Shore stiffness) at 20 °C. We chose a longer period of 2 weeks for safety reasons. 3.2. Ageing conditions All strengthened slabs were exposed to saturated humidity in a climatic chamber (relative humidity of 95%, at least), at a temperature of 40 °C. Specimens were periodically removed from the chamber in order to perform mechanical tests on the bonded FRP overlays. Details of the interval periods between successive test sessions are given in Table 4. 3.3. Testing methods Adhesive bond characterizations were carried-out at room temperature by pull-off tests according to EN 1542 European standard [27]. A partial core was first drilled around the test zone, with an approximate depth of 4 mm within the concrete substrate. A cylindrical steel body of diameter 50 mm was then glued to the test zone using an epoxy adhesive. Finally, a tensile loading was applied to the steel body by mean of a dynamometer device (Fig. 1c), until debonding occurred. The adhesive bond strength or pull-off strength (MPa) was simply deduced by dividing the failure load (N) by the cross sectional area of the steel body (mm2). Series of six tests were systematically carried-out to obtain statistical information (average values and standard deviations). Tensile properties of the bulk epoxies A and B were also determined in the initial state, i.e. for unaged samples, according to EN ISO 527-1. Tests were performed on an Inströn testing machine with a displacement rate of 0.01 mm s1 (6 repeated tests). 3.4. Experimental results and discussions

Fig. 1. Concrete slabs strengthened either by bonded carbon fibre sheets (a) or by CFRP laminates (b), and dynamometric device for bond strength assessment (c).

All specimens were stored at room temperature for 2 weeks before the beginning of the test campaign, to allow polymerization of the epoxy adhesive to proceed. In the technical datasheets provided by suppliers, periods of 4–7 days were prescribed to achieve

Time evolutions of the pull-off strengths during ageing are depicted in Fig. 2a–d for the eight types of concrete/composite bonded interfaces previously introduced. Last data were collected after an ageing period of 20 months (628 days). Mean values are obtained from the average of six test results, and their standard deviations are also indicated on the figures. Table 5 makes it possible to compare the average values of the pull-off strength in the initial state and after 628 days ageing, and reports the corresponding variations. A large scatter is observed for bond strength values, which is a common feature for pull-off tests conducted on concrete substrates

Table 1 Description and properties of concrete materials used in the fabrication of the various series of FRP strengthened specimens (series A and B are considered in Section 3, while those of series C and D refer to Section 4). Concrete used for specimens of series A

Concrete used for specimens of series B

Concrete used for specimens of series C

Concrete used for specimens of series D

Composition

 CEM I 52.5 type cement  0/4 and 5/20 mm sillicocalcareous aggregates  W/C = 0.5

 CEM I 52.5 type cement  0/4 and 5/20 mm sillicocalcareous aggregates  W/C = 0.5

 CEM II 32.5 type cement  6/10 and 10/20 mm sillicocalcareous aggregates  W/C = 0.55

 CEM II 32.5 type cement  6/10 and 10/20 mm sillicocalcareous aggregates  W/C = 0.55

Surface treatments of concrete blocks

Four types of surface treatments: – Sand-blasting – Grinding – Sand-blasting + carbonation – Grinding + carbonation

Four types of surface treatments: – Sand-blasting – Grinding – Sandblasting + carbonation – Grinding + carbonation

Sand-blasting

Grinding

35 ± 2 MPa

32 ± 4 MPa

37 ± 5 MPa

Not available

3.6 ± 0.1 MPa

Not available

Mechanical properties (unaged samples)  Compressive strength in MPa (EN 35 ± 2 MPa 12390-3 standard)  Tensile strength from splitting tests in Not available MPa (EN 12390-6 standard)

526

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

Table 2 Description and properties of the various epoxy systems used in the present study (Systems A and B are considered in Section 3, systems C and D in Section 4). Em, rrm and erm stand respectively for the tensile Young’s modulus, and the ultimate stress and strain of the polymer adhesives. Epoxy A

Epoxy B

Epoxy C

Epoxy D

Description of the adhesive systems

 Bi-component, ambient curing  Wet lay-up process  Viscosity at 25 °C: 17,000 mPa s

 Bi-component, ambient curing  Bonding of CFRP plates  Thixotropic paste

 Bi-component, ambient curing  bonding of CFRP plates  thixotropic paste

 Bi-component, ambient curing  Wet lay-up process  Viscosity at 23 °C: 6000 mPa s

Main organic constituents in resin and hardener*

 Resin part

 Resin part

 Resin part

 Resin part

– DGEBA

– DGEBA

– Bisphenol A epichlorhydrin oxirane  Hardener part – Alkyletheramines – Diethylenetriamine (DETA)

 Hardener part – Polyamide resin

– Phtalate  Hardener part – Polyamide resin

– Benzylic alcohol

– Benzylic alcohol

– Alkylaminophenol – Triethylenetetramine (TETA)

– Bisphenol A epichlorhydrin oxirane  Hardener part – Triméthylhexane-1,6diamine

– Alkylaminophenol – Alkylaminophenol

Filler content in the mix (wt.%)** and nature of fillers

25% calcite, silica

50% calcite

50% calcite, silica, magnesite

32% calcite

Glass transition temperature*** (°C)

51 ± 2

50 ± 2

51 ± 2

46 ± 2

Mechanical properties (tension tests on control samples according to EN 527-1 standard)

Em = 2.3 ± 0.2 GPa rrm = 29.3 ± 1.2 MPa erm = 2.4 ± 0.3%

Em = 4.9 ± 0.2 GPa rrm = 29.5 ± 1.0 MPa erm = 0.7 ± 0.1%

Em = 4.9 ± 0.2 GPa rrm = 25 ± 1.0 MPa erm = 0.6 ± 0.1%

Em = 4.5 ± 0.2 GPa rrm = 49 ± 2.0 MPa erm = 1.5 ± 0.2%

3.76 ± 0.05

2.48 ± 0.02

2.49 ± 0.02

2.08 ± 0.03

3.98  107

3.17  107

3.21  107

1.06  107

Sorption characteristics at 40 °C and 95% RH****  Maximum mass uptake (%)  Diffusion coefficient (mm2/s) *

As indicated in the product datasheets. As determined after centrifugation of the formulated adhesive in solution with methyl ethyl ketone. As determined by DSC analyses (1.5 °C/min) on bulk epoxy samples cured at room temperature for 15 days. **** Mass uptake as measured on parallelelipedic samples using a precision balance, and diffusion coefficient as derived from the slope of the mass uptake versus (ageing time)1/2 curves according to Fick’s theory (see details in Section 4.3.1.2). **

***

Table 3 Description and properties of the various composite reinforcements used in the present study (Composites A and B are considered in Section 3, while composites C and D refer to Section 4). Ef, rrf and erf denote respectively the tensile Young’s modulus, and the ultimate stress and strain of the FRP reinforcements (in the 0° direction). Composite A

Composite B

Composite C

Composite D

Description of the FRP reinforcement

 Carbon fibre sheet (CFS) impregnated onsite  Bidirectional (70% warp fibres and 30% weft fibres)

 Pultruded CFRP Plate (carbon/epoxy)  Unidirectional (orientation 0°)

 Pultruded CFRP Plate (carbon/epoxy)  Unidirectional (orientation 0°)

 Carbon fibre sheet (CFS) impregnated onsite  Unidirectional (99% warp fibres)

Thickness (mm) Width of the plate or strip (mm)

0.27 (Fabrics), 0.43 (laminate)* –

1.2 (CFRP plate) 50

1.5 (CFRP plate) 100

0.13 (Fabrics), 1.0 (laminate)* 80

Fibre content (vol.%)

65% in the laminate*

>68% in the CFRP plate*

68% in the CFRP plate*

10% in the laminate*

Ef = 162 GPa

Ef = 26.5 GPa (for a 1 mm thick laminate)

rrf P 2800 MPa* erf P 1.6%*

rrf P 350 MPa* erf = 1.15%*

Mechanical properties tension tests (0° direction) for unaged samples according to EN 2561

*

*

*

Ef = 105 GPa (for a 0.43 mm thick laminate)

Ef = 160 GPa

rrf P 1400 MPa* erf not available

rrf P 3000 MPa* erf P 1.7%*

As indicated in the product datasheets. These values are either average values given by the suppliers, or guaranteed minimum values (preceded by P).

[26]. However, several trends can be drawn from these experiments since they were observed throughout the test campaign and appear to be significant with regard to standard deviations.

A focus on composite strengthened specimens prepared from non carbonated concrete surfaces (Fig. 2a and b), leads to the following remarks:

– –

             – – – – – – – –

 – – – –  – – –  – – – – – – – – – – – – –  – – – – –  – – – –  – – – –  – – –  –  – – – – – – – – – –           

– the surface preparation of concrete has a slight effect on the strength level of the bonded interfaces: bond strength values were globally higher with sand-blasted concrete surfaces, and this trend is observed for both initial and ageing specimens. As evoked previously, sand-blasting yields a rougher surface finish as compared to grinding. According to several authors, a substantial roughness of the cementitious substrate promotes mechanical anchoring of the polymer adhesive and tends to increase the contact surface area, both resulting in a higher bond strength [14,28,29], although other factors can be involved such as the presence of cracks on the machined concrete surfaces [30,31]. – hydrothermal ageing induces a progressive and significant decrease in the pull-off strengths of bonded interfaces based on CFS and CFRP strengthening systems. Similar detrimental effects were observed by several authors for analogous ageing conditions [14,19,21,23]. In the present study, the most detrimental effects are observed for ‘‘concrete/CFRP plate” interfaces, especially with the ground concrete substrate (see Table 2). The latter specimen exhibits indeed a drop of properties of approximately 58% after an ageing period of 628 days, whereas the other specimens are ‘‘limited” to a decrease of 46–49%; these relative variations should be considered with caution since large standard deviations are obtained on measured pull-off strengths. The presumed worst ageing behaviour of ‘‘concrete/CFRP plate” interfaces may be related to the high viscosity of epoxy adhesives used for the bonding of pultruded plates, compared to resins used in the wet lay-up process for impregnation of carbon fabrics (cf. Table 2). Indeed, a highly viscous or thixotropic texture of the polymer does not promote a deep penetration into the porous structure of concrete [32], which may shorten the diffusion path of water (from concrete) towards the interfacial region, as compared with the CFS case. – whereas the failure mode was initially cohesive within the concrete substrate for all specimens (Fig. 3a), an increasing number of mixed or adhesive failures was observed over ageing time, as shown in Fig. 3b and c. This is also consistent with a substantial deterioration of the adhesive bond or to a decrease in the epoxy joint properties due to humid ageing [23].

– values of the pull-off strength are globally higher than those of specimens prepared from non carbonated slabs, both in the initial and aged states. This result is not surprising since the carbonation process is known to decrease the porosity of the superficial concrete layer, hence leading to a significant improvement in the mechanical properties of the substrate [33]. – because concrete carbonation took place after the sand-blasting or grinding treatments, the influence of those surface treatments is not discernable any more in the present case, – a major feature is that the decrease of the bond strength during humid ageing is much more limited (11 to 21%) than for specimens based on non carbonated substrates. And even, no significant variation is observed for the ‘‘sand-blasted and carbonated concrete/CFRP plate” interface. In that case, CFRP strengthened specimens seem to perform slightly better than CFS strengthened ones, which is the opposite trend with respect to non carbonated slabs. This may be essentially explained by restricted water diffusion from the concrete substrate towards the adhesive joint, since the carbonation treatment has filled porosities of the superficial concrete layer to a large extend. Thus, in this case, sorption phenomena are mostly related to

                 – – – – – – – – – – – – – – –  – – – – – – – – – – – – – – – 0 12 20 39 56 75 91 110 137 166 201 250 334 444 549 628 T0 T1 T2 T3 T4 T5 T6 T7 T8 T9 T10 T11 T12 T13 T14 T15

Adhesive Epoxy: Concrete: bond: pullcompression tensile tests (6) off tests (6) tests (3)

527

When considering experimental data for strengthened specimens prepared from carbonated concrete slabs (Fig. 2c and d, and Table 2), it is found that:

0 9 23 37 53 83 119 148 175 265 353

Concrete: Concrete: compression tensile splitting tests (3) (3)

          

0 7 27 40 54 84 114 150 179 206 296 394

           

Composite: Adhesive bond: Epoxy: pull-off tests (6) tensile tensile and shear tests (3) tests (6) tests (3) Concrete: Concrete: compression tensile splitting tests (3) (3)

Exposure Type of test performed time (days) Exposure Type of test performed time (days)

Composite: Adhesive bond: Epoxy: pull-off tests (6) tensile tensile and shear tests (3) tests (6) tests (3)

Series D Series C

Test Series A and B session Exposure Type of test performed time (days)

Table 4 Outline of the experimental campaigns showing the details of the test sessions (with corresponding exposure times) and the various types of mechanical characterizations performed at each test session. Values in brackets represent the number of repetitions for each test performed.

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

528

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

Fig. 2. (a)–(d) Pull-off strength versus ageing time for the eight model concrete/composite interfaces. Mean values are averages of six test results.

Table 5 Average values of the pull-off strength in the initial state and after an ageing period of 628 days, and corresponding variations for the eight model bonded interfaces. Type of bonded interface

Initial strength (MPa)

Strength after 628 days ageing (MPa)

Variation (%)

Sand-blasted concrete/CFS Ground concrete/CFS Sand-blasted concrete/CFRP plate Ground concrete + CFRP plate Sand-blasted and carbonated concrete/CFS Ground and carbonated concrete/CFS Sand-blasted and carbonated concrete/CFRP plate Ground and carbonated concrete + CFRP laminate

3.3 ± 0.4 2.7 ± 0.5 3.0 ± 0.8 2.5 ± 0.8 3.8 ± 0.5

1.8 ± 0.3 1.4 ± 0.5 1.5 ± 0.6 1.0 ± 0.4 3.1 ± 0.5

46 48 49 58 18

3.7 ± 0.4 3.7 ± 1.4

2.9 ± 0.4 3.8 ± 0.9

21 +3

3.6 ± 0.8

3.2 ± 1.4

11

direct moisture diffusion from the saturated air environment towards the free edges of the adhesive joint (for the CFRP series) or towards the entire surface of the composite laminate (for the CFS series). This might explain the observed trend reversal. However, the main point is that all series based on carbonated substrates show globally a better ageing behaviour than their counterparts based on non carbonated slabs.

Differences in the ageing behaviours of specimens based on carbonated and non carbonated slabs are even more striking in Fig. 4, which shows a comparative histogram of the pull-off strength levels in the initial state and after 628 days ageing. This result provides interesting information on driving factors responsible for the deterioration process during hydrothermal ageing, and suggests that moisture diffusion from the concrete superficial layer (i.e., diffusion of interstitial pore solution) towards the bonded interface and the adhesive joint is involved, possibly in a larger measure than direct diffusion from the wet air environment. In the case of carbonated substrates, indeed, such a diffusion of the pore solution is significantly restrained due to the lower porosity of the 10 mm thick carbonated layer, hence the limited decrease in pull-off strength over ageing time. Further investigation is needed to confirm the direct causality.

4. Durability of concrete/composite interfaces: second experimental study based on both pull-off and shear characterizations In the framework of a second experimental campaign, complementary investigations were carried-out on the accelerated ageing behaviour of concrete/composite bonded interfaces. In particular, it was proposed:

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

529

Fig. 3. (a) and (b) Examples of mixed failure mode observed for model concrete/composite interfaces subjected to hydrothermal ageing (characterization by pull-off test).

Fig. 4. Comparative histogram of the pull-off strength values in the initial state and after 628 days of ageing, for the eight concrete/composites bonded interfaces under study.

530

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

– to use two characterization methods of the adhesive bond, i.e., pull-off and shear loading tests, in order to provide comparative data and evaluate the influence of the testing conditions or the sensitivity of the methods, – to study two types of FRP strengthening systems, i.e., unidirectional carbon fibre sheets (CFS) and CFRP plates, which were not provided by the same suppliers as materials used in Section 3, – to assess property changes of the constitutive materials (concrete, epoxy adhesive, composite reinforcement itself) due to accelerated ageing, in order to evaluate their relative contributions to the adhesive bond characteristics. 4.1. Preparation of the test specimens A specific design was chosen for the FRP strengthened concrete elements, which makes it possible to perform, on a same specimen, compression testing of concrete as well as pull-off and shear testing of the concrete/FRP bonded interface (Fig. 5a). Test specimens were prepared as follows – concrete blocks of dimensions 21  21  41 cm3 were cast from a CEM II 32.5 cement and silico-calcareous aggregates (6–10 and 10–20 mm), with a water-to-cement ratio of 0.55. A maturation period of at least 28 days was observed. – for half of these concrete blocks, a sand-blasting treatment was made on the upper side and strips of CFRP plates of width 100 mm and thickness 1.5 mm were bonded using an ambient temperature curing epoxy adhesive (denoted epoxy C), which was prescribed by the manufacturer. On each specimen, a testing zone was intended for a single-lap-joint shear test: a single

composite layer was glued to the concrete prism, with a total bonded length of 200 mm as shown in Fig. 5a. In order to prevent or limit edge effects, the lap joint started 50 mm away from the front side of the concrete block (see detail in Fig. 6c). A second testing zone devoted to the pull-off tests was installed at the back of the specimens. Here again, implementation was performed by a specialized staff from the FRP supplier. The aspect of the final specimens is shown in Fig. 5b. – for the other half concrete blocks, a grinding treatment was achieved on the upper surface, and CFS strips of width 80 mm were bonded in a similar way, using a second commercial epoxy adhesive (denoted epoxy D). Final CFS strengthened specimens are seen in Fig. 5c. Additional samples (dumbbell shaped specimens or parallelepipeds of dimension 5  5  40 mm3) of the bulk epoxy adhesives C and D were also prepared in order to study separately the ageing behaviours of the pure polymer materials. All specimens were stored for 2 weeks at room temperature before starting the test campaign. Characteristics of the various materials used in the preparation of the FRP strengthened concrete specimens of series C and D (concrete batches, commercial adhesives, composites reinforcements) are listed in Tables 1–3. 4.2. Ageing conditions The same accelerated ageing conditions were used as in the previous experimental campaign, i.e., a temperature of 40 °C and a relative humidity higher than 95%. All specimens were stored on

Fig. 5. (a) Design of the composite strengthened test specimens and series of specimens reinforced by CFS (b) or by CFRP plates (c).

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

@C @2C ¼ D 2 @t @x

531

ð1Þ

An approximate solution of Eq. (1) is given by:

mt 4 ¼ m1 h

Fig. 6. (a) Picture of the shear test device, (b) clamping conditions of the concrete block, and (c) locations of the five strain gages along the bonded interface (distances are expressed in mm from the free edge of the adhesive joint). It should be noted that the free edge of the adhesive layer is located 50 mm away from the concrete corner.

shelves in a 10 m2 climatic room, and some of them were removed at periodical terms for characterizations. Details of the interval periods between successive test sessions (T0–T11) are given in Table 4, together with the types of mechanical tests performed at each session. 4.3. Characterization methods 4.3.1. Characterization of the constitutive materials (concrete, bulk adhesives, CFRP) 4.3.1.1. Characterization of the concrete material. To assess possible changes in the concrete properties during hydrothermal ageing, compression tests were carried-out on core samples (diameter 110 mm, length 210 mm) removed from the ageing composites strengthened specimens, according to EN 12390-3 European standard [34]. Compressive strength values were averaged on three repeated tests. In addition, the initial tensile strength of concrete (at test session T0) was assessed by means of tensile splitting tests according to EN 12390-6 European standard [35] (average on three repeated tests). However, this test was not used to assess the property-evolution of concrete during the experimental campaign. 4.3.1.2. Characterizations on bulk adhesives. The sorption kinetics of the various epoxy adhesives were investigated by monitoring mass uptakes of parallelelipedic samples during hydrothermal ageing. Samples were periodically removed from the climatic chamber and weighed using a precision electronic balance (Mettler AE 240). Moisture diffusion coefficients were then derived according to Fick’s theory; the second law provides the change in the concentration C of diffusing species (water) at a distance x from the contacting surface, as a function of time t and diffusion coefficient D:

rffiffiffiffiffiffi Dt

p

ð2Þ

where mt is the mass uptake of the polymer at time t, m1 is the mass uptake at equilibrium, and h is the sample thickness. Thermal analyses were also carried-out by modulated differential scanning calorimetry (MDSC), in order to quantify changes in the glass transition temperatures Tg induced by ageing. Such experiments were made using a MDSC Q100 apparatus from TA Instrument. Small chips of polymer (3–4 mg) were removed from the sample surface and set in hermetic aluminium pans, which were then heated from 30 to 200 °C at a linear heating rate of 1.5 °C/ min with a superimposed harmonic signal (amplitude of 0.4 °C and period of 60 s). Temperature modulation made it possible to separate the contributions of reversible phenomena (glass transition) from that of kinetic processes (structural relaxation, water evaporation) to the overall heat flow signal. The glass transition phenomenon resulted in a jump in the specific heat capacity on the reversible heat flow thermogram, and the Tg value was determined using the midpoint-by-half-height method of the Universal Analysis software. In addition, tensile tests (EN ISO 527-1 [36]) were performed on the dumbbell shaped polymer samples in order to assess eventual changes in the mechanical behaviours of the bulk epoxies due to hydrothermal ageing. Tests were carried-out with an Instron testing machine at a displacement rate of 0.01 mm/min. Deformation of materials was monitored by strain gages glued on the test specimens.

4.3.1.3. Characterization of the CFRP plates. Few tensile tests were conducted on the CFRP plates (in the fibre longitudinal direction) according to EN ISO 527-5 [37] in order to monitor eventual changes in the material Young’s modulus induced by accelerated ageing. Tests were not conducted up to the failure. The CFS composite formed by the wet-lay-up process was only tested in the initial state (unaged material). Unfortunately, data regarding ageing samples are not available.

4.3.2. Pull-off and shear characterizations of the concrete/FRP adhesive bond Pull-off tests were carried-out in a similar way as in Section 3.3, according to EN 1542 European standard. Mean values of the pulloff strength were calculated from six repeated tests. Single-lap-joint shear tests were performed at room temperature using a home-made tension machine equipped with a hydraulic jack of capacity 100 kN (Fig. 6a), which has been extensively described in several papers [38–41]. Very similar test set-ups are being developed by other teams [25,42]. In our case, adequate clamping conditions of the concrete blocks (Fig. 6b) have been chosen to prevent any rotation of the specimen during the test. Tests were conducted at a constant displacement rate of 6 lm/s and the applied force and displacement of the grip (using a LVDT sensor) were monitored continuously, until complete debonding of the FRP overlay occurred. Mean values of the maximum shear load were averaged on three repeated tests. In order to determine the strain profiles along the bonded interface during the shear test as well as the load transfer length, specimens were instrumented by five strain gages glued to the surface of the composite material, as shown in Fig. 6c.

532

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

4.4. Experimental results and discussions Next sections will focus respectively on the ageing behaviours of the constitutive materials (concrete, bulk polymer adhesives, CFRP plate), the concrete–CFRP interface, and finally the concrete–CFS interface. 4.4.1. Ageing behaviour of the constitutive materials 4.4.1.1. Ageing behaviour of the concrete material. Fig. 7 depicts the evolutions versus ageing time of the concrete compressive strength, as measured on core samples removed from the CFRP and CFS strengthened specimens. A slight difference in the initial strength levels is observed for the two types of specimens, since concrete blocks were cast from two different batches of fresh mix (compressive strengths of 32 ± 4 MPa and 37 ± 5 MPa for initial concrete specimens of batches C and D, respectively). Then, over ageing time, a slight and progressive increase in the strength values is obtained for all concrete specimens, which was expected

due to the continuation of the hydration process in the warm and wet ageing environment. After 13 months of exposure, the compressive strengths were respectively 40.0 ± 0.3 MPa and 43 ± 3 MPa for concrete specimens of batches C and D. 4.4.1.2. Ageing behaviour of the bulk epoxy adhesives. Moisture uptake characteristics of the bulk adhesives C and D subjected to hydrothermal ageing at 40 °C and 95% R.H. are reported in Table 2. System C used for the bonding of CFRP plates exhibits both the highest mass uptake at saturation and the highest diffusion coefficient among the two adhesives, hence the fastest sorption kinetics. With regard to DSC experiments, evolutions of the glass transition temperatures of the two systems versus ageing time are depicted in Fig. 8. Apparent evolutions are the resultant of a competition between two antagonistic mechanisms, i.e., the ongoing cross-linking process at 40 °C and the plasticization effect due to moisture ingress into the polymer network: – systems C and D both exhibit an initial increase in Tg (from 50 to 57 °C for epoxy C, and from 46 to 67 °C for epoxy D) due to the rise of the storage temperature from ambient to 40 °C, which promotes a continuation of the cross-linking process, – for epoxy C, the plasticization effect becomes rapidly predominant and a sharp drop of Tg, down to 31 °C, is observed after few hours ageing. Such a steep evolution is consistent with the initially fast sorption kinetics shown previously. Then, after 12 days ageing, a re-increase in Tg is observed, possibly due to the slowing down of the sorption kinetics and again a predominance of cross-linking,

Fig. 7. Evolutions of the concrete compressive strength versus ageing time for cylindrical core samples removed from the CFRP plate and CFS strengthened specimens.

Fig. 8. Evolutions of the glass transition temperature Tg versus ageing time for epoxies C and D (modulated DSC analyses with a heating rate of 1.5 °C min1).

Fig. 9. Tensile loading tests for bulk epoxies C (a) and D (b). Stress–strain curves are plotted for initial samples and specimens aged up to 8 months at 40 °C and 95% R.H.

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

533

– with regard to epoxy D, a slight plasticization effect is observed with a very limited decrease in Tg down to 63 °C and a stabilisation around this value. Such behaviour is in line with the lower moisture uptake characteristics observed for this matrix system.

4.4.1.3. Ageing behaviour of the CFRP plates. Tensile tests performed on the pultruded CFRP plates provided a value of 162 GPa for the longitudinal Young’s modulus and did not reveal any significant evolution of the material stiffness after an ageing period of 4 months.

Fig. 9 shows the stress–strain curves for epoxies C and D submitted to tensile loading, both in the initial state and after various ageing periods:

4.4.2. Ageing behaviour of the concrete–CFRP plate interface Results of the pull-off and shear loading characterizations regarding specimens strengthened with CFRP plates (bonded with epoxy C) are presented in Figs. 10 and 11, respectively. Evolutions of the ultimate strength (or capacity) and evolutions of the failure mode induced by accelerated ageing are illustrated for each test configuration. As regards the failure mode, the following remarks can be done:

– initially, the two bulk adhesives both exhibit an elastic behaviour; however, ultimate mechanical properties of system C (tensile strength rrm = 25 ± 1 MPa, ultimate strain erm = 0.6 ± 0.1%, Young’s modulus Em = 4.9 ± 0.2 GPa) are much lower than those of system D (rrm = 49 ± 2 MPa, erm = 1.5 ± 0.2% and Em = 4.5 ± 0.2 GPa), – during hydrothermal ageing, both mechanical behaviours evolve and become elasto-plastic, with a progressive decrease in the tensile strength and stiffness and an overall increase in ductility. After 8 months ageing, the mechanical properties are as follows: for system C (rrm = 6.6 ± 0.5 MPa, erm = 1.4 ± 0.3%, Em = 1.3 ± 0.2 GPa) and for system D (rrm = 26 ± 1 MPa, erm = 1.9 ± 0.4% and Em = 3.1 ± 0.2 GPa, respectively). It is to note that the two epoxy adhesives are not affected in the same degree by hydrothermal ageing: tensile strength and elastic modulus are both decreased by a factor 3.8 for epoxy C, whereas strength is roughly decreased by 2 and modulus by 1.5 for system D. Hence, epoxy C is found to be more sensitive to the chosen ageing conditions, which is in line with the larger moisture uptake and plasticization effect shown previously. In addition, for this matrix system, the tensile strength of the 8- month-aged specimens becomes close to that of the concrete material as evaluated from tensile splitting tests (3.6 ± 0.1 MPa).

– a very specific phenomenon is observed for specimens tested in shear at session T0 (unaged state): an adhesive failure mode at the polymer/FRP interface is obtained which is characteristic of a poor adherence of epoxy C on the composite plate (Fig. 11c). It may be explained by a relatively low cross-linking degree of epoxy C due to an incomplete curing at room temperature. Nevertheless, specimens tested in shear at session T1, after 7 days exposure at 40 °C and 95% RH, show again a typical concrete failure mode (Fig. 11b and d) indicating that the level of adherence has increased due to a post-curing effect of epoxy C. This is consistent with the rise in the glass transition temperature of epoxy C observed during the first days of exposure, as shown by DSC experiments (cf. Fig. 8). This initial phenomenon was not observed in the pull-off configuration, neither for the other epoxies considered in this study. – then, for both test methods, it is found that exposure at 40 °C and 95% R.H. leads to a progressive evolution of the failure mode from a pure concrete failure (as shown in Figs. 10c and 11d for pull-off and shear series, respectively) towards a cohe-

Fig. 10. Pull-off results for concrete/CFRP interfaces: (a) pull-off strength versus ageing time and (b) histogram of the fracture modes. Pictures of fractured surfaces for initial (c) and aged (d) specimens (tests sessions T0 and T10, respectively).

534

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

Fig. 11. Results of the shear tests for concrete/CFRP interfaces: (a) maximum shear load versus ageing time and (b) histogram of the fracture modes. Pictures of fractured surfaces for specimens tested in shear at sessions T0, T1 and T10 (c–e).

sive failure within the epoxy joint (Figs. 10d and 11e). Such a trend can be correlated to the large decrease in the mechanical properties of ageing epoxy C, since the other constitutive materials (concrete and composite) do not exhibit remarkable changes in their characteristics, as shown in Section 4.4.1. With regard to the mechanical properties, the following evolutions can be pointed out: – in the case of pull-off tests, a decrease in strength is observed (Fig. 10a), which is in fair agreement with the evolution of the failure mode, – on the contrary, for the shear tests, no significant evolution of the maximum shear load was detected (Fig. 11a), which seems contradictory, at a first approach, with the observed change in failure mode. In order to clear up this latter issue and explain why the shear capacity remains unchanged whilst the failure mode evolves, signals provided by the strain gages during shear experiments were examined. Fig. 12 shows the strain profiles of the CFRP plate along the bonded interface, at various load levels. Diagrams are plotted for both the initial and aged specimens (test sessions T1 and T10, respectively). Those profiles correspond to a single specimen, however very similar diagrams were obtained for the other specimens of the same test sessions. With regard to the strain profile of the initial specimen (Fig. 12a), it is found that shear loads are mainly transferred through the initial part of the adhesive joint, near the free edge. A transfer length of approximately 130 mm can be defined, beyond which the joint is not solicited (except in the damage domain at high load levels).

From Fig. 12b, it is found that hydrothermal ageing strongly affects the strain profile along the bonded interface and leads to a significant increase in the transfer length (>200 mm). Since transfer lengths are determined graphically, it is difficult to assess their level of uncertainty; nevertheless, from the various repeated tests, it is found that this apparent increase in the transfer length value is quite significant and representative. Such an evolution means that a more uniform load transfer is achieved over the entire joint length; for a given level of applied load, one can notice a substantial decrease in the strain values near the edge of the adhesive joint, as compared to the unaged specimen. It is proposed that such a modification of the load transfer mechanism during accelerated ageing results from the large change in the mechanical properties of the epoxy adhesive C, especially the decrease in the polymer stiffness and the enhanced ductility that both facilitate plastic deformation of the joint. This hypothesis is supported by a finite element modelling which will be presented in a next paper. The fact that the maximum load capacity (as measured by the single-lap-joint shear test, cf. Fig. 11a) is little affected by humid ageing could be explained by the evolution of the interfacial load transfer, which roughly counterbalances the effect of the joint weakening. Similar effects were reported by Leone et al. [43], in the case of CFRP strengthened specimens tested under doublelap-joint shear configuration at elevated temperature. A rise in the test temperature led to a large increase in the transfer length, while the maximum load remained constant. However, once the failure is initiated, crack propagation occurs in the weakest part of the bonded assembly which becomes progressively the aged epoxy material, hence the observed change in the fracture mode. Additionally, the level of mobilization of the CFRP plate in the shear tests can be assessed, by comparing the rupture strain of the CFRP plate (1.6% according to the product datasheet, with neg-

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

Fig. 12. Strain profiles of the composite along the bonded interface at various load levels during the shear test, for initial (a) and aged (b) CFRP strengthened specimens. The two profiles were recorded at test sessions T1 and T10, respectively (see Table 4).

ligible evolution during ageing) and the ultimate strain of the adhesive bond as provided by strain gages near the free edge of the joint (0.23% and 0.22% in the initial and aged states, respectively – cf. Fig. 12). It leads to a level of mobilization around 14% in the two cases, which is quite low. 4.4.3. Ageing behaviour of the concrete-CFS interface With regard to specimens strengthened with CFS (bonded with epoxy D), results of the pull-off and shear loading characterizations are presented in Figs. 13 and 14, in terms of ultimate strength (or capacity) and failure mode evolutions during hydrothermal ageing: – in that case, pull-off experiments do not reveal any significant change in the pull-of strength (Fig. 13a) nor any evolution of the failure mode, which always remains cohesive within the concrete substrate (Fig. 13b and c), – as regards the shear experiments, it is to note that the maximum shear load also remains constant over ageing time (Fig. 14a). However, an evolution of the failure mode is observed from a concrete failure towards a failure in the glue or even an interfacial failure (Fig. 14b–d), which is a direct evidence for substantial deterioration of the adhesive bond during ageing. The fact that the maximum shear load is not affected despite changes in the adhesive bond can be discussed again considering the strain profiles presented in Fig. 15. In this case, the transfer length is much shorter than in the case of the CFRP plates. Near the free edge of the adhesive joint, the strain gage alignment is not tight enough to provide clear information about the evolution of this transfer length during ageing. How-

535

Fig. 13. Pull-off results for concrete/CFS interfaces: (a) pull-off strength versus ageing time and pictures of fractured surfaces for initial (b) and aged (c) specimens (test sessions T0 and T11, respectively).

ever, some differences can be noticed between the two profiles at high load levels: (i) for the unaged specimen, one can observe a domain of constant deformation (horizontal line) near the free edge of the adhesive joint, which corresponds to a partial debonding. The debonded length seems to increase progressively as the test goes on, and the load is redistributed over a 70– 90 mm transfer length on the remaining bonded segment. (ii) Differently, the ageing specimen exhibits higher strain levels in the free edge region and there is no clear indication of a gradual debonding process. As the test goes on, the load seems progressively redistributed over a larger length of the lap joint. These features suggest the existence of plastic deformations of the polymer joint, favored by the increased ductility and reduced stiffness of aged epoxy D (cf. Fig. 9). This may possibly explain the constancy of the ultimate load as determined by shear tests. In addition, the level of mobilization of the CFS reinforcement can be assessed again, by comparing the ultimate strain of the CFS laminate (1.15% as reported by the manufacturer, and supposed to be little affected by ageing to a first approach), and the ultimate strain of the adhesive bond provided by strain gages (respectively 0.67% and 0.82% in the initial and aged states – cf. Fig. 15). It provides levels of mobilization of 58% and 71% for the initial and aged specimens, respectively. These levels are globally much higher than those obtained for CFRP specimens, and an apparent increase is observed after ageing. Another important finding is that the pull-off test may be less sensitive that the shear test since no evolution of the failure mode

536

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

Fig. 14. Results of shear tests for concrete/CFS interfaces: (a) Maximum shear load versus ageing time and (b) histogram of the fracture modes. Pictures of fractured surfaces for initial (c) and aged (d) specimens (test sessions T0 and T11, respectively).

is detected with the first method. It is believed that the pull-off test is able to discriminate the ageing behaviour of the adhesive bond (in terms of strength or fracture mode), only if the polymer adhesive exhibits large property variations and/or if its tensile strength becomes the same order as that of the concrete substrate. This was the case for epoxy C in the previous section, and probably for epoxy A and B in the first experimental campaign (since mass uptake behaviours were very similar to that of epoxy C, as reported in Table 1), but not for matrix system D in the present section. From these results, it is proposed that test methods based on shear solicitation should be promoted for the characterization of adhesively bonded joints. A portable device allowing to perform in situ shear characterizations or quality control on working sites is suggested as an alternative to the pull-off method. Moreover, the ageing behaviour of concrete/composite interfaces exposed to 40 °C and 95% is found to be primarily dependant on the moisture sensitivity and plasticization phenomenon of the epoxy system, in agreement with other authors [21,23]. Considering the reported chemical compositions given in Table 2 and the various experimental characterizations, it is suggested that epoxy systems based on polyamide hardeners (i.e., containing amide groups AC(@O)ANHA, such as epoxy B and C) or on alkyletheramine hardeners (i.e., containing ether groups RAOAR0 , such as epoxy A) may be more sensitive to hydrothermal ageing than systems based on conventional diamine hardeners (epoxy D). Indeed, extensive sorption phenomena and large property evolutions (cf. Table 2 and Fig. 9, respectively) are observed for these systems, probably due to a higher polarity of the polymer chains, and thus

a pronounced hydrophilic character. Further work is needed on this point. 5. Conclusion In the first part of this study, the ageing behaviour of model concrete-composite interfaces was investigated and the influence of several parameters was evaluated using pull-off characterizations. It was found that: – humid ageing causes a progressive and significant decrease in the pull-off strength of the bonded interfaces for CFS and CFRP strengthened specimens prepared from non carbonated concrete slabs. Moreover, the failure mode evolves from a substrate failure towards a mixed or interfacial failure, – the surface treatment of the concrete slabs has a slight effect on mechanical tests, since higher pull-off strengths are globally obtained with sand-blasting compared to grinding, – for strengthened specimens prepared from carbonated concrete slabs, the pull-off properties were only little affected by humid ageing. This result suggests that deterioration observed for non carbonated specimens is caused at a large extend, by moisture diffusion from the concrete surface towards the adhesively bonded joint. Such a diffusion process is restrained in the case of carbonated concrete substrates. In the second part, the ageing behaviour of CFS and CFRP plate strengthened concrete blocks was assessed using both pull-off and shear loading characterizations.

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

537

Accelerated ageing behaviours described in this study are only valid for the chosen exposure conditions. They give indications on the possible deterioration mechanisms under wet environments, but do not provide direct correlation with durability of the adhesive bond under actual in-service conditions. It is to note that additional protective coatings (acrylic systems or hydraulic binders) are usually prescribed for outdoor exposure conditions. Acknowledgments Authors wish to acknowledge the technical staffs of the ‘‘Laboratoire Régional des Ponts et Chaussées d’Autun” and the ‘‘Laboratoire Régional de l’Est Parisien” for their contributions to the experimental campaigns. References

Fig. 15. Strain profiles of the composite along the bonded interface at various load levels during the shear test, for initial (a) and aged (b) CFS strengthened specimens. The two profiles were recorded at test sessions T0 and T10, respectively (see Table 4).

– epoxy adhesives used for the bonding exhibited a decrease in mechanical strength during humid ageing, as well as a pronounced elasto-plastic behaviour. An extensive plasticization of the polymer network (evidenced by a sharp drop of Tg) was also observed for the epoxy system used to bond CFRP materials. It is to note that adhesive systems containing polar constituents (hardener with ether or amide groups) seem more sensitive to sorption phenomena and subsequent plasticization effects than conventional epoxy/diamine systems, – shear loading tests revealed an evolution of the failure mode, from a substrate failure towards a cohesive failure within the polymer joint in the case of CFRP strengthened specimens, and even towards interfacial failure in the case of CFS reinforced blocks. Therefore, it is proposed that ageing mechanism is mainly based on polymer plasticization in the first case, and both on plasticization and bond degradation in the latter case, all induced by water diffusion within the interfacial areas and the overall joint. Despite these changes in the failure mode, no evolution of the maximum shear loads was observed during ageing regardless the type of specimen. Strain profiles provided by strain gages revealed that the decrease in the joint properties was counterbalanced by a redistribution of the load transfer along the bonded interface, possibly resulting in an increased transfer length and local plastic deformations at the front of the adhesive joint, – finally, results of the pull-off characterizations were not always consistent with those of the shear experiments; it is believed that the latter method has a higher sensitivity and should be promoted for the characterization of adhesively bonded joints in construction.

[1] Van Den Einde L, Zhao L, Seiblec F. Use of FRP composites in civil structural applications. Constr Build Mater 2003;17:389–403. [2] Karbhari VM. Fiber reinforced composite bridge systems–transition from the laboratory to the field. Compos Struct 2004;6(1–4):5–16. [3] Pendhari SS, Kant T, Desai YM. Application of polymer composites in civil construction: a general review. Compos Struct 2008;84:114–24. [4] Hamelin P. Renforcement des ouvrages d’art par matériaux composites. Techniques de l’Ingénieur 2002; AM5615: 1-10 [in French]. [5] Karbhari VM, Chin JW, Hunston D, Benmokrane B, Juska T, Morgan R, et al. Durability gap analysis for fiber-reinforced polymer composites in civil infrastructure. J Compos Construct 2003;7(3):238–47. [6] Gangarao HVS, Taly N, Vijay PV. Reinforced concrete design with FRP composites. CRC Press: Taylor and Francis Group; 2006. [7] Hamelin P. AFGC: recommendations concerning repairing and strengthening of concrete structures by composite materials. In: Teng JG, (Ed.), FRP composites in civil engineering, vol. 1; 2001. p. 165–75. [8] Karbhari VM, Abanilla MA. Design factors, reliability, and durability prediction of wet layup carbon/epoxy used in external strengthening. Composites Part B 2007;38:10–23. [9] Chin JW, Aouadi K, Haight MR, Hugues WL, Nguyen T. Effects of water, salt solution and simulated concrete pore solution on the properties of composite matrix resins used in civil engineering applications. Polym Compos 2001;22:282–97. [10] Yang QA, Xian GJ, Karbhari VM. Hygrothermal ageing of an epoxy adhesive used in FRP strengthening of concrete. J Appl Polym Sci 2008;107(4):2607–17. [11] Hulatt J, Hollaway L, Thorne A. Preliminary investigations on the environmental effects on new heavyweight fabrics for use in civil engineering. Composites Part B 2002;33:407–14. [12] Abanilla MA, Li Y, Karbhari VM. Durability characterization of wet layup graphite/epoxy composites used in external strengthening. Composites Part B 2006;37(2–3):200–12. [13] Frigione M, Aiello MA, Naddeo C. Water effects on the bond strength of concrete/concrete adhesive joints. Constr Build Mater 2006;20:957–70. [14] Wisner G, Böhm S, Dilger K. Polymer adhesive bonding of ultra-high performance concrete (UHPC). Proc Adhes Soc Meet 2009;88–90:1086–9506. [15] Toutanji HA, Gomez W. Durability characteristics of concrete beams externally bonded with FRP composite sheets. Cem Concr Compos 1997;19:351–8. [16] Mukhopadhyaya P, Swarmy RN, Lynscale CJ. Influence of aggressive exposure conditions on the behaviour of adhesive bonded concrete–GFRP joints. Constr Build Mater 1998;12:427–46. [17] Aiello MA, Focacci F, Nanni A. Effects of thermal loads on concrete cover of fiber-reinforced polymer reinforced elements: theoretical and experimental analysis. ACI Mater J 2001;98(4):332–9. [18] Malvar LJ, Joshi NR, Bearn JA, Novinson T. Environmental effects on the short term bond of carbon fiber reinforced polymer (CFRP) composites. ASCE J Conpos Construct 2003;7(1):58–63. [19] Grace NF, Singh SB. Durability of carbon fiber-reinforced polymer strengthened concrete beams. ACI Struct J 2005;102(1):40–51. [20] Silva MAG, Biscaia H. Degradation of bond between FRP and RC beams. Compos Struct 2008;85:164–74. [21] Marouani S, Curtil L, Hamelin P. Composites realized by hand lay-up process in a civil environment: initial properties and durability. Mater Struct 2008;41(5):831–51. [22] Gavari MM, Rad AY, Gavari MM. Environmental durability of materials and bonded joints involving fiber reinforced polymers and concrete. In: IVth International conference on times of polymers (top) and composites, AIP conference proceedings, vol. 1042; 2008. p. 187–9. [23] Karbhari VM, Gosh K. Comparative durability evaluation of ambient temperature cured externally bonded CFRP and GFRP composites for repair bridges. Composites Part A 2009;40:1353–63. [24] Momayeza A, Ehsanib MR, Ramezanianpoura AA, Rajaie H. Comparison of methods for evaluating bond strength between concrete substrate and repair materials. Cem Concr Res 2005;35:748–57.

538

K. Benzarti et al. / Construction and Building Materials 25 (2011) 523–538

[25] Aiello MA, Leone M. Interface analysis between FRP RBR system and concrete. Composites Part B 2008;39:618–26. [26] Bonaldo E, Barros JAO, Lourenço PB. Bond characterization between concrete substrate and repairing SFRC using pull-off testing. Int J Adhes Adhes 2005;25:463–74. [27] EN 1542. Products and systems for the protection and repair of concrete structures. Test methods. Measurement of bond strength by pull-off; 1999. p. 1–10. [28] Abu-Tair AI, Lavery D, Nadjai A, Rigden SR, Ahmed TMA. A new method for evaluating the surface roughness of concrete cut for repair or strengthening. Constr Build Mater 2000;14:171–6. [29] Toutanji H, Ortiz G. The effect of surface preparation on the bond interface between FRP sheets and concrete members. Compos Struct 2001;53:457–62. [30] Garbacz A, Courard L, Kostana K. Characterization of concrete surface roughness and its relation to adhesion in repair systems. Mater Character 2006;56:281–9. [31] Karbhari VM, Navada R. Investigation of durability and surface preparation associated defect criticality of composites bonded to concrete. Composites: Part A 2008;39:997–1006. [32] Djouani F, Connan C, Delamar M, Chehimi MM, Benzarti K. Cement paste– epoxy adhesive interactions. Construct Build Mater 2010;25:411–23. [33] Jerga J. Physico-mechanical properties of carbonated concrete. Constr Build Mater 2004;18:645–52. [34] NF EN 12390-3. Testing hardened concrete. Part 3: Compressive strength of test specimens; 2003.

[35] NF EN 12390-6. Testing hardened concrete. Part 6: tensile splitting strength of test specimens; 2000. [36] EN ISO 527-1. Plastics – determination of tensile properties – Part 1: general principles; 1993. [37] EN ISO 527-5. Plastics – determination of tensile properties – Part 5: test conditions for unidirectional fibre-reinforced plastic composites; 2009. [38] Chataigner S, Caron JF, Benzarti K, Quiertant M, Aubagnac C. Characterization of composite-to-concrete interface: use of the cohesive zone approach. In: Proceedings of the fourth international conference on FRP composites in civil engineering – CICE2008. Switzerland: Zurich; 2008 [CD-Rom paper]. [39] Chataigner S, Caron JF, Benzarti K, Quiertant M, Aubagnac C. Characterization of composite to concrete bonded interface. Description of the single lap shear test. Eur J Environ Civil Eng 2009;13(9):1073–82. [40] Chataigner S, Caron JF, Benzarti K, Quiertant M, Aubagnac C. Use of a single lap shear test to characterize composite to concrete or steel bonded interfaces. Construct Build Mater 2010;25:468–78. [41] Benzarti K, Quiertant M, Aubagnac C, Chataigner S, Nishizaki I, Kato Y. Durability of CFRP strengthened concrete structures under accelerated or environmental ageing conditions. In: Concrete repair, rehabilitation and retrofitting II. Boca Raton: CRC Press-Taylor & Francis Group; 2009. p. 421–2. [42] Mazzotti C, Savoia M, Ferracuti M. A new single-shear set-up for stable debonding of FRP–concrete joints. Constr Build Mater 2009;23:1529–37. [43] Leone M, Matthys S, Aiello M-A. Effect of elevated service temperature on bond between FRP EBR systems and concrete. Composites Part B 2009;40:85–93.